Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Webinar Series 2019Next-generation sequencing for drug-resistant TB
Webinar 1: Basic principles, practices and guidance5 February 2019
Dr Christopher Gilpin, PhD, MPH Senior Scientist
Diagnostics to support treatment of DR-TB:Next-generation sequencing technologies for the detection
of mutations associated with drug resistance in Mycobacterium tuberculosis complex.
Key changes to WHO guidelines on treatment and management of DR–TB
Diagnostics to support treatment of DR-TB
Basic principles of next generationsequencing
Overview of the accuracy of sequencing compared with phenotypic DST
Presentation Outline
WHO guidelines on treatment & management of drug-resistant TB, 1996 +
2016
2016
2017 2018
2019
ConsolidatedDR-TB
treatmentguidelines
prefinal texthttps://www.who.int/tb/publications/2018/WHO.2018.MDR-TB.Rx.Guidelines.prefinal.text.pdf
annexeshttps://www.who.int/tb/publications/2018/Annexes_8-10.pdf
“The new WHO recommendations, based on the most recent available evidence, signal an important departure from previous approaches to treat MDR/RR-TB. Injectable agents are no longer among the priority medicines when designing longer MDR-TB regimens and fully oral regimens should thus become the preferred option for most patients”
Dr Tereza KasaevaDirector of WHO’s Global TB Programme
GROUP MEDICINES
Group A Levofloxacin OR Moxifloxacin Lfx / MfxBedaquiline BdqLinezolid Lzd
Group B Clofazimine CfzCycloserine OR Terizidone Cs/ Trd
Group C Ethambutol EDelamanid DlmPyrazinamide ZImipenem-cilastatin OR Meropenem Ipm-Cln / MpmAmikacin (OR Streptomycin) Am (S)Ethionamide OR Prothionamide Eto / Ptop-aminosalicylic acid PAS
Longer MDR-TB regimensRevised classification of component medicines
Technical documents on diagnostics to support new treatment regimens
https://www.who.int/tb/areas-of-work/laboratory/policy_statements/en/
Technical Expert Consultation
In 2017, FIND conducted systematic review of available minimum inhibitory concentration (MIC) data for phenotypically wild type (pWT) strains, as well as genotypically non-wild type (gNWT) strains, including strains from allelic-exchange experiments for second-line antituberulosis medicines.
The medicines included in the review were the second-line injectable agents (kanamycin, amikacin and capreomycin); clofazimine and bedaquiline; cycloserine and terizidone; linezolid; delamanid; and the fluoroquinolones (ofloxacin, levofloxacin, gatifloxacin and moxifloxacin).
Media reviewed: Löwenstein-Jensen, Middlebrook 7H10/7H11; BACTEC™ Mycobacterial Growth Indicator Tube™ 960 (MGIT).
Critical concentrations (CC) and clinical breakpoints (CB) for medicines recommended for the treatment of RR-TB and MDR-TB
New Technical Guide
.
Next-generation sequencing (NGS) has great potential as a method for rapidly diagnosing drug-resistant TB (DR-TB) in diverse clinical reference laboratory settings worldwide
What technologies are available?
What is our current knowledge of the relationship between genotypic and phenotypic DST?
Targeted NGS vs whole genome sequencing (WGS)
Targeted next-generation sequencing (NGS) –NGS of specific genomic regions in a genome. Generally, targeted NGS is focused on sequencing a select set of genes or gene regions that have known or suspected associations with a specific pathogen (e.g. Mycobacterium tuberculosis) or a specific phenotype (e.g. drug resistance).
Whole genome sequencing (WGS) – The process of determining the complete genome sequence for a given organism at one time through NGS. This method can determine the order of all nucleotides in a given genome and detect any variations relative to a reference genome using bioinformatics analyses.
.
https://www.cdc.gov/pulsenet/pathogens/wgs.html
Next generation sequencing
Extracted genomic material for any sample must be of sufficient yield, purity and integrity to generate high quality sequencing data.
MTB has a complex cell wall that is difficult to lyse using conventional bacteriology methods,
Efficient extraction methods and DNA purification are crucial to ensure high quality DNA for NGS applications, especially WGS
The DNA extraction method should sufficiently weaken the cell wall, releasing the DNA, and allowing for effective DNA purification in following steps.
DNA extraction (1)
WGS applications rely upon an enrichment step to further promote sequencing success, either through MTB culture prior to DNA extraction or through additional processing, such as targeted enrichment steps, following extraction to ensure high DNA yield.
Targeted NGS, in contrast, is designed to amplify specific MTBC markers and resistance genes, therefore avoiding some of these issues by design. For this reason, targeted NGS applications may generate valid results even for DNA inputs of minimal quantity and quality.
DNA extraction (2)
DNA library, or a collection of specifically fragmented DNA to which oligonucleotide adaptors have been attached, must be prepared and quantified for both targeted NGS and WGS applications.
