Can whole genome sequencing replace AST? Matthew J Ellington Antimicrobial Resistance & Healthcare Associated Infections (AMRHAI) Reference Unit Crown copyright
EUCAST Subcommittee on the role of whole genome sequencing (WGS) in AST of bacteria 1. Review literature describing the role of WGS in AST of bacteria 2. Assess the sensitivity and specificity of WGS vs phenotypic AST 3. Consider how WGS for AST may be applied in clinical micro labs Likely implications for pheno and other geno methods in use 4. Consider the epidemiological implications of using WGS 5. Consider the clinical implications of WGS for the selection of R x 6. Consider the principles of how the results of WGS for AST could be presented to clinical users to describe the drivers and barriers to routine use of WGS Report out for consultation http://www.eucast.org/documents/consultations/ 2 RaMi-NGS, 10 th June 2016 Crown Copyright
What can WGS offer? Measures susceptibility Resistance mechanisms ECOFF (WT vs. non-wt) Clinical resistance (S vs. R) Additional data Suitable speed Cost Phenotypic AST (limited) (most) (e.g. TB) WGS-based AST? (must be inferred) (most) (e.g. TB) 3 RaMi-NGS, 10 th June 2016 Crown Copyright
Most appropriate AST comparators What criteria should WGS data be assessed against? clinical breakpoints indicate likelihood of therapeutic success (S) or failure (R) of antibiotic treatment based on microbiological findings ECOFFs (epidemiological cut-off values) differentiate wild-type (WT) from nonwild-type (NWT) isolates with an acquired resistance mechanism 4 RaMi-NGS, 10 th June 2016 Crown Copyright
Focus on WHO priority organisms Organisms Enterobacteriaceae E. coli 3GC, FQs Priority resistances K. pneumoniae 3GC, carbapenems Non-typhoidal Salmonella Shigella spp. FQs FQs S. aureus - MRSA S. pneumoniae - Penicillin N. gonorrhoeae - 3GCs Plus others M. tuberculosis, C. difficile, A. baumannii, P. aeruginosa 5 RaMi-NGS, 10 th June 2016 Crown Copyright
A growing literature 501 isolates; S. aureus; 5112 143 AST isolates;2 results; 98.8% species; WGS 1001 concordance AST results; 96.7% WGS concordance 388 isolates; 1 species; 1158 AST results; 88.9% WGS concordance 6 RaMi-NGS, 10 th June 2016 Crown Copyright
Evidence reports e.g. Enterobacteriaceae The relatively limited number of acquired resistance genes and resistanceassociated mutations that dominate epidemiologically in the Enterobacteriaceae High levels of accuracy of genotype-phenotype correlation in published studies; means that well-informed screening approaches can be very accurate. Predicting AST results will be harder for some than for others to improve the levels of accuracy across large genetically diverse datasets we need to understand more of: The full range of mechanisms + better detection of unknowns (e.g. mcr-1) Interplay between mechanisms 7 RaMi-NGS, 10 th June 2016 Crown Copyright
Complex interplays determine an MIC External [drug] V Entry + V Efflux Periplasmic [drug] V Hydrolysis V Binding It s a lot more complex than gene presence / absence 8 RaMi-NGS, 10 th June 2016 Crown Copyright
Combinatorial resistance: WGS vs. AST 9 RaMi-NGS, 10 th June 2016 Crown Copyright Reuter, Ellington, Cartwright et al., 2013. JAMA Intern Med 12;173:1397-404
Non-organism based factors (1) Data quality Before WGS can be routinely implemented into accredited clinical practice there is a need to establish necessary minimum QC-thresholds Currently there are no international QC standards GMI driving WGS proficiency testing project with 50 laboratories worldwide Such initiatives are important first steps towards setting QC thresholds 10 RaMi-NGS, 10 th June 2016 Crown Copyright
Non-organism based factors (2) AMR gene database: A single, standardised reference db Need better standardisation of annotation of AMR genes Will help improve accuracy in public databases Need a single, regularly updated challenge database containing all validated AMR genes and chromosomal SNPs linked with AMR Minimum standards for including new determinants in the standard db Need international consensus on criteria to define novel or variant genes Inextricably linked to issues of gene nomenclature 11 RaMi-NGS, 10 th June 2016 Crown Copyright
Non-organism based factors (3) Systematic sources of error for pheno / geno correlation Flaws with phenotypic AST An inadequate limit of detection of WGS i.e. when detection is direct from clinical specimens e.g. TB Incomplete knowledge base for genotypic basis of phenotypic resistance Affects sensitivity of WGS prediction (resulting in very major errors) In development phase - anticipate many gaps in the knowledge-base problematic bacteria; problematic antibiotics 12 RaMi-NGS, 10 th June 2016 Crown Copyright
Impacts (incl. clinical) of WGS-based genotypic AST Further evidence needed, but could soon replace much surveillance AST low impact of the low error rate Further evidence needed, but could soon reduce AST tests in reference labs unless to guide treatment for agents with poorest genotypic/phenotypic concordance comparative in-vitro activity of new agents longer for a paradigm shift to WGS to guide clinical decision making very major errors - gene absence cannot always predict susceptibility robust evidence will be needed surveillance of treatment failure +/- novel resistance mechanisms education needed and major behavioural change for prescribers 13 RaMi-NGS, 10 th June 2016 Crown Copyright
Concluding comments Currently: insufficient data to present a definitive document on the topic Review describes current state-of-the art and will require updates Quantity / quality of evidence for pheno/geno concordance must improve An MIC reflects more than gene presence / absence: ECOFFs first comparator, breakpoints will be tougher but are necessary Algorithms vary, but for comparisons they: should use the same centralised database of loci should have been calibrated and shown equivalent Advocating a global standard method is unrealistic at this time Learn lessons from the past and avoid multiple parallel routine systems Balance need to standardise with academic drivers for bioinformaticians to develop and improve their own tools 14 RaMi-NGS, 10 th June 2016 Crown Copyright
Acknowledgements & live consulatation Gunnar Kahlmeter and Derek Brown for setting the challenge Committee members for excellent contributions and discussions: Frank M. Aarestrup (Denmark) Rafael Canton (Spain) Michel Doumith (UK) Oskar Ekelund (Sweden, co-editor) Christian Giske (Sweden) Hajo Grundman (Germany) Henrik Hasman (Denmark) Katie L. Hopkins (UK) Matt Holden (UK) Jon Iredell (Australia) Gunnar Kahlmeter (Sweden) Claudio U. Koser (UK) Alasdair MacGowan (UK) Dik Mevius (Netherlands) Mike Mulvey (Canada) Thierry Naas (France) Tim Peto (UK) Jean-Marc Rolain (France) Ørjan Samuelsen (Norway) Neil Woodford (UK, Chair) Report for comment: http://www.eucast.org/documents/consultations/ 15 RaMi-NGS, 10 th June 2016 Crown Copyright