Sources and potential transmission of Clostridium

Sources and potential transmission of Clostridium

difficile in the community of Western Australia

Su Chen Lim

BSc, MInfDis

This thesis is presented for the degree of Doctor of Philosophy of The University of

Western Australia

School of Biomedical Sciences

Microbiology and Immunology

2018

ii

THESIS DECLARATION

I, Su Chen Lim, certify that:

This thesis has been substantially accomplished during enrolment in the degree.

This thesis does not contain material which has been submitted for the award of any

other degree or diploma in my name, in any university or other tertiary institution.

No part of this work will, in the future, be used in a submission in my name, for any

other degree or diploma in any university or other tertiary institution without the prior

approval of The University of Western Australia and, where applicable, any partner

institution responsible for the joint-award of this degree.

This thesis does not contain any material previously published or written by another

person, except where due reference has been made in the text and, where relevant, in the

Declaration that follows.

The work(s) are not in any way a violation or infringement of any copyright, trademark,

patent, or other rights whatsoever of any person.

The following approvals were obtained prior to commencing the relevant work

described in the thesis: human research ethics approval from The University of Western

Australia for a retrospective case-control study (RA/4/1/8222).

This thesis contains published work and/or work prepared for publication, some of

which has been co-authored.

Su Chen Lim

22 October 2018

iii

ABSTRACT

Clostridium difficile is a well-known cause of healthcare-related diarrhoea. However,

C. difficile infection (CDI) is increasingly being diagnosed among individuals in the

community, many of whom lack the traditional risk factors of old age, recent

hospitalisation and antimicrobial exposure. In the Northern Hemisphere, genetically

indistinguishable C. difficile strains have been reported from humans, animals, food and

the environment, suggesting a possible zoonotic, foodborne and/or environmental

source for community-associated CDI (CA-CDI). Little is known about the

epidemiology of C. difficile in Australian foods and environments. The main goal of this

thesis was to identify potential sources of CA-CDI in Western Australia (WA) with a

focus on animal-to-human transmission through food and the environment. Knowing

how epidemic strains of C. difficile disseminate from food animals, reach the food

supply and contaminate the environment will help develop strategies to reduce the risk

of contamination and, therefore, transmission to humans. To achieve this goal, studies

were designed to:

(i) determine the prevalence and genotypes of C. difficile in the retail foods and

environment of WA;

(ii) examine the susceptibility of food and/or environmental C. difficile isolates

to antimicrobial agents and food preservatives;

(iii) define the extent of genetic overlap and potential transmission between

human, food and environmental C. difficile population;

(iv) investigate possible risk factors for predominantly community-associated

CDI; and,

(v) explore the relatedness of C. difficile from Australia to those that originated

in Asia.

To achieve the first aim of this study, vegetables (n = 280), meat (n = 37), compost (n =

81) and lawn (n = 311) in Perth, Western Australia (WA) were sampled. Selective

enrichment culture, toxin profiling and PCR ribotyping were performed on strains of

C. difficile recovered. C. difficile was isolated from 14.6%, 0.0%, 27.2% and 58.5% of

vegetables, meat, compost and lawn, respectively. Nearly half the isolates were

toxigenic (46.3%, 126/272), predominantly with the toxin profile A+B+CDT- (91.3%,

115/126). Collectively, the five most common RTs were 014/020 (28.3%, 77/272),

followed by 010 (14.3%, 39/272), QX 077 (5.9%, 16/272), 051 (2.9%, 8/272) and 056

(2.9%, 8/272). Many of the isolated C. difficile RTs were common among Australian

iv

human clinical samples and production animals, suggesting possible transmission of

C. difficile from animals to humans through food and environmental sources.

Susceptibility of food and environmental C. difficile isolates to fidaxomicin,

vancomycin, metronidazole, rifaximin, clindamycin, erythromycin,

amoxicillin/clavulanate, moxifloxacin and tetracycline was determined by the agar

dilution method. Phenotypic resistance to at least one antimicrobial agent was observed

in 103 (37.6%) isolates, predominantly from compost (38.9%, 14/36) and lawn (45.1%,

82/182). Compared to isolates from food and lawn, compost isolates displayed a higher

proportion of resistance to erythromycin (< 2.0% vs. 19.4%) and tetracycline (< 2.0%

vs. 13.9%). The compost isolates were likely of animal origin due to the common

practice of composting animal manure for agricultural use, and the acquisition of

antimicrobial resistance (AMR) genes is undoubtedly driven by the heavy use of

macrolides and tetracycline-based antimicrobials in the Australian livestock industry.

Both compost and lawn isolates shared AMR/susceptibility patterns similar to those

reported in human-derived isolates. The role of food and the environment in the

transmission of CDI is uncertain but both are suspected to play an intermediary role

between animals and humans. This study provides an important baseline for future

surveillance of AMR in food and environmental C. difficile in Australia.

The susceptibility of C. difficile to common food preservatives used in processed meats

was determined using broth microdilution. The MIC50/MIC90 of sodium nitrite, sodium

nitrate and sodium metabisulphite against C. difficile strains (n = 11) were 250/500

µg/ml, > 4, 000/> 4, 000 µg/ml and 1, 000/1, 000 µg/ml, respectively, which are all

higher than the maximum permitted level allowed in processed meats. No bactericidal

activity against C. difficile was observed. This finding indicates that processed meats

have the potential to be a source of C. difficile.

C. difficile RT 056 is well-established in both human and cattle populations in Australia.

It is also one of the most common RT in Australian food and the environment. Whole

genome sequencing (WGS) analysis of 29 strains of C. difficile RT 056 from human,

food, compost, lawn and veterinary origin demonstrated two clusters of human and

food/compost strains with ≤ 2 single nucleotide variant (SNV) differences, suggesting a

very recent shared ancestry. The core genome SNV analysis found 14% of human

strains showed a clonal relationship with one or more food/compost strains, consistent

with recent transmission. Clonal groups included strains from CA-CDI and hospital-

associated infection healthcare facility onset cases, as well as vegetable and compost

v

strains. These findings support the hypothesis of foodborne and environmental

transmission of C. difficile in WA and highlight the importance of the “One Health”

concept in monitoring the emergence of infectious diseases.

A retrospective case-controlled study was conducted with patients with CDI caused by

C. difficile RT 012 (a predominantly community-associated RT in WA) and compared

to controls of the same gender and similar age, with CDI caused by any other RTs

between January 2014 and March 2015. By univariate analysis, eating raw non-organic

homegrown vegetables and having contact with an individual who was living in a

nursing home were significantly associated with CDI caused by RT 012 (OR 16.50,

95% CI 1.09 – 250.18, p = 0.04 for both). However, with a response rate of only 14.8%

(23/155), this study lacked the statistical power to truly detect any differences in risk

factors between cases and controls. Multivariate analysis was not performed due to the

small number of enrolled cases and the broad confidence intervals for the only two

significant variables.

In response to a recent study that identified C. difficile RT 012 as one of the most

common RTs causing CDI in Asia, WGS was performed to examine the relatedness of

Australian food and clinical RT 012 isolates to those of Asian origin. SNV analysis of

11 C. difficile RT 012 strains from Australia (food, n = 1; human, n = 4) and Asia

(human, n = 6) revealed two clusters of human strains (≤ 2 SNVs difference). The first

clonal group consisted of human isolates from Shanghai and Hong Kong, with 0 SNV

differences in their core genome. The second clonal group comprised human isolates

from WA and Thailand, with 2 SNV differences in their core genome. The food isolate

from WA was genetically very distant from all other sequenced isolates, with an

average of 884 SNVs difference. With an overall prevalence of just 1.1% among WA’s

food and environmental isolates and a complete absence in Australian production

animals, the local establishment of RT 012 in WA community now seems unlikely.

Taken together, the findings suggest CDI caused by RT 012 may be of Asian origin

where the prevalence is high.

This thesis provides novel and critical insights into the prevalence, molecular

epidemiology, AMR, and potential transmission of food and environmental C. difficile

in WA. The studies presented here show food and the environment are a source for

C. difficile in WA, predominated by strains prevalent in both humans and animals with

comparable AMR profiles. For the first time, phylogeny analysis of non-recombinant

SNVs on C. difficile suggested that (i) over an extended period there has been a long-

vi

range transmission of C. difficile RT 056 in WA between humans, food and the

environment, supporting the theory of foodborne and environmental transmission of

CDI; and (ii) that the emergence of RT 012 in WA might be from Asia. Food and

environmental sampling studies for C. difficile in Australia are relatively new. In

coming years, further epidemiological studies and WGS will continue to provide greater

insights into the potential transmission of C. difficile between humans, animals, food

and the environment. Ultimately, a better understanding of CDI transmission is still

required. However, the evidence presented in this thesis adds weight to the hypothesis

that CDI is a zoonotic disease and that a collaborative One Health approach is needed to

reduce the overall burden of CDI.

vii

TABLE OF CONTENTS

CHAPTER 1 INTRODUCTION ................................................................................... 1

1.1 INTRODUCTION ................................................................................................ 2

1.2 C. DIFFICILE AND C. DIFFICILE INFECTION .............................................. 2

1.2.1 History ....................................................................................................... 2

1.2.2 Clinical presentation and traditional risk factors ...................................... 3

1.2.3 Detection of C. difficile in clinical cases ................................................... 6

1.2.4 Molecular typing method .......................................................................... 6

1.2.5 Health and economic burden ..................................................................... 9

1.2.6 Changing epidemiology ............................................................................ 9

1.2.6.1 Emergence of the hypervirulent C. difficile RT 027 strain ....................... 9

1.2.6.2 Emergence of CA-CDI ............................................................................ 14

1.2.7 Transmission of C. difficile in the hospital setting .................................. 16

1.3 POTENTIAL SOURCES OF C. DIFFICILE OUTSIDE OF THE HOSPITAL

SETTING ............................................................................................................ 18

1.3.1 Methods for detection of C. difficile in production animals, food and the

environment............................................................................................. 18

1.3.1.1 Production animals .................................................................................. 18

1.3.1.2 Food ........................................................................................................ 18

1.3.1.3 Environment ............................................................................................ 20

1.3.2 Prevalence and molecular epidemiology of C. difficile .......................... 20

1.3.2.1 Production animals .................................................................................. 20

1.3.2.2 Food ........................................................................................................ 21

1.3.2.3 Environment ............................................................................................ 27

1.3.3 Genetic relationship between C. difficile strains from various sources .. 27

1.4 C. DIFFICILE IN AUSTRALIA ........................................................................ 31

1.4.1 C. difficile in humans .............................................................................. 31

1.4.2 C. difficile in Australian livestock ........................................................... 32

1.4.2.1 C. difficile in Australian cattle and calves ............................................... 35

1.4.2.2 C. difficile in Australian sheep and lambs ............................................... 37

1.4.2.3 C. difficile in Australian neonatal pigs .................................................... 37

1.4.3 Long-range interspecies transmission of RT 014 between Australian pigs

and humans ............................................................................................. 38

1.4.4 Gaps in knowledge .................................................................................. 40

1.5 THESIS AIMS AND OBJECTIVES .................................................................. 40

viii

CHAPTER 2 MATERIALS AND METHODS .......................................................... 41

2.1 MATERIALS ...................................................................................................... 42

2.1.1 Culture media .......................................................................................... 42

2.1.2 Buffer and solutions ................................................................................ 42

2.1.3 Commercial kits ...................................................................................... 42

2.1.4 PCR reagents ........................................................................................... 42

2.2 GENERAL BACTERIAL CULTURE METHODS ........................................... 43

2.2.1 Enrichment culture .................................................................................. 43

2.2.2 Direct culture ........................................................................................... 43

2.2.3 C. difficile identification.......................................................................... 44

2.3 MOLECULAR CHARACTERISATION OF C. DIFFICILE ISOLATES ........ 44

2.3.1 DNA extraction ....................................................................................... 44

2.3.2 Toxin profiling ........................................................................................ 44

2.3.3 PCR ribotyping ....................................................................................... 46

2.4 TESTING FOR C. DIFFICILE CONTAMINATION OF FOODS AND THE

ENVIRONMENT ............................................................................................... 46

2.4.1 Retail vegetables and meats .................................................................... 47

2.4.1.1 Settings and sample collection ................................................................ 47

2.4.1.2 Sample preparation for culture ................................................................ 47

2.4.1.3 Probability of one vegetable being positive for C. difficile .................... 48

2.4.2 Composted products ................................................................................ 48

2.4.2.1 Setting, sample collection and sample preparation for culture ............... 48

2.4.3 Lawn ........................................................................................................ 48

2.4.3.1 Setting, sample collection and sample preparation for culture ............... 48

2.4.3.1.1 Differentiation of “young” lawn from “old” lawn .................................. 50

2.4.3.1.2 Quantification of C. difficile in lawn samples ......................................... 50

2.5 ANTIMICROBIAL SUSCEPTIBILITY TESTING........................................... 50

2.5.1 Agar incorporation method ..................................................................... 50

2.5.1.1 Statistical analysis ................................................................................... 51

2.5.2 Broth microdilution method .................................................................... 56

2.5.3 Checkerboard assay ................................................................................. 58

2.6 WHOLE-GENOME SEQUENCING ................................................................. 58

2.6.1 Strain collection ...................................................................................... 58

2.6.2 Genomic DNA extraction ....................................................................... 59

2.6.3 Genomic DNA sequencing ..................................................................... 59

2.6.4 Analysis of WGS data ............................................................................. 63

2.6.4.1 In silico multilocus sequence typing and AMR gene profiling............... 63

ix

2.6.4.2 De novo assembly and annotation ........................................................... 63

2.6.4.3 SNV analysis ........................................................................................... 63

2.7 A RETROSPECTIVE CASE-CONTROL STUDY OF C. DIFFICILE RT 012

IN WA ................................................................................................................. 64

2.7.1 Ethics approval ........................................................................................ 64

2.7.2 Study setting and design ......................................................................... 64

2.7.3 Data collection and analysis .................................................................... 65

CHAPTER 3 C. DIFFICILE IN FOOD ...................................................................... 66

3.1 INTRODUCTION .............................................................................................. 67

3.2 RESULTS ........................................................................................................... 67

3.2.1 Prevalence and molecular epidemiology of C. difficile on WA-grown

root vegetables ........................................................................................ 67

3.2.2 Prevalence and molecular epidemiology of C. difficile on imported

vegetables and vegetables of unknown country of origin ....................... 72

3.2.3 Prevalence of C. difficile in meat products available in WA .................. 76

3.3 DISCUSSION ..................................................................................................... 76

3.3.1 C. difficile on the retail vegetables available in WA ............................... 76

3.3.2 C. difficile in retail meats available in WA ............................................. 80

3.4 CHAPTER SUMMARY ..................................................................................... 81

CHAPTER 4 C. DIFFICILE IN THE ENVIROMENT ............................................ 83

4.1 INTRODUCTION .............................................................................................. 84

4.1.1 Compost manufacturing process in Australia ......................................... 85

4.2 RESULTS ........................................................................................................... 85

4.2.1 Prevalence and molecular epidemiology of C. difficile in compost

products available in WA ........................................................................ 85

4.2.2 Prevalence and molecular epidemiology of C. difficile in the public lawn

of WA ...................................................................................................... 87

4.2.2.1 Enumeration of C. difficile in public lawn .............................................. 91

4.3 DISCUSSION ..................................................................................................... 95

4.4 CHAPTER SUMMARY ..................................................................................... 98

CHAPTER 5 SUSCEPTIBILITY OF FOOD AND ENVIRONMENTAL

C. DIFFICILE TO ANTIMICROBIALS AND PRESERVATIVES ..................... 100

5.1 ANTIMICROBIAL SUSCEPTIBILITY OF FOOD AND ENVIRONMENTAL

C. DIFFICILE IN WA ...................................................................................... 101

5.1.1 Introduction ........................................................................................... 101

5.1.2 Results ................................................................................................... 102

5.1.3 Discussion ............................................................................................. 108

5.2 SUSCEPTIBILITY OF C. DIFFICILE TO FOOD PRESERVATIVES ......... 111

x

5.2.1 Introduction ........................................................................................... 111

5.2.2 Results ................................................................................................... 112

5.2.3 Discussion ............................................................................................. 115

5.3 CHAPTER SUMMARY ................................................................................... 118

CHAPTER 6 ASPECTS OF THE EPIDEMIOLOGY OF CDI IN WESTERN

AUSTRALIA ............................................................................................................... 119

6.1 WGS OF C. DIFFICILE RT 056 ...................................................................... 120

6.1.1 Introduction ........................................................................................... 120

6.1.2 Results ................................................................................................... 120

6.1.2.1 MLST .................................................................................................... 120

6.1.2.2 SNV analysis ......................................................................................... 121

6.1.2.3 In silico and in vitro AMR profiling ..................................................... 124

6.1.3 Discussion ............................................................................................. 124

6.2 C. DIFFICILE RT 012 IN WA ......................................................................... 129

6.2.1 Introduction ........................................................................................... 129

6.2.2 Results ................................................................................................... 132

6.2.2.1 Case-control study ................................................................................. 132

6.2.2.2 Relatedness of Australian and Asian isolates using WGS .................... 132

6.2.2.2.1 MLST .................................................................................................... 132

6.2.2.2.2 SNV analysis ......................................................................................... 132

6.2.2.2.3 In silico AMR profiling ......................................................................... 140

6.2.3 Discussion ............................................................................................. 140

6.3 CHAPTER SUMMARY ................................................................................... 143

CHAPTER 7 KEY FINDINGS AND GENERAL DISCUSSION .......................... 145

7.1 INTRODUCTION ............................................................................................ 146

7.2 KEY FINDINGS AND GENERAL DISCUSSION ......................................... 146

7.3 STUDY LIMITATION AND DIRECTION FOR FUTURE RESEARCH ..... 163

7.4 CONCLUSIONS ............................................................................................... 164

REFERENCES .............................................................................................................. 165

APPENDICES .............................................................................................................. 194

APPENDIX I QUESTIONNAIRE USED IN CASE-CONTROL STUDY .............. A1

APPENDIX II THESIS-RELATED PUBLICATIONS ............................................. A9

APPENDIX III INDIRECTLY RELATED PUBLICATION .................................. A32

xi

LIST OF TABLES

Table 1.1 Global prevalence of C. difficile in production animals ......................... 22

Table 1.2 Global prevalence of C. difficile in foods ............................................... 25

Table 1.3 Global prevalence of C. difficile in the environment .............................. 28

Table 1.4 Clusters of highly similar human and animal C. difficile RT 078 isolates

identified by Knetsch et al. using core genome analysis ........................ 30

Table 2.1 Primers used in toxin gene PCR assay and PCR ribotyping ................... 45

Table 2.2 Antimicrobial concentration tested and solvents used in the preparation

of antimicrobial stock solution ................................................................ 55

Table 2.3 The source and molecular characteristics of C. difficile strains used in

testing the effect of food preservatives ................................................... 57

Table 2.4 Strain collection used for WGS............................................................... 60

Table 3.1 Prevalence and toxin profiles of C. difficile cultured from WA-grown

root vegetables ........................................................................................ 69

Table 3.2 Summary of C. difficile PCR RTs isolated from WA-grown root

vegetables, by store ................................................................................. 71

Table 3.3 Prevalence and toxin profiles of C. difficile from imported vegetables

and vegetables of unknown country of origin ......................................... 73

Table 3.4 Summary of C. difficile PCR RTs isolated from imported vegetables and

vegetables of unknown country of origin, by store ................................. 75

Table 4.1 Prevalence and toxin gene profiles of C. difficile cultured from compost

products in WA ....................................................................................... 88

Table 4.2 Summary of C. difficile PCR RTs isolated from compost products, sorted

by the sample........................................................................................... 89

Table 4.3 Prevalence and risk factor analysis for C. difficile positivity in public

lawns ....................................................................................................... 92

Table 4.4 Enumeration of C. difficile in soil from the lawn samples ...................... 96

Table 5.1 Antimicrobial susceptibilities of 274 C. difficile isolates from food,

compost and lawn .................................................................................. 103

Table 5.2 Antimicrobial susceptibilities of the most prevalent food and

environmental C. difficile RTs .............................................................. 107

Table 5.3 Susceptibility of C. difficile to food preservatives ................................ 113

Table 5.4 In vitro activities of sodium nitrite and sodium nitrate individually and in

combination against C. difficile............................................................. 114

Table 6.1 Antimicrobial susceptibilities of 29 C. difficile RT 056 ....................... 125

xii

Table 6.2 The 10 most common C. difficile RTs in WA, based on HISWA data,

2011 – 2014 ........................................................................................... 130

Table 6.3 Characteristics of participants and people who refused to participate .. 133

Table 6.4 Characteristics of CDI patients with RT 012 (cases) and CDI patients

with other RTs (controls) ..................................................................... 134

Table 6.5 Risk factor analysis for C. difficile RT 012........................................... 135

LIST OF FIGURES

Figure 1.1 Bristol stool chart ...................................................................................... 4

Figure 1.2 Endoscopic and histological visualization of colitis due to CDI .............. 5

Figure 1.3 Transmission events of epidemic fluoroquinolone-resistant C. difficile

RT 027, inferred from phylogeographic analysis ................................... 11

Figure 1.4 Genetic map of the C. difficile toxin loci ................................................ 13

Figure 1.5 Timeline for definitions of C. difficile-associated disease exposure....... 15

Figure 1.6 Relationships between 957 C. difficile isolates obtained between

September 2007 – March 2011 as estimated by SNVs, and compared to

epidemiological links .............................................................................. 17

Figure 1.7 Hospital ward relationships between cases with genetically closely

related C. difficile .................................................................................... 19

Figure 1.8 Maximum-likelihood phylogeny of 248 C. difficile RT 078 based on

Average Nucleotide Identity (ANI) calculated by pairwise comparison of

genome assemblies .................................................................................. 29

Figure 1.9 C. difficile RT diversity in humans, in Australia, as reported by seven

separate studies........................................................................................ 33

Figure 1.10 C. difficile RTs in Australian cattle, sheep and pigs ............................... 36

Figure 1.11 Maximum-likelihood phylogeny of 44 C. difficile RT 014 from pigs and

humans .................................................................................................... 39

Figure 2.1 Sampling points of lawn in a public park ............................................... 49

Figure 2.2 Origin of C. difficile RTs tested in antimicrobial study .......................... 52

Figure 2.3 Dendrogram of the 5 most prevalent C. difficile RTs, their associated

toxin profiles and distribution ................................................................. 54

Figure 3.1 C. difficile PCR ribotyping patterns, toxin gene profiles and prevalence

on WA-grown organic and non-organic root vegetables ........................ 70

Figure 3.2 C. difficile PCR ribotyping patterns for strains isolated from imported

vegetables and vegetables of unknown country of origin ....................... 74

xiii

Figure 4.1 Schematic diagram of the composting process and sampling points for

compost products .................................................................................... 86


in compost products in WA .................................................................... 90


in the public lawn of WA ........................................................................ 93

Figure 4.4 Geographical distribution of C. difficile RTs in the public lawn of WA 94

Figure 4.5 Incidence of CA-CDI cases in WA from 2010 and 2014, sorted by

residential postcode ................................................................................. 99

Figure 6.1 SNV analysis of 29 C. difficile RT 056 ................................................ 122

Figure 6.2 Heatmap of pairwise core genome SNV differences between 29

C. difficile RT 056 isolates .................................................................... 123

Figure 6.3 C. difficile RT 012 cases reported to HISWA in 2011 – 2014 .............. 131

Figure 6.4 Heatmap of pairwise core genome SNV differences between 11

C. difficile RT 012 isolates, sorted by origin ........................................ 139

Figure 7.1 C. difficile RT diversity in WA (predominantly), including amongst

humans, cattle and calves, sheep and lamb, piglets, vegetables, compost

and lawn ................................................................................................ 151

Figure 7.2 Potential sources of C. difficile implicated in the transmission of CDI in

the community of WA ........................................................................... 160

xiv

ACKNOWLEDGEMENTS

This research was supported by the Australian Government Research Training Program.

The completion of this PhD would not be possible without the support of many

colleagues, friends, and family members. First, I would like to thank my coordinating

supervisors Professor Thomas V. Riley for years of support, guidance and for funding

all of my conference travel for which I am immensely grateful. I thank my co-

supervisor Dr Niki F. Foster for her guidance, as well as her time and effort in providing

much insightful and invaluable feedback that always greatly improved my writing

skills. Lastly, I would like to thank Dr Kate Hammer for inheriting me as a PhD student

when Niki left our research group to join a diagnostic laboratory in 2015.

I would like to acknowledge that many people contributed to the studies presented in

this thesis, namely: Dr Peter Moono, with whom I collaborated in the compost and lawn

study; Ms Grace Androga, with whom I had a lovely time working on the antimicrobial

study; Dr Lauren Bloomfield, who provided the epidemiological data on clinical

C. difficile isolates that were of interest to me; Dr Johanna Dups, who helped me

designed the case-control study, and Dr Daniel Knight, who kindly performed the WGS

analysis on my behalf. I would also like to express my gratitude to Dr Deirdre Collins

and Dr Alethea Rea for helping me with my statistical analysis.

This PhD has been an amazing learning experience enriched by members of the Riley

group who provided a positive working environment for me to conduct my research.

Thank you for all the friendly banter, lunchtime chit-chat, the seemingly endless supply

of snacks, and weekend outings for dim sum and hiking. These past and present Riley

group members are: Ms Stacey Hong, Dr Papanin Putsathit, Ms Grace Androga, Dr

Deirdre Collins, Ms Niloufar Roshan Hesari, Dr Nikki Man, Dr Niki F. Foster, Dr Kate

Hammer, Dr Michele Squire, Dr Kerry Carson, Dr Peter Moono, Mr Alan McGovern,

Ms Sicilia Perumalsamy, Dr Johanna Dups, Dr Brian Kullin, Dr Korakrit Imwattana,

Mr Niraj Shivaperumal, Ms Ashleigh Hale and Dr Lauren Bloomfield.

To my family, this PhD would never have been completed without your support and

unconditional love. I dedicate this work to my colleagues, friends and family who were

with me throughout the course of my PhD journey. Lastly, I would like to thank the

ladies who worked at the coffee cart. The hot chocolate they made was instrumental in

fuelling my research and thesis writing.

xv

AUTHORSHIP DECLARATION: CO-AUTHORED PUBLICATIONS

This thesis contains work that has been published and/or prepared for publication. In all

the papers listed below, authorship was based on the 2007 Australian Code for the

Responsible Conduct of Research, developed jointly by the National Health and

Medical Research Council (NHMRC), the Australian Research Council (ARC) and

Universities Australia.

I hereby certify that I made a significant contribution to (i) conception and design of

studies, (ii) analysis and interpretation of data, (iii) drafting the articles or revising it

critically for important intellectual content, and (iv) final approval of the version to be

published. For all publications on which I am the first author, I took primary

responsibility for the overall management of the drafting and submission process with

editorial assistance from my co-author(s).

Su Chen Lim

22 October 2018

I, Professor Thomas V. Riley certify that the student statements regarding their

contribution to each of the works listed below are correct.

Professor Thomas V. Riley

11 March 2019

xvi

The following publications are directly related to my thesis research, arose during the

course of my PhD candidature and are located in APPENDIX II.

PUBLICATION 1

Details of the work: Su-Chen Lim, Niki F. Foster, Briony Elliott, Thomas V.

Riley

High prevalence of Clostridium difficile on retail root

vegetables, Western Australia.

Journal of Applied Microbiology 2018; 124: 585-590.

Full version: http://dx.doi.org/10.1111/jam.13653

Location in thesis: Chapter 3 and Appendix II

Student contribution: Designed the study, performed all the experimental work

and data analysis; wrote the paper (80%).

Co-author signatures

and dates:

Niki F. Foster Briony Elliott Thomas V. Riley

PUBLICATION 2

Details of the work:

Peter Moono, Su-Chen Lim, Thomas V. Riley

High prevalence of toxigenic Clostridium difficile in public

space lawns in Western Australia.

Scientific Reports 2017; 7: 41196.

Full version: http://www.ncbi.nlm.nih.gov/pubmed/28145453


Student contribution: Performed all molecular work and data analysis; assisted

with drafting the paper and provided critical revision of the

final manuscript (50%).


and dates:

Peter Moono Thomas V. Riley

http://dx.doi.org/10.1111/jam.13653

http://www.ncbi.nlm.nih.gov/pubmed/28145453

xvii

PUBLICATION 3

Details of the

work:

Su-Chen Lim, Grace O. Androga, Daniel R. Knight, Peter Moono,

Niki F. Foster, Thomas V. Riley

Antimicrobial susceptibility of Clostridium difficile isolated from

food and environmental sources in Western Australia.

International Journal of Antimicrobial Agents 2018; 52: 411-415.

Full version: https://doi.org/10.1016/j.ijantimicag.2018.05.013

Location in

thesis:

Chapter 5 and Appendix II

Student

contribution:

Designed the study, contributed to half of the experimental work,

performed all data analysis; wrote the paper (75%).

Co-author

signatures and

dates:

Grace O. Androga Daniel R. Knight Peter Moono

Niki F. Foster Thomas V. Riley

PUBLICATION 4


Su-Chen Lim, Niki F. Foster, Thomas V. Riley

Susceptibility of Clostridium difficile to the food

preservatives sodium nitrite, sodium nitrate and sodium

metabisulphite.

Anaerobe 2016; 37: 67-71.

Full version: https://doi.org/10.1016/j.anaerobe.2015.12.004


Student contribution: Assisted with study design, performed all experimental

work and data analysis and wrote the paper (75%).


and dates:


https://doi.org/10.1016/j.ijantimicag.2018.05.013

https://doi.org/10.1016/j.anaerobe.2015.12.004

xviii

This publication arose during the course of my PhD candidature, is indirectly related to

my thesis research and is located in APPENDIX III.

PUBLICATION 5 (INDIRECTLY THESIS RELATED)


Grace O. Androga, Daniel R. Knight, Su-Chen Lim,

Niki F. Foster, Thomas V. Riley

Antimicrobial resistance in large clostridial toxin-

negative, binary toxin-positive Clostridium difficile

ribotypes.

Anaerobe 2018; 54: 55-60.

Full version: https://doi.org/10.1016/j.anaerobe.2018.07.007

Location in thesis: Appendix III

Student contribution to

work:

Performed half the experimental work and provided

critical revision of the final manuscript (35%).

Co-author signatures and

dates:

Grace O. Androga Daniel R. Knight


PRESENTATIONS

1. Su-Chen Lim, Niki F. Foster, Thomas V. Riley. Susceptibility of Clostridium

difficile to food preservatives: sodium metabisulphite, sodium nitrate and

sodium nitrite. 5th International Clostridium difficile Symposium, 2015, Bled,

Slovenia.

2. Su-Chen Lim, Niki F. Foster, Thomas V. Riley. High prevalence of Clostridium

difficile found on Australian retail vegetables. West Coast Microbiome Network,

2016, Perth, Australia.

3. Su-Chen Lim, Niki F. Foster, Thomas V. Riley. High prevalence of Clostridium

difficile found on Australian retail vegetables. Australian Society for

Microbiology Annual Scientific Meeting and Exhibition, 2016, Perth, Australia.

https://doi.org/10.1016/j.anaerobe.2018.07.007

xix

4. Su-Chen Lim, Peter Moono, Thomas V. Riley. Clostridium difficile found in

gardening products: innocent bystander or the cause of community-acquired

C. difficile infection through contamination of foods and the environments?

Australian Society for Microbiology Annual Scientific Meeting and Exhibition,

2016, Perth, Australia.

5. Su-Chen Lim. Antimicrobial susceptibility of Clostridium difficile from diverse

environmental sources. Australian Society for Microbiology Annual Scientific

Meeting and Exhibition, 2017, Hobart, Australia.

6. Su-Chen Lim, Daniel R. Knight, Niki F. Foster, Thomas V. Riley. Reservoirs of

Clostridium difficile in the community. Australian Society for Microbiology

Annual Scientific Meeting and Exhibition, 2018, Brisbane, Australia.

7. Su-Chen Lim, Daniel R. Knight, Niki F. Foster, Thomas V. Riley. Whole-

genome sequencing reveals potential spread of Clostridium difficile between

humans, foods and the environment of Western Australia. 6th International

Clostridium difficile Symposium, 2018, Bled, Slovenia.

8. Su-Chen Lim, Johanna Dups, Daniel R. Knight, Lauren Bloomfield, Deirdre

Collins, Niki F. Foster, Thomas V. Riley. Emergence of predominantly

community-associated Clostridium difficile PCR ribotype 012 in Western

Australia: Risk factors and relatedness to strains from Asia. 6th International

Clostridium difficile Symposium, 2018, Bled, Slovenia.

9. Thomas V. Riley, Su-Chen Lim, Peter Moono, Sicilia Perumalsamy, Niki F.

Foster. High prevalence of Clostridium difficile in the Western Australian

environment. Australasian College for Infection Prevention and Control Annual

Conference, 2018, Brisbane, Australia.

INDIRECTLY RELATED PRESENTATION

1. Grace O. Androga, Su-Chen Lim, Daniel R. Knight, Niki F. Foster, Thomas

Riley. Antibiogram patterns of non-toxigenic, CDT producing Clostridium

difficile ribotypes. International Congress of Chemotherapy and Infection, 2017,

Taipei, Taiwan.

xx

ABBREVIATIONS

AMC Amoxicillin/clavulanate

AMR Antimicrobial resistant/resistance

ANI Average Nucleotide Identity

BA Blood agar

BHIB Brain heart infusion broth

bp Base pair(s)

BSA Bovine serum albumin

°C Degree Celsius

CA Community-associated / California

CAI Community-associated infection

CCFA Cycloserine-cefoxitin-fructose agar

CCNA Cell cytotoxicity neutralisation assay

CDC Centers for Disease Control and Prevention

CDI C. difficile infection

CDRN C. difficile Ribotype Network

CDT Binary toxin

cdtA Binary toxin subunit A

cdtB Binary toxin subunit B

CdtLoc Binary toxin locus

cfu Colony forming unit

CG Clonal group

CI Confidence interval

CLI Clindamycin

CLN Clinic

CLSI Clinical and Laboratory Standards Institute

cm Centimetre(s)

CO-HCFA Community-onset HCFA

CRT Carrot

DMSO Dimethyl sulfoxide

DNA Deoxyribonucleic acid

DL Digestate liquid

dNTP Deoxynucleoside triphosphate

EIA Enzyme immunoassay

EMA European Medicines Agency

ENA European Nucleotide Archive

ERY Erythromycin

EUCAST European Committee on Antimicrobial Susceptibility Testing

°F Degree Fahrenheit

FDX Fidaxomicin

FIC Fractional inhibitory concentration

FICI FIC index

FQR Fluoroquinolone-resistant

g Gram(s) / Gravity

GDH Glutamate dehydrogenase

GM Geometric mean / Garden mixes

h Hour(s)

HA Hospital-associated

HAI Hospital-associated infection

xxi

HB Human biosolids

HCFA Healthcare facility-associated

HCFO/HO Healthcare facility onset

HISWA Healthcare Infection Surveillance WA

I Intermediate

kg Kilogram(s)

L Litre

LOS Length of stay

MBC Minimum bactericidal concentration

M Mulches

MDR Multidrug resistance

MEM Meropenem

mg Milligram(s)

MgCl2 Magnesium chloride

MIC Minimum inhibition concentration

min Minute(s)

ml Millilitre(s)

MLST Multilocus sequence typing

MLSB Macrolide-lincosamide-streptogramin B

MLVA Multilocus variable number tandem repeat analysis

MTZ Metronidazole

MXF Moxifloxacin

NA No data available

NaCl Sodium Chloride

NAP North America pulsotype

NNATs Nucleic acid amplification tests

NSW New South Wales

OH Ohio

ON Ontario

OR Odd ratio(s)

PaLoc Pathogenicity locus

PBS Phosphate-buffered saline

PBST PBS with 0.1% Tween 20

PCR Polymerase chain reaction

PD Patient day

PE Paired-end

PFGE Pulsed-field gel electrophoresis

PMC Pseudomembranous colitis

POT Potato

ppm Parts per million

PST Peptone saline with 0.1% Tween 20

QX Internal nomenclature

R Resistance

REA Restriction endonuclease analysis

RFLP Restriction fragment length polymorphism

RFX Rifaximin

RNA Ribonucleic acid

rRNA Ribosomal RNA

RT Ribotype

RTE Ready-to-eat

xxii

s Second(s)

S Susceptible

SA South Australia

SC Soil conditioner

SM Sheep manure

SNV Single nucleotide variant

spp. Species

ST Sequence type

tcdA/TcdA Toxin A

tcdB/TcdB Toxin B

tcdC PaLoc negative regulator

TET Tetracycline

UK United Kingdom

UNK Unknown

UP Ultra-pure

US United States

USA United States of America

UV Ultraviolet

VAN Vancomycin

vs Versus

v/v Volume to volume ratio

WA Western Australia

WGS Whole-genome sequencing

w/v Weight/volume

yr Year(s)

µg Microgram(s)

µl Microlitre(s)

µM Micromolar(s)

× g Times gravitational force

Chapter 1: Introduction

1

CHAPTER 1 INTRODUCTION


2

1.1 INTRODUCTION

Clostridium difficile is a spore-forming anaerobe bacterium that is a well-known cause

of infectious diarrhoea in hospitalised patients. Since 2003, there has been a change in

the epidemiology of C. difficile infection (CDI) in several locations around the world1.

Strains of fluoroquinolone-resistant C. difficile have emerged and spread throughout

North America and Europe, causing significant morbidity and mortality1, 2. A number of

countries have also reported increases in previously rare community-associated CDI

(CA-CDI), up to 26% in Australia3, 36% in The Netherlands4 and 41% in the United

States of America (USA)5.

The literature review presented below is separated into four main sections.

Section 1.2 offers a general introduction to C. difficile and CDI covering the following

topics; history, clinical presentation, risk factors, molecular typing methods, health and

economic burden, changes in epidemiology, transmission within a hospital setting and

the emergence of CA-CDI.

Section 1.3 presents a comprehensive review of the prevalence and molecular

epidemiology of C. difficile in reservoirs outside a hospital setting; focusing on

production animals, food and the environment.

In Section 1.4, a summary of previous surveillance studies on Australian livestock and

people is presented. This provided context for the research aim of this thesis (Section

1.5) which is to identify reservoirs of C. difficile in the community of Western Australia

(WA).

1.2 C. DIFFICILE AND C. DIFFICILE INFECTION

1.2.1 History

C. difficile was first identified in 1935 by Hall and O’Toole, in the meconium and

faeces of a healthy infant6. Although C. difficile produced toxins that were pathogenic in

mice and guinea pig models, further investigation showed no evidence of pathogenicity

in humans7, 8. For the next three decades, C. difficile received little attention and was

regarded as part of the normal gastrointestinal microbiota of human infants. In the

1960s, an increase in the number of cases of pseudomembranous colitis (PMC) created

a surge of interest9-11. During the 1970s, there were reports of C. difficile isolation from

hospitalised patients who had undergone antimicrobial therapy12, 13, however, it was not

until 1978 that C. difficile was identified as the causative agent of PMC and


3

antimicrobial-associated diarrhoea14-16. This finding was reported almost simultaneously

in three separate studies from the United Kingdom (UK) and USA. In March 1978,

George et al. suggested that C. difficile caused PMC after isolating C. difficile in the

faeces of all patients with PMC in Birmingham Children’s Hospital, UK14. In April

1978, a study from the USA reported the isolation of C. difficile from patients with

clindamycin-associated PMC and demonstrated the presence of faecal toxin and

toxigenicity of isolates using a tissue culture assay15. Last, in May 1978, another UK

study suggested C. difficile as responsible for PMC and that previous antibiotic therapy

was a risk factor for infection16. The authors concluded that changes in gastrointestinal

microbiota in response to antibiotic therapy might be the contributing factor. During the

1980s and 1990s, the introduction and widespread use of broad-spectrum third-

generation cephalosporins in hospitals contributed to a significant increase in the

incidence of CDI17, 18. By the early 2000s, the emergence of fluoroquinolone-resistant

C. difficile strains had caused major changes in the epidemiology of C. difficile with

outbreaks occurring throughout Europe and North America1, 2, 19-21. This change in

epidemiology is further discussed in Section 1.2.6.

1.2.2 Clinical presentation and traditional risk factors

CDI is a toxin-mediated disease of the colon; involvement of the small intestine is rare.

Clinical presentation of CDI ranges from asymptomatic colonisation to mild diarrhoea

with soft or watery stool (type 5 – 7 of the Bristol Stool Chart, see Figure 1.1), and

abdominal pain and fever20. In severe cases, signs can include (i) PMC, numerous, small

(2 – 10 mm), raised, round, yellowish plaques covering large portions of the intestinal

mucosa, with neutrophils erupting through the surface epithelium causing inflammation

(Figure 1.2); (ii) toxic megacolon, dilated and swollen colon; and (iii) bowel

perforation20-24. Extra-intestinal manifestations of CDI are rare, but sepsis, shock and

death may occur.

C. difficile is transmitted by the faecal-oral route. Thus, ingestion of dormant C. difficile

spores can lead to symptomatic infection if these spores germinate and proliferate

within the colon to produce toxins which mediate intestinal epithelium injury,

inflammation and colitis25-27. CDI is intrinsically resistant to multiple antimicrobials.

Recent antimicrobial usage is the most important risk factor for CDI as it disrupts the

normal gastrointestinal microbiota that normally prevents the colonisation and

outgrowth of C. difficile20, 28, 29. Broad-spectrum antimicrobials such as clindamycin,

cephalosporins, penicillins and fluoroquinolones, are commonly associated with


4

Figure 1.1 Bristol stool chart. The amount of time the stool spent in the bowel

decreases from type 1 – 7. A normal stool would be type 3 or 4, depending on the

normal bowel habits of an individual. Type 5 – 7 are soft, fluffy or watery stool

consistent with the mild diarrhoea symptom of CDI30.


5

Figure 1.2 Endoscopic and histological visualisation of colitis due to CDI. (A)

Endoscopic visualisation of a healthy colonic mucosa31. (B) Endoscopic visualisation of

a sigmoid colon with PMC23. (C) Histological visualisation of a normal colonic

mucosa24. (D) Histological visualisation of a colon with PMC24.


6

CDI28, 29, 32. Other risk factors associated with CDI include recent hospitalisation, age >

65 years old, the presence of underlying diseases (inflammatory bowel disease, irritable

bowel syndrome, Crohn’s disease and renal failure), and exposure to proton pump

inhibitors33, 34.

1.2.3 Detection of C. difficile in clinical cases

Several laboratory diagnostic tests are available for CDI and can be broadly grouped

into those that detect toxins [cell cytotoxicity neutralisation assay (CCNA) and enzyme

immunoassay (EIA)], and those that detect the presence of C. difficile [glutamate

dehydrogenase (GDH), culture of toxigenic C. difficile from stool and stool nucleic acid

amplification tests (NAATs) for toxin genes]35-38.

Since the early 1980s, the detection of C. difficile toxins in the stool of patients by

CCNA has been the gold standard for identifying toxigenic C. difficile39. However, due

to its long turnaround time, EIA kits that can detect both toxins A and B became

popular, especially in smaller laboratories that lacked culturing facilities40. GDH is an

enzyme produced by both toxigenic and non-toxigenic C. difficile, as well as

C. botulinum and C. sporogenes41. The GDH test can detect the presence of bacterium

with a high sensitivity and is comparable with stool culture but detection of C. difficile

toxin is still necessary for confirmation of CDI. Regardless, it remains useful as a rapid

screening test where only GDH positive samples are further investigated using stool

culture and NAATs42, 43. Until recently, cycloserine-cefoxitin-fructose agar (CCFA) was

the most commonly used culture medium for C. difficile44. Currently, many laboratories

have adopted ChromID C. difficile agar, a chromogenic medium developed by

bioMérieux (Marcy-l’Etoile, France) that is more selective and has a 12% higher

recovery rate at 24 h compared to CCFA at 48 h45. The optimal diagnostic approach

remains controversial and testing guidelines vary from country to country43, 46, 47.

1.2.4 Molecular typing method

Molecular typing of isolates is important for surveillance and outbreak investigation.

Several typing methods are currently in use for C. difficile. Some focus on macro

analysis of the genome [restriction endonuclease analysis (REA) and pulsed-field gel

electrophoresis (PFGE)], or small regions within the genome [polymerase chain

reaction (PCR) ribotyping, toxinotyping and multilocus variable number tandem repeat

analysis (MLVA)], and others differentiate between strains using nucleotide sequence


7

analysis at multiple loci [multilocus sequence typing (MLST)] or the complete bacterial

genome [whole-genome sequencing (WGS)].

The most commonly used REA method for C. difficile typing relies on frequent cutting

by restriction enzyme HindIII48. The resulting fragments are separated by

electrophoresis in an agarose gel. This method is highly discriminatory but visual

assessment of a large number of fragments in a single gel may be difficult and

compromises the ability for comparison of inter-laboratory results48, 49. PFGE is used

for C. difficile in the USA. Restriction enzymes, typically SmaI or SacII, are used to

infrequently cleave the bacterial DNA43, 49. This generates approximately 7 – 20

fragments that are generally large in size, requiring separation by pulsed field gel

electrophoresis to minimise shearing. The resulting banding patterns are highly

discriminatory43, 49. In the USA, specific patterns are designated with a North American

Pulsotype (NAP) and a number (e.g. NAP1).

PCR ribotyping is the dominant typing method in Europe and the rest of the world. It is

based on the variation in size and number of the 16S – 23S rDNA intergenic spacer

regions49, 50. The Public Health Service Anaerobe Reference Unit, Cardiff (UK),

established a ribotyping reference database for C. difficile however, currently, the

responsibility of collating existing and assigning new ribotypes (RTs) is performed by

the C. difficile Ribotype Network (CDRN) in Leeds (UK), which has > 600 different

RTs in the CDRN database43, 49. To facilitate inter-laboratory comparison of typing

results and to make PCR ribotyping less time-intensive, high-resolution, sequenced-

based capillary gel electrophoresis was developed and banding patterns are referred to

as RTs (e.g. RT 014)51. A web-based database (https://webribo.ages.at/), unaffiliated

with the CDRN library, was developed in Austria to allow users to upload their

ribotyping data and receive a RT identification51.

Toxinotyping is a PCR restriction fragment length polymorphism (RFLP) -based

method for differentiating C. difficile strains according to the variation in their toxin

genes when compared to the reference strain VPI 10463 (toxinotype 0)48. This method

involves PCR amplification and restriction enzyme digestion of 10 regions covering the

entire pathogenicity locus (PaLoc) [five open reading frames, two toxin genes (tcdA and

tcdB) and three additional genes (tcdC, tcdD and tcdE)]52. Later, B1 and A3 fragments

of the tcdB and tcdA genes were selected as markers for differentiation of variant

C. difficile strains as variation in these fragments alone enabled detection and

differentiation of strains53. B1 RFLP (types 1 – 11) are determined according to a

https://webribo.ages.at/


8

HincII/AccI restriction profile and A3 RFLP (types 1 – 16) are determined according to

deletions/insertions and an EcoRI restriction profile53. The combination of variants in

the B1 and A3 fragments determines the toxinotype. Currently, there are 34 known

toxinotypes (I to XXXIV) and toxinotypes with the same B1 and A3 combination but

with differences in the other regions of the PaLoc can be further differentiated into

subtypes (designated a, b, c, etc)53.

MLVA utilises the naturally occurring variation in the number of tandem repeat DNA

sequences found at different loci in the C. difficile genome49. This typing method

involves amplification of a selection of these loci using a multiplex PCR assay. The

number of repeats at each locus is calculated from the fragment sizes and are combined

to provide an MLVA profile. This method is highly discriminatory and produces

consistent results that are comparable between laboratories49, 54.

For C. difficile, MLST involves the comparison of the nucleotide sequences of seven

highly conserved housekeeping genes (adk, atp, dxr, recA, sodA, tpi and glyA) to

characterise strains55. Sequence types (STs) are assigned based on allelic variants at

each of these genes. The resulting data are unambiguous and can be easily compared

between laboratories using a web-based MLST database

(https://pubmlst.org/cdifficile/)55.

The highest level of discriminatory power is achieved through sequencing the whole

C. difficile genome49, 54. The sequenced whole genome is compared at the nucleotide

level. Since 2006, several closed genome of C. difficile ranging from 4.1 – 4.3 Mbp

have been sequenced and annotated, including strain 630, a RT 012 strain isolated in

Switzerland in 1982; CD37 (RT 009 isolated in the USA in 1980); M68 (RT 017

isolated in Ireland in 2006); M120 (RT 078 isolated from the UK in 2007); G46 (RT

027 isolated from the UK in 2006); R20291 (RT 027 isolated from the UK in 2006) and

BI1 (RT 027 isolated in the USA in 1988)56-58. As an unambiguous and contiguous

sequence of known nucleotide spanning the entire chromosome and plasmids (if

present), these C. difficile genomes play an important role in the next-generation

sequencing data analysis pipeline by providing an extremely high-quality reference for

mapping of draft genomes. C. difficile genomes have an extremely low level of

conservation, with a shared core genome estimated to be as low as 16%59-62. Therefore,

the gold standard for determining the relatedness of C. difficile strains using WGS data

is to compare the non-repetitive, core genome using single nucleotide variant (SNV)

analysis and this has remained so since first performed on C. difficile by Didelot and

https://pubmlst.org/cdifficile/)55


9

colleagues in 201263. The size of the core genome varied depending on the analysed

strains, making it unreliable to compare the results of different analysis. For

comparative purposes, annotated and raw reads of previously sequenced C. difficile

genomes can be downloaded, for inclusion in bioinformatics analyses, from GenBank

(https://www.ncbi.nlm.nih.gov/genbank/) and European Nucleotide Archive (ENA,

https://www.ebi.ac.uk/ena), respectively. By analysing the evolution and genetic

diversity of C. difficile within symptomatic patients, the molecular clock of C. difficile

was estimated by several studies to be 1 – 2 SNVs per-core-genome per-year63-66. This

has revolutionised the investigation of C. difficile transmission, as closely-related strains

with ≤ 2 SNVs differences in their core-genome are now regarded as clonal and

suggestive of direct transmission or a recent common source.

1.2.5 Health and economic burden

The Centers for Disease Control and Prevention (CDC) rated C. difficile as one of the

most important antimicrobial resistance (AMR) threats to public health in the USA.

They estimated that in 2013 in the USA CDI contributed to 250,000 infections and

14,000 deaths67, while a later study estimated that CDI caused around 500,000

infections and 16,000 to 42,000 deaths per annum68. Since 2003, severe CDI cases have

become more common and mortality rates have risen to as high as 17%69. In 2005, the

incidence of CDI in the USA was 11.2 cases per 10,000 population70, while it is lower

in Australia and Europe at 4.0 and 7.0 cases per 10,000 population in 2012 and 2011 –

2013, respectively3, 71. The length of stay (LOS) was significantly longer for patients

with CDI than those without (17 vs 4 days in Ireland72 and 17 vs 6 days in Australia73).

Prolonged LOS (based on bed costs) accounts for ~94% of the total costs associated

with CDI, while the remaining cost was attributed to laboratory testing and antibiotic

treatment72, 74. In the USA, healthcare costs due to CDI have been estimated to range

from US$1 billion to US$3 billion per annum67, 75, 76.

1.2.6 Changes in epidemiology

1.2.6.1 Emergence of the hypervirulent C. difficile RT 027 strain

In 2003, Quebec, Canada, experienced the first CDI outbreak caused by hypervirulent

fluoroquinolones-resistant C. difficile RT 027/toxinotype III/NAP11 (different

terminology reflects different molecular typing methods as discussed in Section 1.2.4).

In Quebec, between 1991 and 2003, the incidence of CDI increased from 35.6 to 156.3

cases per 100,000 population per year, and patients > 65 years old with CDI had

https://www.ncbi.nlm.nih.gov/genbank/

https://www.ebi.ac.uk/ena


10

increased by a massive 8.5 times (102.0 to 866.5 cases per 100,000 population per

year)1. At the same time, severe CDI cases with one or more of the following outcomes:

megacolon, perforation, colectomy, shock or death within 30 days of diagnosis,

increased from 7.1% (12/169) to 18.2% (71/390)1. Furthermore, patients who died

within 30 days of diagnosis surged from 4.7% (8/169) in 1991 to 13.8% (54/390) in

20031.

Shortly after, similar reports of hypervirulent C. difficile RT 027 followed in the USA

and Europe21, 70, 77, 78. In the USA, CDI exceeded 300,000 cases in 2006, a significant

increase from < 150,000 cases in 200020. In Germany, the incidence of CDI increased

from 3.8 cases per 100,000 population in 2002 to 14.8 cases per 100,000 in 200679. In

Finland, the incidence of CDI increased from 16 cases per 100,000 population in 1996

to 34 cases per 100,000 population in 200480, while in Spain, the incidence rate of CDI

in patients aged > 65 years old increased by three-fold between 1997 and 200581. Since

then, C. difficile RT 027 has been documented in all provinces of Canada, in most

European countries and in hospitals of ≥ 40 states in USA20. The incidence of RT 027 in

European countries increased drastically from being the sixth most common RT (5% of

389 isolates) in 200882 to the most prevalent (12.5% of 953 isolates) in 2011 – 201483.

In 2013, the evolutionary history of 151 C. difficile RT 027 predominantly from

hospitalised patients in the USA, Europe, UK, Australia and Asia, isolated between

1985 and 2010, was investigated and phylogenetic analysis was performed65. Two

distinct epidemic lineages of fluoroquinolone-resistant (FQR1 and FQR2) RT 027 were

demonstrated emerging in North America via independently acquiring an identical point

mutation (Thr82Ile) in DNA gyrase subunit A. Bayesian phylogeographic analysis

indicated that the FQR1 lineage emerged in the USA with the earliest isolate originating

from Pittsburgh, Pennsylvania in 2001, and subsequently being transmitted to South

Korea and Switzerland to cause sporadic CDI (Figure 1.3). The FQR2 lineage also

originated in North America (59% probability from the USA and 33% probability from

Canada) and became more geographically widespread with trans-continental

dissemination to Europe and Australia. None of the RT 027 isolates from Japan and

Singapore were part of the epidemic FQR1 or FQR2 lineages nor associated with any

major hospital outbreaks, suggesting they were historical C. difficile RT 027 population

prior to the acquisition of the fluoroquinolone-resistance gene. Fluoroquinolone-

resistant RT 027 was isolated in Hong Kong in 200984 and in Australia in 201185, but


11

Figure 1.3 Transmission events of epidemic fluoroquinolone-resistant C. difficile RT 027, inferred from phylogeographic analysis. (a) Global

spread of two fluoroquinolone-resistant RT 027 lineages [FQR1 (red arrow) and FQR2 (blue arrow)]65. Countries where fluoroquinolone-resistant and

fluoroquinolone-sensitive RT 027 has been reported are in pink and yellow, respectively. (b) Inferred arrivals (blue dashed arrow) and transmission

(red arrow) of FQR2 into and within the UK based on phylogeographic analysis65.


12

has failed to establish in these regions.

C. difficile produces two major virulence factors, toxin A (TcdA) and toxin B (TcdB),

encoded by tcdA and tcdB situated within the PaLoc (Figure 1.4A)86. Epidemic

C. difficile RT 027 like many other C. difficile strains produces both TcdA and TcdB

but reports suggest this strain produces more potent TcdB56, 87. In 2005, McDonald et al.

suggested that due to deletions in the negative regulator gene tcdC, epidemic RT 027

produced more toxin in vitro compared to historical strains collected from 1984 to

199087. However, in 2010, Merrigan et al. quantified C. difficile toxins from epidemic

RT 027 and non-epidemic C. difficile strains, and no significant differences in toxin

production were observed88. In 2012, by restoring the ∆117 frameshift mutation and 18-

bp nucleotide deletion which occurred naturally in the tcdC gene of epidemic RT 027

(R20291, isolated from the UK in 2006) and deleting and restoring the naturally intact

tcdC gene of historic C. difficile RT 012 (strain 630, isolated from Switzerland in 1982)

Cartman et al. found no association between tcdC and toxin production89. In 2009,

Stabler et al. showed that TcdB from epidemic RT 027 was a more potent cytotoxin

than TcdB from previous circulating RTs56. In this study, eight cell lines were

challenged separately with purified TcdB from epidemic RT 027 (R20291), non-

epidemic historic RT 027 (CD196, isolated from France in 1985) and historic RT 012

(strain 630) strains. Compared to historic RT 012, TcdB from epidemic RT 027 was

more potent against all eight cell lines and non-epidemic RT 027 was more potent in six

of the eight cell lines56. This differences between strains in TcdB cytotoxicity was

attributed to variation within the tcdB sequence in the PaLoc, specifically at the 3’

region which encodes the toxin-binding domain56, 90. Epidemic RT 027 strains also

sporulated earlier and produced significantly more spores than non-epidemic C. difficile

strains88.

In addition to TcdA and TcdB, RT 027 strains like several other C. difficile also produce

an extra toxin known as binary toxin (CDT), encoded by cdtA and cdtB on the binary

toxin locus (CdtLoc) (Figure 1.4B)86. The role of binary toxin is still unclear, but the

toxin is postulated to assist with colonisation. A recent report suggested that CDT

facilitates bacterial adhesion to intestinal epithelia by depolymerising the cytoskeleton

to produce microtubule protrusions91. In general, CDT-producing strains have been

linked to increased disease severity92.


13

A

B

Figure 1.4 Genetic map of the C. difficile toxin loci, (A) the 19.6 kb PaLoc and (B)

the 6.2 kb CdtLoc86. Boxes indicate open reading frames with arrows showing the

direction of transcription. Toxin genes are shaded in green, regulatory genes are in red,

tcdE is in blue and genes located outside of the PaLoc is in grey.


14

1.2.6.2 Emergence of CA-CDI

CDI was historically considered as a nosocomial or healthcare-associated infection,

with infection occurring almost exclusively within the hospitalised population. Since the

early 2000s, when CDI outbreaks were being documented across the globe, large-scale

surveillance studies have been implemented in numerous developed countries in an

effort to understand and control the spread of infection. It became apparent from these

efforts that in addition to an increased incidence rate (as discussed in Section 1.2.6.1), a

growing proportion of CDI cases were being acquired in the community. In Australia,

26% of all CDI cases in 2011 were classified as CA-CDI3 an increase of four-fold since

199593. In the USA in 2005, CA-CDI accounted for 41% of all cases also an increase of

four-fold from 19915.

Current classification of CA-CDI and hospital-associated CDI (HA-CDI) cases are

based on guidelines proposed in 2007 by McDonald et al. from the CDC94. These

guidelines define CA-CDI as cases with (i) symptom onset in the community and no

history of hospitalisation within the previous 12 weeks or (ii) symptom onset in ≤ 48 h

after hospital admission. For surveillance purposes, these guidelines also define

“healthcare facility onset, healthcare facility-associated (HO-HCFA)”, “community-

onset, healthcare facility-associated (CO-HCFA)” and “indeterminate CDI cases”

(Figure 1.5).

Notably, CA-CDI cases often lack the traditional risk factors for CDI. CA-CDI cases

are generally younger, female and healthy, with no prior contact with a healthcare

setting or hospitalised patients, and often with no history of antimicrobial usage29, 95.

Other risk factors associated with CA-CDI include having contact with an infant under

the age of 2 years and using proton pump inhibitor96, 97. Depending on the geographical

region, strains that are commonly associated with CA-CDI can differ from those

predominant in the hospital setting. For example, RT 027 which plagued hospitals

across the northern hemisphere is rarely reported among CA-CDI cases in these areas68,

98. In The Netherlands, compared to RT 027, RT 078/toxinotype V/NAP7 is more

frequently CA (6.7% vs 17.5%, OR 2.98 and 95% CI 2.11 – 8.02) and common among

younger population (73.5 vs 67.4 of age, p < 0.01)98. Similar findings have been made

in North America with one study reporting 46% of all RT 078 isolates being CA99. In

2005, RT 078 was the eleventh most common RT (~3%) in 14 different European

countries100 and, by 2008, it had risen to the third most common RT (~8%) in 34

different European countries82 and remained so until 201283.


15

Figure 1.5 Timeline for definitions of C. difficile-associated disease exposure.

Patients with symptom onset during hospitalisation [marked by an asterisk (*)] or

discharged from a healthcare facility within the previous 4 weeks would be classified as

having community-onset, healthcare facility-associated disease (CO-HCFA). Patients

discharged from a healthcare facility between the previous 4 – 12 weeks would be

classified as indeterminate cases. Patient with no history of hospitalisation within the

previous 12 weeks would be classified as community-associated C. difficile associated

disease (CA-CDAD) or also known as CA-CDI. Healthcare facility-onset, healthcare

facility-associated disease (HO-HCFA) are cases with symptom onset happening

between > 48 h of hospitalisation and discharge94.


16

RT 078 is common in production animals. In North America and Europe, it is

consistently one of the most predominant RTs in cattle101-111 and pigs103, 104, 112-117. In

2014, Knetsch et al. reported identical (by WGS) C. difficile RT 078 isolates from

asymptomatic farmers and pigs, suggestive of zoonosis and/or shared environmental

transmission of C. difficile66. While RTs 027 and 078 caused multiple outbreaks in

North America and Europe, they are relatively rare in the Asia-Pacific region and are

reported only sporadically. In the Asia-Pacific, a strain that is commonly associated

with CA-CDI is RT 01295, 118. RT 012 is the first and second most common RT among

CA-CDI cases in South Korea95 and China118, respectively. Recently, it was reported as

the most frequent toxigenic strain in pregnant woman in China119. In Australia, RT 012

is also associated with CA-CDI (L Bloomfield and TV Riley, unpublished data), but the

majority of the RTs causing CA-CDI are also common in hospital-associated cases. The

diversity of C. difficile RTs circulating in humans in Australia is further discussed in

Section 1.4.1.

1.2.7 Transmission of C. difficile in the hospital setting

The transmission of HA-CDI is generally assumed to be either directly from

symptomatic patients, or via healthcare workers or contaminated surfaces. Previous

studies on the molecular epidemiology of C. difficile using MLVA, MLST,

toxinotyping and PCR ribotyping supported this theory due to insufficient

discriminatory power62, 120. However, recent advances in WGS technology has allowed

ultra-fine genomic comparison of strains previously deemed indistinguishable. In 2013,

Eyre et al. performed WGS on 957 C. difficile isolated from all symptomatic patients

identified in the hospitals and community of Oxfordshire between September 2009 and

March 201164. Surprisingly, 45% of the strains had > 10 SNVs difference in their core-

genome (Figure 1.6). Furthermore, of the 333 isolates (35%) with ≤ 2 SNVs difference

from at least one earlier case, only 38% of them had direct hospital contact with a

symptomatic patient. Meanwhile, genetically diverse strains of C. difficile continued to

be identified throughout the study period. Overall, the results of this study suggested

that the sources of C. difficile in a hospital setting are diverse and more varied than just

symptomatic patients. A similar finding was also reported by Didelot et al., who

performed WGS on 486 C. difficile strains isolated from CDI cases in Oxfordshire

between September 2006 and June 201063. In that study, 15 closely-related groups were

identified by phylogenetic analysis but information on patient ward movement

suggested that majority of the genetically-related cases were epidemiologically


17

Figure 1.6 Relationships between 957 C. difficile isolates obtained between

September 2007 – March 2011 as estimated by (A) SNVs, and compared to (B)

epidemiological links64.


18

unrelated (Figure 1.7). The evidence so far suggests that symptomatic patients are not a

prominent reservoir for HA-CDI and a diverse source of C. difficile from asymptomatic

carrier and/or outside of the hospital is plausible. A comprehensive review of C. difficile

in sources outside of a hospital setting is presented below, in Section 1.3.

1.3 POTENTIAL SOURCES OF C. DIFFICILE OUTSIDE OF THE

HOSPITAL SETTING

Reports of increases CA-CDI in North America, Europe and the Asia-Pacific region

have prompted the investigation of other potential sources/reservoirs of C. difficile

outside of the hospital setting, particularly production animals, food and the

environment. The methods for detection and molecular epidemiology of C. difficile in

production animals, food and the environment are described here.

1.3.1 Methods for detection of C. difficile in production animals, food and

the environment

1.3.1.1 Production animals

Currently, there are no guidelines for testing for CDI in animals. However, toxigenic

culture has been popular among production animal studies. Culturing methods were

similar across studies and as previously described121, 122. Molecular typing of C. difficile

isolates was performed as described in Section 1.2.4.

1.3.1.2 Food

Currently, there are also no international guidelines for testing C. difficile in foods. To

date, only two studies have performed quantification/direct culture of C. difficile spores

in foods123, 124. In the first study, an entire piece of chicken was added to 50 ml of

phosphate-buffered saline (PBS), thoroughly mixed by hand and 100 µl aliquots were

inoculated onto selective agar123. In the second study, 25 g of meat was rinsed with 25

ml of PBS and thoroughly massaged by hand124. Serial dilutions were performed and

100 µl aliquots were inoculated onto selective agar for incubation. According to these

studies the levels of C. difficile contamination in retail meats were very low with most

positive samples only positive on enrichment culture and quantifiable samples typically

have 20 – 240 spores/g123, 124.

The most widely used method for C. difficile detection in foods are direct incubation of

the samples into enrichment broths for a median of 10 days with a median food to broth

ratio of 1:10102, 110, 123-142. This is followed by alcohol shock and centrifugation at a


19

Figure 1.7 Hospital ward relationships between cases with genetically closely

related C. difficile. Each row represents a different closely related group of CDI cases

based on SNV analysis. Cases were linked by horizontal lines if genetic analysis

determined that transmission was possible. Within each group, cases that are

epidemiologically linked with another are indicated by circles of the same unique

colour, and cases with no identified epidemiological connections are shown in grey63.


20

minimum of 3,800 × g for 10 min, and the resulting pellet was subcultured onto a

selective agar for anaerobic incubation. With the exception of a longer incubation

period, this method is similar to the consensus method recommended by Limbago et al.

in 2012, where the efficacy of C. difficile recovery from meat with three different

culture methods was evaluated140.

1.3.1.3 Environment

The detection of C. difficile in water samples by direct culture generally involves

membrane filtration. Samples from lake, river, sea and puddle, ranging from 100 ml to

1000 ml, were filtered through 0.2 – 0.45 µm pore size cellulose filter membranes and

placed on selective agars136, 143-147. For enrichment culture of water samples, these

filtered membranes were immerged into broths for a median of 6 days136, 143.

Most studies that detect C. difficile in soil samples performed direct culture by alcohol

shocking 5 g of the samples143-145, 148. This was either followed by inoculating the soil

sediments onto selective agars143, 145 or dipping a sterile cotton swab into the alcohol-

soil mixture and swabbing it onto a selective agar144, 148. Only one study has

investigated the prevalence of C. difficile in soil sediment using enrichment culture136.

In that study, 20 g of the samples was added to 80 ml of enrichment broths for a 10 days

anaerobic incubation136.

Of the three studies that investigated the prevalence of C. difficile in household

environments149-151, only one performed direct culture on the samples151. In that study,

household surfaces were swabbed with a sponge and then homogenised in a stomacher

bag with 50 ml sterile PBS151. Centrifugation was performed on the suspension, 45 ml

of the supernatant was removed, the pellet was resuspended and 200 µl of which was

spread plated onto a selective agar. Enrichment was carried out via immersing the

sponges in broths for 5 days150 or homogenised with 50 ml of sterile PBS, followed by

centrifugation and inoculation of resuspended pellet into enrichment broth151.

