Upload
aditya-k
View
215
Download
0
Embed Size (px)
Citation preview
Onychomycosis: a critical study of techniques and criteria
for confirming the etiologic significance of
nondermatophytes
RICHARD C. SUMMERBELL*, ELIZABETH COOPER$, URSULA BUNN%, FRANCES JAMIESON% &
ADITYA K. GUPTA$§
*Centraalbureau voor Schimmelcultures, Utrecht, the Netherlands, $Mediprobe Laboratories, Toronto, %LaboratoriesBranch, Ontario Ministry of Health, Toronto, §Division of Dermatology, Department of Medicine, Sunnybrook and Women’sCollege Health Sciences Center and the University of Toronto, Toronto, Ontario, Canada
Nondermatophytic filamentous fungi (NDF) other than Scytalidium species are
variously said to cause between 0 and 50% of all toenail onychomycoses, though
most estimates are in the 2�/5% range. Due to the difficulty of obtaining ‘gold
standard’ control data for comparison, the accuracy of many laboratory evaluation
processes used to deal with potential NDF onychomycoses has never been
rigorously measured, thus allowing use of differing interpretations of the
significance of cultures. To allow valid comparison of these procedures and
interpretations, a large series of patients who declined treatment were sampled on
multiple occasions from all apparently onychomycotic toenails until adequate
certainty had been attained that all etiologic agents had been isolated and, where
necessary, confirmed as etiologic via consistent repeated isolation. This informa-
tion was used to evaluate results that had been obtained in the initial direct
microscopy and culture studies for the same patient population, as such results are
strongly relied on in routine diagnosis. Direct microscopy (KOH) was found to be
73.8% sensitive for dermatophytes and 67.2% sensitive for proven etiologic NDF
(difference not significant). Culture of the initial specimen coincidentally had a
sensitivity of 74.6% for both fungal groups. KOH and culture in tandem were
83.9% sensitive for indicating fungal etiology based on the first specimen. Different
evaluative frameworks currently used to interpret NDF isolations were contrasted.
The ‘classic’ evaluation procedure, in which all NDF considered etiologic must be
isolated from at least two successive nail specimens, at least one of which must be
microscopic positive for compatible fungal filaments, had a sensitivity of 59.5% but
a specificity of 100% for true NDF infections and mixed NDF/dermatophyte
infections. The most widely used ‘simple association’ evaluation criterion, in which
NDF outgrowth is considered etiologic whenever direct microscopy is positive for
fungal elements and no dermatophyte grows out from the initial specimen, had a
sensitivity of 53.6% and a specificity of 70.3% for NDF infections. A frequently
criticized, but in some quarters still advocated, variant of the simple association
criterion considers NDF outgrowth to be significant whenever the corresponding
specimen is positive for fungal filaments in direct microscopy; application of this
Correspondence: Richard C. Summerbell, Centraalbureau voor
Schimmelcultures, Uppsalalaan 8, 3584 CT, Utrecht, the
Netherlands. Tel: �/31 3021 22688; Fax: �/31 3025 12097; E-mail:
Received 16 September 2003; Accepted 8 January 2004
– 2005 ISHAM DOI: 10.1080/13693780410001712043
Medical Mycology February 2005, 43, 39�/59
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
criterion yielded a sensitivity of 60.7% for true infections but a specificity of only
42%. With the aid of two standard notes soliciting repeat specimens, the classic
criterion was able to attain 92.7% accuracy in recognizing all NDF etiologic agents
and 100% accuracy in disregarding all contaminants from initial specimens that
were positive in direct microscopy and yielded a filamentous fungus in initial
culture. Even in exhaustive longitudinal study, only 20.2% of NDF infections were
found to be associated with a concurrent dermatophytosis. In auxiliary studies,
some nails remained NDF-infected after dermatophytes had been successfully
eliminated by therapy.
Introduction
Both dermatophyte fungi and some members of the
Scytalidium dimidiatum (synanamorph name Nattras-
sia mangiferae ) clade are specialized for pathogenesis of
human nails and skin. Other nondermatophytic fila-
mentous fungi (NDF) rarely cause skin infection, but
approximately 35 species have been rigorously impli-
cated in causation of onychomycosis [1]. The precise
role of these NDF in onychomycosis is imperfectly
understood, and there is considerable controversy
about how best to interpret clinical findings in many
cases in which an NDF grows from an abnormal nail.
This uncertainty has led to disparity in different
researchers’ interpretative frameworks, with the result
that statistics published in various studies on the
relative prevalence of nondermatophytes in onychomy-
cosis differ to an extreme degree. Selected examples are
shown in Table 1; others are reviewed by Greer [2]. The
same degree of discrepancy also applies to infections
attributed to yeasts, but this topic is beyond the scope
of this article.One major factor causing difficulty in NDF status
evaluation is that the reliability of dermatophyte
growth from true-positive samples is far from ideal:
various studies have found that dermatophytes causing
onychomycosis fail to grow from about 15�/50% of
true-positive samples, even though their filaments can
often be seen in the same samples in potassium
hydroxide (KOH)-based direct microscopy [3�/6]. Gen-
tles [7], in a study in which each patient yielding a
filament-positive sample was followed up with repeated
sampling until the corresponding dermatophyte was
grown, showed that 73% grew a culture on the first
sampling occasion, while another 15.7% first grew a
culture on the second sampling occasion. Disturbingly,
5.6% failed to grow a culture until the third sampling,
while another 5.6% grew a culture only from the fourth
or a later sample. The underlying reason, as seen when
vital stains are used on such samples [8], is that
dermatophyte growth in nails is modular and growth
fronts in nail tissue eventually die out, leaving dead
filaments behind in extensive tracts of nutritionally
depleted nail tissue. Needless to say, dermatophyte
cultures cannot be grown from these dead fronts. From
each such sample cultured, however, there is a risk that
a contaminating mold may be grown and misinter-
preted as the agent of the infection. This hazard
connects with another factor making NDF evaluation
vexatious, which is that many of the fungi most
commonly causing NDF onychomycoses are also
among the fungi most commonly obtained in culture
as innocuous contaminants of feet and nails [1]. Not
only are they common indoor fungi in general but also,
the same affinity for nail chemical constituents that
allows them to cause onychomycosis may facilitate
their growth on chemically related, environmentally
proximal materials such as dead skin flakes in shoes
and on floors (when these floors are damp or humid),
as well as moist, decaying shoe leather.
Table 1 Contribution of different groups of organisms to the etiology of onychomycosis as attributed in various studies (in % of total etiologic
agents isolated). Study [reference no.] and geographic area
Attributed etiology [43] UK [14] USA [10] Canada [11] UK [13] Italy
Dermatophytes only 80.5 15.6 90.5 55.7 23.2
Yeasts only 17.0 8.9 5.5 33.2 59.1
Moulds only 2.5 48 4 11.0 17.6
Mixed: dermatophyte�/other NS 19.1 NS NS NS
NS, not stated.
– 2005 ISHAM, Medical Mycology, 43, 39�/59
40 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
In all laboratories assessing onychomycosis, an NDF
is generally considered significant in the rare cases inwhich direct nail microscopy shows very distinctive
nondermatophytic structures (e.g. Scopulariopsis con-
idia or Aspergillus conidiophores) occurring in addi-
tion to fungal filaments in or connectedly emergent
from the affected nail tissue.
Laboratories may differ profoundly, however, in how
they interpret NDF growth from a nail infection in
which fungal filaments of a nondescript nature are seenin direct microscopy, as occurs in the majority of
dermatophytic and NDF infections. A taxonomic
approach to onychomycosis studies suggests that in
cases in which an NDF grows from a sample, authors
generally use one of three major interpretative criteria
to establish that the isolated NDF is a sole causal agent
and not a contaminant. First, the classic , rigorous
procedure for NDF onychomycosis confirmation de-mands at least one demonstration of compatible fungal
filaments in direct microscopy of the relevant nail
tissue, as well as at least two independent isolations
of the putative NDF causal agent from specimens
taken at different times at least a week apart [9,10]. In
the case of a possible mixed infection involving both a
dermatophyte and a nondermatophyte, an additional
(third) isolation of the nondermatophyte may berecommended as an ideal procedure. Second, in an
attempt to circumvent the troublesome need for a
second specimen in evaluating NDF outgrowth, Walshe
and English [11] proposed an inoculum-counting pro-
cedure (later slightly modified by English [12]) in which
an NDF was declared causal when (i) compatible
fungal filaments were present in direct nail microscopy;
(ii) no dermatophyte grew in culture; and (iii) the samenondermatophytic fungus grew from at least five of 20
nail fragments planted from scrapings or drillings of
the lesion. The possibility of recognizing mixed infec-
tions was generally excluded when this criterion was
used. Finally, peer-reviewed publications regularly
appear in which NDF etiology is evaluated with
minimal complexity by means of a simple association
criterion, that is by a criterion in which an NDF istreated as causal if it grows from a single nail specimen
in which fungal filaments are seen in direct microscopy,
but from which no dermatophyte grows (see, for
example, ref. 13). A variant version of this criterion
accepts all NDF that grow from filament-positive nail
specimens as causal, regardless of whether a dermato-
phyte also grows or not (see, for example, ref. 14). In
the former case, authors may be stated to use aconservative simple association criterion, and in the
latter a liberal variant of this criterion. Data generated
by studies using an explicit liberal simple-association
criterion are very similar to data generated in studies
that simply list all fungal isolations from onychomyco-
tic nails without definitively pronouncing on the
significance of the nondermatophytic isolations (see,
for example, ref. 15). Such agnostic simple association
studies tend to be presented in a way that causes them
to be routinely interpreted by later authors as if they
documented nondermatophytic infections rather than
mere isolations (e.g. see handling of such a study done
by Budak et al. [15] in a review by Greer [2]); they thus
form a category of publication that is de facto distinct
from studies that explicitly purport to document the
incidence of probably etiologically insignificant fungi in
association with nails or feet (see, for example, refs
16,17). The simple association criteria have long been
subjected to criticism, but this criticism is not universal;
indeed, one such study recently was awarded the 2001
William J. Stickel Bronze Award by the American
Podiatric Medical Association [14]. It generally agreed
that, as Ellis [18] stated, ‘there is considerable con-
troversy on the significance of nondermatophyte moulds
and yeasts when they are identified in the presence of a
dermatophyte.’A thorough review of past onychomycosis studies
and their relation to these analytical categories is
needed, but is beyond the scope of the present paper.
In addition to the above-listed common interpretive
criteria about which some controversy may exist, there
are also two definitive high-tech criteria, namely
differential flow cytometry and specific in-situ immu-
nohistochemistry of infected nail tissue [19,20]. Though
these techniques appear to give unequivocal results, the
labor and cost of setting them up has so far deterred
use in both routine and other reference laboratories.
The merits and deficiencies of inoculum-counting
techniques for analyzing NDF in onychomycosis have
been elaborated in a previous analysis [21]. The purpose
of the present study was to evaluate the classic and
simple-association NDF sole-agent and mixed-infec-
tion evaluation criteria with a data set derived from a
large longitudinal study of onychomycosis cases. In the
study, untreated onychomycosis patients yielded an
extended series of nail specimens over a period of up
to 3 years. The multiple specimens taken per nail, and
the long-term follow-up involved, provided gold stan-
dard data that could be used to distinguish both causal
and incidental fungi. Comparison of this decisive
information with the ways in which initial specimens
from the same patients would be interpreted by routine
laboratories using the various common confirmation
criteria allowed a determination of how well these
criteria perform as interpretative frameworks for
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 41
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
determining etiologic significance in putative NDF
onychomycoses.