Kits meant for targeted NGS produce libraries that include specific, resistance-conferring genomic regions or lineage markers, while kits intended for WGS produce libraries comprised of fragments of the entire MTB genome.
NGS Library preparation (1)
Photo: Public Health Microbiology, Forensic and Scientific Services, Queensland Department of Health
The general steps of library preparation include DNA fragmentation, ligation to sequencing adapters, PCR, library purification and quantification mostly using enzyme-based methods for DNA fragmentation.
It is critical to check the quality and quantity of the DNA before and after library preparation. This assessment helps to avoid the waste of reagents and consumables during library preparation and sequencing when starting from a poor-quality DNA.
NGS Library preparation (2)
Photo: Public Health Microbiology, Forensic and Scientific Services, Queensland Department of Health.
Illumina MiSeqTM
Employs the ‘sequencing by synthesis’ approach, where a DNA polymerase synthesizes complementary strands to single-stranded DNA targets with synthesis halted by terminator nucleotides.
Capable of processing 1-96 samples per run, and can perform both targeted NGS and WGS applications.
Deeplex®-MycTB targeted NGS workflow GenoScreen (Lille, France) for the Illumina MiSeq.
ThermoFisher Scientific Ion Personal Genome MachineTM
Based on ‘sequencing by synthesis,’ using semiconductor sequencing technology to detect pH changes caused by hydrogen ions released during DNA synthesis.
Compared to Illumina sequencing, an Ion Torrent sequencing run is generally shorter (hours versus days) and the reads are longer.
Read coverage in AT- or GC-rich (as is the case for MTB) regions has been noted to be less efficient with this technology, compared to MiSeq.
Oxford Nanopore MinION
Photo: A handheld DNA sequencer being used in Madagascar to identify TB resistance in sputum samples Institut Pasteur de Madagascar
MinION is a small (90g, hand-held) device that may be connected to a computer through a USB.
Identifies DNA bases by measuring changes in electric conductivity generated as DNA strands pass through a biological nanopore.
Portability of the device (MinION), combined with real-time detection of signals as the sequencing takes place, makes this technology ideal for real-time diagnostics.
Higher cost per sample, but it offers longer reads compared with Illumina.
NGS data analysis (1)The computational resources required for generating, analyzing, storing and managing sequencing output data, including the bandwidth for file transferring if such analyses are outsourced, is one of the main bottlenecks in the successful introduction and implementation of NGS technologies in a given setting.
Photo: Public Health Microbiology, Forensic and Scientific Services, Queensland Department of Health.
NGS data analysis (2) Laboratories should note that the analysis and interpretation of
both targeted NGS and WGS data to detect TB drug resistance
directly from clinical samples requires a standardized, validated,
coordinated, and collaborative analytical approach.
Critical Path to TB Drug Regimens (CPTR) initiative, the Relational
Sequencing TB Data Platform (ReSeqTB). Platform being
relocated to WHO as ReSeqWHO.
For any bioinformatics pipeline for the analysis of TB sequencing
data, prospective users should also consider how they would
validate a given analysis.
Current analytical tools were developed based on the current
knowledge of the molecular basis of TB drug resistance.
Correlation of phenotypic and genotypic DST
WHO commissioned FIND to perform a systematic review of available MIC data for phenotypically wild type and non-wild type strains, including associated sequencing data for relevant resistance genes.
Collated data were used to calculate the frequency of each mutation in resistant and susceptible MTBC isolates and to derive a likelihood ratio and odds ratio for the association with phenotypic resistance.
Mutations were classified as having either high, moderate or minimal confidence for being associated with resistance, or as indeterminate or “not associated with resistance”
Confidence in mutations
Summary table of accuracy
Drug Genes Accuracy Comment
INH katG, inhA , mabA (fabG1)
ahpC-oxyR
Se: 84% / Sp: 98%
All mutations (including
not confidence-graded)
RIF rpoB Se: 96% / Sp: 99%
FQ gyrA, gyrB Se: 89% / Sp: 100%
PZA pncA Se: 83% / Sp: 98%
AMK eis, rrs Se: 79% / Sp: 100%
Strengths and Limitations
Sequencing is the most accurate molecular test available High throughput: up to ~ 200 strains per run/3-4 days Cheaper than standard phenotypic testing and increasingly
being used directly on specimens Systems open to adaption when new knowledge of resistance
emerges---------------------------------------------------------------------------------------- Molecular basis of resistance for the newer medicines not
completely determined (Bedaquiline, Clofazimine, Linezolid, Delamanid)
There is a need to better understanding the reasons for discordant results between phenotypic and genotypic DST results including
The relative contribution of hetero-resistance The importance of efflux pumps or compensatory mechanisms
Next generation sequencing
Currently applicable for use as a tool for the surveillance of DR-TB with the potential application as a diagnostic tool.