1.3.2 Prevalence and molecular epidemiology of C. difficile

1.3.2.1 Production animals

C. difficile has been isolated from wild and domesticated animals, but most notably,

from food production animals (calves, cattle, sheep, lambs, piglets, pigs and poultry)112,

121, 122, 148, 152-155. Production animals, especially cattle, sheep and pigs, are now viewed

as significant reservoirs of C. difficile. The prevalence and circulating strains in


21

livestock varies between studies, most likely influenced by the age of animals tested and

geographical region of the study. The average reported prevalence of C. difficile in

calves was 20.7% (calculated from 17 reports), ranging from ~0.5% in Switzerland110 to

~20.0% in Belgium104 and the USA111 to ~50.0% in Australia122 and Canada107 (Table

1.1). The average prevalence of C. difficile in cattle was 8.2% (calculated from 12

reports), ranging from ~0.5% in the Netherlands156 and USA157 to ~10.0% in Belgium103

and Slovenia158. The average prevalence of C. difficile in sheep was 10.4% (calculated

from five reports), ranging from 0.6% in Australia121 to ~20.0% in Netherlands156 and

Turkey159. The prevalence of C. difficile in lambs was only investigated in one study,

6.5% in Australia121. The average prevalence of C. difficile in piglets was considerably

higher at 66.9% (calculated from 18 reports), ranging from ~25.0% in Spain160 to

~50.0% in Japan161, Slovenia152 and USA162 to 100.0% in the Netherlands116 and

Spain163. The average prevalence of C. difficile in pigs was 13.7% (calculated from 20

reports), ranging from 0.0% in Belgium104 and Switzerland110 to ~20.0% in Ireland114

and Taiwan164 to ~50.0% in the USA165. Meanwhile, the average prevalence of

C. difficile reported for broiler chickens was 15.7% (calculated from two reports),

ranging from 2.3% in the USA128 to 29.0% in Zimbabwe148. The average prevalence of

C. difficile reported for hens was 22.7% (calculated from three reports), but varied

greatly from 0.1% in the USA157 to 62.3% in Slovenia166. RT 078 predominated in most

of these studies and, in some countries, RTs 027 and 014 were the most prevalent

(Table 1.1). All three RTs commonly cause humans CDI in these regions82, 83.

1.3.2.2 Foods

C. difficile has been isolated from retail meats (veal, beef, pork, lamb, chicken and

turkey), seafood (clams, salmon, shrimp and mussels) and vegetables (lettuce, pea

sprouts, ginger, carrots, salad) throughout North America, Europe and the Middle East

(Table 1.2). The prevalence of C. difficile in meats ranges up to 42.0% in the USA134,

but most studies report a much lower prevalence, especially in Europe (~2.0%)110, 127,

167. The prevalence of C. difficile in seafood has varied considerably from < 5.0% in

Canada131, USA168 and Wales143 to ~50.0% in Italy133, largely depending on the level of

sewage contamination in local rivers133. The overall prevalence of C. difficile in

vegetables has been low (~4.0%)132, 169. The most commonly isolated strains in these

studies were once again clinically important RTs 014, 027 and 078.


22

Table 1.1 Global prevalence of C. difficile in production animals.

Host Country Year Prevalence, n (%) Predominant strains Ref

Bovine

Calves Australia 2013 203/360 (56.4) RTs 127/033/056 122

Calves Belgium 2012 4/18 (22.2) RT 078 104

Calves Belgium 2012 21/300 (7.0) RTs 078/126/045 105

Calves Belgium 2017 8/71 (11.3) RT 015 170

Calves Canada 2006 31/278 (11.2) RTs 017/078/027/014 106

Calves Canada 2011 88/172 (51.2) RTs 078 107

Calves Canada 2012 36/874 (4.1) RTs 078 101

Calves Germany 2013 176/999 (17.6) RTs 033/078/045 108

Calves Italy 2015 87/440 (20.7) RTs 078/012/126 109

Calves Iran 2014 90/150 (60.0) NA 171

Calves Netherland 2012 6/100 (6.0) RT 012 156

Calves Slovenia 2009 4/42 (9.5) RT 077 153

Calves Slovenia 2016 243/2442 (10.0) RT 033 158

Calves Switzerland 2010 1/204 (0.5) RT 078 110

Calves Switzerland 2012 6/47 (12.7) RTs 033/066/070/003 172

Calves United States 2008 64/253 (25.3) RT 078 111

Calves United States 2012 56/200 (28.0) NA 173

Cattle Belgium 2012 14/202 (6.9) UCL118a† 104

Cattle Belgium 2013 10/101 (9.9) RT 078 103

Cattle Belgium 2013 8/101 (7.9) UCL5a†/UCL16u† 103

Cattle Belgium 2017 6/109 (5.5) RT 020/UCL24† 170

Cattle Iran 2015 2/50 (4.0) IR12†/IR42† 174

Cattle Netherlands 2012 1/100 (1.0) RT 012 156

Cattle Slovenia 2016 54/540 (10.0) RT 071 158

Cattle Turkey 2018 83/247 (33.6) RT 027 159

Cattle United States 2011 24/186 (12.9) A† 175

Cattle United States 2011 220/4290 (5.1) NA 176




23


Ovine

Lambs Australia 2013 14/215 (6.5) RT 101 121

Sheep Australia 2013 1/156 (0.6) QX 137† 121

Sheep Iran 2015 1/50 (2.0) IR46† 174

Sheep Netherlands 2012 2/11 (18.2) RTs 015/097 156

Sheep Slovenia 2014 6/105 (5.7) Toxinotype V 178

Sheep Turkey 2018 78/308 (25.3) RT 027 159

Porcine

Piglets Australia 2013 114/185 (61.6) RT 237 179

Piglets Australia 2015 154/229 (67.2) RTs 014/033/QX 009† 154

Piglets Belgium 2012 18/23 (78.3) RT 078 104

Piglets Canada 2010 90/121 (74.4) RT 078 112

Piglets Czech 2012 19/30 (63.3) Toxinotype 0 180

Piglets Germany 2013 147/201 (73.1) RTs 078/126 113

Piglets Ireland 2017 34/44 (77.3) RT 078 114

Piglets Japan 2014 69/120 (57.5) P1†/P2†/P3† 161

Piglets Netherlands 2011 71/71 (100.0) RT 078 116

Piglets Slovenia 2008 133/257 (51.8) Toxinotype V/0 152

Piglets Slovenia 2009 247/485 (50.9) RT 066 153

Piglets Spain 2009 140/541 (25.9) NA 160

Piglets Spain 2009 144/144 (100.0) RT 078 163

Piglets Sweden 2014 45/67 (67.2) RT 046 181

Piglets Taiwan 2016 114/134 (85.1) RTs 078/126/127 182

Piglets United States 2009 61/122 (50.0) NA 162

Piglets United States 2010 241/513 (47.0) Toxinotype V 183

Piglets United States 2012 183/251 (72.9) Toxinotype V 165

Pigs Belgium 2012 0/194 (0.0) 104

Pigs Belgium 2013 1/100 (1.0) RT 078 103

Pigs Ireland 2017 33/156 (21.2) RT 078 114

Pigs Japan 2013 2/250 (0.8) NA 184

Pigs Korea 2015 2/659 (0.3) RT 078 115

Pigs Netherlands 2009 33/100 (33.0) RT 078 117



24




Pigs Spain 2017 34/398 (8.5) RTs 078/005/126 187

Pigs Switzerland 2010 0/165 (0.0) 110

Pigs Taiwan 2017 21/92 (22.8) RT 126 164

Pigs United States 2009 15/382 (3.9) NA 162



Pigs United States 2011 252/2936 (8.6) Toxinotype V 188





Poultry

Broilers United States 2011 7/300 (2.3) Toxinotype V/NAP7 128

Broilers Zimbabwe 2008 29/100 (29.0) NA 148

Hens Netherlands 2012 7/121 (5.8) RT 014 156

Hens Slovenia 2008 38/61 (62.3) Toxinotype 0 166

Hens United States 2014 1/680 (0.1) NA 157

†, internal nomenclature; NA, data not available; NAP7, North American Pulse Type 7.

Strains commonly associated with human CDI are shown in red. Other less common

strains that have been associated with human CDI are shown in green.


25

Table 1.2 Global prevalence of C. difficile in foods.

Origin Country Year Prevalence, n (%) Predominant strains Ref

Raw meat

Veal Canada 2007 1/7 (14.3) M31† 126

Veal Canada 2009 3/65 (4.6) M26†/J†/K† 138

Veal Netherlands 2011 0/19 (0.0) 127

Beef Belgium 2014 3/133 (2.3) RTs 014/078 167

Beef Canada 2007 11/53 (20.8) RTs 077/M26†/M31† 126

Beef Canada 2009 10/149 (6.7) RTs 077/014/M26† 138

Beef Canada 2009 14/115 (12.2) RT 078 124

Beef Costa Rica 2013 1/67 (1.5) RT 029 135

Beef Iran 2014 2/121 (1.7) RT 078 125

Beef Netherlands 2011 0/145 (0.0) 127

Beef Saudi Arabia 2018 7/200 (3.5) NA 190

Beef/pork‡ Switzerland 2010 0/46 (0.0) 110

Beef United States 2009 13/26 (50.0) RTs 027/078 134

Pork Belgium 2014 5/107 (4.7) RT 014 167

Pork Canada 2009 14/115 (12.2) RT 078 124

Pork Costa Rica 2013 2/66 (3.0) RT 029 135

Pork Netherlands 2011 0/63 (0.0) 127

Pork Taiwan 2017 17/62 (27.4) RTs 126/127/014 164

Pork United States 2009 6/20 (30.0) RTs 027/078 134

Pork United States 2011 23/243 (9.5) Toxinotype V/NAP7 129

Pork United States 2012 2/102 (2.0) RT 078 130

Chicken Canada 2010 26/203 (12.8) RT 078 123

Chicken Costa Rica 2013 1/67 (1.5) RT 029 135

Chicken Netherlands 2011 7/257 (2.7) RT 003 127

Chicken United States 2011 7/96 (7.3) Toxinotype V/NAP7 128

Buffalo Iran 2014 6/67 (9.0) RT 078 125

Goat Iran 2014 3/92 (3.3) RT 078 125

Lamb Netherlands 2011 1/16 (6.3) RT 045 127

Sheep Saudi Arabia 2018 2/200 (1.0) NA 190

Turkey United States 2009 4/9 (44.4) RT 078 134


26


Ready-to-eat meat

Beef Cote d’Ivoire 2014 30/223 (13.5) NA 191

Pork Taiwan 2017 15/65 (23.1) RT 014 164

Sausages United States 2009 11/23 (47.8) RTs 027/078 134

Seafood

Clams Italy 2012 10/19 (52.3) RTs 014/020/001/078 133

Clams Italy 2015 3/13 (23.1) RTs 010/100/204 192

Fishes Wales 1996 0/107 (0.0) 143

Mussels Italy 2012 16/33 (48.5) RTs 014/020/001/078 133

Mussels Italy 2015 33/912 (3.6) RTs 126/010 192

Seafood‡ Canada 2011 5/119 (4.8) RT 078 131

Shellfish Italy 2011 9/21 (42.9) RTs 003/010 136

Shellfish United States 2014 3/67 (4.5) Toxinotype V/XII 168

Vegetable

Leafy‡ United States 2014 3/65 (4.6) RT 027 169

Leafy‡ Wales 1996 7/300 (2.4) NA 143

Root Canada 2010 5/111 (4.5) 078 132

Root United States 2014 0/39 (0.0) 169

Salads France 2013 3/104 (2.9) RTs 001/014/020/077 137

Salads Iran 2015 6/106 (5.7) NA 193

Salads Scotland 2009 3/40 (7.5) RTs 017/001 194

†, internal nomenclature; ‡, unspecified food type; NA, data not available; NAP7, North

American Pulse Type 7. Strains commonly associated with human CDI are shown in

red. Other less common strains that have been associated with human CDI are shown in

green.


27

1.3.2.3 Environment

C. difficile has been isolated from environmental samples, including from natural

settings (soil, rivers, sea, lakes and sediments) to household areas (toilets, floors,

bathroom sinks and soles of shoes) (Table 1.3). The average prevalence of C. difficile

reported in the natural environment was ~30.0%, ranging from 0.0% of 100 soil and 100

water samples in Lagos State of Nigeria145 to ~40.0% of five water samples from the

Southern Tyrrhenian Sea in Italy136 and ~40.0% of 146 soil samples from a rural

community in Zimbabwe144 to 87.5% of 16 river samples in Wales143. The average

prevalence of C. difficile in households was slightly higher at ~35.0%, ranging from

17.2% on bathroom sinks of patients with recurrent CDI151 to 52.5% in sandboxes for

children and dogs149. RTs 014 and 078 predominated in most of these environmental

studies.

1.3.3 Genetic relationship between C. difficile strains from various sources

To obtain a better understanding of the transmission routes of C. difficile, several

studies have compared C. difficile strains from various sources using typing methods

that offer greater discriminatory power than PCR ribotyping or PFGE. In 2010, 158

C. difficile RT 078 isolates from humans (n = 102) and pigs (n = 56) were investigated

using MLVA; 85.0% of the isolates were genetically related, irrespective of porcine or

human origin195. In 2014, Knetsch et al. used WGS to compare 65 C. difficile RT 078

from pigs (n = 19), asymptomatic farmers (n = 15) and hospitalised patients (n = 31) in

The Netherlands66. A phylogenetic tree based on core genome SNV analysis showed

clustering of pig isolates with isolates from farmers. This included indistinguishable

strains (0 SNV differences) from pigs and farmers on the same farm, providing the first

evidence of interspecies transmission or common recent exposure. In 2017, 248

C. difficile RT 078 isolates from humans and animals from 22 countries (North

America, Europe, Australia and Asia) were analysed using WGS196. Core genome

maximum likelihood phylogeny was performed and six clusters of human and animal

strains from various countries were identified (Figure 1.8 and Table 1.4), suggesting

that C. difficile RT 078 frequently crosses international borders. To date, no study has

used WGS to investigate the genetic similarity of C. difficile from the environment with

C. difficile from humans, animals or foods.


28

Table 1.3 Global prevalence of C. difficile in the environment.


Nature

Lake Wales 1996 7/15 (46.7) NA 143

River Slovenia 2010 42/69 (60.9) RT 014 146

River Wales 1996 14/16 (87.5) NA 143

Sediment† Canada 2014 25/64 (39.1) RT 078 197

Sediment† Italy 2011 0/5 (0.0) 136

Soil‡ Nigeria 2014 0/100 (0.0) 145

Soil‡ Slovenia 2016 29/79 (36.7%) RTs 014/020, 010 147

Soil‡ Wales 1996 22/104 (21.2) NA 143

Soil§ Zimbabwe 2006 54/146 (37.0) NA 144

Soil¶ Zimbabwe 2008 22/100 (22.0) NA 148

Sea Italy 2011 2/5 (40.0) RT 003 136

Sea Wales 1996 7/15 (46.7) NA 143

Water‡ Nigeria 2014 0/100 (0.0) 145

Waterǭ Slovenia 2016 15/104 (14.4%) RTs 014/020, 010 147

Water¶ Zimbabwe 2006 14/234 (6.0) NA 144

Household

Sandbox Spain 2018 21/40 (52.5) RTs 014/039 149

Shoeδ United States 2014 25/63 (39.7) RT 001 150

Toiletδ United States 2014 5/15 (33.3) RT 001 150

Floor dustδ United States 2014 4/12 (33.3) RT 001 150

Toiletε United States 2016 8/30 (26.7) NAP7/NAP6 151

Vacuumε United States 2016 11/27 (40.7) NAP4/NAP11 151

Sinkε United States 2016 5/29 (17.2) NAP7/NAP6 151

†, sediment collected from river or sea bank; ‡, unspecified type; §, collected from

homesteads in a rural community; ǭ, collected from puddle water; ¶, collected from

market places; δ, collected from households (n = 30) in the community; ε, collected from

household of patients (n = 16) with recurrent CDI and geographically- and age- matched

controls (n = 8); NA, data not available; NAP7, North American Pulse Type 7. Strains

commonly associated with human CDI are shown in red. Other less common strains that

have been associated with human CDI are shown in green.


29

Figure 1.8 Maximum-likelihood phylogeny of 248 C. difficile RT 078 based on

Average Nucleotide Identity (ANI) calculated by pairwise comparison of genome

assemblies. Branch and taxa colouring/labelling are as follow: dark blue, human; red,

animal; orange, food; light blue, environment; dark green, Europe; purple, North

America; pink, Asia; and light green, Australia. Closely related clusters (see Table 1.4)

containing both human and animal isolates are labelled 1 – 6 and highlighted in

yellow196.


30

Table 1.4 Clusters of highly similar human and animal C. difficile RT 078 isolates

identified by Knetsch et al. using core genome analysis196. ANI for human isolate

compared to the animal isolate is shown.

ENA ID, European Nucleotide Archive identification code; ANI, Average Nucleotide

Identity.


31

1.4 C. DIFFICILE IN AUSTRALIA

CDI is not a notifiable disease in Australia so there is no legal requirement for medical

practitioners to inform state Health Departments of cases. However, in 2010 the

Australian Commission on Safety and Quality in Healthcare mandated hospital

surveillance in all States and Territories to be undertaken as a requirement for hospital

accreditation. A comprehensive summary of previous C. difficile surveillance studies in

Australia is presented here.

1.4.1 C. difficile in humans

Between 1983 and 1992, the incidence of CDI in WA significantly increased from 22

cases/100,000 patient days (PD) to 50 cases/100,000 PD17. This was attributed to the

increased use of third-generation cephalosporins in the same time period. From 1993 to

1998, the incidence of CDI in WA remained relatively stable, averaging at 2 – 3 cases

per 1,000 discharges18. However, in November 1998, a change in antimicrobial policy

in WA hospitals saw a marked decrease in the prescription of third-generation

cephalosporins and the incidence of CDI dropped to 0.87 and 0.7 cases per 1,000

discharges in 1999 and 2000, respectively18.

While North America and Europe were plagued with outbreaks of CDI caused by RT

027 in the early 2000s, it was not until 2008 that Australia had its first isolation of RT

027198. Moreover, it was a recurrent infection acquired when the patient was travelling

in the USA. The first local acquisition of fluoroquinolone resistant RT 027 was in 2010

in Melbourne85. This was a hospital-associated infection and there were at least two

other subsequent confirmed cases of CDI caused by RT 027 in the same hospital at this

time. Fortunately, RT 027 has failed to establish in Australia; most likely due to

Australia’s conservative policy on fluoroquinolone use in both humans and animals199.

In 2012, the incidence of hospital-identified CDI in WA increased from 32.5/100,000

PD in 2011 to 40.3/100,000 PD, with peaks occurring at the end of the year (spring and

summer period of the southern hemisphere)3. The adaptation of more sensitive

diagnostic testing at this time may have contributed to the increase but was unlikely to

be the sole reason for the increased incidence rate as it would not account for those

observed peaks. Similar trends of seasonality were also observed in other states200-202. In

a cross-sectional study conducted in two Australian hospitals between 2012 and 2014,

in WA and Queensland, asymptomatic C. difficile colonisation also peaked in the

summer period200. In Victoria, seasonality was evident with HA-CDI peaking in


32

summer and early autumn (January – March) and lowest in winter (June –

September)201. Between 2010 and 2014, the incidence of HA-CDI in Victoria was

diminishing (79.8% in 2010, 76.5% in 2011, 73.3% in 2012, 68.3% in 2013 and 67.1%

in 2014) while the incidence of CA-CDI increased (16.4% in 2010, 22.6% in 2011,

26.1% in 2012, 31.0% in 2013 and 32.3% in 2014), indicating the growing need for

enhanced surveillance in the community setting201. Meanwhile, in late 2011, Tasmania

experienced a 53% increase in the number of CDI cases202. In an attempt to understand

the reasons behind these increases in incidence rate, a population-based study was

performed, and all cases of laboratory identified CDI that occurred between 2010 and

2011 were identified. It revealed that the rate of CA-CDI increased from 10/100,000

population in 2010 (95% CI, 7.5 – 13.2) to 17/100,000 population in 2011 (95% CI 14 –

21.5)202. Overall, nearly 30% of all CDI cases had no traditional risk factors and were

classified as CA-CDI3.

Around 86% of C. difficile isolated from clinical specimens in Australia have the toxin

profile A+B+CDT-203. Binary toxin-producing strains are rare. The most common RTs

both locally (in WA) and nationally were frequently RTs 014/020, 002 and 056 (Figure

1.9). C. difficile strains that predominated in HA-CDI (Figure 1.9F) also predominated

in CA-CDI (Figure 1.9C and 1.9G)203, 204.

1.4.2 C. difficile in Australian livestock

Reports of indistinguishable C. difficile from production animals and humans in North

America and Europe have raised concerns regarding potential zoonotic transmission of

C. difficile. However, until recently, little was known about the prevalence of C. difficile

in Australian production animals. The first study was by Squire et al. in 2009 and they

found C. difficile in pig herds in WA179. The circulating C. difficile strains in those

piggeries were dominated by a unique type, RT 237 (A-B+CDT+) which was of the

same clade (clade 5) and ST 11 as RT 078, the major production animal RT in the

northern hemisphere and associated with CA-CDI. There are at least six distinct clades

of C. difficile (clades 1, 2, 3, 4, 5 and C-I) circulating in the world, identified via

phylogenetic analysis based on in silico MLST205. C. difficile RT 237 was commonly

isolated from neonatal piglets with idiopathic diarrhoea (scouring). To date, RT 078 has

not been isolated from Australian livestock.


33

Figure 1.9 C. difficile RT diversity in humans, in Australia, as reported by seven

separate studies: (A) A WA surveillance study from October 2011 – September 2012

(747 isolates)206. (B) A cross-sectional study of CDI from December 2011 – May 2012

in two tertiary hospitals in WA (70 isolates)207. (C) A study that investigated the

epidemiology of CA-CDI from July 2013 – June 2014 across six emergency

departments of WA hospitals (27 isolates)204. (D) A WA study in July 2014 where all

faecal samples submitted for enteric testing were surveyed for C. difficile (59


34

isolates)208. (E) A month-long national study in 2010 (330 isolates)209. A study that

compare the circulating RTs of C. difficile among (F) HA-CDI (96 isolates) and (G)

CA-CDI (152 isolates) cases in two Australian States over a 3-year period (2012 –

2014)203. (H) A national study where 151 isolates sampled from August – September

2014, February – March 2015 and August – September 2015 were randomly selected

for ribotyping210. RTs are prefixed with either the official CDRN number or internal

nomenclature (QX).


35

In an attempt to understand the molecular epidemiology of C. difficile in Australian

production animals across the country, in 2012 and 2013, our research laboratory

undertook three national surveillance studies of C. difficile in Australian production

animals121, 122, 154. Summaries of these studies are provided below.

1.4.2.1 C. difficile in Australian cattle and calves

In 2012, a cross-sectional study investigated the prevalence and molecular

epidemiology of C. difficile in Australian cattle and calves at slaughter122. A total of 975

samples were collected from carcass washings, gastrointestinal contents and faeces from

cattle and calves at abattoirs across five Australian States. Carcass washings of adult

cattle were collected after hosing of the whole carcasses were completed.

Gastrointestinal contents of adult cattle were collected directly from the large intestine

within the viscera processing area of the slaughter line. Direct and enrichment culture

was performed and C. difficile isolates were characterised by toxin gene profiling and

PCR ribotyping. The prevalence of C. difficile in faeces was high in young calves and

decreased significantly as the animal aged: 56.4% (203/360) in < 7-day-old calves,

3.8% (1/26) in 2- to 6- month-old calves and 1.8% (5/280) in adult cattle. No C. difficile

was isolated from samples of carcass washings (n = 151) or gastrointestinal contents (n

= 158); all of which were from adult cattle only.

Nearly all of the isolates were toxigenic (98.6%, 206/209). Over half of the isolates

(54.5%, 114/209) were positive for toxin genes tcdA, tcdB, cdtA and cdtB

(A+B+CDT+). The remaining isolates yielded the following toxin profiles: A+B+CDT-

(22.0%, 46/209), A-B-CDT+ (19.6%, 41/209), A-B+CDT+ (1.9%, 4/209), A-B-CDT-

(1.4%, 3/209) and A-B+CDT- (0.5%, 1/209). Twenty-one RTs were identified (Figure

1.10A) and RT 078 was not found. The most common RTs were 127, 033, 056 and 126,

comprising 83.3% (174/209) of isolates. These RTs have been found in humans in

Australia (Figure 1.9), with RTs 014 and 056 frequently being in the top three most

commonly isolated RT in human CDI in Australia203, 204, 206-209. These findings suggest

that young cattle could be potential reservoirs for toxigenic C. difficile strains that cause

disease in humans. In Australia, male calves at the age of < 7-day-old are slaughtered as

a waste product of dairy farming at a time when they apparently have a high rate of

C. difficile carriage. It is not known how this might pose a transmission risk.


36

Figure 1.10 C. difficile RTs in (A) Australian cattle, (B) sheep and (C) pigs, as

reported from a cross-sectional study of cows and calves at slaughter in 2013 (209

isolates)122, a national ovine study in 2011 (15 isolates)121 and a national porcine study

in 2014 (154 isolates)154, respectively. RTs are prefixed with either the official CDRN

number or internal nomenclature (QX). RTs that are commonly associated with human

CDI in Australia is shown in red (RTs 014, 018 and 056).


37

1.4.2.2 C. difficile in Australian sheep and lambs

Also, in 2012, a study was conducted to investigate the prevalence and molecular

epidemiology of C. difficile in Australian sheep and lambs121. A total of 300 faecal

samples from sheep and lambs were collected from abattoirs in South Australia,

Victoria and New South Wales. A further 71 faecal samples from lambs were collected

from two farms in WA. Enrichment culture was performed and C. difficile isolates were

characterised by toxin gene profiling and PCR ribotyping. The prevalence of C. difficile

was 6.5% (14/215) in lambs (< 1 year of age) and 0.6% (1/156) in sheep (≥ 1 year of

age). No significant differences in prevalence were observed between lambs at slaughter

and on farms (4.2% vs 11.2%, p = 0.08), both < 1 year of age.

All of the isolates were toxigenic; 93.3% (14/15) were positive for toxin genes tcdA and

tcdB but negative for binary toxin genes cdtA and cdtB (A+B+CDT-) and one isolate

had the toxin profile A-B+CDT+. Seven RTs were identified. Neither epidemic RTs

027 nor 078 were found. The most common RTs were 101 (40.0%, 6/15), followed by

QX 049 (13.3%, 2/15), QX 098 (13.3%, 2/15), 137 (13.3%, 2/15), QX 029 (6.7%,

1/15), 056 (6.7%, 1/15) and QX 137 (6.7%, 1/15) (Figure 1.10B). With the exception of

RT 056, all other RTs were not associated with CDI in humans. These findings

suggested that Australian sheep and lambs were unlikely to be a major reservoir of

human infection. The prevalence of C. difficile in < 7-day-old Australian lambs was not

investigated at the time, but it is likely to be high as recent studies have shown that

C. difficile is more common in neonatal animals211.

1.4.2.3 C. difficile in Australian neonatal pigs

From 2012 to 2013, nationwide surveillance was conducted to investigate the

prevalence and molecular epidemiology of C. difficile in Australian pigs154. In total, 229

faecal samples were collected from neonatal piglets (aged < 7-day-old) from 21 farms

across five Australian States. Direct and enrichment culture was performed and

C. difficile isolates were characterised by toxin gene profiling and PCR ribotyping.

Overall, 67.2% (154/229) of faecal samples were positive for C. difficile with no

significant differences in prevalence between piglets with or without diarrhoea, or

between farms with or without a history of idiopathic diarrhoea.

Overall, 84.4% (130/154) of the isolates were toxigenic. The majority of isolates

(43.5%, 67/154) were positive for toxin genes tcdA and tcdB but negative for binary

toxin genes cdtA and cdtB (A+B+CDT-). The remaining isolates yielded the following


38

toxin profiles: A-B-CDT+ (29.2%, 45/154), A-B-CDT- (15.6%, 24/154), A-B+CDT+

(10.4%, 16/154) and A+B+CDT+ (1.3%, 2/154). Similar toxin profile of isolates were

recovered from piglets with no history of idiopathic neonatal diarrhea for ≥ 6 months

and with idiopathic neonatal diarrhea for ≥ 6 months, except non-toxigenic strains (A-

B-CDT-) were more prevalent in non-diarrheic farms than in the diarrheic farms (34.6%

vs 5.9%, p = 0.001)154. Twenty-three RTs were identified. RT 014 (A+B+CDT-) was

the most prevalent RT, comprising 23.4% of the isolates (Figure 1.10C). This is of

significance as RT 014 is currently the most prevalent RT in human infections in

Australia, accounting for ~30% of all CDI cases200, 203, 204, 207-210, in both hospital-

(Figure 1.9F) and community-associated cases (Figure 1.9C and 1.9G)203. It is also

consistently one of the most common RTs causing CDI in Europe82, 83 and Asia212, 213.

These findings demonstrated that the colonisation of C. difficile in Australian neonatal

piglets is widespread among both symptomatic and asymptomatic piglets, and

Australian piglets are a potential reservoir for CDI in humans. How this could be the

case is currently unclear.

1.4.3 Long-range interspecies transmission of RT 014 between Australian

pigs and humans

By analysing the evolution and genetic diversity of C. difficile strains within

symptomatic patients, the molecular clock of C. difficile was previously estimated by

several studies to be 1 – 2 SNVs per-genome per-year, and strains with ≤ 2 SNVs

differences in their core genome are deemed clonal and direct transmission is possible63-

66. In 2017, WGS and core genome phylogenetic analysis was performed on a collection

of 40 C. difficile RT 014 isolates from Australian pigs and humans, from four

Australian States214. Phylogenies based on MLST and core orthologous genes showed

clustering of strains from Australian pigs and humans, indicative of a very recent shared

ancestry. SNV analysis based on non-recombinant core genome showed 42% of human

strains had a clonal relationship (0 – 2 SNVs) with one or more pig strains, consistent

with recent inter-species transmission (Figure 1.11). Interestingly, some of the

C. difficile human and porcine clones were from different states (separated by thousands

of kilometres) and collected months apart (Figure 1.11 CG2 and CG3). Therefore,

direct transmission between pigs and humans was unlikely. In addition to separation by

vast distances, 50% of the human isolates in these clonal groups (CGs) had no recent

healthcare exposure, suggesting acquisition of C. difficile from unknown sources in the

community. Collectively, these data strongly suggest that over an extended period of


39

Figure 1.11 Maximum-likelihood phylogeny of 44 C. difficile RT 014 from pigs and

humans based on non-recombinant SNVs (n = 1287) identified after mapping all

sequence reads against the CD630 reference genome (accession AM180355). Branch

and taxa colouring/labelling for RT 014 strains are as follow: teal, human (H); purple,

porcine (P); red, ST2 (n = 21); green, ST13 (n = 16); blue, ST49 (n = 7). Taxa labels

include ID: ORIGIN-SITE, ISOLATION DATE, and ACQUISITION STATUS (if

known). Black boxes indicate a CG where all isolates differ by ≤ 2 SNVs. To enhance

the visual resolution of the relative evolutionary distances between test genomes,

CD630 was omitted from the final phylogeny tree214.


40

time there has been long-range interspecies transmission of C. difficile RT 014 between

Australian pigs and humans or a shared common source. The transmission pathway

remains unknown.

1.4.4 Gaps in knowledge

Overall, these studies clearly demonstrate that Australian livestock is a reservoir for

C. difficile in Australia, including strains of clinical importance (RTs 014/020 and 056).

However, like many overseas studies, closely-related and in some cases

indistinguishable animal and human strains in Australia were separated by vast

distances and with no known history of prior contact between the two hosts. Thus, the

transmission pathway between humans, production animals and/or a common source

remains unclear.

Following the isolation of C. difficile in foods, soil and water in North America and

Europe, it was hypothesised that C. difficile might be contributing to CDI (especially

CA-CDI) through contaminated foods and/or the environments; possibly due to

agricultural recycling of effluent and manure from the livestock industry. To date, little

is known about the prevalence of C. difficile in Australian retail foods and no study has

investigated the prevalence of C. difficile in the Australian environment.

1.5 THESIS AIMS AND RESEARCH OBJECTIVES

The aim of this PhD project was to identify the potential sources/reservoirs of

C. difficile in the community that might be contributing to CDI in humans, with a focus

on food and the environment.

The specific research objectives were:

i. determine the prevalence and genotypes of C. difficile in the retail foods and

environment of WA;

ii. examine the susceptibility of food and/or environmental C. difficile isolates to

antimicrobial agents and food preservatives;

iii. define the extent of genetic overlap and potential transmission between human,

food and environmental C. difficile population;

iv. investigate possible risk factors for predominantly community-associated CDI;

and,

v. explore the relatedness of C. difficile from Australia to those that originated in

Asia.

Chapter 2: Materials and methods

41

CHAPTER 2

MATERIALS AND METHODS


42

2.1 MATERIALS

2.1.1 Culture media

All agar plates and broths except for ChromIDTM and supplemented Brucella broth were

prepared by Pathwest Laboratory Medicine Media (Pathwest Media), Mt Claremont,

WA. These include:

i. Blood agar (BA).

ii. Brain heart infusion broth (BHIB) supplemented with 15% glycerol.

iii. BHIB supplemented with 1 g/L taurocholate, 250 mg/L cycloserine, 8 mg/L

cefoxitin, 1 g/L L-cysteine and 5 g/L yeast extract.

iv. Brucella agar supplemented with 5 µg/ml hemin, 1 µg/ml vitamin K1, 5%

v/v laked sheep blood and varying concentration of antibiotics. Brucella agar

plates were prepared according to the Clinical and Laboratory Standards

Institute (CLSI) guideline215.

ChromIDTM C. difficile agar (bioMérieux, Marcy I’Etoile, France) was purchased pre-

prepared from bioMérieux. Brucella broth supplemented with 5 µg/ml hemin (Sigma

Aldrich, Castle Hill, NSW, Australia), 1 µg/ml vitamin K1 (Sigma Aldrich, Castle Hill,

NSW, Australia) and 10% lysed horse blood was prepared according to the CLSI

guideline215.

2.1.2 Buffers and solutions

0.85% NaCl solution……………… Pathwest Media

Chelex-100………………………… Sigma Aldrich, Castle Hill, NSW, Australia

Ultra-pure (UP) water……………… Fisher Biotec, Perth, WA, Australia

A 5% w/v Chelex-100 solution was prepared by suspending 50g of Chelex-100 resin in

1 L of UP water.

2.1.3 Commercial kits

MinElute PCR Purification Kit……. QIAGEN, Venlo, the Netherlands

QuickGene DNA tissue kit S……… Kurabo, Osaka, Japan

Lysing Matrix B…………………… MP Biomedicals, Solon, OH, USA

2.1.4 PCR reagents

AmpliTaq Gold® DNA Polymerase Applied Biosystems, Foster City, CA, USA

Bovine serum albumin (BSA)……... Sigma Aldrich, Castle Hill, NSW, Australia


43

MgCl2……………………………… Applied Biosystems, Foster City, CA, USA

Gene Amp 10× PCR Buffer II……... Applied Biosystems, Foster City, CA, USA

dNTPs……………………………… Fisher Biotec, Wembley, WA, Australia

Primers……………………………... GeneWorks, Thebarton, SA, Australia

2.2 GENERAL BACTERIAL CULTURE METHODS

Unless noted, all anaerobic incubation was performed at 35 °C in an A35 anaerobic

chamber (Don Whitley Scientific Ltd, Shipley, West Yorkshire, UK), containing an

atmosphere of 80% nitrogen, 10% hydrogen and 10% carbon dioxide, with 75% relative

humidity. Solid media, with the exception of ChromID C. difficile agar, was pre-

reduced for ≥ 2 h before use for culture. BHIB supplemented with taurocholate,

cycloserine and cefoxitin was pre-reduced for ≥ 12 h prior to use. Pure cultures were

stored at -70° C in cryovials containing 1 ml BHIB with 15% glycerol (Pathwest

Media).

2.2.1 Enrichment culture

This enrichment procedure was performed on vegetable, meat and environmental

samples using BHIB supplemented with taurocholate, cycloserine and cefoxitin (as

described in Section 2.1.1). Approximately 25 g of meats, ≤ 10 g of vegetable, 5 g of

compost or 5 g of lawn samples were inoculated into a 90 ml pre-reduced supplemented

BHIB and incubated anaerobically at 35 °C for 7 – 10 days. A negative control (25 ml,

10 ml, 5 ml and 5 ml of PBS added to 90 ml of supplemented BHIB for meat, vegetable,

compost and lawn samples, respectively) was included in each round of sampling to

monitor potential contamination. Alcohol shock was performed on enrichment broths by

adding an equal volume of absolute ethanol to an aliquot of broth, 2 ml for meat,

vegetable and compost samples, and 5 ml for lawn samples. After leaving at room

temperature for 1 h, the suspension was centrifuged at 3,800 × g for 10 min and 10 µl

loop-full of sediment was plated on ChromID agar. The plates were further incubated

anaerobically for 48 h and colonies examined as described in Section 2.2.2.

2.2.2 Direct culture

ChromID C. difficile agar was used for the selective culture of C. difficile. A 10 µl loop

of sample was streaked for single colonies on ChromID agar and plates incubated

anaerobically for 48 h. Presumptive C. difficile colonies which appeared as dark grey,


44

black or white, raised umbonate colonies with smooth or filamentous edges44 were sub-

cultured onto BA as described in Section 2.2.3.

2.2.3 C. difficile identification

All presumptive C. difficile colonies on ChromID were sub-cultured onto a pre-reduced

BA and incubated anaerobically for 48 h. C. difficile colonies were identified based on

their BA colony morphology of ground glass appearance, opaque, greyish-white and

non-haemolytic, characteristic chartreuse fluorescence under long-wave UV light (360

nm) and characteristic horse dung odour. The identity of uncertain isolates was

confirmed by the presence of L-proline aminopeptidase activity using Diatabs (Rosco

Diagnostica, Tasstrup, Denmark) and aminopeptidase reagent (Rosco Diagnostica,

Tasstrup, Denmark) as per the manufacturer’s instructions.

2.3 MOLECULAR CHARACTERISATION OF C. DIFFICILE ISOLATES

2.3.1 DNA extraction

A 1 µl loop of a 48 h C. difficile culture on BA was emulsified in 100 µl of 5% w/v

Chelex-100 solution. The mixture was vortexed, heated at 100 °C for 12 min then

centrifuged at 14,000 g for 12 min at 4 °C. The resulting supernatant was transferred to

a sterile RNase- and DNase-free 1.5 ml microcentrifuge tube and stored at -20 °C until

the subsequent PCR steps.

2.3.2 Toxin profiling

The presence of C. difficile toxin genes (tcdA, tcdB, cdtA and cdtB) was determined by

PCR assays as previously described216-218 with minor modifications. Briefly, a multiplex

PCR was used to detect both the non-repeating and repeating region of tcdA. Isolates

were considered tcdA positive when they yielded both a 252 bp product from primers

NK2-NK3 and a 193 bp product from primers tcdA1-tcdA2. The primers used to

amplify fragments of the target genes are listed in Table 2.1.

The reaction mixes consisted of 2 mM MgCl2, 1× buffer II, 0.01% w/v BSA, 0.2 mM

dNTP, 0.75 units of AmpliTaq Gold DNA polymerase and 0.2 µM of each primer.

Except for mixes containing primers tcdA1-tcdA2, where a 4 µl aliquot of the DNA

template (obtained as described in Section 2.3.1) was added to 16 µl of each PCR mix,

a 2 µl aliquot of the DNA template was added into 18 µl of each PCR mix. C. difficile

RT 027 which contains all four target toxin genes was used as a positive control strain

for all toxin gene PCR assays. A negative control consisting of UP water and reaction


45

Table 2.1 Primers used in toxin gene PCR assay and PCR ribotyping

Gene/target Primer Sequence (5’ to 3’) Position Product size (bp) References

tcdA NK2 CCCAATAGAAGATTCAATATTAAGCTT 2479-2505 252 216

NK3 GGAAGAAAAGAACTTCTGGCTCACTCAGGT 2254-2283

tcdA rep tcdA1 CAGTCACTGGATGGAGAATT 5825-5844 193 (Elliott et al.

unpublished) tcdA2 AAGGCAATAGCGGTATCAG 6017-5999

tcdB NK104 GTGTAGCAATGAAAGTCCAAGTTTACGC 2945-2972 203 218

NK105 CACTTAGCTCTTTGATTGCTGCACCT 3123-3148

cdtA cdtA pos TGAACCTGGAAAAGGTGATG 507-526 375 217

cdtA rev AGGATTATTTACTGGACCATTTG 882-860

cdtB cdtB pos CTTAATGCAAGTAAATACTGAG 368-389 510 217

cdtB rev AACGGATCTCTTGCTTCAGTC 878-858

16S-23S rRNA CD 16S CTGGGGTGAAGTCGTAACAAGG 114-1466 variable 50

CD 23S GCGCCCTTTGTAGCTTGACC 20-1


46

mix was also included in each round. The inoculated reaction mixes were cycled on the

following program: 95 °C for 10 min then 35 cycles of 94 °C for 30 s, 55 °C for 30 s,

and 72 °C for 90 s, then a final extension at 72 °C for 7 min. Toxin PCR products were

visualised using capillary gel electrophoresis on the QIAxcel Advanced system

(QIAGEN, Venlo, the Netherlands).

2.3.3 PCR ribotyping

PCR ribotyping was performed as previously described50. The primers used targeted the

3’ end of the 16S rRNA gene and the 5’ end of the 23S rRNA of C. difficile (Table 2.1).

The reaction mixes consisted of 4 mM MgCl2, 1× buffer II, 0.024% w/v BSA, 0.4 mM

dNTP, 3.75 units of AmpliTaq Gold DNA polymerase and 0.4 µM of each primer. A 10

µl aliquot of the DNA template (obtained as described in Section 2.3.1) was added to 40

µl of each PCR mix. A positive control strain (C. difficile RT 027), and a negative

control of UP water and reaction mixes were included in each round. The inoculated

reaction mixes were cycled on the following program: a denaturation step at 95 °C for

10 min, followed by 25 cycles of 94 °C for 1 min, 55 °C for 1 min and 72 °C for 2 min,

and then a final extension step of 72 °C for 7 min.

Following PCR, the products were purified and concentrated using the MinElute® PCR

Purification kit (QIAGEN). The purified PCR products were visualised by capillary gel

electrophoresis on the QIAxcel Advanced system (QIAGEN). The analysis of PCR

ribotyping products was performed using the BioNumerics software package v.6.5

(Applied Maths, Saint-Martens-Latem, Belgium). Dendrograms were generated for all

isolates using Pearson correlation and Dice coefficient to assess the diversity in the

population. PCR RTs were identified by comparison with the banding patterns in our

reference library which consist of 54 internationally recognised RTs including 15

reference strains from the European Centre for Disease Prevention and Control, and

various unique RTs currently circulating in Australia assigned with internal

nomenclature, prefixed with QX (Riley TV, unpublished data).

2.4 TESTING FOR C. DIFFICILE CONTAMINATION OF FOODS AND

THE ENVIRONMENT

Surveys of C. difficile on vegetable, meat, compost and lawn samples were conducted in

Perth metropolitan area and/or the south west of WA.


47

2.4.1 Retail vegetables and meats

2.4.1.1 Settings and sample collection

A survey of retail WA-grown root vegetables was conducted from March to November

2015. A total of 300 root vegetables including 81 organic potatoes, 57 organic carrots,

54 organic beetroots, 54 organic onions and 54 non-organic potatoes were purchased

from 31 retail stores, farmers’ markets and major supermarkets.

A survey of imported vegetables and vegetables of unknown country of origin was

conducted in April 2016. A total of 540 vegetables (90 spring onions, 90 English

spinach, 90 coriander, 90 asparaguses, 90 garlic and 90 pak choy) were purchased from

39 Asian grocery stores and 31 major supermarkets. Garlic and asparagus purchased from

major supermarkets were imported from China and Peru or Mexico, respectively. Spring

onion, English spinach, coriander and pak choy were purchased from Asian grocery stores

and the country of origin was unknown.

Three of each vegetable purchased from the same location were treated as one sample.

To prevent cross-contamination, vegetables were placed into separate bags when

purchased.

A survey of retail meat was conducted from November 2014 to February 2015. A total

of 37 meat samples (32 ready-to-eat salami, four uncooked 18-to-20 week old veal

meats and one uncooked marinated baby goat meat) were purchased from seven retail

stores and major supermarkets. Each meat sample was treated as one sample and they

were individually packaged either by manufacturers or retailers.

2.4.1.2 Sample preparation for culture

To reduce the possibility of laboratory contamination, the initial preparation of the

vegetables and meat was performed before transportation to the laboratory for

enrichment culture. For vegetables, potato skins were peeled using a peeler, the basal

plate and roots of onions and garlic, and the crown and taproot of carrots and beetroots,

the roots of spring onions, English spinach and coriander, the outer stems and leaves of

pak choy, and the stems of asparagus were harvested using a sterile disposable scalpel

on a non-porous tile as a chopping board. Three vegetables of the same kind purchased

from the same stores/markets were treated as a single sample. For meats, 25 g of sample

was cut using a sterile disposable scalpel on a non-porous tile and weighed on a scale in

a sterile petri dish. In between samples, the peeler and non-porous tile were soaked with

bleach solution (6,000 ppm free chlorine; 1 min) to kill spores and a new scalpel and


48

petri dish were used. Samples were then transported to the laboratory and enrichment

cultures were performed ≤ 24 h after collection using supplemented BHIB as described

in Section 2.2.1.

2.4.1.3 Probability of one vegetable being positive for C. difficile

The probability that one out of a sample of three vegetables was positive for C. difficile was

determined using the formula:

3*(1-p)^2*p + 3(1-p)*p^2 + p^3 = r where the sum of three ways for one vegetable, three

ways for two vegetables and one way for all three vegetables to be contaminated with

C. difficile would equal to r, the prevalence of samples that tested positive for C. difficile in

that vegetable type and solving p would represents the probability of one vegetable being

positive for C. difficile.

2.4.2 Composted products

2.4.2.1 Setting, sample collection and sample preparation for culture

From August 2015 to January 2016, a total of 81 composted products were sampled (4

human biosolids, 3 pre-digestate liquid, 3 digestate liquid, 37 soil conditioners, 12

mulches and 22 garden mixes) from one of Australia’s leading garden product

manufacturing companies and 12 smaller suppliers in WA. All samples were collected

in sterile 250 ml containers. Samples were maintained at ambient temperature during

transportation, and enrichment cultures were performed ≤ 24 h after collection using

supplemented BHIB as described in Section 2.2.1.

2.4.3 Lawn

2.4.3.1 Setting, sample collection and sample preparation for culture

From February to June 2016, a total of 311 lawn samples were collected from 20 public

parks across 10 postcodes in the Perth metropolitan area and 1 postcode in the south

west region of WA. Each public park was divided into four quadrants with the centre

generally being a children’s play facility area. Four samples were obtained per quadrant

starting from the centre and moving outwards as shown in Figure 2.1, with the

exception of a few small parks where less than four samples per quadrant were

obtained. The lawn samples weighed approximately 50 g and consisted of grass and its

root system with attached soil. In between samples, a new set of sterile examination

gloves and a new sterile tongue depressor were used to dig out the grass and its root

system. All samples were collected into sterile 250 ml containers and transported at


49

Figure 2.1 Sampling points of lawn in a public park. The park was visually divided

into four quadrants and four samples per quadrants were obtained as indicated ( ).


50

ambient temperature to the laboratory. Enrichment cultures were performed ≤ 24 h after

collection using supplemented BHIB as described in Section 2.2.1.

2.4.3.1.1 Differentiation of “young” lawn from “old” lawn

The turf industry in Australia considers lawns to be “young” up to 4 months after laying

down. This young lawn can be identified by the presence of a visible gap line between

two or more strips of laid lawn which remains visible for up to 4 months219. In the

present study, lawns that did not have visible gap lines were classified as “old”. This

technique for categorising lawn by age has been widely used in the USA220.

2.4.3.1.2 Quantification of C. difficile in lawn samples

In October 2016, a total of 10 lawn samples were collected from parks that previously

tested positive for C. difficile (as previous samples were discarded following the

relocation of our laboratory in May 2016). C. difficile viable counts were performed on

lawn samples using previously described methods with minor modifications221. Briefly,

approximately 5 g of soil were aseptically obtained from lawn sample by shaking the

root system over a stomacher bag (Colworth, London, UK). The soil was suspended in

25 ml of PBS with 0.1% Tween 20 (PBST) or peptone saline with 0.1% Tween 20

(PST) and manually mixed for 60 s. A 100 µl aliquot was spread evenly over ChromID

agar and incubated anaerobically for 48 h. The remaining volume was centrifuged at

3,800 g for 10 min. The pellet was resuspended in 1 ml of either PBST or PST, then 100

µl was inoculated onto ChromID agar and incubated anaerobically for 48 h. To monitor

for potential contamination, a negative control of either 25 ml PBST or PST was added

into a stomacher bag and processed alongside other lawn samples as described above.

The presumptive C. difficile colonies were counted and confirmed as C. difficile

according to Section 2.2.2. This process equated to a detection limit of 2 cfu/g of soil.

2.5 ANTIMICROBIAL SUSCEPTIBILITY TESTING

2.5.1 Agar incorporation method

A collection of 274 C. difficile strains isolated from food (n = 56), compost (n = 36) and

lawn (n = 182) in WA during 2015 and 2016 was tested against fidaxomicin,

vancomycin, metronidazole, rifaximin, clindamycin, erythromycin,

amoxicillin/clavulanate, moxifloxacin, meropenem and tetracycline and minimum

inhibition concentrations (MICs) were determined by the agar incorporation method as

described by the CLSI215, 222. A summary of the RTs, toxin gene profiles and sources of


51

the 274 C. difficile isolates is shown in Figure 2.2. PCR ribotyping patterns for the five

most prevalent RTs in the collection (010, 014/020, 051, 056 and QX 077), which

comprised 55.1% of isolates (n = 151), are shown in Figure 2.3. The toxin profiles

represented included A-B-CDT- (n = 145; 52.9%), A+B+CDT- (n = 117; 42.7%), A-

B+CDT+ (n = 7; 2.6%), A+B+CDT+ (n = 2; 0.7%), A-B-CDT+ (n = 2; 0.7%) and A-

B+CDT- (n = 1; 0.4%).

Briefly, a 10 µl loop of C. difficile from the frozen stock was streaked for single

colonies on a pre-reduced BA and plates were incubated anaerobically for 48 h. To

check for culture purity, phenotypic characterisation of C. difficile on BA was assessed

as described in Section 2.2.3. An inoculum with turbidity equal to a 0.5 McFarland

standard was prepared in 0.85% NaCl solution. The inocula were transferred to the

antimicrobial-containing agar plates using a 52-pin inoculum replicator which deposited

approximately 1-2 µl of each inoculum/plate. The antimicrobial concentration tested

and solvents used are shown in Table 2.2.

Non-toxigenic C. difficile ATCC 700057, Eubacterium lentum ATCC 43055,

Bacteroides fragilis ATCC 25285 and Bacteroides thetaiotaomicron ATCC 29741 were

included in each assay as control strains. The clinical breakpoints for vancomycin and

metronidazole were those recommended by EUCAST223. For fidaxomicin, the European

Medical Agency proposed breakpoint of 1 mg/L was used224. Rifaximin resistance (≥ 32

mg/L) was as described by O’Connor et al.225. CLSI breakpoints were used for

clindamycin, erythromycin, amoxicillin/clavulanate, moxifloxacin, meropenem and

tetracycline222.

2.5.1.1 Statistical analysis

The Kruskal-Wallis rank sum test and Dunn test were performed to compare the

geometric means of MICs between C. difficile of food, compost and lawn origins.

Fisher’s exact test, Chi-square test and post hoc test were used to compare the resistance

rates of C. difficile from different origins.


52

Figure 2.2 Origin of C. difficile RTs tested in antimicrobial study (n = 274). RT others (*) QX 001 (A+B+CDT-), QX 026 (A+B+CDT-), QX 067

(A-B-CDT-), QX 076 (A+B+CDT-), QX 121 (A-B-CDT-), QX 122 (A-B-CDT-), QX 140 (A-B-CDT-), QX 274 (A+B+CDT+), QX 327 (A-B-CDT-),


53

QX 399 (A+B+CDT+), QX 409 (A+B+CDT-), QX 449 (A-B-CDT-), QX 463 (A-B-CDT-), QX 519 (A+B+CDT-), QX 525 (A-B-CDT-), QX 546 (A-

B-CDT-), QX 547 (A-B+CDT+), QX 550 (A-B-CDT-), QX 597 (A-B-CDT-), QX 598 (A-B-CDT-), QX 599 (A-B-CDT-), QX 600 (A-B-CDT+), QX

602 (A-B-CDT-), QX 603 (A-B-CDT-), QX 604 (A-B-CDT-), QX 605 (A-B-CDT-), QX 606 (A-B-CDT-), QX 607 (A-B-CDT-), 005 (A+B+CDT-),

009 (A-B-CDT-), 017 (A-B+CDT-), 018 (A+B+CDT-), 033 (A-B-CDT+), 046 (A+B+CDT-), 064 (A+B+CDT-), 070 (A+B+CDT-), 077 (A-B-CDT-),

080 (A-B+CDT+), 106 (A+B+CDT-), 137 (A+B+CDT-) and 584 (A-B+CDT+).


54

Figure 2.3 Dendrogram of the 5 most prevalent C. difficile RTs, their associated toxin profiles and distribution. PCR ribotyping pattern analysis

was performed by creating neighbour joining tree using Pearson correlation (5% optimization, 1% curve smoothing). R, reference strain (for

comparative purposes only).


55

Table 2.2 Antimicrobial concentration tested and solvents used in the preparation

of antimicrobial stock solution.

Antimicrobial agent Concentration range (mg/L) Solventa

Fidaxomicin 0.002-4 DMSO

Vancomycin 0.015-4 Water

Metronidazole 0.015-4 DMSO and water

Rifaximin 0.0005-64 Methanol

Clindamycin 0.015-32 Water

Erythromycin 0.03-256 Ethanol and water

Amoxicillin/clavulanateb 0.03-16 Buffered water/water

Moxifloxacin 0.12-32 Water

Meropenem 0.25-16 Water

Tetracycline 0.03-64 Water

a DMSO, dimethyl sulfoxide; b, amoxicillin stock solution was prepared in buffered

water (water + Na2HPO4 + KH2PO4) and clavulanate stock solution was prepared in

water


56

2.5.2 Broth microdilution method

The MICs and minimum bactericidal concentrations (MBCs) of C. difficile to common

food preservatives, sodium nitrite (Thermo Fisher, Australia), sodium nitrate (BDH,

Australia) and sodium metabisulphite (BDH, Australia), were determined by the broth

microdilution method as described by the CLSI215. C. difficile strains (Table 2.3) of

food origin were kindly provided by Professor Scott Weese, University of Guelph, ON,

Canada as C. difficile was yet to be isolated from Australian foods at the time of testing.

C. difficile strains of animal origin were from our local collection.

Briefly, a 10 µl loop of C. difficile from the frozen stock was streaked for single

colonies on a pre-reduced BA or supplemented Brucella agar. The plates were incubated

anaerobically for 48 h. A series of two-fold dilutions of each preservative with final

concentrations ranging from 0 to 4,000 µg/ml was made in a 96-well microtitre plate

using pre-reduced supplemented Brucella broth. An inoculum with turbidity equal to a

0.5 McFarland standard (approximately 2×107 cfu/ml) was prepared in 0.85% NaCl

solution, then diluted by one in ten in supplemented Brucella broth, before adding 100

µl of the inoculum to each well to results in a final inoculum concentration of 1×106

cfu/ml. A viable count was performed to confirm the final inoculum concentration and

purity checks were performed on the inoculums using BA incubated aerobically and

anaerobically. Approximately 15% of the C. difficile cells were spores as shown by

Schaffer/Fulton spore stain. Briefly, a drop of the inoculum was spread onto a sterile

slide and dried over a flame. The slide was then flooded with 5% malachite green at

room temperature for 10 min and washed under running water before staining with

0.5% safranin for 30 – 60 s. The slide was rinsed with water and blotted dry before

viewing under a microscope. After 24 h incubation, MICs were determined visually as

the lowest concentration of food preservative resulting in an optically clear microtitre

tray well. MBCs were determined at 24 h by sub-culturing 10 µl from each well onto

ChromID agar. After 24 h incubation, colonies on ChromID agar were counted and

compared to the original inoculum. The MBC was defined as the lowest concentration

of food preservative resulting in ≥ 99.9% death of the inoculum226.

Vancomycin, antibiotic-free, and Staphylococcus aureus ATCC 29213 controls were

included in each assay. The expected MIC (≤ 512 µg/ml) and MBC (≤ 512 µg/ml) of

sodium metabisulphite against S. aureus ATCC 29213 were as described by Frank and


57

Table 2.3 The source and molecular characteristics of C. difficile strains used in

testing the effect of food preservatives.

Strains Source Country Ribotype Toxin profile References

ATCC

700057

- - RT 038 A-B-CDT- -

Food origin

Cd1001 Pork meet Canada E A+B+CDT- 124

Cd1002 Ground beef Canada RT 027 A+B+CDT+ 124

Cd1006 Ground beef Canada RT 078 A+B+CDT+ 124

Cd1009 Chicken meat Canada RT 078 A+B+CDT+ 123

Cd1017 Ground beef Canada C (RT 251) A+B+CDT+ 124

Cd1106 Chicken meat Canada RT 014 A+B+CDT- -

Food animal origin

AI35 Piglet Australia RT 237 A-B+CDT+ 179

AI204 Calf Australia RT 033 A-B-CDT+ 122

AI218 Calf Australia RT 127 A+B+CDT+ 122

AI273 Calf Australia RT 126 A+B+CDT+ 122


58

Patel227. At least three independent replicates were performed and modal MICs and

MBCs were determined.

2.5.3 Checkerboard assay

This checkerboard assay was used to determine the combined effect of sodium nitrite

and sodium nitrate, food preservatives commonly used in combination in ready-to-eat

meats.

Serial two-fold dilutions of 8,000 µg/ml sodium nitrite were made in a vertical

orientation in a 96-well microtitre plates using 100 µl of sterile distilled water. Nine

doubling dilutions of 16,000 µg/ml sodium nitrate, at four times the intended final

concentration, were prepared using sterile distilled water. Fifty microliter aliquots of

each sodium nitrate dilution were added in a horizontal orientation so that the plate

contained various concentration combinations of the two preservatives, both with final

concentration ranging from 0 to 4,000 µg/ml. An inoculum with turbidity equal to a 2.0

McFarland standard (approximately 4×107 cfu/ml) was prepared in 0.85% NaCl

solution, then diluted by one in ten in 4× concentrated supplemented Brucella broth.

Each well was inoculated with 50 µl of the inoculum resulting in a final inoculum

concentration of 1×106 cfu/ml in each well before anaerobic incubation for 24 to 48 h.

Purity checks and a vancomycin control were included in each assay. Three independent

replicates were performed and modal MICs were determined.

The fractional inhibitory concentration (FIC) was calculated by dividing the MIC of the

sodium nitrite and sodium nitrate combination by the MIC of sodium nitrite or sodium

nitrate alone. The FIC index (FICI) was obtained by adding both FICs. When the FICI

was ≤ 0.5, synergy was indicated; FICI of > 0.5 to 4.0 was defined as indifferent

interaction; and FICI of > 4.0 was defined as antagonism interaction228.

2.6 WHOLE-GENOME SEQUENCING

2.6.1 Strain collection

To investigate the transmission of CDI in WA, a total of 29 C. difficile RT 056 strains

[humans-WA (n = 21), food-WA (n = 4), environmental-WA (n = 4)] and 11 C. difficile

RT 012 strains [food-WA (n = 1), human-Australia (n = 4) and human-Asia (n = 6)]

were sequenced. C. difficile RT 056 was selected for WGS because it was one of the

most prevalent RTs in Australian clinical samples200, 207-209 and production animals122 as

well as in our food and environmental surveys (Chapters 3 and 4). Eleven C. difficile


59

RT 012 strains from Australia and Asia was selected for WGS because it was an

emerging RT in WA (Chapter 6) and commonly found in Asia95, 119, 229-231.

Classification of human isolates from WA as CA-CDI or HA-CDI was provided by L

Tracey and TV Riley according to McDonald et al.94

Human isolates from WA were from CDI patients identified as part of the WA

Department of Health Healthcare Infection Surveillance WA (HISWA) program. Food,

compost and lawn isolates were isolated in WA between November 2015 and June 2016

as described in Section 2.4. The environmental isolate from a veterinary clinic was from

our research laboratory collection isolated in 2007.

The human isolates from Asia were from our research laboratory collection and isolated

during a C. difficile prevalence study in the Asia Pacific region that involved 13

countries (Australia, China, Hong Kong, India, Indonesia, Japan, Malaysia, Philippines,

Singapore, South Korea, Taiwan, Thailand and Vietnam) sponsored by Otsuka

Pharmaceutical (Tokyo, Japan). Information on each sequenced strain is available in

Table 2.4.

2.6.2 Genomic DNA extraction

C. difficile frozen stocks were streaked for single colonies on pre-reduced BA and plates

were incubated anaerobically for 48 h. A sterile disposable hockey stick was used to

harvest the growth and then spread plate was performed on another pre-reduced BA for

a further 48 h of anaerobic incubation. Then, genomic DNA was extracted using lysing

matrix B (MP Biomedicals) and QuickGene DNA tissue kit S (Kurabo) as per the

manufacturer’s instructions.

2.6.3 Genomic DNA sequencing

Genomic DNA extracts were sequenced by the Institute for Immunology and Infectious

Diseases at Murdoch University. Multiplexed paired-end (PE) genome libraries were

constructed using standard Nextera XT protocols (Illumina Inc., San Diego, CA, USA)

and sequencing was performed on a MiSeq platform (Illumina) that generated 250 PE

reads. Sequencing yielded a median PE read count of 99% ≥ Q30, resulting in a

theoretical fold coverage of 99× across all isolates. Fastq files were trimmed for quality

and adapter content using Trimmomatic v0.33232.


60

Table 2.4 Strain collection used for WGS.

Lab ID Host Sample type Country Source Hospital/store† CDI exposure Ribotype Sample date

FI007 Food Carrot Australia WA W2 - 056 Mar-2015

FI020 Food Potato Australia WA W13 - 056 Apr-2015

FI040 Food Potato Australia WA W2 - 056 Aug-2015

FI043 Food Potato Australia WA W7 - 056 Oct-2015

CI014 Environmental Compost Australia WA - - 056 Dec-2015

CI015 Environmental Compost Australia WA - - 056 Dec-2015

ENV035 Environmental Lawn Australia WA - - 056 May-2016

EN017 Environmental Veterinary clinic Australia WA - - 056 2007

WA0707 Human Adult-CDI Australia WA FTH HAI-HCFO 056 Dec-2011

WA0732 Human Adult-CDI Australia WA ARMA CAI 056 Nov-2011

WA0959 Human Adult-CDI Australia WA ARMA HAI-HCFO 056 Feb-2012

WA1091 Human Adult-CDI Australia WA SCGH CAI 056 Mar-2012

WA1487 Human Adult-CDI Australia WA SCGH HAI-HCFO 056 Aug-2012

WA1554 Human Adult-CDI Australia WA FTH HAI-HCFO 056 Sep-2012

WA1589 Human Adult-CDI Australia WA FTH HAI-HCFO 056 Sep-2012

WA1647 Human Adult-CDI Australia WA RPH HAI-HCFO 056 Oct-2012

WA1703 Human Adult-CDI Australia WA ARMA CAI 056 Nov-2012


61


WA1932 Human Adult-CDI Australia WA SWAN CAI 056 Jan-2013

WA 1945 Human Adult-CDI Australia WA PMH HAI-HCFO 056 Jan-2013

WA2779 Human Adult-CDI Australia WA SWAN CAI 056 Dec2013

WA2942 Human Adult-CDI Australia WA SWAN CAI 056 Jan-2014

WA3095 Human Adult-CDI Australia WA SCGH HAI-HCFO 056 Apr-2014

WA3219 Human Adult-CDI Australia WA PMH CAI 056 May-2014

WA3223 Human Adult-CDI Australia WA SCGH CAI 056 May-2014

WA3282 Human Adult-CDI Australia WA ROCK CAI 056 May-2014

WA3461 Human Adult-CDI Australia WA ROCK CAI 056 Jul-2014

WA3613 Human Adult-CDI Australia WA RPH HAI-HCFO 056 Sep-2014

PW12757 Human Adult-CDI Australia WA FSH UNK 056 Aug-2015

PW13018 Human Adult-CDI Australia WA FTH UNK 056 Aug-2015

FI032 Food Beetroot Australia WA W13 - 012 Apr-2015

WA2952 Human Adult-CDI Australia WA FTH CAI 012 Feb-2014

WA3410 Human Adult-CDI Australia WA FTH CAI 012 Jul-2014

WA3835 Human Adult-CDI Australia WA SCGH HAI-HCFO 012 Nov-2014

SQ477 Human Adult-CDI Australia SA HSP UNK 012 May-2013


62


OT0066 Human Adult-CDI Hong Kong Hong Kong PWH UNK 012 Jul-2014

OT0088 Human Adult-CDI Taiwan Kaohsiung EDH UNK 012 Jul-2014

OT0104 Human Adult-CDI Singapore Changi CGH UNK 012 Aug-2014

OT0171 Human Adult-CDI Thailand Bangkok RH UNK 012 Sep-2014

OT0184 Human Adult-CDI South Korea Daejeon CNUH UNK 012 Sep-2014

OT0405 Human Adult-CDI China Shanghai HSH UNK 012 Nov-2014

WA, Western Australia; SA, South Australia; CDI, Clostridium difficile infection; CAI, community-associated infection; HAI-HCFO, hospital-

associated healthcare facility onset; UNK, unknown; †, de-identified hospital labs and stores


63

2.6.4 Analysis of WGS data

WGS data was kindly assembled, annotated and analysed by Dr Daniel Knight, School

of Biomedical Sciences, The University of Western Australia, as previously

described214. The methods used are described briefly below

2.6.4.1 In silico multilocus sequence typing and AMR gene profiling

MLST and acquired AMR genes were determined using pubMLST and ARG-ANNOT

(http://en.mediterraneeinfection.com) databases, respectively, and compiled within

SRST2 v0.1.855, 233, 234. A maximum-likelihood tree was generated from MUSCLE

aligned concatenated allele sequences (seven loci, 3501 bp) using PhyML v3.0 with a

Hasegawa Kishino-Yano evolutionary model and 1000 random bootstrap replicates235,

236.

2.6.4.2 De novo assembly and annotation

Trimmed reads were assembled using SPAdes v3.6237 or if contiguity was low, the A5

pipeline238. ABACAS v1.3.1239 was used to order and orientate contigs relative to the

genome reference strain CD630 (GenBank accession AM180355.1, clade 1) for

C. difficile RT 012 and 056. For gap closure and contig extension, GMcloser v1.3240

was used. Finally, ab initio annotation was performed using Prokka v1.11241.

2.6.4.3 SNV analysis

Short read mapping, variant calling, and filtering were performed using methods

developed for transmission analysis of S. aureus242. This pipeline has since been

developed and widely implemented in the microevolutionary studies of C. difficile63, 64,

66, 243.

Trimmed PE reads from each isolate were mapped to the finished reference strains

using Smalt v0.7.6 (http://www.sanger.ac.uk/science/tools/smalt-0). Core genome

SNVs were identified across all mapped sites using a Bayesian statistical framework

implemented by the algorithms mpileup and view within SAMtools v0.1.12-10244. To

remove false positives and to extract only high-quality bonna fide variant sites for

subsequent downstream analyses, VCFtools v0.1.13245, SnpEff v4.2246 and in-house

Unix scripts were performed on the raw base calls.

SNVs were of high quality with Phred-scaled QUAL score ≥ 200, supported by a read

consensus of 75%, a minimum of five reads (including one in each direction) and

homozygous SNVs under a diploid model (GT = 1/1). SNVs occurring in regions of

http://en.mediterraneeinfection.com/

http://www.sanger.ac.uk/science/tools/smalt-0


64

unusual depth (> threshold of 3× median depth for that isolates) were not called. Indels

were removed and only SNVs in the non-repeating region of the reference chromosome

were called. To alleviate the confounding effect of homologous recombination in the

SNV data set, consensus fasta files were produced for each sample with variant sites

positioned on the appropriate reference strain backbone and input into Gubbins

v1.4.5247. Gubbins scans the sequenced alignment, identifying regions of heightened

base substitution density which were then removed resulting in a final set of high

quality concatenated SNVs in “clonal frame”63.