Materials and methods
Patient population
Details of the selection and composition of the patient
population have been given by Gupta et al. [21]. In
brief, patients were enrolled in the study on the basis of
having at least one abnormal, apparently lesional
toenail that had not been treated with antifungal drugs
at least 6 months prior to the study and that, by
consent, would not be treated over the course of the
study. In total, 473 patients (mean age 64, standarddeviation (SD)9/15 years) were included in the study, of
whom 341 yielded sufficient specimens to be fully
evaluated in the context of the present study. Each
fully evaluated patient was examined at least three
times within a 3-year interval, and the majority were
examined on five or more occasions.
Sampling and laboratory study
Sampling was done with a sterile #15 scalpel blade or
curette according to standard procedures [22]. The area
of nail to be sampled was disinfected with ethanol andsuperficial material was abraded away and discarded
before more deeply situated material was collected for
fungal analysis.
Laboratory studies
The techniques recommended by Kane and Summer-
bell [23] were followed for inoculation of nail scrapings.
To facilitate growth of dermatophytes, Sabouraud agar
amended with a mixture of 100 mg/ml cycloheximide,
100 mg/ml chloramphenicol, and 50 mg/ml gentamicin
(CCG) was used in combination with Casamino acids�/
erythritol�/albumin (CEA) medium [24] plus CCG. Tofacilitate mold outgrowth, a cycloheximide-free med-
ium, Littman oxgall agar (Difco Laboratories, Detroit,
MI) amended with 100 mg/ml streptomycin, was
employed. Media with no fungal growth were declared
negative at 21 days. In a deviation from routine
practice, all mold species growing from the specimens
were identified at least to genus level, and all potentially
causal molds except Acremonium spp. were identifiedto the species level. Molecular analysis of the Acre-
monium isolates involved is currently in progress.
Infection classification
Individually for each affected toe, lesional areas
compatible with superficial white onychomycosis were
sampled separately from areas compatible with distal�/
lateral subungual onychomycosis in order to assurethat all etiologic agents involved were elucidated. The
records obtained were pooled for each patient in the
present study, but a breakdown by disease type is given
by Gupta et al. [21]. Records for each patient were
examined and when a distinct pattern of fungal
isolation became clear the patient was classified as to
whether he or she supported an uncomplicated derma-
tophytosis, an NDF infection, a mixed NDF/dermato-phyte infection or another infection type. Those who
grew a dermatophyte at least once but did not
consistently grow any particular mold after two or
more visits were classed as uncomplicated dermatophy-
tosis patients. Those who (i) grew a particular mold
consistently on three successive occasions; (ii) yielded a
specimen showing consistent filaments in direct micro-
scopy at least once; and (iii) continued to grow the samemold consistently in most or all specimens during
follow-up of the same nail (e.g. left hallux) and nail
stratum (i.e. subungual or superficial), without growing
a dermatophyte on any occasion, were classified as
NDF onychomycosis patients. Those who grew a
dermatophyte on one or more occasions, and also
grew a mold with the same consistency, site specificity,
and (where occurring alone in an individual nail orstratum) direct microscopic verifiability as indicated
above for NDF onychomycosis, were classed as mixed-
infection patients. S. dimidiatum and Scytalidium
hyalinum sole-agent and mixed infections were consid-
ered to constitute a separate category; any outgrowth of
these species, which do not occur as contaminants in
the study area, was considered etiologically significant.
Patients who grew neither a dermatophyte nor aconsistent mold, and who were negative for fungal
filaments in direct microscopy, were classed as having
lesions of nonmycotic etiology. Cases in which a patient
yielded positive direct microscopy at least once but also
failed to grow a dermatophyte or a consistent mold in
multiple specimens were considered to be of unresolved
etiology.
Evaluation of laboratory practices
The cases studied were evaluated in terms of how they
would be understood based on reports generated in
normal laboratory practice, both in routine labora-
tories and in the reference laboratory of R.C.S. As
routine practice relies heavily on the results of the initial
patient specimen, which is often the only one taken, theresults of this specimen were focused on and were
contrasted with the overall conclusion about the
patient’s infection status after multiple samples had
– 2005 ISHAM, Medical Mycology, 43, 39�/59
42 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
been analyzed. In the classic evaluation procedure for
NDF onychomycoses, the results of the first twospecimens are usually taken as the deciding point;
therefore, in evaluating the success of this criterion, the
results deriving from this pair of specimens were
considered. Only successive specimens from the same
nail were taken into consideration in this evaluation.
Sensitivities, specificities, and positive predictive values
were generated where appropriate both for discrete
types of laboratory results, such as elucidation of adermatophyte in a dermatophytosis patient, and for
judgment criteria used to make inferences about the
status of infections (e.g. criteria stating that, given
certain findings, patients should be deemed to have or
not have an NDF infection). Sensitivity was calculated
by the formula TP}/(TP�/FN), where TP�/true posi-
tives and FN�/false negatives. Specificity was TN}/
(TN�/FP), where TN�/true negatives and FP�/falsepositives. Positive predictive value was reckoned as
TP}/(TP�/FP). In the case of evaluation criteria
applied to NDF growing from specimens, a correct
affirmation of etiologic status was considered a true
positive, a correct exclusion of etiologic status a true
negative, and so on.
We also investigated the efficacy of two notes used in
the R.C.S.’s laboratory in an attempt to elicit themaximal number of NDF onychomycosis confirma-
tions according to classic criteria. These notes are
attached to outgoing report sheets for nail specimens
that have grown a potentially significant NDF.
‘Note 1’
Note 1 is similar to those sent out by many laboratories
for potentially significant NDF that grow from speci-
mens where direct microscopy has revealed the presence
of nonspecific (i.e. not recognizable with certainty as
dermatophytic or nondermatophytic) fungal filaments,
and reads as follows.
This common, normally nonpathogenic fungus, when isolated from
filament-positive nail material, is usually a contaminant overgrowing
nonviable dermatophyte inoculum. Uncommonly, it causes onycho-
mycosis and can be confirmed as an etiologic agent by consistently
repeated isolation in the absence of a dermatophyte. Rare mixed
infections will show consistently repeated isolation in successive
samples along with at least one isolation of a dermatophyte.
This note is not used for all NDF outgrowth, as
many such species are constantly occurring contami-
nants that never cause onychomycosis; its use isrestricted to cases of isolation of species well docu-
mented [1] via the classic confirmation procedure (often
supplemented with additional repeat specimens) as
being potentially significant (Table 2). A few potentially
significant species, such as Chaetomium globosum and
Alternaria alternata , are excluded from the list ofelicitors of Note 1, because they are constant nail
contaminants that only very rarely cause onychomy-
cosis, and because actual onychomycosis cases caused
by these fungi are often, though not always, associated
with dark or otherwise unusual filaments in direct
microscopy. An efficient way of dealing routinely with
such fungi in rare cases where they produce only
nondescript filaments in infected nail tissue is yet tobe designed.
‘Note 2’
Note 2 is as far as we know an innovation of our own
based on the observation that true NDF onychomy-
cosis cases are more likely than dermatophytoses to be
associated with patchy nail invasion leading to anunusually high chance of fortuitously negative direct
microscopic findings in readings of single specimens
[21]. It appears that these fortuitously negative readings
Table 2 Fungi routinely reported via Notes 1 and 2 (see main text)
as potential agents of opportunistic onychomycosis when encountered
in outgrowth from nail specimens. Corresponding teleomorph or
anamorph names are given where these alternative states are
commonly seen in clinical laboratory cultures and aid in routine
identification
Acremonium potronii
A. sclerotigenum
A. strictum
Aphanoascus fulvescens (anamorph Chrysosporium state of A.
fulvescens )
Aspergillus candidus
A. chevalieri (teleomorph Eurotium chevalieri )
A. flavus
A. fumigatus
A. nidulans (teleomorph Emericella nidulans )
A. niger
A. sydowii
A. terreus
A. ustus
A. versicolor
A. vitis (teleomorph Eurotium amstelodami )
Curvularia lunata
Fusarium oxysporum
F. proliferatum
F. solani
F. verticillioides (formerly called F. moniliforme )
Lasiodiplodia theobromae
Microascus cinereus (anamorph Scopulariopsis state of M. cinereus )
M. cirrosus (anamorph Scopulariopsis state of M. cirrosus )
Onychocola canadensis
O. kanei
Scopulariopsis brevicaulis
S. brumptii
S. candida
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 43
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
are not generally based on reader error; in fact, the
reading is generally accurate but the nail fragmentschosen for reading happen to lack fungal elements by
chance alone. Exhaustive analysis shows, however, that
a minority of other fragments in the same specimen
packet generally do contain these elements [21]. Thus a
single negative direct microscopic reading cannot be
taken to rule out NDF onychomycosis when a Table 2
mold grows out in culture, especially when this growth
is at a high inoculum level [21].To cover this eventuality in routine practice where
the physician and patient must decide if submitting a
second specimen is worthwhile, we composed Note 2 as
follows.
In cases like the present one where direct microscopic examination
shows no mycelium, this isolation would ordinarily be considered
surface contamination. Since the species isolated does occasionally
cause onychomycosis, however, in addition to being a common
contaminant, this isolation is noted for your records. In a small
proportion of such cases, a follow-up specimen from the affected nail
will yield positive direct microscopy and regrow the same fungal
species, confirming opportunistic onychomycosis.
Results
Infections
Of 341 patients yielding three or more evaluable
specimens, 311 yielded results that allowed unequivocal
classification of the infection status of their nails (Table
3). The remainder showed fungal filaments in direct
microscopy on one occasion but grew neither a
dermatophyte nor a consistent mold in up to eight
successive specimens; these patients were considered tobe of unresolved status. Of the 311 patients with
resolved status, 177 (56.9%) were uncomplicated der-
matophytosis patients, 17 (5.5%) had dermatophytosis
mixed with NDF infection; 67 (21.5%) had NDF
infection and 47 (15.1%) were considered to have nail
lesions for which fungal etiology had been ruled out.
There was also one probable yeast infection, one
pure S. dimidiatum infection, and one mixedT. mentagrophytes �/Scytalidium hyalinum infection.
Species identification of most of the NDF implicated
in causing the infections scrutinized in this study is
given in our previous publication [21] in connection
with relevant inoculum count data. The dermatophytes
isolated in uncomplicated infections included 121
Trichophyton rubrum , 50 T. mentagrophytes complex,
one Microsporum gypseum, three mixtures of T. rubrum
and T. mentagrophytes complex, one mixture of
T. rubrum and T. violaceum , and one mixture of
T. rubrum , M. gypseum and T. mentagrophytes com-
plex. Mixed dermatophyte/mold infections included, as
dermatophyte components, nine T. rubrum , seven T.
mentagrophytes complex, and one mixture of T. rubrum
and T. mentagrophytes complex. Just one case could be
tentatively classed as a yeast infection, with persistent
colonization seen by Candida parapsilosis. In this case,
no pseudohyphae were seen in direct microscopy and
the nail disruption was concluded to have been caused
by adjacent paronychia.