Development of the WHO technical guide presents current knowledge of the correlation of certain mutations with phenotypic DST. With broader implementation of NGS in low- and middle-income countries this knowledge base will continue grow and our understanding will be refined.
Targeted sequencing and WGS approaches
Accuracy of DNA sequencing compared to phenotypic DST - isoniazid
The high-confidence katG S315 and inhA c-15t double mutants.
Using only the high, medium and low confidence-graded mutations as predictors of phenotypic resistance the sensitivity and specificity for INH resistance detection were estimated to be 78% (95% CI 77.6-79.1%) and 100.0% (95% CI 99.7-100.0%), respectively.
Estimate excludes mutations associated with phenotypic INH resistance that could not be confidence-graded due to insufficient data.
If these mutations were also considered in addition to the confidence-graded mutations, the sensitivity and specificity for INH resistance detection would be 84% (95% CI 82.9-84.3%) and 98%(95% CI 96.8-98.4%), respectively.
Accuracy of DNA sequencing compared to phenotypic DST - rifampicin
Using only the high, medium and low confidence-graded mutations as predictors of phenotypic resistance the sensitivity and specificity for RIF resistance detection were estimated to be 91% (95% CI 89.8-91.1%) and 100% (95% CI 99.9-100.0%), respectively.
Estimate excludes mutations associated with phenotypic RIF resistance that could not be confidence-graded due to insufficient data.
If these mutations were also considered in addition to the confidence-graded mutations, the sensitivity and specificity for RIF resistance detection would be 96% (95% CI 95.2-96.1%) and 99%(95% CI 98.1-98.8%), respectively.
Accuracy of DNA sequencing compared to phenotypic DST - fluoroquinolones
The sensitivity and specificity of gyrA mutations for predicting phenotypic resistance to FQs in general, and MFX specifically, has been underestimated in past reports due to phenotypic reference standard critical concentrations that were set too high.
These have recently been corrected and will likely increase the accuracy of these mutations as predictors of phenotypic resistance.
Using only the high, medium and low confidence-graded mutations as predictors of phenotypic resistance the sensitivity and specificity for MFX resistance detection were estimated to be 89%(95% CI 86.4-91.7%) and 100% (95% CI 99.2-100.0%), respectively.
Accuracy of DNA sequencing compared to phenotypic DST - pyrazinamide
The known resistance-conferring mutations are diverse and distributed over several hundred base pairs of the pncA gene and its promoter.
This gene region also includes many mutations that do not confer PZA resistance.
When all the mutations associated with resistant phenotypes were considered the sensitivity and specificity for PZA resistance detection were 83% (95% CI 81.7-84.5%) and 94% (95% CI 92.9-94.7%), respectively.
Manually curating the mutation lists the sensitivity and specificity for PZA resistance detection would be 83% (95% CI 81.9-84.7%) and 98%(95% CI 96.9-98.1%).
The reduced sensitivity of sequencing of the pncA gene for the detection of PZA resistance is likely due to the limited reproducibility of the phenotypic test.
• .
Accuracy of DNA sequencing compared to phenotypic DST - Amikacin
Using only the high, medium and low confidence-graded mutations as predictors of phenotypic resistance the sensitivity and specificity for AMK resistance detection were estimated to be 80% (95% CI 76.7-82.3%) and 98% (95% CI 96.8-98.4%), respectively.
However, this estimate excludes mutations associated with phenotypic AMK resistance that could not be confidence-graded due to insufficient data.
If these mutations were also considered in addition to the confidence-graded mutations, the sensitivity and specificity for AMK resistance detection would be 79% (95% CI 75.8-81.5%) and 100%(95% CI 99.7-100.0%), respectively.
Three most effective agents grouped together and strongly recommended for all patients.
Most patients may start treatment with at least 4 agents likely to be effective and continue with 3 once bedaquiline is stopped.
An all-oral regimen is possible in most patients.
Amikacin and streptomycin lower down in priority ranking (if DST is negative and audiometry possible).
Kanamycin and capreomycin linked to poorer outcomes in IPD meta-analysis and no longer recommended.
Longer MDR-TB regimens (1)Significant changes
Longer MDR-TB regimensRegimen composition
Choice of medicines depends upon the: • expected balance of effectiveness and harms• preference for oral over injectable agents• the results of drug-susceptibility testing (DST)• reliability of DST methods• population drug resistance levels• history of previous use of medicine in a patient• drug tolerability• potential drug-drug interactions
• Programmes already implementing the shorter MDR-TB regimen with good results and capacity to monitor for ototoxicity • switch from Km to Am• while Km is used close follow-up for non-response to
treatment or early relapse• apply a low threshold to switch non responders to a new
longer regimen
• Programmes planning to offer the shorter regimen to newly-diagnosed patients are advised to continue only if they have DST capacity to exclude at least FQ and SLI resistance
• Programmes considering modified shorter regimens (e.g. replacing injectable with BDQ) can do so as operational research
Shorter MDR-TB regimen