Finally, SNVs were annotated using SnpEff246 and pairwise SNV differences between

all isolates of interest was calculated using a custom python script provided by Dr

David Eyre, University of Oxford. A final alignment of concatenated SNVs was used

for ML inference in RAxML v7.0.4 with a general time reversible model of evolution

and a CAT approximation for substitutional heterogeneity248.

2.7 A RETROSPECTIVE CASE-CONTROL STUDY OF C. DIFFICILE RT

012 IN WA

2.7.1 Ethics approval

This study was approved by the UWA Human Research Ethics Committee (approval

reference number RA/4/1/8222) on 27 July 2016.

2.7.2 Study setting and design

A retrospective case-control study on emerging C. difficile RT 012 infection was

conducted in WA. Cases were defined as hospital patients who previously had

C. difficile RT 012 cultured from their stool samples and confirmed by molecular typing

(as described in Section 2.3) as part of the HISWA program between January 2014 and

March 2015. HISWA is a state-wide program that collects data related to healthcare-

associated infections, including CDI, across WA. Infection rates are published quarterly

as aggregates and patient identifiable data are not released249. Each case was matched

with two controls for age (± 2 y) and gender. Controls were hospital patients identified

through the HISWA program in the same period of time, who previously had C. difficile

infection by an RT other than RT 012.


65

2.7.3 Data collection and analysis

A study information sheet, consent form, and structure questionnaire was posted to 155

potential participants up to three times between September and December 2016. The

documentation was posted a second and third time approximately 1 month apart if the

potential participants had not replied to an earlier posting. If a potential participant did

not reply to any of the postings within a month of the third posting, it was considered a

refusal to participate in the study. A unique de-identified number was assigned to each

potential case and control. A copy of the questionnaire used for data collection is shown

in Appendix I.

Data were collated in Microsoft Office Excel and statistical analysis was performed

using RStudio v1.0.153. Characteristics of cases and controls were compared using the

Fisher’s exact test. Univariate odds ratios (ORs) were calculated and used to identify

risk factors for C. difficile RT 012. Assistance with data analysis was kindly provided

Dr Alethea Rea, Centre for Applied Statistics, The University of Western Australia.

Chapter 3: C. difficile in food

66

CHAPTER 3

C. DIFFICILE IN FOOD

Part of the work presented in this chapter has been published:

Lim SC, Foster NF, Elliot B, Riley TV. High prevalence of Clostridium difficile on

retail root vegetables, Western Australia. Journal of Applied Microbiology. 2018; 124:

585-90.


67

3.1 INTRODUCTION

Previously regarded as a nosocomial or hospital-acquired infection, CDI is increasingly

being diagnosed globally as CA in individuals with no traditional risk factors20. Like in

the rest of the world, the incidence of CA-CDI in Australia has risen and is currently

accounting for approximately 30% of all CDI cases3. Reports of genetically highly

related C. difficile in humans, animals and retail foods have raised concerns that CA-

CDI might have a zoonotic or foodborne aetiology66, 117, 167, 195, 214. This is a possibility

as other spore-forming Clostridia such as C. perfringens are well known causes of

foodborne illness250.

In 2011, genetically highly related C. difficile (≤ 4 SNVs differences in their core

genomes) strains, recovered from predominantly CA-CDI cases with no evidence of

geographical clustering, emerged across four Australian states. This led to speculation

that there might have been foodborne spread involving the national food chain206.

Unlike the northern hemisphere where C. difficile has been isolated from retail meats124,

127-130, 138, previous studies on Australian ground meats found no C. difficile

contamination (NF Foster and TV Riley, unpublished data). Prior to this PhD project,

no other studies have investigated C. difficile in Australian food.

This chapter describes three food sampling surveys examining the prevalence and

molecular epidemiology of C. difficile in and on retail vegetables and meats available in

WA. It addresses the first aim of the thesis: to determine the prevalence and molecular

epidemiology of C. difficile on retail foods available in WA. Sample collection and

preparation procedures were as described in Section 2.4, and enrichment culture and

molecular typing were performed as described in Section 2.2 and 2.3, respectively.

3.2 RESULTS

3.2.1 Prevalence and molecular epidemiology of C. difficile on WA-grown

root vegetables

A total of 300 WA-grown root vegetables; organic potatoes (n = 81), organic carrots (n

= 57), organic beetroots (n = 54), organic onions (n = 54) and non-organic potatoes (n =

45), were purchased from 31 organic stores, weekend farmers’ markets and major

supermarkets in WA from March to November 2015. Three of each vegetable

purchased from the same location were treated as a single sample. Overall, C. difficile

was isolated from 30.0% (30/100) of the root vegetable samples using enrichment


68

culture (Table 3.1). The highest prevalence was observed in organically-grown potatoes

(55.6%, 15/27), followed by non-organic potatoes (50.0%, 9/18), organic beetroots

(22.2%, 4/18), organic onions (5.6%, 1/18) and organic carrots (5.3%, 1/19). The

overall probability of one root vegetable being positive for C. difficile was 11.2%;

23.7% for organic potatoes, 20.6% for non-organic potatoes, 8.0% for organic beetroots,

1.9% for organic onions and 1.8% for organic carrots (calculated as described in

Section 2.4.1.3). There was no significant difference between prevalence in potatoes

grown organically and potatoes grown non-organically, however, a greater proportion of

potato samples was positive for C. difficile compared to other root vegetables (Fisher’s

exact, p < 0.0001).

Approximately half of the isolated strains were toxigenic (51.2%, 22/43). Of the

toxigenic isolates recovered, 81.8% (18/22) were A+B+CDT-, 9.1% (2/22) were A-

B+CDT+, 4.5% (1/22) was A+B+CDT+ and 4.5% (1/22) was A-B-CDT+.

The RT banding patterns of isolated C. difficile are shown in Figure 3.1. Of the 43

isolates, 23 (53.5%) were internationally recognised RTs 051 (n = 6), 056 (4), 014/020

(2), 101 (2), 002 (1), 010 (1), 012 (1), 033 (1), 064 (1), 070 (1), 137 (1), 237 (1) and 584

(1). The remaining 46.5% of isolates (20/43) did not match any international reference

strains in our laboratory; 70.0% (14/20) were RTs currently circulating in Australia [QX

145 (n = 5), QX 393 (3), QX 049 (2), QX 142 (2), QX 072 (1) and QX 274 (1)] and

30.0% (6/20) were novel to our laboratory [QX 545 (n = 2), QX 518 (1), QX 519 (1),

QX 525 (1) and QX 551 (1)]. C. difficile RTs 027 and 078 were not found. Only 6.0%

(6/100) of the root vegetable samples were contaminated with more than one strain of

C. difficile; 11.1% (3/27) of organic potatoes, 11.1% (2/18) of non-organic potatoes and

5.6% (1/18) of organic beetroot.

Of the 31 stores, 19 (61.3%) were selling root vegetables that were contaminated with

C. difficile (Table 3.2). Of those 19, five (26.3%) were selling more than one type of

vegetable that was contaminated with C. difficile. Ten stores (32.3%) had at least two

different C. difficile RTs isolated from their vegetables, with a median of three. Store

W2 was the only store from which the same RT was isolated twice, RT 056 on organic

potatoes and organic carrots. Diversity in C. difficile strains was also observed between

stores with 62.5% (15/24) of all isolated RTs not shared with any other stores.


69

Table 3.1 Prevalence and toxin profiles of C. difficile cultured from WA-grown root vegetables.

No. of isolates with toxin profiles below, n (%)

Origin Prevalence, n (%) A-B-CDT- A+B+CDT- A-B+CDT+ A+B+CDT+ A-B-CDT+ Total

Organic

Potatoes 15/27 (55.6) 11 (52.4) 9 (50.0) 0 (0.0) 1 (100.0) 1 (100.0) 22 (51.2)

Beetroots 4/18 (22.2) 4 (19.0) 1 (5.6) 1 (50.0) 0 (0.0) 0 (0.0) 6 (14.0)

Onions 1/18 (5.6) 1 (4.8) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 1 (2.3)

Carrots 1/19 (5.3) 0 (0.0) 1 (5.6) 0 (0.0) 0 (0.0) 0 (0.0) 1 (2.3)

Non-organic

Potatoes 9/18 (50.0) 5 (23.8) 7 (38.9) 1 (50.0) 0 (0.0) 0 (0.0) 13 (30.2)

Total 30/100 (30.0) 21 (48.8) 18 (41.9) 2 (4.7) 1 (2.3) 1 (2.3) 43 (100.0)


70


on WA-grown organic (α) and non-organic (β) root vegetables. RTs 027 and 078 were

included for reference (R). PCR ribotyping pattern analysis was performed by creating a

neighbour-joining tree, using Dice coefficient (optimisation, 5%; curve smoothing, 1%).

QX, internal nomenclature.


71

Table 3.2 Summary of C. difficile PCR RTs isolated from WA-grown root

vegetables, by store.

Store Origin No. of isolates, n (%) PCR ribotype

W1 Potatoesα 3 (7.0) 051* and 101

Beetrootsα 3 (7.0) QX 393†, QX 551† and 237†

W2 Potatoesα 5 (11.6)

QX 142†, QX 274†, QX 518,

014/020 and 056

Carrotsα 1 (2.3) 056

W5 Onionsα 1 (2.3) QX 072

W6 Potatoesα 1 (2.3) QX 145

W7 Potatoesα 5 (11.6) QX 145, 002, 051†, 056† and 064



Beetrootsα 1 (2.3) 010

W11 Potatoesα 2 (4.7) QX 519 and 014/020


W13 Potatoesα 2 (4.7) 033† and 056†

Beetrootsα 1 (2.3) 012

W15 Beetrootsα 1 (2.3) 051

W20 Potatoesβ 1 (2.3) 051


Potatoesβ 1 (2.3) QX 049

W22 Potatoesβ 2 (4.7) QX 145† and 137†


W26 Potatoesβ 3 (7.0) QX 049†, QX 393† and QX 545†

W28 Potatoesβ 1 (2.3) QX 545

W29 Potatoesβ 3 (7.0) QX 142, 584 and 070


α, organic; β, non-organic; *, two RT 051 strains were isolated on two varieties of

organic potatoes; †, RT isolated from the same sample.


72

3.2.2. Prevalence and molecular epidemiology of C. difficile on imported

vegetables and vegetables of unknown country of origin

A total of 540 vegetables, including spring onions (n = 90), English spinach (n = 90),

coriander (n = 90), asparagus (n = 90), garlic (n = 90) and pak choy (n = 90), were

purchased from 39 Asian grocery stores and 31 major supermarkets in WA in April

2016. Spring onions, English spinach, coriander and pak choy were purchased from

Asian grocery stores and their country of origin was unknown. The garlic and asparagus

were purchased from major supermarkets and the countries of origin were China, and

Peru or Mexico, respectively. Three of each vegetable purchased from the same location

were treated as a single sample. Overall, C. difficile was isolated from 6.1% (11/180) of

the vegetable samples using enrichment culture (Table 3.3). The highest prevalence

was observed in spring onions (20.0%, 6/30), followed by English spinach (13.3%,

4/30) and coriander (3.3%, 1/30), while no C. difficile was isolated from asparagus

(0/30), garlic (0/30) or pak choy (0/30). The overall probability of one imported

vegetable or vegetable of unknown country of origin being positive for C. difficile was

2.1%; 7.2% for spring onions, 4.6% for English spinach, 1.1% for coriander and 0.0%

for asparagus, garlic and pak choy (calculated as described in Section 2.4.1.3).

Most (81.8%, 9/11) of the isolated strains were non-toxigenic, 9.1% (1/11) were

A+B+CDT- and 9.1% (1/11) were A-B-CDT+. The 11 isolates belonged to 10 different

C. difficile RTs (Figure 3.2), with two (18.2%) isolates identified as RT 286. The

remaining 81.8% (9/11) isolates did not match any international reference strains,

44.4% (4/9) of which were RTs currently circulating in Australia [QX 054 (n = 1), QX

076 (1), QX 122 (1) and QX 393 (1)] and 55.6% (5/9) were novel to our laboratory [QX

551 (1), QX 598 (1), QX 599 (1), QX 600 (1) and QX 604 (1)].

Of the 70 stores, 10 (14.3%) were selling vegetables that were contaminated with

C. difficile (Table 3.4). C. difficile-positive samples were only from the vegetables

sampled from Asian grocery stores. Different C. difficile RTs were isolated from each

of the 10 stores, except for 286 which was isolated from an English spinach sample

from store W32 and a coriander sample from store W35. Store W37 was the only store

with more than one C. difficile-positive vegetable. The spring onion and English

spinach samples from store W37 were contaminated with C. difficile RTs QX 393 and

QX 054, respectively.


73

Table 3.3 Prevalence and toxin profiles of C. difficile from imported vegetables and vegetables of unknown country of origin.

No. of isolates with toxin profiles below, n (%)

Origin Prevalence, n (%) A-B-CDT- A+B+CDT- A-B-CDT+ Total

Spring onions 6/30 (20.0) 5 (55.6) 1 (100.0) 0 (0.0) 6 (54.5)

English spinach 4/30 (13.3) 3 (33.3) 0 (0.0) 1 (100.0) 4 (36.4)

Coriander 1/30 (3.3) 1 (11.1) 0 (0.0) 0 (0.0) 1 (9.1)

Asparagus 0/30 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Garlic 0/30 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Pak choy 0/30 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Total 11/180 (6.1) 9 (81.8) 1 (9.1) 1 (9.1) 11 (100.0)


74

Figure 3.2 C. difficile PCR ribotyping patterns for strains isolated from imported vegetables and vegetables of unknown country of origin.

Corresponding toxin gene profile, prevalence and strain origin are shown. For comparison purposes, epidemic reference RTs 027 and 078 are also

added (R). PCR ribotyping pattern analysis was performed by creating a neighbour-joining tree, using Dice coefficient (optimisation, 5%; curve

smoothing, 1%). QX, internal nomenclature.


75

Table 3.4 Summary of C. difficile PCR RTs isolated from imported vegetables and

vegetables of unknown country of origin, by store.

Store Origin No. of isolates, n (%) PCR ribotype

W32 English spinach 1 (9.1) 286

W33 English spinach 1 (9.1) QX 551

W34 Spring onion 1 (9.1) QX 604

W35 Coriander 1 (9.1) 286



English spinach 1 (9.1) QX 054

W38 English spinach 1 (9.1) QX 600





76

3.2.3. Prevalence of C. difficile in meat products available in WA

None of 32 ready-to-eat (RTE) salami samples, four uncooked 18-to-20-week-old veal

calf meat samples and one uncooked marinated baby goat meat sample tested positive

for C. difficile.

3.3 DISCUSSION

The emergence of CA-CDI and growing evidence supporting CDI as a zoonotic and/or

foodborne infection has ignited great interest in identifying reservoirs of C. difficile in

the community. Given the lack of data on C. difficile in Australian foods, the studies

reported in this chapter aimed to describe the prevalence and molecular epidemiology of

C. difficile isolated from the retail vegetables and meats available in WA.

3.3.1 C. difficile on the retail vegetables available in WA

The combined prevalence of C. difficile on vegetables in these studies (14.6%, 41/280)

was high compared to other reports. To date, only limited studies have investigated the

contamination of C. difficile on vegetables. A meta-analysis conducted in 2014 reported

a mean C. difficile prevalence of 2.1% (range 0.0 – 7.5%) on vegetables169. Only peer-

reviewed publications reporting the prevalence of C. difficile on vegetables in the past

20 years were included in the meta-analysis and six studies spanning Africa, Europe and

North America were identified. None of 100 vegetables (ugwu leaves, ewedu, efo tete,

onion and tomato) in Nigeria tested positive for C. difficile145, although this is likely to

be due to their direct culturing method of simply placing unwashed vegetables on agar

culture media and then removing the vegetables prior to incubation of agar plates in

anaerobic conditions. Enrichment culture, which offers a greater chance of detecting

C. difficile on vegetables, was not performed and, with the exception of onions, all other

tested vegetables were cultivated above ground. Intuitively, vegetables grown above

ground are more likely to have a lower prevalence of C. difficile on their surfaces

compared to root-vegetables. In France, only three vegetable samples (2.9%) tested

positive for C. difficile, two RTE salads and one pea sprout sample137. In Scotland, 7.5%

(3/40) of RTE salads were positive for C. difficile, all of which were packaged and

marked as imported from European Union countries194. Direct culture using contact

plates was performed in Wales on 300 unwashed vegetables with seven (2.4%) [potato

(n = 2), onion (1), carrot (1), radish (1), mushroom (1) and cucumber (1)] contaminated

with C. difficile143. Of the 111 root vegetables and leafy greens tested in Canada, 4.5%


77

(n = 5) were positive for C. difficile [carrot (n = 1), ginger (3) and eddoes (1)]132.

Interestingly, and perhaps unexpectedly, none of the 39 root vegetables (onions, carrots,

potatoes, beetroots and parsnip) tested in Ohio were positive for C. difficile despite

purposely sampling vegetables with soil residue on their surfaces. However, C. difficile

was isolated from the outer leaves of iceberg lettuce, green peppers and eggplant, giving

a 2.4% (3/125) prevalence of C. difficile on vegetables available in Ohio169. After the

publication of the meta-analysis in 2014, which included all studies mentioned above,

only one other study investigated the prevalence of C. difficile on vegetables, in which

5.7% (6/106) of RTE salads were positive in Iran193.

The higher C. difficile prevalence in the WA vegetable surveys compared to other

surveys elsewhere may be due to several factors. The types of vegetables tested could

have influenced the prevalence. With potatoes, for example, sampling a greater surface

area covered in soil may have led to better detection of C. difficile contamination. In this

study, the prevalence of C. difficile on root vegetables was significantly higher than on

imported vegetables and vegetables of unknown country of origin which mostly

comprised of non-root vegetables (Chi-squared, p < 0.0001). The culture method is also

likely to have been more sensitive as three vegetables of the same kind were pooled and

treated as one sample, so one C. difficile positive vegetable would result in a positive

sample. However, the probability that only one of three vegetables was positive for

C. difficile was still higher than most overseas studies at overall 5.1%, 11.2% on WA-

grown root vegetables and 2.1% on imported vegetables and vegetables of unknown

country of origin. In addition, the culture method also used a bigger volume of

enrichment broth compared to other studies, as well as a longer incubation time.

Currently, there are no international standards for isolating C. difficile from food,

although the method used was similar to the USA CDC-recommended method for meat

with the exception of a longer incubation time140. However, it is also possible that the

higher prevalence of C. difficile on vegetables in WA reflects seasonality138, and

different vegetable growing and processing practices which may have resulted in

contamination of produce by animal manure. The extent of animal manure usage as

fertiliser in farming is not known in other countries; but in Australia, approximately 1.7

million tonnes per year of animal manure is applied to agricultural land as fertiliser,

6.8% (115, 989 tonnes) of which is used for farming in WA251.

Laboratory contamination is unlikely to be responsible for the high prevalence of

C. difficile found as all samples were prepared using aseptic techniques prior to


78

transportation to the laboratory for enrichment culture, and a diverse range of C. difficile

PCR RTs was isolated including RTs novel to our laboratory, QX 518 (A-B-CDT-), QX

519 (A+B+CDT-), QX 525 (A-B-CDT-), QX 545 (A+B+CDT-), QX 551 (A-B-CDT-),

QX 598 (A-B-CDT-), QX 599 (A-B-CDT-), QX 600 (A-B-CDT+) and QX 604 (A-B-

CDT-). Such diversity is highly unlikely to represent laboratory contamination.

Thus, in addition to the higher prevalence of C. difficile on vegetables available in WA,

these studies showed a much greater strain diversity than earlier studies outside

Australia, with 32 different RTs among 54 isolated strains from 41 C. difficile positive

samples. In France, only three RTs were isolated from three C. difficile positive

samples, RTs 001, 014/020/077 and 015137. In Canada, two RTs among five strains

were isolated from five positive samples, three were RT 078 and two were of an

unidentified RT previously isolated in human CDI132. In the USA, two RTs were

isolated from three positive samples, two isolates were RT 027 and one was an

A+B+CDT- strain that did not match any of their reference collection169. Only three

strains were isolated from RTE salads in Scotland, two were RT 017 and one was RT

001194, both RTs were common in European clinical isolates252, 253. The six isolates

from the Iranian study were not typed, but five (83.3%) of them were non-toxigenic and

one was A+B+CDT-193. While C. difficile RTs 027 and 078 have dominated vegetable

prevalence surveys in North America, neither has been found in Australian food studies.

This is not surprising as RT 027 has rarely been isolated in Australia85 and RT 078 has

never been isolated from Australian production animal although the closely related RT

126 has121, 122, 154.

Of the RTs represented in both the vegetable surveys reported in this chapter,

C. difficile RT 014/020 was also the most common strain in Australian piglets

(23.4%)154 and in human clinical samples, being the cause of CDI in 30.0% (99/330) of

the cases209 and 22.1% (23/104) of asymptomatic colonisation200. Although only two

RT 014/020 isolates were recovered in these surveys, the finding of such a clinically

important RT still indicates a possible risk of transmission through vegetables.

C. difficile RT 056, the third most prevalent RT (7.4%; 4/54) found, was also the fourth

most prevalent strain in humans (3.9%; 13/330)209 and the third most prevalent strain in

Australian veal calves (7.7%; 16/209)122. Other isolated C. difficile RTs with an

epidemiological link to Australian livestock included RTs 101, 137, 033 and 237. RTs

101 (40%; 6/15) and 137 (13.3%; 2/15) were the two most prevalent RTs in lambs121,

RT 033 was the second most prevalent RT in both veal calves (19.6%; 41/209)122 and


79

piglets (13%; 20/154)154, while RT 237 is a pig strain unique to one piggery in WA154.

All of these C. difficile RTs have been isolated from human CDI cases in Australia203,

254, 255.

These studies are the first to provide data on C. difficile contamination of retail

vegetables in Australia. Vegetable samples were purchased from a total of 101 stores in

the Perth metropolitan area while previous studies outside Australia generally only

sampled from between 4 and 11 stores. Another strength of the WA studies is that a

large number of each of 11 vegetable types was tested, while previous studies either

sampled large numbers of a single vegetable type (RTE salads)137, 193, 194 or ≤ 10 of a

certain vegetable type132, 169.

The surveys reported here have several limitations. The degree of C. difficile

contamination on the vegetables was not determined. However, the concentration of

C. difficile spores on contaminated foods is generally presumed to be low, with studies

that report positive samples detecting C. difficile by enrichment culture only123 or, if

positive by direct culture, counts of 20 – 240 spores/g124. Another limitation was that all

of the vegetables tested were purchased in WA. For the smaller vendors vegetables

would almost certainly have been grown in WA, however, larger retail chains transport

vegetables throughout Australia and it is quite possible that vegetables from Eastern

Australia were included in the surveys. For a more thorough understanding of the

prevalence and molecular types of C. difficile on retail vegetables in Australia, larger

studies involving samples from multiple States and Territories are necessary.

Furthermore, it is unclear if the differences in C. difficile prevalence between our two

vegetable surveys are due to differences in country of origin, vegetable types (root

vegetables vs. mostly non-root vegetables) or time of year sampled. These issues should

be addressed in any future vegetable surveillance study.

These studies are the first to show such great diversity in C. difficile strains from

vegetables with many of the RTs also common in production animals and human

clinical cases. These C. difficile strains are likely of animal origin due to the common

practice of using animal manure as fertiliser in growing vegetables. The high prevalence

of C. difficile in retail vegetables available in WA, especially root vegetables such as

locally-grown potatoes, heightens concerns about possible foodborne transmission of

C. difficile and contamination of kitchen while cleaning or processing the vegetables. In

January 2016, a small pilot study was performed by our research group to determine the


80

level of C. difficile contamination in the kitchen of 35 households in WA (TV Riley,

unpublished data). The kitchen was swabbed with a sponge and enrichment culture was

performed. In this pilot study, 8.6% (3/35) of the samples were positive for C. difficile.

Two of the isolates were RT 014 and the other was RT 002, both were consistently

among the top three most common RTs in causing human CDI in Australia (Figure

1.9)203, 204, 206-210.

3.3.2 C. difficile in retail meats available in WA

Unlike reports from the northern hemisphere, a study with over 300 Australian ground

meat samples found no C. difficile contamination in WA (NF Foster and TV Riley,

unpublished data). However, most retail ground meats in the Australian market are from

adult food animals which are known to have a low gastrointestinal prevalence of

C. difficile at the time of slaughter122. An exception to this generalisation is that

Australian veal calves under the age of 7-days may be slaughtered to make RTE meats

such as salami (Meat and Livestock Australia, personal communication). These neonatal

animals are more likely to carry C. difficile than older animals. In 2013, a cross-

sectional study in Australia reported 56.4% (203/360), 3.8% (1/26) and 1.8% (5/280)

C. difficile prevalence in the faeces of < 7-days old calves (predominantly male dairy

calves), 2-to-6-months old calves and older calves, respectively122. In 2016, a

longitudinal study in a WA pig farm investigated C. difficile gastrointestinal carriage in

piglets over time and found the prevalence to be 40.0% (8/20), 50.0% (10/20), 20.0%

(4/20), 0.0% (0/20) and 0.0% (0/20) on days 1, 7, 13, 20 and 42 post-farrowing,

respectively256. Furthermore, 25.3% (76/300) of Australian veal calf carcasses from

animals aged < 14 days were contaminated with C. difficile221, with a median count of 7

cfu/cm2.

RTE meats made from veal calf carcasses are commonly cured with sodium nitrite and

sodium nitrate, and preserved either by fermentation, air-drying or smoking257, all of

which are incapable of killing spores. The concentration of sodium nitrite and sodium

nitrate able to kill or inhibit growth of C. difficile was investigated in Chapter 5258.

C. difficile spores can survive recommended cooking temperatures for meat (71 ºC) for

over 2 h259 and resist desiccation at ambient temperature for over 5 months260, so air-

drying at room temperature or smoking meats until an internal temperature of 65 ºC

(150 ºF) for 10 min (as required by the Meat and Livestock Australia)257 would not kill

the spores. This prompted the need to investigate the prevalence of C. difficile in RTE


81

meats available in Australia. Five uncooked intact types of meat from young animals

and 32 salami were sampled, however, C. difficile was not found in any of the 37 meat

samples tested.

Limited studies have investigated the presence of C. difficile in RTE meats as most

studies focused on the presence of C. difficile in uncooked ground meats and uncooked

intact meats124, 125, 127-130, 134, 138. C. difficile prevalence in meats appears to vary greatly

between studies. In North America, the reported prevalence of C. difficile in meats was

as high as 42.0%124, 134, 138, while studies in Europe indicated a much lower prevalence

of < 3%127, 167. In 2009, a study in the USA reported C. difficile in 47.8% (11/23) of

RTE meats, including beef summer sausage (14.3%, 1/7) and pork Braunschweiger

(62.5%, 10/16), and found RTs 027 (36.4%, 4/11) and 078 (63.6%, 7/11)134. In Taiwan,

23.1% (15/65) of RTE meats, including braised pork skin (25.7%, 9/35) and braised

pork colon (20.0%, 6/30), were positive for C. difficile 164. These studies are not directly

comparable to the present study due to differences in the types of meats sampled,

however, the methods used in all studies were similar whereby samples were enriched

(7 – 10 days) in broth containing a spore germinant.

Besides the choice of types of meat sampled, the absence of C. difficile in WA meat

samples could be due to the small sample size and high meat processing and importing

standards in Australia. Sampling the environment of slaughterhouses where veal calves

are processed and sampling the supply chain of RTE meats from the manufacturing

plants to the retail stage would be a more direct way of assessing the prevalence of

C. difficile in such facilities and meats. This was not performed due to time constraints

and the current sensitivity in Australia around slaughtering very young veal calves.

Their separation from their mother at < 7-days old for slaughter as a “waste product” of

dairy farms creates on-going tension between the Australian dairy industry and animal

protection organisations261.

3.4 CHAPTER SUMMARY

A high prevalence of C. difficile was found on retail vegetables available in WA,

especially locally-grown root vegetables (30.0%), while no C. difficile was isolated

from any sampled meat product. This finding is in contrast with reports from North

America where a high prevalence of C. difficile ranging up to 42.0% was found in

meats134, 138, while the prevalence of C. difficile in vegetables was generally low,

< 5.0%132, 169. In WA, with the isolation of clinically important C. difficile strains, it


82

seems that contaminated root vegetables are likely to be a potential source of CA-CDI.

The cleaning and processing C. difficile-contaminated vegetables in a kitchen is

suspected to be contributing to the contamination of C. difficile in household kitchen.

Foodborne and environmental transmission of CDI are yet to be proven, but ingestion of

toxigenic C. difficile spores on vegetables and/or household environment by at risk

populations in the community (such as those taking antibiotics) may lead to CDI. The

C. difficile strains on vegetables available in WA are likely to be of animal origins due

to the common practice of using animal manure as fertiliser for growing vegetables.

Despite ongoing effort by our research group, no C. difficile was isolated from retail

meats.

Future research may include (i) larger studies involving samples from multiple States

and Territories, (ii) identifying any potential seasonality by using a more structured

sampling regime, (iii) determining the source of contamination such as contaminated

animal manure, soils, or irrigation water, as well as inadequate food handling

procedures by workers, downstream contamination of transport vehicles, kitchens and

supermarkets, and (iv) a more discriminatory typing methods, such as WGS, are also

necessary to confirm the association of animal and human CDI with contaminated foods

and kitchens.

Chapter 4: C. difficile in the environment

83

CHAPTER 4

C. DIFFICILE IN THE

ENVIRONMENT

Part of the work presented in this chapter has been published:

Moono P, Lim SC, Riley TV. High prevalence of Clostridium difficile in public space

lawns in Western Australia. Scientific Reports 2017; 7: 41196.


84

4.1 INTRODUCTION

The recent isolation of C. difficile on retail vegetables available in WA, including RTs

common to animals and humans in the region (Chapter 3), raises concern that

vegetables might be a source for CA-CDI. One possibility was that composted animal

manure being used as fertiliser in farming was contaminating crops with C. difficile of

animal origin. As it was not possible to collect soil samples from market gardens

growing vegetables, compost products destined to be used in farming, landscaping and

gardening were collected from one of Australia’s leading garden products

manufacturers and 12 smaller suppliers in the Perth metropolitan area, WA.

In Australia, human biosolids from wastewater treatment plants and manure from

production animal farms are composted to be used as gardening products, as described

in Section 4.1.1. Recently, several studies have demonstrated the survival of C. difficile

through wastewater treatment plants197, 262, 263 and the composting process264. However,

only one study has investigated the prevalence of C. difficile in compost products with

35.7% (5/14) of composted pig manure from piggeries testing positive for C. difficile265.

A limitation of that study was that the temperature and other composting conditions

were not monitored so it was not known whether the composting process used complied

with the standards for commercial products. This chapter represents the first to

investigate the prevalence of C. difficile in commercially composted products.

In addition, it became apparent that pig manure from WA’s piggeries was being used

directly as fertiliser in the production of roll-out lawns (turf) for both domestic and

public space areas. The production of roll-out lawns for use in the community has

become extremely popular in Western society over the last few decades as a result of

new housing and suburb development266. This lawn could be contaminated with

C. difficile of pig origin and acting as a reservoir for CA-CDI.

This chapter describes two studies that examine the prevalence and molecular

epidemiology of C. difficile in the environment, in (i) commercially composted products

and (ii) the public lawn of WA (this work was carried out in collaboration with Dr Peter

Moono). It addresses the second aim of the thesis: to determine the prevalence and

molecular epidemiology of C. difficile in the environment in WA. The sample collection

and preparation procedure were as described in Section 2.4, and enrichment culture and

molecular typing were performed as described in Sections 2.2 and 2.3, respectively.


85

4.1.1 Compost manufacturing process in Australia

Compost is defined as an organic product that has undergone controlled aerobic and

thermophilic biological transformation through the composting process to achieve

pasteurization. The raw material can include vegetation and crop residues, food and

garden wastes, manures and biosolids. The Australian Standards AS 3743267 and

4454268 state that appropriate turning of the outer material to the inside of the compost

pile is required so the whole mass is subjected to a minimum of three turns with the

internal temperature reaching a minimum of 55 °C for 3 consecutive days before each

turn, 1 – 2 turns per week. When higher risk materials are included in the compost

feedstock (including manures, animal waste, human biosolids, food or grease trap

waste), a minimum of five turns with the core temperature of the compost pile

maintained at 55 °C or higher for a minimum of 15 days are required268, 269. In addition

to sufficient turning/aeration and heating, moisture content is also crucial to the

breakdown of the raw material and should be maintained between 45% – 65% by

adding water (if it is too dry) or dry ingredients (if it is too wet)267, 268. Upon completion

of the pasteurization process, the compost piles are cured for a minimum of 2 months to

ensure maturity267, 268. This process is known as windrow composting. It is occurring at

the sites of all compost manufacturer in Australia. However, at the manufacturing site of

one of Australia’s leading gardening companies, approximately 55, 000 tonnes/yr of

food waste from supermarkets, abattoirs, agricultural companies and food

manufacturing facilities is recycled via anaerobic digestion in tanks to produce methane

gas for clean energy and liquid digestate as biofertiliser. Instead of water, this digestate

is added to compost piles in between each turn to maintain a moisture content of 45% –

65%.

4.2 RESULTS

4.2.1 Prevalence and molecular epidemiology of C. difficile in compost

products available in WA

A total of 81 samples were collected of which, 29 were from four points in the

composting process (Figure 4.1) at one of Australia’s leading garden products

manufacturer (4 human biosolids, 3 pre-digestate liquid, 3 digestate liquid, 13 soil

conditioners, 4 mulches and 2 garden mixes). The remaining samples (24 soil

conditioners, 8 mulches and 20 garden mixes) were final products sampled from the

suppliers in the Perth metropolitan area, WA. Overall, 27.2% (22/81) of the samples


86

Figure 4.1 Schematic diagram of the composting process and sampling points for compost products ( ).


87

were contaminated with C. difficile. The highest C. difficile prevalence was observed in

the digestate liquid (100.0%, 3/3), followed by human biosolids (75.0%, 3/4), soil

conditioners (29.7%; 11/37), mulches (16.7%; 2/12) and garden mixes (13.6%; 3/22),

while no C. difficile was isolated from pre-digestate liquid (0.0%, 0/3) (Table 4.1).

Of the 22 C. difficile-positive samples, 10 (45.5%) were positive for toxigenic

C. difficile strains. In general, toxigenic C. difficile were found in all (3/3) of the

digestate liquid, 25.0% (1/4) of human biosolids, 13.5% (5/37) of soil conditioners and

4.5% (1/22) of garden mixes.

In total, 36 C. difficile strains were isolated from the 22 C. difficile-positive samples.

Eight samples [4 soil conditioners (CP009, CP031, CP054 and CP071), 2 human

biosolids (CP065 and CP067), 1 liquid digestate (CP017) and 1 garden mixes (CP072)]

were contaminated with two C. difficile strains (Table 4.2), one digestate liquid sample

(CP016) was contaminated with three C. difficile strains (RTs QX 546, QX 547 and

237) and one human biosolid (CP064) was contaminated with five C. difficile strains

(RTs QX 463, QX 551, 010, 039 and 125). Each of the remaining samples was only

contaminated by one strain of C. difficile.

Nearly half of the isolated strains were toxigenic (44.4%, 16/36); of which, 62.5%

(10/16) were A+B+CDT-, 31.3% (5/16) were A-B+CDT+ and 6.3% (1/16) were A-

B+CDT-.Twenty-three C. difficile RTs were identified and RT banding patterns are

shown in Figure 4.2. The non-toxigenic C. difficile RT 010 (13.9%, 5/36) was the most

commonly isolated RT followed by RTs 014/020 (11.1%, 4/36), 237 (8.3%, 3/36), 039

(5.6%, 2/36), 051 (5.6%, /36), 056 (5.6%, 2/36) and QX 608 (5.6%, 2/36). Other

internationally recognised RTs identified were RTs 005 (1), 012 (1), 017 (1), 046 (1),

080 (1) and 125 (1). The remaining 33.3% (12/36) isolates did not match any

international reference strains, 58.3% (7/12) of which were RTs currently circulating in

Australia [QX 026 (n = 1), QX 077 (1), QX 140 (1), QX 210 (1), QX 327 (1), QX 463

(1) and QX 551 (1)] and 41.7% (5/12) were novel to our laboratory [QX 608 (n = 2),

QX 546 (1), QX 547 (1) and QX 597 (1)].

4.2.2 Prevalence and molecular epidemiology of C. difficile in the public

lawn of WA

A total of 311 lawn samples were collected from public parks within 30 km north and

80 km south of Perth, WA. Overall, C. difficile was isolated in 58.5% (182/311) of lawn

samples. Compared to old established lawn, the prevalence of C. difficile in the newly


88

Table 4.1 Prevalence and toxin gene profiles of C. difficile cultured from compost products in WA.

No. of isolates, n (%)

Origin Prevalence, n (%) A-B-CDT- A+B+CDT- A-B+CDT+ A-B+CDT- Total

Composting / digesting

Digestate liquid 3/3 (100.0) 1 (2.8) 0 (0.0) 5 (13.9) 0 (0.0) 6 (16.7)

Human biosolids 3/4 (75.0) 7 (19.4) 2 (5.6) 0 (0.0) 0 (0.0) 9 (25.0)

Pre-digestate liquid 0/3 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Final composted products

Soil conditioners 11/37 (29.7) 8 (22.2) 6 (16.7) 0 (0.0) 1 (2.8) 15 (41.7)

Mulches 2/12 (16.7) 2 (5.6) 0 (0.0) 0 (0.0) 0 (0.0) 2 (5.6)

Garden mixes 3/22 (13.6) 2 (5.6) 2 (5.6) 0 (0.0) 0 (0.0) 4 (11.1)

Total 22/81 (27.2) 20 (55.6) 10 (27.8) 5 (13.9) 1 (2.8) 36 (100.0)


89

Table 4.2 Summary of C. difficile PCR RTs isolated from compost products, sorted

by the sample.

Sample Origin No. of isolates, n (%) PCR ribotype

CP001 Soil conditioner 1 (2.8) QX 608

CP002 Garden mixes 1 (2.8) QX 608

CP008 Digestate liquid 1 (2.8) 237

CP009 Soil conditioner 2 (5.6) 014/020 and 051

CP016 Digestate liquid 3 (8.3) QX 546, QX 547 and 237

CP017 Digestate liquid 2 (5.6) 080 and 237


CP031 Soil conditioner 2 (5.6) QX 597 and 010

CP034 Soil conditioner 1 (2.8) 056


CP041 Garden mixes 1 (2.8) QX 210

CP046 Mulches 1 (2.8) 010

CP054 Soil conditioner 2 (5.6) 012 and 014/020

CP061 Mulches 1 (2.8) 051

CP064 Human biosolids 5 (13.9) QX 463, QX 551, 010,

039 and 125

CP065 Human biosolids 2 (5.6) QX 140 and 010

CP067 Human biosolids 2 (5.6) QX 026 and 046


CP071 Soil conditioner 2 (5.6) 014/020 and 017

CP072 Garden mixes 2 (5.6) 005 and 014/020




90


in compost products in WA. RTs 027 and 078 were included as reference strains (R).

PCR ribotyping pattern analysis was performed by creating a neighbour-joining tree,

using Dice coefficient (optimisation, 5%; curve smoothing, 1%). QX, internal

nomenclature; DL, digestate liquid; HB, human biosolids; SC, soil conditioner; M,

mulches; GM, garden mixes.


91

laid lawn was significantly higher (46.9% vs. 65.2%, Chi-square p = 0.0017) (Table

4.3). The newly laid lawn has a visible gap between two or more strips of laid down

lawns which generally remain visible up to 4 months after laying down270. Lawns that

did not have this visible gap were assumed to be old and established. By univariate

analysis, C. difficile was twice more likely to be detected in the newly laid lawn than in

old lawn (OR = 2.11, 95% CI: 1.3 – 3.4). The size, location of park and season were not

associated with the detection of C. difficile in lawn, however, the study period was too

short to effectively observe any effect of seasonality on recovery.

Of the 182 C. difficile-positive lawn samples, 86 (47.3%) were positive for toxigenic

C. difficile strains. Overall, toxigenic C. difficile were found in 30.3% (60/198) of

sampled newly laid lawn and 23.0% (26/113) of old established lawn, no significant

differences were observed between the two (Chi-square p = 0.167).

One isolate per sample was cultured. The 182 isolated strains belongs to 33 different

C. difficile RTs (Figure 4.3). Almost half of the strains were toxigenic (47.3%, 86/182),

all of which were A+B+CDT-, and all remaining strains were non-toxigenic (52.7%,

96/182). C. difficile RT 014/020 (39.0%, 71/182) was the most commonly isolated RT,

followed by RTs 010 (21.4%, 39/182), QX 077 (8.2%, 15/182), QX 189 (3.8%, 7/182)

and 039 (2.7%, 5/182). Among toxigenic strains, C. difficile RT 014/020 predominated

in both newly laid lawn (80.0%, 48/60) and old established lawn (88.5%, 23/26). It was

also the most widely distributed in WA and predominate in nearly all sampled

postcodes (Figure 4.4). Overall, 70.3% (128/182) of the isolates were internationally

recognised RTs 014/020 (n = 71), 010 (39), 039 (5), 002 (4), 054 (2), 056 (2), 009 (1),

012 (1), 106 (1), 108 (1) and 125 (1). The remaining 29.7% (54/182) isolates did not

match any international reference strains, 72.2% (39/54) of which were RTs currently

circulating in Australia [QX 077 (15), QX 189 (7), QX 142 (4), QX 518 (3), QX 393

(2), QX 608 (2), QX 001 (1), QX 054 (1), QX 072 (1), QX 121 (1), QX 210 (1) and QX

409 (1)] and 27.8% (15/54) were novel to our laboratory [QX 601 (4), QX 610 (2), QX

611 (2), QX 549 (1), QX 550 (1), QX 602 (1), QX 603 (1), QX 605 (1), QX 606 (1) and

QX 607 (1)].

4.2.2.1 Enumeration of C. difficile in public lawn

Viable counts was performed on 10 random lawn samples collected from public parks

that previously tested positive for C. difficile by enrichment culture. Of the 10 samples,

five tested positive by direct culture. The highest viable count of C. difficile was


92

Table 4.3 Prevalence and risk factor analysis for C. difficile positivity in public

lawns.

Variable Prevalence, n (%) Univariate OR (95% CI)

Age*

New lawn 129/198 (65.2) 2.11 (1.32 – 3.40)

Old lawn 53/113 (46.9) Referent

Size of park†

Small 43/72 (59.7) 0.89 (0.47 – 1.71)

Medium 60/101 (59.4) 0.88 (0.49 – 1.59)

Large 26/53 (49.1) 0.58 (0.28 – 1.16)

Extra-large 53/85 (62.4) Referent

Location‡

South 84/150 (56.0) 1.22 (0.78 – 1.92)

North 98/161 (60.9) Referent

Season

Winter 47/87 (54.0) 0.77 (0.47 – 1.28)

Autumn 135/224 (60.3) Referent

* New lawn (has a visible gap between two or more strips of laid down lawns), ≤ 4

months, while old lawn (does not have a visible line of gap between strips of lawn), > 4

months; † small (0.5 – 1 km2), medium (1.1 – 2 km2), large (2.1 – 2.9 km2) and extra-

large (≥ 3 km2); ‡ south or north of Perth City, WA.


93


in the public lawn of WA. RTs 027 and 078 were included as reference strains (R). PCR

ribotyping pattern analysis was performed by creating a neighbour-joining tree, using

Dice coefficient (optimisation, 5%; curve smoothing, 1%). QX, internal nomenclature.


94

Figure 4.4 Geographical distribution of C. difficile RTs in the public lawn of WA.

Pie charts show the proportion of the most common RTs per postcode. The number in

the centre of the pie charts is the prevalence of C. difficile in that postcode.


95

1,200 cfu/g and the lowest was 50 cfu/g (Table 4.4). Two different diluents were used

to enumerate C. difficile in lawn, PBST recovered C. difficile in five of the sample (50 –

1,200 cfu/g), while PST recovered C. difficile in four (50 – 250 cfu/g).

4.3 DISCUSSION

Traditionally regarded as a hospital-acquired infection, the emergence of CA-CDI has

prompted a need to investigate other potential sources of C. difficile outside of the

hospital setting. This chapter was the first to describe the isolation and characterisation

of C. difficile in the environment of Australia and to the best of our knowledge the first

to investigate the presence of C. difficile in commercially composted products and

public space lawn. Only one other study, in Japan in 2017, has investigated the

prevalence of C. difficile in compost products, in which 35.7% (5/14) of composted pig

manure (a type of soil conditioner) tested positive for C. difficile265. Although the

composting procedure was not monitored in that study, and thus it is not possible to

know if it complied with international standards, the finding was comparable to the

results reported here [29.7% (11/37) in the soil conditioners]. C. difficile-contaminated

compost can contribute to the widespread dissemination of C. difficile as they are

widely used in farming, gardening and landscaping. Products of the national brand are

being shipped and sold in leading retailers across the country. Recent studies by Xu et

al. and Usui et al. in 2016 and 2017, respectively, showed that C. difficile spores are

resilient and can survive the composting process264, 265. Although complete elimination

is implausible, Xu et al. reported that windrow composting is an effective means for

reducing C. difficile spores in compost, from 3.7 log10 cfu/g down to 0.3 log10 cfu/g,

with the greatest reduction occurring during the curing phase264. Of the limited number

of studies that have investigated the survival of C. difficile during composting264 and its

presence in the final composted products265; one stark difference in WA is the adding of

anaerobically digested liquid to compost piles as a mean to maintain appropriate

moisture content. The anaerobic digestive tanks which operate at a mesophilic

temperature (incapable of killing spores) would be an ideal environment for C. difficile

to flourish. This inference was supported by a complete absence of C. difficile in the

pre-digestate liquid despite enrichment culture, but a 100.0% prevalence in digestate

liquid (Table 4.1). To determine if the addition of digestate liquid is, in fact, re-

introducing C. difficile into the compost and if so, by how much, quantification of

C. difficile at various stages of the composting process is required. By exploring ways to


96

Table 4.4 Enumeration of C. difficile in soil from the lawn samples, using two

different diluents.

Viable count (cfu/g of soil)

Sample PBST PST

1 0 0

2 0 0

3 0 0

4 50 800

5 50 200

6 250 100

7 0 0

8 0 50

9 0 0

10 100 1200

PBST, phosphate buffer solution with 0.1% Tween 20; PST, peptone saline with 0.1%

Tween 20.


97

reduce the level of C. difficile in the digestate liquid, one could potentially lower the

level of C. difficile contamination in compost and limit its exposure to susceptible

individuals. This could be achieved by increasing the internal temperature of the

anaerobic digestive tanks to 96 °C, as shown by Rodriguez-Palacios et al. to inhibit

99.9% of C. difficile spores within 1 – 2 min (a 6.5 log10 reduction)271.

No other studies have investigated the presence of C. difficile in lawn, as other

environmental studies focused mainly on soil143-145, 147, 148, water136, 143-147 and household

samples150, 151, 265 (Section 1.3.2.3). Previous studies between 1996 and 2016 found

0.0% ‒ 37.0% of soil samples in the community to contain C. difficile143-145, 147, 148. This

is considerably lower than the prevalence found in the public lawn of WA (58.5%,

182/311), which consist of grass with soil still attached to its root system. This disparity

could be due to (i) differences in culturing method, as most C. difficile environmental

studies used direct culture instead of enrichment143-145, 147, 148; and/or (ii) the practice of

using pig manure as fertiliser in the agricultural production of Australian roll-out lawns

has resulted in widespread contamination of lawn with C. difficile from pigs which

according to a nationwide surveillance study in Australia in 2013 have a high

prevalence [67.2% (154/229)] of C. difficile154. Both of these possibilities are likely to

be true.

Together, C. difficile RTs 014/020 [11.1% (4/36) in compost and 39.0% (71/182) in

lawn] and 010 [13.9% (5/36) in compost and 21.4% (39/182) in lawn] accounted for

over half (53.7%, 117/218) of the isolated environmental C. difficile strains in WA. RT

014/020 was the second and first most prevalence RT in compost and lawn,

respectively. It had a median prevalence of 44.4% in parks across 10 postcodes in WA

(Figure 4.4). RT 014/020 is of significance in Australia as it is the most prevalent

disease-causing RT in both humans (~30.0%)203, 204, 206-210 and pigs (23.4%)154 (as

previously discussed in Section 1.4). This overlap of C. difficile RT 014/020 between

humans, animals and the environment could infer potential transmission of RT 014/020

between origins; but further research using WGS and high-resolution core genome

phylogenetic analyses are required to determine the relatedness of RT 014 from

humans, animals, compost and lawn. In 2017, WGS and high-resolution core genome

phylogenetic analyses were performed on 40 C. difficile RT 014 isolates of human and

porcine origins from four Australian states214. Two CGs with closely related (≤ 2 SNVs

differences in their core genome) human and porcine strains were identified, consistent

with a recent inter-species transmission. However, some of these closely related clonal


98

strains were separated by thousands of kilometres with no evidence of prior contact

between the hosts. This suggests a persistent community reservoir of C. difficile RT 014

with the potential for long-range dissemination. From this chapter, roll-out lawn

contaminated with pig manure is suspected of contributing to the widespread

contamination of RT 014 in WA. Interestingly, the majority of CA-CDI cases in WA

reside in newly developed suburbs in the outskirt of Perth City (Figure 4.5) where roll-

out lawn remains a popular choice for the yards and fields of residential homes and

parks (L Bloomfield and TV Riley, unpublished data). In 2017, RT 014/020 was also

revealed as the most common cause of CA-CDI in WA, accounting for 40.7% of

cases204.

Currently, the infective dose of C. difficile is not known, although the suggestion has

been made that it could be between 100 – 1000 spores, depending on the host

susceptibility272. The highest count of C. difficile in WA’s lawn was 1,200 cfu/g of

sample, however, it is unclear how much exposure to the contaminated environment is

required for someone to acquire sufficient spores that could lead to CDI. Furthermore,

ingestion of C. difficile spores does not always lead to symptomatic infection due to the

complex pathogenesis of CDI that requires the gut microbiota to be disturbed29.

4.4 CHAPTER SUMMARY

In conclusion, this chapter examined the prevalence and molecular epidemiology of

C. difficile in composted products and public lawns in WA. The high prevalence and

isolation of clinically important C. difficile strains suggest that the environment might

be a potential source of C. difficile outside of the hospital setting. C. difficile RT

014/020, a C. difficile lineage of emerging One Health importance and the most

prevalent RT in both humans and porcine in Australia, predominated in the

environmental samples. This finding supports the hypothesis of environmental

transmission of C. difficile, although more discriminatory tying methods such as WGS

are necessary to examine the extent of genetic overlap between strains of human, animal

and environmental origins.

The practice of direct fertilisation of roll-out lawn with pig manure and the addition of

anaerobically digested liquid onto compost pile present significant and perhaps under-

appreciated risk factors for contamination of lawn and compost products. Both compost

and roll-out lawn have the capacity to cross vast distances and may contribute to clonal

CDI in individuals with no evidence of an epidemiological link.


99

Figure 4.5 Incidence of CA-CDI cases in WA from 2010 and 2014, sorted by

residential postcode (L Bloomfield and TV Riley, unpublished data). Data were

collected as part of the HISWA study. The incidence rate of CDI is available in HISWA

reports273-276, but information on CA-CDI was not published.

Chapter 5: Susceptibility to antimicrobials / preservatives

100

CHAPTER 5

SUSCEPTIBILITY OF FOOD AND

ENVIRONMENTAL C. DIFFICILE TO

ANTIMICROBIALS AND

PRESERVATIVES

The work presented in this chapter has been published:

Lim SC, Foster NF, Riley TV. Susceptibility of Clostridium difficile to the food

preservatives sodium nitrite, sodium nitrate and sodium metabisulphite. Anaerobe 2016;

37: 67-71.

Lim SC, Androga GO, Knight DR, Moono P, Foster NF, Riley TV. Antimicrobial

susceptibility of Clostridium difficile isolated from food and environmental sources in

Western Australia. International Journal of Antimicrobial Agents 2018; 52:411-15.


101

This chapter described two separate studies that investigate (i) the susceptibility of food

and environmental C. difficile isolates from WA to 10 antimicrobial agents (Section

5.1), and (ii) the susceptibility of food and animal C. difficile isolates to three food

preservatives commonly used in RTE meat (Section 5.2).

5.1 ANTIMICROBIAL SUSCEPTIBILITY OF FOOD AND

ENVIRONMENTAL C. DIFFICILE IN WA

5.1.1 Introduction

Recent studies have found genetically-related, and in some cases indistinguishable,

C. difficile strains from animals and humans, indicative of zoonotic or anthropozoonotic

transmission66, 117, 195, 214. However, related isolates were almost always separated by

vast geographical distances and there was no known prior contact between hosts,

making direct contact transmission unlikely to be occurring. However, indirect contact

with animal C. difficile strains through contaminated food and/or the environment is a

possibility and could be a potential source of C. difficile in the community. This could

happen through land and farm application of animal manure as in the common practice

of recycling biowaste in developed countries197, 251, 277. In Australia, each year

approximately 1.7 million tonnes of animal manure is applied to agricultural land as

fertiliser, 6.8% (115 989 tonnes) of which is used for farming in WA251.

Similar antimicrobial profiles between animal and human C. difficile isolates have been

reported, supporting the hypothesis that animal and human C. difficile may be from a

common source214, 278, 279. In Australia, animal and human isolates of clade 5 origin

(RTs 033, 078, 126, 127, 237, 281 and 288) showed little variation and no significant

difference in MICs of fidaxomicin, vancomycin, metronidazole, rifaximin,

amoxicillin/clavulanate, meropenem, piperacillin/tazobactam, ceftriaxone and

trimethoprim278. Significant variation in resistance between animal and human isolates

was observed with clindamycin, erythromycin, moxifloxacin and tetracycline but this

was limited to RTs 012 and 127. Similar resistance and susceptibility profiles were also

observed with animal and human C. difficile strains in Slovenia and 26 of the 30 tested

antimicrobials279.

Animal-to-human transmission of C. difficile through food is possible as shown with

other enteric pathogens of animal origin that cause human colonisation or infection via

ingestion of food280-282. In 2001, ingestion of a glycopeptide-resistant animal

Enterococcus faecium strain via chicken or pork resulted in detection in human stool for


102

up to 14 days after ingestion281. A study in Denmark documented an outbreak of

quinolone-resistant Salmonella enterica through pork283. Source tracking led the

outbreak investigation to a slaughterhouse, and pulsed-field gel electrophoresis

identified the source of infection as two herds of swine that recently shared a shed.

While human and animal C. difficile isolates are occasionally surveyed for antimicrobial

susceptibility, data on AMR in food and environmental C. difficile are sparse. In this

section, 274 food and environmental C. difficile strains [272 strains as described in

Chapters 3 and 4, and two additional strains (one RTs QX 399 and one 014/020)

isolated from vegetables purchased from a major local distributor] were tested against a

panel of 10 antimicrobial agents (fidaxomicin, vancomycin, metronidazole, rifaximin,

clindamycin, erythromycin, amoxicillin/clavulanate, moxifloxacin, meropenem and

tetracycline) using an agar incorporation method (Section 2.5.1). This collection

represented 70 different RTs, many of which overlapped with strains commonly

isolated from symptomatic animals and humans. Strains were predominantly 014/020

(28.5%, 78/274) and non-toxigenic 010 (15.7%, 43/274) (Figure 2.2 and Figure 2.3).

The toxin profiles included A-B-CDT- (n = 145; 52.9%), A+B+CDT- (n = 117; 42.7%),

A-B+CDT+ (n = 7; 2.6%), A+B+CDT+ (n = 2; 0.7%), A-B-CDT+ (n = 2; 0.7%) and A-

B+CDT- (n = 1; 0.4%). As food and environmental C. difficile may be of animal origin

and act as a source of human CDI, the hypothesis was that the antimicrobial profiles

would be comparable to those of animals and/or humans.

5.1.2 Results

Fidaxomicin was the most active agent, showing potent in vitro activity against all

isolates (MIC50/MIC90, 0.06/0.12 mg/L; MIC range < 0.002 – 0.5 mg/L) (Table 5.1).

This activity was superior to the recommended first-line treatment agents for CDI,

vancomycin (MIC50/MIC90, 1/2 mg/L) and metronidazole (MIC50/MIC90, 0.25/0.5

mg/L). However, no metronidazole resistance was observed and only two (0.73%)

isolates were resistant to vancomycin (MIC = 4 mg/L, resistant breakpoint > 2).

Phenotypic resistance to at least one antimicrobial agent was observed in 103 (37.6%)

of the 274 isolates, predominantly those from compost (14/36, 38.9%) and lawn

(82/182, 45.1%). Only 7/56 food isolates (12.5%) exhibited resistance. All seven AMR

food isolates were different RTs isolated from vegetables [five potatoes (RTs 051, 101,

QX 145, QX 518, QX 519), one spring onion (RT QX 393) and one beetroot (RT 012)]

bought from different stores.


103

Table 5.1 Antimicrobial susceptibilities of 274 C. difficile isolates from food, compost and lawn.

Breakpoints* MIC range

(mg/L)

MIC50

(mg/L)

MIC90

(mg/L)

GM

(mg/L)

Resistant

% (n)

Non-resistant

% (n)

Agent Origin n S I R p-value†

FDX Food 56 - - ≥ 1 0.004 - 0.12 0.06 0.06 0.03‡ 0 (0) 100 (56)

Compost 36 0.004 - 0.5 0.06 0.12 0.06 0 (0) 100 (36)

Lawn 182 < 0.002 - 0.12 0.06 0.12 0.06 0 (0) 100 (182)

Total 274 <0.002 – 0.5 0.06 0.12 0.05 0 (0) 100 (274) 1

VAN Food 56 ≤ 2 - > 2 0.5 - 2 2 2 1.41 0 (0) 100 (56)

Compost 36 1 - 2 1 2 1.26 0 (0) 100 (36)

Lawn 182 0.25 - 4 1 2 1.27 1.1 (2) 98.9 (180)

Total 274 0.25 - 4 1 2 1.29 0.7 (2) 99.3 (274) 1

MTZ Food 56 ≤ 2 - > 2 0.25 - 1 0.5 0.5 0.40§ 0 (0) 100 (56)

Compost 36 0.25 - 2 0.25 0.5 0.29 0 (0) 100 (36)

Lawn 182 0.12 - 1 0.25 0.5 0.30 0 (0) 100 (182)

Total 274 0.12 - 1 0.25 0.5 0.32 0 (0) 100 (274) 1

RFX Food 56 - - ≥ 32 0.004 - 64 0.008 8 0.04 3.6 (2) 96.4 (54)

Compost 36 0.004 - > 64 0.008 0.25 0.02 2.8 (1) 97.2 (35)

Lawn 182 0.002 - 16 0.004 1 0.01‡ 0 (0) 100 (182)

Total 274 0.002 - > 64 0.008 4 0.02 1.1 (3) 98.9 (271) 0.037


104


(mg/L)

MIC50

(mg/L)

MIC90

(mg/L)

GM

(mg/L)

Resistant

% (n)

Non-resistant

% (n)


CLI Food 56 ≤ 2 4 ≥ 8 0.12 - 8 2 4 1.97‡ 7.1 (4) 92.9 (52)

Compost 36 1 - > 32 4 > 32 4.49 30.6 (11) 69.4 (25)

Lawn 182 0.12 - > 32 4 16 3.71 42.3 (77) 57.7 (105)

Total 274 0.12 - > 32 4 16 3.34 33.6 (92) 66.4 (182) < 0.0001

ERY Food 56 - - > 8 0.25 - 2 1 1 0.74 0 (0) 100 (56)

Compost 36 0.5 - > 256 1 > 256 2.94§ 19.4 (7) 80.6 (29)

Lawn 182 0.06 - > 256 1 2 0.81 1.1 (2) 98.9 (180)

Total 274 0.06 - > 256 1 2 0.94 3.3 (9) 96.7 (265) < 0.0001

AMC Food 56 ≤ 4 8 ≥ 16 0.125 - 1 0.5 0.5 0.41 0 (0) 100 (56)

Compost 36 0.06 - 4 0.25 1 0.44 0 (0) 100 (36)

Lawn 182 0.125 - 8 0.5 1 0.66§ 0 (0) 100 (182)

Total 274 0.06 - 8 0.5 1 0.57 0 (0) 100 (274) 1

MXF Food 56 ≤ 2 4 ≥ 8 0.5 - 4 2 2 1.43 0 (0) 100 (56)

Compost 36 1 - 4 2 2 1.56 0 (0) 100 (36)

Lawn 182 0.25 - 8 2 2 1.40 1.1 (2) 98.9 (180)

Total 274 0.25 - 8 2 2 1.43 0.7 (2) 99.3 (272) 1


105


(mg/L)

MIC50

(mg/L)

MIC90

(mg/L)

GM

(mg/L)

Resistant

% (n)

Non-resistant

% (n)


MEM Food 56 ≤ 4 8 ≥ 16 1 - 4 2 4 2.73 0 (0) 100 (56)

Compost 36 1 - > 16 4 4 3.30¶ 5.6 (2) 94.4 (34)

Lawn 182 0.5 - 16 2 4 2.42 0.5 (1) 99.5 (181)

Total 274 0.5 - > 16 2 4 2.58 1.1 (3) 98.9 (271) 0.071

TET Food 56 ≤ 4 8 ≥ 16 0.06 - 64 0.12 0.25 0.15 1.8 (1) 98.2 (55)

Compost 36 0.06 - 64 0.12 16 0.49§ 13.9 (5) 86.1 (31)

Lawn 182 0.03 - 16 0.12 0.5 0.17 1.1 (2) 98.1 (180)

Total 274 0.03 - 64 0.12 0.5 0.19 2.9 (8) 97.1 (266) 0.002

*, Breakpoints were recommended by EMA (report WC500119707, http://www.ema.europe.eu/)224, EUCAST223, O’Connor et al.225, and CLSI215; †,

Chi-square / Fisher exact to compare C. difficile resistant % of food, compost and lawn origins; ‡, GM is significantly lower than the other two origins

(FDX, p < 0.0001; RFX, p < 0.005; CLI, p < 0.0001); §, GM is significantly higher than the other two origins (MTZ, p < 0.0001; ERY, p < 0.005;

AMC, p < 0.0001; TET, p < 0.0001); ¶ , GM is significantly higher than the lawn isolates, but not the food isolates (MEM, p < 0.005); FDX,

fidaxomicin; VAN, vancomycin; MTZ, metronidazole; RFX, rifaximin; CLI, clindamycin; ERY, erythromycin; AMC, amoxicillin/clavulanate; MXF,

moxifloxacin; MEM, meropenem; TET, tetracycline; S, susceptible; I, intermediate; R, resistance; MIC, minimum inhibition concentration; GM,

geometric mean.

http://www.ema.europe.eu/


106

Multidrug resistance (MDR), defined as resistance to at least one agent in three or more

antimicrobial categories, was observed in 4 (3.9%) of the 103 resistant isolates and all

were of compost origin; one isolate each of RTs QX 327 (A-B-CDT-), QX 140 (A-B-

CDT-) and 046 (A+B+CDT-) was resistant to clindamycin (MIC = > 32 mg/L),

erythromycin (MIC = > 256 mg/L) and tetracycline (MIC = 64 mg/L, 16 mg/L and 32

mg/L, respectively), and one isolate of RT 012 (A+B+CDT-) was resistant to rifaximin

(MIC = > 64 mg/L), erythromycin (MIC = > 256 mg/L) and tetracycline (MIC = 64

mg/L). Almost one-third (4/14) of the resistant compost isolates were MDR. Of the four

MDR isolates, both RTs 046 and QX 140 were isolated from human biosolids, QX 327

was from chicken manure sampled from a chicken farm and RT 012 was from a bag of

commercial organic compost.

There was a significant association between the source of isolates and resistance to

rifaximin (3.6% from food, 2.8% from compost, 0.0% from lawn; Fisher’s exact p =

0.037), clindamycin (7.1% from food, 30.6% from compost, 42.3% from lawn; Chi-

square p = < 0.0001), erythromycin (0.0% from food, 19.4% from compost, 1.1% from

lawn; Fisher’s exact p = < 0.0001) and tetracycline (1.8% from food, 13.9% from

compost and 1.1% from lawn; Fisher’s exact p = 0.002) (Table 5.1). Compared to food

and lawn isolates, compost isolates were more often resistant to erythromycin (compost

vs. food, post hoc test p = 0.001; compost vs. lawn, post hoc test p = 0.0002) and

tetracycline (compost vs. food, post hoc test p = 0.049; compost vs. lawn, post hoc test p

= 0.005) (Table 5.1). The proportion of lawn isolates resistant to clindamycin was

similar to that of compost isolates (post hoc test p = 0.26); however, both were

significantly higher than food isolates (food vs. compost, post hoc test p = 0.01; food vs.

lawn, post hoc test p = < 0.0001). Susceptibility to fidaxomicin, vancomycin,

metronidazole, amoxicillin/clavulanate, moxifloxacin and meropenem did not

significantly vary between isolates from different sources. Compared to other rare RTs,

C. difficile RT 014/020 was more commonly resistant to clindamycin (28.5%, 35/123

vs. 44.9%, 35/78, respectively) (Chi-square p = 0.01) (Table 5.2). No significant

variation in clindamycin-resistance was observed between C. difficile RTs 010, QX 077,

051, 056 and other rare strains (Chi-square/Fisher exact p > 0.05).


107

Table 5.2 Antimicrobial susceptibilities of the most prevalent food and

environmental C. difficile RTs.

% resistance in various RTs (n)*

Agent

014/020

(n = 78)

010

(n = 43)

QX 077

(n = 14)

051

(n = 8)

056

(n = 8)

Rare RTs‡

(n = 123)

FDX 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

VAN 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1.6 (2)

MTZ 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

RFX 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 2.4 (3)

CLI 44.9 (35)† 37.2 (16) 35.7 (5) 12.5 (1) 0 (0) 28.5 (35)

ERY 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 7.3 (9)

AMC 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

MXF 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1.6 (2)

MEM 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 2.4 (3)

TET 1.3 (1) 0 (0) 0 (0) 0 (0) 0 (0) 5.7 (7)

* Breakpoints were recommended by EMA (report WC500119707,

http://www.ema.europe.eu/)224, EUCAST223, O’Connor et al.225, and CLSI215; †, higher

resistance compared to rare strains, Chi-square p-value = 0.01; ‡, RTs that were isolated

≤ 7 times; FDX, fidaxomicin; VAN, vancomycin; MTZ, metronidazole; RFX,

rifaximin; CLI, clindamycin; ERY, erythromycin; AMC, amoxicillin/clavulanate; MXF,

moxifloxacin; MEM, meropenem; TET, tetracycline.



108

5.1.3 Discussion

While there has been an increase in publications on C. difficile in food and the

environment, few reports have investigated the antimicrobial susceptibility of isolates

recovered130, 132, 137, 138, 146. This study is the first to determine the AMR patterns of food

and environmental C. difficile isolates in Australia.

As macrolides and tetracycline-based antimicrobial agents constitute approximately

40% (112.2 tonnes per year) of all antimicrobials used in Australian food animals284,

and waste from these production facilities is used in the manufacture of compost

products, it was not surprising to find erythromycin and tetracycline resistance among

C. difficile isolates from compost. Although clindamycin is not approved for use in food

animals, previous studies on animal C. difficile strains have frequently reported

intermediate or resistant MICs of clindamycin148, 155, 214, 279. A previous study from our

laboratory on C. difficile RT 014 from Australian pigs showed high (69%) non-

susceptibility to clindamycin, erythromycin and tetracycline214, and 100% susceptibility

to fidaxomicin, vancomycin, metronidazole, rifaximin, amoxicillin/clavulanate,

moxifloxacin and meropenem, in agreement with the resistance patterns of compost

isolates in this study. These findings, taken together, support the theory that compost

isolates are likely to be of animal origin.