Direct microscopy
In initial specimens, cases of dermatophytosis, includ-
ing mixed infections, were signaled by positive direct
microscopy (fungal filaments seen) in 144 of 195 cases
(test sensitivity of 73.8%). In the long-term, positive
direct microscopy was confirmed for 100% of derma-
tophytosis patients.In uncomplicated NDF infection, the initial speci-
men was microscopic positive in 45 of 67 cases (67.2%).
This proportion of microscopic positives is statistically
similar to that seen with dermatophytes in the initial
specimen (P�/0.68, chi-squared). In the remaining 22
cases, the initial specimen was microscopic-negative,
though in all cases multiple later specimens were found
to be positive. Including correctly negative readings fortrue nonmycotic nails along with positive readings for
both dermatophyte- and NDF-infected nails, direct
microscopy of the initial specimen gave an accurate
picture of the nail’s infection status in 235 of 311
evaluable cases (75.6%).
Some physicians may treat patients for dermatophy-
tosis on receiving or generating a report that fungal
filaments are present in direct microscopy of nails. Forthe resolved cases in the present study, however, only
131 of 190 instances of positive direct microscopy in the
initial specimen signaled dermatophytosis (i.e. this test,
if used as an indicator of uncomplicated dermatophy-
tosis, has a positive predictive value of only 70.0%, at
least in populations similar to the one studied here). If
mixed dermatophyte infections (including NDF and
Scytalidium mixtures) are included as valid diagnosesof dermatophytosis, despite any differences in thera-
peutic strategy that might be entailed by mixtures, the
positive predictive value of this test would still be just
75.8%.
Scopulariopsis brevicaulis cases are occasionally
associated with the appearance of recognizable Scopu-
lariopsis elements, usually conidia, in direct microscopy
along with fungal filaments [1]. In the present study,distinctive Scopulariopsis direct microscopic presenta-
tion was seen in only two of 13 S. brevicaulis cases in
direct examination of the initial specimen, and addi-
– 2005 ISHAM, Medical Mycology, 43, 39�/59
44 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
Table 3 Correspondence of patient infection categories, as determined by multiple follow-up samples, with the results of initial nail specimens
Outgrowth from first
specimen
Direct
microscopy
for first
specimen
Infection category (final classification based on longitudinal study)
Dermatophytosis
alone
Dermatophyte-
mold mixed
infection
Dermatophyte-
Scytalidium
mixed
infection
Mold infection
alone
Scytalidium
infection
alone
Yeast
infection
Nonmycotic
nail
abnormality
Unresolved:
KOH�/once,
no significant
culture in four
or more
specimens
Dermatophyte only �/ 67 3 �/ �/ �/ �/ �/ �/
�/ 20 �/ �/ �/ �/ �/ �/ �/
yc 1 �/ �/ �/ �/ �/ �/ �/
Mold only �/ 17 3 �/ 43 (incl. 1 insignif.)* �/ �/ �/ 8
�/ 3 2 �/ 10 (incl. 2 insignif.)* �/ �/ 19 7
yc �/ �/ �/ �/ �/ �/ 1 �/
Dermatophyte plus mold �/ 28 6 �/ �/ �/ �/ �/ �/
�/ 7 �/ �/ �/ �/ �/ �/ �/
Dermatophyte plus yeast �/ 8 �/ �/ �/ �/ �/ �/ �/
�/ 1 �/ �/ �/ �/ �/ �/ �/
Scytalidium alone �/ �/ �/ 1 �/ �/ �/ �/ �/
Yeast alone �/ 1 �/ �/ 1 �/ �/ �/ 3
�/ 2 �/ �/ 2 �/ �/ 6 1
Mold plus yeast �/ 1 �/ �/ �/ �/ �/ �/ �/
�/ �/ �/ �/ �/ �/ 1 �/ �/
yc �/ �/ �/ �/ �/ �/ 1 �/
Negative (no fungus grown) �/ 9 �/ �/ 1 1 �/ �/ 4
�/ 12 3 �/ 10 �/ �/ 20 7
Total (n�/341) �/ 177 17 1 67 1 1 47 30
Direct microscopy: �/, fungal filaments present; �/, no fungal elements seen; yc, budding yeast cells seen; *insignificant in the number of cases specified, a mold grew in the initial specimen,
but it was etiologically insignificant and in fact a different mold species was ultimately shown to be the agent of infection (see comments in text about wrong mold isolations).
–2005
ISHA
M,M
edical
Myco
logy,
43,39�
/59
Onych
om
ycosis
45
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
tional examinations in connection with later samples
revealed such in-situ elements in only one additional
case (Table 4). Scopulariopsis conidia were not seen in
any of the three well-established mixed S. brevicaulis /
dermatophytosis cases examined.
Culture
To illustrate the types of diagnostic decisions that were
necessary in this study, a small number of selected
culture records from our patients are shown in Table 5.
A typical, relatively uncomplicated NDF case is pre-
sented (patient OH-1) as well as an uncomplicated
dermatophytosis yielding a few insignificant NDF
colonies (patient MA-14). In addition, more complex
types of cases are shown (e.g. patients OL-1, SI-1 and
BO-1) for whom initial specimen results gave little
evidence of the actual type of infection involved.
Dermatophytes
In uncomplicated dermatophytosis patients, the initial
specimen yielded 132 of the 177 etiologic dermato-
phytes that were ultimately detected (test sensitivity of
74.6%; Table 6). Another 14.1% of the infecting
dermatophytes first grew out in the second specimen,
while in 11.4% of cases the infecting dermatophyte
grew out only in the third or a later specimen. A similar
pattern was seen in mixed dermatophytosis/mold
infections, where a small majority of etiologic derma-
tophytes grew out from the initial specimen, but nearly
half first grew from a repeat specimen. In two cases,
dermatophyte growth was not seen until the fifth
successive specimen or later.
Specimens yielding positive direct microscopy and
growth of only a dermatophyte are generally inter-
preted as straightforward cases of dermatophytosis.
Longitudinal study in the present series of patients,
however, showed that three of 70 such initial specimen
records (4.3%) were actually derived from mixed
dermatophyte/NDF cases.
NDF
In uncomplicated NDF cases, the NDF species ulti-
mately judged to be etiologic in the case grew from the
first specimen in 50 of the 67 cases studied (test
sensitivity of 74.6%). In three additional cases in which
an NDF was judged causal, one or more NDF grew in
the initial specimen but the species ultimately found to
be causal did not do so; the outgrowing NDF from the
initial specimens in these cases were purely incidental.
To minimize the considerable confusion this minor
factor causes, these are noted hereafter as cases of
‘wrong mold’ outgrowth from initial specimens. If the
17 NDF from mixed dermatophytosis/NDF infection
are included along with the uncomplicated NDF cases
in the assessment of initial culture reliability, then
results show that overall, 61 etiologically significant
NDF (72.6%) grew in the initial specimen, while 23
were first grown from the second or a later specimen.In a total of 140 cases in which an NDF grew in the
initial specimen (Table 7), this mold was causally
significant in either an uncomplicated NDF infection
or mixed dermatophytosis/NDF infection in 61 cases
but was insignificant in 79 cases, giving initial-specimen
culture of an NDF a positive predictive value for
etiology of just 43.6%. Though most of the etiologically
significant initial-specimen NDF outgrowths were
combined with positive direct microscopy (51 of 61
cases; 83.6%), 10 cases showed negative direct micro-
scopy in the initial specimen. (As mentioned above,
positive direct microscopic readings were obtained in
later specimens for all such cases in which an NDF was
found to be causal). In cases in which all NDF growing
in the initial specimen were insignificant, direct micro-
scopy was positive in 47 cases, including 46 dermato-
phytosis cases and one of the three instances of wrong
mold outgrowth mentioned above. The causal derma-
tophyte itself grew from the initial specimen in 28 of the
46 dermatophytosis cases, leaving 18 true dermatophy-
tosis cases in which an insignificant NDF was the sole
filamentous fungus growing out from a specimen
microscopic positive for fungal hyphae.
Table 4 Appearance of Scopulariopsis -specific structures in direct microscopy in infections caused by S. brevicaulis
Infection category Direct microscopic
result classification
Finding in first
specimen
Finding in total number
of specimens taken
(at least three per patient)
Pure Scopulariopsis cases conidia present 2 3
conidia absent 8 7
Mixed dermatophytosis/Scopulariopsis infection cases conidia present �/ �/
conidia absent 3 3
– 2005 ISHAM, Medical Mycology, 43, 39�/59
46 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
Table 5 Chronological sample series for six patients with onychomycosis. Records are arranged vertically under coded patient designators, with initial sample at top of column for each
patient and later samples towards the bottom. A final diagnosis for each patient is given at the bottom of the column of records
OH�/1* PH�/1 MA�/14 OL�/1 SI�/1 BO�/1
4/96$, L3R3%, SW§, �/,
Acremonium sp. (14) **
4/96 L1R1%, DL§, �/, Aspergillus terreus (7) 5/96, R1, DL, �/,
Trichophyton
rubrum
4/96, R1, DL, �/,
Paecilomyces
lilacinus (1),
Aspergillus nidu-
lans (1)
5/96, L1, DL, �/,
Aspergillus versicolor
(1), Arthrographis
kalrae (3)
5/96, R1, DL, �/, Candida albicans
6/96, L3R3, SW, �/,
Acremonium sp. (15�/)
4/96, L2, SW, �/, nfg%% 10/96, R1, DL, �/,
Fusarium solani (1)
6/96, R1, DL, �/,
A. nidulans (1)
11/96, L1, DL, �/,
Fusarium solani (8),
Aspergillus
fumigatus (1),
Alternaria alternata
(1)
6/96, R1, DL, �/, Scopulariopsis
brevicaulis (n.c. §§), Candida sp.
6/96, R1, DL, �/,
Acremonium sp. (2),
Scedosporium sp. (1)
8/96, L1, DL, �/, A. terreus (n.c.) 11/96, R1, DL, �/, T.
rubrum
10/96, R1, DL, �/,
A. nidulans (8)
4/97, L1, DL, �/,
Fusarium solani (10)
6/96, L4, SW, �/, Aspergillus candidus
(1), Alternaria alternata (1)
8/96, R3, SW, �/,
Acremonium sp. (15�/)
8/96, R1&2, SW, �/, Onychocola canadensis (9) 1/97, R1, DL, �/, T.
rubrum , Candida sp.
1/97, R1, DL, �/,
A. nidulans (1)
9/97, L1, DL, �/,
Fusarium solani (10),
A. fumigatus (7)
9/96, R1, DL, �/, S. brevicaulis (n.c.)
8/96, R1, DL, �/,
Acremonium sp. (15),
Candida sp.
10/96, L1R1, DL, �/, A. terreus (15�/) 3/97, R1, DL, �/, nfg 3/97, R1, SW, �/,
A. nidulans (3)
10/97, L1, DL, �/,
Fusarium solani (n.c.),
Penicillium sp. (n.c.)
9/96, L4, SW, �/, nfg
11/96, L3, SW, �/,
Acremonium sp. (5), Sco-
pulariopsis brevicaulis (1)
4/97, R2, DL, �/, O. canadensis (15�/) 6/97, R1, DL, �/,
Scopulariopsis
brumptii (4)
5/97, R1, DL, �/,
Acremonium sp.
