Bringing TB genomics to the clinic: A comprehensive pipeline to predict antimicrobial susceptibility from genomic data, validated and accredited to ISO standards.

Abstract

BackgroundWhole genome sequencing (WGS) is increasingly contributing to the clinical management of tuberculosis. Whilst the availability of bioinformatic tools for analysis and clinical reporting of Mycobacterium tuberculosis sequence data is improving, However, there remains a need for accessible, flexible bioinformatic tools that can be easily tailored for clinical reporting needs in different settings and are suitable for accreditation to international standards. MethodsWe developed tbtAMR, a flexible yet comprehensive tool for analysis of Mycobacterium tuberculosis genomic data, including inference of phenotypic susceptibility and lineage calling. Validation was undertaken using local and publicly-available real-world data (phenotype and genotype) and synthetic genomic data to determine the appropriate quality control metrics and extensively validate the pipeline for clinical use. FindingstbtAMR accurately predicted lineages and phenotypic susceptibility for first- and second-line drugs, with equivalent computational and predictive performance compared to other bioinformatics tools currently available. tbtAMR is flexible with modifiable criteria to tailor results to users needs. InterpretationThe tbtAMR tool is suitable for use in clinical and public health microbiology laboratory settings, and can be tailored to specific local needs by non-programmers. We have accredited this tool to ISO standards in our laboratory, and it has been implemented for routine reporting of AMR from genomic sequence data in a clinically relevant timeframe. Reporting templates, validation methods and datasets are provided to offer a pathway for laboratories to adopt and seek their own accreditation for this critical test, to improve the management of tuberculosis globally. FundingVictorian Government Department of Health; Australian National Health and Medical Research Council and Medical Research Futures Fund. Research in contextO_ST_ABSEvidence before this studyC_ST_ABSWe searched PubMed for studies using the search terms: "Mycobacterium tuberculosis", "clinical", "bioinformatics", "genomics", "drug resistance (OR antimicrobial)", published prior to 31st July 2024 without language restrictions (n=258). We considered all studies from this search that used genomics to infer likely drug-resistance in M. tuberculosis. Many of these studies highlight the challenges in making meaningful interpretations from genomic data for the purpose of inferring AMR for clinical applications. Despite the development of bioinformatics tools and compilation of catalogues of resistance conferring mutations, few of these studies directly address the challenges specific to implementation of a whole genome sequencing (WGS) and bioinformatics pipelines to deliver validated and accredited results for use in clinical applications and none provide sustainable solutions. Added value of this studyTo the best of our knowledge, this study is the first to detail practical solutions to challenges to validating and accrediting the routine inference of AMR from WGS data for clinical applications for treatment of M. tuberculosis. We developed, validated and accredited a bioinformatics tool, tbtAMR, which is robust and flexible, using a data-driven approach to inferring resistance, allowing for simple customisation of behaviour on a large collection of in-house data supplemented with publicly available data. Furthermore, we have made this pipeline and the accompanying data available for use by the wider public health community to aid others in implementing similar programs. Consultation with infectious disease clinicians, microbiologists and epidemiologists has allowed us to develop a robust workflow, that encompasses sequencing, analysis, interpretation, reporting, discrepancy resolution and ongoing verification. Implications of all the available evidenceThis study addressed the specific needs of reporting genomic AMR for M. tuberculosis in a clinical and public health laboratory. Furthermore, this work makes available a pipeline and data for laboratories to utilise to implement genomic AMR for M. tuberculosis. Through the use and further development of publicly available and open-source bioinformatics software, we have demonstrated the feasibility of clinical reporting of genomic AMR results for M. tuberculosis in a low-incidence high-income setting. We believe that this work lays the foundation for making reporting of genomic AMR for M. tuberculosis more accessible to public health and reference laboratories in general.