The majority of food, compost and lawn isolates were non-resistant to fidaxomicin

(100%), vancomycin (99.3%), metronidazole (100%), rifaximin (98.9%),

amoxicillin/clavulanate (100%), moxifloxacin (99.3%) and meropenem (98.9%). More

variation in resistance profiles was found against clindamycin (7.1% in food, 30.6% in

compost and 42.3% in lawn), erythromycin (0.0% in food, 19.4% in compost and 1.1%

in lawn) and tetracycline (1.8% in food, 13.9% in compost and 1.1% in lawn). Although

not indicative of transmission to humans, both compost and lawn isolates shared

AMR/susceptibility patterns similar to those reported in human-derived isolates210. In

2015, antimicrobial susceptibility testing was performed on 440 human C. difficile

isolates collected across five Australian states210. In that study, the majority of

C. difficile isolates were susceptible to fidaxomicin (100%), vancomycin (100%),

metronidazole (100%), rifaximin (100%), amoxicillin/clavulanate (100%), moxifloxacin

(96.1%) and meropenem (99.5%), similar to the findings for food, compost and lawn

isolates. In addition, the rate of resistance for clindamycin in human C. difficile was

84.3%210. In this study, both compost (30.6%) and lawn (42.3%) isolates displayed

relatively high levels of clindamycin resistance and the MICs were comparable to those


109

of humans (MIC50/MIC90, 8/> 32 mg/L; MIC range 0.5 – > 32 mg/L)210. Unfortunately,

the Australian human isolates were not tested against erythromycin and tetracycline so

it was not possible to compare their antimicrobial profile to that of isolates of food,

compost and lawn origin. However, based on current knowledge, the similarities in

antimicrobial profile between human, compost and lawn isolates support the hypothesis

that compost and lawn might be potential reservoirs for human CDI in Australia,

especially CA-CDI. This is in agreement with Janezic et al. where the majority of their

environmental C. difficile strains from puddle water and soil shared a comparable

antimicrobial profile with their previously published data on human and animal

isolates147, 279. In their study, similarity in resistant profile between environmental (E),

human (H) and animal (A) isolates could be observed for amoxicillin (E: 0.0%, H:

0.0%, A: 0.0%), metronidazole (E: 0.0%, H: 0.0%, A: 0.0%), vancomycin (E: 0.0%, H:

0.0%, A: 0.0%), tigecycline (E: 0.0%, H: 1.1%, A: 2.1%), rifampicin (E: 0.0%, H:

2.2%, A: 1.0%), erythromycin (E: 8.6%, H: 13.0%, A: 8.3%), clindamycin (E: 28.6%,

H: 42.4%, A: 38.5%), imipenem (E: 37.1%, H: 53.3%, A: 28.1%) and linezolid (E:

0.0%, H: 0.0%, A: 0.0%). Further research using whole-genome sequencing, which

offers greater discrimination than conventional typing methods, is necessary to truly

determine the relatedness of these environmental isolates to those of humans,

particularly in the community.

Globally, C. difficile RT 014/020 is consistently one of the most frequently isolated

toxigenic RTs in humans82, 83, 207, 209, 212. It is the most prevalent RT in Australia,

responsible for 30.0% of CDI cases across all Australian States and Territories209. It is

one of the most frequently isolated clinical RTs in Europe82, 83, and the most common

RT in Slovenia (32.0%) and France (22.0%)83. In Asia, RT 014/020 is one of the most

prevalent RT alongside RTs 017 and 018212. Of public health interest, in this study,

C. difficile RT 014/020 (35/78, 44.9%) was significantly more likely to be resistant to

clindamycin compared to other less common RTs. More importantly, exposure to

clindamycin has been reported as a risk factor for CA-CDI (OR 16.80, 95% CI 7.48 –

37.76)285. It is unclear if the tendency of C. difficile RT 014/020 to be clindamycin-

resistant could have contributed to its widespread success in causing CDI, or whether

other factors are important. Furthermore, two of the four MDR isolates from compost

were RTs 012 and 046, RTs that are toxigenic and have previously been associated with

human CDI206, 209. This suggests that the environment could be a source of clinically

relevant C. difficile strains harbouring AMR genes.


110

Based on AMR patterns, C. difficile strains from compost were more resistant to

erythromycin, tetracycline and/or clindamycin, compared to strains from food and lawn.

This may reflect of a sampling bias as it is possible that the vegetables and lawn came

from farms that use different sources of composted animal manure as fertiliser. Due to

issues of confidentiality, it was impossible to determine from which animal farm

compost samples originally came. It is likely that acquisition of resistance may have

occurred within the Australian food animal facilities; and loss of resistance genes was

likely driven by differences in antimicrobial selective pressure between animal facilities

and the environment. For example, sub-therapeutic doses and prolonged use of

antimicrobials such as macrolides and tetracycline as growth promoters in food animals

created an ideal selective pressure for the propagation of resistant strains, but when

effluent goes to farms in the form of composted animal manure, the lack of selective

pressure could over time result in the loss of AMR genes when/if C. difficile proliferates

in the environment. Study on the persistence of C. difficile in biosolids-amended soil by

Xu et al. has shown that the level of C. difficile in soil does fluctuate across seasons,

decreasing during summer and increasing in the fall264. The acquisition and loss of

AMR genes does occur readily in C. difficile in response to selective pressure, as

C. difficile has a diverse and highly flexible accessory genome comprising a range of

mobile genetic elements conferring resistance to macrolide/lincosamide

[Tn6194/Tn5398 (ermB)] and tetracycline [Tn916/Tn5397 (tetM)], many of which are

capable of inter- and intra-species transfer in vitro62, 214, 286.


111

5.2 SUSCEPTIBILITY OF C. DIFFICILE TO FOOD PRESERVATIVES

5.2.1 Introduction

In recent years, C. difficile has been isolated from retail meats, seafood and vegetables

with prevalence up to 42.0%110, 128, 131-134, 137, 138, 194, 287. C. difficile RT 078, a

predominant strain in food animals in the northern hemisphere which is associated with

CA-CDI, was isolated in ground meat and poultry in the region123, 124. In Scotland, RTE

salads were contaminated with C. difficile RTs 001 and 017, both common clinical

isolates in Scotland and Europe194, 252, 288. Indistinguishable RTs have been found in

food, food animals and humans289. Collectively, these findings have led to concern that

C. difficile might be associated with food, contributing to the rise in CA-CDI cases.

Unlike in the northern hemisphere where C. difficile strains have been isolated in retail

meats110, 128, 134, 138, a study of Australian fresh ground meats found no C. difficile

contamination (NF Foster and TV Riley, unpublished data). This is likely due to the fact

that most retail meats in the Australian market are from adult animals290 which have a

very low prevalence of C. difficile at the time of slaughter122. The only neonatal animals

in Australia being consumed by the public are male veal calves under the age of 7-

days261, 290. These Australian veal calves have a high prevalence (56.4% ) of C. difficile

in their gut at the time of slaughter122 and the meat is used to make RTE products such

as cured and fermented meats (Meat and Livestock Australia, personal communication).

In Australia, some RTE meats (such as roast meat, cured and then cooked meats, and

cooked fermented meats) are heat-treated until the core reaches at least 65 °C for 10 min

which kills vegetative pathogens257, 291, 292. After cooking, a two-stage cooling process

takes place where the meat products are cool from 52 °C to 12 °C (in 6 h and 7.5 h for

uncured and cured products, respectively) and to 5 °C within 24 h of cooking257, 292. For

the non-heat treated RTE meats (dried meats, slow cured meats and uncooked

fermented meats), as the meat matures, pH decreases which inhibits bacterial growth257.

In addition to food preservatives, salt is also added to non-heat treated RTE meats to

lower the water available for microorganisms to grow. Pure water has a water activity of

1.00 and raw meat has a water activity of 0.99257 and once the water activity drops

below 0.95, many foodborne bacteria cannot replicate.

Non-heat treated RTE meats contain food preservatives to prevent spoilage. Sodium

nitrite (E250) and sodium nitrate (E251) are food preservatives widely used in

combination in non-heat treated RTE meats. Nitrite helps to develop flavour, gives the


112

characteristic pink colour of processed meat and inhibits the growth of spoilage and

pathogenic bacteria such as Clostridium botulinum293. Sodium nitrate acts as a reservoir

for nitrite production through the nitrate reductase enzyme produced by microorganisms

such as staphylococci and lactobacilli. Sodium metabisulphite (E223) is a food

preservative and effective antioxidant commonly used in raw sausages. It also acts as a

reducing agent for preventing the grey-brown discolouration of meat in raw sausages294.

The use of food preservatives in foods is strictly regulated for food safety reasons. In

Australia and New Zealand, the use of nitrite, nitrate and metabisulphite in meat

products is restricted to 125 ppm (equivalent to 125 mg/L), 500 ppm and 500 ppm,

respectively295, 296. In the USA, the concentration of nitrite and nitrate in meat products

is not allowed to exceed 200 ppm and 500 ppm, respectively297; while the European

Union set a maximum of 150 ppm nitrate and nitrite added to dry-fermented sausages

and 250 ppm nitrate in long-cured products when no nitrite is added298.

This section addresses the fourth aim of the thesis: to determine the susceptibility of

C. difficile to food preservatives commonly used in non-heat treated RTE meats. Broth

microdilution and checkerboard assays described in Sections 2.5.2 and 2.5.3 were used

to investigate the independent and combined activities of sodium nitrite, sodium nitrate

and sodium metabisulphite against 11 strains of C. difficile. The sources and molecular

characteristics of the C. difficile strains are shown in Table 2.3.

5.2.2 Results

The modal MICs of sodium nitrite, sodium nitrate and sodium metabisulphite against

C. difficile were 250 mg/L, > 4, 000 mg/L and 1, 000 mg/L, respectively (Table 5.3).

The MIC50/MIC90 of sodium nitrite, sodium nitrate and sodium metabisulphite against

C. difficile were 250/500 mg/L, > 4, 000/> 4, 000 mg/L and 1, 000/1, 000 mg/L,

respectively. All three replicates yielded similar MICs with ≤ 2-fold variation, with the

exception of three isolates (Cd1001, Cd1002 and Cd1017) with sodium nitrite where 3 –

4 fold variations were observed between replicates. No bactericidal activity against

C. difficile was observed for all three food preservatives at the highest tested

concentration (4, 000 mg/L). Checkerboard assays of sodium nitrate and sodium nitrite

gave an FICI ranging from 1.5 – 3, indicating an indifferent interaction between the two

preservatives (Table 5.4).


113

Table 5.3 Susceptibility of C. difficile to food preservatives.

C. difficile

isolates

Sodium nitrite Sodium nitrate Sodium metabisulphite

MIC* (mg/L) MBC* (mg/L) MIC* (mg/L) MBC* (mg/L) MIC* (mg/L) MBC* (mg/L)

ATCC 700057 250 1000 > 4000 > 4000 1000 2000 (1000 – 4000)†

Cd1001 125 (15.33 – 250)†, ‡ 250 (62.5 – 1000)† 2000 > 4000 1000 1000

Cd1002 500 > 4000 > 4000 > 4000 1000 > 4000

Cd1006 500 > 4000 > 4000 > 4000 1000 > 4000

Cd1009 250 (125 – 500)†, ‡ > 4000 > 4000 > 4000 1000 > 4000

Cd1017 125 (31.25 – 500)†, ‡ > 4000 > 4000 > 4000 500 > 4000

Cd1106 250 (125 – 500)† > 4000 > 4000 > 4000 1000 > 4000

AI35 500 > 4000 > 4000 > 4000 1000 > 4000

AI204 250 (125 – 500)† > 4000 > 4000 > 4000 1000 > 4000

AI218 125 > 4000 > 4000 > 4000 1000 > 4000

AI273 250 (125 – 500)† > 4000 > 4000 > 4000 1000 > 4000

Mode 250 > 4000 > 4000 > 4000 1000 > 4000

*, the values are the mode of three separate experiments; †, median (range) was reported when independent replicates yielded different MICs, hence no

mode could be determined; ‡, more than 2-fold variation in MICs was observed between replicates; MIC, minimum inhibitory concentration; MBC,

minimum bactericidal concentration.


114

Table 5.4 In vitro activities of sodium nitrite and sodium nitrate individually and in combination against C. difficile.

MIC (mg/L)

C. difficile isolates Sodium nitrate Sodium nitrite Sodium nitrate-sodium nitrite combination FICI Interactive category

ATCC 700057 ˃ 4000 125 ˃ 4000/125 2 Indifferent

Cd1001 ˃ 4000 125 ˃ 4000/125 2 Indifferent

Cd1002 ˃ 4000 125 ˃ 4000/125 2 Indifferent

Cd1006 ˃ 4000 250 ˃ 4000/250 2 Indifferent

Cd1009 ˃ 4000 250 ˃ 4000/250 2 Indifferent

Cd1017 ˃ 4000 125 ˃ 4000/250 3 Indifferent

Cd1106 ˃ 4000 250 ˃ 4000/250 2 Indifferent

AI35 ˃ 4000 500 ˃ 4000/500 2 Indifferent

AI204 ˃ 4000 500 ˃ 4000/250 2 Indifferent

AI218 ˃ 4000 250 ˃ 4000/250 1.5 Indifferent

AI273 ˃ 4000 500 ˃ 4000/500 2 Indifferent

MIC, minimum inhibitory concentration; FICI, fractional inhibitory concentration index.


115

To ensure testing quality, vancomycin was included as a positive control in each run

and had a modal MIC of 1.0 mg/L, within the acceptable range of 0.5 – 4 mg/L as

defined by CLSI215. To ensure the methodology was effective for quantifying the

inhibitory effect of food preservatives, Staphylococcus aureus ATCC 29213, which has

a reported MIC and MBC of ≤ 512 mg/L for sodium metabisulphite, was included as a

positive control227. In this study, the MIC and MBC of sodium metabisulphite against

S. aureus ATCC 29213 was 500 mg/L, in agreement with Frank and Patel (2007)227. No

growth was observed in the negative control (PBS) which was included in each

replicate.

5.2.3 Discussion

Sodium nitrite inhibited the growth of all tested C. difficile isolates at MIC50/MIC90

values of 250/500 mg/L which is higher than the maximum permitted level allowed in

RTE meats in the USA (200 ppm), Europe (150 ppm), Australia (125 ppm) and New

Zealand (125 ppm). This raises potential food safety concerns as many commercially

produced RTE meats containing these preservatives have not undergone a “kill step”

such as cooking. Even if it did, cooking until the core of the meat products reaches the

recommended cooking temperature of RTE meat (65 °C for 10 min) would not kill

C. difficile spores as they survive 71 °C for 2 h259. There are no other published data on

the effects of food preservatives on C. difficile in broth or meats, but the recorded MICs

were similar to those reported for Clostridium perfringens299. Labbe et al. (1970)299

reported that 200 – 400 mg/L of nitrite was required to inhibit C. perfringens spore

outgrowth at pH 6.

Fluctuation of MIC results between replicates of Cd1001, Cd1002 and Cd1017 against

sodium nitrite was likely due to age variation of the C. difficile cultures within the

recommended 24 – 48 h as defined by the CLSI guideline215. In hindsight, all replicates

of any future susceptibility study involving spore-forming bacteria should be

approximately the same age, ± 1 h, to ensure consistency in spore numbers. The

inhibitory mechanisms are not well understood, but previous work has shown that nitrite

inhibits microbes more effectively at low pH suggesting that the antimicrobial action is

likely associated with the generation of nitric oxide or nitrous acid299. Studies on

Clostridium sporogenes and C. perfringens spores have shown that nitrite prevented

vegetative cell division and inhibited the emergence of vegetative cells from spores

while allowing germination to occur299, 300. This is likely to also be the case for

C. difficile spores.


116

Our results suggest that nitrites can inhibit C. difficile growth but are unable to kill the

organism. This is consistent with a study on C. botulinum spores where C. botulinum

was recovered from artificially seeded, commercially formulated and cooked sausages

after storage at 8 °C for 5 weeks regardless of the concentration of nitrite added (0, 75

and 120 mg/kg), with the highest C. botulinum counts detected in nitrite-free products

and evidence of botulinum toxigenicity after 3 or 5 weeks of storage301. While

C. botulinum was detected in sausages formulated with 75 and 120 mg/kg of sodium

nitrite, none presented toxigenicity during the 5-week study period suggesting that

although the spores were able to survive the cooking and cooling process, they were

unable to produce toxigenic vegetative cells in the presence of sodium nitrite.

Sodium nitrate showed no inhibitory effect on the growth of C. difficile, even at the

maximum tested concentration of 4, 000 mg/L which is at least eight times the

concentration normally used in commercially-produced RTE meats. This is consistent

with a previous in vitro study where nitrate had no effect on spore germination and

outgrowth of C. sporogenes NCA 3679300. However, the lack of inhibitory effect by

sodium nitrate was not unexpected as nitrate needs to be reduced to nitrite by

commensal microorganisms in order to show some inhibitory effect. In meat matrices,

commensal microorganisms such as Staphylococcus and Micrococcus reduce nitrate to

nitrite302, altering their relative concentrations and therefore their inhibitory ability.

Although sodium nitrate and sodium nitrite show indifferent interactions in the

checkerboard assay, in the presence of commensal microorganisms or starter culture in

the case of fermented RTE meats, nitrate might show some effect on the growth of

C. difficile as it has on delaying botulinum toxin production303, 304.

Sodium metabisulphite inhibited the growth of C. difficile at 1, 000 mg/L, which is

higher than the allowed level of 500 ppm in Australia. No bactericidal activity against

C. difficile was observed at the highest concentration tested of 4, 000 mg/L. There are

no other published data on the effect of sodium metabisulphite on spore-forming

bacteria. Given that sodium metabisulphite is the only food preservatives used in raw

sausages and at a sub-inhibitory concentration to C. difficile, the possibility of

C. difficile contamination in raw sausages needs to be investigated.

In this study, although the reported MICs of tested food preservatives were higher than

their maximum permitted levels allowed in RTE meats, it is doubtful that C. difficile

would be able to grow when stored appropriately at 5 °C292. It would require the meats

to be stored at an abusive temperature for a long period of time for C. difficile to


117

potentially grow. Examples of abusive temperature leading to a foodborne clostridia

outbreak occurred in Ohio and Virginia in 1993305 where 156 individuals suffered

C. perfringens gastroenteritis after eating RTE corned beef on St. Patrick’s Day. The

corned beef was cooked but allowed to cool for an extended period of time at room

temperature before refrigeration. Prior to serving, the refrigerated corned beef was kept

warm at 48.8 °C. A combination of slow cooling and inadequate re-heating likely

allowed the C. perfringens spores that survived the cooking process to germinate and

replicate.

Certainly, foodborne transmission of other Clostridium spp. such as C. perfringens and

C. botulinum has been well documented in commercially produced RTE meats305-308. In

the USA, C. perfringens is the third most common pathogen contributing to foodborne

illnesses, with nearly one million cases of mild diarrhoea per year309. Being a spore-

forming bacterium, C. perfringens can survive the recommended cooking temperature

of RTE meats. As the meat cools, spores that survived the cooking process germinate to

vegetative cells that produce toxins. Fortunately, high number of C. perfringens and

C. botulinum are needed to produce enough toxin to cause symptoms upon ingestion

and their growth during the cooling period can be hindered by having an acidic pH of <

5.5, and a low water activity of < 0.93 which can be achieved by the addition of salt257.

As mentioned, nitrite is effective in delaying spore germination and toxin production303,

304. Therefore, the combined effect of low pH, low water activity, nitrite and rapid

cooling through the growth range (12 °C – 50 °C) has led to a low risk of C. perfringens

foodborne outbreaks257. Nonetheless, when allowed to grow in RTE meat that is not

adequately cooked or refrigerated, C. perfringens can produce toxins to cause foodborne

illnesses upon ingestion305-308. Currently, no study has determined if C. difficile can

produce toxins in food matrices. It is widely assumed that CDI is caused by the

ingestion of spores which germinate in the gastrointestinal tract when conditions are

favourable for growth, depending on host susceptibility such as the level of commensal

bowel microbiota, previous antimicrobial usage or old age/immune senescence96, 310.

Perhaps, future studies could investigate if C. difficile can produce toxins in meat that is

stored in abusive temperatures or inadequately cooked and refrigerated. With the recent

findings of C. difficile in raw meats at a prevalence up to 42%110, 128, 134, 138 coupled with

the ability of C. difficile to survive in the presence of food preservatives, it is a concern

if these contaminated meats are used to make RTE products. Furthermore, the infectious


118

dose of C. difficile is currently unknown but suspected to be low (100 – 1000 spores),

again depending on host susceptibility272.

To date, only two published studies have investigated the prevalence of C. difficile in

RTE meats, where 62.5% of pork Braunschweiger sausage134 and 23.1% of braised pork

skin and colon164 were positive for C. difficile. The prevalence of C. difficile in RTE

meats available in WA was investigated in Chapter 3.

A limitation of this study is that it used C. difficile cultures which at the time of testing

contained approximately 15% spores. Further research is required to study the effect of

food preservatives on C. difficile spores and vegetative cells separately. The impact of

pH, temperature and other commensal microorganisms on the activity of food

preservatives, possibly through the use of meat matrices, also needs exploring.

5.3 CHAPTER SUMMARY

This chapter presents the first investigation of the antimicrobial susceptibilities of food

and environmental C. difficile in Australia. AMR was uncommon among isolates from

food; however, clindamycin-resistance was common in isolates from lawn; while,

isolates from compost showed high resistance to clindamycin, erythromycin and

tetracycline. All MDR isolates originated from compost. Both compost and lawn

isolates shared antimicrobial profiles similar to those reported in animal and human

isolates, supporting the hypothesis of animal-to-human transmission through a

contaminated environment. Future studies using WGS are required to determine the

relatedness of these environmental isolates to those of humans, especially from CA-CDI

cases. This study provides valuable baseline data for future antimicrobial testing of food

and environmental C. difficile strains in Australia.

Also for the first time, the effect of food preservatives on C. difficile was investigated.

Although the relevance of contaminated food in the epidemiology of CDI is yet to be

confirmed, the ability of C. difficile to survive in the presence of food preservatives

indicates a need to further investigate the prevalence of C. difficile in commercially

produced RTE meats as they may serve as a source of viable C. difficile and contribute

to CDI in both hospitals and the community.

Chapter 6: Aspects of the epidemiology of CDI in WA

119

CHAPTER 6

ASPECTS OF THE EPIDEMIOLOGY

OF CDI IN WESTERN AUSTRALIA


120

This chapter described two separate studies that investigate the transmission routes of

C. difficile RTs 056 and 012 in WA, using WGS and/or case-control study.

6.1 WGS OF C. DIFFICILE RT 056

6.1.1 Introduction

C. difficile RT 056 (A+B+CDT-) is a strain that has been reported as among the most

frequent toxigenic isolates in a European hospital survey, associated with complicated

disease outcomes82. In 2016, RT 056 was predominantly CA in England311. It is

consistently one of the most prevalent CDI-causing RTs in Australia, accounting for up

to 9.6% of human CDI cases200, 203, 204, 206-208, 312. In 2013, a study of the prevalence of

C. difficile in veal calves was conducted in 25 abattoirs across five Australian States. In

this study C. difficile RT 056 was the third most common RT in the gastrointestinal

tracts of < 7-day-old calves, contributing to 7.7% (16/209) of isolates122. In the work

presented earlier in this thesis, C. difficile RT 056 was one of the most prevalent RTs in

Australian foods and the environment (Chapters 3 and 4). The presence of this RT in

humans, animals, food and the environment in Australia suggests a possible common

transmission pathway such as animal-to-human transmission through contaminated

foods and/or the environment.

To date, assessment of genetic overlap has been limited to low resolution genotyping

tools such as analysis of the 16S – 23S rRNA intergenic spacer region48, i.e. ribotyping.

In this section, to detect evidence of potential transmission routes, WGS and high-

resolution core genome phylogenetics were performed on a collection of 29 C. difficile

RT 056 of human (n = 21), food (n = 4) and environmental (n = 4) origins. Information

on this strain collection is available in Section 2.6.1. DNA extraction, sequencing and

analysis of WGS data were performed as described in Section 2.6.

6.1.2 Results

6.1.2.1 MLST

All 29 C. difficile RT 056 displayed allelic conservation in the seven housekeeping

genes (adk1, atpA5, dxr7, glyA1, recA1, sodA3 and tpi1). These sequences were

consistent with clade 1 and ST 34.


121

6.1.2.2 SNV analysis

A SNV-based maximum-likelihood phylogenetic tree of the 29 C. difficile RT 056 in

clonal frame is shown in Figure 6.1. A heatmap of pairwise SNV differences between

the 29 genomes is shown in Figure 6.2. Applying a fixed-rate molecular clock of 1-2

SNVs per-genome per-year, as calculated for C. difficile in a number of studies that use

clinical isolates from the same patient over a period of time63, 64, three CGs were

identified (strains differing by ≤ 2 SNVs in their core genome). Two of these CGs

included both human and food or compost strains. Overall, 14% of human strains

showed a clonal relationship with one or more food/compost strains.

CG II comprised of two human (WA0959 and WA1091) isolates and one food (organic

potato, FI040) isolate collected over a 2.5-year period (February 2012 – August 2015).

Of the two human strains, one was classified as CA-CDI (WA1091) and the other as

HA-CDI (WA0959) with healthcare facility onset. The HA-CDI isolate was from a

patient living in a rural area, while the CA-CDI isolates came from a Perth metropolitan

area resident (separated by 130.98 km). These were recovered 33 days apart and from

two different hospitals separated by 27.9 km, while the food strain was isolated from an

organic potato over 2 years later. The food isolate was more closely related to WA0959

(1 SNV difference in their core genome) than to WA1091 (3 SNVs difference).

CG III comprised of two human (WA1945 and WA2779) isolates, two compost (CI014

and CI015) isolates and one food (organic potato; FI043) isolate collected over a 3-year

period (January 2013 – December 2015). The human strains were from CA-CDI

(WA2779) and HA-CDI (WA1945) healthcare facility onset cases, isolated almost 12

months apart and from two different hospitals separated by 17.8 km. They were both

isolated from CDI cases living in the Perth metropolitan area. The food and compost

isolates were recovered 2 – 3 years after the human isolates. No SNV differences were

observed between the two compost isolates despite one having been cultured from a

residential home that was giving out free composted sheep manure and the other from a

major commercial supplier in WA. All other isolates in this CG were 2 SNVs away

from one another.

CG I comprised of two human isolates collected 2-year apart (September 2012 and

September 2014). Both were from HA-CDI healthcare facility onset cases but originated

from different hospitals, 13.0 km apart. The isolates had a 1 SNV difference in their


122

Figure 6.1 SNV analysis of 29 C. difficile RT 056. Maximum-likelihood phylogeny

was based on non-recombinant SNVs identified after mapping all sequence reads

against the CD630 reference genome (accession AM180355, 4,290,252 bp). Taxa labels

include ID: origin, state, isolation date and acquisition status (if known). CRT, carrot;

POT, potato; CLN, clinic; SC, soil conditioner; SM, sheep manure; CAI, community-

associated infection; HAI HCFO, hospital-associated infection healthcare facility onset;

UNK, unknown. The three CGs where isolates differed by ≤ 2 SNVs are coloured

green, red and blue, respectively. CD630 was omitted from the final phylogeny to

enhance the visual resolution of the relative evolutionary distances between genomes.


123

Figure 6.2 Heatmap of pairwise core genome SNV differences between 29 C. difficile RT 056 isolates, sorted by origin: brown, environmental (n

= 4); red, food (n = 4) and blue, human (n = 21).


124

core genome.

6.1.2.3 In silico and in vitro AMR profiling

Sequenced RT 056 genomes were examined for the presence of acquired AMR genes.

SRST2 identified only one AMR gene, the methyltransferase gene ermB in a single

human isolate (WA0707). The resistance elements tetM, tetW, tetA(P) and tetB(P) were

not found in any of the sequenced isolates.

In vitro antimicrobial activity for fidaxomicin, vancomycin, metronidazole, rifaximin,

clindamycin, amoxicillin/clavulanate, moxifloxacin, meropenem and tetracycline was

largely in agreement with the results of resistance gene profiling (Table 6.1). Strain

WA0707 showed phenotypic resistance to clindamycin (resistant breakpoint ≥ 8 mg/L)

and erythromycin (resistant breakpoint > 8 mg/L), consistent with the genotype (ermB),

with an MIC of > 32 mg/L and > 256 mg/L, respectively. While 15/29 (51.7%) isolates

displayed phenotypic resistance against clindamycin, all but WA0707 had an MIC of

only 8 mg/L. Two isolates (WA3223 and WA3461) showed phenotypic resistance to

erythromycin (MIC = 128 mg/L) with no known AMR genes.

6.1.3 Discussion

In 2013, Eyre et al. performed WGS on 957 C. difficile isolates, collected from

September 2007 to March 2011, from 1250 CDI patients in hospitals and the

community of Oxford, UK64. In that study, 45% (n = 428) of the isolates were

genetically distinct with > 10 SNVs differences in their core genome and 36%

(120/333) of closely related (≤ 2 SNVs) isolates had no known epidemiological link to

another symptomatic case or exposure to a hospital setting. The authors concluded that

the strains likely originated from outside of the hospital setting, possibly from food

and/or the environment. That was not the only study that showed the majority of the

seemingly indistinguishable clinical isolates typed by a conventional method were

actually distantly related using WGS. In 2012, 625 pairs of human C. difficile isolates of

the same ST and sampled within a month of each other in the Oxford University

Hospitals (Oxford, UK) were sequenced using WGS63. With the exception of ST1, only

19% (67/358) of the paired genomes shared a common ancestor where direct

transmission was possible. These findings confirmed that transmission between

symptomatic C. difficile patients was not the major contributor to CDI in hospitals and

that other transmission routes needed to be explored. To detect evidence of potential

foodborne and environmental transmission of C. difficile in WA, the extent of genetic


125

Table 6.1 Antimicrobial susceptibilities of 29 C. difficile RT 056.

Agent

Breakpoints† MIC range

(mg/L)

MIC50

(mg/L)

MIC90

(mg/L)

GM

(mg/L)

Resistant

% (n)

Non-resistant

% (n) S I R

Fidaxomicin - - ≥ 1 0.002 – 0.12 0.03 0.06 0.03 0 (0) 100 (29)

Vancomycin ≤ 2 - > 2 1 – 2 2 2 1.86 0 (0) 100 (29)

Metronidazole ≤ 2 - > 2 0.12 – 0.5 0.25 0.25 0.26 0 (0) 100 (29)

Rifaximin - - ≥ 32 0.002 – 0.008 0.004 0.004 0.004 0 (0) 100 (29)

Clindamycin ≤ 2 4 ≥ 8 0.12 – > 32 4 8 4.41 51.7 (15) 48.3 (14)

Erythromycin - - > 8 0.5 – > 256 1 2 1.45 10.3 (3) 89.7 (26)

Amoxicillin/clavulanate ≤ 4 8 ≥ 16 0.25 – 0.5 0.5 0.5 0.45 0 (0) 100 (29)

Moxifloxacin ≤ 2 4 ≥ 8 1 – 2 2 2 1.86 0 (0) 100 (29)

Meropenem ≤ 4 8 ≥ 16 2 – 4 4 4 3.23 0 (0) 100 (29)

Tetracycline ≤ 4 8 ≥ 16 0.06 – 0.5 0.25 0.5 0.21 0 (0) 100 (29)

S, susceptible; I, intermediate; R, resistance; MIC, minimum inhibition concentration; GM, geometric mean.

† Breakpoints for fidaxomicin recommended by EMA (report WC500119707, http://www.ema.europe.eu/)224, for vancomycin and metronidazole

recommended by EUCAST223, rifaximin as described by O’Connor et al.225, and the remaining as recommended by CLSI215.



126

overlap between C. difficile RT 056 strains of food, environmental and human origin is

described using WGS. Phylogeny based on the alignment of orthologous genes and on

SNVs in the core genome provided ultra-fine scale resolution of this C. difficile

population.

In the study, two CGs of closely related (1 – 2 SNVs) human and food/environmental

isolates were identified (CGs II and III) (Figure 6.1). The occurrence of closely related

human and food/environmental strains, collected 2 – 3 years apart, suggests that over an

extended period of time there has been a long-range transmission of C. difficile RT 056

from food and the environment to humans and/or from humans to food and the

environment, likely through the agricultural recycling of human biosolids. The faecal-

oral route of C. difficile acquisition makes it logical for the directionality of CDI

transmission to be from food and the environment to humans. Moreover, 50% of the

human strains within CGs II and III were cases classified as CA-CDI, which represents

acquisition from outside of the hospital system. In CGs II and III, the food and

environmental isolates were collected 2 – 3 years after the corresponding human clonal

isolates. The isolation date could infer contamination of foods and the environment by

human strains but that might not be true due to the ability of C. difficile to form spores

that can survive months or even years in the environment, during which evolution is

effectively suspended in time20, 313.

The molecular clock for C. difficile was estimated by several studies to be 1 – 2 SNVs

per-genome per-year through analysing the evolution and genetic diversity of

C. difficile within symptomatic patients63-66. In publications by Didelot et al. and Eyre et

al., consecutive isolates of the same ST were sampled over 1 – 561 days from 91 and

145 CDI patients, respectively63, 64. WGS was performed and the within-host

evolutionary molecular clock was estimated to be an average of 1.4 SNVs per-genome

per-year (95% CI, 0.6 – 2.3) and 0.74 SNVs per-genome per-year (95% CI, 0.22 –

1.40), respectively. These are consistent with He et al. and Knetsch et al., where the

mutation rates of C. difficile RTs 027 (1 – 2 SNVs per-genome per-year) and 078 (1.1

SNVs per-genome per-year), respectively, across populations were estimated using

programs such as BEAST and Path-O-Gen v1.365, 66. However, food and environmental

C. difficile are widely assumed to be in spore form124, so it is possible that they may

have a different molecular clock with a much slower mutation rate. The differences in

mutation rates among reproducing vegetative C. difficile and metabolically dormant

spores makes it difficult to infer transmission events as closely related cases may be


127

separated by a long interval. Even in the case of CG II where the phylogenetic tree

indicated that human isolate WA0959 was ancestral to both FI040 (food origin) and

WA1091 (human origin), it does not necessarily indicate directionality such as WA0959

causing contamination of food, or human CDI. While WA0959, WA1091 and FI040 do

shared a recent common ancestor, from that moment, the bacterium could evolve

separately inside different hosts, humans, animals or environmental.

In addition to food and the environment acting as a potential reservoir for human CDI, it

is equally important to emphasise that more than half (65.5%) of the sequenced

genomes were not clonal, with 3 – 24 SNVs difference in their core genome. While 3 –

5 SNVs differences could suggest a potential transmission event occurring a few years

ago, clinical and food/environmental isolates with more than 10 SNVs difference were

genetically so diverse that direct transmission was improbable. This suggests exposure

to multiple reservoirs of C. difficile. C. difficile is ubiquitous and found almost

everywhere in the environment (such as food138, 169, 194, water146, 147, 314, soil147 and

lawn315) as well as in humans2, 20, food animals121, 152, 154 and pets316-318.

Currently, several studies have shown C. difficile strains from animals and humans to be

genetically closely-related, suggesting zoonotic transmission66, 195, 196, 214. In one study,

WGS of 247 C. difficile RT 078 strains (predominantly from humans and animals) from

22 countries showed limited geographical clustering, suggesting repeated international

transmission196. Moreover, extensive clustering of animal and human strains was

reported, including a cluster of an animal strain from Canada with multiple human

strains from the UK. Given the lack of contact between hosts, the mode of transmission

between animals and humans was unclear. A similar finding was reported in our

laboratory’s previous study on C. difficile RT 014 where 42% of human strains showed

a clonal relationship (≤ 2 SNVs) with one or more pig strains, consistent with recent

zoonotic transmission214. However, the clonal strains were separated by vast distances

across multiple states. Again, this suggests a persistent community reservoir with long-

range dissemination potential. Through agricultural recycling of composted animal

manure and/or inadequate cleaning of meat products in a slaughterhouse, C. difficile of

animal origins could contaminate food in national and international food chains, as well

as the environment. This study provided support for the hypothesis of foodborne and

environmental transmission of C. difficile, and could potentially explain closely-related

human and animal strains with no epidemiological link.


128

Resistance to macrolide-lincosamide-streptogramin B (MLSB) antimicrobials in

C. difficile is commonly due to ermB genes which confer resistance to both clindamycin

and erythromycin through methylation of 23S rRNA. However, in this study, two

isolates without ermB genes were highly resistant to erythromycin (MIC = 128 mg/L)

but not clindamycin. This is not uncommon as an increased number of ermB-negative

C. difficile MLSB resistant strains have been identified in Europe. The mechanism of

action is yet to be identified but likely involve an unknown efflux pump286. Significant

differences in antimicrobial profile between humans, foods and environmental

C. difficile RT 056 were not observed.

There are several limitations to this study; (i) a relatively low number of C. difficile RT

056 isolates (n = 29) was investigated, all of which originated from one state (WA), (ii)

an even smaller number of food (n = 4) and environmental (n = 4) isolates was included

in the analysis, (iii) epidemiological data was not available such as the consumption of

any potentially C. difficile contaminated food, exposure to composted products or

contact with any animals prior to having CDI, and (iv) animal strains of C. difficile RT

056 were not included in the WGS. C. difficile RT 056 is predominant in Australian

calves122. Isolates from < 7-day old calves and clinical isolates from numerous

Australian States and Territories would greatly enhance our understanding of the

complex transmission of C. difficile at a national level and allow us to demonstrate more

links between animals, humans, food and the environment. This was not performed due

to lack of food and environmental isolates from other states and territories.

Finally, the transmission route of CDI is complex and development of disease depends

largely on the host susceptibility272. Our current understanding of C. difficile

transmission, particularly in the role of food and the environment, is still in its infancy.

This study demonstrated multiple clonal populations of human and food/environmental

C. difficile strains in WA indicative of a recent shared ancestry and support the

hypothesis of foodborne and environmental transmission of C. difficile. On-going

surveillance of C. difficile is still needed, especially in food and the environment. A

food and environmental sampling study, using standardise culturing method, involving

multiple states and territories is paramount to understanding C. difficile transmission.

Performing WGS and SNV analysis on a larger collection of human, animal, food and

environmental isolates from multiple states and territories would paint a clearer picture

of transmission in Australia.


129

6.2 C. DIFFICILE RT 012 IN WA

6.2.1 Introduction

Prompted by the marked increase in the incidence of CDI3, 273, in 2010 the Australian

Commission in Safety and Quality in Healthcare mandated CDI surveillance in every

hospital in all States and Territories, as CDI was not a notifiable disease in Australia. By

October 2011, all public and private hospitals in WA were submitting all C. difficile-

positive faecal samples for PCR ribotyping of isolated strains. Culture and ribotyping of

all hospital-identified C. difficile from October 2011 to March 2015 was performed as

part of the WA Department of Health, HISWA program273-276, 319. Through this

program, C. difficile RT 012 was detected as an emerging strain in WA with increasing

numbers (TV Riley, unpublished data). Since C. difficile RT 012 was first identified in

September 2012, this previously uncommon RT rose to the equal seventh most

prevalent RT in 2013 and the equal second most prevalent RT by the end of 2014

(Table 6.2). The increased numbers of C. difficile RT 012 cases in WA are shown in

Figure 6.3. Interestingly, C. difficile RT 012 cases in WA were mostly CA and female

(L Bloomfield and TV Riley, unpublished data). Following the isolation of C. difficile

RT 012 from WA-grown organic beetroot in April 2015 (Chapter 3), foodborne

transmission of RT 012 was suspected.

A retrospective case-control study was carried out to determine whether consumption of

certain vegetables or exposure to certain environmental factors increased the risk of CDI

caused by RT 012. The study design was described in Section 2.7. A subsequent study

examining the relatedness of Australian food and clinical RT 012 isolates to Asian

clinical RT 012 is also reported here. This was conducted in response to a prevalence

study from 2015 that identified this RT as one of the most common causing CDI in

Asia213.


130

Table 6.2 The 10 most common C. difficile RTs in WA, based on HISWA data, 2011 – 2014.

2011 2012 2013 2014

RT n (%) RT n (%) RT n (%) RT n (%)

1 014/020 41 (31) 014/020 111 (25) 014/020 50 (16) 014/020 182 (39)

2 054 12 (9) 002 56 (13) 056 23 (7) 002 31 (7)

3 002 8 (6) 056 27 (6) 002 16 (5) 012 31 (7)

4 018 8 (6) 053 15 (3) QX 076 15 (5) 056 29 (6)

5 015/193 7 (5) 010 14 (3) 054 14 (5) 046 14 (3)

6 056 7 (5) 015/ 193 14 (3) 046 12 (4) 103 12 (3)

7 064 5 (4) 017 13 (3) 005 11 (4) 010 9 (2)

8 244 5 (4) 018 12 (3) 012 11 (4) 018 8 (2)

9 046 4 (3) 005 11 (2) 018 11 (4) 054 8 (2)

10 QX 001 3 (2) 046 11 (2) 017 10 (3) 070 8 (2)

Data was collected as part of the HISWA study. Incidence rate of CDI is available in HISWA reports273-276, but information on RTs was not published.


131

Figure 6.3 C. difficile RT 012 cases reported to HISWA in 2011 – 2014. CAI, community-associated infection; HAI CO, healthcare-associated

infection community-onset; HAI HCFO, healthcare-associated infection healthcare facility onset; INDETER, indeterminate. Cases were defined

according to McDonald et al.94. Incidence rate of CDI is available in HISWA reports273-276, but information on RTs was not published.

0

1

2

3

4

5

6

O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D

2011 2012 2013 2014

Nu

mb

er o

f R

T 0

12 c

ase

s

Infection year and month

CAI HAI CO HAI HCFO INDETER.


132

6.2.2. Results

6.2.2.1. Case-control study

In total, 155 potential participants were contacted by questionnaire. Only 23 individuals

responded to the survey and were enrolled in the study, a response rate of only 14.8%.

No significant differences in demographic characteristics were observed between those

that were enrolled and those that were not enrolled (Table 6.3). C. difficile RT 012

cases were more likely to be living in rural areas (65.4%, 34/52) compared to other CDI

cases (26.2%, 27/103), although these differences did not reach statistical significance

(Fisher exact p = 0.08). There were no significant differences in age, gender or area of

residence between enrolled cases and controls (Table 6.4). By univariate analysis,

eating raw non-organic home-grown vegetables (cases 75.0% vs. control 15.4%) and

having contact with an individual that was living in a nursing home (cases 75.0% vs.

control 15.4%) were associated with C. difficile RT 012 (OR 16.50, 95% CI 1.09 –

250.18 for both) (Table 6.5). Multivariate analysis was not performed due to the small

number of enrolled cases and the broad confidence intervals for the only two significant

variables.

6.2.2.2. Relatedness of Australian and Asian isolates using WGS

6.2.2.2.1. MLST

All 11 C. difficile RT 012 (food-WA, n = 1; human-Australia, n = 4 and human-Asia, n

= 6) displayed allelic conservation in the seven housekeeping genes (adk1, atpA4, dxr7,

glyA1, recA1, sodA3 and tpi3). These sequences were consistent with clade 1 and ST

54.

6.2.2.2.2. SNV analysis

A heatmap of pairwise SNV differences between the 11 genomes is shown in Figure

6.4. Two CGs were observed. First, a human isolate from Shanghai, China (OT0405)

had zero SNV differences with a human isolate from Hong Kong (OT0066). These were

isolated 5 months apart and separated by 1,226 km. Second, a human isolate from WA

(WA3835) had a clonal relationship of ≤ 2 SNVs difference in their core genome with a

human isolate from Thailand (OT0171). The food isolate from WA was genetically very

distant from all other sequenced isolates, with an average SNV difference of 884.


133

Table 6.3 Characteristics of participants and people who refused to participate.

Characteristics

Potential cases Potential controls

Participant

(n = 4)

Did not participate

(n = 48)

p-value† Participant

(n = 19)

Did not participate

(n = 84)

p-value†

Female 3 (75.0) 38 (79.2) 1.00 14 (73.7) 67 (79.8) 0.55

Median age (IQR) 71.0 (61.5 – 75.3) 70.5 (57.5 – 86.3) 0.74 69.0 (59.5 – 79.5) 71.0 (54.3 – 87.0) 0.78

Residence: Metro 3 (75.0) 15 (31.3) 0.11 14 (73.7) 62 (73.8) 1.00

Note: Data are number (%) of patients unless otherwise stated; IQR, interquartile range; †, cases and controls were compared using Fisher’s exact test.


134

Table 6.4 Characteristics of CDI patients with RT 012 (cases) and CDI patients

with other RTs (controls).

Characteristic Cases (n = 4) Control (n = 19) p-value†

Female 3 (75.0) 14 (73.7) 1.00

Median age (IQR) 71.0 (61.5 – 75.3) 69.0 (59.5 – 79.5) 0.84

Residence: Metro 3 (75.0) 14 (73.7) 1.00

Note: Data are number (%) of patients unless otherwise stated; IQR, interquartile range;

†, cases and controls were compared using Fisher’s exact test.


135

Table 6.5 Risk factor analysis for C. difficile RT 012.

Variable Cases Controls Univariate OR (95% CI) p-

value

Shopped at

Coles 4/4 (100.0) 13/18 (72.2) ˃ 1000 (0.00 – undefined) 1.00

Woolworths 3/4 (75.0) 11/18 (61.1) 1.91 (0.16 – 22.20) 0.61

IGA 3/4 (75.0) 8/18 (4.4) 3.75 (0.33 – 43.31) 0.29

Fremantle market 1/4 (25.0) 0/18 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Asian market 1/4 (25.0) 1/18 (5.6) 5.67 (0.27 – 117.45) 0.26

Butcher shop 1/4 (25.0) 3/18 (16.7) 1.67 (0.13 – 22.00) 0.70

Weekend market 1/4 (25.0) 2/18 (11.1) 2.67 (0.18 – 39.63) 0.48

Organic shop 1/4 (25.0) 0/18 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Non-organic shop 2/4 (50.0) 1/18 (5.6) 17.00 (1.02 – 283.01) 0.05

Ate cooked organic

Onion 0/3 (0.0) 3/19 (15.8) 0.00 (0.00 – undefined) 1.00

Beetroot 0/3 (0.0) 0/18 (0.0) Nil Nil

Carrot 0/3 (0.0) 2/18 (11.1) 0.00 (0.00 – undefined) 1.00

Potato 0/3 (0.0) 2/18 (11.1) 0.00 (0.00 – undefined) 1.00

Spring onion 0/3 (0.0) 1/17 (5.9) 0.00 (0.00 – undefined) 1.00

Garlic 0/3 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Sprouts 0/3 (0.0) 0/18 (0.0) Nil Nil

Asparagus 0/3 (0.0) 0/18 (0.0) Nil Nil

Celery 0/3 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Lettuce 0/3 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Pre-packed salads 0/3 (0.0) 2/19 (10.5) 0.00 (0.00 – undefined) 1.00

Home-grown

vegetables

0/3 (0.0) 1/19 (5.3) 0.00 (0.00 – undefined) 1.00

Ate raw organic

Onion 0/3 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Beetroot 0/3 (0.0) 0/17 (0.0) Nil Nil

Carrot 0/3 (0.0) 1/17 (5.9) 0.00 (0.00 – undefined) 1.00

Potato 0/3 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Spring onion 0/3 (0.0) 0/17 (0.0) Nil Nil

Garlic 0/3 (0.0) 0/17 (0.0) Nil Nil


136


value

Sprouts 1/4 (25.0) 0//17 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Asparagus 0/3 (0.0) 0/17 (0.0) Nil Nil

Celery 0/3 (0.0) 0/17 (0.0) Nil Nil

Lettuce 0/3 (0.0) 2/18 (11.1) 0.00 (0.00 – undefined) 1.00

Pre-packed salads 0/3 (0.0) 3/19 (15.8) 0.00 (0.00 – undefined) 1.00

Home-grown

vegetables

1/4 (25.0) 1/18 (5.6) 5.67 (0.27 – 117.45) 0.26

Ate cooked non-organic

Onion 4/4 (100.0) 13/16 (81.3) ˃ 1000 (0.00 – undefined) 1.00

Beetroot 2/3 (66.7) 6/12 (50.0) 2.00 (0.14 – 28.42) 0.61

Carrot 4/4 (100.0) 13/15 (86.7) ˃ 1000 (0.00 – undefined) 1.00

Potato 4/4 (100.0) 15/16 (93.8) ˃ 1000 (0.00 – undefined) 1.00

Spring onion 3/4 (75.0) 6/12 (50.0) 3.00 (0.24 – 37.67) 0.40

Garlic 3/4 (75.0) 11/14 (78.6) 0.82 (0.06 – 11.00) 0.88

Sprouts 2/4 (50.0) 5/12 (41.7) 1.40 (0.14 – 13.57) 0.77

Asparagus 2/3 (66.7) 3/12 (25.0) 6.00 (0.39 – 92.28) 0.20

Celery 4/4 (100.0) 8/14 (57.1) ˃ 1000 (0.00 – undefined) 1.00

Lettuce 3/4 (75.0) 9/13 (69.2) 1.33 (0.10 – 17.10) 0.83

Pre-packed salads 1/4 (25.0) 7/15 (46.7) 0.38 (0.03 – 4.55) 0.45

Home-grown

vegetables

3/4 (75.0) 6/13 (46.2) 3.50 (0.28 – 43.16) 0.33

Ate raw non-organic

Onion 4/4 (100.0) 4/13 (30.8) ˃ 1000 (0.00 – undefined) 1.00

Beetroot 2/4 (50.0) 2/13 (15.4) 5.50 (0.46 – 65.16) 0.18

Carrot 3/4 (75.0) 5/13 (38.5) 4.80 (0.39 – 59.90) 0.22

Potato 1/4 (25.0) 1/13 (7.7) 4.00 (0.19 – 84.20) 0.37

Spring onion 3/4 (75.0) 2/12 (16.7) 15.00 (0.98 – 228.90) 0.05

Garlic 1/4 (25.0) 2/13 (15.4) 1.83 (0.12 – 27.80) 0.66

Sprouts 2/4 (50.0) 3/13 (23.1) 3.33 (0.32 – 34.83) 0.32

Asparagus 1/4 (25.0) 2/13 (15.4) 1.83 (0.12 – 27.80) 0.66

Celery 4/4 (100.0) 8/15 (53.3) ˃ 1000 (0.00 – undefined) 1.00


137


value

Lettuce 4/4 (100.0) 8/14 (57.1) ˃ 1000 (0.00 – undefined) 1.00

Pre-packed salads 1/4 (25.0) 4/12 (33.3) 0.67 (0.51 – 8.64) 0.76

Home-grown

vegetables

3/4 (75.0) 2/13 (15.4) 16.50 (1.09 – 250.18) 0.04

Ate processed meat

Salami 2/3 (66.7) 3/14 (21.4) 7.33 (0.48 – 111.19) 0.15

Ham 4/4 (100.0) 9/17 (52.9) ˃ 1000 (0.00 – undefined) 1.00

Meatball 1/3 (33.3) 0/14 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Sausage roll 3/3 (100.0) 3/16 (18.8) ˃ 1000 (0.00 – undefined) 1.00

Other cured meat 2/3 (66.7) 6/15 (40.0) 3.00 (0.22 – 40.93) 0.41

Contact with animal

Own household

pets

2/4 (50.0) 11/18 (61.1) 0.64 (0.07 – 5.61) 0.64

Other people’s

pets

2/4 (50.0) 6/16 (37.5) 1.67 (0.18 – 15.13) 0.65

Farm/wild animals 0/4 (0.0) 1/18 (5.6) 0.00 (0.00 – undefined) 1.00

Do gardening work 3/4 (75.0) 4/16 (25.0) 9.00 (0.72 – 113.02) 0.09

Contact with

Garden mix 3/4 (75.0) 4/13 (30.8) 5.75 (0.53 – 86.56) 0.14

Mulch 2/4 (50.0) 2/12 (16.7) 5.00 (0.42 – 59.66) 0.20

Compost 1/4 (25.0) 2/11 (18.2) 1.50 (0.10 – 23.07) 0.77

Animal manure 1/4 (25.0) 1/18 (5.6) 5.67 (0.27 – 117.45) 0.26

Organic fertiliser 2/4 (50.0) 1/12 (8.3) 11.00 (0.65 – 187.17) 0.10

Non-organic

fertiliser

0/4 (0.0) 2/13 (15.4) 0.00 (0.00 – undefined) 1.00

Newly laid lawn

turf

0/4 (0.0) 1/12 (8.3) 0.00 (0.00 – undefined) 1.00

Human contact

Child < 2 years old 3/3 (100.0) 4/15 (26.7) ˃ 1000 (0.00 – undefined) 1.00

Healthcare worker 1/3 (33.3) 11/13 (84.6) 0.09 (0.01 – 1.55) 0.10

Childcare worker 1/3 (33.3) 2/13 (15.4) 2.75 (0.16 – 46.79) 0.48


138


value

Person with

diarrhoea

1/3 (33.3) 1/7 (14.3) 3.00 (0.12 – 73.64) 0.50

Person who has

been hospitalised

1/3 (33.3) 5/11 (45.5) 0.60 (0.04 – 8.73) 0.71

Person who lived

in a nursing home

3/4 (75.0) 2/13 (15.4) 16.50 (1.09 – 250.18) 0.04

Person who works

with animals

0/3 (0.0) 4/14 (28.6) 0.00 (0.00 – undefined) 1.00

Visited

Hospital 1/3 (33.3) 15/17 (88.2) 0.07 (0.00 – 1.12) 0.06

Medical clinic 1/3 (33.3) 12/15 (80.0) 0.13 (0.01 – 1.89) 0.13

Aged care facility 3/4 (75.0) 3/12 (25.0) 9.00 (0.66 – 122.79) 0.10

Childcare facility 1/3 (33.3) 0/12 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Pre-school 1/3 (33.3) 0/12 (0.0) ˃ 1000 (0.00 – undefined) 1.00

Veterinary clinic 0/3 (0.0) 1/13 (7.7) 0.00 (0.00 – undefined) 1.00

Zoo 0/3 (0.0) 0/12 (0.0) Nil Nil

Farm 0/3 (0.0) 2/12 (16.7) 0.00 (0.00 – undefined) 1.00

OR, odd ratio; CI, confidence interval


139

Figure 6.4 Heatmap of pairwise core genome SNV differences between 11 C. difficile RT 012 isolates, sorted by origin: red, food from WA (n =

1); blue, human from Australia (n = 4) and green, human from Asia (n = 6).


140

6.2.2.2.3. In silico AMR profiling

Sequenced C. difficile RT 012 were surveyed for the presence of AMR genes. SRST2

identified AMR genes in 72.7% (n = 8) of the isolates. AMR genes encoding resistance

to tetracycline (tetM), aminoglycosides (Ant6-Ia, Aac6-Aph2, Sat4A and Aph3-III), and

macrolides-lincosamide-streptogramins (ermB) were detected in a food isolate from

WA (FI032). Of the four human isolates from Australia, only WA3835 carried AMR

genes (ermB, tetM and Aac6-Aph2). The human isolates from China, Hong Kong, Korea

and Thailand (OT0405, OT0066, OT0184 and OT0171, respectively) carried genes

encoding resistance to aminoglycosides (Aac6-Aph2) and macrolides-lincosamide-

streptogramins (ermB). Meanwhile, the Singapore (OT 0104) and Taiwan (OT0088)

isolates were carrying ermB, tetM and Aac6-Aph2 genes.

6.2.3. Discussion

To identify the risk factors for the predominantly CA infection with C. difficile RT 012

in WA, a retrospective case-control study was conducted. In total, 52 RT 012 cases

were identified between January 2014 and March 2015, and were matched with 103

controls who had CDI caused by other RTs during the same period of time. Suspecting

that RT 012 might have a foodborne transmission following its isolation from organic

beetroot (Chapter 3), a questionnaire was designed to compare differences in exposure

to certain foods and environmental factors. While the response rate was low (14.8%),

consuming raw non-organic home-grown vegetables and having contact with an

individual that was living in a nursing home were significantly associated with RT 012

infection (p = 0.04 for both). In addition, a number of other variables had elevated odds

ratios with p-values approaching significance (p < 0.05). These included purchasing

vegetables from a non-organic shop (OR 17.00, 95% CI 1.02 – 283.01, p = 0.05),

consuming raw non-organic spring onion (OR 15.00, 95% CI 0.98 – 228.90), p = 0.05)

and performing gardening work (OR 9.00, 95% CI 0.72 – 113.02, p = 0.09). It is

possible that the retail non-organic vegetables (including spring onions) might be

contaminated with C. difficile RT 012, as there was a high prevalence of C. difficile

from retail vegetables available in WA (Chapter 3). In fact, 20.0% of spring onions in

WA tested positive for C. difficile, although none of the strains isolated was RT 012

(Chapter 3). With consuming home-grown vegetables significantly associated with

CDI with RT 012, it was not unexpected to note that RT 012 cases were more likely to

perform gardening work compared to other CDI cases (75.0% vs. 25.0%), although this

differences was not statistically significant.


141

There were several limitations to our case-control study. Despite sending out three

rounds of the questionnaire, the response rate for our study was still too low to detect

any potential risk factors for C. difficile RT 012. Furthermore, a retrospective case-

control study is prone to recall bias and some of the participants had to recall their

exposure to certain foods and environmental factors from up to 1 – 2 years previously.

We chose to perform the case-control study retrospectively due to time limitations as

cases could be identified from pathology testing already conducted. Recently, new

studies have reported a high incidence of C. difficile RT 012 in Asian countries. In light

of this new information, future case-control study on RT 012 should include questions

regarding travel history and consumption of foods imported from Asian countries prior

to CDI.

Following the commencement of this case-control study, C. difficile RT 012 was

isolated from WA compost products and lawn samples (Chapter 4). However, with an

overall prevalence of just 1.1% (3/274) among Australian foods and environmental

isolates and a complete absence in Australian production animals121, 122, 154, 221, 256, local

establishment and wild-spread dissemination of C. difficile RT 012 in the community

seems unlikely.

While the prevalence of C. difficile RT 012 across Europe is relatively low (between

2.0% – 4.0%)82, 83, 320, studies since 2016 have shown that RT 012 is among the most

prevalent RTs in Asia, accounting for nearly 20.0% of all CDI cases. RT 012 was the

first and third most common RT in Beijing and Changsha with a prevalence of 14.0%

(3/21) and 16.4% (19/116), respectively229, 321. In a multicentre study that involved

hospitals across northern, eastern and southern China, C. difficile RT 012 was the

second most common RT (15.7%, 28/178)230. In Hong Kong, it represented 9.2%

(26/284) of all isolated C. difficile strains and was the fourth most common RT in

humans231. It was the second most prevalent disease-causing RT in Singapore (18.0%,

11/61)322. In 2017, C. difficile RT 012 was reported as the most prevalent RT in CA-

CDI cases in South Korea, accounting for 18.3% (11/60) of CA-CDI and 4.9% (22/445)

of HA-CDI cases (p = 0.001)95. In 2018, C. difficile RT 012 was reported as a

community-associated RT in southwest China, more commonly found in adults than in

children118. In another recent prevalence study of C. difficile in the Asia Pacific region,

RT 012 was among the top five most common RTs in circulation (16.7% in Hong Kong,

13.6% in Singapore, 11.4% in China, 6.8% in Taiwan and 5.3% in Thailand)213.


142

Knowing what we know now (an absence of C. difficile RT 012 in local production

animals and a low prevalence in meat, vegetables and environmental samples), it is

possible that C. difficile RT 012 cases may be acquired from Asia where the prevalence

is high, either in the form of returned travellers and/or local foodborne transmission

through seafood imported from Asia. Australia is a country that produces the majority

of its retail meat and vegetables. However, Australia imports a large quantity of low-

priced seafood from Asia (especially from China, Vietnam, Malaysia and Thailand)323.

Currently, little is known about the prevalence of C. difficile in seafood available in

WA. In 2015, Putsathit et al. investigated the prevalence of C. difficile in 97 prawns,

seven mussels and two clams from Thailand, available at retail stores in WA; all of

which were purchased peeled, cooked and frozen324. No C. difficile was isolated from

any of the samples, possibly due to the processing procedure that the seafood had

already received prior to packaging. Future studies could target unprocessed raw

seafood imported from Asia.

The SNV analysis (Figure 6.4) of 11 isolates lends some support to the hypothesis that

C. difficile RT 012 may have been acquired from Asia. A human isolate (WA3835) of

from a WA HAI-HCFO case had a clonal relationship with a Thai (OT0171) clinical

isolate, suggesting international transmission. WA3835 was also more closely related to

Asian isolates (average 10.3 SNVs differences) than any other WA isolates (average 25

SNVs differences). Unfortunately, information on the patient’s travel history and recent

consumption of imported seafood (if any) was not collected. All three human-WA

isolates were within 16 – 26 SNVs of each other. Notably, the WA-grown beetroot

(FI032) isolate was very distantly related to all sequenced human WA isolates with an

average of 872 SNVs differences in their core genome. The two C. difficile RT 012

isolates from compost products and lawn samples (Chapter 4) were not sequenced, as

they were not available at the time of sequencing.

In this study, multiple AMR genes were detected in 72.7% of all C. difficile RT 012

consistent with reports from Slovenia325, China229, Hong Kong231 and Europe253. High

phenotypic resistance against clindamycin, erythromycin, rifampicin and tetracycline

was observed in 60% of C. difficile RT 012 in Slovenia325. While being responsible for

the most CDI cases in a Beijing hospital (16.4%), all of their C. difficile RT 012 isolates

(n = 19) also displayed co-resistance to clindamycin and erythromycin229. Similar to the

study in Beijing, 96% of C. difficile RT 012 in Hong Kong were resistant to both

clindamycin and erythromycin231. Meanwhile, in Europe, RT 012 was among the top


143

three most common RTs displaying MDR, defined as resistance to at least three classes

of antimicrobial253.

Ongoing surveillance of CDI in humans is in place in Australia, but identifying risk

factors can be challenging as exposure to C. difficile does not necessarily lead to CDI

and infective dose is largely depending on host susceptibility272. Furthermore, CDI is

not a notifiable disease in Australia; thus, making it more difficult to investigate the

source of infection, monitor trends and control the spread of CDI in the community. In

this study, the number of RT 012 isolates sequenced (n = 11) was low. Further

sequencing, including of other human, animal, food and environmental isolates from

Australia and Asia, could help to identify relationships needed to better understand the

transmission pathways of C. difficile. Knowing how C. difficile strains disseminate both

nationally and internationally across country borders could help develop strategies to

reduce the risk of infection.

6.3 Chapter summary

In this chapter, novel insights on the transmission routes of C. difficile RTs 056 and 012

in WA were explored using WGS and a case-control study. RT 056 is well-established

in Australian cattle and human populations and widespread in the community through

contaminated food and the environment. For the first time, multiple clonal populations

of human and food/compost RT 056 strains were seen suggesting a very recent shared

ancestry, consistent with recent transmission or lack of strain diversity. Moreover, these

findings provide some support for the hypothesis of foodborne and environmental

transmission for RT 056 in WA, however, further epidemiological support is needed.

While food and environmental C. difficile RT 056 isolates were identified as being

closely related to local clinical RT 056 isolates, the analysis of C. difficile RT 012

strains revealed greater diversity locally, and a close relationship to a strain from an

Asian case. RT 012 is abundant in Asia, has not been reported from any Australian

production animals, and was rare in WA food and environmental samples. The clonal

relationship between a human WA and Thai isolate by WGS suggests that C. difficile

crosses geographical barriers, conceivably through international travel and trade

between Asia and Australia. Our understanding of CDI transmission, particularly in

food and the environmental setting, is still in its infancy, especially in developing

nations. Ongoing molecular typing and surveillance in humans, animals, foods and the

environment are needed to develop disease prevention and management strategies.


144

WGS will continue to provide the discriminatory power that conventional typing

methods lack. Tackling the increased incidence of CDI will require a “One Health”

approach where collaborative efforts from researchers, doctors, veterinarians, industry

partners and government bodies are needed.

Chapter 7: Key findings and general discussion

145

CHAPTER 7

KEY FINDINGS AND GENERAL

DISCUSSION


146

7.1 INTRODUCTION

C. difficile is an opportunistic pathogen of human and animals. Over recent decades, the

global incidence and severity of CDI have grown, resulting in a significant burden to the

global healthcare system both financial and non-financial. In addition, the previously

rare CA-CDI has become common and currently contributes to ~35% of all CDI cases3-

5.

In Europe and North America, the isolation of genetically indistinguishable strains of

C. difficile from humans, production animals, food and the environment has fuelled

speculation that CDI, especially CA-CDI, may have a zoonotic, foodborne or

environmental aetiology66, 132, 147, 214, 289. In Australia, while production animals have

been identified as an important amplification reservoir for clinically important

C. difficile strains121, 122, 154, nothing is known about C. difficile in foods and the

environment.

In this PhD project, the prevalence and molecular epidemiology of C. difficile in retail

foods and the environment of WA external to hospitals were investigated and,

subsequently, identified as a source of clinically important C. difficile strains that were

also prevalent in Australian production animals. The research objectives were:

(i) to determine the prevalence and genotypes of C. difficile in the retail foods

and environment of WA;

(ii) to examine the susceptibility of food and/or environmental C. difficile

isolates to antimicrobial agents and food preservatives;

(iii) to define the extent of genetic overlap and potential transmission between

human, food and environmental C. difficile population;

(iv) to investigate possible risk factors for predominantly community-associated

CDI and

(v) to explore the relatedness of C. difficile from Australia to those that

originated in Asia.

7.2 KEY FINDINGS AND GENERAL DISCUSSION

In Chapter 3, the molecular epidemiology of C. difficile on retail vegetables and meats

was investigated in WA. Overall, C. difficile was isolated from 14.6% of the vegetables

[55.6% (15/27) of organic potatoes, 50.0% (9/18) of non-organic potatoes, 22.2% (4/18)

of organic beetroots, 20.0% (6/30) of spring onions, 13.3% (4/30) of English spinach,


147

5.6% (1/18) of organic onions, 5.3% (1/19) of organic carrots, 3.3% (1/30) of coriander,

0.0% (0/30) of asparagus, 0.0% (0/30) of garlic and 0.0% (0/30) of pak choy], while

none of the meats [32 salami, 4 uncooked 18-to-20 week old veal calf meat and 1

uncooked baby goat meat] tested positive for C. difficile. Compared to vegetables

cultivated above ground (spring onions, English spinach, coriander, asparagus and pak

choy), a significantly greater prevalence of C. difficile was found on root vegetables

(organic potatoes, non-organic potatoes, organic beetroots, organic onions, organic

carrots and garlic) (7.9% vs. 30.0%, p = 0.0002). A total of 41/280 vegetable samples

tested positive for C. difficile by enrichment culture, 22 samples were contaminated

with toxigenic C. difficile strains: 10 organic potatoes, 7 non-organic potatoes, 2 organic

beetroots, 1 organic carrot, 1 English spinach and 1 spring onion. The proportions of

toxigenic C. difficile strains on organic potatoes, non-organic potatoes, organic

beetroots, organic carrot, English spinach and spring onions were 37.0%, 38.9%, 11.1%,

5.3%, 3.3% and 3.3%, respectively. From a public health standpoint, this finding

indicates that vegetables destined for human consumption were contaminated with

C. difficile capable of producing toxins, and therefore disease. However, both organic

and non-organic potatoes sampled in this study were covered in soil and washing with

water in a household kitchen would be essential prior to cooking. This food preparation

process is likely to reduce the number of C. difficile spores on the surface of the

potatoes as a previous study has reported that cleaning a lettuce with water resulted in a

0.7 log reduction in bacteria attached to the surface of the vegetable326. C. difficile is

common in soil and likely attached to the soil particles143-145, 147, 148, so it stands to

reason that the removal of soil would greatly reduce the number C. difficile on the

surface of the potatoes. Furthermore, potatoes are commonly baked at approximately

220 °C for 45 – 60 min or fried at 190 °C for 5 – 6 min; well above the 96 °C for 1 – 2

min that resulted in a large 6 log reduction in C. difficile spores as reported by

Rodriguez-Palacios et al. in 2011327. Taken together, the risk of a consumer ingesting

high numbers of viable C. difficile on potatoes is likely to be low. Nevertheless,

contamination of household kitchens is likely to occur while washing contaminated

vegetables. If so, whether household contamination could result in CDI is not known as

this also depends heavily on host factors, however, this is certainly possible in the same

way that contamination of hospitals is an important consideration in HA-CDI328, 329.

