Tuberculosis is the leading cause of death among infectious diseases worldwide. Detection of Mycobacterium tuberculosis (Mtb), using routine culture-based methods is time consuming resulting in delayed diagnosis and poor treatment outcomes. Currently available molecular tests provide faster diagnosis but are able to screen only limited hot-spot mutations. Whole genome sequencing from direct sputum offers a potential solution, however, due to the presence of other microbes and host DNA its use in diagnostic testing remains challenging.
In this study, we present a targeted Mtb-enrichment assay for lineage-4 coupled with an improved analysis pipeline that uses 1657 bacterial taxa as background for reducing non-Mtb genome from sputum DNA. This method drastically improved the recovery of Mtb DNA from sputum (Mtb alignment increased from 3% to >65%) as compared to non-enrichment-based sequencing. We obtained >99% Mtb genome coverage as compared to 49% in non-enriched sputum sequencing. We were able to identify Mtb positive samples from controls with 100% accuracy using Mpt64 gene coverage. Our method not only achieved 100% sensitivity to resistance variants profiled by line probe assay (LPA), but also outperformed LPA in determining drug resistance based on phenotypic drug susceptibility tests for 6 anti-tuberculosis drugs (accuracy of 97.7% and 92.8% by enriched WGS and LPA, respectively).