In this study we investigated how rapid NGS in conjunction with targeted amplification of genomic regions of interest could be utilised as a potential diagnostic to identify drug resistant tuberculosis in clinical samples.

Phenotypic drug susceptibility testing (DST) for tuberculosis (TB) requires weeks to yield results. Although molecular tests rapidly detect drug resistance-associated mutations (DRMs), they are not scalable to cover the full genome and the many DRMs that can predict resistance. Whole-genome sequencing (WGS) methods are scalable, but if conducted directly on sputum, typically require a target enrichment step, such as nucleic acid amplification. We developed a targeted isothermal amplification-nanopore sequencing workflow for rapid prediction of drug resistance of TB isolates. We used amplification by consensus (ABC) to perform targeted isothermal amplification of three regions within the M. tuberculosis genome (37°C, 90 minutes), followed by nanopore sequencing on the MinION. We tested 29 Mycobacterial genomic DNA extracts from patients with DR-TB and compared our results to WGS by Illumina and phenotypic DST to evaluate the accuracy of resistance to rifampicin and isoniazid prediction. Amplification by ABC showed equivalent fidelity compared to high fidelity PCR (100% concordance). Nanopore sequencing generated identical DRM predictions to WGS, with considerably faster sequencing run times of minutes rather than days. Sensitivity and specificity of rifampicin resistance prediction for our workflow were 96.3% (95% CI 81.0 – 99.9%) and 100.0% (95% CI 15.8 – 100.0%), respectively. For isoniazid resistance prediction, sensitivity and specificity were 100.0% (95% CI 86.3 – 100.0%) and 100.0% (95% CI 39.8 – 100.0%) respectively. The workflow consumable costs per sample is less than £100. Our rapid and low-cost drug resistance genotyping workflow provides accurate prediction of rifampicin and isoniazid resistance, making it appropriate for use in resource-limited settings.