Apart from potatoes, all other vegetables that tested positive for C. difficile in this

project (beetroots, carrots, English spinach and spring onions) are frequently consumed

raw with minimal processing and, thus, could pose a risk of infection to the consumer.


148

The strain population of C. difficile on WA vegetables was predominated by RTs 051

(A-B-CDT-), QX 145 (A-B-CDT-), 056 (A+B+CDT-), QX 393 (A-B-CDT-) and

014/020 (A+B+CDT-), comprising 39.3% of the isolates. The finding of RTs 014/020

and 056 in retail vegetables is significant as they are among the most common RTs in

causing CDI in both humans and production animals in Australia.

Chapter 4 examined the prevalence and genetic diversity of C. difficile in the

environment of WA, specifically in compost and lawn samples. The prevalence of

C. difficile in compost was 60.0% in samples still undergoing composting/digesting

[100.0% (3/3) of digestate liquid, 75.0% (3/4) of human biosolids and 0.0% (3/3) of

pre-digestate liquid] and 22.5% in the final composted product [29.7% (11/37) of soil

conditioners, 16.7% (2/12) of mulches and 13.6% (3/22) of garden mixes]. This finding

was not surprising as Xu et al. demonstrated that windrow composting, a process used

by Australian compost manufacturing companies, resulted in a 3.4 log reduction in

C. difficile spores with the largest reduction occurring during the curing phase of

composting prior to maturation of products264. Overall, a total of 22/81 samples tested

positive for C. difficile, only 10 samples were contaminated with toxigenic strains: 3

digestate liquid, 1 human biosolid, 5 soil conditioner and 1 garden mixes. Soil

conditioner, garden mixes and mulches (final composted products) had a 13.5% (5/37),

4.5% (1/22) and 0.0% (0/12) prevalence of toxigenic C. difficile strains, respectively.

Importantly, these final composted products sampled from the suppliers in the Perth

metropolitan area are being sold to the public. Thus, the level of contamination at this

stage is crucial and reflects a risk of dissemination of C. difficile in the community. RTs

014/020 (A+B+CDT-) and 056 (A+B+CDT-) were the predominating toxigenic strains

in the final products, and both RTs are a common cause of HA- and CA- CDI in

Australia203, 204, 206-210.

The overall prevalence of C. difficile in WA lawn was high (58.5%) with a load of up to

1200 cfu/g of lawn. Newly laid roll-out lawn less than 4 months old had a significantly

higher prevalence of C. difficile compared to older established lawn (65.2% vs 46.9%, p

= 0.015). A total of 182/311 lawn samples from 10 postcodes tested positive for

C. difficile; 47.3% (86/182) of which were contaminated with toxigenic C. difficile

strains. All toxigenic strains were positive for toxin genes tcdA and tcdB, but not binary

toxin genes (cdtA and cdtB; A+B+CDT-). Clinically important RT 014/020 was the

most prevalent (39.0%, 71/182) and widely distributed RT as it predominated in 9/10 of

the sampled postcodes (Figure 4.4), ranging from 30 km north to 80 km south of Perth


149

City. The widespread contamination of lawn by RT 014/020 suggested a possible

porcine origin as RT 014 is the most prevalent RT in Australian pigs (23.4%) and pig

manure was known to have been applied directly as fertiliser in the production of the

roll-out lawn154. Roll-out lawn is popular in Australia as it establishes a lot faster than

growing grass by seed. New housing and suburb development over the past decades in

WA has seen increased popularity for roll-out lawn for parks and residential lawn as it

did in the USA266. Coincidentally, the majority of CA-CDI cases in WA between 2010

and 2014 reside in newly developed suburbs in the outskirt of Perth City (Figure 4.5)

(L Bloomfield and TV Riley, unpublished data). It is not known if those cases were

acquired from contaminated lawn, however, this is being investigated further. In 2017,

Knight et al. identified clonal C. difficile RT 014 of human and porcine origins in

Australia, consistent with a recent inter-species transmission214. Interestingly, some of

these closely related clonal strains were separated by thousands of kilometres across

Australia with no evidence of prior contact. This suggests a persistent community

reservoir of RT 014 with the potential for long-range dissemination. The use of roll-out

lawn contaminated with RT 014 of porcine origin could be a mechanism contributing to

the widespread dissemination of RT 014 in the community and clonal CDI cases with

no apparent epidemiological link. Future genomic studies are necessary to determine the

extent of genetic overlap between RT 014 from humans, pigs and lawn. Despite a large

number of samples of lawn (n = 311) being collected, a limitation of this study was that

a relatively small number of public parks (n = 20) was sampled with most being located

in the metropolitan area of Perth. Further, more expansive, studies are required.

Overall, a diverse population of C. difficile was isolated from the retail vegetables and

environmental samples of WA. However, RTs 027 and 078 were not found and CDT+

strains only accounted for 3.7% of the isolates. The apparent absence of RTs 027 and

078 was not surprising as RT 027 has failed to establish in Australia following the first

report of a case in 2009198 and, to date, RT 078 has not been isolated from Australian

livestock121, 122, 154 and is uncommon in local human cases203, 204, 206-210. The small

number of human RT 078 cases that have been reported in Australia203, 204, 209, 210 can be

explained by possible importation of cases. The possibility also exists that

contaminated pork has been brought into the country. Although all fresh pork sold in

Australia is locally produced, virtually all processed pork products sold in Australia

(ham, bacon, salami, etc.) are made from imported pork which may be contaminated. As

described in Chapter 5, and reiterated below, preservatives used in such products have


150

no impact on C. difficile. The extent of overlap in C. difficile RTs in Australia between

humans, production animals, food, compost and lawn is presented in Figure 7.1. RTs

014/020 and 056 predominated in humans, production animals and retail foods, as well

as the environment of WA. In WA, RT 014/020 accounted for a median of 36% of all

CDI cases (range 31% – 41%) (Figure 7.1A1 – A4) and, nationally, a median of 31% of

all CDI cases (range 29% – 35%) (Figure 7.1A5 – A8). Previous surveillance studies

also identified RT 014 as the most common strain in Australian neonatal pigs,

accounting for 23% of all isolates (Figure 7.1D). This PhD study provides a first and

valuable insight into the prevalence and molecular epidemiology of C. difficile in retail

foods and the environment of WA, with RT 014/020 identified as the most prevalent RT

among retail vegetables and the environment [5.4% on vegetables (Figure 7.1E), 11.1%

in compost (Figure 7.1F) and 39.0% in lawn (Figure 7.1G)].

It is possible that one of the reasons for such a high prevalence of C. difficile RT

014/020 in humans, pigs, food and the environment is the poor resolution of

conventional PCR ribotyping with this group51. In Austria, in 2008, 146 C. difficile

isolates belonging to 39 different RTs typed by agarose gel-based PCR ribotyping were

further characterised by Indra et al. using capillary gel electrophoresis-based PCR

ribotyping51. In this study, 24 isolates of RT 014 were further divided into 7 different

subgroups (sr014/0 to sr014/6). In Australia, RT 014 has been differentiated into 3 STs

using in silico MLST: ST 2, 13 and 49, and interspecies clustering has been reported

between human and porcine RT 014 ST lineages 2 and 13 by Knight et al. in 2017214.

RT 056 was another RT common in human, production animal, retail food and the

environment of WA (Figure 7.1). Between 2011 and 2015, RT 056 was frequently

among the top three most common RTs in human CDI (Figure 7.1A1 – A8), accounting

for a median of 6% of all CDI cases (ranges from 2% – 13%). It has an 8% and 7%

prevalence in Australian cows and calves, respectively, at slaughter (Figure 7.1B) and

in Australian sheep and lambs (Figure 7.1C), respectively121, 122. On retail vegetables

(Figure 7.1E) and compost (Figure 7.1F) of WA, RT 056 has a prevalence of 7% and

6%, respectively. Other RTs isolated from food and/or the environment of WA with a

known history of causing CDI in human and production animal include RTs 033 (A-B-

CDT+) and 237 (A-B+CDT+)278, 330. Reports of RT 033 strains in humans are

uncommon but not unprecedented in Australia330. In fact, the prevalence of RT 033 in

humans is likely underreported as A-B-CDT+ strains of C. difficile are not

detected/reported by most diagnostic methods, such as the illumigene C. difficile


151


152

Figure 7.1 C. difficile RT diversity in WA (predominantly), including amongst (A) humans, (B) cattle and calves, (C) sheep and lamb, (D) piglets,

(E) vegetables, (F) compost and (G) lawn. Data on humans are from eight separate studies: (A1) A WA surveillance study from October 2011 –

September 2012 (747 isolates)206, (A2) a cross-sectional study of CDI from December 2011 – May 2012 in two tertiary hospitals in WA (70

isolates)207, (A3) a study that investigated the epidemiology of CA-CDI from July 2013 – June 2014 across six emergency departments of WA

hospitals (27 isolates)204, (A4) a WA study in July 2014 where all faecal samples submitted for enteric testing were surveyed for C. difficile (59

isolates)208, (A5) a month-long national study in 2010 (330 isolates)209, a study that compared the circulating RTs of C. difficile among (A6) HA-CDI

(96 isolates) and (A7) CA-CDI (152 isolates) cases in two Australian States over a 3-year period (2012 – 2014)203, and (A8) a national study where 151

isolates from August – September 2014, February – March 2015 and August – September 2015 were randomly selected for ribotyping210. Data on

production animals are from three separate studies: (B) a cross-sectional study of cows and calves at slaughter in 2013 (209 isolates)122, (C) a national

ovine study in 2011 (15 isolates)121, and (D) a national porcine surveillance study in 2014 (154 isolates)154. Data on (E) vegetables, (F) compost and

(G) lawn are from this thesis (Chapters 3 and 4). RTs are colour-coded and prefixed with either the official CDRN number or internal nomenclature

(QX).


153

assay (Meridian Bioscience, Inc.) and Xpert C. difficile/Epi assay (Cepheid)254. In

animals, RT 033 was the second most prevalent RTs in both calves (20%) (Figure

7.1B) and neonatal pigs (13%) (Figure 7.1D)122, 154. Meanwhile, RT 237 is a pig strain

unique to WA piggeries154, 179 and, until 2012, it had not been reported in humans255.

Overall, in WA, clinically important C. difficile RTs with clear epidemiological links to

production animals have been isolated from retail vegetables and the environment.

While clinical and animal isolates are occasionally surveyed for antimicrobial

susceptibility, less is known about the AMR profile of food and environmental

C. difficile isolates. Thus, the antimicrobial susceptibility of food and environmental

C. difficile isolates in WA was explored in Chapter 5.1. Compared to isolates from

food and lawn, a higher proportion of compost isolates displayed resistance to

erythromycin (< 2.0% vs 19.4%) and tetracycline (< 2.0% vs 13.9%) (Table 5.1). These

compost isolates were likely of animal origin due to the common practice of composting

animal manure for agricultural use, and the acquisition of AMR genes is undoubtedly

driven by the heavy use of macrolides and tetracycline-based antimicrobials in the

Australian livestock industry (112.2 tonnes per year)284. The differences in AMR

profiles between vegetables, compost and lawn could be due to (i) different sources of

animal manure were used as fertiliser in the vegetable and lawn farms, and/or (ii) the

loss of resistance genes driven by differences in antimicrobial selective pressure

between animal facilities and the environment. The AMR profiles of compost isolates

were in agreement with a previous study on Australian porcine C. difficile RT 014

which showed high (69%) non-susceptibility to clindamycin, erythromycin and

tetracycline, and 100% susceptibility to fidaxomicin, vancomycin, metronidazole,

rifaximin, amoxicillin/clavulanate, moxifloxacin and meropenem214. Meanwhile,

C. difficile of human origin in Australia has previously been reported with a high

(84.3%) clindamycin resistance rate210. In this thesis, clindamycin resistance was

observed among both compost (30.6%) and lawn (42.3%) isolates with MICs

comparable to humans. Unfortunately, no information was available on clindamycin

susceptibility of C. difficile of animal origin in Australia as it was not tested in previous

animal studies. In addition to the isolation of indistinguishable RTs, the finding of

comparable AMR profiles between food and environmental C. difficile with human

and/or animal isolates further supports the spilling-over of C. difficile between

reservoirs, potentially from animals to vegetable and the environment via the

agricultural recycling of animal waste.


154

In Australia, suckling pigs (2 – 6 weeks old) and male neonatal dairy calves, also known

as bobby calves, (< 7-day old) are the only neonatal animals slaughter for meat290, 331.

Suckling pigs are relatively rarely eaten in Australia, prepared for special occasions and

gatherings, with a low demand from the consumer market. Thus, there is likely to be

limited exposure of contaminated suckling pig meat to the domestic consumer.

Meanwhile, male dairy calves are considered surplus to the dairy industry and are a by-

product of the process to ensure continual milk supply from its mother 261, 290. Between

2014 and 2015, approximately 630,000 calves were slaughtered in Australia, producing

37,800 tonnes of veal (carcase weight)331. In Australia, 70% of the total beef and veal

production (1.5 million tonnes) was exported to Asia, North America, Europe, the

Middle East and Africa331. Of the beef and veal consumed domestically (640,000

tonnes), bobby calf veal likely represents a small percentage of the total veal

production331. To distance themselves from the controversial welfare issues of

slaughtering newborn calves, raw bobby calf veal is not available in any of the

Australian major supermarkets, but it is used to make RTE meat products. In Australia,

RTE meat products are either cooked at a temperature incapable of killing C. difficile

spores (65 °C for 10 mins) or preserved using non-heat treatment methods such as

curing257. Given C. difficile prevalence is at its highest during the neonatal period122, 256,

the slaughter age of dairy calves represents a unique risk factor for the contamination of

carcasses and meat products. C. difficile has been detected on 25% of carcasses of

Australian neonatal veal calves, likely a result of contamination from faecal and

gastrointestinal content during the slaughtering process221. In Chapter 5.2, the

susceptibility of C. difficile to food preservatives commonly used in RTE meats was

investigated. In turn, the MICs of sodium nitrite (250 mg/L), sodium nitrate (> 4000

mg/L) and sodium metabisulphite (1000 mg/L) were higher than the maximum

permitted level allowed in processed meats and no bactericidal activity was observed.

C. difficile ability to survive in the presence of food preservatives coupled with its

remarkable resilience to extreme temperatures (-80 °C to 71 °C) creates the need to

investigate the prevalence of C. difficile in RTE meats259, 332. In WA, C. difficile has not

been recovered from retail ground meat from adult animals (NF Foster and TV Riley,

unpublished data). In Chapter 3, a small sample of 32 RTE meats was purchased from

major supermarkets and local butchers in WA. The majority of these products was

produced in WA. Although the sample size was small, it was larger than a study in the

USA that reported C. difficile in 47.8% (11/23) of RTE meats134 and comparable to

another study in Taiwan that reported C. difficile on 25.7% (9/35) of pork skin and


155

20.0% (6/30) of pork colon164. The culturing protocols used were similar with

enrichment broth supplemented with taurocholate to encourage spore germination134, 164.

Perhaps due to the low proportion of Australian dairy farms in WA (2.5%)333 and the

fact that major Australian supermarkets do not sell products made from bobby veal

calves, true bobby calf veal could not be sourced in any of the butcher shops (many of

which make their own RTE meats) and supermarkets in WA. With 68% of Australian

dairy farms located in VIC, QLD and southern NSW, sampling bobby veal and RTE

meats made in those states may have a higher chance of detecting C. difficile333.

As previously mentioned, C. difficile RT 056 is one of the most prevalent CDI-causing

RTs in Australia, accounting up to 13%, 8% and 7% of CDI in human203, 204, 206-210,

cattle and calves122, and sheep and lamb121, respectively. It was also the second most

prevalent toxigenic C. difficile strain in the food and environmental samples in WA,

second to RT 014 (Chapters 3 and 4). The presence of RT 056 in humans, animals,

food and the environment raised the possibility that CDI may have a zoonotic,

foodborne or environmental aetiology. To date, the assessment of genetic overlap for

RT 056 in Australia has been limited to low resolution genotyping tools such as PCR

ribotyping. In Chapter 6.1, to detect evidence of potential transmission, WGS and

high-resolution core genome phylogenetics were performed on a collection of 29

C. difficile RT 056 of human (n = 21), food (n = 4) and environmental (n = 4) origins in

WA. All isolates displayed allelic conservation in the seven housekeeping genes (adk1,

atpA5, dxr7, glyA1, sodA3 and tpi1, belonging to ST 34. This phylogeny based on non-

recombinant core genome SNVs provided ultra-fine scale resolution on the RT 056

population and found clustering of human, food and/or environmental strains indicative

of a recent shared ancestry. SNV analysis identified three distinct CGs that comprised

34.5% (10/29) of the sequenced strains. Two CGs, CGs II and III, encompassed clones

from human, food and/or environmental origins (CG II: 2 human and 1 food clones;

CGIII: 2 human, 1 food and 2 environmental clones) (Figure 6.1). Overall, 14% (3/21)

of the sequenced human strains had a clonal relationship of ≤ 2 SNVs difference with

one or more food or environmental strains. Half (2/4) of the human isolates in CGs II

and III were classified as CA-CDI cases. All genetically closely-related food and

environmental clones in CGs II and III were isolated 2 – 3 years after their

corresponding human clones. This could demonstrate directionality in which the food

and environment were contaminated by RT 056 from humans, possibly because of

agricultural recycling of composted human biosolids. But, it is important to remember


156

that food and environmental C. difficile are widely believed to be in spore form, during

which evolution is effectively suspended in time124. Therefore, food and environmental

C. difficile may have a much slower molecular clock than vegetative C. difficile such as

those actively causing diseases in symptomatic patients. To date, no study has

investigated the molecular clock of food and environmental C. difficile spores. Overall,

these data suggest that over an extended period of time there has been a long-range

transmission of C. difficile RT 056 in WA between humans, food and the environment,

however, a common source cannot be ruled out. The food and environmental sampling

studies in Chapters 3 and 4 occurred while our research group was undertaking a

C. difficile surveillance study that cultured and typed all hospital-identified C. difficile

in WA. Thus, laboratory contamination of food and environmental samples with RT

056 from humans was possible but this was unlikely because (i) the corresponding

human clones in CGs II and III were cultured and typed up to 3 years prior to the

commencement of the food and environmental sampling studies, and (ii) a diverse range

of food and environmental C. difficile strains was isolated (70 different RTs), including

22 (31.4%) RTs that were novel to our research group the majority of which (81.8%,

18/22) were non-toxigenic.

In terms of CDI transmission in a hospital setting, there is increasing evidence that

suggests importation of strains from the community. In the UK, 45% of C. difficile

cases in hospital were genetically distinct with > 10 SNVs from all previous cases,

indicating diverse sources other than symptomatic patients may be playing a major part

in C. difficile transmission, possibly asymptomatic carriers and/or reservoirs outside of

the hospital64. In Australia, mathematical modelling suggested that HA-CDIs are

sustained by the importation of C. difficile from the community334. In WA, in 2017, 8%

(104/1380) of asymptomatic patients in hospitals were colonised with C. difficile, and

toxigenic C. difficile strains were isolated from 6% (76/1380) of asymptomatic patients

with RT 014/020, 018 and 056 being the most common200. All three RTs have been

known to cause CDI in humans200, 203, 204, 206-210. Furthermore, 59% of environmental

samples from the immediate outdoor environment of WA hospitals were recently found

to be contaminated with C. difficile (S Perumalsamy and TV Riley, unpublished data).

Asymptomatic patients and shoes of healthcare staff/visitors could be carrying

C. difficile into the hospitals but this is yet to be investigated. In this thesis, the HA-CDI

cases in CGs II and III were clonal (1 – 2 SNVs) with at least one food and/or

environmental C. difficile isolates. Hence, food and environmental sources may play a


157

role in (i) the importation of C. difficile into a hospital setting and/or (ii) acquisition of

C. difficile in the community prior to hospitalisation.

Since 2013, WA has experienced an increase in CDI caused by RT 012, predominantly

community-associated (L Bloomfield and TV Riley, unpublished data). In Chapter 6.2,

a retrospective case-control study was performed to determine the risk factors of RT

012 CDI in WA. Suspecting community reservoirs might have played a role in the

transmission and dissemination of RT 012, a survey was designed to ascertain the

association of CDI by RT 012 with exposure to a variety of food, environment, animal

and human factors. Unfortunately, out of the 155 potential participants (52 cases and

103 controls), only 23 (14.8%) individuals responded to the survey. By univariate

analysis, eating raw non-organic home-grown vegetables and having contact with an

individual living in a nursing home were associated with CDI by RT 012 (OR 16.50,

95% CI 1.09 – 250.18 for both). Due to the small number of enrolled cases and the

broad confidence intervals for the only two significant variables, multivariate analysis

was not performed.

Following the commencement of this case-control study, C. difficile RT 012 was

subsequently isolated from compost (n = 1) and lawn (n = 1) in WA. With an overall

prevalence of just 1.1% (3/274) among WA food and environmental isolates (Chapters

3 and 4) and a complete absence in Australian production animals121, 122, 154, the local

establishment of RT 012 in WA community was thought to be unlikely. However, a

2015 prevalence study in the Asia-Pacific region by Collins et al. identified RT 012 as

one of the most prevalent RTs causing CDI in Asia (16.7% in Hong Kong, 13.6% in

Singapore, 11.4% in China, 6.8% in Taiwan and 5.3% in Thailand)213. In an exploratory

study to examine the relatedness of RT 012 from Australia with those from Asia, WGS

was performed. A small selection of C. difficile RT 012 isolates from WA organic

beetroot (n = 1), Australian clinical samples (n = 4) and Asian clinical samples (n = 6)

was sequenced. The two C. difficile RT 012 isolates from compost and lawn samples

(Chapter 4) were not sequenced, as they were yet to be isolated at the time of

sequencing. All 11 isolates displayed allelic conservation in the seven housekeeping

genes (adk1, atpA4, dxr7, glyA1, sodA3 and tpi3), belonging to ST 54. A phylogeny

based on non-recombinant core genome SNV analysis revealed a clonal relationship

between a WA human strain (WA3835) and a human strain from Thailand (OT0171),

with only 2 SNVs differences in their core genome (Figure 6.4). Interestingly, WA3835

was more closely related to Asian isolates (median of 6.5 SNVs differences) than any


158

other WA isolates (median of 25 SNVs differences) and was classified as HAI-HCFO.

Notably, the RT 012 from WA beetroot was very distantly related to all other sequenced

isolates, with a median SNVs difference of 869. AMR genes were identified in 72.7%

(8/11) of the isolates, 50% of which were MDR with AMR genes for tetracycline,

aminoglycosides and macrolides-lincosamide-streptogramins. This is consistent with

findings from Europe and Asia where RT 012 with MDR traits is increasingly being

reported229, 231, 253, 325. In 2016, there were other reports of RT 012 as one of the most

prevalent RTs causing human CDI in Asia (16.4% in Beijing229, 14.0% in Changsha321,

15.7% in northern, eastern and southern China230 and 9.2% in Hong Kong231). In 2017,

it was identified as a predominantly CA RT in South Korea, accounting for 18.3% of

CA-CDI vs 4.9% of HA-CDI cases95. In 2018, RT 012 was also identified as a CA RT

in southwest China and, for an unknown reason, more commonly isolated in adults than

children118. Collectively, these studies suggest that the C. difficile RT 012 in WA may

have originated in Asia where the prevalence is high, and got to WA either via returned

travellers and/or through food imported from Asia. For future case-control studies

aiming to determine the risk factors for CDI caused by RT 012, questions regarding

recent travel history and consumption of any food of Asian origins should be included.

In the case of C. difficile transmission in Australia, there is growing evidence that

challenges the long-standing notion that CDI is a healthcare-acquired infection. First,

young Australian production animals are significant reservoirs of C. difficile (56%

prevalence in < 7-day old calves122, 7% prevalence in <1-year old lambs121 and 67%

prevalence in < 7-day old neonatal pigs154). Second, studies have shown an abundance

of C. difficile spores in treated biosolids (73.0%)197, animal effluent (100.0%)335 and

animal manure (median of 10.0% in bovines101, 103-111, 122, 153, 156-159, 170-177, 6.1% in

ovines121, 156, 159, 174, 178, 31.3% in porcines103, 104, 110, 112-117, 152-154, 156, 157, 160-165, 176, 179-189

and 5.8% in poultry128, 148, 156, 157, 166) which are recycled to agriculture either through

composting or direct fertilisation of crops and lawn turf121, 122, 154, 197, 262, 264. Third,

C. difficile has been isolated from Australian vegetables (14.6%), compost (27.2%) and

lawn (58.5%). Forth, genomic analysis reveals long-range interspecies transmission of

C. difficile between Australian pigs and humans with no epidemiological links,

suggesting a persistent community reservoir either food or environmental214. Fifth,

mathematical modelling has indicated that importation of C. difficile from the

community is the driving force of HA-CDI in Australia334. Lastly, the use of

antimicrobial agents as a disease preventive measure in Australian livestock and the


159

practice of using unpasteurised animal manure in farming are likely a driving force for

(i) C. difficile colonisation and disease onset in animals, (ii) amplification and

persistence of C. difficile in animal farms, (iii) the spilling-over of C. difficile to crops

and the environment and, subsequently, (iv) C. difficile colonisation and onset of CDI in

the community.

Numerous studies have reported the isolation of clinically relevant C. difficile strains

from production animals, food and the environment. This presented several hypotheses

regarding the transmission of C. difficile in the community, all involving zoonotic

transmission as suggested by several others272, 336. Scenarios for how this might occur in

Australia are shown in Figure 7.2. Production animals are recognised reservoirs of

C. difficile. Table 1.1 summarised the prevalence and predominant strains of C. difficile

circulating in the production animals of various countries, including Australia. The

overall reported prevalence of C. difficile was a median of 12.7% in calves, 6.2% in

adult cattle, 6.5% in lambs, 5.7% in sheep, 67.2% in piglets and 8.6% in pigs. Due to

lack of colonisation resistance, neonatal animals generally have a higher prevalence of

C. difficile that decreases as the animals age256. Several reports have demonstrated

shedding of C. difficile at slaughter and contamination of carcasses and meats due to gut

content spillage103, 122, 221. To date, the majority of studies that report C. difficile in foods

focus on meat with prevalence ranging from a median of 2% in Europe110, 127, 167 up to

42% in the North America134. Meanwhile, in WA, none of the adult meat (NF Foster

and TV Riley, unpublished data) or RTE meat (Chapter 3) samples tested positive for

C. difficile, despite using an enrichment protocol similar to the consensus method

proposed by Limbago et al. but with a longer incubation time140. The reason for this

disparity between regions is unknown. However, given most of the retail meats in

Australia are from adult animals which at the time of slaughter have a low prevalence of

C. difficile, the likelihood of contaminated meats playing a role in the transmission of

CDI in Australia is likely to be low. Nevertheless, genetically related C. difficile RT 014

of human and pig origins, indicative of recent shared ancestry, has been reported in

Australia by Knight et al., with no apparent epidemiological link214.

Manure from production animals is commonly used as organic fertiliser for crops. In

Australia, the application of manure to farmland is suspected to contaminate vegetables

with C. difficile of animal origin. C. difficile has been isolated on 14.6% of retail

vegetables available in WA (Chapter 3). Thus, minimally processed or uncooked


160


161

Figure 7.2 Potential sources of C. difficile implicated in the transmission of CDI in the community of WA. Young production animals are common

reservoirs for C. difficile. Manure from production animals can contaminate vegetables and the environment through direct application to vegetable and

lawn farms as fertiliser. Composted human biosolids, from sewage treatment plant, and animal manure can contaminate crops and the environment

when used in gardening and landscaping. Anaerobic digestion of food waste in tanks offers an ideal growing environment for C. difficile spores to

amplify and subsequently be re-introduced to composted products in liquid fertiliser. In addition, acquisition of foreign C. difficile strains can occur

during international travel.


162

vegetables could be a potential vehicle for CDI. In Australia, unpasteurized pig manure

is also used as fertiliser in the production of roll-out lawn. With 58.5% of public lawn in

WA contaminated with C. difficile, predominantly the most common RT in Australian

pigs (RT 014), environmental transmission of C. difficile from lawns to susceptible

hosts is possible (Chapter 4).

As an enteric pathogen, C. difficile is commonly encountered in raw sewage and studies

by Xu et al. and Romano et al. have demonstrated C. difficile is capable of surviving the

wastewater treatment process to be recovered in dewatered human biosolids and treated

wastewater197, 262. In Australia, human biosolids and animal manure are commercially

composted to produce gardening products. While composting has been shown by Xu et

al. to be an effective mean of reducing the load of C. difficile spores, complete

elimination was doubtful264. Furthermore, to be eco-friendly, food waste from

Australian supermarkets is recycled and digested anaerobically in tanks to produce

clean energy and liquid fertiliser. The anaerobic tanks are likely providing an optimal

conditions for the amplification of anaerobes, including C. difficile, either from

contaminated food waste and/or contaminated environment at the compost

manufacturing company. The subsequent addition of liquid fertiliser from the digestive

tanks, which have a 100% prevalence of C. difficile, to all composting piles is suspected

of re-introducing C. difficile into the compost products. In WA, 22.5% of the final

composted products tested positive for C. difficile. The sale and use of these

contaminated products in gardening and landscaping are undoubtedly contributing to

widespread environmental contamination in the community.

In addition to local acquisition, a susceptible host can also be colonised or infected with

C. difficile during international travel. An example of this happening was seen in 2009

when the first isolation of hypervirulent C. difficile RT 027 in Australia was reported in

a 43-year old woman who was infected while travelling in the USA and suffered a

recurrent infection upon her arrival back in Australia198. Furthermore, as previously

mentioned, the emergence of RT 012 in WA is likely to be of Asian origin, acquired

while travelling and/or consuming contaminated food from Asia (Chapter 6).

Nevertheless, while clinically important toxigenic C. difficile strains have been detected

in food and the environment in support of foodborne and environmental transmission of

C. difficile, currently, there are no documented outbreaks of CDI linked to contaminated

food or the environment. C. difficile has been found in humans, animals, food, soil,

rivers, lakes, compost, lawns and household environments. It is reasonable to suspect


163

that we are exposed to toxigenic C. difficile on a daily basis, but exposure to C. difficile

does not necessarily lead to CDI. The infectious dose of C. difficile in humans is still

unknown but it is agreed to vary greatly between healthy individuals and vulnerable

population with predisposing risk factors. A history of antimicrobial usage is the most

widely reported risk factor for CDI in humans, especially agents with activity against

commensal bowel microbiota such as clindamycin, aminopenicillins, extended-

spectrum cephalosporins and fluoroquinolones62. A disrupted intestinal microbiota

offers less colonisation resistance which upon ingestion of C. difficile spores could lead

to overgrowth of vegetative C. difficile in the colon. The use of proton pump inhibitors

reduces stomach acid which could facilitate the survival of spores, vegetative cells and

toxins337. There is also evidence that proton pump inhibitors disrupt the host immune

system that would normally reduce the risk of CDI338.

7.3 STUDY LIMITATIONS AND DIRECTION FOR FUTURE RESEARCH

There are limitations to this work that must be acknowledged.

In the vegetable sampling studies, it was impossible to acquire information on the farms

where the vegetables were grown. Information such as the location of farms, if animal

manure was used as fertiliser and, if so, from what type of animals and the level of

environmental contamination in said farms would be useful in providing additional

insight into the source of C. difficile contamination and the relative risk of particular

farming practices. These should be considered in future studies but would require the

collaboration of Australian farmers. For a more thorough understanding of the

prevalence and molecular type of C. difficile in the retail foods of Australia, studies

involving food laboratories from different states and territories are necessary. Currently,

there is no international standard for culturing C. difficile from foods. Thus, culturing

methods would need to be standardised across laboratories prior to the commencement

of studies.

In the compost sampling study, a small number of samples was collected during the

composting/digesting process, as most of the samples were end products ready to be

sold to consumers. Quantification of C. difficile in samples from various stages of the

composting process would have provided information on the effectiveness of the

composting procedure at reducing C. difficile spores and perhaps, more importantly, if

the addition of digestate liquid to composting piles re-introduces C. difficile into the

products. This should be investigated in future studies.


164

In Chapter 6.1, the number of RT 056 isolates (29) investigated was low compared to

this RT contribution to human CDI and its prevalence in Australian cattle. Animal

isolates were not included, which may have provided additional information about

transmission. A greater number of isolates from humans, animals, food and the

environment would have greatly enhanced our understanding of the complex

transmission dynamics of this C. difficile population. Large national studies involving

multiple states and territories are necessary to broaden our collection of food and

environmental isolates.

In Chapter 6.2, the retrospective case-control study on RT 012 had a low response rate.

The number of RT 012 isolates (11) investigated using WGS was small and the

classification of CDI exposure for the sequenced Asian isolates was not available.

Given RT 012 is common in Asia and increasingly being reported in Europe and South

America, including more isolates from these countries could provide additional

information on the global dissemination and transmission of RT 012. Future

epidemiological studies aiming to identify the risk factors for acquiring RT 012 should

include questions regarding recent travel history and consumption of any foods of Asian

origin.

7.4 CONCLUSIONS

This thesis provides novel and critical insights into the prevalence, molecular

epidemiology, AMR, and potential transmission of food and environmental C. difficile

in WA. The studies presented here show food and the environment are a source for

C. difficile in WA, predominated by strains prevalent in both humans and animals with

comparable AMR profiles. For the first time, phylogeny analysis of non-recombinant

SNVs on C. difficile suggested that (i) over an extended period there has been a long-

range transmission of C. difficile RT 056 in WA between humans, food and the

environment, supporting the theory of foodborne and environmental transmission of

CDI; and (ii) that the emergence of RT 012 in WA might be from Asia. Food and

environmental sampling studies for C. difficile in Australia are relatively new. In

coming years, further epidemiological studies and WGS will continue to provide greater

insights into the potential transmission of C. difficile between humans, animals, food

and the environment. Ultimately, a better understanding of CDI transmission is still

required. However, the evidence presented in this thesis adds weight to the hypothesis

that CDI is a zoonotic disease and that a collaborative One Health approach is needed to

reduce the overall burden of CDI.

References

165

REFERENCES

[1] Pepin J, Valiquette L, Alary ME, et al. Clostridium difficile-associated

diarrhea in a region of Quebec from 1991 to 2003: a changing pattern of disease

severity. Canadian Medical Association Journal. 2004; 171: 466-72.

[2] Freeman J, Bauer MP, Baines SD, et al. The changing epidemiology of

Clostridium difficile infections. Clinical Microbiology Review. 2010; 23: 529-49.

[3] Slimings C, Armstrong P, Beckingham WD, et al. Increasing incidence of

Clostridium difficile infection, Australia, 2011-2012. The Medical Journal of Australia.

2014; 200: 272-6.

[4] Paltansing S, van den Berg RJ, Guseinova RA, et al. Characteristics and

incidence of Clostridium difficile-associated disease in The Netherlands, 2005. Clinical

Microbiology and Infection. 2007; 13: 1058-64.

[5] Khanna S, Pardi DS, Aronson SL, et al. The epidemiology of community-

acquired Clostridium difficile infection: a population-based study. The American

Journal of Gastroenterology. 2012; 107: 89-95.

[6] Hall IC, O'Toole E. Intestinal flora in new-borin infants: with a description of a

new pathogenic anaerobe, Bacillus difficilis. American Journal of Diseases of Children.

1935; 49: 390-402.

[7] Snyder ML. The normal fecal flora of infants between two weeks and one year

of age. Journal of Infectious Diseases. 1940; 66: 1-16.

[8] Smith LD, King EO. Occurrence of Clostridium difficile in infections of man.

Journal of Bacteriology. 1962; 84: 65-67.

[9] Gelfand MD, Krone CL. Non-staphylococcal pseudomembranous colitis. The

American Journal of Digestive Diseases. 1969; 14: 278-81.

[10] Valberg LS, Truelove SC. Noninfective pseudomembranous colitis following

antibiotic therapy. American Journal of Digestive Diseases. 1960; 5: 728-38.

[11] McGovern V, Goulston SJ. Pseudomembranous colitis. Gut. 1965; 6: 207-12.

[12] Hafiz S, Morton RS, McEntegart MG, et al. Clostridium difficile in urogenital

tract of males and females. Lancet. 1975; 1: 420-21.

[13] Hafiz S, Oakley CL. Clostridium difficile: isolation and characteristics. Journal

of Medical Microbiology. 1976; 9: 129-36.

[14] George RH, Symonds JM, Dimock F, et al. Identification of Clostridium

difficile as a cause of pseudomembranous colitis. British Medical Journal. 1978; 1: 695-

95.

References

166

[15] George WL, Goldstein EJC, Sutter VL, et al. Etiology of antimicrobial-agent

associated colitis. Lancet. 1978; 1: 802-03.

[16] Larson HE, Price AB, Honour P, et al. Clostridium difficile and etiology of

pseudomembranous colitis. Lancet. 1978; 1: 1063-66.

[17] Riley TV, O'Neill GL, Bowman RA, et al. Clostridium difficile -associated

diarrhoea: epidemiological data from Western Australia. Epidemiology and Infection.

1994; 113: 13-20.

[18] Thomas C, Stevenson M, Williamson DJ, et al. Clostridium difficile-

associated diarrhea: epidemiological data from Western Australia associated with a

modified antibiotic policy. Clinical Infectious Diseases. 2002; 35: 1457-62.

[19] Belmares J, Johnson S, Parada JP, et al. Molecular epidemiology of

Clostridium difficile over the course of 10 years in a tertiary care hospital. Clinical

Infectious Diseases. 2009; 49: 1141-7.

[20] Rupnik M, Wilcox MH, Gerding DN. Clostridium difficile infection: new

developments in epidemiology and pathogenesis. Nature Reviews Microbiology. 2009;

7: 526-36.

[21] Kuijper EJ, Coignard B, Tull P. Emergence of Clostridium difficile-associated

disease in North America and Europe. Clinical Microbiology and Infection. 2006; 12: 2-

18.

[22] Khalaf N, Crews JD, DuPont HL, et al. Clostridium difficile: an emerging

pathogen in children. Discovery Medicine. 2012; 14: 105-13.

[23] Carrion AF, Hosein PJ, Cooper EM, et al. Severe colitis associated with

docetaxel use: a report of four cases. World Journal of Gastrointestinal Oncology.

2010; 2: 390-94.

[24] Carpenter HA, Talley NJ. The importance of clinicopathological correlation in

the diagnosis of inflammatory conditions of the colon: histological patterns with clinical

implications. The American Journal of Gastroenterology. 2000; 95: 878-96.

[25] Mitchell TJ, Ketley JM, Haslam SC, et al. Effect of toxin A and B of

Clostridium difficile on rabbit ileum and colon. Gut. 1986; 27: 78-85.

[26] Riegler M, Sedivy R, Pothoulakis C, et al. Clostridium difficile toxin B is

more potent than toxin A in damaging human colonic epithelium in vitro. The Journal

of Clinical Investigation. 1995; 95: 2004-11.

[27] Kuehne SA, Cartman ST, Heap JT, et al. The role of toxin A and toxin B in

Clostridium difficile infection. Nature. 2010; 467: 711-3.

References

167

[28] Owens RC, Donskey CJ, Gaynes RP, et al. Antimicrobial-associated risk

factors for Clostridium difficile infection. Clinical Infectious Diseases. 2008; 46: S19-

S31.

[29] Kuntz JL, Chrischilles EA, Pendergast JF, et al. Incidence of and risk factors

for community-associated Clostridium difficile infection: a nested case-control study.

BMC Infectious Diseases. 2011; 11: 194.

[30] B&B. Bristol stool form scale. Bladder and Bowel Community (B&B). Web

citation: https://www.bladderandbowel.org (last accessed August 2018).

[31] Mantas A. Endoscopy colon ascending normal. Mantas MD Gasteroenterology.

Web citation: https://mantasmd.com/ (last accessed August 2018).

[32] Pakyz AL, Jawahar R, Wang Q, et al. Medication risk factors associated with

healthcare-associated Clostridium difficile infection: a multilevel model casecontrol

study among 64 US academic medical centres. Journal of Antimicrobial Chemotherapy.

2014; 69: 1127-31.

[33] Nicholson MR, Thomsen IP, Slaughter JC, et al. Novel risk factors for

recurrent Clostridium difficile infection in children. Journal of Pediatric

Gastroenterology and Nutrition. 2015; 60: 18-22.

[34] Bloomfield MG, Sherwin JC, Gkrania-Klotsas E. Risk factors for mortality in

Clostridium difficile infection in the general hospital population: a systematic review.

Journal of Hospital Infection. 2012; 82: 1-12.

[35] Planche T, Wilcox M. Reference assays for Clostridium difficile infection: one

or two gold standards? Journal of Clinical Pathology. 2011; 64: 1-5.

[36] Swindells J, Brenwald N, Reading N, et al. Evaluation of diagnostic tests for

Clostridium difficile infection. Journal of Clinical Microbiology. 2010; 48: 606-08.

[37] Planche T, Aghaizu A, Holliman R, et al. Diagnosis of Clostridium difficile

infection by toxin detection kits: a systematic review. The Lancet Infectious Diseases.

2008; 8: 777-84.

[38] Delmée M, Van Broeck J, Simon A, et al. Laboratory diagnosis of Clostridium

difficile-associated diarrhoea: a plea for culture. Journal of Medical Microbiology.

2005; 54: 187-91.

[39] Bowman RA, Riley TV. Laboratory diagnosis of Clostridium difficile-

associated diarrhoea. European Journal of Clinical Microbiology and Infectious

Diseases. 1988; 7: 476-84.

[40] Wilkins TD, Lyerly DM. Clostridium difficile testing: after 20 years, still

challenging. Journal of Clinical Microbiology. 2003; 41: 531-34.

https://www.bladderandbowel.org/

https://mantasmd.com/

References

168

[41] Lyerly DM, Barroso LA, Wilkins TD. Identification of the latex test-reactive

protein of Clostridium difficile as glutamate dehydrogenase. Journal of Clinical

Microbiology. 1991; 29: 2639-42.

[42] Planche TD, Davies KA, Coen PG, et al. Differences in outcome according to

Clostridium difficile testing method: a prospective multicentre diagnostic validation

study of C. difficile infection. The Lancet Infectious Diseases. 2013; 13: 936-45.

[43] Collins DA, Elliott B, Riley TV. Molecular methods for detecting and typing of

Clostridium difficile. Pathology. 2015; 47: 211-8.

[44] Boseiwaqa LV, Foster NF, Thean SK, et al. Comparison of ChromID C.

difficile agar and cycloserine-cefoxitin-fructose agar for the recovery of Clostridium

difficile. Pathology. 2013; 45: 495-500.

[45] Carson KC, Boseiwaqa LV, Thean SK, et al. Isolation of Clostridium difficile

from faecal specimens: a comparison of chromID C. difficile agar and cycloserine-

cefoxitin-fructose agar. Journal of Medical Microbiology. 2013; 62: 1423-7.

[46] Cohen SH, Gerding DN, Johnson S, et al. Clinical practice guidelines for

Clostridium difficile infection in adults: 2010 update by the Society for Healthcare

Epidemiology of America (SHEA) and the Infectious Diseases Society of America

(IDSA). Infection Control and Hospital Epidemiology. 2010; 31: 431-55.

[47] Cheng AC, Ferguson JK, Richards MJ, et al. Australasian Society for

Infectious Diseases guidelines for the diagnosis and treatment of Clostridium difficile

infection. Medical Journal of Australia. 2011; 194: 353-58.

[48] Huber CA, Foster NF, Riley TV, et al. Clostridium difficile typing methods:

callenges for standardization. Journal of Clinical Microbiology. 2013; 51: 2810.

[49] Rodriguez C, Van Broeck J, Taminiau B, et al. Clostridium difficile infection:

early history, diagnosis and molecular strain typing methods. Microbial Pathogenesis.

2016; 97: 59-78.

[50] Stubbs SLJ, Brazier JS, O'Neill GL, et al. PCR targeted to the 16S-23S rRNA

gene intergenic spacer region of Clostridium difficile and construction of a library

consisting of 116 different PCR ribotypes. Journal of Clinical Microbiology. 1999; 37:

461-63.

[51] Indra A, Huhulescu S, Schneeweis M, et al. Characterization of Clostridium

difficile isolates using capillary gel electrophoresis-based PCR ribotyping. Journal of

Medical Microbiology. 2008; 57: 1377-82.

References

169

[52] Rupnik M, Avesani V, Janc M, et al. A novel toxinotyping scheme and

correlation of toxinotypes with serogroups of Clostridium difficile isolates. Journal of

Clinical Microbiology. 1998; 36: 2240-47.

[53] Rupnik M, Janezic S. An update on Clostridium difficile toxinotyping. Journal

of Clinical Microbiology. 2016; 54: 13-18.

[54] Knetsch CW, Lawley TD, Hensgens MP, et al. Current application and future

perspectives of molecular typing methods to study Clostridium difficile infections.

Eurosurveillance. 2013; 18: 57-67.

[55] Griffiths D, Fawley W, Kachrimanidou M, et al. Multilocus sequence typing

of Clostridium difficile. Journal of Clinical Microbiology. 2010; 48: 770-78.

[56] Stabler RA, He M, Dawson L, et al. Comparative genome and phenotypic

analysis of Clostridium difficile 027 strains provides insight into the evolution of a

hypervirulent bacterium. Genome Biology. 2009; 10: R102.

[57] He M, Sebaihia M, Lawley TD, et al. Evolutionary dynamics of Clostridium

difficile over short and long time scales. Proceedings of the National Academy of

Sciences of the United States of America. 2010; 107: 7527-32.

[58] Brouwer MSM, Allan E, Mullany P, et al. Draft genome sequence of the

nontoxigenic Clostridium difficile strain CD37. Journal of Bacteriology. 2012; 194:

2125-26.

[59] Janvilisri T, Scaria J, Thompson AD, et al. Microarray identification of

Clostridium difficile core components and divergent regions associated with host origin.

Journal of Bacteriology. 2009; 191: 3881-91.

[60] Sebaihia M, Wren BW, Mullany P, et al. The multidrug-resistant human

pathogen Clostridium difficile has a highly mobile, mosaic genome. Nature Genetics.

2006; 38: 779-86.

[61] Stabler RA, Gerding DN, Songer JG, et al. Comparative phylogenomics of

Clostridium difficile reveals clade specificity and microevolution of hypervirulent

strains. Journal of Bacteriology. 2006; 188: 7297-305.

[62] Knight DR, Elliott B, Chang BJ, et al. Diversity and evolution in the genome

of Clostridium difficile. Clinical Microbiology Reviews. 2015; 28: 721-41.

[63] Didelot X, Eyre DW, Cule M, et al. Microevolutionary analysis of Clostridium

difficile genomes to investigate transmission. Genome Biology. 2012; 13.

[64] Eyre DW, Cule ML, Wilson DJ, et al. Diverse sources of C. difficile infection

identified on whole-genome sequencing. New England Journal of Medicine. 2013; 369:

1195-205.

References

170

[65] He M, Miyajima F, Roberts P, et al. Emergence and global spread of epidemic

healthcare-associated Clostridium difficile. Nature Genetics. 2013; 45: 109-13.

[66] Knetsch CW, Connor TR, Mutreja A, et al. Whole genome sequencing

reveals potential spread of Clostridium difficile between humans and farm animals in

the Netherlands, 2002 to 2011. Eurosurveillance. 2014; 19: 30-41.

[67] CDC. 2013. Antibiotic resistance threats in the United States, 2013. Centers for

Disease Control and Prevention (CDC): Atlanta, United States. Web citation:

https://www.cdc.gov/drugresistance/threat-report-2013/ (last accessed April 2018).

[68] Lessa FC, Mu Y, Bamberg WM, et al. Burden of Clostridium difficile

infection in the United States. New England Journal of Medicine. 2015; 372: 825-34.

[69] Pepin J, Valiquette L, Cossette B. Mortality attributable to nosocomial

Clostridium difficile-associated disease during an epidemic caused by a hypervirulent

strain in Quebec. Canadian Medical Association Journal. 2005; 173: 1037-42.

[70] Zilberberg MD, Shorr AF, Kollef MH. Increase in adult Clostridium difficile–

related hospitalizations and case-fatality rate, United States, 2000–2005. Emerging

Infectious Diseases. 2008; 14: 929-31.

[71] Davies KA, Longshaw CM, Davis GL, et al. Underdiagnosis of Clostridium

difficile across Europe: the European, multicentre, prospective, biannual, point-

prevalence study of Clostridium difficile infection in hospitalised patients with

diarrhoea (EUCLID). Lancet Infectious Diseases. 2014; 14: 1208-19.

[72] Al-Eidan FA, McElnay JC, Scott MG, et al. Clostridium difficile-associated

diarrhoea in hospitalised patients. Journal of Clinical Pharmacy and Therapeutics.

2000; 25: 101-09.

[73] Bond SE, Boutlis CS, Yeo WW, et al. The burden of healthcare-associated

Clostridium difficile infection in a non-metropolitan setting. Journal of Hospital

Infection. 2017; 95: 387-93.

[74] Wilcox MH, Cunniffe JG, Trundle C, et al. Financial burden of hospital-

acquired Clostridium difficile infection. Journal of Hospital Infection. 1996; 34: 23-30.

[75] Dubberke ER, Wertheimer AI. Review of current literature on the economic

burden of Clostridium difficile infection. Infection Control and Hospital Epidemiology.

2009; 30: 57-66.

[76] McGlone SM, Bailey RR, Zimmer SM, et al. The economic burden of

Clostridium difficile. Clinical Microbiology and Infection. 2012; 18: 282-89.

https://www.cdc.gov/drugresistance/threat-report-2013/

References

171

[77] Redelings MD, Sorvillo F, Mascola L. Increase in Clostridium difficile–related

mortality rates, United States, 1999–2004. Emerging Infectious Diseases. 2007; 13:

1417-19.

[78] Jones AM, Kuijper EJ, Wilcox MH. Clostridium difficile: a European

perspective. The Journal of Infection. 2013; 66: 115-28.

[79] Burckhardt F, Friedrich A, Beier D, et al. Clostridium difficile surveillance

trends, Saxony, Germany. Emerging Infectious Diseases. 2008; 14: 691-92.

[80] Lyytikainen O, Turunen H, Sund R, et al. Hospitalizations and deaths

associated with Clostridium difficile infection, Finland, 1996-2004. Emerging Infectious

Diseases. 2009; 15: 761-5.

[81] Soler P, Nogareda F, Cano R. Rates of Clostridium difficile infection in

patients discharged from Spanish hospitals, 1997-2005. Infection Control and Hospital

Epidemiology. 2008; 29: 886-89.

[82] Bauer MP, Notermans DW, van Benthem BHB, et al. Clostridium difficile

infection in Europe: a hospital-based survey. Lancet. 2011; 377: 63-73.

[83] Freeman J, Vernon J, Morris K, et al. Pan-European longitudinal surveillance

of antibiotic resistance among prevalent Clostridium difficile ribotypes. Clinical

Microbiology and Infection. 2015; 21: 248.e9–48.e16.

[84] Cheng VCC, Yam WC, Chan JFW, et al. Clostridium difficile ribotype 027

arrives in Hong Kong. International Journal of Antimicrobial Agents. 2009; 34: 492-93.

[85] Richards M, Knox J, Elliott B, et al. Severe infection with Clostridium

difficile PCR ribotype 027 acquired in Melbourne, Australia. Medical Journal of

Australia. 2011; 194: 369-71.

[86] Carter GP, Rood JI, Lyras D. The role of toxin A and toxin B in the virulence

of Clostridium difficile. Trends in Microbiology. 2012; 20: 21-29.

[87] McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin gene–

variant strain of Clostridium difficile. New England Journal of Medicine. 2005; 353:

2433-41.

[88] Merrigan M, Venugopal A, Mallozzi M, et al. Human hypervirulent

Clostridium difficile strains exhibit increased sporulation as well as robust toxin

production. Journal of Bacteriology. 2010; 192: 4904-11.

[89] Cartman ST, Kelly ML, Heeg D, et al. Precise manipulation of the

Clostridium difficile chromosome reveals a lack of association between tcdC genotype

and toxin production. Applied and Environmental Microbiology. 2012; 78: 4683.

References

172

[90] Stabler RA, Dawson LF, Phua LTH, et al. Comparative analysis of

BI/NAP1/027 hypervirulent strains reveals novel toxin B-encoding gene (tcdB)

sequences. Journal of Medical Microbiology. 2008; 57: 771-75.

[91] Cowardin CA, Buonomo EL, Saleh MM, et al. The binary toxin CDT

enhances Clostridium difficile virulence by suppressing protective colonic eosinophilia.

Nature Microbiology. 2016; 1: 1-10.

[92] Barbut F, Decre D, Lalande V, et al. Clinical features of Clostridium difficile-

associated diarrhoea due to binary toxin (actin-specific ADP-ribosyltransferase)-

producing strains. Journal of Medical Microbiology. 2005; 54: 181-5.

[93] Riley TV, Cooper M, Bell B, et al. Community-acquired Clostridium difficile-

associated diarrhea. Clinical Infectious Diseases. 1995; 20: S263-S65.

[94] McDonald LC, Coignard B, Dubberke E, et al. Recommendations for

surveillance of Clostridium difficile-associated disease. Infection Control and Hospital

Epidemiology. 2007; 28: 140-45.

[95] Kwon SS, Gim JL, Kim MS, et al. Clinical and molecular characteristics of

community-acquired Clostridium difficile infections in comparison with those of

hospital-acquired C. difficile. Anaerobe. 2017; 48: 42-46.

[96] Bloomfield LE, Riley TV. Epidemiology and risk factors for community-

associated Clostridium difficile infection: a narrative review. Infectious Diseases and

Therapy. 2016; 5: 231-51.

[97] Wilcox MH, Mooney L, Bendall R, et al. A case-control study of community-

associated Clostridium difficile infection. The Journal of Antimicrobial Chemotherapy.

2008; 62: 388-96.

[98] Goorhuis A, Bakker D, Corver J, et al. Emergence of Clostridium difficile

infection due to a new hypervirulent strain, polymerase chain reaction ribotype 078.

Clinical Infectious Diseases. 2008; 47: 1162-70.

[99] Jhung MA, Thompson AD, Killgore GE, et al. Toxinotype V Clostridium

difficile in humans and food animals. Emerging Infectious Diseases. 2008; 14: 1039-45.

[100] Barbut F, Mastrantonio P, Delmee M, et al. Prospective study of Clostridium

difficile infections in Europe with phenotypic and genotypic characterisation of the

isolates. Clinical Microbiology and Infection. 2007; 13: 1048-57.

[101] Costa MC, Reid-Smith R, Gow S, et al. Prevalence and molecular

characterization of Clostridium difficile isolated from feedlot beef cattle upon arrival

and mid-feeding period. BMC Veterinary Research. 2012; 8: 38.

References

173

[102] Kalchayanand N, Arthur TM, Bosilevac JM, et al. Isolation and

characterization of Clostridium difficile associated with beef cattle and commercially

produced ground beef. Journal of Food Protection. 2013; 76: 256-64.

[103] Rodriguez C, Avesani V, Van Broeck J, et al. Presence of Clostridium difficile

in pigs and cattle intestinal contents and carcass contamination at the slaughterhouse in

Belgium. International Journal of Food Microbiology. 2013; 166: 256-62.

[104] Rodriguez C, Taminiau B, Van Broeck J, et al. Clostridium difficile in young

farm animals and slaughter animals in Belgium. Anaerobe. 2012; 18: 621-25.

[105] Zidaric V, Pardon B, Dos Vultos T, et al. Different antibiotic resistance and

sporulation properties within multiclonal Clostridium difficile PCR ribotypes 078, 126,

and 033 in a single calf farm. Applied and Environmental Microbiology. 2012; 78:

8515-22.

[106] Rodriguez-Palacios A, Stampfli HR, Duffield T, et al. Clostridium difficile

PCR ribotypes in calves, Canada. Emerging Infectious Diseases. 2006; 12: 1730-36.

[107] Costa MC, Stampfli HR, Arroyo LG, et al. Epidemiology of Clostridium

difficile on a veal farm: prevalence, molecular characterization and tetracycline

resistance. Veterinary Microbiology. 2011; 152: 379-84.

[108] Schneeberg A, Neubauer H, Schmoock G, et al. Presence of Clostridium

difficile PCR ribotype clusters related to 033, 078 and 045 in diarrhoeic calves in

Germany. Journal of Medical Microbiology. 2013; 62: 1190-98.

[109] Magistrali CF, Maresca C, Cucco L, et al. Prevalence and risk factors

associated with Clostridium difficile shedding in veal calves in Italy. Anaerobe. 2015;

33: 42-7.

[110] Hofer E, Haechler H, Frei R, et al. Low occurrence of Clostridium difficile in

fecal samples of healthy calves and pigs at slaughter and in minced meat in Switzerland.

Journal of Food Protection. 2010; 73: 973-75.

[111] Hammitt MC, Bueschel DA, Kee AK, et al. A possible role for Clostridium

difficile in the etiology of calf enteritis. Veterinary Microbiology. 2008; 127: 343-52.

[112] Weese JS, Wakeford T, Reid-Smith R, et al. Longitudinal investigation of

Clostridium difficile shedding in piglets. Anaerobe. 2010; 16: 501-4.

[113] Schneeberg A, Neubauer H, Schmoock G, et al. Clostridium difficile

genotypes in piglet populations in Germany. Journal of Clinical Microbiology. 2013;

51: 3796-803.

[114] Stein K, Egan S, Lynch H, et al. PCR-ribotype distribution of Clostridium

difficile in Irish pigs. Anaerobe. 2017; 48: 237-41.

References

174

[115] Cho A, Byun JW, Kim JW, et al. Low prevalence of Clostridium difficile in

slaughter pigs in Korea. Journal of Food Protection. 2015; 78: 1034-36.

[116] Hopman NEM, Keessen EC, Harmanus C, et al. Acquisition of Clostridium

difficile by piglets. Veterinary Microbiology. 2011; 149: 186-92.

[117] Debast SB, van Leengoed LA, Goorhuis A, et al. Clostridium difficile PCR

ribotype 078 toxinotype V found in diarrhoeal pigs identical to isolates from affected

humans. Environmental Microbiology. 2009; 11: 505-11.

[118] Liao F, Li W, Gu W, et al. A retrospective study of community-acquired

Clostridium difficile infection in southwest China. Scientific Reports. 2018; 8: 3992.

[119] Ye GY, Li N, Chen YB, et al. Clostridium difficile carriage in healthy pregnant

women in China. Anaerobe. 2016; 37: 54-57.

[120] Killgore G, Thompson A, Johnson S, et al. Comparison of seven techniques

for typing international epidemic strains of Clostridium difficile: restriction

endonuclease analysis, pulsed-field gel electrophoresis, PCR-ribotyping, multilocus

sequence typing, multilocus variable-number tandem-repeat analysis, amplified

fragment length polymorphism, and surface layer protein A gene sequence typing.

Journal of Clinical Microbiology. 2008; 46: 431-7.

[121] Knight DR, Riley TV. Prevalence of gastrointestinal Clostridium difficile

carriage in Australian sheep and lambs. Applied and Environmental Microbiology.

2013; 79: 5689-92.

[122] Knight DR, Thean S, Putsathit P, et al. Cross-sectional study reveals high

prevalence of Clostridium difficile non-PCR ribotype 078 strains in Australian veal

calves at slaughter. Applied and Environmental Microbiology. 2013; 79: 2630-5.

[123] Weese JS, Reid-Smith RJ, Avery BP, et al. Detection and characterization of

Clostridium difficile in retail chicken. Letters in Applied Microbiology. 2010; 50: 362-5.

[124] Weese JS, Avery BP, Rousseau J, et al. Detection and enumeration of

Clostridium difficile spores in retail beef and pork. Applied and Environmental

Microbiology. 2009; 75: 5009-11.

[125] Rahimi E, Jalali M, Weese JS, et al. Prevalence of Clostridium difficile in raw

beef, cow, sheep, goat, camel and buffalo meat in Iran. BMC Public Health 2014: 119.

[126] Rodriguez-Palacios A, Staempfli HR, Duffield T, et al. Clostridium difficile

in retail ground meat, Canada. Emerging Infectious Diseases. 2007; 13: 485-87.

[127] de Boer E, Zwartkruis-Nahuis A, Heuvelink AE, et al. Prevalence of

Clostridium difficile in retailed meat in the Netherlands. International Journal of Food

Microbiology. 2011; 144: 561-4.

References

175

[128] Harvey RB, Norman KN, Andrews K, et al. Clostridium difficile in poultry

and poultry meat. Foodborne Pathogens and Disease. 2011; 8: 1321-3.

[129] Harvey RB, Norman KN, Andrews K, et al. Clostridium difficile in retail meat

and processing plants in Texas. Journal of Veterinary Diagnostic Investigation. 2011;

23: 807-11.

[130] Curry SR, Marsh JW, Schlackman JL, et al. Prevalence of Clostridium

difficile in uncooked ground meat products from Pittsburgh, Pennsylvania. Applied and

Environmental Microbiology. 2012; 78: 4183-6.

[131] Metcalf D, Avery BP, Janecko N, et al. Clostridium difficile in seafood and

fish. Anaerobe. 2011; 17: 85-86.

[132] Metcalf DS, Costa MC, Dew WMV, et al. Clostridium difficile in vegetables,

Canada. Letters in Applied Microbiology. 2010; 51: 600-02.

[133] Pasquale V, Romano V, Rupnik M, et al. Occurrence of toxigenic Clostridium

difficile in edible bivalve molluscs. Food Microbiology. 2012; 31: 309-12.

[134] Songer JG, Trinh HT, Killgore GE, et al. Clostridium difficile in retail meat

products, USA, 2007. Emerging Infectious Diseases. 2009; 15: 819-21.

[135] Quesada-gomez C, Mulvey MR, Vargas P. Isolation of a toxigenic and

clinical genotype of Clostridium difficile in retail meats in Costa Rica. Journal of Food

Protection. 2013; 76: 348-51.

[136] Pasquale V, Romano VJ, Rupnik M, et al. Isolation and characterization of

Clostridium difficile from shellfish and marine environments. Folia Microbiologica.

2011; 56: 431-37.

[137] Eckert C, Burghoffer B, Barbut F. Contamination of ready-to-eat raw

vegetables with Clostridium difficile in France. Journal of Medical Microbiology. 2013;

62: 1435-8.

[138] Rodriguez-Palacios A, Reid-Smith RJ, Staempfli HR, et al. Possible

seasonality of Clostridium difficile in retail meat, Canada. Emerging Infectious

Diseases. 2009; 15: 802-5.

[139] Bouttier S, Barc M-C, Felix B, et al. Clostridium difficile in ground meat,

France. Emerging Infectious Diseases. 2010; 16: 733-35.

[140] Limbago B, Thompson AD, Greene SA, et al. Development of a consensus

method for culture of Clostridium difficile from meat and its use in a survey of U.S.

retail meats. Food Microbiology. 2012; 32: 448-51.

References

176

[141] Sophie Marie Martirani Von A, Frida K, Gunilla Trowald W, et al. Low

occurrence of Clostridium difficile in retail ground meat in Sweden. Journal of Food

Protection. 2009; 72: 1732-34.

[142] Visser M, Sepehrim S, Olson N, et al. Detection of Clostridium difficile in

retail ground meat products in Manitoba. Canadian Journal of Infectious Diseases and

Medical Microbiology. 2012; 23: 28-30.

[143] Al Saif N, Brazier JS. The distribution of Clostridium difficile in the

environment of South Wales. Journal of Medical Microbiology. 1996; 45: 133-37.

[144] Simango C. Prevalence of Clostridium difficile in the environment in a rural

community in Zimbabwe. Transactions of the Royal Society of Tropical Medicine and

Hygiene. 2006; 100: 1146-50.

[145] Adegboyega T. Environmental sources Clostridium difficile in Lagos State,

Nigeria. Journal of Health, Medicine and Nursing. 2014; 3: 1-5.

[146] Zidaric V, Beigot S, Lapajne S, et al. The occurrence and high diversity of

Clostridium difficile genotypes in rivers. Anaerobe. 2010; 16: 371-5.

[147] Janezic S, Potocnik M, Zidaric V, et al. Highly divergent Clostridium difficile

strains isolated from the environment. PloS one. 2016; 11: e0167101.

[148] Simango C, Mwakurudza S. Clostridium difficile in broiler chickens sold at

market places in Zimbabwe and their antimicrobial susceptibility. International Journal

of Food Microbiology. 2008; 124: 268-70.

[149] Orden C, Neila C, Blanco JL, et al. Recreational sandboxes for children and

dogs can be a source of epidemic ribotypes of Clostridium difficile. Zoonoses and

Public Health. 2018; 65: 88-95.

[150] Alam MJ, Anu A, Walk ST, et al. Investigation of potentially pathogenic

Clostridium difficile contamination in household environs. Anaerobe. 2014; 27: 31-33.

[151] Shaughnessy MK, Bobr A, Kuskowski MA, et al. Environmental

contamination in households of patients with recurrent Clostridium difficile infection.

Applied and Environmental Microbiology. 2016; 82: 2686-92.

[152] Pirs T, Ocepek M, Rupnik M. Isolation of Clostridium difficile from food

animals in Slovenia. Journal of Medical Microbiology. 2008; 57: 790-2.

[153] Avbersek J, Janezic S, Pate M, et al. Diversity of Clostridium difficile in pigs

and other animals in Slovenia. Anaerobe. 2009; 15: 252-5.

[154] Knight DR, Squire MM, Riley TV. Nationwide surveillance study of

Clostridium difficile in Australian neonatal pigs shows high prevalence and

References

177

heterogeneity of PCR ribotypes. Applied and Environmental Microbiology. 2015; 81:

119-23.

[155] Thitaram SN, Frank JF, Siragusa GR, et al. Antimicrobial susceptibility of

Clostridium difficile isolated from food animals on farms. International Journal of Food

Microbiology. 2016; 227: 1-5.

[156] Koene MGJ, Mevius D, Wagenaar JA, et al. Clostridium difficile in Dutch

animals: their presence, characteristics and similarities with human isolates. Clinical

Microbiology and Infection. 2012; 18: 778-84.

[157] Rodriguez-Palacios A, Barman T, LeJeune JT. Three-week summer period

prevalence of Clostridium difficile in farm animals in a temperate region of the United

States (Ohio). Canadian Veterinary Journal. 2014; 55: 786-89.

[158] Bandelj P, Blagus R, Briski F, et al. Identification of risk factors influencing

Clostridium difficile prevalence in middle-size dairy farms. Veterinary Research. 2016;

47: 41.

[159] Hampikyan H, Bingol EB, Muratoglu K, et al. The prevalence of Clostridium

difficile in cattle and sheep carcasses and the antibiotic susceptibility of isolates. Meat

Science. 2018; 139: 120-24.

[160] Alvarez-Perez S, Blanco JL, Bouza E, et al. Prevalence of Clostridium difficile

in diarrhoeic and non-diarrhoeic piglets. Veterinary Microbiology. 2009; 137: 302-5.

[161] Usui M, Nanbu Y, Oka K, et al. Genetic relatedness between Japanese and

European isolates of Clostridium difficile originating from piglets and their risk

associated with human health. Frontiers in Microbiology. 2014; 5: 513.

[162] Norman KN, Harvey RB, Scott HM, et al. Varied prevalence of Clostridium

difficile in an integrated swine operation. Anaerobe. 2009; 15: 256-60.

[163] Pelaez T, Alcala L, Blanco JL, et al. Characterization of swine isolates of

Clostridium difficile in Spain: a potential source of epidemic multidrug resistant strains?