(1)
12/97, L1, DL, �/,
Fusarium solani (15)
10/96, R1, DL, �/, S. brevicaulis (n.c.),
Candida sp.
11/96, R1, DL, �/, nfg 7/97, R1, DL, �/, O. canadensis (15�/), Alternaria
alternata (2)
12/97, R1, DL, �/,
nfg
10/97, R1, DL, �/,
A. nidulans (8)
2/97, R1, DL, �/, Onychocola
canadensis (15�/), Candida sp.
4/97, L3, SW, �/,
Acremonium sp. (3)
5/98, R2, DL, �/, O. canadensis (15�/) 5/98, R1, DL, �/, T.
rubrum
1/98, R1, DL, �/,
A. nidulans (7)
5/97, R1, DL, �/, C. albicans
4/97, R1, DL, �/,
Acremonium sp. (13)
5/98, R1, DL, �/, O. canadensis (15�/),
Trichophyton mentagrophytes
7/98, R1, DL, �/,
Penicillium sp. (2)
5/97, L4, SW, �/, A. candidus (1)
10/97, R3, SW, �/,
Acremonium sp. (1)
5/98, L1, DL, �/, A. terreus (15�/),
T. mentagrophytes
10/98, R1, DL, �/,
Phoma sp. (n.c.)
6/97, R1, DL, �/, O. canadensis (15),
Penicillium sp. (2), Chaetomium
globosum (1), S. brevicaulis (1)
10/97, R1, DL, �/,
Acremonium sp. (2),
Scopulariopsis brevicaulis
(1)
7/98, L1, DL, �/, A. terreus (10) 6/97, L4, SW, �/, nfg
2/98, L2&3, SW, �/,
Acremonium sp. (15�/),
11/98, R1, DL, �/, O. canadensis (15) 6/97, R1, DL, �/, O. canadensis (15),
C. globosum (1), S. brevicaulis (2),
Microascus cirrosus (3)
–2005
ISHA
M,M
edical
Myco
logy,
43,39�
/59
Onych
om
ycosis
47
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
Direct microscopy plus culture of first specimen
Studies comparing alternative techniques such as nail
histopathology to KOH-based direct microscopy and
culture (see Discussion below) often contrast the
sensitivities of various techniques for revealing the
presence of any and all etiologic filamentous fungi inabnormal nails in the first specimen taken. For
comparison with such studies, it should be noted
that in the present study, for the first specimen alone,
the overall sensitivity of KOH examination and culture
together to unequivocally demonstrate the involve-
ment of a filamentous fungal etiologic agent, whether
with a demonstration of fungal filaments present in
KOH or with culture positive for either a dermato-phyte or a Scytalidium , was 83.9% (219 definite
positives in first specimens for 263 patients with
known filamentous fungal infection; all relevant data
included in Table 3). The FN figure used in the
denominator of this calculation was based on gold-
standard knowledge of the total number of patients
who could ultimately be found to have at least one
etiologic filamentous fungus; this figure has not beendetermined in any previous study of which we are
aware.
Evaluation criteria evaluated
The existing evaluation criteria differ mainly in their
treatment of specimens in which nondermatophytic
fungi grow out. No dermatologic mycology laboratory
system provides physicians with information poten-
tially prejudicing an interpretation of specimens that
are both direct microscopic negative and culture
negative, or of specimens that are microscopic positive
and culture-negative. Thus analysis of the overallaccuracy of these interpretative systems per se must
generally be focused on specimens yielding various
categories of cultured organisms.
Simple association: conservative variant
Considering only cases with positive direct microscopy
and outgrowth of a filamentous fungus in culture, the
conservative variant of the simple association criterion
for judging initial specimens (all NDF considered
etiologic if growing in the absence of a dermatophyte
from a specimen microscopically positive for fungal
filaments; any dermatophytes growing always consid-ered to be sole agents of infection) would provide a
correct assessment in 145 cases (81.9%) and an
incorrect assessment in 32 cases (18.1%), consisting
of nine cases in which an etiologic NDF involved in a
mixed dermatophytosis infection would be missedTa
ble
5(C
on
tin
ued
)
OH� /
1*
PH�/
1M
A�/
14
OL�/
1S
I�/1
BO�/1
2/9
8,
R1
,D
L,�/,
Acr
emo
niu
msp
.(8
)
11
/98
,L
1,
DL
,�
/,A
.te
rreu
s(1
5�
/)9
/97
,L
4,
SW
,�/
,S
.b
revi
cau
lis
(1),
Asp
erg
illu
sve
rsic
olo
r(1
1),
Alt
ern
ari
a
alt
ern
ata
(1)
5/9
8,
L2&
3,
SW
,� /,
Acr
emo
niu
msp
.(1
5�
/)
5/9
9,
R4
,D
L,� /
,O
.ca
na
den
sis
(15
)6
/98
,R
1,
DL
,�/
,n
fg
5/9
8,
R1
,D
L,� /,
Acr
emo
niu
msp
.(1
5�
/)
11
/99
,R
4,
DL
,�
/,O
.ca
na
den
sis
(15
),
T.
men
tag
rop
hyte
s
6/9
8,
L4,
SW
,�/,
S.
brevi
cauli
s(1
3),
C.
alb
ica
ns
11
/99
,R
2,
DL
,�
/,O
.ca
na
den
sis
(15�
/),8
/98
,R
1,
DL
,�
/,O
nych
oco
laca
na
-
den
sis
(15
),C
an
did
asp
.
9/9
7,
L4
,S
W,�
/,n
fg
Dia
gn
osi
s
Acr
emo
niu
msp
.D
Lin
R1
an
dS
Win
L3
an
dR
3,
po
ssib
lya
lso
L2
A.
terr
eus
an
dT
.m
enta
gro
phy
tes
DL
inL
1;
O.
cana
den
sis
an
dT
.m
enta
gro
phy
tes
DL
inR
1a
nd
R4
;O
.ca
na
den
sis
DL
inR
2;
O.
can
ad
ensi
sS
Win
R1
an
dR
2
T.
rub
rum
DL
inR
1A
.n
idu
lan
sD
L
an
dS
Win
R1
F.
sola
ni
DL
inL
1S
.b
revi
cauli
sa
nd
O.
cana
den
sis
DL
inR
1;
SW
inL
4o
fu
nce
rta
inca
use
,
po
ssib
lyA
.ca
nd
idus
an
d/o
rS
.
brevi
cauli
s
*P
ati
ent
cod
ed
esig
nati
on
;$t
ime
sam
ple
tak
en(m
on
th/y
ear)
;%n
ail
(L,
toe
of
left
foo
t;R
,ri
gh
t);
§str
atu
min
volv
ed(D
L,
dis
tal�
/late
ral
sub
un
gu
al
on
ych
om
yco
sis;’
dir
ect
mic
rosc
op
icre
sult
;
**
gro
wth
den
sity
(nu
mb
ero
fin
ocu
law
ith
po
siti
ve
ou
tgro
wth
ou
to
f1
5p
lan
ted
);§§
n.c
.,n
oin
ocu
lum
cou
nt
ma
de;
%%n
.f.g
.,n
ofu
ng
us
gro
wn
;S
W,
sup
erfi
cia
lw
hit
eo
nych
om
yco
sis)
.
– 2005 ISHAM, Medical Mycology, 43, 39�/59
48 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
(three of the nine cases) or incorrectly ignored (six
cases), 17 cases in which an etiologic dermatophyte
involved in uncomplicated dermatophytosis would be
missed and an incidental NDF considered etiologic
instead, and four cases in which an etiologic dermato-
phyte involved in a mixed infection would be missed
and sole etiology would be ascribed to the etiologic
NDF involved. There was also one case in which an S.
hyalinum isolate involved in a mixed infection with a
dermatophyte would be incorrectly judged a sole agent
as a result of the dermatophyte’s failure to grow in the
initial specimen, and one case in which NDF infection
would be attributed to an incidental wrong mold NDF
instead of the actual etiologic species. The figures for
correct and incorrect judgments in regard to the
conservative simple association criterion are based on
the premise that the user of the criterion would not
make a final judgment about cases in which positive
direct microscopy was accompanied by no fungal
growth or by growth of only a yeast. If the user
interpreted the outgrowing yeasts as etiologic, the
percentage of erroneous judgments involved would
increase.
When conservative simple-association is analyzed
specifically for its ability to assess the significance of
NDF outgrowth from microscopic-positive specimens
from which only an NDF grew in the initial specimen
(or an NDF plus a yeast, assuming the latter would be
ignored as not corresponding to the elements seen in
direct microscopy), it can be seen that with the data in
Table 3 this practice would generate correct attributions
of NDF etiology (to the correct NDF species rather
than to a wrong mold outgrowth) in 42 cases (65.6%),
whereas it would attribute causality to an insignificant
NDF in 19 cases (29.7%) and would attribute sole
causality to three molds actually involved in a mixedinfection with a dermatophyte (4.7%).
In order to obtain the overall sensitivity of con-
servative simple-association applied to etiologic NDF
(including sole-agent NDF and mixed dermatophytosis
components), one needs to determine the proportion of
the true cases that would be correctly recognized by a
criterion accepting only etiologic NDF growing from
microscopic positive initial specimens (i.e. 51), minussix cases of etiology incorrectly dismissed because of a
co-occurring dermatophyte, that is, 45 of 84 cases
(53.6%). The positive predictive value of this criterion
for our study population, 70.3%, is determined by
dividing its 45 correct calls into a total composed of
these calls plus the total of its false-positive calls,
including the above-mentioned 18 cases in which an
insignificant NDF was the sole filamentous fungusgrowing out from a microscopic-positive specimen, plus
one incorrect call based on wrong mold outgrowth
from a microscopic-positive specimen. The specificity
of conservative simple-association for correctly recog-
nizing NDF etiology is 76.5%, based on correctly
calling 62 nonetiologic NDF ‘negative’ and falsely
calling another 19 ‘positive’.
Simple association: liberal variant
The liberal simple-association criterion (all NDF
considered etiologic if growing from a specimen
microscopically positive for fungal filaments), when
applied to microscopic-positive, filamentous-fungal-
culture-positive initial specimens in our data, wouldprovide a correct judgment for 123 cases (69.5%),
including 75 uncomplicated dermatophytosis cases,
Table 6 Serial specimen first yielding dermatophyte in pure dermatophytosis and mixed-infection patients. Specimen first yielding dermatophyte
Infection category First Second Third Fourth Fifth or later
Dermatophytosis 132 (74.6)* 25 (14.1) 7 (4.0) 4 (2.3) 9 (5.1)
Mixed dermatophyte-mold infection 9 (52.9) 4 (23.5) 2 (11.8) �/ 2 (11.8)
Mixed dermatophyte-Scytalidium infection 1
*Percentages of total dermatophytes within each category (sole dermatophytosis/mixed dermatophytosis) are in parentheses.
Table 7 Non-Scytalidium molds growing from the initial specimen: direct microscopy reading and etiologic significance*
Filaments in initial specimen
direct microscopy
Cases in which a significant
NDF grew in 1st specimen
Cases in which one or more
insignificant NDF grew in
first specimen
Total
�/ 51 47 98
�/ 10 32 42
Total 61 79 140
*Cases with ‘unresolved’ final status and cases with only yeast cells in direct microscopy of initial specimen are excluded.