Anaerobe. 2013; 22: 45-49.

[164] Wu Y-C, Chen C-M, Kuo C-J, et al. Prevalence and molecular

characterization of Clostridium difficile isolates from a pig slaughterhouse, pork, and

humans in Taiwan. International Journal of Food Microbiology. 2017; 242: 37-44.

[165] Fry PR, Thakur S, Abley M, et al. Antimicrobial resistance, toxinotype, and

genotypic profiling of Clostridium difficile isolates of swine origin. Journal of Clinical

Microbiology. 2012; 50: 2366-72.

References

178

[166] Zidaric V, Zemljic M, Janezic S, et al. High diversity of Clostridium difficile

genotypes isolated from a single poultry farm producing replacement laying hens.

Anaerobe. 2008; 14: 325-27.

[167] Rodriguez C, Taminiau B, Avesani V, et al. Multilocus sequence typing

analysis and antibiotic resistance of Clostridium difficile strains isolated from retail

meat and humans in Belgium. Food Microbiology. 2014; 42: 166-71.

[168] Norman KN, Harvey RB, Andrews K, et al. Survey of Clostridium difficile in

retail seafood in College Station, Texas. Food Additives and Contaminants. 2014; 31:

1127-9, 2014.

[169] Rodriguez-Palacios A, Ilic S, LeJeune JT. Clostridium difficile with

moxifloxacin/clindamycin resistance in vegetables in Ohio, USA, and prevalence meta-

analysis. Journal of Pathogens. 2014; 2014: 158601.

[170] Rodriguez C, Hakimi DE, Vanleyssem R, et al. Clostridium difficile in beef

cattle farms, farmers and their environment: Assessing the spread of the bacterium.

Veterinary Microbiology. 2017; 210: 183-87.

[171] Doosti A, Mokhtari-Farsani A. Study of the frequency of Clostridium difficile

tcdA, tcdB, cdtA and cdtB genes in feces of Calves in south west of Iran. Annals of

Clinical Microbiology and Antimicrobials. 2014; 13: 21.

[172] Romano V, Albanese F, Dumontet S, et al. Prevalence and genotypic

characterization of Clostridium difficile from ruminants in Switzerland. Zoonoses and

Public Health. 2012; 59: 545-48.

[173] Houser BA, Soehnlen MK, Wolfgang DR, et al. Prevalence of Clostridium

difficile toxin genes in the feces of veal calves and incidence of ground veal

contamination. Foodborne Pathogens and Disease. 2012; 9: 32-36.

[174] Esfandiari Z, Weese JS, Ezzatpanah H, et al. Isolation and characterization of

Clostridium difficile in farm animals from slaughterhouse to retail stage in Isfahan, Iran.

Foodborne Pathogens and Disease. 2015; 12: 864-6.

[175] Rodriguez-Palacios A, Pickworth C, Loerch S, et al. Transient fecal shedding

and limited animal-to-animal transmission of Clostridium difficile by naturally infected

finishing feedlot cattle. Applied and Environmental Microbiology. 2011; 77: 3391-7.

[176] Thitaram SN, Frank JF, Lyon SA, et al. Clostridium difficile from healthy

food animals: optimized isolation and prevalence. Journal of Food Protection. 2011;

74: 130-33.

References

179

[177] Rodriguez-Palacios A, Koohmaraie M, LeJeune JT, et al. Prevalence,

enumeration, and antimicrobial agent resistance of Clostridium difficile in cattle at

harvest in the United States. Journal of Food Protection. 2011: 1618-24, 2011 Oct.

[178] Avbersek J, Pirs T, Pate M, et al. Clostridium difficile in goats and sheep in

Slovenia: characterisation of strains and evidence of age-related shedding. Anaerobe.

2014; 28: 163-67.

[179] Squire MM, Carter GP, Mackin KE, et al. Novel molecular type of

Clostridium difficile in neonatal pigs, Western Australia. Emerging Infectious Diseases.

2013; 19: 790-2.

[180] Goldova J, Malinova A, Indra A, et al. Clostridium difficile in piglets in the

Czech Republic. Folia Microbiologica. 2012; 57: 159-61.

[181] Noren T, Johansson K, Unemo M. Clostridium difficile PCR ribotype 046 is

common among neonatal pigs and humans in Sweden. Clinical Microbiology and

Infection. 2014; 20: O2-O6.

[182] Tsai BY, Ko WC, Chen TH, et al. Zoonotic potential of the Clostridium

difficile RT078 family in Taiwan. Anaerobe. 2016; 41: 125-30.

[183] Baker AA, Davis E, Rehberger T, et al. Prevalence and diversity of toxigenic

Clostridium perfringens and Clostridium difficile among swine herds in the midwest.

Applied and Environmental Microbiology. 2010; 76: 2961-7.

[184] Asai T, Usui M, Hiki M, et al. Clostridium difficile isolated from the fecal

contents of swine in Japan. Journal of Veterinary Medical Science. 2013; 75: 539-41.

[185] Hopman NEM, Oorburg D, Sanders I, et al. High occurrence of various

Clostridium difficile PCR ribotypes in pigs arriving at the slaughterhouse. Veterinary

Quarterly. 2011; 31: 179-81.

[186] Keessen EC, van den Berkt AJ, Haasjes NH, et al. The relation between farm

specific factors and prevalence of Clostridium difficile in slaughter pigs. Veterinary

Microbiology. 2011; 154: 130-34.

[187] Andres-Lasheras S, Bolea R, Mainar-Jaime RC, et al. Presence of

Clostridium difficile in pig faecal samples and wild animal species associated with pig

farms. Journal of Applied Microbiology. 2017; 122: 462-72.

[188] Norman KN, Scott HM, Harvey RB, et al. Prevalence and genotypic

characteristics of Clostridium difficile in a closed and integrated human and swine

population. Applied and Environmental Microbiology. 2011; 77: 5755-60.

References

180

[189] Susick EK, Putnam M, Bermudez DM, et al. Longitudinal study comparing

the dynamics of Clostridium difficile in conventional and antimicrobial free pigs at farm

and slaughter. Veterinary Microbiology. 2012; 157: 172-78.

[190] Bakri M. Prevalence of Clostridium difficile in raw cow, sheep, and goat meat

in Jazan, Saudi Arabia. Saudi Journal of Biological Sciences. 2018; 25: 783-85.

[191] Kouassi KA, Dadie AT, N'Guessan KF, et al. Clostridium perfringens and

Clostridium difficile in cooked beef sold in Cote d'Ivoire and their antimicrobial

susceptibility. Anaerobe. 2014; 28: 90-4.

[192] Troiano T, Harmanus C, Sanders IM, et al. Toxigenic Clostridium difficile

PCR ribotypes in edible marine bivalve molluscs in Italy. International Journal of Food


[193] Yamoudy M, Mirlohi M, Isfahani BN, et al. Isolation of toxigenic Clostridium

difficile from ready-to-eat salads by multiplex polymerase chain reaction in Isfahan,

Iran. Advanced Biomedical Research. 2015; 4: 87.

[194] Bakri MM, Brown DJ, Butcher JP, et al. Clostridium difficile in ready-to-eat

salads, Scotland. Emerging Infectious Diseases. 2009; 15: 817-18.

[195] Bakker D, Corver J, Harmanus C, et al. Relatedness of human and animal

Clostridium difficile PCR ribotype 078 isolates determined on the basis of multilocus

variable-number tandem-repeat analysis and tetracycline resistance. Journal of Clinical

Microbiology. 2010; 48: 3744-9.

[196] Knetsch CW, Kumar N, Forster SC, et al. Zoonotic transfer of Clostridium

difficile harboring antimicrobial resistance between farm animals and humans. Journal

of Clinical Microbiology. 2018; 56: e01384-17.

[197] Xu C, Weese JS, Flemming C, et al. Fate of Clostridium difficile during

wastewater treatment and incidence in Southern Ontario watersheds. Journal of Applied

Microbiology. 2014; 117: 891-904.

[198] Riley TV, Thean S, Hool G, et al. First Australian isolation of epidemic

Clostridium difficile PCR ribotype 027. Medical Journal of Australia. 2009; 190: 706-

08.

[199] Riley TV. Is Clostridium difficile a threat to Australia's biosecurity? Medical

Journal of Australia. 2009; 190: 661-62.

[200] Furuya-Kanamori L, Clements AC, Foster NF, et al. Asymptomatic

Clostridium difficile colonization in two Australian tertiary hospitals, 2012-2014:

prospective, repeated cross-sectional study. Clinical Microbiology and Infection. 2017;

23: 48 e1-48 e7.

References

181

[201] Worth LJ, Spelman T, Bull AL, et al. Epidemiology of Clostridium difficile

infections in Australia: enhanced surveillance to evaluate time trends and severity of

illness in Victoria, 2010-2014. Journal of Hospital Infection. 2016; 93: 280-85.

[202] Mitchell BG, Wilson F, McGregor A. An increase in community onset

Clostridium difficile infection: a population-based study, Tasmania, Australia.

Healthcare Infection. 2012; 17: 127-32.

[203] Furuya-Kanamori L, Riley TV, Paterson DL, et al. Comparison of

Clostridium difficile ribotypes circulating in Australian hospitals and communities.

Journal of Clinical Microbiology. 2017; 55: 216-25.

[204] Collins DA, Selvey LA, Celenza A, et al. Community-associated Clostridium

difficile infection in emergency department patients in Western Australia. Anaerobe.

2017; 48: 121-25.

[205] Dingle KE, Elliott B, Robinson E, et al. Evolutionary history of the

Clostridium difficile pathogenicity locus. Genome Biology and Evolution. 2014; 6: 36-

52.

[206] Eyre DW, Tracey L, Elliott B, et al. Emergence and spread of predominantly

community-onset Clostridium difficile PCR ribotype 244 infection in Australia, 2010 to

2012. Eurosurveillance. 2015; 20: 21059.

[207] Foster NF, Collins DA, Ditchburn SL, et al. Epidemiology of Clostridium

difficile infection in two tertiary-care hospitals in Perth, Western Australia: a cross-

sectional study. New Microbes and New Infections. 2014; 2: 64-71.

[208] Collins DA, Riley TV. Routine detection of Clostridium difficile in Western

Australia. Anaerobe. 2016; 37: 34-37.

[209] Cheng AC, Collins DA, Elliott B, et al. Laboratory-based surveillance of

Clostridium difficile circulating in Australia, September - November 2010. Pathology.

2016; 48: 257-60.

[210] Knight DR, Giglio S, Huntington PG, et al. Surveillance for antimicrobial

resistance in Australian isolates of Clostridium difficile, 2013-14. The Journal of

Antimicrobial Chemotherapy. 2015; 70: 2992-9.

[211] Moono P, Foster NF, Hampson DJ, et al. Clostridium difficile infection in

production animals and avian species: A review. Foodborne Pathogens and Disease.

2016; 13: 647-55.

[212] Collins DA, Hawkey PM, Riley TV. Epidemiology of Clostridium difficile

infection in Asia. Antimicrobial Resistance and Infection Control. 2013; 2: 21.

References

182

[213] Collins DA. Epidemiology of Clostridium difficile infecion in the Asia-Pacific

region. The University of Western Australia, 2016.

[214] Knight DR, Squire MM, Collins DA, et al. Genome analysis of Clostridium

difficile PCR ribotype 014 lineage in Australian pigs and humans reveals a diverse

genetic repertoire and signatures of long-range interspecies transmission. Frontiers in

Microbiology. 2017; 7: 2138.

[215] CLSI. 2011. Methods for antimicrobial susceptibility testing of anaerobic

bacteria-seventh edition; approved standard – seventh edition M11-A7. Clinical and

Laboratory Standards Institute (CLSI): Wayne, PA. 2011.

[216] Kato N, Ou CY, Kato H, et al. Identification of toxigenic Clostridium difficile

by the polymerase chain reaction. Journal of Clinical Microbiology. 1991; 29: 33-37.

[217] Stubbs S, Rupnik M, Gibert M, et al. Production of actin-specific ADP-

ribosyltransferase (binary toxin) by strains of Clostridium difficile. FEMS Microbiology

Letters. 2000; 186: 307-12.

[218] Kato H, Kato N, Watanabe K, et al. Identification of toxin A-negative, toxin

B-positive Clostridium difficile by PCR. Journal of Clinical Microbiology. 1998; 36:

2178-82.

[219] Scotts Gardening Team. 2017. The Aussie homeowner's guide to a lush green

lawn. Scotts Gardening Team. Web citation: https://www.scottsaustralia.com.au/diy-

garden-guides-advice/lawns/guide-to-a-lush-green-lawn/guide-to-a-lush-green-lawn/

(last accessed November 2017).

[220] Giner NM, Polsky C, Pontius RG, et al. Understanding the social determinants

of lawn landscapes: a fine-resolution spatial statistical analysis in suburban Boston,

Massachusetts, USA. Landscape and Urban Planning. 2013; 111: 25-33.

[221] Knight DR, Putsathit P, Elliott B, et al. Contamination of Australian newborn

calf carcasses at slaughter with Clostridium difficile. Clinical Microbiology and

Infection. 2016; 22: 266.e1–66.e7.

[222] CLSI. 2013. Performance standards for antimicrobial susceptibility testing;

twenty-third informational supplement M100-S23. Clinical and Laboratory Standards

Institute (CLSI): Wayne, PA. 2013.

[223] EUCAST. 2017. Clinical breakpoint tables for interpretation of MICs and zone

diameters; version 7.1. European Committee on Antimicrobial Susceptibility Testing

(EUCAST): London, United Kingdom. Web citation: http://www.eucast.org/ (last

accessed November 2017).

https://www.scottsaustralia.com.au/diy-garden-guides-advice/lawns/guide-to-a-lush-green-lawn/guide-to-a-lush-green-lawn/

https://www.scottsaustralia.com.au/diy-garden-guides-advice/lawns/guide-to-a-lush-green-lawn/guide-to-a-lush-green-lawn/

http://www.eucast.org/

References

183

[224] EMA. 2011. Assessment report, Dificlir fidaxomicin EMA/857570/2011.

European Medicines Agency (EMA): London, United Kingdom. Web citation:

www.ema.europe.eu (last accessed September 2011).

[225] O'Connor JR, Galang MA, Sambol SP, et al. Rifampin and rifaximin

resistance in clinical isolates of Clostridium difficile. Antimicrobial Agents and

Chemotherapy. 2008; 52: 2813-7.

[226] CLSI. 1999. Determining bactericidal activity of antimicrobial agents; approved

guidelines M26-A. Clinical and Laboratory Standards Institute (CLSI): Wayne, PA.

1999.

[227] Frank KL, Patel R. Activity of sodium metabisulfite against planktonic and

biofilm Staphylococcus species. Diagnostic Microbiology and Infectious Disease. 2007;

57: 355-9.

[228] Odds FC. Synergy, antagonism, and what the chequerboard puts between them.

Journal of Antimicrobial Chemotherapy. 2003; 52: 1.

[229] Cheng JW, Xiao M, Kudinha T, et al. Molecular epidemiology and

antimicrobial susceptibility of Clostridium difficile isolates from a university teaching

hospital in China. Frontiers in Microbiology. 2016; 7: 1621.

[230] Gao Q, Wu S, Huang HH, et al. Toxin profiles, PCR ribotypes and resistance

patterns of Clostridium difficile: a multicentre study in China, 2012-2013. International

Journal of Antimicrobial Agents. 2016; 48: 736-39.

[231] Chow VCY, Kwong TNY, So EWM, et al. Surveillance of antibiotic resistance

among common Clostridium difficile ribotypes in Hong Kong. Scientific Reports. 2017;

7: 17218.

[232] Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina

sequence data. Bioinformatics. 2014; 30: 2114-20.

[233] Inouye M, Dashnow H, Raven LA, et al. SRST2: rapid genomic surveillance

for public health and hospital microbiology labs. Genome Medicine. 2014; 6: 90.

[234] Gupta SK, Padmanabhan BR, Diene SM, et al. ARG-ANNOT, a new

bioinformatic tool to discover antibiotic resistance genes in bacterial genomes.

Antimicrobial Agents and Chemotherapy. 2014; 58: 212-20.

[235] Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and

high throughput. Nucleic Acids Research. 2004; 32: 1792-97.

[236] Guindon S, Delsuc F, Dufayard JF, et al. Estimating maximum likelihood

phylogenies with PhyML Bioinformatics for DNA Sequence Analysis 2009; 113-37.


References

184

[237] Bankevich A, Nurk S, Antipov D, et al. SPAdes: a new genome assembly

algorithm and its applications to single-cell sequencing. Journal of Computational

Biology. 2012; 19: 455-77.

[238] Coil D, Jospin G, Darling AE. A5-miseq: an updated pipeline to assemble

microbial genomes from Illumina MiSeq data. Bioinformatics. 2015; 31: 587-89.

[239] Assefa S, Keane TM, Otto TD, et al. ABACAS: algorithm-based automatic

contiguation of assembled sequences. Bioinformatics. 2009; 25: 1968-69.

[240] Kosugi S, Hirakawa H, Tabata S. GMcloser: closing gaps in assemblies

accurately with a likelihood-based selection of contig or long-read alignments.

Bioinformatics. 2015; 31: 3733-41.

[241] Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics.

2014; 30: 2068-69.

[242] Harris SR, Feil EJ, Holden MTG, et al. Evolution of MRSA during hospital

transmission and intercontinental spread. Science. 2010; 327: 469-74.

[243] Mac Aogain M, Moloney G, Kilkenny S, et al. Whole-genome sequencing

improves discrimination of relapse from reinfection and identifies transmission events

among patients with recurrent Clostridium difficile infections. Journal of Hospital

Infection. 2015; 90: 108-16.

[244] Li H, Handsaker B, Wysoker A, et al. The sequence alignment/map format

and SAMtools. Bioinformatics. 2009; 25: 2078-79.

[245] Danecek P, Auton A, Abecasis G, et al. The variant call format and VCFtools.

Bioinformatics. 2011; 27: 2156-58.

[246] Cingolani P, Platts A, Wang LL, et al. A program for annotating and

predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome

of Drosophila melanogaster strain w(1118); iso-2; iso-3. Fly. 2012; 6: 80-92.

[247] Croucher NJ, Page AJ, Connor TR, et al. Rapid phylogenetic analysis of

large samples of recombinant bacterial whole genome sequences using Gubbins.

Nucleic Acids Research. 2015; 43: e15.

[248] Stamatakis A. RAxML-VI-HPC: maximum likelihood-based phylogenetic

analyses with thousands of taxa and mixed models. Bioinformatics. 2006; 22: 2688-90.

[249] WA Health. 2017. Quarterly report- Healthcare infection surveillance Western

Australia (HISWA) Quarter 4, April - June, 2016-2017. Department of Health,

Government of Western Australia. Web citation:

https://ww2.health.wa.gov.au/Articles/F_I/Infectious-disease-data/Healthcare-Infection-

Surveillance-Western-Australia-HISWA-reports (last accessed November 2017).

https://ww2.health.wa.gov.au/Articles/F_I/Infectious-disease-data/Healthcare-Infection-Surveillance-Western-Australia-HISWA-reports


References

185

[250] Grass JE, Gould LH, Mahon BE. Epidemiology of foodborne disease

outbreaks caused by Clostridium perfringens, United States, 1998-2010. Foodborne

Pathogens and Disease. 2013; 10: 131-6.

[251] ABS. 2011. Land management and farming in Australia, 2009-10. Australian

Bureau of Statistics (ABS), Australian Government. Web citation:

http://www.abs.gov.au/Ausstats/[email protected]/0/DBBD03A1790B04DDCA257B970015E5

F4?OpenDocument (lass accessed November 2017).

[252] Mutlu E, Wroe AJ, Sanchez-Hurtado K, et al. Molecular characterization and

antimicrobial susceptibility patterns of Clostridium difficile strains isolated from

hospitals in south-east Scotland. Journal of Medical Microbiology. 2007; 56: 921-9.

[253] Spigaglia P, Barbanti F, Mastrantonio P. Multidrug resistance in European

Clostridium difficile clinical isolates. Journal of Antimicrobial Chemotherapy. 2011; 66:

2227-34.

[254] Androga GO, McGovern AM, Elliott B, et al. Evaluation of the Cepheid

Xpert C. difficile/Epi and meridian bioscience illumigene C. difficile assays for

detecting Clostridium difficile ribotype 033 strains. Journal of Clinical Microbiology.

2015; 53: 973-5.

[255] Mc Govern AM, Foster NF, Pereira LA, et al. Human Clostridium difficile

infection caused by a livestock-associated PCR ribotype 237 strain in Western

Australia. JMM Case Reports. 2016; 3: e005062.

[256] Moono P, Putsathit P, Knight DR, et al. Persistence of Clostridium difficile

RT 237 infection in a Western Australian piggery. Anaerobe. 2016; 37: 62-66.

[257] MLA. 2015. Guidelines for the safe manufacturing of smallgoods; 2nd edition.

Meat and Livestock Australia (MLA): New South Wales, Australia. Web citation:

https://www.mla.com.au/globalassets/mla-corporate/research-and-

development/program-areas/food-safety/pdfs/guidelines-for-the-safe-manufacture-of-

smallgoods_2nd-edition.pdf (last accessed May 2017).

[258] Lim SC, Foster NF, Riley TV. Susceptibility of Clostridium difficile to the

food preservatives sodium nitrite, sodium nitrate and sodium metabisulphite. Anaerobe.

2016; 37: 67-71.

[259] Rodriguez-Palacios A, Reid-Smith RJ, Staempfli HR, et al. Clostridium

difficile survives minimal temperature recommended for cooking ground meats.

Anaerobe. 2010; 16: 540-42.

http://www.abs.gov.au/Ausstats/[email protected]/0/DBBD03A1790B04DDCA257B970015E5F4?OpenDocument

http://www.abs.gov.au/Ausstats/[email protected]/0/DBBD03A1790B04DDCA257B970015E5F4?OpenDocument

https://www.mla.com.au/globalassets/mla-corporate/research-and-development/program-areas/food-safety/pdfs/guidelines-for-the-safe-manufacture-of-smallgoods_2nd-edition.pdf



References

186

[260] Fekety R, Kim KH, Brown D, et al. Epidemiology of antibiotic-associated

colitis: isolation of Clostridium difficile from the hospital. American Journal of

Medicine. 1981; 70: 906-08.

[261] RSPCA. What happens to bobby calves? Royal Society for the Prevention of

Cruelty to Animals (RSPCA). Web citation: https://www.rspca.org.au/ (last accessed

November 2017).

[262] Romano V, Pasquale V, Krovacek K, et al. Toxigenic Clostridium difficile

PCR ribotypes from wastewater treatment plants in southern Switzerland. Applied and

Environmental Microbiology. 2012; 78: 6643-6.

[263] Steyer A, Gutierrez-Aguirre I, Racki N, et al. The detection rate of enteric

viruses and Clostridium difficile in a waste water treatment plant effluent. Food and

Environmental Virology. 2015; 7: 164-72.

[264] Xu C, Wang D, Huber A, et al. Persistence of Clostridium difficile in

wastewater treatment-derived biosolids during land application or windrow composting.

Journal of Applied Microbiology. 2016; 120: 312-20.

[265] Usui M, Kawakura M, Yoshizawa N, et al. Survival and prevalence of

Clostridium difficile in manure compost derived from pigs. Anaerobe. 2017; 43: 15-20.

[266] Robbins P, Birkenholtz T. Turfgrass revolution: measuring the expansion of

the American lawn. Land Use Policy. 2003; 20: 181-94.

[267] Standards Australia. 2003. Potting mixes AS 3743-2003. Standards Australia;

New South Wales, Australia. Web citation: https://www.standards.org.au/ (last accessed

May 2016).

[268] Standards Australia. 2003. Composts, soil conditioners and mulches AS 4454-

2003. Standards Australia; New South Wales, Australia. Web citation:

https://www.standards.org.au/ (last accessed May 2016).

[269] EPA. 2013. Compost guideline. Environment Protection Authority (EPA), South

Australia, Australia. Web citation: http://www.epa.wa.gov.au/ (last accessed May 2016)

[270] Scotts Gardening Team. 2017. The Aussie homeowner's guide to a lush green

lawn. Scotts Gardening Team. Web citation: https://www.scottsaustralia.com.au/diy-

garden-guides-advice/lawns/guide-to-a-lush-green-lawn/guide-to-a-lush-green-lawn/

(last accessed November 2017).

[271] Rodriguez-Palacios A, Lejeune JT. Moist-heat resistance, spore aging, and

superdormancy in Clostridium difficile. Applied and Environmental Microbiology.

2011; 77: 3085-91.

https://www.standards.org.au/

https://www.standards.org.au/

http://www.epa.wa.gov.au/

References

187

[272] Warriner K, Xu C, Habash M, et al. Dissemination of Clostridium difficile in

food and the environment: significant sources of C. difficile community-acquired

infection? Journal of Applied Microbiology. 2017; 122: 542-53.

[273] WA Health. 2011. Healthcare infection surveillance Western Australia- Annual

report 2010-2011. Department of Health, Government of Western Australia. Web

citation: https://ww2.health.wa.gov.au/Articles/F_I/Infectious-disease-data/Healthcare-

Infection-Surveillance-Western-Australia-HISWA-reports. (last accessed November

2017).





2017).





2017).





2017).

[277] Zaleski KJ, Josephson KL, Gerba CP, et al. Survival, growth, and regrowth

of enteric indicator and pathogenic bacteria in biosolids, compost, soil, and land applied

biosolids. Journal of Residuals Science and Technology. 2005; 2: 49-63.

[278] Knight DR, Riley TV. Clostridium difficile clade 5 in Australia: antimicrobial

susceptibility profiling of PCR ribotypes of human and animal origin. The Journal of

Antimicrobial Chemotherapy. 2016; 71: 2213-7.

[279] Pirs T, Avbersek J, Zdovc I, et al. Antimicrobial susceptibility of animal and

human isolates of Clostridium difficile by broth microdilution. Journal of Medical

Microbiology. 2013; 62: 1478-85.

[280] Marshall BM, Levy SB. Food animals and antimicrobials: impacts on human

health. Clinical Microbiology Reviews. 2011; 24: 718-33.









References

188

[281] Sorensen TL, Blom M, Monnet DL, et al. Transient intestinal carriage after

ingestion of antibiotic-resistant Enterococcus faecium from chicken and pork. New

England Journal of Medicine. 2001; 345: 1161-66.

[282] Donabedian SM, Thal LA, Hershberger E, et al. Molecular characterization

of gentamicin-resistant enterococci in the United States: evidence of spread from

animals to humans through food. Journal of Clinical Microbiology. 2003; 41: 1109-13.

[283] Molbak K, Baggesen DL, Aarestrup FM, et al. An outbreak of multidrug-

resistant, quinolone-resistant Salmonella enterica serotype typhimurium DT104. New

England Journal of Medicine. 1999; 341: 1420-25.

[284] APVMA. 2014. Quality of antimicrobial products sold for veterinary use in

Australia. Australian Pesticides and Veterinary Medicines Authority (APVMA);

Australian Government. Web citation: https://apvma.gov.au/node/11816 (last accessed

November 2017).

[285] Brown KA, Khanafer N, Daneman N, et al. Meta-analysis of antibiotics and

the risk of community-associated Clostridium difficile infection. Antimicrobial Agents

and Chemotherapy. 2013; 57: 2326-32.

[286] Spigaglia P. Recent advances in the understanding of antibiotic resistance in

Clostridium difficile infection. Therapeutic Advances in Infectious Disease. 2016; 3: 23-

42.

[287] Lim SC, Foster NF, Elliott B, et al. High prevalence of Clostridium difficile on

retail root vegetables, Western Australia. Journal of Applied Microbiology. 2018; 124:

585-90.

[288] van den Berg RJ, Claas ECJ, Oyib DH, et al. Characterization of toxin A-

negative, toxin B-positive Clostridium difficile isolates from outbreaks in different

countries by amplified fragment length polymorphism and PCR ribotyping. Journal of

Clinical Microbiology. 2004; 42: 1035-41.

[289] Marsh JW, Tulenko MM, Shutt KA, et al. Multi-locus variable number

tandem repeat analysis for investigation of the genetic association of Clostridium

difficile isolates from food, food animals and humans. Anaerobe. 2011; 17: 156-60.

[290] The Aussie Farms Repository. 2014. Age of animals slaughtered. The Aussie

Farms Repository: New South Wales, Australia. Web citation:

https://www.aussiefarms.org.au/kb/abattoirs/48-age-animals-slaughtered (last accessed

December 2017).

[291] Food Standards Australia and New Zealand. 2015. Food Standard Code-

Standard 4.2.3- Production and processing standard for meat. Food Standards Australia

https://apvma.gov.au/node/11816

References

189

and New Zealand: Canberra, Australia. Web citation: http://www.foodstandards.gov.au

(last accessed January 2018).

[292] CSIRO. 2007. Australian standard for the hygienic production and

transportation of meat and meat products for human consumption- FRSC Technical

Report No. 3- AS 4696:2007. Commonwealth Scientific and Industrial Research

Organisation (CSIRO): Canberra, Australia. Web citation: https://www.csiro.au/ (last

accessed May 2017).

[293] Sindelar JJ, Milkowski AL. Human safety controversies surrounding nitrate

and nitrite in the diet. Nitric Oxide. 2012; 26: 259-66.

[294] Food Science Australia. 2006. Meat technology- information sheet- Sulphur

dioxide, sulphites in meat products. Food Science Australia: Canberra, Australia. Web

citation: http://www.meatupdate.csiro.au (last accessed November 2016).

[295] Food Standards Australia and New Zealand. 2018. Food Standard Code-

Standard 1.3.1- Food Additives, 2007-2008. Food Standards Australia and New

Zealand: Canberra, Australia. Web citation: http://www.foodstandards.gov.au (last

accessed June 2015).

[296] Food Standards Australia and New Zealand. 2005. Initial Assessment

Report- Proposal P298- Benzoate and sulphite permissions in food of 3 August 2005.

Food Standards Australia and New Zealand: Canberra, Australia. Web citation:

http://www.foodstandards.gov.au (last accessed June 2015).

[297] FDA. 2015. Code of Federal Regulations Title 21 Volume 3 Part 170- Food

additives. Food and Drug Administration (FDA): Maryland, United States. Web

citation: https://www.fda.gov/default.htm (last accessed June 2015).

[298] EU Law and Publications. 2006. Directive 2006/52/EC of the European

Parliament and of the Council of 5 July 2006 amending Directive 95/2/EC on food

additives other than colours and sweeteners and Directive 94/35/EC on sweeteners for

use in foodstuffs, L204 of 26 July 2006. EU Law and Publications: Brussels, Belgium.

Web citation: https://publications.europa.eu/en/home (last accessed June 2015).

[299] Labbe RG, Duncan CL. Growth from spores of Clostridium perfringens in

presence of sodium nitrite. Applied Microbiology. 1970; 19: 353-59.

[300] Duncan CL, Foster EM. Effect of sodium nitrite, sodium chloride and sodium

nitrate on germination and outgrowth of anaerobic spores. Applied Microbiology. 1968;

16: 406-11.

https://www.csiro.au/

http://www.meatupdate.csiro.au/

References

190

[301] Keto-Timonen R, Lindstrom M, Puolanne E, et al. Inhibition of toxigenesis

of group II (nonproteolytic) Clostridium botulinum type B in meat products by using a

reduced level of nitrite. Journal of Food Protection. 2012; 75: 1346-49.

[302] Skovgaard N. Microbiological aspects and technological need- technological

needs for nitrates and nitrites. Food Additives and Contaminants. 1992; 9: 391-97.

[303] Hustad GO, Cerveny JG, Trenk H, et al. Effect of sodium nitrite and sodium

nitrate on botulinal toxin production and nitrosamine formation in wieners. Applied


[304] Christiansen LN, Johnston RW, Kautter DA, et al. Effect of nitrite and

nitrate on toxin production by Clostridium botulinum and on nitrosamine formation in

perishable canned comminuted cured meat. Applied Microbiology. 1973; 25: 357-62.

[305] Zimomra J, Wenderoth T, Snyder A, et al. Clostridium perfringens

gastroenteritis associated with corned beef served at St. Patrick’s Day meals – Ohio and

Virginia, 1993. Morbidity and Mortality Weekly Report. 1994; 43: 137-38.

[306] Golden NJ, Crouch EA, Latimer H, et al. Risk assessment for Clostridium

perfringens in ready-to-eat and partially cooked meat and poultry products. Journal of

Food Protection. 2009; 72: 1376-84.

[307] Gillespie I, Little C, Mitchell R. Microbiological examination of cold ready-to-

eat sliced meats from catering establishments in the United Kingdom. Journal of

Applied Microbiology. 2000; 88: 467-74.

[308] McKay KA, Kumchy PI. Incidence and significance of Clostridia isolated from

ready-to-eat meats and rendered products. Canadian Journal of Comparative Medicine.

1969; 33: 307-09.

[309] CDC. 2011. Burden of foodborne illness: Findings. Centers for Disease Control

and Prevention (CDC): Atlanta, United States. Web citation:

https://www.cdc.gov/foodborneburden/2011-foodborne-estimates.html (last accessed

January 2018).

[310] Hensgens MPM, Keessen EC, Squire MM, et al. Clostridium difficile

infection in the community: a zoonotic disease? Clinical Microbiology and Infection.

2012; 18: 635-45.

[311] Fawley WN, Davies KA, Morris T, et al. Enhanced surveillance of

Clostridium difficile infection occurring outside hospital, England, 2011 to 2013.

Eurosurveillance. 2016; 21: 30295.

[312] Huber CA, Hall L, Foster NF, et al. Surveillance snapshot of Clostridium

difficile infection in hospitals across Queensland detects binary toxin producing

References

191

ribotype UK 244. Communicable Diseases Intelligence Quaterly Report. 2014; 38:

E279-84.

[313] Gerding DN, Muto CA, Owens RC, Jr. Measures to control and prevent

Clostridium difficile infection. Clinical Infectious Diseases. 2008; 46: S43-S49.

[314] Kotila SM, Pitkanen T, Brazier J, et al. Clostridium difficile contamination of

public tap water distribution system during a waterborne outbreak in Finland.

Scandinavian Journal of Public Health. 2013; 41: 541-5.

[315] Moono P, Lim SC, Riley TV. High prevalence of toxigenic Clostridium

difficile in public space lawns in Western Australia. Scientific Reports. 2017; 7: 41196.

[316] Stone NE, Sidak-Loftis LC, Sahl JW, et al. More than 50% of Clostridium

difficile isolates from pet dogs in Flagstaff, USA, carry toxigenic genotypes. PloS one.

2016; 11: e0164504.

[317] Weese JS, Finley R, Reid-Smith RR, et al. Evaluation of Clostridium difficile

in dogs and the household environment. Epidemiology and Infection. 2010; 138: 1100-

04.

[318] Riley TV, Adams JE, Oneill GL, et al. Gastrointestinal carriage of Clostridium

difficile in cats and dogs attending veterinary clinics. Epidemiology and Infection. 1991;

107: 659-65.


summary report 2014-2015. Department of Health, Government of Western Australia.

Web citation: https://ww2.health.wa.gov.au/Articles/F_I/Infectious-disease-

data/Healthcare-Infection-Surveillance-Western-Australia-HISWA-reports (last

accessed November 2017).

[320] Krutova M, Matejkova J, Drevinek P, et al. Increasing incidence of

Clostridium difficile ribotype 001 associated with severe course of the infection and

previous fluoroquinolone use in the Czech Republic, 2015. European Journal of

Clinical Microbiology and Infectious Diseases. 2017; 36: 2251-58.

[321] Hawkey PM, Marriott C, Liu WE, et al. Molecular epidemiology of

Clostridium difficile infection in a major Chinese hospital: an underrecognized problem

in Asia? Journal of Clinical Microbiology. 2013; 51: 3308-13.

[322] Tan XQ, Verrall AJ, Jureen R, et al. The emergence of community-onset

Clostridium difficile infection in a tertiary hospital in Singapore: a cause for concern.

International Journal of Antimicrobial Agents. 2014; 43: 47-51.



References

192

[323] ESRI Australia. 2017. Where does Australia import its food from? ESRI

Australia: Queensland, Australia. Web citation:

http://esriau.maps.arcgis.com/home/index.html (last accessed Feb 2018).

[324] Putsathit P. Epidemiology of Clostridium difficile infection in Thailand. The

University of Western Australia, 2017.

[325] Bandelj P, Golob M, Ocepek M, et al. Antimicrobial susceptibility patterns of

Clostridium difficile isolates from family dairy farms. Zoonoses and Public Health.

2017; 64: 213-21.

[326] Seymour IJ, Burfoot D, Smith RL, et al. Ultrasound decontamination of

minimally processed fruits and vegetables. International Journal of Food Science and

Technology. 2002; 37: 547-57.

[327] Rodriguez-Palacios A, LeJeune JT. Moist heat resistance, spore aging, and

superdormancy in Clostridium difficile. Applied and Environmental Microbiology.

2011.

[328] McFarland LV, Mulligan ME, Kwok RYY, et al. Nosocomial acquisition of

Clostridium difficile infection. New England Journal of Medicine. 1989; 320: 204-10.

[329] Weber DJ, Rutala WA, Miller MB, et al. Role of hospital surfaces in the

transmission of emerging health care-associated pathogens: norovirus, Clostridium

difficile, and Acinetobacter species. American Journal of Infection Control. 2010; 38:

S25-S33.

[330] Androga GO, Hart J, Foster NF, et al. Infection with toxin A-negative, toxin

B-negative, binary toxin-positive Clostridium difficile in a young patient with ulcerative

colitis. Journal of Clinical Microbiology. 2015; 53: 3702-04.

[331] MLA. 2015. Market information and industry insights - Australian cattle

industry projections 2015. Meat and Livestock Australia (MLA): New South Wales,

Australia. Web citation: http://www.mla.com.au/About-MLA/News-and-media/Media-

releases/2015-cattle-industry-projections-released (last accessed October 2015).

[332] Deng K, Plaza-Garrido A, Torres JA, et al. Survival of Clostridium difficile

spores at low temperatures. Food Microbiology. 2015; 46: 218-21.

[333] Legendairy. 2018. Our industry, Legendaitry: Victoria, Australia. Web citation:

http://www.legendairy.com.au/dairy-farming/our-industry/our-regions (last accessed

July 2018).

[334] McLure A, Clements ACA, Kirk M, et al. Healthcare-associated Clostridium

difficile infections are sustained by disease from the community. Bulletin of

Mathematical Biology. 2017; 79: 2242-57.

References

193

[335] Squire MM. Clostridium difficile and idiopathic neonatal diarrhoea in

Australian piglets. The University of Western Australia, 2015.

[336] Rupnik M. Is Clostridium difficile-associated infection a potentially zoonotic

and foodborne disease? Clinical Microbiology and Infection. 2007; 13: 457-59.

[337] Jump RLP, Pultz MJ, Donskey CJ. Vegetative Clostridium difficile survives

in room air on moist surfaces and in gastric contents with reduced acidity: a potential

mechanism to explain the association between proton pump inhibitors and C. difficile-

associated diarrhea? Antimicrobial Agents and Chemotherapy. 2007; 51: 2883.

[338] Larcombe S, Hutton ML, Lyras D. Involvement of bacteria other than

Clostridium difficile in antibiotic-associated diarrhoea. Trends in Microbiology. 2016;

24: 463-76.

Appendices

194

APPENDICES

Page 1 of 8

Clostridium difficile Questionnaire

Answering this questionnaire is voluntary. All information you provide will be confidential.

Instructions: Please complete all questions in the questionnaire by writing in the space provided or ticking the box. If you are uncertain about how to respond please leave blank and move on to the next question. If you are completing this questionnaire on behalf of the patient, please indicate below. Where the question refers to ‘you’, please provide the patient’s information.

You are being invited to participate in this study because C. difficile was found in the faecal sample that you submitted, for processing at PathWest, when you visited your local GP and/or the emergency department of a hospital in between January 2014 to March 2015 because of diarrhoea.

A. About the participant

A1. Who is completing this questionnaire?

Patient

Family/relative/friend/carer of the patient

A2. Year of birth:

A3. Gender:

Male

Female

A4. Where do you normally live?

Home

Nursing home

Other. Please specify: _______________________________________

A5. Postcode:

A6. What is your occupation? _______________________

Official use only:

Unique ID:

A1

Page 2 of 8

B. Food B1. Where do you normally buy your food? (Tick all boxes that apply)

Coles supermarket

Woolworths supermarket

IGA supermarket

Fremantle market

Asian supermarket

Butcher shop

Non-organic vegetable shop

Weekend vegetable market. Please specify: __________________________________________

Organic vegetable shop. Please specify: __________________________________________

Other. Please specify: __________________________________________ B2. Did you eat any cooked organic vegetables in the 2 weeks before getting C. difficile-related-diarrhoea? Organic vegetable: e.g. vegetable brought from organic shop, organic vegetable section of Woolworths supermarket and stall in farmer market that claimed to sell organic vegetables.

Yes (Please go to B3)

No (Please go to B4) B3. How often do you eat the following cooked organic vegetables before getting C. difficile-related-diarrhoea? Please fill only one box in every line.

Cooked organic

1 time

2 - 3 times

1 - 2 times

3 - 6 times

< 1 times

1 - 3 times

Never

per day per week per month

Onions Beetroot Carrots Potatoes Spring onions Garlic Sprouts Asparagus Celery Lettuce Pre-packed salads Home grown vegetables Others B4. Did you eat any raw organic vegetables in the 2 weeks before getting C. difficile-related-diarrhoea? Raw organic vegetable: e.g. uncooked raw organic salad and fresh organic vegetable juice.


No (Please go to B6)

A2

Page 3 of 8

B5. How often do you eat the following raw organic vegetables before getting C. difficile-related-diarrhoea? Please fill only one box in every row.

Raw organic

1 time

2 - 3 times

1 - 2 times

3 - 6 times

< 1 times

1 - 3 times

Never


Onions Beetroot Carrots Potatoes Spring onions Garlic Sprouts Asparagus Celery Lettuce Pre-packed salads Home grown vegetables Others B6. Did you eat any cooked non-organic vegetables in the 2 weeks before getting C. difficile-related-diarrhoea? Non-organic vegetable: All vegetables that were not labelled as organic.


No (Please go to B8) B7. How often do you eat the following cooked non-organic vegetables before getting C. difficile-related-diarrhoea? Please fill only one box in every row.

Cooked non-organic

1 time

2 - 3 times

1 - 2 times

3 - 6 times

< 1 times

1 - 3 times

Never


Onions Beetroot Carrots Potatoes Spring onions Garlic Sprouts Asparagus Celery Lettuce Pre-packed salads Home grown vegetables Others

A3

Page 4 of 8

B8. Did you eat any raw non-organic vegetables in the 2 weeks before getting C. difficile-related-diarrhoea?


No (Please go to B10) B9. How often do you eat the following raw non-organic vegetables before getting C. difficile-related-diarrhoea? Please fill only one box in every row.

Raw non-organic

1 time

2 - 3 times

1 - 2 times

3 - 6 times

< 1 times

1 - 3 times

Never


Onions Beetroot Carrots Potatoes Spring onions Garlic Sprouts Asparagus Celery Lettuce Pre-packed salads Home grown vegetables Others B10. Did you eat any processed meat in the 2 weeks before getting C. difficile-related-diarrhoea? Processed meat: Ready-to-eat meat such as salami, ham and cooked sausage.


No (Please go to C1) B11. How often do you eat the following processed meat before getting C. difficile-related-diarrhoea? Please fill only one box in every row.

Processed meat

1 time

2 - 3 times

1 - 2 times

3 - 6 times

< 1 times

1 - 3 times

Never


Salami Ham Meatball Sausage roll Other cured meats

A4

Page 5 of 8

C. Animal contact C1. Did you have any household pets living with you in the 2 weeks before getting C. difficile-related-diarrhoea?

Yes (Please go to C2)

No (Please go to C3)

C2. What household pets did you have?

Dog

Cat

Rabbit

Guinea pig / hamster

Mouse / rat

Bird

Chickens / other poultry

Horse / pony / donkey

Fish

Turtle

Reptile (e.g. snake)

Other. Please specify: __________________ C3. Did you have any physical contact with other people’s household pets in the 2 weeks before getting C. difficile-related-diarrhoea?


No (Please go to C5) C4. What household pets did you have physical contact with in the 2 weeks before getting C. difficile-related-diarrhoea?

Dog

Cat

Rabbit


Mouse / rat

Bird

Chickens / other poultry


Fish

Turtle

Reptile (e.g. snake)

Other. Please specify: __________________ C5. Did any of the household pets have diarrhoea at the time you had contact with them?


No (Please go to C6)

A5

Page 6 of 8

C6. Did you live on a farm with animals in the 2 weeks before getting C. difficile-related-diarrhoea?


No (Please go to C7) C7. Did you have any physical contact with any other animals (including farm animals and wild animals) in the 2 weeks before getting C. difficile-related-diarrhoea?

Yes (Please go to C8 and C9)

No (Please go to C10) C8. Which of the following animals did you have contact with in the 2 weeks before getting C. difficile-related-diarrhoea?


Chickens / poultry

Cattle (including dairy cows / calves)

Pigs / piglets

Sheep / lamb

Deer

Working dog

Farm cat

Birds

Other animals in a zoo or wildlife park

Other. Please specify: ____________________ C9. What kind of contact did you have with the animals, was it:

General farm work with animals

Job related (e.g. slaughterhouse worker, veterinarian, and animal trainer)

Feeding

Petting / stroking

Cleaning / bathing the animals, grooming

Cleaning cages / kennels / stables

Picking up manure / faeces

Giving medication

Other. Please specify: ___________________________ C10. Did you have any intentional or accidental contact with animal manure / faeces or compost containing animal manure / faeces in the 2 weeks before getting C. difficile-related-diarrhoea?


No (Please go to D1)

C11. What type of animal manure / faeces was this?

Bird droppings

Cattle / calf

Sheep / lamb


A6

Page 7 of 8

Chicken / poultry

Pigs

Dog

Cat

Rabbit

Mouse / rat


Compost from garden centre

Fertilizer or compost made from animal manure

Other. Please specify: ____________________

D. Contact with gardening products D1. Did you do any gardening in the 2 weeks before getting C. difficile-related-diarrhoea?

Yes (Please go to D2)

No (Please go to D2) D2. Did you have any contact with the following gardening products in the 2 weeks before getting C. difficile-related-diarrhoea?

Gardening products Yes No Unsure

Garden mix (Example: potting mix and vegetable mix) Mulch Compost (Example: organic compost and mushroom compost) Manure

Cow manure Sheep manure Chicken manure Horse manure Pig manure Blended manure (Example: mix of cow and sheep manure) Other

Blood and bone fertiliser Organic fertiliser Non-organic fertiliser Newly laid lawn turf

A7

Page 8 of 8

E. Human contact E1. Did you have any contact with any of the following people in the 2 weeks before getting C. difficile-related-diarrhoea?

Human contact Yes No Unsure

Baby or young child under the age of 2 years old Healthcare worker Childcare worker A person who had diarrhoea in the last 3 months A person who has been to hospital in the last 3 months A person who lived in a nursing home in the last 3 months A person who works with animals

F. Other factors F1. Did you visit any of the following places in the 2 weeks before getting C. difficile-related-diarrhoea?

Factors Yes No Unsure

Hospital Medical clinic Aged care facility Childcare facility Pre-school Veterinary clinic Animal farm Zoo Farmland F2. Is there any other information that you would like to share, which you think might be related to your C. difficile-related-diarrhoea? Example: Any underlying illness, antibiotic usage and domestic or international travel.

F3. Would you be willing to be contacted by telephone if we have any further questions?

Yes. Please provide your phone number: ______________________________________________

No

Thank you for taking the time to complete this questionnaire. Please place your completed questionnaire into the pre-paid envelope and post it back to us. Please keep the participant information form for your own records.

A8

ORIGINAL ARTICLE

High prevalence of Clostridium difficile on retail rootvegetables, Western AustraliaS.C. Lim1 , N.F. Foster2, B. Elliott3 and T.V. Riley1,2,3,4

1 The University of Western Australia, Nedlands, WA, Australia

2 PathWest Laboratory Medicine, Nedlands, WA, Australia

3 Edith Cowan University, Joondalup, WA, Australia

4 Murdoch University, Murdoch, WA, Australia

Keywords

animals, Australia, Clostridium difficile,

diarrhoea, prevalence, root vegetables.

Correspondence

Thomas V. Riley, Department of Microbiology,

PathWest Laboratory Medicine, Queen Eliza-

beth II Medical Centre, Nedlands 6009, WA,

Australia.

E-mail: [email protected]

2017/1683: received 24 August 2017, revised

30 October 2017 and accepted 3 November

2017

doi:10.1111/jam.13653

Abstract

Aims: The incidence of community-associated Clostridium difficile infection

(CA-CDI) in Australia has increased since mid-2011. With reports of clinically

important C. difficile strains being isolated from retail foods in Europe and

North America, a foodborne source of C. difficile in cases of CA-CDI is a

possibility. This study represents the first to investigate the prevalence and

genotypes of C. difficile in Australian retail vegetables.

Methods and Results: A total of 300 root vegetables grown in Western

Australia (WA) were collected from retail stores and farmers’ markets. Three

vegetables of the same kind bought from the same store/market were treated as

one sample. Selective enrichment culture, toxin profiling and PCR ribotyping

were performed. Clostridium difficile was isolated from 30% (30/100) of pooled

vegetable samples, 55�6% of organic potatoes, 50% of nonorganic potatoes,

22�2% of organic beetroots, 5�6% of organic onions and 5�3% of organic

carrots. Over half (51�2%, 22/43) the isolates were toxigenic. Many of the

ribotypes of C. difficile isolated were common among human and Australian

animals.

Conclusions: Clostridium difficile could be found commonly on retail root

vegetables of WA. This may be potential sources for CA-CDI.

Significance and Impact of the Study: This study enhances knowledge of

possible sources of C. difficile in the Australian community, outside the

hospital setting.

Introduction

Clostridium difficile is a well-known cause of healthcare-

associated infectious diarrhoea (Slimings et al. 2014).

Over the past decade, the incidence of community-asso-

ciated C. difficile infection (CA-CDI) has increased

worldwide (Freeman et al. 2010) and CA-CDI currently

accounts for c. 26% of all CDI cases in Australia (Slim-

ings et al. 2014). Similar changes have been seen in other

parts of the developed world (Freeman et al. 2010).

Recent changes in the global epidemiology of CDI have

been mainly attributed to the emergence of so-called

‘hypervirulent’ strains of C. difficile, however, the reasons

for an increase in CA-CDI in individuals without the

traditional risk factors such as old age, recent hospital

stay and antimicrobial exposure remain unclear (Slimings

et al. 2014).

One possible source of C. difficile in the community is

food contaminated with C. difficile. C. difficile ribotype

(RT) 078 is the predominant strain found colonizing the

gastrointestinal tracts of production animals in the

Northern Hemisphere, and this strain has been isolated

from retail foods in Canada and the United States (Weese

2010a). Over the last decade, C. difficile RT 078 has

become a common RT found in human infection in Eur-

ope (Bakker et al. 2010) and a rising cause of CDI in the

United States (Limbago et al. 2009). Although some pub-

lications report up to 42% prevalence of clinically

Journal of Applied Microbiology 124, 585--590 © 2017 The Society for Applied Microbiology 585

Journal of Applied Microbiology ISSN 1364-5072

A9

important C. difficile strains in retail foods (Songer et al.

2009), most give much lower prevalence figures, particu-

larly in Europe (Lund and Peck 2015). Nonetheless, it

has been hypothesized that CA-CDI could be a foodborne

infection (Weese 2010a).

Recent reports have indicated a high rate of gastroin-

testinal carriage of C. difficile in Australian cattle (Knight

et al. 2013a) and pigs (Knight et al. 2015). To date, no

study has investigated the prevalence of C. difficile in

Australian foods although significant contamination of

veal calf carcasses has been shown to occur at slaughter

(Knight et al. 2013a). In this study, we investigated the

prevalence of C. difficile contaminating Australian-grown

root vegetables. Isolates of C. difficile were characterized

by PCR ribotyping and toxin profiling.

Materials and methods

Sample collection and preparation

A total of 300 root vegetables grown in Western Australia

(WA), including organic potatoes (n = 81), organic carrots

(n = 57), organic beetroots (n = 54), organic onions

(n = 54) and nonorganic potatoes (n = 54), were pur-

chased from 24 retail stores and seven farmers’ markets

between March 2015 and November 2015. From each retail

store and farmers’ market, three of each vegetable were

purchased depending on availability, except store (W1)

where two varieties of organic potatoes were purchased

(three each) on the same day. All vegetables were placed

into separate bags to prevent cross-contamination. To

reduce the possibility of laboratory contamination, the ini-

tial preparation of the vegetables was performed before

transportation to the laboratory. A vegetable peeler was

used to collect potato skin, and a sterile disposable scalpel

to harvest the basal plate and roots of onions, and the

crown and taproot of carrots and beetroots. Three vegeta-

bles of the same kind bought from the same store or mar-

ket, were then pooled together and treated as one sample.

The peeler was soaked with bleach solution (6000 ppm free

chlorine) for 1 min between samples as this concentration

of bleach and exposure time has been shown to be effective

in killing C. difficile spores (Hacek et al. 2010), and a new

scalpel was used between samples for other vegetables. A

nonporous tile was used as a chopping board and treated

with bleach solution between samples.

Culture, isolation and identification of C. difficile

Briefly, vegetable parts and 10 g of potato skin were trans-

ferred to 90 ml of BHIB supplemented with 5 g l�1 yeast

extract, 1 g l�1 L-cysteine, 1 g l�1 taurocholic acid,

250 mg l�1 cycloserine and 8 mg l�1 cefoxitin (PathWest

Media, Mt Claremont, WA, Australia). A negative control,

10 ml of phosphate-buffered saline added to 90 ml of

enrichment broth, was included in each round of sampling

to monitor potential contamination. After anaerobic

incubation for 10 days, alcohol shock was performed on

2 ml of enrichment broth by the addition of 2 ml of

absolute alcohol. After 1 h, the suspension was centrifuged

at 3800 g for 10 min and 10 ll of sediment plated on

C. difficile ChromID™ agar (BioM�erieux, Marcy I’ Etoile,

France) as described previously (Boseiwaqa et al. 2013).

Plates were incubated anaerobically in a Don Whitley Sci-

entific Ltd (Otley, Yorkshire, UK) A35 anaerobic chamber

with an atmosphere of 10% hydrogen, 10% carbon diox-

ide and 80% nitrogen, and examined at 24 and 48 h. Ten

presumptive C. difficile colonies; small, irregular and with

a raised umbonate profile, coloured or not (Boseiwaqa

et al. 2013), were subcultured per ChromID plate onto

separate prereduced blood agar plates for identification

based on their colony morphology of ground glass appear-

ance, opaque, greyish-white and nonhaemolytic, character-

istic chartreuse fluorescence under long-wave UV light

(360 nm) and characteristic horse dung odour (Knight

et al. 2013a). The identity of uncertain isolates was

confirmed by the presence of L-proline aminopeptidase

activity (Rosco Diagnostica, Tasstrup, Denmark).

Toxin profiling and PCR ribotyping

PCR toxin profiling and ribotyping were performed as

previously described (Knight et al. 2013a) with slight

modification. Briefly, a multiplex PCR was used to detect

tcdA. The primers included NK2 and NK3 from Kato

et al. (1991) to detect the A1 region and novel primers

(B. Elliott et al. unpublished data), tcdA-1 (CAGT-

CACTGGATGGAGAATT) and tcdA-2 (AAGGCAA-

TAGCGGTATCAG), to detect the A3 region. Each

reaction consisted of 4 ll template DNA, 2 mmol l�1

MgCl2, 19 buffer II, 0�01% (w/v) BSA, 200 lmol l�1

each dNTP, 0�75 U of Taq polymerase and

0�2 lmol l�1 of each primer in a final volume of 20 ll.The PCR program consisted of 95°C for 10 min then 35

cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 90 s,

then a final extension at 72°C for 7 min. Detection of

both tcdA regions was required for an isolate to be con-

sidered tcdA+. Isolates which did not correspond to any

internationally recognized RTs in our library collection of

54 internationally recognized RTs were assigned an inter-

nal nomenclature prefixed with QX.

Statistical analysis

Fisher’s exact test was performed to compare the preva-

lence of C. difficile on vegetables.

Journal of Applied Microbiology 124, 585--590 © 2017 The Society for Applied Microbiology586

C. difficile on root vegetables S.C. Lim et al.

A10

Results

Clostridium difficile was found in 30% of the pooled root

vegetable samples; 55�6% (15 of 27 samples) of organic

potatoes, 50% (9 of 18) of nonorganic potatoes, 22�2%(4 of 18) of organic beetroots, 5�6% (1 of 18) of organic

onions and 5�3% (1 of 19) of organic carrots. There was

no significant difference between prevalence in potatoes

grown organically and potatoes not grown organically,

however, a greater proportion of potato samples was pos-

itive for C. difficile compared to other vegetables

(P < 0�0001). Of the positive vegetables, 20�0% (3 of 15)

of organic potatoes, 22�2% (two of nine) of nonorganic

potatoes and 25�0% (one of four) of organic beetroots

were contaminated with more than one C. difficile

ribotype.

Twenty-four RTs of C. difficile were identified (Fig. 1),

13 of which were internationally recognized RTs. Half of

the isolated strains were toxigenic (51�2%, 22 of 43), pre-

dominantly A+B+CDT� (81�8%, 18 of 22). Nontoxigenic

strains comprised 48�8% (21 of 43) of isolates. Clostrid-

ium difficile RT QX 274 was the only isolate with toxin

profile A+B+CDT+. RT 027 and RT 078 strains were not

found. The most common RT was nontoxigenic RT 051

which comprised 14�0% (6 of 43) of isolates. The next

most prevalent RTs were QX 145 (11�6%), 056 (9�3%),

QX 393 (7�0%), 014/020 (4�7%), 101 (4�7%), QX 049

(4�7%), QX 142 (4�7%) and QX 545 (4�7%; Fig. 1).

Discussion

The prevalence of C. difficile contamination of root veg-

etables in the present study was higher than reported in

other countries, ranging from at least 10% (30 of 300

vegetables) to 30% (90 of 300 vegetables) due to pooling

of three vegetables into a single sample. In France, 2�9%(3 of 104) of ready-to-eat salads and pea sprouts were

positive for C. difficile (Eckert et al. 2013); while 4�5% (5

of 111) and 7�5% (3 of 40) of retail vegetables in Canada

(Metcalf et al. 2010) and Scotland (Bakri et al. 2009),

respectively, were contaminated with C. difficile. The

higher prevalence in the present study may be due to a

number of factors. First, this was a study of root vegeta-

bles while other studies were mainly of leaf vegetables or

Sample (n)

Onion α(1)

Potato α(1)

Potato α(1)

Potato α(1)

Potato α(1)

Potato α(1)

Potato α(1)

Potato α(1)

Potato α(2)

Potato β(2)

Beetroot α(1)

Beetroot α(1)

Potato α(3) β(2); beetroot α(1)

Potato α(1) β(1)

Potato a(1) β(1)

Potato β(1)

Potato α(1) β(1); beetroot α(1)

Potato α(3) β(1)

Beetroot α(1)

Beetroot α(1)

Potato β(1)

Potato α(1) β(1)

Potato α(3); carrot α(1)

Potato β(1)

Store code (n)

W5: (1)

W7: (1)

W13: (1)

W22: (1)

W11: (1)

W10: (1)

W21: (1); W26: (1)

W2: (1)

W7: (1)

W2: (2)‡; W7: (1); W13: (1)

W2: (1)

W2: (1); W11: (1)

W26: (1); W28: (1)

W13: (1)

W10: (1)

W1: (2)+; W7: (1); W15: (1); W20: (1); W30: (1)

W6: (1)+; W7: (1); W8: (1); W12: (1); W22: (1)

W1: (1); W24: (1)

W29: (1)

W1: (1)

W2: (1); W29: (1)

W29: (1)

W1: (1)

11·6

2·3

2·3

2·3

4·7

4·7

4·7

4·7

4·7

14·0

2·3

2·3

2·3

2·3

9·3

2·3

2·3

2·3

2·3

2·3

2·3

2·31

1

1

1

1

1

1

1

1

1

1

6

2

1

3

1

2

1

1

5

43Total

–

–

–

–

–

–

–

–

–

–

–

–

–

–

+

–

–

–––

––

–

–

–

–

+

–

––

–– –

–

++

+

–

––

–

–

–

–

–

+

+

+

–

–

+

+

+

+

+

+

+

+

++

+

+

+

+

+

+

+

+

+

+

++

4

2

2

2

7·0

2·3

W1: (1);W21: (1); W26: (1)

%n

Toxin profile

PCR ribotype

QX 072

60 80 100

UK 002

UK 033*

UK 137*

QX 519*

QX 525

QX 049

QX 518

UK 064

UK 056*

QX 274

QX 545

UK 012

UK 010

UK 051

UK 101*

UK 070

QX 393

UK 237*

QX 142

QX 551

QX 145

UK 584

UK 014/020*

tcdA tcdB cdtA/cdtB

Figure 1 Clostridium difficile PCR ribotypes and toxin gene profiles isolated from Western Australian vegetables. PCR ribotype pattern analysis

was performed by creating a neighbour-joining tree, using the Pearson correlation (optimization, 5%; curve smoothing, 1%). *Clostridium diffi-

cile PCR ribotypes common in humans and/or Australian production animals; a, organic; b, nonorganic; †, two varieties of organic potatoes (pur-

chased from W1 on the same day in October 2015) were positive for UK 051; ‡a sample of organic carrots (purchased in March 2015) and

organic potatoes (purchased in August 2015) from W2 were positive for C. difficile UK 056. Isolates which did not correspond to any internation-

ally recognized RTs in our library collection were given an internal nomenclature prefixed with QX.


S.C. Lim et al. C. difficile on root vegetables

A11

vegetables cultivated above ground. Root vegetables are

likely to have more soil residue on their surfaces and this

may have contributed to the high prevalence, as C. diffi-

cile can be abundant in agricultural soil (AlSaif and Bra-

zier 1996; Simango 2006). With potatoes, for example,

sampling a greater surface area covered in soil may have

led to better detection of C. difficile contamination. The

culture method used is likely to have been more sensitive

as three vegetables of the same kind were pooled and

treated as one sample, so as little as one C. difficile-posi-

tive vegetable would result in a positive sample. However,

even if that was the case, the prevalence reported in this

study would still be higher (≥10%) compared to other

studies. The present study also used a bigger volume of

enrichment broth compared to other studies, as well as a

longer incubation time (Bakri et al. 2009; Metcalf et al.

2010; Eckert et al. 2013). Currently, there are no interna-

tional standards for isolating C. difficile from food,

although the method used was similar to the US

CDC-recommended method for meat (Limbago et al.

2012) except for the longer incubation time.

It is also possible that the higher prevalence of C. diffi-

cile on vegetables reflects different vegetable growing and

processing practices in Australia which may have resulted

in contamination of produce by animal manure. In

Australia, c. 1�7 million tonnes per year of animal manure

is applied to agricultural land as fertilizer, 6�8%(115 989 t) of which is used for farming in WA (Aus-

tralian Bureau of Statistics 2011). This includes manure

from chicken, cattle, pig and sheep farms. In this study,

many of the C. difficile RTs identified on the vegetables

are commonly found in Australian production animals.

Of the RTs represented among the 43 food isolates

(Fig. 1), C. difficile RT 014/020 was also the most com-

mon strain in Australian piglets (23�4%) in a nationwide

surveillance study (Knight et al. 2015). In addition, RT

014/020 was the most prevalent strain in human clinical

samples in WA between October 2011 and September

2012, accounting for 30% (99 of 330) of CDI cases

(Cheng et al. 2016). Clostridium difficile RT 056, which

constituted 9�3% (4 of 43) of the food isolates, was the

fourth most prevalent strain in humans (3�9%; 13 of 330;

Cheng et al. 2016) and the third most prevalent strain in

Australian veal calves (7�7%; 16 of 209; Knight et al.

2013a). Other C. difficile RTs isolated with an epidemio-

logical link to Australian livestock include RTs 101, 137,

033 and 237. Ribotypes 101 (40%; 6 of 15) and 137

(13�3%; 2 of 15) were the two most prevalent RTs in

lambs (Knight and Riley 2013b), RT 033 was the second

most prevalent ribotype in both veal calves (19�6%; 41 of

209; Knight et al. 2013a) and piglets (13%; 20/154;

Knight et al. 2015), while RT 237 is a pig strain unique

to one piggery in WA (Knight et al. 2015). All of these

RTs of C. difficile have been isolated from human cases

of CDI in Australia (Androga et al. 2015; Furuya-Kana-

mori et al. 2016; McGovern et al. 2016).

Neither C. difficile RT 027, RT 078 nor RT244 were

found in this study. It is not surprising that RT 027 was

not isolated as it has been isolated infrequently in Aus-

tralia (Richards et al. 2011) and RT 078 has never been

isolated from an Australian production animal (Knight

and Riley 2013b; Knight et al. 2013a, 2015). RT244

emerged in Australia in 2011/2012 (Eyre et al. 2015) as a

cause of severe community-acquired infection and the

pattern of disease across Australia suggested a foodborne

outbreak. During this period RT244 was the third most

common RT of C. difficile detected in Australia (Huber

et al. 2014). It has subsequently declined to undetectable

levels implying that there is no longer a source or reser-

voir in Australia. However, the isolation of other clini-

cally important C. difficile strains from retail vegetables in

WA suggests ongoing foodborne transmission of CDI is

likely. With many of the isolated RTs being common in

animals, the most likely source of contamination is

through the use of animal manure as fertilizer in agricul-

tural farming. However, there are other possible points of

contamination including downstream food processing, at

wholesale, during transport and at the retail market.

Laboratory contamination is unlikely to be responsible

for the high prevalence of C. difficile in the present study

as all samples were prepared using aseptic techniques

prior to transportation to the laboratory for enrichment

culture, and a diverse range of C. difficile PCR RTs was

isolated including RTs novel to our laboratory, QX

518 (A�B�CDT�), QX 519 (A+B+CDT�), QX 525

(A�B�CDT�), QX 545 (A+B+CDT�) and QX 551

(A�B�CDT�). Such diversity is highly unlikely to repre-

sent laboratory contamination.

This study has several limitations. Firstly, the level of

C. difficile contamination on the vegetables was not deter-

mined. The concentration of C. difficile spores on con-

taminated foods is generally presumed to be low, with

studies that report positive samples detecting C. difficile

by enrichment culture only (Weese et al. 2010b) or, if

positive by direct culture, counts of 20–240 spores per g

(Weese et al. 2009). The infective dose of C. difficile is

currently unknown, although suggestion has been made

that it could be low (100–1000 spores) depending on host

susceptibility (Hensgens et al. 2012; Warriner et al.

2017). Secondly, some strains were given internal nomen-

clature (QX types) because they did not correspond to

any internationally recognized ribotypes in our library

collection. Thus, making them impossible to compare

with RTs reported in other studies. Another limitation of

this study was that all the vegetables tested were grown in

WA. For a more thorough understanding of the



A12

prevalence and molecular types of C. difficile on retail

foods in Australia, larger studies involving samples from

multiple States and Territories are necessary. A more

structured sampling regime could identify any potential

seasonality.

This study represents the first to determine the preva-

lence of C. difficile in Australian retail foods. A higher

prevalence of C. difficile was found on retail vegetables in

WA (ranging from ≥10–30%) in comparison to reports

from Europe and North America. The finding of diverse

C. difficile RTs common to livestock and humans suggests

that CDI, especially CA-CDI, might have a foodborne

transmission component in Australia and that the strains

of C. difficile involved may be of animal origins. How-

ever, whether transmission is via food per se, or food

contaminates the environment (e.g. a home kitchen),

remains to be determined. To show foodborne transmis-

sion of CDI, further studies with more discriminatory

typing methods, such as whole-genome sequencing, are

necessary to compare the relatedness of these food

isolates with those of human CDI.