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 49
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
six mixed infections (with both NDF and dermatophyte
component correctly recognized), and 42 uncompli-cated NDF infections. On the other hand, it would
generate incorrect judgments for 54 cases (30.5%),
including 18 cases of incidental NDF growing from
uncomplicated dermatophytosis patients being falsely
considered etiologic, 28 cases of incidental NDF
growing along with dermatophytes in initial specimens
falsely being considered participatory in mixed infec-
tions, three cases of NDF truly involved in mixedinfections with dermatophytes falsely being considered
sole agents, and three cases of NDF involved in mixed
infections with dermatophytes being missed with re-
sulting attribution of the infection to the dermatophyte
alone. Also, as in the case of the conservative simple-
association criterion, there would be one case in which
an S. hyalinum isolate involved in a mixed infection
with a dermatophyte would be incorrectly judged a soleagent, and one case in which NDF infection would be
attributed to the wrong mold.
Applied to cases of NDF outgrowth from micro-
scopic-positive specimens from which only an NDF
grew in the initial specimen, liberal simple-association
would generate judgments identical to those made by
its conservative counterpart (i.e. it would be 65.6%
accurate for our sample).However, in relation to all true NDF infections (pure
and mixed), regardless of their initial direct microscopic
result, this criterion, in which all true cases initially
presenting as microscopic-negative are missed, would
have a sensitivity (for NDF only, ignoring dermato-
phyte components of mixed infections) of 51/84
(60.7%). This is higher than the sensitivity of the
conservative variant; NDF occurring in mixed infec-tions are not ignored in the liberal variant. The liberal
variant, however, has a positive predictive value based
on a numerator of 51 correct calls divided by a
denominator including these calls plus 18 false sole-
agent NDF attributions, 28 false mixed NDF attribu-
tions, and one attribution of infection to a wrong mold;
this ratio equals 52.0%. In other words, in populations
similar to that studied here, this criterion is indis-tinguishably better than a coin toss for determining
whether an isolated NDF is etiologic. More dramati-
cally still, the specificity of this criterion in terms of
correctly evaluating NDF is 42%.
Classic evaluation
The classic criterion (NDF infection accepted whendirect microscopy is compatibly positive at least once
and the same NDF is isolated from at least two
distinctly temporally separated samples of the same
nail), when applied to microscopic-positive, filamen-
tous-fungal-culture-positive initial specimens in ourdata, would provide an appropriate judgment for 164
cases (92.7%), including 121 pure dermatophytoses, 36
uncomplicated NDF infections and six mixed-infection
cases. On the other hand, it would miss or fail to
confirm at least one involved pathogenic type in six
pure NDF cases and six mixed infections (including
five in which the NDF was unconfirmed or undetected
and one in which the dermatophyte was undetected), aswell as one Scytalidium -dermatophyte mix and one
case with wrong mold outgrowth in the initial speci-
men. The 121 successfully handled pure dermatophy-
toses include nine cases that are considered
appropriately handled because even though the derma-
tophyte failed to grow in the first specimen (or, in cases
where Note 1 was issued, in the first two specimens),
the NDF that grew were not reported in a mannersuggestive of etiologic significance. This refraining from
misattribution caused these specimens to have the same
status as common direct-microscopic-positive, culture-
negative specimens, which are generally interpreted as
highly likely to be dermatophytic.
An NDF (whether significant or not) grew from a
microscopic-positive initial specimen in 70 cases in the
present study. In a system using the classic confirma-tion criterion for NDF follow-up investigation, culture
and direct microscopic findings from the initial speci-
men would trigger the sending out of Note 1 (see
above) in all of these cases except those in which the
NDF species growing did not occur in Table 2. In fact,
Note 1, in our regular dermatologic mycology practice,
would have been sent out in response to 61 micro-
scopic-positive specimens from the present study grow-ing a fungus from Table 2 in culture. As seen in the top
four rows of Table 8, 51 true cases of NDF or mixed
NDF onychomycosis would have been included in the
number of specimens for which initial examination
generated this response. In the remaining cases, the
note would have been sent out in response to a
noncausal NDF; in all such cases in the present study,
the second specimen grew a dermatophyte, a discor-dant, incidental NDF, an incidental yeast or no fungus.
This would have allowed the NDF from the first
specimen to be correctly classed as noncausal. In
evaluation of the initial specimen plus the first addi-
tional specimen submitted in response to Note 1
(assuming an ideal response by patients and physicians,
i.e. assuming that such a confirmatory specimen was
sent in by the attending physician each time the notewas issued), this note-sending practice would have led
to confirmation of 36 uncomplicated NDF infections
and NDF components of seven mixed infections. These
– 2005 ISHAM, Medical Mycology, 43, 39�/59
50 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
Table 8 True cases of nondermatophyte onychomycosis, including mixed infections: accuracy of diagnostic classification after study of two serial specimens, and effectiveness of two
informative notes used to solicit confirmatory specimens from physicians
Results of first two direct miroscopy
and culture tests
Conventional interpretation in classic system Infection category
D1 C1 D2 C2 Pure mold
infection
Mixed mold/dermatophyte
infection
�/ �/ �/ �/ Confirmed via note 1 22 6
�/ �/ �/ �/ Unconfirmed despite issue of note 1 �/ 2
�/ �/ �/ �/ Confirmed via note 1 14 1
�/ �/ �/ �/ Unconfirmed despite issue of note 1 6 �/
�/ �/ �/ �/ Unconfirmed; note 1 issued for common agents in report for 2nd specimen* 3 2
�/ �/ �/ �/ Unsuspected �/ 1
�/ �/ �/ �/ Confirmed via note 2 5 2
�/ �/ �/ �/ Unconfirmed despite issue of note 2 3 �/
�/ �/ �/ �/ Unconfirmed; note 1 issued for common agents in report for 2nd specimen 8 3
�/ �/ �/ �/ Unconfirmed; note 2 issued for common agents in report for 2nd specimen 3 �/
�/ �/ �/ �/ Unsuspected 3 �/
Totals a�/67 a�/17
a Confirmed by classic criterion 41 9
a Not confirmed by classic criterion within two specimens 26 8
C1, culture for specimen 1; C2, culture for specimen 2; D1, direct microscopy for specimen 1; D2, direct microscopy for specimen 2; ‘Confirmed’ means that the truly infecting
nondermatophytic agent or co-agent of the infection is confirmed as etiologically significant. The dermatophytic component of mixed infections is not referred to in this Table. For text of
Notes 1 and 2, see Materials and methods; *’Note 1 issued for common agents in report for second specimen’ means that though the infection in question would not be confirmed after two
successive specimens, the second specimen’s result, including a culture record of a species from Table 2, would trigger issuance of a note suggesting that an additional (third) specimen be
taken.
–2005
ISHA
M,M
edical
Myco
logy,
43,39�
/59
Onych
om
ycosis
51
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
numbers together constitute 84.3% of the 51 true NDF
or mixed infections eliciting the note. Six of the 42 trueuncomplicated NDF infections (14.3%) and two of the
nine mixed infections stimulating use of note one would
not be correctly confirmed due to chance failure of the
causal NDF to grow from the second specimen.
Note 2, used in cases where the initial specimen is
microscopic negative but grows a known potential
agent of onychomycosis (Table 2), would have been
issued in 21 cases in this study. This practice would haveresulted in second-specimen confirmation of uncom-
plicated NDF infection in five cases and mixed
dermatophytosis in two cases (Table 8, row 7), but
would falsely have missed three additional cases of
NDF infection. The missing of these cases would have
derived from the chance failure of the fungal elements
present in the infected areas to be seen in direct
microscopy in both the first and second specimens(Table 8, row 8). A further 17 true uncomplicated NDF
infections and six mixed infections would not have been
elucidated by procedures relying on use of Notes 1 and
2, as the results of the initial specimen would not have
triggered the issuing of either of these notes. If, in these
cases, the physician had spontaneously decided to
submit a second specimen despite not receiving a
laboratory note recommending this (as not infrequentlyoccurs in the Canadian health care system if onycho-
mycosis is strongly suspected but not confirmed by the
first specimen), 19 of these cases of missed infection
would have elicited Note 1 or Note 2 in examination of
the second specimen, and would thus have been set on
track for potential correct diagnosis despite the un-
helpful results of the first specimen. Only four of the
true NDF or mixed onychomycosis cases would stillhave been completely unsuspected after a second
specimen had been examined and cultured. In just
one of these four cases, a direct microscopic-positive
reading in one of the two specimens might have led the
physician to make a default clinical presumption of
dermatophytosis, in view of the two negative culture
trials that had occurred; in this case, in fact, a mixed
dermatophytosis/NDF infection was involved (Table 8,row 6). Importantly, no spurious, false attributions of
uncomplicated NDF or mixed NDF infection would
have been made as a result of reporting the outcomes of
the first two specimens, assuming that the submitting
physicians understood the classic confirmation criter-
ion.
It is salient to integrate the above results with the
remaining results for resolved cases in order to judgethe overall performance of the KOH-and-culture sys-
tem in conjunction with classic interpretive criteria and
the two advisory notes. Looking at the quality of first-
specimen results combined with any second-specimen
results directly elicited by use of the notes, one findsthat 230 of 311 cases (74%) would be judged completely
correctly, with all involved etiologic agents reported in
each case. These cases would break down as 136 of the
177 pure dermatophytoses ultimately recognized, in-
cluding four dermatophytoses culturally confirmed
only because Note 1 would have been sent out in
response to noncausal NDF, plus five cases in which
both the dermatophyte and the NDF components ofthe 17 total mixed infections would be fully confirmed
as a result of sending out Notes 1 or 2, 41 of 67
uncomplicated NDF cases, zero of two cases involving
Scytalidium, one yeast infection (detected as constant
C. parapsilosis growth only because Note 2 would have
been sent out in response to an insignificant S.
brevicaulis growth, eliciting a second specimen that
would not otherwise have been called for) and 47 of 47nonfungal cases. In addition, the dermatophyte com-
ponent but not the NDF component would be
recognized for six mixed infections, the NDF but not
the dermatophyte component for four mixed infections,
and the S. hyalinum but not the co-occurring derma-
tophyte in one mixed Scytalidium case, meaning that at
least one etiologic agent would be correctly recognized,
or the lack of fungal etiology correctly recognized, in241/311 (77.5%) of cases. It should be stressed that in
contrast to the 92.7% correctness figure given above for
microscopic-positive first specimens yielding a filamen-
tous fungal culture, this 77.5% correctness figure
reflects the dilution of accurate analysis by numerous
true-positive cases with negative direct microscopy and/
or a negative culture in the initial specimen.