Acknowledgements

We thank Daniel Knight, Katherine Hammer and Stacey

Hong for their help in collecting samples, and Papanin

Putsathit and Stacey Hong for laboratory assistance.

Conflict of Interest

No conflict of interest declared.

References

AlSaif, N. and Brazier, J.S. (1996) The distribution of

Clostridium difficile in the environment of South Wales.

J Med Microbiol 45, 133–137.Androga, G.O., Hart, J., Foster, N.F., Charles, A., Forbes, D.

and Riley, T.V. (2015) Infection with toxin A-negative,

toxin B-negative, binary toxin-positive Clostridium difficile

in a young patient with ulcerative colitis. J Clin Microbiol

53, 3702–3704.Australian Bureau of Statistics (2011) Land management and

farming in Australia, 2009-10.

Bakker, D., Corver, J., Harmanus, C., Goorhuis, A., Keessen,

E.C., Fawley, W.N., Wilcox, M.H. and Kujiper, E.J. (2010)

Relatedness of human and animal Clostridium difficile PCR

ribotype 078 isolates determined on the basis of

multilocus variable-number tandem-repeat analysis and

tetracycline resistance. J Clin Microbiol 48, 3744–3749.Bakri, M.M., Brown, D.J., Butcher, J.P. and Sutherland, A.D.

(2009) Clostridium difficile in ready-to-eat salads, Scotland.

Emerg Infect Dis 15, 817–818.

Boseiwaqa, L.V., Foster, N.F., Thean, S.K., Squire, M.M., Riley,

T.V. and Carson, K.C. (2013) Comparison of ChromID C.

difficile agar and cycloserine-cefoxitin-fructose agar for the

recovery of Clostridium difficile. Pathology 45, 495–500.Cheng, A.C., Collins, D.A., Elliott, B., Ferguson, J.K., Paterson,

D.L., Thean, S. and Riley, T.V. (2016) Laboratory-based

surveillance of Clostridium difficile circulating in Australia,

September-November 2010. Pathology 48, 257–260.Eckert, C., Burghoffer, B. and Barbut, F. (2013)

Contamination of ready-to-eat raw vegetables with

Clostridium difficile in France. J Med Microbiol 62, 1435–1438.

Eyre, D.W., Tracey, L., Elliott, B., Slimmings, C., Huntington,

P.G., Stuart, R.L., Korman, T.M., Kotsiou, G. et al. (2015)

Emergence and spread of predominantly community-onset

Clostridium difficile PCR ribotype 244 infection in

Australia, 2010 to 2012. Euro Surveill 20, 21059.

Freeman, J., Bauer, M.P., Baines, S.D., Corver, J., Fawley,

W.N., Goorhuis, B., Kujiper, E.J. and Wilcox, M.H. (2010)

The changing epidemiology of Clostridium difficile

infections. Clin Microbiol Rev 23, 529–549.Furuya-Kanamori, L., Riley, T.V., Paterson, D.L., Foster, N.F.,

Huber, C.A., Hong, S., Harris-Brown, T., Robson, J. et al.

(2016) A comparison of Clostridium difficile ribotypes

circulating in Australian hospitals and communities. J Clin

Microbiol 55, 216–225.Hacek, D.M., Ogle, A.M., Fisher, A., Robicsek, A. and

Peterson, L.R. (2010) Significant impact of terminal room

cleaning with bleach on reducing nosocomial Clostridium

difficile. Am J Infect Control 38, 350–353.Hensgens, M.P.M., Keessen, E.C., Squire, M.M., Riley, T.V.,

Koene, M.G.J., de Boer, E., Lipman, L.J.A. and Kujiper,

E.J. (2012) Clostridium difficile infection in the

community: a zoonotic disease? Clin Microbiol Infect 18,

635–645.Huber, C.A., Hall, L., Foster, N.F., Gray, M., Allen, M.,

Richardson, L.J., Robson, J., Vohra, R. et al. (2014)

Surveillance snapshot of Clostridium difficile infection in

hospitals across Queensland detects binary toxin

producing ribotype UK 244. Commun Dis Intell Q Rep 38,

279–284.Kato, N., Ou, C.Y., Kato, H., Bartley, S.L., Brown, V.K.,

Dowell, V.R. and Ueno, K. (1991) Identification of

toxigenic Clostridium difficile by the polymerase chain-

reaction. J Clin Microbiol 29, 33–37.Knight, D.R. and Riley, T.V. (2013b) Prevalence of

gastrointestinal Clostridium difficile carriage in Australian

sheep and lambs. Appl Environ Microbiol 79, 5689–5692.Knight, D.R., Thean, S., Putsathit, P., Fenwick, S. and Riley,

T.V. (2013a) Cross-sectional study reveals high prevalence

of Clostridium difficile non-PCR ribotype 078 strains in

Australian veal calves at slaughter. Appl Environ Microbiol

79, 2630–2635.Knight, D.R., Squire, M.M. and Riley, T.V. (2015) Nationwide

surveillance study of Clostridium difficile in Australian


S.C. Lim et al. C. difficile on root vegetables

A13

neonatal pigs shows high prevalence and heterogeneity of

PCR ribotypes. Appl Environ Microbiol 81, 119–123.Limbago, B.M., Long, C.M., Thompson, A.D., Killgore, G.E.,

Hannett, G.E., Havil, N.L., Mickelson, S., Lathrop, S. et al.

(2009) Clostridium difficile strains from community-

associated infections. J Clin Microbiol 47, 3004–3007.Limbago, B., Thompson, A.D., Greene, S.A., MacCannell, D.,

MacGowen, C.E., Jolbitado, B., Hardin, H.D., Estes, S.R.

et al. (2012) Development of a consensus method for

culture of Clostridium difficile from meat and its use in a

survey of U.S. retail meats. Food Microbiol 32, 448–451.Lund, B.M. and Peck, M.W. (2015) A possible route for

foodborne transmission of Clostridium difficile? Foodborne

Pathog Dis 12, 177–182.McGovern, A.M., Foster, N.F., Pereira, L.A., Knight, D.R.,

Elliott, B., Chang, B. and Riley, T.V. (2016) Human

Clostridium difficile infection caused by a livestock-

associated PCR ribotype 237 strain in Western Australia.

JMM Case Rep 3, 1–4.Metcalf, D.S., Costa, M.C., Dew, W.M. and Weese, J.S. (2010)

Clostridium difficile in vegetables, Canada. Lett Appl

Microbiol 51, 600–602.Richards, M., Knox, J., Elliott, B., Mackin, K., Lyras, D.,

Waring, L.J. and Riley, T.V. (2011) Severe infection with

Clostridium difficile PCR ribotype 027 acquired in

Melbourne, Australia. Med J Aust 194, 369–371.Simango, C. (2006) Prevalence of Clostridium difficile in the

environment in a rural community in Zimbabwe. Trans R

Soc Trop Med Hyg 100, 1146–1150.Slimings, C., Armstrong, P., Beckingham, W.D., Bull, A.L.,

Hall, L., Kennedy, K.J., Marquess, J., McCann, R. et al.

(2014) Increasing incidence of Clostridium difficile

infection, Australia, 2011-2012. Med J Aust 200, 272–276.Songer, J.G., Trinh, H.T., Killgore, G.E., McDonald, L.C. and

Limbago, B.M. (2009) Clostridium difficile in retail meat

products, USA, 2007. Emerg Infect Dis 15, 819–821.Warriner, K., Xu, C., Habash, M., Sultan, S. and Weese, S.J.

(2017) Dissemination of Clostridium difficile in food and the

environment: significant sources of C-difficile community-

acquired infection? J Appl Microbiol 122, 542–553.Weese, J.S. (2010a) Clostridium difficile in food—innocent

bystander or serious threat? Clin Microbiol Infect 16, 3–10.Weese, J.S., Avery, B.P., Rousseau, J. and Reid-Smith, R.J. (2009)

Detection and enumeration of Clostridium difficile spores in

retail beef and pork. Appl Environ Microbiol 75, 5009–5011.Weese, J.S., Reid-Smith, R.J., Avery, B.P. and Rousseau, J.

(2010b) Detection and characterization of Clostridium

difficile in retail chicken. Lett Appl Microbiol 50, 362–365.



A14

1Scientific RepoRts | 7:41196 | DOI: 10.1038/srep41196

www.nature.com/scientificreports

High prevalence of toxigenic Clostridium difficile in public space lawns in Western AustraliaPeter Moono1,*, Su Chen Lim1,* & Thomas V. Riley2,3,4

Clostridium difficile is a well-established hospital pathogen. Recently, it has been detected increasingly in patients without hospital contact. Given this rise in community associated infections with C. difficile, we hypothesized that the environment could play an important role in transmission of spores outside the hospital. Lawn samples (311) collected in public spaces in the metropolitan area of Perth, Western Australia, from February to June 2016 were cultured for C. difficile. C. difficile was isolated from the samples by direct and enrichment culture, and characterized by standard molecular methods using toxin gene PCR and ribotyping. The overall prevalence of C. difficile was 59%, new lawn (≤4 months old) was twice as likely as old lawn (>4 months old) to test positive (OR = 2.3; 95%CI 1.16–4.57, p = 0.015) and 35 C. difficile ribotypes were identified with toxigenic ribotype 014/020 (39%) predominating. The highest viable count from lawn soil samples was 1200 CFU/g. These results show that lawns in Perth, Western Australia, harbor toxigenic C. difficile, an important finding. The source of lawn contamination is likely related to modern practice of producing “roll-out” lawn. Further work should focus on identifying specific management practices that lead to C. difficile contamination of lawn to inform prevention and control measures.

Clostridium difficile is a well-established hospital pathogen associated with outbreaks of severe gastroenteritis1. C. difficile infection (CDI) in patients is acquired by oro-fecal ingestion of spores from the environment. The major virulence factors of C. difficile are two exotoxins, toxins A and B2, with some strains producing a third binary toxin3. There has been an increase in the incidence of community-associated CDI (CA-CDI) both in Australia4 and elsewhere5. Because of this increase in CA-CDI4–6, food sources have emerged as a potential res-ervoir of C. difficile. This paradigm is supported by studies that have reported detection of C. difficile in ani-mals7, meat8,9, and root10, and leaf11 vegetables. However, the overall prevalence of C. difficile in foods is low12. Although foodborne transmission of CDI is plausible it has never been proven, and the environment may be another important source of CA-CDI13,14. This is evident from many studies where various ribotypes (RTs) of C. difficile have been detected from the environment including so-called hypervirulent strains (RT 001, 014, 027, 045, 066, 078, and 126)15–17. It is possible that environmental contamination with C. difficile spores may be a more important source for CA-CDI than food12.

Outbreaks of infectious organisms, other than C. difficile, associated with manure have been reported18–21. In the USA, 50% of all biosolids produced each year is used in agriculture and landscaping22–24. C. difficile has been isolated from raw and treated human biosolids16, and the environment15. The mesophilic treatment of biosolids does not reduce C. difficile spore load25,26 and, therefore, application of biosolids to landscaping including private and public space lawns could transmit C. difficile in the community. In Perth, Western Australia, the construction of new housing and the development of new suburbs has seen an expansion of public and private lawns as has occurred in the USA27.

The aim of this study was to determine the prevalence and concentration of C. difficile in newly established (NL) and older (OL) lawns in public spaces in Perth, and to characterize any C. difficile isolates by phenotypic and genotypic techniques. Second, we investigated factors such as the location and size of the lawn to see if they could predict C. difficile status.

1Microbiology & Immunology, School of Pathology and Laboratory Medicine, The University of Western Australia, Nedlands, WA, 6009 Australia. 2School of Medical & Health Sciences, Edith Cowan University, Joondalup, WA, 6027 Australia. 3School of Veterinary & Life Sciences, Murdoch University, Murdoch, WA, 6150 Australia. 4PathWest Laboratory Medicine (WA), Nedlands, WA, 6009 Australia. *These authors contributed equally to this work. Correspondence and requests for materials should be addressed to T.V.R. (email: [email protected])

Received: 17 November 2016

Accepted: 15 December 2016

Published: 01 February 2017

OPEN

A15

www.nature.com/scientificreports/


ResultsPrevalence of C. difficile. The overall prevalence of C. difficile in lawn samples was 59% (182/311) varying from 0% to 70% by postcode. The prevalence of C. difficile was significantly higher in NLs 65% (129/198) com-pared to 47% (53/113) in OLs (Table 1). In the unadjusted univariable model, C. difficile was more likely to be detected in NL than OL (OR = 2.11, 95% CI: 1.32–3.4, p = 0.001) compared to the adjusted model (OR = 2.30, 95% (CI: 1.16–4.57, p = 0.015) (Table 1). In the multivariable analysis, age of lawn was also the only variable that was significantly associated with detection of C. difficile (OR = 1.92, 95% (CI: 1.15–3.21, p = 0.013). The size and location of the lawns were not associated with detection of C. difficile and the study period was too short to observe any effect of season on recovery.

Quantitation of C. difficile in lawn. Viable counts were performed on 10 lawn samples taken from one positive area by direct culture on C.diff ChromID agar (Table 2). Samples were also concentrated by centrif-ugation. Using PBST or PST diluent, the highest number of C. difficile positive cultures was 8 out of 10 after centrifugation. C. difficile recovery was higher in PST than PBST by both direct culture (50% vs 40%) and after centrifugation (80% vs 60%). The highest viable count of C. difficile detected was 1200 CFU/g of soil and the lowest 50 CFU/g of soil (Table 2). There were three ribotypes detected, all non-toxigenic, QX 189, QX 601, and UK 010.

Molecular characterization. There were 35 unique RTs detected in this study. Of the toxigenic (A+ B+ CDT−) RTs detected, 11 isolates harboured tcdA and tcdB genes but no binary toxin genes (Table 3). Toxigenic RTs accounted for 47% (86/182) of the total and included internationally recognized RTs associated with hos-pital infections. The 24 non-toxigenic (A− B− CDT−) RTs accounted for 53% (96/182). The toxigenic strain RT 014/020 was the most predominant (39.0%), followed by the non-toxigenic RT 010 (20.3%) (Table 3). Toxigenic RTs isolated from NL accounted for 33.5% (61/182) of all isolates compared to 14.3% (26/182) in OL. In the NL among toxigenic strains only, RT 014/020 was over represented 78.7% (48/61), followed by RTs 054, 056, 002, 018, QX 610, and QX 611 (all <5%). In the OL among toxigenic strains only, RT 014/020 was also over represented at 88.5% (23/26), followed by RTs 002, 106 and QX 409 (all < 5%).

DiscussionThe aim of this study was to investigate the prevalence of C. difficile in public space lawns in Perth, Western Australia. The reasons for undertaking the project were three-fold. First, there has been an increase in the inci-dence of CA-CDI in Australia4. Second, although many production animals carry C. difficile in their gut, particu-larly young animals, it is unlikely that meat contamination plays a major role in CA-CDI12. Last, anecdotally we learned that manure from pig farms was being used by turf farms for the production of lawn.

Variable Variable categories C. difficile number isolated (%)

Univariable model Covariate Odds ratios (95% CI)*

Odds ratios (95% CI)† Sampling site P value¶

Age‡ Old lawn (n = 113) 53 (47) Referent

New lawn (n = 198) 129 (65) 2.11 (1.32–3.4) 2.30 (1.16–4.57) 0.015#

Area Extra-large (n = 85) 53 (62) Referent

Large (n = 53) 26 (49) 0.58 (0.28–1.16) 0.49 (0.16–1.49) 0.7

Medium (n = 101) 60 (59) 0.88 (0.49–1.59) 1.02 (0.42–2.51) 0.7

Small (n = 72) 43 (60) 0.89 (0.47–1.71) 0.88 (0.32–2.43) 0.7

Location North (n = 161) 98 (60.9) Referent

South (n = 150) 84 (56) 1.22 (0.78–1.92) 1.25 (0.61–2.59) 0.99

SeasonAutumn (n = 224) 135 (60.3) Referent

Winter (n = 87) 47 (54) 0.77 (0.47–1.28) 0.67 (0.28–1.62) 0.52

Table 1. The relationship between the prevalence of C. difficile in lawn and the age of the lawn, its size, sampling site, location, postcode, and season in Perth. *CI; The 95% confidence interval of odds ratio for the covariate estimates. †CI; The 95% confidence interval of odds ratio for the univariate estimates. ‡Age of lawn; new lawn ≤ 4 months, old lawn > 4 months. ¶P values are based on likelihood ratios and P < 0.05 was considered significant#. Univariable logistic regression model with random effect (site of sampling and postcode). The random effect term for postcode was not included in all the models because its addition or removal did not change the model estimates significantly.

Diluent

Sample (viable count cfu/g of soil) No. samples positive1 2 3 4 5 6 7 8 9 10

PBST 0 0 0 50 50 250 0 0 0 100 4

PST 0 0 0 800 200 100 0 50 0 1200 5

Table 2. Viable counts of Clostridium difficile in soil from lawn samples using either phosphate buffer solution (pH 7.4) (PBST) or peptone saline (PST), both containing 0.1% Tween 20, as a diluent.

A16



The overall prevalence of C. difficile in lawn samples was 59%, varying from 0% to 70% by postcode. In the adjusted univariable model, C. difficile was more likely to be detected in NL than OL (OR = 2.30, 95% CI: 1.16–4.57, p = 0.015) (Table 1). Other studies have reported a high prevalence (30–68%) of C. difficile in the household environment and biosolid treatment plants15,16, similar to this study. The reason for the lower preva-lence of C. difficile in OL (> 4 months) is not clear, although an effect of natural exposure of germinated spores to oxygen or ultraviolet light cannot be excluded. It is unlikely that C. difficile can multiply to any great extent in the lawn and the counts detected were not very high. A recent study has shown that ultraviolet light devices can be used as sanitizers in hospitals to reduce pathogenic organisms including C. difficile28. Another study found that C. difficile spores could survive up to ~5 months in the environment29. Therefore, the difference in prevalence between NL and OL could be explained by environmental factors such rain, wind, and ultraviolet light over time, and contact with people and animals which could help disperse the initial spore load, particularly on the surface of the lawn. The majority of the NL in this study was located within recently built suburbs in Perth. The relation-ship between new suburb buildings and lawn expansion is well studied27. Our findings suggest that new suburbs and consequent expansion of NL could be contributing to dispersal of C. difficile in the environment and variables other than NL were not significantly associated with isolation of C. difficile.

Nearly 25% of global disease burden has been attributed to environmental risk factors30. Animal manure is widely used in both agriculture and landscaping22–24,31, and anecdotal evidence suggests this practice is wide-spread in Australia. In the USA, manure that has been composted is exempted from federal or state rule regarding pathogen levels if it is going for land application31. The major problem which is often overlooked in applying ani-mal manure or biosolids to land is consideration of the duration of persistence of pathogens on plants31. Animal effluent treatment involves removal of solid wastes and then holding liquid effluent in anaerobic ponds under mesophilic anaerobiosis. The treatment method for human raw effluent in Switzerland involved grid separation, primary sedimentation, and secondary biological treatment (an activated sludge process)16. The major problem in animal manure treatment is lack of separation of feces from young animals versus older animals. Young animals are associated with shedding high levels of pathogenic organisms including, and in particular, C. difficile8,31,32. Currently, there is pressure on the turf industry to meet the demand for lawns in newly built suburbs both pri-vate and public lawns. This has led the industry to develop the lawn farming system of roll-out lawns which can easily be transported to clients on demand. The lawn, like other plants, is grown on manure or biosolids for rapid growth and early maturation. The fact that even composted manure or treated animal effluent does not inactivate C. difficile spores implies that lawns could be heavily contaminated with C. difficile and our data suggest this is the case. Even compost, manure or biosolids that have been produced with specific standards23 have caused dis-ease outbreaks in the USA18–21. In this regard, manure, compost and biosolids should be screened for pathogenic microorganisms such as C. difficile before being applied to land for recreational or agricultural use.

Although the infectious dose for C. difficile is unknown, there are suggestions that it could be low (100–1000 spores)33. In addition, in some high-risk individuals infection could be driven mainly by host factors. In this study, we found that the highest C. difficile spore concentration in the lawn samples was 1200 CFU/g of soil (Table 2). The concentration of C. difficile in feces of cattle has been reported as 2.5 × 104 CFU/mL9 and in human feces as 6.66 + 0.62 log10 CFU/mL34. It is unclear how much exposure to lawns is required for a human or animal to be contaminated with sufficient spores that could lead to infection with C. difficile. The fact that ingestion of C. difficile from the environment does not quickly result in infection is due to the complex nature of the patho-genesis of disease that requires the gut microflora to be perturbed, usually in the form of antibiotic exposure5. Although we did not find any seasonal fluctuations of C. difficile in lawn samples in this study, it is likely that recreational exposure to C. difficile could be higher in some seasons when people regularly use public spaces and lower in low activity periods. Therefore, studies that quantitate spore load on fomites such as shoes are needed because they may help explain the rise of CA-CDI.

There were 35 unique RTs detected in this study, including the clinically important RTs 014/020, 056, 054, 002, 018, 012 and 043 (Table 3). The high prevalence of RT 014/020 in this study was an interesting and important

Ribotype

Toxin gene profile

Number (%)tcdA tcdB cdtA/B

014/020 + + − 71 (39.0)

010 − − − 37 (20.3)

QX 077 − − − 13 (7.1)

QX 189 − − − 7 (3.9)

039 − − − 5 (2.7)

QX 601 − − − 4 (2.2)

393 − − − 4 (2.2)

QX 142 − − − 4 (2.2)

002 + + − 4 (2.2)

Others* Various Various − 33 (18.1)

Table 3. The frequency and toxin gene profile of C. difficile ribotypes in lawns in Perth, Western Australia. *Others: QX 518, QX 611, QX 608, QX 610, RT 056, RT 054, QX 607, QX 606, QX 605, QX 603, QX 602, QX 550, QX 449, QX 409, QX 393, QX 210, RT 125, QX 121, RT 106, QX 072, QX 067, QX 054, RT 018, RT 012, RT 009 and RT 043.

A17



finding given that RT 014/020 is the most predominant RT in laboratory samples from diarrheic patients in Australia6,35,36. RT 014/020 is also the main ribotype currently found in pigs in Australia37 suggesting that pig manure may be responsible for the contamination of lawn in Perth. Other researchers have reported detection of RT 014/020 in river water17 and treated sludge16.

The detection of RTs 012, 002, and 018 was unexpected because they are not routinely detected from the environment16,17, although they are increasingly associated with CA-CDI35,36. Recently, RTs 056, 010, 078, 213, 009 and 020 were detected in dogs38. RT 056 was also recently detected in gastrointestinal tracts of veal calves at slaughter32. Several of the other RTs detected in our study (056, 054, 018 and 002) have been associated with human CDI in Europe39, Japan40 and Australia6. These findings suggest that all lawns may be getting contami-nated with animal strains of C. difficile and that this could be contributing to an increase in CA-CDI.

Interestingly, no binary toxin producing C. difficile isolates were cultured in this study although such strains have been recovered previously in high numbers from production animals in Australia32,37,41. The reasons for this are unclear. Our earlier animal studies have suggested that different strains of C. difficile tend to infect/col-onise herds of animals in particular geographic areas32,37 and these are not always binary toxin producers, such as RT014/020 in pigs37. The lack of binary toxin producers in the current study may just reflect a relatively small sample size and the fact this study was undertaken in Western Australia only.

The lawns sampled were located within the metropolitan area of Perth, suggesting that people and dogs walking on lawns could transfer C. difficile into the household42. A high prevalence (32.3%) of C. difficile in household environs has been reported, particularly on the soles of shoes15. It is also possible that shoes could be vehicles by which C. difficile could be introduced in health care facilities. The detection of C. difficile in lawn could help explain the substantial proportion of cases (45%) of CDI originating from unknown reservoirs as recently described by others43,44.

This study has some limitations. First, we only sampled a relatively small number of public spaces (n = 20) and the study was undertaken only in Western Australia which may impact its generalisability to other regions of Australia and the world. However, all lawn/turf producing companies have access to the same sources of manure in Western Australia, suggesting that the problem could be similar throughout the metropolitan area of Perth. The age of some of the lawns was estimated with a series of photos of lawns with a known age and this could have resulted in misclassification bias. However, this would not have affected the overall outcomes of the study as both OL and NL had relatively high prevalence of C. difficile. We investigated management practices of commercial lawn farms, however, the frequency of maintenance of the lawns was not established.

In conclusion, the high prevalence of C. difficile in lawn is potentially an important finding, however, it is difficult to estimate the risk of CDI through exposure to C. difficile in lawns. There is a need for further detailed and structured studies to examine how modern lawns are grown, and to investigate the role of any manure and biosolids applied as fertiliser as a source of CA-CDI. The relevance of these findings should be further con-firmed by studies comparing the relatedness of isolates in this study to those CA-CDI cases using whole genome sequencing.

MethodsBackground. Lawn in Western Australia can either be raised as seeding lawn or “instant turf ” in the form of rolls of pre-grown lawn45. Seeded lawn has lower cost to plant compared to instant turf, however, it takes longer to mature and is more labour intensive than turf rolls. The advantage of turf rolls is that they are fast to grow and easy to lay, and can be used within a week, hence the current high demand. Soft leaved buffalo grass is a widely used variety in Western Australia because it can withstand drought; most lawns are established in early autumn or late summer to avoid extreme temperatures in summer45.

Lawns in Australia are considered young up to 4 months by the turf industry45. Therefore, we used this time-point to distinguish between newly established (NL) and older (OL) lawns [NL (≤ 4 months) and OL (> 4 months)]. From autumn through winter, older lawns tend to lose colour and become dormant. We initiated this study during this period when it should have been relatively easy to identify lawns that were new.

Study location and samples. NL and OL samples were collected from 20 public spaces within 11 post-codes in Perth and surrounding areas in February–June 2016. When information on the age of lawns in some areas was unavailable, photographic identification was employed. A NL was identified by the presence of a visible line between two or more strips of laid down lawn which remained visible for up to 4 months. Lawns that did not have these were assumed to be old45. This involved comparing photographs with a series of photographs taken of lawns of known age (NL appeared bright green, luxuriant and with sharp edges). This technique has been widely used in the USA while some researchers have even employed remote sensing imaging46. We sampled within 5–30 km north and 5–80 km south of Perth city. We categorized public spaces by size (small [0.5–1 km2], medium [1.1–2 km2], large [2.1–2.9 km2], and extra-large [≥ 3 km2]) based on area data from local council authority web-sites in each study area or, where no information was available, we estimated the area of the space in relation to other known areas. All the public spaces that met the size cut off and had a grass variety that was commercially available were included in the sampling frame, however, sampling was performed on the basis of convenience. We estimated that 150 samples were appropriate to detect 18% difference in C. difficile prevalence between NL and OL (α = 0.05, power 80%, 95 CI). To account for sampling clustering, we multiplied 150 samples obtained by simple random sampling by 2 design effect. A total of 311 NL or OL samples were collected in sterile 200 mL specimen jars. Each public space was divided into four quadrants with the centre generally being a children’s play facility. Four samples were obtained per quadrant starting from the centre and moving outwards. The lawn sample consisted of grass and its root system with attached soil, and was obtained using a new set of sterile examination

A18



gloves and a sterile tongue depressor which was used to dig out grass and its root system. Each lawn sample weighed approx. 50 g and these were transported at ambient temperature to the laboratory.

C. difficile isolation. Ninety mL brain heart infusion broths supplemented with cycloserine and cefoxitin, were pre-reduced for 4 h in an anaerobic chamber (A35, Don Whitley Scientific Ltd., Shipley, West Yorkshire, UK) at 37 °C. The lawn samples were aseptically cut into ~5 cm2 pieces weighing approx. 5 g, inoculated into the broth and gently shaken, and then incubated at 35 °C in an A35 Anaerobic Workstation (Don Whitley Scientific Ltd., UK) for 5 days with the lids loose. A 5 mL aliquot of culture broth was mixed with 5 mL anhydrous ethanol (96%) and incubated for 1 h before being centrifuged at 3000 g for 10 min. The pellet was plated onto C. diff ChromID agar (bioMérieux, Marcy l’Etoile, France) and the plates incubated anaerobically for 48 h. C. difficile was identified on the basis of its characteristic chartreuse fluorescence detected with UV light (~360 nm wavelength), morpho-logical characteristic (ground glass appearance) and odor (horse dung)32. Identification of uncertain isolates was achieved by Gram staining and detection of L-proline aminopeptidase (Remel Inc., Lenexa, KS, USA).

Quantitation of C. difficile in lawn samples. A viable count of C. difficile was performed on a subset of lawn samples using previously described methods with some modifications9,34. Briefly, approx. 5 g of soil was aseptically obtained from an approx. 10 g lawn sample by shaking the root system and placed in a Stomacher bag (Colworth, London, UK) containing 25 mL of phosphate buffered saline at pH 7.4 supplemented with 0.1% Tween 20 (PBST) or peptone saline also containing 0.1% Tween 20 (PST) and shaken for 60 s. A 100 μ L aliquot was removed and directly inoculated onto C. diff ChromID agar, spread using a sterile hockey stick (InterPath Services Pty Ltd, West Victoria, Australia) and incubated anaerobically. The remaining volume was centrifuged at 3000 g for 10 min, the supernatant discarded, the pellet resuspended in 1000 μ L of either PBST or PST and 100 μ L directly inoculated on C. diff ChromID. Viable colony counts were performed at 48 h. A negative control (either PBST or PST) was used in each assay to monitor for potential contamination. We performed molecular typing on all positive samples using toxin PCR and PCR-ribotyping to confirm the isolates. The detection limit for the viable count was > 1 CFU per gram of soil.

Molecular characterization. All isolates were characterized by PCR to determine the presence of toxin A (tcdA), B (tcdB), and binary toxin (cdtA and cdtB) genes47,48. Novel primers were multiplexed with tcdA primers NK2 and NK346 to detect the tcdA repeat region A3 fragment. Briefly, 4 μ L of DNA extract was added to a PCR mixture containing 2 mM MgCl2, 50 mM KCl, 10 mM Tris–HCl (pH 8.3), 200 mM of dNTP, 0.2 mM of each primer NK2, NK3 and BE-tcdA-1 (FP: 5′CAGTCACTGGATGGAGAATT-3′ and BE-tcdA-2 (RP: 5′ -AAGGCAATAGCGGTATCAG-3′ [B. Elliott, unpublished data]. Detection of both the tcdA1 and tcdA3 fragment was required for an isolate to be considered tcdA positive. Amplifications were carried out in a 2720 Thermal Cycler (Applied Biosystems, Foster City, USA).

PCR ribotyping was performed on isolates as described elsewhere49. RTs were identified by comparing their banding patterns with those in our reference library from the Anaerobe Reference Laboratory (ARL, Cardiff, UK) ribotypes that included 15 reference strains from the European Centre for Disease Prevention and Control (ECDC) and the most prevalent PCR ribotypes currently circulating in Australia (B. Elliott, T. V. Riley, unpub-lished data). Isolates that could not be identified with the international reference library were designated with local (QX) nomenclature.

Statistical analysis. Univariable random effect logistic regression analysis was used to describe the relation-ship between C. difficile culture status (response variable), and independent variables: age of lawn (new vs. old), sampling site, location (north vs south), postcode, size of playground (small, medium, large, and extra-large) and season (autumn: March, April, May and winter: June, July August). Two random effects, postcode and sampling site were included in the model to account for clustering or autocorrelation among samples from the same sam-pling site and postcode. Univariable models were constructed including one main effect and two random effects per model, then a backward stepwise multivariable model including the interactions terms, main effects and random effects. In the multivariable model, variables that were not statistically significant at α = 0.05 were excluded from the model with the assumptions that they were not confounders. Variables were retained in the model if they were significant or part of their interaction terms was significant or they were considered as con-founders. Random effects were excluded from the model if they explained very little from the models based on the change in Akaike’s Information Criterion (AIC), Bayesian Information Criterion and deviance χ2 50. Model selection was also evaluated by the likelihood ratios using the generalized linear model (GLM) framework using R version 3.3.1.

References1. Rupnik, M., Wilcox, M. H. & Gerding, D. N. Clostridium difficile infection: new developments in epidemiology and pathogenesis.

Nat. Rev. Microbiol. 7, 526–536 (2009).2. Kuehne, S. A. et al. The role of toxin A and toxin B in Clostridium difficile infection. Nature 467, 711–713 (2010).3. Cowardin, C. A. et al. The binary toxin CDT enhances Clostridium difficile virulence by suppressing protective colonic eosinophilia.

Nat Microbiol 1, 16108–16118 (2016).4. Slimings, C. et al. Increasing incidence of Clostridium difficile infection, Australia, 2011–2012. Med. J. Aust. 200, 272–276 (2014).5. Kuntz, J. L., Chrischilles, E. A., Pendergast, J. F., Herwaldt, L. A. & Polgreen, P. M. Incidence of and risk factors for community-

associated Clostridium difficile infection: A nested case-control study. BMC Infect. Dis. 11, 194–2001 (2011).6. Eyre, D. W. et al. Emergence and spread of predominantly community-onset Clostridium difficile PCR ribotype 244 infection in

Australia, 2010 to 2012. Eurosurveil. 20, pii21059, 1–10 (2015).7. Keessen, E. C., Harmanus, C., Dohmen, W., Kuijper, E. J. & Lipman, L. J. Clostridium difficile infection associated with pig farms.

Emerg. Infect. Dis. 19, 1032–1034 (2013).

A19



8. Houser, B. A. et al. Prevalence of Clostridium difficile toxin genes in the feces of veal calves and incidence of ground veal contamination. Foodborne Pathog. Dis. 9, 32–36 (2012).

9. Knight, D. R., Putsathit, P., Elliott, B. & Riley, T. V. Contamination of Australian newborn calf carcasses at slaughter with Clostridium difficile. Clin. Microbiol. Infect. 22, 266.e261–266.e267 (2016).

10. Eckert, C., Burghoffer, B. & Barbut, F. Contamination of ready-to-eat raw vegetables with Clostridium difficile in France. J. Med. Microbiol. 62, 1435–1438 (2013).

11. Bakri, M. M., Brown, D. J., Butcher, J. P. & Sutherland, A. D. Clostridium difficile in ready-to-eat salads, Scotland. Emerg. Infect. Dis. 15, 817–818 (2009).

12. Limbago, B. et al. Development of a consensus method for culture of Clostridium difficile from meat and its use in a survey of US retail meats. Food Microbiol. 32, 448–451 (2012).

13. Bauer, M. P. & Kuijper, E. J. Potential sources of Clostridium difficile in human infection. Infect. Dis. Clin. North Am. 29, 29–35 (2015).

14. Squire, M. M. & Riley, T. V. Clostridium difficile infection in humans and piglets: A ‘One Health’ Opportunity. Curr. Top. Microbiol. Immunol. 365, 299–314 (2013).

15. Alam, M. J., Anu, A., Walk, S. T. & Garey, K. W. Investigation of potentially pathogenic Clostridium difficile contamination in household environs. Anaerobe 27, 31–33 (2014).

16. Romano, V. et al. Toxigenic Clostridium difficile PCR ribotypes from wastewater treatment plants in southern Switzerland. Appl. Environ. Microbiol. 78, 6643–6646 (2012).

17. Zidaric, V., Beigot, S., Lapajne, S. & Rupnik, M. The occurrence and high diversity of Clostridium difficile genotypes in rivers. Anaerobe 16, 371–375 (2010).

18. Doyle, M. & Erickson, M. Summer meeting 2007–the problems with fresh produce: an overview. J. Appl. Microbiol. 105, 317–330 (2008).

19. Gould, L. H. et al. Surveillance for foodborne disease outbreaks–United States, 1998–2008. MMWR. Surveill. Summ. 62, 1–34 (2013).

20. Painter, J. A. et al. Attribution of foodborne illnesses, hospitalizations, and deaths to food commodities by using outbreak data, United States, 1998–2008. Emerg. Infect. Dis. 19, 407–415 (2013).

21. Gerba, C. P. & Smith, J. E. Jr. Sources of pathogenic microorganisms and their fate during land application of wastes. J. Environ. Qual. 34, 42–48 (2005).

22. USEPA. Sewage sludge (biosolids): frequently asked questions. http://water.epa.gov/polwaste/wastewater/treatment/biosolids/genqa.cfm (Date of access:01/05/2016) (2012).

23. Iranpour, R. et al. Regulations for biosolids land application in US and European Union. J. Residuals Sci. Tech. 1, 209–222 (2004).24. North East Biosolids Residuals Association (NEBRA). A national biosolids regulation, quality, end use & disposal survey. http://www.

nebiosolids.org/uploads/pdf/NtlBiosolidsReport-20July07.pdf. (Date of access:01/05/2016) (2007).25. Nikaeen, M., Dehnavi, H. A., Hssanzadeh, A. & Jalali, M. Occurrence of Clostridium difficile in two types of wastewater treatment

plants. J. Formos. Med. Assoc. 114, 663–665 (2015).26. Xu, C., Salsali, H., Weese, J. S. & Warriner, K. Inactivation of Clostridium difficile in sewage sludge by anaerobic thermophilic

digestion. Can. J. Microbiol. 62, 16–23 (2015).27. Robbins, P. & Birkenholtz, T. Turfgrass revolution: measuring the expansion of the American lawn. Land Use Policy 20, 181–194

(2003).28. Weber, D. J. et al. Effectiveness of ultraviolet devices and hydrogen peroxide systems for terminal room decontamination: Focus on

clinical trials. Am. J. Infect. Control 44, e77–e84 (2016).29. Kramer, A., Schwebke, I. & Kampf, G. How long do nosocomial pathogens persist on inanimate surfaces? A systematic review. BMC

Infect. Dis. 6, 130–137 (2006).30. Prüss-Üstün, A. & Corvalán, C. How much disease burden can be prevented by environmental interventions? Epidemiology 18,

167–178 (2007).31. Suslow, T. et al. Production practices as risk factors in microbial food safety of fresh and fresh‐cut produce. Compr. Rev. Food Sci.

Food Saf. 2, 38–77 (2003).32. Knight, D. R., Thean, S., Putsathit, P., Fenwick, S. & Riley, T. V. Cross-sectional study reveals high prevalence of Clostridium difficile

non-PCR ribotype 078 strains in Australian veal calves at slaughter. Appl. Environ. Microbiol. 79, 2630–2635 (2013).33. Warriner, K. et al. Dissemination of Clostridium difficile in food and the environment: significant sources of C. difficile community

acquired infection? J. Appl. Microbiol. doi: 10.1111/jam.13338 (2016).34. Housman, S. T., Banevicius, M. A., Lamb, L. M. & Nicolau, D. P. Isolation and quantitation of Clostridium difficile in aqueous and

fecal matter using two types of selective media. J. Microbiol. Immunol. Infect. 49, 445–447 (2015).35. Cheng, A. C. et al. Laboratory-based surveillance of Clostridium difficile circulating in Australia, September–November 2010.

Pathology 48, 257–260 (2016).36. Collins, D. A. & Riley, T. V. Routine detection of Clostridium difficile in Western Australia. Anaerobe 37, 34–37 (2016).37. Knight, D. R., Squire, M. M. & Riley, T. V. Nationwide surveillance study of Clostridium difficile in Australian neonatal pigs shows

high prevalence and heterogeneity of PCR ribotypes. Appl. Environ. Microbiol. 81, 119–23 (2015).38. Alvarez-Perez, S. et al. Faecal shedding of antimicrobial-resistant Clostridium difficile strains by dogs. J. Small Anim. Pract. 56,

190–195 (2015).39. Fawley, W. et al. Surveillance and outbreak report: Enhanced surveillance of Clostridium difficile infection occurring outside

hospital, England, 2011 to 2013. Eurosurveil. 21, pii = 30295. http://dx.doi.org/10.2807/1560-7917.ES.2016.21.29.30295 (2016).40. Senoh, M. et al. Predominance of PCR-ribotypes, 018 (smz) and 369 (trf) of Clostridium difficile in Japan: a potential relationship

with other global circulating strains? J. Med. Microbiol. 64, 1226–1236 (2015).41. Knight, D. R., Squire, M. M. & Riley, T. V. Laboratory detection of Clostridium difficile in piglets in Australia. J. Clin. Microbiol. 52,

3856–3862 (2014).42. Weese, J. S., Finley, R., Reid-Smith, R., Janecko, N. & Rousseau, J. Evaluation of Clostridium difficile in dogs and the household

environment. Epidemiol. Infect. 138, 1100–1104 (2010).43. Eyre, D. W. et al. Diverse sources of C. difficile infection identified on whole-genome sequencing. New Engl. J. Med. 369, 1195–1205 (2013).44. Kubota, H. et al. Longitudinal investigation of carriage rates, counts, and genotypes of toxigenic Clostridium difficile in early infancy.

Appl. Environ. Microbiol. 82, 5806–5814 (2016).45. Scotts Gardening Team. The Aussie homeowner’s guide to a lush green lawn. http://www.scottsaustralia.com.au/diy-garden-guides-

advice/lawns/maintaining-repairing-new-lawns/guide-to-a-lush-green-lawn/ (Date of access: 29/11/2016) (2016).46. Giner, N. M., Polsky, C., Pontius, R. G. & Runfola, D. M. Understanding the social determinants of lawn landscapes: A fine-

resolution spatial statistical analysis in suburban Boston, Massachusetts, USA. Landscape Urban Plan. 111, 25–33 (2013).47. Kato, N. et al. Identification of toxigenic Clostridium difficile by the polymerase chain reaction. J. Clin. Microbiol. 29, 33–37 (1991).48. Stubbs, S. et al. Production of actin‐specific ADP‐ribosyltransferase (binary toxin) by strains of Clostridium difficile. FEMS

Microbiol. 186, 307–312 (2000).49. Stubbs, S. L., Brazier, J. S., O’Neill, G. L. & Duerden, B. I. PCR targeted to the 16S–23S rRNA gene intergenic spacer region of

Clostridium difficile and construction of a library consisting of 116 different PCR ribotypes. J. Clin. Microbiol. 37, 461–463 (1999).50. Dohoo, I. R., Martin, W. & Stryhn, H. Mixed models for discreet data. (2nd ed.) 579–606 (VER Inc, 2009).

A20



AcknowledgementsPeter Moono is a recipient of an International Postgraduate Scholarship awarded by The University of Western Australia. This work was supported in part by a grant from the Australian Research Council (DPI 50104670).

Author ContributionsT.V.R. conceived the study. P.M. collected samples and performed culture and DNA extraction. S.C.L. performed molecular characterization and contributed to drafting the manuscript. T.V.R. and P.M. performed data analysis and interpretation of results and drafted the manuscript. All authors approved the final manuscript.

Additional InformationCompeting financial interests: The authors declare no competing financial interests.How to cite this article: Moono, P. et al. High prevalence of toxigenic Clostridium difficile in public space lawns in Western Australia. Sci. Rep. 7, 41196; doi: 10.1038/srep41196 (2017).Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license,

unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ © The Author(s) 2017

A21

International Journal of Antimicrobial Agents 52 (2018) 411–415

Contents lists available at ScienceDirect

International Journal of Antimicrobial Agents

journal homepage: www.elsevier.com/locate/ijantimicag

Short Communication

Antimicrobial susceptibility of Clostridium difficile isolated from food

and environmental sources in Western Australia

Su-Chen Lim

a , Grace O. Androga

a , b , Daniel R. Knight a , c , Peter Moono

a , Niki F. Foster d , Thomas V. Riley

a , c , d , e , ∗

a School of Biomedical Sciences, The University of Western Australia, Nedlands, WA, Australia b Department of Microbiology, PathWest Laboratory Medicine, Nedlands, WA, Australia c School of Veterinary and Life Sciences, Murdoch University, Murdoch, WA, Australia d OzFoodNet, Communicable Diseases Control Directorate, Department of Health, Government of Western Australia, Perth, WA, Australia e School of Medical and Health Sciences, Edith Cowan University, Joondalup, WA, Australia

a r t i c l e i n f o

Article history:

Received 13 January 2018

Accepted 15 May 2018

Keywords:

Clostridium difficile

Antimicrobial

Resistance

Food

Environmental

Australia

a b s t r a c t

We recently reported a high prevalence of Clostridium difficile in retail vegetables, compost and lawn

in Western Australia. The objective of this study was to investigate the antimicrobial susceptibility of

previously isolated food and environmental C. difficile isolates from Western Australia. A total of 274 C.

difficile isolates from vegetables, compost and lawn were tested for susceptibility to a panel of 10 an-

timicrobial agents (fidaxomicin, vancomycin, metronidazole, rifaximin, clindamycin, erythromycin, amox-

icillin/clavulanic acid, moxifloxacin, meropenem and tetracycline) using the agar incorporation method.

Fidaxomicin was the most potent agent (MIC 50 /MIC 90 , 0.06/0.12 mg/L). Resistance to fidaxomicin and

metronidazole was not detected and resistance to vancomycin (0.7%) and moxifloxacin (0.7%) was low.

However, 103 isolates (37.6%) showed resistance to at least one agent, and multidrug resistance was ob-

served in 3.9% of the resistant isolates (4/103), all of which came from compost. A significantly greater

proportion of compost isolates were resistant to clindamycin, erythromycin and tetracycline compared

with food and/or lawn isolates. Clostridium difficile ribotype (RT) 014/020 showed greater clindamycin re-

sistance than other less common RTs ( P = 0.008, χ2 ). Contaminated vegetables, compost and lawn could

be playing an intermediary role in the transmission of C. difficile from animals to humans. Environmental

strains of C. difficile could also function as a reservoir for antimicrobial resistance genes of clinical rele-

vance. This study provides a baseline for future surveillance of antimicrobial resistance in environmental

C. difficile isolates in Australia.

© 2018 Elsevier B.V. and International Society of Chemotherapy. All rights reserved.

1. Introduction

Clostridium difficile infection (CDI) is the leading cause of life-

threatening infectious diarrhoea in humans and is a major public-

health issue in many developed countries [1] . Clostridium difficile

causes a wide range of symptoms, from mild diarrhoea to severe

pseudomembranous colitis and, in rare cases, fulminant colitis that

may lead to intestinal perforation or megacolon and sepsis [1] .

The major risk factor for developing CDI is exposure to antimi-

crobials, particularly agents with activity against commensal bowel

flora such as clindamycin, aminopenicillins, extended-spectrum

cephalosporins and fluoroquinolones [1] . Since 20 0 0, a substan-

∗ Corresponding author. Present address: Department of Microbiology, PathWest

Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, WA 6009, Aus-

tralia. Tel.: + 61 8 6457 3690; fax: + 61 8 9382 8046.

E-mail address: [email protected] (T.V. Riley).

tial increase in the incidence of CDI has been observed world-

wide, including community-associated CDI (CA-CDI). Currently, ca.

30% of all CDI cases in Australia are CA-CDI with no traditional

risk factors such as hospital stay, previous antimicrobial use or old

age/immune senescence [2] .

Some studies have reported genetically-related, and in some

cases indistinguishable, C. difficile strains from animals and humans

[3] , suggestive of zoonotic transmission. However, most related iso-

lates were separated by vast geographical distances and there was

no known prior contact between hosts, making direct transmission

unlikely. Thus, following the isolation of clinically important C. dif-

ficile strains from food and the environment, it has been hypoth-

esised that some CA-CDI might be foodborne or that transmission

from an environmental source occurs in some other way [4] .

Recently, we found a high prevalence of C. difficile on retail root

vegetables (30.0%; 30/100) [5] , in compost (27.2%; 22/81) destined

for use in farming and landscaping (S.C. Lim et al., unpublished

https://doi.org/10.1016/j.ijantimicag.2018.05.013

0924-8579/© 2018 Elsevier B.V. and International Society of Chemotherapy. All rights reserved.

A22

412 S.-C. Lim et al. / International Journal of Antimicrobial Agents 52 (2018) 411–415

Fig. 1. Origin of Clostridium difficile ribotypes (RTs) tested in this study ( n = 274). ∗ Other RTs include QX 001 (A + B + CDT–), QX 026 (A + B + CDT–), QX 067 (A–B–CDT–),

QX 076 (A + B + CDT–), QX 121 (A–B–CDT–), QX 122 (A–B–CDT–), QX 140 (A–B–CDT–), QX 274 (A + B + CDT + ), QX 327 (A–B–CDT–), QX 399 (A + B + CDT + ), QX 409 (A + B + CDT–),

QX 449 (A–B–CDT–), QX 463 (A–B–CDT–), QX 519 (A + B + CDT–), QX 525 (A–B–CDT–), QX 546 (A–B–CDT–), QX 547 (A–B + CDT + ), QX 550 (A–B–CDT–), QX 597 (A–B–CDT–),

QX 598 (A–B–CDT–), QX 599 (A–B–CDT–), QX 600 (A–B–CDT + ), QX 602 (A–B–CDT–), QX 603 (A–B–CDT–), QX 604 (A–B–CDT–), QX 605 (A–B–CDT–), QX 606 (A–B–CDT–),

QX 607 (A–B–CDT–), UK 005 (A + B + CDT–), UK 009 (A–B–CDT–), UK 017 (A–B + CDT–), UK 018 (A + B + CDT–), UK 033 (A–B–CDT + ), UK 046 (A + B + CDT–), UK 064 (A + B + CDT–),

UK 070 (A + B + CDT–), UK 077 (A–B–CDT–), UK 080 (A–B + CDT + ), UK 106 (A + B + CDT–), UK 137 (A + B + CDT–) and UK 584 (A–B + CDT + ).

data) and in public lawns (58.5%; 182/311) [6] in Perth, Western

Australia. It is possible that these contaminated items could be

playing a role in the transmission of CDI, especially CA-CDI. It is

common practice in Australia to use composted animal manure as

fertiliser for vegetable and lawn (turf) farming, which could result

in contamination of vegetables and lawn with C. difficile of ani-

mal origin and lead to genetically highly-related human CDI in the

community.

Whilst human clinical isolates are occasionally surveyed for an-

timicrobial susceptibility, less is known about antimicrobial resis-

tance in C. difficile from other sources. The aims of this study were:

(i) to determine the antimicrobial susceptibility of C. difficile iso-

lated from food, compost and lawn to a panel of 10 antimicrobial

agents; and (ii) to compare the antimicrobial resistance profile of

these strains with published animal- and human-derived isolates

from Australia.

2. Materials and methods

2.1. Bacterial isolates

In 2015 and 2016, studies were carried out in Western Australia

to determine the prevalence of C. difficile from food and environ-

mental sources. All C. difficile isolates from those studies ( n = 274)

were tested in the current investigation: 56 from vegetables [43

from root vegetables [5] and 13 from imported vegetables and/or

vegetables of unknown country of origin (unpublished data)], 36

from compost (unpublished data) and 182 from lawn [6] . The iso-

lates had been characterised by toxin gene profiling and PCR ribo-

typing [5,6] using a reference library consisting of a collection of

54 internationally recognised UK ribotypes (RTs), which included

15 reference strains from the European Centre for Disease Preven-

tion and Control (ECDC), as well as various RTs currently circu-

lating in Australia assigned with internal nomenclature, prefixed

with QX. A summary of the RTs, toxin gene profiles and sources

of the 274 C. difficile isolates is shown in Fig. 1 . Clostridium diffi-

cile RT 014 and 020 were grouped together due to nearly identical

RT banding patterns, differing by only one band. Clostridium difficile

RT 014/020, which is the most common RT isolated from Australian

pigs and humans [3] , was the most common food and environmen-

tal RT in the collection. The toxin profiles represented included A–

B–CDT– ( n = 145; 52.9%), A + B + CDT– ( n = 117; 42.7%), A–B + CDT +

( n = 7; 2.6%), A + B + CDT + ( n = 2; 0.7%), A–B–CDT + ( n = 2; 0.7%) and

A–B + CDT– ( n = 1; 0.4%).

2.2. Minimum inhibitory concentration (MIC) determination by agar

incorporation

MICs of a panel of 10 antimicrobial agents were determined

by the agar incorporation method as described by the Clinical

and Laboratory Standards Institute CLSI [7,8] . The panel com-

prised the first-line CDI therapies vancomycin and metronidazole

as well as fidaxomicin, rifaximin, clindamycin, erythromycin, amox-

icillin/clavulanic acid (AMC), moxifloxacin, meropenem and tetra-

cycline. The clinical breakpoints for vancomycin and metronidazole

were those recommended by the European Committee on Antimi-

crobial Susceptibility Testing (EUCAST) ( http://eucast.org ). For fi-

daxomicin, the European Medical Agency (EMA) proposed break-

point of ≥1 mg/L was used (report WC500119707; http://www.

ema.europa.eu/ ). Rifaximin resistance ( ≥32 mg/L) was as described

by O’Connor et al. [9] , and the breakpoints for clindamycin, ery-

thromycin, AMC, moxifloxacin, meropenem and tetracycline were

those provided by the CLSI [8] .

2.3. Statistical analysis

The Kruskal–Wallis rank-sum test and Dunn’s test were per-

formed to compare the geometric mean MICs between C. difficile

of food, compost and lawn origins. Fisher’s exact test, χ2 test and

post-hoc test were used to compare the resistance rates of C. diffi-

cile from different origins.

A23

S.-C. Lim et al. / International Journal of Antimicrobial Agents 52 (2018) 411–415 413

Table 1

Summary minimum inhibitory concentration (MIC) data of 274 Clostridium difficile isolates from food, compost and lawn.

Breakpoints (mg/L)

Agent Origin N S I R MIC range

(mg/L)

MIC 50

(mg/L)

MIC 90

(mg/L)

GM

(mg/L)

%R ( n ) %NR ( n ) P -value †

FDX a Food 56 – – ≥1 0.004–0.12 0.06 0.06 0.03 ‡ 0 (0) 100 (56)

Compost 36 0.004–0.5 0.06 0.12 0.06 0 (0) 100 (36)

Lawn 182 < 0.002–0.12 0.06 0.12 0.06 0 (0) 100 (182)

Total 274 < 0.002–0.5 0.06 0.12 0.05 0 (0) 100 (274) 1

VAN

b Food 56 ≤2 – > 2 0.5–2 2 2 1.41 0 (0) 100 (56)

Compost 36 1–2 1 2 1.26 0 (0) 100 (36)

Lawn 182 0.25–4 1 2 1.27 1.1 (2) 98.9 (180)

Total 274 0.25–4 1 2 1.29 0.7 (2) 99.3 (272) 1

MTR b Food 56 ≤2 – > 2 0.25–1 0.5 0.5 0.40 § 0 (0) 100 (56)

Compost 36 0.25–2 0.25 0.5 0.29 0 (0) 100 (36)

Lawn 182 0.12–1 0.25 0.5 0.30 0 (0) 100 (182)

Total 274 0.12–1 0.25 0.5 0.32 0 (0) 100 (274) 1

RFX c Food 56 – - ≥32 0.004–64 0.008 8 0.04 3.6 (2) 96.4 (54)

Compost 36 0.004 to > 64 0.008 0.25 0.02 2.8 (1) 97.2 (35)

Lawn 182 0.002–16 0.004 1 0.01 ‡ 0 (0) 100 (182)

Total 274 0.002 to > 64 0.008 4 0.02 1.1 (3) 98.9 (271) 0.037

CLI d Food 56 ≤2 4 ≥8 0.12–8 2 4 1.97 ‡ 7.1 (4) 92.9 (52)

Compost 36 1 to > 32 4 > 32 4.49 30.6 (11) 69.4 (25)

Lawn 182 0.12 to > 32 4 16 3.71 42.3 (77) 57.7 (105)

Total 274 0.12 to > 32 4 16 3.34 33.6 (92) 66.4 (182) < 0.0 0 01

ERY d Food 56 – – > 8 0.25–2 1 1 0.74 0 (0) 100 (56)

Compost 36 0.5 to > 256 1 > 256 2.94 § 19.4 (7) 80.6 (29)

Lawn 182 0.06 to > 256 1 2 0.81 1.1 (2) 98.9 (180)

Total 274 0.06 to > 256 1 2 0.94 3.3 (9) 96.7 (265) < 0.0 0 01

AMC d Food 56 ≤4 8 ≥16 0.125–1 0.5 0.5 0.41 0 (0) 100 (56)

Compost 36 0.06–4 0.25 1 0.44 0 (0) 100 (36)

Lawn 182 0.125–8 0.5 1 0.66 § 0 (0) 100 (182)

Total 274 0.06–8 0.5 1 0.57 0 (0) 100 (274) 1

MFX d Food 56 ≤2 4 ≥8 0.5–4 2 2 1.43 0 (0) 100 (56)

Compost 36 1–4 2 2 1.56 0 (0) 100 (36)

Lawn 182 0.25–8 2 2 1.40 1.1 (2) 98.9 (180)

Total 274 0.25–8 2 2 1.43 0.7 (2) 99.3 (272) 1

MEM

d Food 56 ≤4 8 ≥16 1–4 2 4 2.73 0 (0) 100 (56)

Compost 36 1 to > 16 4 4 3.30 ¶ 5.6 (2) 94.4 (34)

Lawn 182 0.5–16 2 4 2.42 0.5 (1) 99.5 (181)

Total 274 0.5 to > 16 2 4 2.58 1.1 (3) 98.9 (271) 0.071

TET d Food 56 ≤4 8 ≥16 0.06–64 0.12 0.25 0.15 1.8 (1) 98.2 (55)

Compost 36 0.06–64 0.12 16 0.49 § 13.9 (5) 86.1 (31)

Lawn 182 0.03–16 0.12 0.5 0.17 1.1 (2) 98.9 (180)

Total 274 0.03–64 0.12 0.5 0.19 2.9 (8) 97.1 (266) 0.002

S, susceptible; I, intermediate; R, resistant; MIC 50/90 , MIC required to inhibit 50% and 90% of the isolates, respectively; GM, geometric mean; NR, non-resistant; FDX,

fidaxomicin; VAN, vancomycin; MTR, metronidazole; RFX, rifaximin; CLI, clindamycin; ERY, erythromycin; AMC, amoxicillin/clavulanic acid; MFX, moxifloxacin; MEM,

meropenem; TET, tetracycline. a Resistance ( ≥1 mg/L) as recommended by the European Medical Agency (EMA) (report WC500119707; http://www.ema.europa.eu/ ). b Breakpoints are those recommended by European Committee on Antimicrobial Susceptibility Testing (EUCAST) ( http://eucast.org ) based on the epidemiological cut-off

values for the ‘wild-type’ population. c Resistance ( ≥32 mg/L) as described by O’Connor et al. [9] . d Breakpoints are those recommended for anaerobes by the Clinical and Laboratory Standards Institute (CLSI) [8] . † χ2 /Fisher’s exact to compare C. difficile %R of food, compost and lawn origins. ‡ GM is significantly lower than the other two origins (FDX, P < 0.0 0 01; RFX, P < 0.005; CLI, P < 0.0001). § GM is significantly higher than the other two origins (MTR, P < 0.0 0 01; ERY, P < 0.005; AMC, P < 0.0001; TET, P < 0.0001). ¶ GM is significantly higher than lawn isolates but not food isolates (MEM, P < 0.005).

3. Results

Fidaxomicin was the most active agent, showing potent in

vitro activity against all isolates (MIC 50 /MIC 90 , 0.06/0.12 mg/L; MIC

range < 0.002–0.5 mg/L) ( Table 1 ). This activity was superior to

the recommended first-line treatment agents for CDI, namely van-

comycin (MIC 50 /MIC 90 , 1/2 mg/L) and metronidazole (MIC 50 /MIC 90 ,

0.25/0.5 mg/L). However, no metronidazole resistance was ob-

served and only two isolates (0.7%) were resistant to vancomycin

(MIC = 4 mg/L; resistant breakpoint > 2 mg/L).

Phenotypic resistance to at least one antimicrobial agent was

observed in 103 (37.6%) of the 274 isolates, predominantly those

from compost (14/36; 38.9%) and lawn (82/182; 45.1%). Only 7

(12.5%) of 56 food isolates exhibited resistance. Multidrug resis-

tance (MDR), defined as resistance to at least one agent in three

or more antimicrobial categories, was observed in 4 (3.9%) of the

103 resistant isolates and all were of compost origin; one iso-

late each of QX 327 (A–B–CDT–), QX 140 (A–B–CDT–) and UK 046

(A + B + CDT–) were resistant to clindamycin (MIC > 32 mg/L), ery-

thromycin (MIC > 256 mg/L) and tetracycline (MICs = 64, 16 and

32 mg/L, respectively), and one isolate of UK 012 (A + B + CDT–) was

resistant to rifaximin (MIC > 64 mg/L), erythromycin (MIC > 256

mg/L) and tetracycline (MIC = 64 mg/L). Almost one-third (4/14) of

the resistant compost isolates were MDR.

There was a significant association between the source of iso-

lates and resistance to rifaximin (3.6% from food, 2.8% from com-

A24

414 S.-C. Lim et al. / International Journal of Antimicrobial Agents 52 (2018) 411–415

post and 0.0% from lawn; P = 0.037, Fisher’s exact test), clindamycin

(7.1% from food, 30.6% from compost and 42.3% from lawn; P <

0.0 0 01, χ2 ), erythromycin (0.0% from food, 19.4% from compost

and 1.1% from lawn; P < 0.0 0 01, Fisher’s exact test) and tetracy-

cline (1.8% from food, 13.9% from compost and 1.1% from lawn;

P = 0.002, Fisher’s exact test) ( Table 1 ). Compared with food and

lawn isolates, compost isolates were more often resistant to ery-

thromycin (compost vs. food, P = 0.001, post-hoc test; compost vs.

lawn, P = 0.0 0 02, post-hoc test) and tetracycline (compost vs. food,

P = 0.049, post-hoc test; compost vs. lawn, P = 0.005, post-hoc test)

( Table 1 ). The proportion of lawn isolates resistant to clindamycin

was similar to that of compost isolates ( P = 0.26, post-hoc test);

however, both were significantly higher than food isolates (food vs.

compost, P = 0.01, post-hoc test; food vs. lawn, P < 0.0 0 01, post-

hoc test). Susceptibility to fidaxomicin, vancomycin, metronidazole,

AMC, moxifloxacin and meropenem did not differ significantly be-

tween isolates from different sources.

4. Discussion

Whilst there has been an increase in publications on the preva-

lence of C. difficile in food and the environment, few reports have

investigated the antimicrobial susceptibility of the C. difficile iso-

lates recovered [10,11] . This study is the first to determine the an-

timicrobial resistance patterns of food and environmental C. difficile

isolates in Australia.

As macrolides and tetracycline-based antimicrobial agents con-

stitute ca. 40% (112.2 tonnes per year) of all antimicrobials used

in Australian food animals [12] , it was not surprising that C. dif-

ficile isolates from compost exhibited resistance to erythromycin

and tetracycline. Although clindamycin is not approved for use in

food animals, previous studies on animal C. difficile strains have

frequently reported intermediate or resistant MICs for clindamycin

[3,13] . Our previous study on C. difficile RT 014 from pigs showed

high (69%) non-susceptibility to clindamycin, erythromycin and

tetracycline but 100% susceptibility to fidaxomicin, vancomycin,

metronidazole, rifaximin, AMC, moxifloxacin and meropenem [3] ,

in agreement with the resistance patterns of compost isolates in

the current study. Taken together, these findings support our the-

ory that compost isolates are likely to be of animal origin as C.

difficile spores will survive the composting process [14] . Although

not indicative of transmission to humans, both compost and lawn

isolates shared antimicrobial resistance/susceptibility patterns sim-

ilar to those reported in human-derived isolates [15] . In 2015, an-

timicrobial susceptibility testing was performed on 440 human

C. difficile isolates collected across five Australian states [15] . In

that study, the majority of isolates were susceptible to fidax-

omicin (100%), vancomycin (100%), metronidazole (100%), rifaximin

(100%), AMC (100%), moxifloxacin (96.1%) and meropenem (99.5%);

similar to the findings for food, compost and lawn isolates. Fur-

thermore, C. difficile that causes human CDI in Australia has a high

prevalence of clindamycin resistance (84.3%) [15] . In the current

study, both compost and lawn isolates had relatively high levels of

clindamycin resistance with MICs comparable with isolates from

humans (MIC 50 /MIC 90 , 8/ > 32 mg/L; MIC range 0.5 mg/L to > 32

mg/L) [15] . Further studies using whole-genome sequencing are

necessary to better determine the relatedness of these compost

and lawn isolates with C. difficile isolated from humans, particu-

larly in the community.

Clostridium difficile RT 014/020 is consistently one of the most

frequently isolated toxigenic RTs in humans worldwide, including

Australia [16–18] . Of public-health interest, in this study, C. difficile

RT 014/020 exhibited significantly greater resistance to clindamycin

(35/78; 44.9%), a reported risk factor for CA-CDI [19] , compared

with other less common RTs (54/196; 27.6%) ( P = 0.008, χ2 ). Fur-

thermore, two of the four MDR isolates (RTs 012 and 046 from

compost) were toxigenic and RTs that have previously been associ-

ated with human CDI [17] . This suggests that environmental C. dif-

ficile could be a reservoir for antimicrobial resistance genes of clini-

cal relevance. In addition, exposure to toxigenic and antimicrobial-

resistant food and environmental C. difficile poses a risk of infec-

tion to susceptible individuals and re-infection of a resolved pa-

tient with a new strain of C. difficile while the gut microbiota is

still compromised.

Based on antimicrobial resistance patterns, the C. difficile strains

from compost appeared quite different to strains from vegeta-

bles and lawn, with greater resistance to erythromycin, tetracycline

and/or clindamycin in compost isolates. This may be a reflection

of a sampling bias as it is possible that the compost used to fer-

tilise the vegetables and lawn came from different farms with dif-

ferent antimicrobial usage. It was impossible to determine which

animal farm the manure originally came from due to issues of

confidentiality. Furthermore, acquisition and loss of antibiotic re-

sistance genes occurs readily in C. difficile in response to selective

pressure, as C. difficile has a diverse and highly flexible accessory

genome comprising a range of mobile genetic elements conferring

resistance to macrolides/lincosamides [Tn 6194 /Tn 5398 ( ermB )] and

tetracycline [Tn 916 /Tn 5397 ( tetM )], many of which are capable of

interspecies and intraspecies transfer in vitro [1,3,20] .

In summary, similarities in antimicrobial resis-

tance/susceptibility patterns were observed between environ-

mental C. difficile isolates and those of animal and human origin.

Future genomic studies are required to determine whether these

isolates do indeed originate from animals and are responsible for

CA-CDI. Nevertheless, this study shows that food or the environ-

ment harbouring toxigenic C. difficile strains could be sources for

CDI in the community. This study provides a baseline for future

surveillance of antimicrobial resistance in food and environmental

C. difficile in Australia.

Acknowledgements

The authors are grateful to Sally Duncan and colleagues at Path-

West Media (Mt Claremont, WA, Australia) for preparation of the

susceptibility testing culture media. D.R.K is funded by a National

Health and Medical Research Council (NHMRC) Early Career Fel-

lowship (APP1138257).

Funding

This study was supported by PathWest internal funding.

Competing interests

None declared.

Ethical approval

Not required.

References

[1] Knight DR , Elliott B , Chang BJ , Perkins TT , Riley TV . Diversity and evolution in the genome of Clostridium difficile . Clin Microbiol Rev 2015;28:721–41 .

[2] Slimings C , Armstrong P , Beckingham WD , Bull AL , Hall L , Kennedy KL , et al. Increasing incidence of Clostridium difficile infection, Australia,

2011–2012. Med J Aust 2014;200:272–6 . [3] Knight DR , Squire MM , Collins DA , Riley TV . Genome analysis of Clostridium

difficile PCR ribotype 014 lineage in Australian pigs and humans reveals a diverse genetic repertoire and signatures of long-range interspecies transmis-

sion. Front Microbiol 2017;7:2138 .

[4] Warriner K , Xu C , Habash M , Sultan S , Weese SJ . Dissemination of Clostridium

difficile in food and the environment: significant sources of C. difficile commu-

nity-acquired infection? J Appl Microbiol 2017;122:542–53 . [5] Lim SC , Foster NF , Elliott B , Riley TV . High prevalence of Clostridium difficile on

retail root vegetables, Western Australia. J Applied Microbiol 2018;124:585–90 .

A25

S.-C. Lim et al. / International Journal of Antimicrobial Agents 52 (2018) 411–415 415

[6] Moono P , Lim SC , Riley TV . High prevalence of toxigenic Clostridium difficile in public space lawns in Western Australia. Sci Rep 2017;7:41196 .