Among cases with resolved status, there were 11microscopic-positive, culture-negative initial specimens
for which judgment would correctly remain pending in
the classic system, as well as in all other systems. Such
microscopic-positive, culture-negative reports, though
not resembling planned advisories like Notes 1 and 2,
may in themselves stimulate any physician wishing to
know the identity of an etiologic agent to consider
submitting a second specimen. If these 11 specimenswere subtracted from the number of inadequate initial
specimen results (i.e. if they were no longer considered
incorrect strictly because a true etiologic agent was left
unidentified), then 77.5% (241/311) of patients could be
said to have been adequately informed of their nails’
etiologic status by the methods used in this study,
exclusive of any results obtained for further specimens
spontaneously submitted by physicians. If the 11partially correct mixed-infection results were added,
as in the preceding paragraph, this would raise the
figure for what might be called ‘reasonably adequately
– 2005 ISHAM, Medical Mycology, 43, 39�/59
52 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
handled cases’ to 81.0%. The most significant compo-
nent of the cases considering inadequately handledwould be the 12 dermatophytoses, three mixed infec-
tions and 10 NDF infections, together comprising 8.0%
of the 311 resolved cases examined, for which both
direct microscopy and culture were completely negative
in the initial specimen. This inadequacy, of course, is
intrinsic to the sampling and/or laboratory techniques
used and does not vary among the interpretive systems
compared here.For contrast with the evaluations given of the simple
association criteria with pure and mixed NDF infec-
tions, the sensitivity and specificity of the classic
criterion should also be summarized in this regard.
The use of Notes 1 and 2 would have resulted in the
confirmation of 41 of 67 uncomplicated NDF infec-
tions and the NDF component of nine of 17 mixed
infections, giving a sensitivity of 59.5%. This is slightlylower than the 60.7% sensitivity of liberal simple-
association, as stringent confirmation requirements
allow chance negative findings in the second specimen
to obstruct a small number of confirmation attempts.
This sensitivity level, however, is partnered with a
positive predictive value and a specificity of 100%, as
this criterion makes false positives extremely unlikely.
In no case was the same incidental, noncausal Table 2NDF isolated from two successive specimens in a way
that would lead to a false report of confirmed NDF or
mixed onychomycosis. To rule out the possibility that
the delineation of gold-standard reference data in our
study might have been compromised by our allowing
two successive, coincidental growths of the same NDF
from a nail region invested with effete, nongrowing
dermatophyte filaments to be falsely interpreted as atrue positive for NDF infection, a third growth of the
same NDF was made a minimal criterion for accep-
tance of gold-standard-level true NDF etiology, and
the long-term record of outgrowth was monitored to
ensure there was no unexplained loss of an apparent
infective component in these untreated patients. Any
dubious cases would have been excluded from the
analysis; in fact, however, this step was only taken ina small number of cases where an insufficient number
of specimens could be obtained to clarify the status of
the outgrowth.
Unresolved cases have been excluded from all
analysis so far. In classic methodology 15 of these 30
cases (i.e. those with negative direct microscopy in the
initial specimen) would be given normal negative
reports. The remainder would be considered mycosesof unknown type; however, as our later records showed,
further attempts to gain insight into the nature of these
cases would be unfruitful even after seven or eight
successive specimens. The cases would therefore remain
unclassified from the viewpoint of laboratory mycol-ogy, although clinically they would likely be treated as
dermatophytosis. In either type of simple-association,
any of these cases yielding an NDF in the initial
specimen (i.e. eight cases in the present study) would be
incorrectly classed as infections caused by the isolated
NDF. With this factor included, and assuming that
none of these unconfirmable attributions of NDF
infection were correct, the positive predictive value ofconservative simple association for correct calls in
regard to NDF outgrowth would fall from 70.3 to
62.5%. That of liberal simple-association would fall to
48.1%. The value for the classic criterion for NDF
infection would remain at 100%.
Discussion
In most laboratory testing, test sensitivities and specifi-
cities below 90% would be considered unreliable,
especially for a final reference test. The extreme
ecological complexity of onychomycosis, however, has
so far engendered a situation in which definitive tests
must generally be judged on the basis of sensitivities,
positive predictive values and/or specificities in the
70�/80% range or below. Some authors, indeed, ignorecriticism and continue to use interpretive frameworks
with specificities below 50%. It should be noted,
however, that in the absence of gold-standard data
from longitudinal studies like the present one, these
accuracy levels have not previously been precisely
specified.
The value of etiologic agent identification in onycho-
mycosis is sometimes questioned, with the reflectionthat this information is superfluous unless drug re-
sponses vary among the etiologic agents identified [25].
The emphasis given in the present study to accurate
identification of the etiologic agent is partially based on
the knowledge that drug susceptibilities and in-vivo
responses often do vary significantly among onycho-
mycotic agents [21,26�/28]. In addition, however, our
viewpoint is that physicians generally prefer accuratelaboratory identification in cases where significantly
varying responses among etiologic agents are possible
(because ethical considerations make inappropriate
therapy even of rare types of cases objectionable,
especially when significant expense or any risk of
adverse effects is involved), and any impulse to de-
emphasize this accuracy in regard to onychomycosis is
simply a practical accommodation to the difficultiesand costs currently experienced. This response thus
reflects a pragmatic strategy that remains compatible
with the ultimate goal of improving accuracy by
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 53
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
gaining a more thorough scientific understanding of
the techniques, organisms and ecological relations thatbecome involved when onychomycosis is diagnosed in
the laboratory.
The patient population studied in the present analy-
sis was predominantly geriatric, and this group is well
known to have more NDF onychomycoses than are
found in younger groups [29]. The proportion of NDF
and mixed infections found in the present study should
not be expected to be duplicated in studies ondemographically different groups. One might even
find, as Ellis et al. [18] did, a stringently selected
clinical trial patient population of over 100 patients
with very few demonstrable NDF infections. In general
laboratory practice in our area, however, classically
confirmed NDF onychomycoses have been found to
make up about 3% of the total number of onychomy-
coses examined [10]. These data were obtained beforeNotes 1 and 2 were devised, and many cases for which
no second specimen had been obtained were excluded
from calculations of etiologic agent prevalence and
were considered to be merely ‘suggestive’ cases based
on initial specimen results. This may indicate that the
true proportion of NDF onychomycoses dealt with was
somewhat higher than that given in our 1989 study [10].
Also, the study in question [10] did not evaluate mixedinfections. A later study [30] facilitated such investiga-
tions by showing that sampling individual nails and
affected nail strata (subungual, superficial or, where
applicable, medial) separately greatly aids in clarifying
interpretation of mixed-infection results, as many
infecting NDF are confined to a single nail within a
single stratum, while other nails and strata may bear
other etiologic agents.The present study numerically confirms the long-
asserted judgment that simple association criteria have
unacceptable specificity in regard to NDF infections.
The great disparity of figures in the literature on
proportions of NDF onychomycosis (Table 1) is
primarily based on the ongoing use of simple-associa-
tion criteria by some authors. The liberal simple-
association criterion in particular greatly exaggeratesthe number of mixed dermatophyte/NDF infections
recognized, and studies employing this criterion can
often be immediately recognized in that numbers of
mixed infections will be nearly as large as, or even
greater than, one or both of the other major infection
categories, that is uncomplicated dermatophytosis and
uncomplicated NDF infection. Extension of the criter-
ion to incidental yeasts results in data showing largenumbers of spurious yeast infections. Delineating such
recognition criteria for these studies can be important,
as authors may state their methods and evaluation
criteria vaguely, leaving their methodology to be
partially inferred from the data collected. It may bethought to be self-evident that a criterion accepting
possible incidental NDF as etiologic would drastically
overestimate NDF infection numbers; however, the
ongoing appearance of studies using these criteria
shows that precise measurement of the degree of error
involved may be needed to convince the users of these
criteria, as well as reviewers and editors, that the
criteria are inadmissible.It should be pointed out that our study on mostly
geriatric patients to a considerable extent flatters the
simple association criteria, in that a reasonable propor-
tion of the NDF infections they recognize turn out to
be genuine on further analysis. In younger patient
populations, the overwhelming majority of NDF infec-
tions recognized by these criteria (e.g. infections judged
to be caused by NDF based on microscopic-positivespecimens yielding only an NDF in culture) would be
spurious.
The gold-standard data definitively resolve another
long-controversial area of onychomycosis, namely, the
matter of how frequently NDF infections are accom-
panied or facilitated by the presence of a dermatophyte.
As dermatophytes can indeed be isolated belatedly in
some mixed infections, after much sampling in whichonly the co-occurring NDF grows, the impression has
been given that perhaps an underlying dermatophyte is
always coextant in NDF infections. The well-known
Pseudeurotium ovale onychomycosis case of English et
al . [31], in which a single colony of a co-occurring T.
rubrum was isolated only after 200 nail fragments, 20
from each toe, had been plated out, may have appeared
to substantiate this impression. In fact, Table 6 clearlyshows that the isolation of dermatophytes from mixed
infections, like the isolation of dermatophytes from
uncomplicated infections, shows the approximate in-
verse-exponential distribution that is ubiquitously seen
in ecological sampling studies (see, for example, ref. 32).
By fortuity, in fact, the numbers of dermatophytes
isolated per sampling occasion in our data for derma-
tophytes from mixed infections falls off almost exactlyas powers of two. Our 17 mixed infections thus
represent all but a very small residual number of as
yet unelucidated dermatophytes growing in association
with NDF in the onychomycoses studied. The notion
that a significantly higher number of the remaining
NDF infections in our study coexist with a tinea
unguium can be excluded. Any impression that derma-
tophytes are always or usually present to facilitateNDF infections appears to be grossly erroneous.
The role of dermatophytes in facilitating mixed
infections can also be evaluated via studies in which
– 2005 ISHAM, Medical Mycology, 43, 39�/59
54 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
these infections are treated. Though no patients
inducted into the present study were treated, some
other patients seen concomitantly were treated, and
some were also repeatedly sampled. Isolations from
three such patients are shown in Table 9. These patients
were treated 12 weeks per os with an antifungal drug or
drugs that cannot be named due to clinical trial-related
confidentiality agreements. As can be seen, patient 1
had a mixed S. brevicaulis/T. rubrum infection,
although the former component was not elucidated
until after treatment had begun. As the other data
collected in the present study show, the failure of the S.
brevicaulis component to grow out in the initial speci-
men can probably be ascribed to the approximately
25% probability of this happening due to chance alone
rather than to the species being absent in the pre-
therapeutic baseline sample. (The sampling physician
may also have become more oriented to S. brevicaulis -
affected lesional areas when purely dermatophyte-
affected areas began to resolve.) After 8 weeks of
treatment, the dermatophyte was apparently eliminated
and did not recur in 6 months of post-treatment follow-
up, but S. brevicaulis continued to be isolated and the
nail continued to be clinically abnormal. Clearly S.
brevicaulis had no dependence on T. rubrum for
ongoing nail colonization. In patient 3, a T. rubrum
isolate occurring concomitantly with S. brevicaulis
apparently recrudesced after treatment, but during the
period in which it was under maximal drug inhibition
the S. brevicaulis component of the infection continued
to grow abundantly. In patient 2, the NDF component
of a Fusarium solani/T. rubrum onychomycosis con-
tinued to grow 3 months after the end of therapy, while
T. rubrum appeared to have been eliminated by therapy.
Though the F. solani did not grow in the final two
follow-up specimens taken in this case, the nail
remained clinically abnormal with no sign of resolution
of infection, and assessors of the case concurred that it
was most likely the two isolation failures and one
negative direct microscopy in the final samples were
chance effects rather than an indication that F. solani
had been eliminated. Unfortunately, the patient did not
appear for the scheduled final follow-up sample. These
three treatment studies, then, reinforce the idea that
many NDF involved in mixed dermatophytosis can
continue to prosper in nails from which the dermato-
phyte has been eliminated, or in which dermatophyte
growth is temporarily strongly inhibited by treatment
effects. This conclusion accords with the finding of
many dermatophyte-free NDF infections in the main
study presented here. Though the sample size involved
in these treatment studies is minimal, even a single
definite case of an NDF participant in a mixed
infection prospering in the long-term absence of its
former dermatophyte cohabitant is sufficient to con-
firm that this possibility is realistic.
Data from the present study on the number of
specimens required to grow the etiologic dermatophytes
from uncomplicated tinea unguium are almost identical
to those acquired previously by Gentles [7]. A relatively
surprising finding of our study, however, is that
etiologic NDF were equally likely to fail to grow out
in the first specimen, and to appear only in the second
or a later specimen. (Though the near-exact equality,
differing only in the second, insignificant decimal place,
seen in percentages of first specimen dermatophyte and
NDF outgrowth in the present study can only be
considered coincidental). Indeed, in all respects, NDF
infections proved just as difficult as dermatophytoses to
detect reliably, whether in direct microscopy or in
culture, even when they were demonstrably established
in longitudinal analysis. The physician submitting a
Table 9 Serial evaluations of fungal outgrowth and direct microscopy for nails with mixed-agent onychomycosis treated 12 weeks per os with
oral antifungal therapy
Sample Patient 1 Patient 2 Patient 3
Direct Culture Direct Culture Direct Culture
Pre-treatment �/ Trichophyton rubrum �/ T. rubrum �/ T. rubrum, S. brevicaulis
1 month on therapy �/ T. rubrum, Scopulariopsis brevicaulis �/ T. rubrum, Fusarium solani �/ S. brevicaulis
2 months on therapy �/ S. brevicaulis �/ F. solani �/ S. brevicaulis
3 months on therapy �/ S. brevicaulis �/ F. solani �/ S. brevicaulis
1 month post therapy �/ S. brevicaulis �/ F. solani �/ S. brevicaulis
2 months post therapy �/ S. brevicaulis �/ �/ �/ S. brevicaulis
3 months post therapy �/ S. brevicaulis �/ F. solani �/ S. brevicaulis
4 months post therapy �/ S. brevicaulis �/ �/ �/ T. rubrum, S. brevicaulis
5 months post therapy �/ S. brevicaulis �/ �/ �/ S. brevicaulis
6 months post therapy �/ S. brevicaulis ND ND �/ S. brevicaulis
ND, not done.
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 55
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
dermatologic mycology sample faces an unusually
difficult situation, in that as much as one-quarter ofthe single-specimen data returned by the laboratory,
though accurate (based on the samples submitted and
the use of recommended techniques) will ultimately
prove to have been incomplete or otherwise misleading.
In response to this difficulty, recent studies have
evaluated the sensitivity and overall diagnostic utility of
histopathological studies of nail biopsies vis-a-vis
conventional and modified KOH-and-culture techni-ques [25,33]. Though at least one small comparative
study has shown a significantly higher number of
positive results for histopathology than for KOH-and-
culture [33], no study so far has attempted to estimate
the overall sensitivity of histopathology against defini-
tive gold-standard data of the type used in the present
study. Also, the 28 cases in study [33] biopsy positive
for fungal hyphae yielded only 14 positive KOH resultsand seven cultures positive for nonyeast, noncontami-
nant fungi, including six dermatophytes and one
etiologically unconfirmed, unidentified Aspergillus.
Treating the biopsies positive for hyphae as signaling
known onychomycosis cases and tentatively accepting
the Aspergillus as causal, one finds that the 50% KOH-
positive rate and the 25% positive etiologic culture rate
are both highly significantly lower (P B/0.01, chi-squared) than parallel rates of over 70% positivity for
both individual techniques in the present study, sug-
gesting some methodological deficiency in the KOH
and culture techniques used.
The idea, advanced more or less tentatively by some
authors [25,33,34], that histopathological studies could
perhaps be used as a single method for the evaluation
of onychomycosis, is not to be recommended, in that itwould clearly lead to many NDF infections being
misdiagnosed as dermatophytoses. Though hyphae of
nondermatophytic fungi are often somewhat thicker,
thinner or more nodulose than those of dermatophytes
[23], it remains clear (contrary to assertions made by
Contet-Audonneau et al. [35] and others) that there is
as yet no truly reliable, objective way to distinguish
dermatophytic from most nondermatophytic elementsseen in histopathology conducted with the normally
used fungal stains (periodic acid-Schiff, Gomori methe-
namine silver). The accuracy of such histopathological
filament classification has never been rigorously tested
with either highly or normally experienced observers.
Our study shows that even the NDF most frequently
seen forming clearly nondermatophytic elements in nail
lesions, S. brevicaulis, only does so in a minority of thecases it causes (3/13 in the present study). With the
patient population seen in the present study, even if we
accorded histopathology a theoretical 100% sensitivity
for detecting fungal elements from all true cases in the
initial specimen, it would still, if it took all nondescriptfilaments to represent dermatophytosis, result in a
misdiagnosis of 17 mixed NDF infections, 64 of 67
pure NDF infections (assuming distinctive S.
brevicaulis elements were elucidated in three cases),
and both Scytalidium infections. Of our 263 patients
with filamentous fungal infection, 31.6% would be
incorrectly diagnosed. This is 26.7% of the whole
patient population; such a rate of misdiagnosis isscarcely acceptable. The same problem would occur if
the recently mooted technique of in-vivo confocal
microscopy [36] were used in the absence of culture to
diagnose presumptive dermatophyte onychomycosis.
Extended studies to determine if histopathology
partnered with culture, including follow-up of possible
NDF or mixed NDF patients for at least one additional
specimen, could significantly improve on conventionalKOH-and-culture would be of interest. Unfortunately,
in current literature, studies combining histopathology
with culture are just as likely as KOH-and-culture
studies to use dubious shortcuts for confirming NDF
etiology. Borkowski et al. [33], apparently adopting
liberal simple-association, accepted all outgrowths as
significant positive cultures, including a Verticillium (a
filamentous taxon with no valid pathogenic record inhuman dermatological mycology) from a nail biopsy
positive only for yeast elements, and a Candida as well
as a ‘Madurella ’ (a purely tropical group of nonspor-
ulating soil fungi not known to cause onychomycosis
and only technically recognizable either by sequencing
or by demonstrating black fungal grains in a case of
mycetoma) from two nails biopsy positive for derma-
tophyte-like, penetrating filaments. Lawry et al. [25], inattempting to confirm etiology for NDF growing in
initial culture from specimens histopathologically po-
sitive for fungal filaments, sought to elicit consistent
NDF outgrowth in a second culture trial done with
residual portions of the original specimen. This in
effect amounted to a single-specimen inoculum-count-
ing method with an even lower standard for acceptance
of etiology (two appearances of the fungus in culture)than the proven unreliable standard [21] of five culture-
positive inocula out of 20 planted as originally
proposed by Walshe and English [11]. Gupta et al.
[21] have shown that many NDF, especially members of
the heavily sporulating genera Aspergillus and Scopu-
lariopsis, can give high growth rates from single nail
specimens in which they are found purely as insignif-
icant contaminants deriving from the specimen (i.e. notderiving from transport materials or from process
contamination in laboratory workup). Splitting the
initial specimen into two portions and culturing the
– 2005 ISHAM, Medical Mycology, 43, 39�/59
56 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
second portion later is highly unlikely to alter this effect
due to transitory, but potentially heavy, body surfacecontamination. Needless to say, such practices are not
conducive to accurate comparative analyses of diag-
nostic methodologies.
Preliminary studies have suggested, again without
reference to gold-standard data, that PCR can improve
the rate of detection of dermatophytes from nail
specimens [37]. Unfortunately, because contaminating
molds can sometimes be present in large quantities onnails, direct PCR amplification with universal fungal
primers has no potential advantage in the investigation
of NDF infections: any attempt to make this applica-
tion would simply reinvent liberal simple-association.
Indeed, modern PCR techniques are so sensitive that
even one or a few contaminating spores of an NDF on
a nail could lead to a positive signal; the confusion
potentially caused by this amplification of contaminat-ing fungi has already been recognized in PCR studies of
suspected deep mycotic lesions and of fungi occurring
on nasal sinus membrane surfaces [38,39]. PCR results
for common environmental fungi detected on normally
unsterile body surfaces, including lesional surfaces, can
only be interpreted in the context of direct microscopy,
culture and other corroborating studies such as follow-
up specimens.The flow cytometry and differential immunohisto-
chemistry techniques investigated by Pierard et al . [20],
if they could be made practicable for routine or even
most reference laboratories, would have clear advan-
tages in sensitivity and first-specimen specificity over
existing routine methods. On the other hand, main-
taining immunohistochemical reagent stocks for all the
possible agents of NDF infection would be arduous,and it is clearly easier to initially detect which species is
involved in onychomycosis by culturing than by screen-
ing a panel of up to 35 or more histopathologic sections
with specific reagents for each possible etiologic fungus.
Of course, specific immunohistochemicals could be
used as a follow-up to culture to determine accurately
if filaments corresponding to a cultured NDF are
present in nail tissue. However, such a system wouldby no means solve all the problems intrinsic to correct
diagnosis of NDF infections: confusion could still be
caused by chance filament-negative samples from true
infections (in histopathologically studied nail dystrophy
patients who grew a dermatophyte in culture, ‘false’-
negative biopsies were found by Suarez et al. [40] in five
of 74 (6.8%) of cases and by chance-negative cultures as
well as wrong mold isolations as defined above. In flowcytometry, defining parameters for all possible agents
of infection would be a challenge. New agents of well-
verified onychomycosis are still being described [41]
and users of techniques such as immunohistochemistry
and flow cytometry must be aware that their judgmentcriteria should allow for the possibility of unknown
fungi, including some that might confound established
differentiating criteria (e.g. via antigenic cross-reac-
tion).
With regard to the dermatophyte components of
onychomycosis, it has been shown that most if not all
untreated cases of T. rubrum infection will show
demonstrable levels of the characteristic mycotoxinxanthomegnin in the nail tissue [42]. This technique
may warrant further investigation and extension to
additional dermatophyte species known to produce
xanthomegnin or closely related metabolites in vitro
[42].
The one modification of the classic system intro-
duced in the present study, namely, the use of Note 2,
increased the sensitivity of the method, allowing sevenmore NDF cases to be correctly recognized than would
have been confirmed otherwise. On the other hand, the
cultured NDF triggering use of this note turned out to
be insignificant in slightly more than half of the cases in
which the note would have been issued. This is in
contrast to Note 1, which in our study population
would have directly rewarded the labor of taking a
second patient sample 67.2% of the time (41 confirmedNDF cases/61 issuances). What must be borne in mind
in conjunction with these numbers is that because in
both our data and those of Gentles [7], around 15% of
causal dermatophytes are shown to be first grown in
the second specimen, and because adding in a second
specimen raises the accuracy of dermatophyte culture
alone from around 75 to around 90%, soliciting a
second specimen in some cases where an NDF grows,even in cases where the NDF turns out later to have
been insignificant, is a practice that in the aggregate
will improve the overall chances of correctly under-
standing the cases investigated. In the present study, the
issuance of Note 1 as per routine practice would have
resulted in the elucidation of four pure dermatophy-
toses that would not have been elucidated otherwise,
as well as the dermatophyte component of one mixedinfection. Use of Note 2 would also have yielded the
dermatophyte component of one mixed infection.