[7] Clinical and Laboratory Standards Institute. M11-A7: methods for antimicrobial susceptibility testing of anaerobic bacteria. seventh edition. Wayne, PA: CLSI;

2011 . [8] Clinical and Laboratory Standards Institute. M100-S23: performance standards

for antimicrobial susceptibility testing; twenty-third informational supplement. Wayne, PA: CLSI; 2013 .

[9] O’Connor JR , Galang MA , Sambol SP , Hecht DW , Vedantam G , Gerding DN ,

et al. Rifampin and rifaximin resistance in clinical isolates of Clostridium difficile . Antimicrob Agents Chemother 2008;52:2813–17 .

[10] Eckert C , Burghoffer B , Barbut F . Contamination of ready-to-eat raw vegetables with Clostridium difficile in France. J Med Microbiol 2013;62:1435–8 .

[11] Zidaric V , Beigot S , Lapajne S , Rupnik M . The occurrence and high diversity of Clostridium difficile genotypes in rivers. Anaerobe 2010;16:371–5 .

[12] Australian Pesticides and Veterinary Medicines Authority. Report on the quality

of antimicrobial products sold for veterinary use in Australia: July 2005 to June 2010, https://apvma.gov.au/node/11816 ; 2014 [accessed 10.01.18] .

[13] Pirš T , Avberšek J , Zdovc I , Krt B , Andlovic A , Lejko-Zupanc T , et al. Antimicro- bial susceptibility of animal and human isolates of Clostridium difficile by broth

microdilution. J Med Microbiol 2013;62:1478–85 . [14] Xu C , Wang D , Huber A , Weese SJ , Warriner K . Persistence of Clostridium

difficile in wastewater treatment-derived biosolids during land application or

windrow composting. J Appl Microbiol 2016;120:312–20 .

[15] Knight DR , Giglio S , Huntington PG , Korman TM , Kotsanas D , Moore CV . Surveillance for antimicrobial resistance in Australian isolates of Clostridium

difficile , 2013–14. J Antimicrob Chemother 2015;70:2992–9 . [16] Bauer MP , Notermans DW , van Benthem BH , Brazier JS , Wilcox MH , Rupnik M ,

et al. Clostridium difficile infection in Europe: a hospital-based survey. Lancet 2011;377:63–73 .

[17] Cheng AC , Collins DA , Elliott B , Ferguson JK , Paterson DL , Thean S , et al. Labora- tory-based surveillance of Clostridium difficile circulating in Australia, Septem-

ber–November 2010. Pathol 2016;48:257–60 .

[18] Collins DA , Hawkey PM , Riley TV . Epidemiology of Clostridium difficile infection in Asia. Antimicrob Resist Infect Control 2013;2:21 .

[19] Brown KA , Khanafer N , Daneman N , Fisman DN . Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection. Antimicrob

Agents Chemother 2013;57:2326–32 . [20] Spigaglia P . Recent advances in the understanding of antibiotic resistance in

Clostridium difficile infection. Ther Adv Infect Dis 2016;3:23–42 .

A26

Susceptibility of Clostridium difficile to the food preservatives sodiumnitrite, sodium nitrate and sodium metabisulphite

Su-Chen Lim a, Niki F. Foster a, b, Thomas V. Riley a, b, *

a Microbiology & Immunology, School of Pathology and Laboratory Medicine, The University of Western Australia, Queen Elizabeth II Medical Centre,Nedlands 6009, Western Australia, Australiab Department of Microbiology, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands 6009, Western Australia, Australia


Article history:Received 30 July 2015Received in revised form30 November 2015Accepted 12 December 2015Available online 15 December 2015

Keywords:Clostridium difficileReady-to-eat meatSodium nitriteSodium nitrateSodium metabisulphite

a b s t r a c t

Clostridium difficile is an important enteric pathogen of humans and food animals. Recently it has beenisolated from retail foods with prevalences up to 42%, prompting concern that contaminated foods maybe one of the reasons for increased community-acquired C. difficile infection (CA-CDI). A number ofstudies have examined the prevalence of C. difficile in raw meats and fresh vegetables; however, fewerstudies have examined the prevalence of C. difficile in ready-to-eat meat. The aim of this study was toinvestigate the in vitro susceptibility of 11 C. difficile isolates of food animal and retail food origins to foodpreservatives commonly used in ready-to-eat meats. The broth microdilution method was used todetermine the minimum inhibitory concentrations (MIC) and minimum bactericidal concentrations(MBC) for sodium nitrite, sodium nitrate and sodium metabisulphite against C. difficile. Checkerboardassays were used to investigate the combined effect of sodium nitrite and sodium nitrate, commonlyused in combination in meats. Modal MIC values for sodium nitrite, sodium nitrate and sodium meta-bisulphite were 250 mg/ml, >4000 mg/ml and 1000 mg/ml, respectively. No bactericidal activity wasobserved for all three food preservatives. The checkerboard assays showed indifferent interaction be-tween sodium nitrite and sodium nitrate. This study demonstrated that C. difficile can survive in thepresence of food preservatives at concentrations higher than the current maximum permitted levelsallowed in ready-to-eat meats. The possibility of retail ready-to-eat meats contaminated with C. difficileacting as a source of CDI needs to be investigated.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Clostridium difficile is an anaerobic, spore-forming, Gram-posi-tive bacillus widely found in soil, water and the gastrointestinaltracts of food animals and humans. It causes mild to severe diarrheaand, occasionally, the more serious pseudomembranous colitis andtoxic megacolon in humans. Since 2000, there has been a globalincrease in C. difficile infection (CDI) with heightened severity andmortality, and a rise in community-acquired infection (CA-CDI) inindividuals without traditional risk factors of old age or antibioticusage [1e5]. This has been mainly due to the emergence of so-called hypervirulent strains of C. difficile, particularly PCR-

ribotypes 027 and 078, that produce binary toxin (CDT) in addi-tion to toxins A and B.

Recently, C. difficile has been found in retail meats, seafoods andvegetables with prevalences up to 42% [6e12]. C. difficile of thesame ribotype has been found in foods, food animals and humans[13]. In Canada, Weese et al. (2009) found C. difficile ribotype 078,common in food animals and a cause of disease in humans, inground meats and poultry [14,15]. In Scotland, ready-to-eat saladswere contaminated with C. difficile ribotypes 017 and 001; both arecommon clinical isolates in Scotland and Europe [16e18]. Thesefindings have led to growing concern that retail foods contami-nated with C. difficile may be one of the reasons for the increasedincidence of CDI, particularly in the community. Despite the po-tential for foodborne transmission of C. difficile, there are only asmall number of studies that have looked at the prevalence ofC. difficile in foods, most of which focused on raw meats and freshvegetables. To our knowledge, only two studies have investigatedthe prevalence of C. difficile in raw sausages and only one study has

* Corresponding author. Microbiology & Immunology, School of Pathology andLaboratory Medicine, The University of Western Australia, Queen Elizabeth IIMedical Centre, Nedlands 6009, Western Australia, Australia.

E-mail address: [email protected] (T.V. Riley).


Anaerobe

journal homepage: www.elsevier .com/locate/anaerobe

http://dx.doi.org/10.1016/j.anaerobe.2015.12.0041075-9964/© 2015 Elsevier Ltd. All rights reserved.

Anaerobe 37 (2016) 67e71

A27

investigated the prevalence of C. difficile in ready-to-eat meats suchas cured, fermented, smoked or cooked meat [7,19].

Commercially produced ready-to-eat meats including non-heattreated meats often contain food preservatives to prevent spoilageand extend shelf life. Sodium nitrite (E250) and nitrate (E251) arefood preservatives widely used in combination in ready-to-eatmeats. Nitrite helps to develop flavor, reacts with myoglobin toproduce nitrosylhaemochrome which gives the characteristic pinkcolour of processed meat and inhibits the growth of spoilage andpathogenic bacteria such as Clostridium botulinum [20]. Nitrate actsas a reservoir for nitrite production through the enzyme nitratereductase produced by microorganisms such as staphylococci andlactobacilli. Sodium metabisulphite (E223) is a food preservativeused in raw sausages. The use of food preservatives in foods isstrictly regulated for food safety reasons. In Australia and NewZealand, the use of nitrite, nitrate and metabisulphite in meatproducts is restricted to 125 ppm (mg/kg), 500 ppm and 500 ppm,respectively [21,22]. In the USA, the use of nitrite and nitrate inmeat products is not allowed to exceed 200 ppm and 500 ppm,respectively [23], while the European Union set a maximum of150 ppm nitrate and nitrite added to dry-fermented sausages and250 ppm nitrate in long-cured products when no nitrite is added[24].

The aim of this studywas to investigate the in vitro susceptibilityof C. difficile to three food preservatives commonly use incommercially produced ready-to-eat meats: sodium nitrite (E250),sodium nitrate (E251) and sodium metabisulphite (E223), todetermine if commercially produced ready-to-eat meats may posea risk for CDI. Checkerboard assays were used to investigate thecombined effect of sodium nitrite and sodium nitrate.

2. Methods and materials

2.1. Strains and growth conditions

Eleven C. difficile isolates were used in this study (Table 1).Isolates were selected to represent a range of sources of C. difficilehaving been isolated from retail foods and food animals. All isolateswere routinely cultured on pre-reduced supplemented Brucellaagar (Becton, Dickinson and company) and blood agar underanaerobic conditions (80% N2, 10% CO2 and 10% H2) at 37 �C in ananaerobic chamber (Don Whitley Scientific) for 24e48 h.

2.2. Food preservatives

The following food preservatives were used in this study: so-dium nitrite (Thermo Fisher, Australia), sodium nitrate (BDH,Australia) and sodium metabisulphite (BDH, Australia). All

preservative solutions were made fresh before use by dissolvingpure substance in sterile distilled water.

2.3. Broth microdilution assay

To determine the susceptibility of C. difficile to food pre-servatives, broth microdilution assays were performed according tothe Clinical and Laboratory Standards Institute methodology [25].Briefly, a series of two-fold dilutions of each preservative with finalconcentrations ranging from 0 to 4, 000 mg/ml was made in a 96-well microtitre plate in pre-reduced supplemented Brucella broth.Suspensions of test organisms cultured on supplemented Brucellaagar were adjusted to 0.5 McFarland in 0.85% saline, and thendiluted in supplemented Brucella broth to correspond to a finalinoculum concentration of approximately 1.0 х 106 cfu/ml. The finalinoculum concentration was confirmed by viable counts. Sporesconstituted approximately 15% of cells, determined by a Schaffer/Fulton spore stain viewed under a light microscope.

After 24 h incubation, minimum inhibitory concentrations(MICs) were determined visually as the lowest concentration offood preservative resulting in an optically clear microtitre well.Minimum bactericidal concentrations (MBCs) were determined bysub-culturing 10 ml from each well of the microtitre plate at 24 h,spot inoculating onto ChromID agar (bioM�erieux) and incubatedanaerobically for 24 h at 37 �C. After incubation, colonies werecounted and compared to the counts of original inoculum. TheMBCwas determined as the lowest concentration of food preservativeresulting in �99.9% death of the inoculum [26]. Purity checks oninoculum suspensions were performed by subculturing an aliquotfrom the inoculatedwell that contained the lowest concentration offood preservative after serial dilutions, onto two blood agar platesfor simultaneous incubation both aerobically and anaerobically. Avancomycin control was included to ensure testing quality. The MICvalues obtained for vancomycin needed to fall within the accept-able range of 0.5e4 mg/ml as indicated by the CLSI guidelines. Toensure the methodology was valid to quantify the inhibitory effectof food preservatives, Staphylococcus aureus ATCC 29213 was usedas a positive control and testing performed according to CLSIguidelines [26,27]. The MIC and MBC values of sodium meta-bisulphite against S. aureus ATCC 29213 needed to be � 512 mg/mlas reported by Frank and Patel [28]. All isolates were tested on atleast three separate occasions and modal MICs and MBCs weredetermined.

2.4. Checkerboard assay

Serial two-fold dilutions of 8000 mg/ml sodium nitrite weremade in a vertical orientation in a 96-well microtitre plate using

Table 1Origin and molecular characteristics of C. difficile isolates used in this study.

C. difficile isolates Origin Ribotype Toxin profile Reference

ATCC 700057 e UK 038 A�B�CDT�Foods originCd1001 Pork meat, Canada E AþBþCDT� [14]Cd1002 Ground beef, Canada UK 027 AþBþCDTþ [14]Cd1006 Ground beef, Canada UK 078 AþBþCDTþ [14]Cd1009 Chicken meat, Canada UK 078 AþBþCDTþ [15]Cd1017 Ground beef, Canada C (UK 251) AþBþCDTþ [14]Cd1106 Chicken meat, Canada UK 014 AþBþCDT�Food animals originAI35 Piglets, Australia UK 237 A�BþCDTþ [39]AI204 Calves, Australia UK 033 A�B�CDTþ [40]AI218 Calves, Australia UK 127 AþBþCDTþ [40]AI273 Calves, Australia UK 126 AþBþCDTþ [40]

S.-C. Lim et al. / Anaerobe 37 (2016) 67e7168

A28

100 ml of sterile distilled water. Nine doubling dilutions of 16,000 mg/ml sodium nitrate, at four times the intended final con-centration, were prepared using sterile distilled water. Fifty-microliter aliquots of each sodium nitrate dilution were added ina horizontal orientation so that the plate contained various con-centration combinations of the two preservatives. Growth from 24to 48 h Brucella agar cultures of the test organisms was adjusted to2.0McFarland in saline, and then diluted in four times concentratedsupplemented Brucella broth to correspond to a final inoculumconcentration of approximately 1.0 � 106 cfu/ml. Each well wasinoculated with 50 ml of inoculum and incubated anaerobically at37 �C for 24e48 h. Purity checks and a vancomycin control wereincluded to ensure testing quality as described above. Three inde-pendent replicates were performed, modal MICs were determined.

The fractional inhibitory concentration (FIC) was calculated bydividing the MIC of sodium nitrite and sodium nitrate in combi-nation by the MIC of sodium nitrite or sodium nitrate alone. The FICindex (FICI) was obtained by adding both FICs. When the FICI was�0.5, synergywas indicated; an FICI of >0.5 to 4.0 was defined as anindifferent interaction; and antagonism was defined as an FICI of>4.0 [29].

3. Results and discussion

The MICs and MBCs of sodium nitrite, sodium nitrate and so-diummetabisulphite against the control S. aureus ATCC 29213 were>4000 mg/ml, >4000 mg/ml and 500 mg/ml, respectively. The MICsand MBCs of sodium nitrite, sodium nitrate and sodium meta-bisulphite against C. difficile are shown in Table 2. Sodium nitriteinhibited the growth of all tested C. difficile isolates at 125e500 mg/ml with an average of 250 mg/ml which is higher than themaximum permitted level allowed in ready-to-eat meats in theUnited States (200 ppm), Europe (150 ppm), Australia (125 ppm)and New Zealand (125 ppm). This raises possible food safety con-cerns as many commercially produced ready-to-eat meats are alsonon-heat treated cured or fermented meats. While there are nopublished data on the effects of food preservatives against C. difficilein broth or in meats, these MICs were similar to those reported forClostridium perfringens [30]. Labbe and Duncan found 200e400 mg/ml of nitrite was required to inhibit spore outgrowth at pH 6. Theinhibitory mechanisms of nitrite are not well understood, but ni-trite inhibits microbes more effectively at low pH suggesting thatthe antimicrobial action is likely associated with the generation ofnitric oxide or nitrous acid [30]. Studies on Clostridium sporogenesand C. perfringens spores showed that nitrite prevented vegetative

cell division and inhibited the emergence of vegetative cells fromspores while allowing germination to occur [30,31].

Our results suggest that nitrites can inhibit C. difficile growth butare unable to kill the organism. This is consistent with a study onC. botulinum spores where C. botulinum was recovered from artifi-cially seeded, commercially formulated and processed sausagesafter storage at 8 �C for 5 weeks regardless of the concentration ofnitrite added, with the highest C. botulinum counts detected innitrite-free products [32].

Sodium nitrate had no measurable effect on the growth ofC. difficile at the maximum permitted levels (Table 2). Even at4000 mg/ml, which is at least eight times the concentration nor-mally used in commercially produced ready-to-eat meats, thepreservative failed to inhibit C. difficile growth. This is consistentwith a previous in vitro study where nitrate had no effect on sporegermination and outgrowth of C. sporogenes NCA 3679 [31].

Sodium metabisulphite, a food preservative used in raw sau-sages, inhibited the growth of C. difficile at 1000 mg/ml (Table 2)which is higher than the allowed level of 500 ppm in Australia. Nobactericidal activity against C. difficile was observed at the highestconcentration tested of 4000 mg/ml. There are no published data onthe effect of sodium metabisulphite on spore-forming bacteria.

Checkerboard assays of sodium nitrate and sodium nitrite gave aFICI in the range of 1.5e3, indicating an indifferent interactionbetween the preservatives (Table 3). However, this is unlikely to berepresentative of the combined effects of these preservatives inmeat products. Inmeatmatrices, other commensal microorganismsthat are present reduce nitrate to nitrite [33], altering their relativeconcentrations and therefore their inhibitory ability. In the pres-ence of these microorganisms, nitrate might show some effect onthe growth of C. difficile as it has on delaying botulinum toxinproduction [34,35].

Although the reported MICs of food preservatives are higherthan their maximum permitted levels allowed in commerciallyproduced ready-to-eat meats, it is doubtful that C. difficilewould beable to grow in ready-to-eat meats when stored at the recom-mended temperature of 5 �C. This would require the meats to bestored at abusive temperatures for a long period of time. Never-theless the recent findings of C. difficile in raw meats with preva-lences ranging up to 42% [6e10] are a cause for concern as thesemay be used to make ready-to-eat meats, especially non-heattreated ready-to-eat meats such as cured or fermented products.The possibility of ready-to-eat meats contaminated with C. difficileneeds to be investigated given that C. difficile can survive and growin the presence of food preservatives commonly used in meat

Table 2Susceptibility of C. difficile to food preservatives.

C. difficile isolates Sodium nitrite Sodium nitrate Sodium metabisulphite

MIC (mg/ml) MBC (mg/ml) MIC (mg/ml) MBC (mg/ml) MIC (mg/ml) MBC (mg/ml)

ATCC 700057 250 1000 >4000 >4000 1000 2000<i>a (1000 e >4000)</i>

Cd1001 125a (15.33e250) 250a (62.5 e 1000) 2000 >4000 1000 1000Cd1002 500 >4000 >4000 >4000 1000 >4000Cd1006 500 >4000 >4000 >4000 1000 >4000Cd1009 250a (125e500) >4000 >4000 >4000 1000 >4000Cd1017 125a (31.25e250) >4000 >4000 >4000 500 >4000Cd1106 250a (12e500) >4000 >4000 >4000 1000 >4000AI35 500 >4000 >4000 >4000 1000 >4000AI204 250a (125e500) >4000 >4000 >4000 1000 >4000AI218 125 >4000 >4000 >4000 1000 >4000AI273 250a (125e500) >4000 >4000 >4000 1000 >4000Mode 250 >4000 >4000 >4000 1000 >4000

MIC, minimum inhibitory concentration; MBC, minimum bactericidal concentration.The values are the mode from at least three separate experiments.

a Median (range) was reported when independent replicates yielded different MICs, hence no mode could be determined.

S.-C. Lim et al. / Anaerobe 37 (2016) 67e71 69

A29

products. There is potential for foodborne transmission of C. difficilethrough commercially produced ready-to-eat meats as has beenreported for other Clostridia spp [36e38].

A limitation of this study is that it used C. difficile cultures whichat the time of testing contained approximately 15% of spores.Further research is required to study the effect of food preservativeson C. difficile spores and vegetative cells separately. The impact ofpH, temperature and other commensal microorganisms, possiblythrough the use of meat matrices, also needs exploring.

This study represents the first to investigate the effect of foodpreservatives on C. difficile. Although the relevance of contaminatedfood in the epidemiology of CDI is not yet known, the ability ofC. difficile to survive in the presence of food preservatives indicatesa need to investigate the prevalence of C. difficile in commerciallyproduced ready-to-eat meats as they may serve as a source ofinfection for people in hospitals and the community.

Acknowledgments

We thank Professor Scott Weese at the Department of Patho-biology, University of Guelph, Guelph, ON, Canada for providing theC. difficile isolates of food origin used in this study.

References

[1] J. Freeman, et al., The changing epidemiology of Clostridium difficile infections,Clin. Microbiol. Rev. 23 (2010) 529e549.

[2] H. Honda, E.R. Dubberke, The changing epidemiology of Clostridium difficileinfection, Curr. Opin. Gastroenterol. 30 (2014) 54e62.

[3] E.J. Kuijper, J.T. van Dissel, M.H. Wilcox, Clostridium difficile: changing epide-miology and new treatment options, Curr. Opin. Infect. Dis. 20 (2007)376e383.

[4] M.H. Wilcox, et al., Changing epidemiology of Clostridium difficile infectionfollowing the introduction of a national ribotyping-based surveillance schemein England, Clin. Infect. Dis. 55 (2012) 1056e1063.

[5] J. Pepin, et al., Clostridium difficile-associated diarrhea in a region of Quebecfrom 1991 to 2003: a changing pattern of disease severity, Can. Med. Assoc. J.171 (2004) 466e472.

[6] E. Hoffer, et al., Low occurrence of Clostridium difficile in fecal samples ofhealthy calves and pigs at slaughter and in minced meat in Switzerland,J. Food Prot. 73 (2010) 973e975.

[7] J.G. Songer, et al., Clostridium difficile in retail meat products, USA, 2007,Emerg. Infect. Dis. 15 (2009) 819e821.

[8] R.B. Harvey, et al., Clostridium difficile in poultry and poultry meat, FoodbornePathog. Dis. 8 (2011) 1321e1323.

[9] A. Rodriguez-Palacios, et al., Possible seasonality of Clostridium difficile in retailmeat, Canada, Emerg. Infect. Dis. 15 (2009) 802e805.

[10] S.M. Von Abercron, et al., Low occurrence of Clostridium difficile in retailground meat in Sweden, J. Food Prot. 72 (2009) 1732e1734.

[11] D. Metcalf, et al., Clostridium difficile in seafood and fish, Anaerobe 17 (2011)85e86.

[12] C. Eckert, B. Burghoffer, F. Barbut, Contamination of ready-to-eat raw vege-tables with Clostridium difficile in France, J. Med. Microbiol. 62 (2013)1435e1438.

[13] A. Rodriguez-Palacios, et al., Clostridium difficile in foods and animals: historyand measures to reduce exposure, Animal Health Res. Rev. 14 (2013) 11e29.

[14] J.S. Weese, et al., Detection and enumeration of Clostridium difficile spores inretail beef and pork, Appl. Environ. Microbiol. 75 (2009) 5009e5011.

[15] J.S. Weese, et al., Detection and characterization of Clostridium difficile in retailchicken, Lett. Appl. Microbiol. 50 (2010) 362e365.

[16] M.M. Bakri, et al., Clostridium difficile in ready-to-eat salads, Scotl. Emerg.Infect. Dis. 15 (2009) 817e818.

[17] E. Mutlu, et al., Molecular characterization and antimicrobial susceptibilitypatterns of Clostridium difficile strains isolated from hospitals in south-eastScotland, J. Med. Microbiol. 56 (2007) 921e929.

[18] R.J. van den Berg, et al., Characterization of toxin A-negative, toxin B-positiveClostridium difficile isolates from outbreaks in different countries by amplifiedfragment length polymorphism and PCR ribotyping, J. Clin. Microbiol. 42(2004) 1035e1041.

[19] R.B. Harvey, et al., Clostridium difficile in retail meat and processing plants inTexas, J. Vet. Diagnostic Invest. 23 (2011) 807e811.

[20] J.J. Sindelar, A.L. Milkowski, Human safety controversies surrounding nitrateand nitrite in the diet, Nitric Oxide-Biol. Chem. 26 (2012) 259e266.

[21] Food Standards Australia and New Zealand, Food Standards Code- Standard1.3.1- Food Additives, 2007e2008. http://www.foodstandards.gov.au(accessed 18.06.15.).

[22] Food Standards Australia and New Zealand, Initial Assessment Report- Pro-posal P298- Benzoate and Sulphite Permissions in Food of 3.8, 2005. http://www.foodstandards.gov.au (accessed 18.06.15.).

[23] Code of Federal Regulations, Title 21, Volume 3 USA, Food and Drugs. Part 170,Food Additives, 170.60, 172.175, 172.70, 2005.

[24] Directive, Directive 2006/52/EC of the European Parliament and of the Councilof 5 July 2006 Amending Directive 95/2/EC on Food Additives Other thanColours and Sweeteners and Directive 95/35/EC on Sweeteners for Use inFoodstuffs, O.J.L204 of 26.7.2006, 2006.

[25] CLSI, Methods for Antimicrobial Susceptibility Testing of Anaerobic Bacreria;Approved Standard. Seventh Edition.CLSI Document M11eA7, 2009 (Wayne,PA).

[26] CLSI, Methods for Determining Bactericidal Activity of Antimicrobial Agents;Approved Guideline. CLSI Document M26-A, 1999 (Wayne, PA).

[27] CLSI, Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria thatGrow Aerobically; Approved Standard. Sixth Edition. CLSI Document M7eA6,2003 (Wayne, PA).

[28] K.L. Frank, R. Patel, Activity of sodium metabisulfite against planktonic andbiofilm Staphylococcus species, Diagnostic Microbiol. Infect. Dis. 57 (2007)355e359.

[29] F.C. Odds, Synergy, antagonism, and what the chequerboard puts betweenthem, J. Antimicrob. Chemother. 52 (2003) 1.

[30] R.G. Labbe, C.L. Duncan, Growth from spores of Clostridium perfringens in thepresence of sodium nitrite, Appl. Microbiol. 19 (1970) 353e359.

[31] C.L. Duncan, et al., Effect of sodium nitrite, sodium chloride, and sodium ni-trate on germination and outgrowth of anaerobic spores, Appl. Microbiol. 16(1968) 406e411.

[32] R. Keto-Timonen, et al., Inhibition of toxigenesis of group II (nonproteolytic)Clostridium botulinum type B in meat products by using a reduced level ofnitrite, J. Food Prot. 75 (2012) 1346e1349.

[33] N. Skovgaard, Microbiological aspects and technological need: technologicalneeds for nitrates and nitrites, Food Addit. Contam. 9 (1992) 391e397.

[34] G.O. Hustad, et al., Effect of sodium nitrite and sodium nitrate on botulinaltoxin production and nitrosamine formation in wieners, Appl. Microbiol. 26(1973) 22e26.

[35] L.N. Christiansen, et al., Effect of nitrite and nitrate on toxin production byClostridium botulinum and on nitrosamine formation in perishable cannedcomminuted cured meat, vol. 25, 1973, p. 357.

[36] N.J. Golden, et al., Risk assessment for Clostridium perfringens in ready-to-eatand partially cooked meat and poultry products, J. Food Prot. 72 (7) (2009)1376e1384.

[37] I. Gillespie, C. Little, R. Mitchell, Microbiological examination of cold ready-to-eat sliced meats from catering establishments in the United Kingdom, J. Appl.Microbiol. 88 (3) (2000) 467e474.

Table 3In vitro activities of sodium nitrate and sodium nitrite individually and in combination against C. difficile.

C. difficile isolates MIC (mg/ml) FICI Interactive category

Sodium nitrate Sodium nitrite Sodium nitrate-sodium nitrite combination

ATCC 700057 >4000 125 >4000/125 2 IndifferentCd1001 >4000 125 >4000/125 2 IndifferentCd1002 >4000 125 >4000/125 2 IndifferentCd1006 >4000 250 >4000/250 2 IndifferentCd1009 >4000 250 >4000/250 2 IndifferentCd1017 >4000 125 >4000/250 3 IndifferentCd1106 >4000 250 >4000/250 2 IndifferentAI35 >4000 500 >4000/500 2 IndifferentAI204 >4000 500 >4000/250 1.5 IndifferentAI218 >4000 250 >4000/250 2 IndifferentAI273 >4000 500 >4000/500 2 Indifferent

S.-C. Lim et al. / Anaerobe 37 (2016) 67e7170

A30

[38] K.A. McKay, P.I. Kumchy, Incidence and significance of Clostridia isolated fromready-to-eat meats and rendered products, Can. J. Comp. Med. 33 (4) (1969)307.

[39] M.M. Squire, et al., Novel molecular type of Clostridium difficile in neonatal

pigs, Western Australia, Emerg. Infect. Dis. 19 (2013) 790e792.[40] D.R. Knight, et al., Cross-sectional study reveals high prevalence of Clostridium

difficile non-PCR ribotype 078 strains in Australian veal calves at slaughter,Appl. Environ. Microbiol. 79 (2013) 2630e2635.

S.-C. Lim et al. / Anaerobe 37 (2016) 67e71 71

A31

Research paperAntimicrobial susceptibility of anaerobic bacteria

Antimicrobial resistance in large clostridial toxin-negative, binarytoxin-positive Clostridium difficile ribotypes

Grace O. Androga a, b, 1, Daniel R. Knight c, 1, Su-Chen Lim a, Niki F. Foster d,Thomas V. Riley a, b, c, e, *

a School of Biomedical Sciences, The University of Western Australia, Western Australia, Australiab PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Western Australia, Australiac School of Veterinary and Life Sciences, Murdoch University, Western Australia, Australiad Communicable Disease Control Directorate, Department of Health, Western Australia, Australiae School of Medical and Health Sciences, Edith Cowan University, Western Australia, Australia


Article history:Received 12 May 2018Received in revised form13 July 2018Accepted 20 July 2018Available online 25 July 2018

Handling Editor: Paola Mastrantonio

Keywords:C. difficileBinary toxinRibotypeAntimicrobial resistanceA�B-CDTþ

a b s t r a c t

Antimicrobial resistance (AMR) is commonly found in Clostridium difficile strains and plays a major role instrain evolution. We have previously reported the isolation of large clostridial toxin-negative, binary toxin-producing (A�B-CDTþ) C. difficile strains from colonised (and in some instances diarrhoeic) food animals, aswell as from patients with diarrhoea. To further characterise these strains, we investigated the phenotypicand genotypic AMR profiles of a diverse collection of A�B-CDTþ C. difficile strains. The in vitro activities of 10antimicrobial agents were determined for 148 A�B-CDTþ C. difficile strains using an agar dilution meth-odology. Whole-genome sequencing and in silico genotyping was performed on 53 isolates to identify AMRgenes. All strains were susceptible to vancomycin, metronidazole and fidaxomicin, antimicrobials currentlyconsidered first-line treatments for C. difficile infection (CDI). Differences in antimicrobial phenotypes be-tween PCR ribotypes (RTs) were observed but were minimal. Phenotypic resistance was observed in 13isolates to tetracycline (TetR, MIC¼ 16mg/L), moxifloxacin (MxfR, MIC¼ 16mg/L), erythromycin (EryR, MIC�128mg/L) and clindamycin (CliR, MIC¼ 8mg/L). The MxfR strain (RT033) possessed mutations in gyrA/B,while the TetR (RT033) strain contained a tetM gene carried on the conjugative transposon Tn6190. All EryRand CliR strains (RT033, QX521) were negative for the erythromycin ribosomal methylase gene ermB,suggesting a possible alternative mechanism of resistance. This work describes the presence of multipleAMR genes in A�B-CDTþ C. difficile strains and provides the first comprehensive analysis of the AMRrepertoire in these lineages isolated from human, animal, food and environmental sources.

© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

Clostridium difficile is a Gram-positive obligately anaerobic ba-cillus that can persist under aerobic conditions in a non-vegetativespore form. C. difficile infection (CDI) is the leading cause ofantimicrobial-associated diarrhoea in most developed countrieswith high rates of healthcare-related infections reported, especially

in European and North American hospitals [1]. The prevalence ofCDI has increased in parallel with the greater use of antimicrobials,most notably clindamycin, cephalosporins and fluoroquinolones[2e4]. Antimicrobial treatment destroys the commensal gut bac-teria that contribute to the inhibition of C. difficile spore outgrowth,creating an imbalance [5]. A continual dysbiosis is exploited byC. difficile, resulting in CDI recurrences that have become a sub-stantial strain on health care systems in regions with high in-cidences of CDI [6].

In the early 2000s, a new generation quinolone (fluo-roquinolone) resistant strain of C. difficile, RT027/NAP1/B1, causedoutbreaks of CDI in Europe and North America leading to theimplementation of antimicrobial stewardship policies in hospitalsettings in these regions [7]. Today, although infections due to

* Corresponding author. Department of Microbiology, PathWest LaboratoryMedicine (WA), Queen Elizabeth II Medical Centre, Nedlands 6009, WesternAustralia, Australia.

E-mail address: [email protected] (T.V. Riley).1 These authors contributed equally to this work.


Anaerobe

journal homepage: www.elsevier .com/locate/anaerobe

https://doi.org/10.1016/j.anaerobe.2018.07.0071075-9964/© 2018 Elsevier Ltd. All rights reserved.

Anaerobe 54 (2018) 55e60

A32

C. difficile RT027 have decreased, fluoroquinolone resistance con-tinues to promote the spread of other C. difficile RTs. In particular,moxifloxacin resistance is commonly found in C. difficile RTs, anindication of the need for frequent review and audit of antimicro-bial stewardship policies [8].

Oral vancomycin and metronidazole are the preferred thera-peutic options for mild or moderate CDI while a combination of oralmetronidazole and intravenous vancomycin is recommended forsevere disease [9]. Both antimicrobials have been linked to diseaserecurrences due to the spore-forming nature of C. difficile, which isresistant to these treatments [9]. Fidaxomicin, a narrow-spectrum,sporicidal macrolide that is highly effective against C. difficile, isoccasionally used as a first-line therapy due to its microbiota pre-serving property that greatly reduces the probability of recurrentCDI [10,11]. Despite it's proven effectiveness, the cost of fidaxomicinis substantially higher than other therapies and, coupled with thesubstantial healthcare costs of CDI, it is not affordable in many partsof the world [12]. In the US, a region highly impacted with CDI, acase of recurrent CDI costs up to $18,000 [12].

Currently, there is global widespread use of antimicrobials inboth hospital and community settings. Approximately 5 out of 6individuals in the US receive a course of antibiotics annually [13]. Asa result, increased antimicrobial resistance (AMR) and reducedsusceptibilities to multiple antimicrobials has become common[1,14]. On the other hand, the use of alternate CDI therapies such asfaecal microbiota transplantation and microbial ecosystem thera-peutics is becoming popular due to excellent recovery rates forrecurrent infections and the lack of dependency on antimicrobials[15]. However, these carry the risk of acquiring AMR genes fromdonors [1,15]. These concerns emphasize the importance of AMRsurveillance in both large clostridial toxin-positive (toxigenic) andlarge clostridial toxin-negative C. difficile strains.

Antimicrobial susceptibility patterns of toxigenic C. difficilestrains have been determined periodically while large clostridialtoxin-negative C. difficile strains have often been ignored. Largeclostridial toxin-negative C. difficile strains lack the main virulencefactors (toxins A and B), however, they may encode a third binarytoxin (CDT), the significance of which is not well understooddespite it being associated with more severe disease [16,17]. CDTshares 80%e85% homology with iota toxin (ɩ-toxin) produced byC. perfringens type E and also possesses an ADP-ribosyltransferaseactivity that modifies actin in the host cells leading to its de-polymerization and inability to form filaments, eventually result-ing in the destruction of the cell cytoskeleton [17]. In vitro experi-ments have confirmed toxicity of CDT and its crucial role inadherence and colonisation [17]. Recently, C. difficile strains pro-ducing only CDT (A�B-CDTþ) have been isolated from diarrhoeicindividuals with recurrent CDI symptoms suggesting the possibilityof CDI in the absence of toxigenic C. difficile strains [18].

Although the role of CDT in infection is unclear, we postulatedthat A�B-CDTþ C. difficile strains may harbour other non-toxinvirulent factors, including antimicrobial resistance markers, thatcontribute to their ability proliferate and cause symptoms ininfected patients. The purpose of this study was to determine theantimicrobial susceptibilities of a collection of A�B-CDTþ C. difficilestrains to a range of antimicrobial agents. In addition, a selection ofthe strains was whole-genome sequenced to corroborate thephenotypic results.

2. Materials and methods

2.1. Bacterial isolates

C. difficile isolates were selected based upon genetic uniquenessusing previous molecular analysis of PCR ribotypes (RTs), toxin

gene profiles and multilocus sequence types (MLST, STs). Thestrains belonged to 10 RTs (033, 238, 239, 288, 585, 586, QX143,QX360, QX444, QX521) and were collected from diverse sources(humans, n¼ 28; foal, n¼ 1; calves, n¼ 52; pigs, n¼ 40; food, n¼ 1;effluent, n¼ 26). The majority (142/148) of isolates were fromAustralia while the remaining six isolates (all RT033) originatedfrom symptomatic patients in France [18]. Table 1 shows thevarious RTs, sequence types (STs) and general characteristics of theisolates analysed.

2.2. Bacterial culture

C. difficile isolates previously frozen at �80 �C using brain heartinfusion broth (supplemented with 15% glycerol) were revived onhorse blood agar (BA) plates incubated anaerobically (A35 Anaer-obic Workstation, Don Whitley Scientific, Shipley, West YorkshireBD17 7SE, United Kingdom) for 48 h to obtain pure cultures.C. difficile colonies were confirmed by their chartreuse fluorescenceunder ultraviolet light.

2.3. Susceptibility testing

The minimum inhibitory concentrations (MICs) of C. difficilestrains were determined using a CLSI-recommended agar dilutionmethod as previously described [19]. A total of 148 A�B-CDTþ

C. difficile isolates were tested against 10 antimicrobials consistingof current CDI therapies (vancomycin, metronidazole, fidaxomicinand rifaximin), antimicrobials associated with high resistance andrisk of CDI development (moxifloxacin, erythromycin, clindamycin)and broad-spectrum antimicrobials frequently that may lead to CDI(meropenem, amoxicillin/clavulanate and tetracycline). The MICswere interpreted using CLSI and EUCASTguidelines where available[20,21]. For fidaxomicin and rifaximin, a European Medical Agencyproposed breakpoint of 1.0mg/L (report WC500119707, http://www.ema.europa.eu/) and recommended breakpoint of �32mg/L[22], respectively, were used.

2.4. DNA sequencing, genome assembly and data analysis

Whole-genome sequencing (WGS) of 53 C. difficile isolatesrepresentative of the 148 A�B-CDTþ isolates was performed usingmethods described by Knight et al. [23]. Bacterial DNA librarieswere generated using standard Nextera XT protocols (Illumina®Inc., San Diego, CA, USA) and paired-end (PE) sequencing wasperformed on the Illumina® Miseq Platform. Quality control andbioinformatic processing of raw reads were performed as describedby Knight et al. [24]. AMR genes and STs were detected in silicousing the ARG-ANNOT and PubMLST databases, respectively,compiled in the short-read typing algorithm SRST2 v0.1.8 [23e25].Draft genomes were assembled and annotated as previouslydescribed [23]. Manual investigation of acquired and intrinsicresistance loci and their underlying genomic context was per-formed using a custom sequence library comprising mobile geneticelements previously identified in C. difficile and other related Fir-micutes, as previously described [23].

3. Results

All isolates were susceptible to fidaxomicin, rifaximin, vanco-mycin, metronidazole, amoxicillin/clavulanate and meropenem(Table 2). Phenotypic resistance was observed in 9.3% (3/28) ofhuman isolates, 38.5% (10/26) of effluent isolates and 0% of cattle(0/53) and pig (0/40) isolates. In total, 13/148 C. difficile isolatesfrom humans (n¼ 3) and effluent (n¼ 10) exhibited phenotypicresistance to tetracycline (TetR, MIC¼ 16mg/L), moxifloxacin

G.O. Androga et al. / Anaerobe 54 (2018) 55e6056

A33

(MxfR, MIC¼ 16mg/L), erythromycin (EryR, MIC �128mg/L) andclindamycin (CliR, MIC¼ 8mg/L). All cattle and pig C. difficile iso-lates were susceptible to all antimicrobial agents tested. In total, 10different RTs were analysed and the resistant isolates belonged totwo RTs only, RTs 033 (n¼ 11, 84.6%) and QX521 (n¼ 2, 15.4%).

Non-synonymous mutations in the DNA gyrase GyrA/B weredetected in the MxfR isolate (RT033) with distinct allele types(GyrA [Lys413Asn], GyrB [Gln160His, Ser366Val, Ser416Ala,Asp426Asn]). The Asp426Asn and Ser416Ala mutations in GyrBcorrelatedwith fluoroquinolone resistance and the othermutationswere non-synonymous mutations that fell outside the quinoloneresistance-determining regions (QRDR) of GyrA and B. The TetRstrain, also belonging to RT033, contained a tetM gene (encoding aribosomal protective protein) carried on a conjugative transposonTn6190, originally discovered in the M120 strain of RT078 (acces-sion NC_017174) isolated from an Irish diabetic patient (Table 3). NoEryR or CliR strain contained methylase erm genes, suggesting apossible alternative mechanism of resistance in these strains.

Eight RT033 isolates also possessed aminoglycoside resistancegenes (aph3-III and sat4A) and harboured a 7269bp fragment of a

multidrug resistance gene cassette from the ruminant facultativeanaerobe Erysipelothrix rhusiopathiae (99% nucleotide seq ID toKP339868.1). Interestingly, this cassette also had a third (syntenic)aminoglycoside gene (ant6-Ia), which was not picked up by SRST2analysis but identified on manual curation of the assembledgenome. Further manual curation of the A�B-CDTþ C. difficile ge-nomes detected genes encoding a b-lactamase inducing penicillin-binding protein (blaR) and a multidrug resistance transporterprotein (cme), loci that have been reported previously in otherC. difficile lineages (Table 3).

4. Discussion

This work illustrates antimicrobial phenotypic resistance andthe presence of multiple AMR genes in A�B-CDTþ C. difficile RTsisolated from human, animal and environmental (effluent) sources.Our collection of C. difficile RT033 strains exhibited resistance tomore antimicrobials of different classes than any other A�B-CDTþ

C. difficile RT tested. This is noteworthy because this RT, despitebeing thought of as not clinically relevant, has been associated with

Table 1Distribution of PCR ribotypes and sequence types (STs) of the various A�B-CDTþ C. difficile isolates (n¼ 148) included in the study.

RIBOTYPE C. difficile MLST genes RIBOTYPE PATTERNS

adk atpA dxr glyA recA sodA tpi ST CLADE SOURCE COUNTRY SYMPTOMATIC/ASYMPTOMATIC

RT 238 5 8 5 26 15 29 8 169 5 Pigs Australia, n¼ 23 NIa

QX 143 5 8 5 28 15 28 59 386 5 Human Australia, n¼ 1 Symptomatic

RT 585 5 15 5 27 15 29 20 164 5 Human Australia, n¼ 4 Symptomatic

Foal Australia, n¼ 1 Symptomatic



Human France, n¼ 6 Symptomatic

Pigs Australia, n¼ 17 NI

Food Australia, n¼ 1 NAb

Effluent Australia, n¼ 10 NA

Calves Australia, n¼ 24 Asymptomatic


RT 288 5 8 5 11 9 11 8 11 5 Calves Australia, n¼ 28 NI

Human Australia, n¼ 1 Symptomatic


QX 521 5 8 5 27 15 28 8 280 5 Effluent Australia, n¼ 16 NA


a ; No information available.b ; Not applicable.

G.O. Androga et al. / Anaerobe 54 (2018) 55e60 57

A34

human infections in Australia, Europe and North America[18,26e28]. We hypothesize that the presence of multiple AMRgenes in this RT may be a factor driving the increased incidence ofRT033 human and animal infections.

C. difficile RT033, also classified as toxinotype XI, is common infood animals, especially piglets and veal calves [29]. It belongs toST11 andMLST clade 5, a clade known to cause significant mortalitythat contains the so-called “hypervirulent” RT078 strain [30].Symptomatic human cases of RT033 infection described in theliterature include single cases from Australia, Italy and NorthAmerica, and four cases from France [18,26e28]. We recently re-ported the discovery of a vanB2-like vancomycin resistance operonfrom an RT033 C. difficile strain isolated from an Australian veal calfat slaughter [31]. Although phenotypically inactive, possibly due tofragmentation in the vanRB gene, the origin of this element invancomycin-resistant Enterococcus species illustrates the possibil-ity that a fully vancomycin-resistant strain of C. difficile mayemerge. None of our RT033 C. difficile isolates contained a vanB2operon and they were all susceptible to vancomycin(MIC¼ 1e2mg/L, Table 2). However, they showed similar pheno-typic resistance characteristics to clinically relevant toxigenicC. difficile strains.

Since the initial association between CDI and antimicrobialtherapy was confirmed, many toxigenic C. difficile strains have beenreported as resistant to clindamycin and erythromycin, oftenrelated to the rRNA adenine N-6-methyltransferase encoded by theermB gene [32,33]. Approximately 17 mobile elements have beenlinked to macrolide-lincosamide-streptogramin B (MLSB) resis-tance in C. difficile but Tn5398 is the most commonly identifiedermB-containing element found in CliR and EryR C. difficile strains[1]. Notably, this non-conjugative element contains two copies ofermB genes [1]. Some of our A�B-CDTþ C. difficile RTs (033, n¼ 8 andQX521, n¼ 2) displayed anMLSB phenotype yet did not harbour anyof the known methylase subclasses (ermB, ermC or ermTR). Thisdiscordance has been observed in C. difficile previously, and publi-cations have suggested that mutations in L4/L22 riboproteins and23s rRNA could explain the MLSB resistance [1]. Analysis of thesequenced genomes showed that both the L4/L22 riboprotein genesand 23s rRNA genes in this population were full-length and wild-type with no variations identified that were found exclusively inMLSBþ strains. However, analysis of the multiple 23s rRNA allelespresent in a typical C. difficile genome was not possible with theIllumina short-read sequencing approach used in this study.

Fluoroquinolone resistance (FQR) in C. difficile has beencontinually documented since the outbreaks caused by two inde-pendently evolved FQR lineages of C. difficile RT027/BI/NAP1 inCanada, USA and Europe between 2002 and 2006 [1,23]. Althoughthe incidence of C. difficile RT027 infections has markedly reducedin some countries, FQR in other C. difficile RTs continues to emerge,most notably in ST11 and RT017 lineages [23]. Mutations within thedefined QRDRs of DNA gyrase subunits GyrA and/or GyrB generallyconfer resistance to FQs, however, non-QRDR polymorphismsresulting in FQR have been observed [33]. We identified both QRDRand non-QRDR mutations in gyrA/B. These mutations were identi-fied in an isolate (RT033) that was phenotypically resistant tomoxifloxacin (MIC¼ 16mg/L). The isolate originated from a patientin France who was considered to have CDI and had only A�B-CDTþ

C. difficile RT033 isolated from stool specimens. The patient fullyrecovered after treatment with oral metronidazole, however, thiscase exemplifies acquisition and possible proliferation of the FQRgenotypes within A�B-CDTþ C. difficile strains [18].

With regard to tetracycline, resistance in C. difficile is thought tobe less common and varies between countries and RTs [34].C. difficile tetracycline resistance genes are commonly carried onTn916 and Tn5397-like mobile elements, however, mobile elementsTa

ble

2Su

scep

tibilityof

A� B

- CDTþ

C.difficile

strainsto

10an

timicrobial

agen

ts.

Antimicrobial

Age

nt(Clin

ical

Break

points)(S/I/R)

Sources

Total(n

¼14

8)%S/I/R

Human

(n¼28

)Fo

od(n

¼1)

Cattle(n

¼54

)Pigs

(n¼39

)Effluen

t(n

¼26

)

MIC

Ran

geMIC

50/90(m

g/L)

MIC

Ran

geMIC

50/90(m

g/L)

MIC

Ran

geMIC

50/90(m

g/L)

MIC

Ran

geMIC

50/90(m

g/L)

MIC

Ran

geMIC

50/90(m

g/L)

MIC

Ran

geMIC

50/90(m

g/L)

VANa

1e2

1/1

11/1

11/1

1e2

1/1

11/1

1e2

1/1

100/-/0

MTZ

a0.25

e0.5

0.5/0.5

0.5

0.5/0.5

0.12

e0.5

0.5/0.5

0.25

e1

0.25

/0.5

0.12

e0.5

0.25

/0.5

0.12

e1

0.5/1

100/-/0

FDXb

0.00

4e0.12

0.03

/0.06

0.01

50.01

5/0.01

50.00

4e0.06

0.01

5/0.06

0.00

4e0.06

0.03

/0.03

0.00

8e0.12

0.03

/0.06

0.00

4e0.12

0.03

/0.12

100/-/0

MXFc

1e16

1/1

11/1

1e2

1/1

1e2

1/2

1e4

1/2

1e2

1/1

98.6/0.7/0.7

CLI

c0.12

e8

0.5/4

0.5

0.5/0.5

0.12

e4

1/4

0.03

e4

0.5/1

0.03

e8

1/4

0.03

e8

0.5/4

92.6/6.7/0.7

ERYc

0.12

e12

81/1

0.5

0.5/0.5

0.25

e4

0.5/0.5

0.5e

41/1

0.25

e12

88/12

80.12

e12

81/4

�/�/

6.7

MER

c1e

42/2

22/2

22/2

2e4

2/4

2e4

2/2

2e4

2/2

100/0/0

AUGc

0.25

e1

0.5/1

0.5

0.5/0.5

0.25

e0.5

0.5/0.5

0.25

e2

0.5/0.5

0.25

e0.5

0.5/0.5

0.25

e2

0.5/1

100/0/0

RFX

d0.00

4e2

0.00

8/0.00

80.00

80.00

8/0.00

80.00

40.00

4/0.00

40.00

4e1

0.00

8/0.00

80.00

4e0.01

50.00

8/0.00

80.00

4e2

0.00

4/0.01

510

0/-/0

TETc

0.03

e0.25

0.12

/0.25

0.12

0.12

/0.12

0.06

e0.12

0.12

/0.12

0.06

e8

0.12

/20.06

e0.12

0.06

/0.12

0.06

e8

0.12

/298

.0/1.3/0.7

S-su

scep

tible,

I-interm

ediate,R

-resistant.Break

points

(minim

um

inhibitoryco

ncentration[m

g/L];S,

I,R)forea

chan

tibiotic

wereas

follo

ws:

VAN-Van

comyc

in(�

2/-/˃2),MET

-Metranidazole(�

2/-/˃2),FD

X-Fidax

omicin

(�/�

/�1),M

XF-

Mox

iflox

acin

(�2/4/�8

),CLI-Clin

dam

ycin

(�2/4/�8

),ER

Y-Erythromyc

in(�

/�/˃8),M

ER-M

erop

enem

(�4/8/�1

6),A

UG-Amox

icillin/clavu

lanate(�

4/8/�1

6),R

FX-Rifax

imin

(�/�

/�32

),TE

T-Te

tracyline(�

4/8/

�16).

aEU

CAST

brea

kpoints

[21].

bResistance

(�1.0mg/L)

asdescribed

byEu

ropea

nMed

ical

Age

ncy

(rep

ortW

C50

0119

707,

http://w

ww.ema.eu

ropa.eu

/).

cBreak

points

asreco

mmen

ded

byCLS

I[20].

dResistance

(�32

mg/L)

asdescribed

byO'Con

nor

etal.[22

].


A35

that carry TetR genes from other bacterial species have beenidentified in C. difficile e.g. tetA/B [23]. The TetR strain in ourA�B-CDTþ C. difficile collection, also an RT033 strain, contained atetM gene carried on a conjugative transposon Tn6190, originallydiscovered in C. difficile RT078 strain M120 and, to date, only re-ported in C. difficile ST11 lineages RT126 and RT078 [35]. Tn6190 is97% homologous to Tn916 and considered to circulate in pigs [36].Our TetR isolate originated from a patient with idiopathic diarrhoeasuggesting possible zoonotic transmission, although a higher-resolution typing approach such as core genome SNP analysiswould be needed to confirm this [35].

In Australia and The Netherlands, bi-directional transmission(zoonotic and anthroponotic) of C. difficile has been demonstratedthat may be facilitating dissemination of AMR genes [23,37].However, in this study, we observed the possible multi-directionaltransmission of AMR genes from human, animal and effluentsources. Ten of 13 resistant isolates (76.9%) came from an envi-ronmental source (effluent from a piggery) and indicated pheno-typic resistance to erythromycin (�128mg/L). These isolatesbelonged to RTs 033 and QX521 (a novel RT). We did not isolateQX521 from any other source, however, C. difficile RT033 wasdetected from all the sources (human, animal, food and effluent)and at least one RT033 isolate from each source (besides food)contained AMR genes (Table 3). Additionally, the RT033 isolatesfrom human and effluent sources exhibited multi-drug resistant(MDR) phenotypes (resistance to two or more antimicrobials) tomoxifloxacin, clindamycin, erythromycin and tetracycline. Theseresults emphasize the importance of a ‘One Health’ approach tocombating AMR in C. difficile [38].

While considerable effort is being made in directing antimi-crobial stewardship, there is increasing concern about the devel-opment of resistance to clinically consequential antimicrobials. Inthis study, we successfully demonstrated that A�B-CDTþ C. difficilestrains from diverse sources are reservoirs of AMR genes that havealso been identified in clinically relevant toxigenic C. difficile strains.

5. Conclusion

AMR is a One Health issue that highlights the importance of theassociation between human health, animal health and the envi-ronment. While the role of A�B-CDTþ C. difficile strains in idiopathic

diarrhoea is still unclear, these strains remain common in foodanimals and could potentially transmit AMR genes. In the future,we will further investigate the evolution and transmission of thesestrains using high-resolution core genome phylogenetics. However,the present study provides a basis for this with a comprehensiveanalysis of AMR profiles of various A�B-CDTþ C. difficile strainsisolated from humans, animals, food and environmental sources.

Funding

This research did not receive any specific grant from fundingagencies in the public, commercial or not-for-profit sectors. GA isfunded by an Australian Postgraduate Award from The University ofWestern Australia. DK is funded by the National Health andMedicalResearch Council of Australia.

Transparency declarations

Professor Riley reports grants and non-financial support fromCepheid, MSD and Otsuka Pharmaceuticals outside the submittedwork. All other authors declare no conflict of interest.

Acknowledgements

We thank Dr. Fr�ed�eric Barbut (National Reference Laboratory forC. difficile, Hopital Saint-Antoine, Paris, France) for providing addi-tional A-B-CDT þ C. difficile strains for this study and the PathWestLaboratory Medicine Media Section for providing the bacteriolog-ical media for the study.

References

[1] P. Spigaglia, Recent advances in the understanding of antibiotic resistance inClostridium difficile infection, Ther Adv Infec Dis 3 (2016) 23e42.

[2] C. Slimings, T.V. Riley, Antibiotics and hospital-acquired Clostridium difficileinfection: update of systematic review and meta-analysis, J. Antimicrob.Chemother. 69 (2014) 881e891.

[3] D.N. Gerding, Clindamycin, cephalosporins, fluoroquinolones, and Clostridiumdifficile-associated diarrhea: this is an antimicrobial resistance problem, Clin.Infect. Dis. 38 (2004) 646e648.

[4] C. Slimings, P. Armstrong, W.D. Beckingham, A.L. Bull, L. Hall, K.J. Kennedy,J. Marquess, R. McCann, A. Menzies, B.G. Mitchell, M.J. Richards, P.C. Smollen,L. Tracey, I.J. Wilkinson, F.L. Wilson, L.J. Worth, T.V. Riley, Increasing incidenceof Clostridium difficile infection, Australia, 2011-2012, Med. J. Aust. 200 (2014)

Table 3AMR genes detected from raw sequence reads of A�B-CDTþ C. difficile strains (n¼ 53).

Phenotype Gene(s) Ribotype Toxinprofile

Source

Aminoglycosideresistancea

aph3-III-sat4A-ant6-Ia RT 033 A�B-CDTþ Human, n¼ 1, pigs, n¼ 3 and effluent, n¼ 4aph3-III-sat4A-npmA- ant6-Ia RT 033 A�B-CDTþ Pig, n¼ 1

b-lactam resistanceb blaRcme

RT 033 A�B-CDTþ Human, n¼ 16, calves, n¼ 6, pigs, n¼ 3, effluent, n¼ 4 and food,n¼ 1

RT 238 A�B-CDTþ Pigs, n¼ 2 and calf, n¼ 1RT 239 A�B-CDTþ Human, n¼ 2RT 288 A�B-CDTþ Human, n¼ 1 and calf, n¼ 3RT 585 A�B-CDTþ Human, n¼ 4 and foal, n¼ 1RT 586 A�B-CDTþ Human, n¼ 1QX 143 A�B-CDTþ Human, n¼ 1QX 444 A�B-CDTþ Human, n¼ 1QX 521 A�B-CDTþ Effluent, n¼ 5QX 629 A�B-CDTþ Human, n¼ 1

Fluoroquinoloneresistance

gyrA (Lys413Asn)gyrB (Gln160His, Ser366Val, Ser416Ala,Asp426Asn)

RT 033 A�B-CDTþ Human, n¼ 1

Glycopeptide resistance van B2 operon RT 033 A�B-CDTþ Calf, n¼ 1Tetracycline resistance TetM RT 033 A�B-CDTþ Human, n¼ 1

a All genomes positive for aminoglycoside resistance genes aph3-III and sat4A harboured a 7269bp fragment of a resistance gene cassette from the ruminant facultativeanaerobe Erysipelothrix rhusiopathiae (99% seq ID to KP339868.1).

b Results obtained by manual curation of all A�B-CDTþ C. difficile genomes.

G.O. Androga et al. / Anaerobe 54 (2018) 55e60 59

A36

272e276.[5] B.B. Lewis, E.G. Pamer, Microbiota-based therapies for Clostridium difficile and

antibiotic-resistant enteric infections, Annu. Rev. Microbiol. 71 (2017)157e178.

[6] A. Deshpande, V. Pasupuleti, P. Thota, C. Pant, D.D. Rolston, A.V. Hernandez,C.J. Donskey, T.G. Fraser, Risk factors for recurrent Clostridium difficile infec-tion: a systematic review and meta-analysis, Infect Con Hosp Epidemiol 36(2015) 452e460.

[7] M. He, F. Miyajima, P. Roberts, L. Ellison, D.J. Pickard, M.J. Martin, T.R. Connor,S.R. Harris, D. Fairley, K.B. Bamford, S. D'Arc, J. Brazier, D. Brown, J.E. Coia,G. Douce, D. Gerding, H.J. Kim, T.H. Koh, H. Kato, M. Senoh, T. Louie, S. Michell,E. Butt, S.J. Peacock, N.M. Brown, T. Riley, G. Songer, M. Wilcox,M. Pirmohamed, E. Kuijper, P. Hawkey, B.W. Wren, G. Dougan, J. Parkhill,T.D. Lawley, Emergence and global spread of epidemic healthcare-associatedClostridium difficile, Nat. Genet. 45 (2013) 109e113.

[8] H. Harvala, E. Alm, T. Akerlund, K. Rizzardi, Emergence and spread ofmoxifloxacin-resistant Clostridium difficile ribotype 231 in Sweden between2006 and 2015, New Microbes New Infect 14 (2016) 58e66.

[9] J.G. Bartlett, Clostridium difficile infection, Infect. Dis. Clin. 31 (2017) 489e495.[10] H.G. Maxwell-Scott, S.D. Goldenberg, Existing and investigational therapies

for the treatment of Clostridium difficile infection: a focus on narrow spectrum,microbiota-sparing agents, Med. Maladies Infect. 48 (2018) 1e9.

[11] N. Eliakim-Raz, J. Bishara, Prevention and treatment of Clostridium difficileassociated diarrhea by reconstitution of the microbiota, Hum. VaccinesImmunother. 1e4 (2018).

[12] H.E. Burton, S.A. Mitchell, M. Watt, A systematic literature review of economicevaluations of antibiotic treatments for Clostridium difficile infection, Phar-macoeconomics 35 (2017) 1123e1140.

[13] A. Srinivasan, Antibiotic stewardship: why we must, how we can, Cleve. Clin. J.Med. 84 (2017) 673e679.

[14] D.A. Leffler, J.T. Lamont, Clostridium difficile infection, N. Engl. J. Med. 372(2015) 1539e1548.

[15] V. Leung, C. Vincent, T. Edens, M. Miller, A.R. Manges, Antimicrobial resistancegene acquisition and depletion following fecal microbiota transplantation(FMT) for recurrent Clostridium difficile infection, Clin. Infect. Dis. 66 (2018)456e457.

[16] G.O. Androga, A.M. McGovern, B. Elliott, B.J. Chang, T.T. Perkins, N.F. Foster,T.V. Riley, Evaluation of the Cepheid Xpert C. difficile/Epi and MeridianBioscience illumigene C. difficile assays for detecting Clostridium difficile ribo-type 033 strains, J. Clin. Microbiol. 53 (2015) 973e975.

[17] B.G. Siles, K. Pradhan, J.M. Fleming, R.P. Samy, H. Barth, M.R. Popoff, Clos-tridium and Bacillus binary enterotoxins: bad for the bowels and eukaryoticbeing, Toxins 6 (2014) 2626e2656.

[18] C. Eckert, A. Emirian, A. Le Monnier, L. Cathala, H. De Montclos, J. Goret,P. Berger, A. Petit, A. De Chevigny, H. Jean-Pierre, B. Nebbad, S. Camiade,R. Meckenstock, V. Lalande, H. Marchandin, F. Barbut, Prevalence and path-ogenicity of binary toxin-positive Clostridium difficile strains that do notproduce toxins A and B, New Microbes New Infect 3 (2014) 12e17.

[19] D.R. Knight, S. Giglio, P.G. Huntington, T.M. Korman, D. Kotsanas, C.V. Moore,D.L. Paterson, L. Prendergast, C.A. Huber, J. Robson, L. Waring, M.C. Wehrhahn,G.F. Weldhagen, R.M. Wilson, T.V. Riley, Surveillance for antimicrobial resis-tance in Australian isolates of Clostridium difficile, 2013-14, J. Antimicrob.Chemother. 70 (2015) 2992e2999.

[20] Clinical and Laboratory Standards Institute, Performance Standards for Anti-microbial Susceptibility Testing: Twenty-third Informational SupplementM100-s23, CLSI, Wayne, PA, USA, 2013.

[21] EUCAST, Breakpoint Tables for Interpretation of MICs and Zone Diameters,Version 5.0. http://www.eucast.org/.

[22] J.R. O'Connor, M.A. Galang, S.P. Sambol, D.W. Hecht, G. Vedantam,D.N. Gerding, S. Johnson, Rifampin and rifaximin resistance in clinical isolatesof Clostridium difficile, Antimicrob. Agents Chemother. 52 (2008) 2813e2817.

[23] D.R. Knight, M.M. Squire, D.A. Collins, T.V. Riley, Genome analysis of Clos-tridium difficile PCR ribotype 014 lineage in Australian pigs and humans re-veals a diverse genetic repertoire and signatures of long-range interspeciestransmission, Front. Microbiol. 7 (2017) 2138.

[24] S.K. Gupta, B.R. Padmanabhan, S.M. Diene, R. Lopez-Rojas, M. Kempf,L. Landraud, J.M. Rolain, ARG-ANNOT, a new bioinformatic tool to discoverantibiotic resistance genes in bacterial genomes, Antimicrob. Agents Che-mother. 58 (2014) 212e220.

[25] M. Inouye, H. Dashnow, L.A. Raven, M.B. Schultz, B.J. Pope, T. Tomita, J. Zobel,K.E. Holt, SRST2: rapid genomic surveillance for public health and hospitalmicrobiology labs, Genome Med. 6 (2014) 90.

[26] G.O. Androga, J. Hart, N.F. Foster, A. Charles, D. Forbes, T.V. Riley, Infectionwith toxin A-negative, toxin B-negative, binary toxin-positive Clostridiumdifficile in a young patient with ulcerative colitis, J. Clin. Microbiol. 53 (2015)3702e3704.

[27] S. Grandesso, F. Arena, F.E. Eseme, S. Panese, L. Henrici De Angelis, P. Spigaglia,F. Barbanti, G.M. Rossolini, Clostridium difficile ribotype 033 colitis in a patientfollowing broad-spectrum antibiotic treatment for KPC-producing Klebsiellapneumoniae infection, Italy, N. Microbiol. 39 (2016) 235e236.

[28] B. Geric, S. Johnson, D.N. Gerding, M. Grabnar, M. Rupnik, Frequency of binarytoxin genes among Clostridium difficile strains that do not produce largeclostridial toxins, J. Clin. Microbiol. 41 (2003) 5227e5232.

[29] D.R. Knight, S. Thean, P. Putsathit, S. Fenwick, T.V. Riley, Cross-sectional studyreveals high prevalence of Clostridium difficile non-PCR ribotype 078 strains inAustralian veal calves at slaughter, Appl. Environ. Microbiol. 79 (2013)2630e2635.

[30] A.S. Walker, D.W. Eyre, D.H. Wyllie, K.E. Dingle, D. Griffiths, B. Shine, S. Oakely,L. O'Connor, J. Finney, A. Vaughan, D.W. Crook, M.H. Wilcox, T.E. Peto, Rela-tionship between bacterial strain type, host biomarkers, and mortality inClostridium difficile infection, Clin. Infect. Dis. 56 (2013) 1589e1600.

[31] D.R. Knight, G.O. Androga, S.A. Ballard, S.A. Howden, T.V. Riley,A phenotypically silent vanB2 operon carried on a Tn1549-like element inClostridium difficile, mSphere 1 (2016) e00177-16.

[32] J.G. Bartlett, T.W. Chang, M. Gurwith, S.L. Gorbach, A.B. Onderdonk, Antibiotic-associated pseudomembranous colitis due to toxin-producing clostridia,N. Engl. J. Med. 298 (1978) 531e534.

[33] G. Ramírez-Vargas, C. Quesada-G�omez, L. Acu~na-Amador, D. Lopez-Ure~na,T. Murillo, M. Del Mar Gamboa-Coronado, E. Chaves-Olarte, N. Thomson,E. Rodríguez-Cavallini, C. Rodríguez, A Clostridium difficile lineage endemic toCosta Rican hospitals is multidrug resistant by acquisition of chromosomalmutations and novel mobile genetic elements, Antimicrob. Agents Chemo-ther. 61 (2017) e02054-16.

[34] D.R. Knight, T.V. Riley, Clostridium difficile clade 5 in Australia: antimicrobialsusceptibility profiling of PCR ribotypes of human and animal origin,J. Antimicrob. Chemother. 71 (2016) 2213e2217.

[35] D.R. Knight, B. Elliott, B.J. Chang, T.T. Perkins, T.V. Riley, Diversity and evolu-tion in the genome of Clostridium difficile, Clin. Microbiol. Rev. 28 (2015)721e741.

[36] J. Corver, D. Bakker, M.S. Brouwer, C. Harmanus, M.P. Hensgens, A.P. Roberts,L.J. Lipman, E.J. Kuijper, H.C. van Leeuwen, Analysis of a Clostridium difficilePCR ribotype 078 100 kilobase island reveals the presence of a novel trans-poson, Tn6164, BMC Microbiol. 12 (2012) 130.

[37] C.W. Knetsch, N. Kumar, S.C. Forster, T.R. Connor, H.P. Browne, C. Harmanus,I.M. Sanders, S.R. Harris, L. Tumer, T. Morris, M. Perry, F. Miyajima, P. Roberts,M. Pirmohamed, J.G. Songer, J.S. Weese, A. Indra, J. Corver, M. Rupnik,B.W. Wren, T.V. Riley, E.J. Kuijper, T.D. Lawley, Zoonotic transfer of Clostridiumdifficile harboring antimicrobial resistance between farm animals andhumans, J. Clin. Microbiol. 56 (2018) e01384-17.

[38] M.M. Squire, T.V. Riley, Clostridium difficile infection in humans and piglets: a'One Health' opportunity, Curr. Top. Microbiol. Immunol. 365 (2013)299e314.


A37

Documents

Sources and potential transmission of Clostridium