For practical purposes at the moment, it is clear that
despite its imperfections there is no known substitute
for classic methodology, inclusive of its handling of
NDF cases via soliciting of additional specimens
(though possibly direct microscopy based on use of
hydroxide with or without fluorescent fungal wallstains could be supplanted by histopathology). As
pointed out by Gupta et al. [21], the classic methodol-
ogy can also be enhanced with minimal extra trouble
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 57
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
and cost by the use of inoculum counting, not in the
manner originally recommended by Walshe and Eng-
lish [11], but rather as a supplementary study to place
probabilities on the likelihood that NDF isolated in
initial specimens will be confirmed as etiologic in
follow-up studies. A chart correlating positive inocu-
lum numbers with probabilities of etiologic significance
for NDF in general is available [21]. Although low
inoculum counts can never be used to rule out an NDF
infection, and though the Walshe and English 5/20 rule
(as detailed above) is highly unreliable [21], high counts
(e.g. 15 inocula positive for the same NDF out of 15
planted) in conjunction with an initial microscopic-
positive specimen can indicate a probability of up to
90% that the isolated NDF is etiologic. In such a case
involving high initial NDF counts, if the suspected
NDF agent were to be missed due to chance sampling
effects alone in a second, confirmatory sample, the
physician would know that a third sample should be
taken or, alternatively, given the original 90% chance of
an NDF infection vis-a-vis the approximately 11%
chance (as shown by the present study and Gentles
[7]) of a dermatophyte growing only from a third or
later specimen, could decide that the patient could
reasonably be treated prospectively as having an NDF
onychomycosis.
However, even when culturing efforts are vigorously
made, it should be recognized that physicians diagnos-
ing cases of onychomycosis are currently obliged to
work to some extent in the dark, knowing that
additional details about at least a few of the cases
they are dealing with could only be elucidated by
taking a number of samples that is not practical.
Therefore the breadth of the activity spectrum of drugs
used in therapy may be important to consider even in
seemingly straightforward cases, provided that this
consideration does not conflict with efficient elimina-
tion of the most prevalent etiologic agents, the derma-
tophytes. The prevailing drugs used in onychomycosis
change over time, as do perceptions of their activity
spectra, and the foregoing statement is not made in
reference to any particular drugs in current use.
Acknowledgements
The authors thank the staff of the Ontario Ministry of
Health mycology laboratory and of Mediprobe La-
boratories who participated in analyzing or collecting
samples in this study, and also thank the patients who
volunteered.
References
1 Summerbell RC. Nondermatophytic fungi causing onychomycosis
and tinea. In: Kane J, Summerbell RC, et al ., eds. Laboratory
Handbook of Dermatophytes. Belmont CA: Star Publishing, 1997:
213�/259.
2 Greer DL. Evolving role of nondermatophytes in onychomycosis.
Int J Dermatol 1995; 34: 521�/524.
3 Achten G, Wanet-Rouard J. Onychomycosis in the laboratory.
Mykosen 1978; 23(Suppl. 1): 125�/127.
4 Crespo Erchiga V, Casanas Carrillo C, Ojeda Martos A, et al .
Examen direct versus culture. Etude sur 1115 cas de dermatomy-
coses. J Mycol Med 1999; 9: 154�/157.
5 Davies RR. Mycological tests and onychomycosis. J Clin Pathol
1968; 21: 729�/730.
6 Midgley G, Moore MK, Cook C, Phan QG. Mycology of nail
disorders. J Am Acad Dermatol 1994; 31: S68�/S74.
7 Gentles JC. Laboratory investigations of dermatophyte infections
of nails. Sabouraudia 1971; 9: 149�/152.
8 Naka W, Hanyaku H, Tajima S, et al . Application of neutral red
staining for evaluation of the viability of dermatophytes and
Candida in human skin scales. J Med Vet Mycol 1994; 32: 31�/35.
9 DiSalvo AF, Fickling AM. A case of nondermatophytic toe
onychomycosis caused by Fusarium oxysporum . Arch Dermatol
1980; 116: 699�/700.
10 Summerbell RC, Kane J, Krajden S. Onychomycosis, tinea pedis
and tinea manuum caused by nondermatophytic fungi. Mycoses
1989; 32: 609�/619.
11 Walshe MM, English MP. Fungi in nails. Br J Dermatol 1966; 78:
198�/207.
12 English MP. Nails and fungi. Br J Dermatol 1976; 94: 697�/701.
13 Mercantini R, Marsella R, Moretto D. Onychomycosis in Rome,
Italy. Mycopathologia 1996; 136: 25�/32.
14 Scherer WP, McCreary JP, Hayes WW. The diagnosis of
onychomycosis in a geriatric population. A study of 450 cases in
south Florida. J Am Podiatr Med Assoc 2001; 91: 456�/464.
15 Budak A, Macura AB, Mazur T, Laskownicka Z. Fungal species
isolated from skin and nail lesions of hands and feet of patients
suspected of mycotic infection. Mykosen 1987; 30: 434�/439.
16 McGinnis MR, Rinaldi MG, Halde C. Mycotic flora of the
human toe web. Mykosen 1976; 19: 84�/86.
17 Zaias N, Oertel I, Elliott DF. Fungi in toe nails. J Invest Dermatol
1969; 53: 140�/142.
18 Ellis D, Watson AB, Marley JE, Williams TG. Nondermatophytes
in onychomycosis of the toenails. Br J Dermatol 1997; 13: 490�/
493.
19 Arrese JE, Pierard-Franchimont C, Greimers R, Pierard GE.
Fungi in onychomycosis. A study by immunohistochemistry and
dual flow cytometry. J Eur Acad Dermatol Venereol 1995; 4: 123�/
130.
20 Pierard GE, Arrese JE, De Doncker P, Pierard-Franchimont C.
Present and potential diagnostic techniques in onychomycosis. J
Am Acad Dermatol 1996; 34: 273�/277.
21 Gupta AK, Cooper EA, McDonald P, Summerbell RC. Inoculum
counting (Walshe/English criteria) in the clinical diagnosis of
onychomycosis caused by nondermatophytic filamentous fungi. J
Clin Microbiol 2000; 39: 2115�/2121.
22 Hull PR, Gupta AK, Summerbell RC. Onychomycosis: an
evaluation of three sampling methods. J Am Acad Dermatol
1998; 39: 1015�/1017.
23 Kane J, Summerbell RC. Dermatologic mycology: examination of
skin, nails and hair. In: Kane J, Summerbell RC, Sigler L, et al .,
– 2005 ISHAM, Medical Mycology, 43, 39�/59
58 Summerbell et al.
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.
eds. Laboratory Handbook of Dermatophytes. Belmont, CA: Star
Publishing, 1997: 33�/44.
24 Summerbell RC, Kane J. Physiological and other special tests for
identifying dermatophytes. In: Kane J, Summerbell RC, Sigler L,
et al ., eds. Laboratory Handbook of Dermatophytes. Belmont, CA:
Star Publishing, 1997: 45�/77.
25 Lawry MA, Haneke E, Strobeck K, et al . Methods for diagnosing
onychomycosis: a comparative study and review of the literature.
Arch Dermatol 2000; 136: 1112�/1116.
26 Gupta AK, Kohli Y. In vitro susceptibility testing of ciclopirox,
terbinafine, ketoconazole and itraconazole against dermatophytes
and nondermatophytes, and in vitro evaluation of combination
antifungal activity. Br J Dermatol 2003; 149: 296�/305.
27 Gupta AK, Ryder JE, Baran R, Summerbell RC. Nondermato-
phyte onychomycosis. Dermatol Clin 2003; 21: 257�/268.
28 De Hoog GS, Guarro J, Gene J, Figueras MJ. Atlas of Clinical
Fungi , 2nd edn. Utrecht: Centraalbureau voor Schimmelcultures,
2000.
29 English MP, Atkinson R. Onychomycosis in elderly chiropody
patients. Br J Dermatol 1974; 91: 67�/72.
30 Gupta AK, Summerbell RC. Combined distal and lateral
subungual and white superficial onychomycosis in the toenails
of an individual. J Am Acad Dermatol 1999; 41: 938�/944.
31 English MP, Harman RRM, Turvey JWJ. Pseudeurotium ovale in
toenails. Br J Dermatol 1967; 79: 553�/556.
32 Gochenaur SE. Fungi of a Long Island oak�/birch forest.
Population dynamics and hydrolase patterns for the soil penicillia.
Mycologia 1984; 76: 218�/231.
33 Borkowski P, Williams M, Holewinski J, Bakotic B. Onychomy-
cosis: an analysis of 50 cases and a comparison of diagnostic
techniques. J Am Podiatr Med Assoc 2001; 91: 351�/355.
34 Weinberg JM, Koestenblatt EK, Tutrone WD, et al . Comparison
of diagnostic methods in the evaluation of onychomycosis. J Am
Acad Dermatol 2003; 49: 193�/197.
35 Contet-Audonneau N, Salvini O, Basile AM, Percebois G.
Onychomycosis provoked by moulds. Importance of the biopsy
for diagnosis. Frequency of the pathogenic species. Sensitivity to
antifungal products. Nouv Dermatol 1995; 14: 330�/340.
36 Hongcharu W, Dwyer P, Gonzalez S, Anderson RR. Confirma-
tion of onychomycosis by in vivo confocal microscopy. J Am Acad
Dermatol 2000; 42: 214�/216.
37 Baek SC, Chae HJ, Houh D, et al . Detection and differentiation
of causative fungi of onychomycosis using PCR amplification and
restriction enzyme analysis. Int J Dermatol 1998; 37: 682�/686.
38 Buzina W, Braun H, Freudenschuss K, et al . Fungal biodiversity
�/ as found in nasal mucus. Med Mycol 2003; 41: 149�/161.
39 Hendolin PH, Paulin L, Koukila-Kahkola P, et al . Panfungal
PCR and multiplex liquid hybridization for detection of fungi in
tissue specimens. J Clin Microbiol 2000; 38: 4186�/4192.
40 Suarez SM, Silvers DN, Scher RK, et al . Histologic evaluation of
nail clippings for diagnosing onychomycosis. Arch Dermatol 1991;
127: 1517�/1519.
41 Gibas CF, Sigler L, Summerbell RC, et al . Arachnomyces kanei
(anamorph Onychocola kanei ) sp. nov., from human nails. Med
Mycol 2002; 40: 573�/580.
42 Gupta AK, Ahmad I, Borst I, Summerbell RC. Detection of
xanthomegnin in epidermal materials infected with Trichophyton
rubrum . J Invest Dermatol 2000; 115: 901�/905.
43 Clayton YM. Clinical and mycological diagnostic aspects of
onychomycoses and dermatomycoses. Clin Exp Dermatol 1992;
17(Suppl. 1): 37�/40.
– 2005 ISHAM, Medical Mycology, 43, 39�/59
Onychomycosis 59
Med
Myc
ol D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y M
ichi
gan
Uni
vers
ity o
n 11
/06/
14Fo
r pe
rson
al u
se o
nly.