Discrepancy between WGS-based genotypic and phenotypic drug susceptibility testing of first and second-line drugs for Mycobacterium tuberculosis: Limited impact of unknown resistance mechanisms.

OBJECTIVES

Accurate detection of Mycobacterium tuberculosis (MTB) drug susceptibility is critical for developing effective treatment regimens. Whole genome sequencing (WGS) based genotypic drug susceptibility testing (gDST) can comprehensively predict MTB drug resistance, but inconsistencies between gDST and phenotypic drug susceptibility tests (pDST) pose a problem for determining drug resistance.

METHODS

To investigate the factors underlying the discrepancies, we tested 509 clinical MTB strains against nine anti-TB drugs. Minimum inhibitory concentration assays and proportional method determined pDST, while WGS-based methods - WHO catalogue (2nd edition), TB-profiler, SAM-TB, Gen-TB and MDCNN - determined gDST.

RESULTS

There were 125 discordant results between the pDST and the 2nd edition WHO catalog-based gDST. After verifying the pDST results, 13.6% (17/125) of the discordant results could be attributed to errors in pDST. Among the remaining 108 discrepancies, 32.4% (35/108) were pDST-R/gDST-S while 67.6% (73/108) were pDST-S/gDST-R. Of the 35 pDST-R/gDST-S discordant results, 51.4% (18/35) were identified as drug resistance using other gDST tools, 28.6% (10/35) were found to be hetero-resistant, and only 20.0% (7/35) were possibly due to unknown drug resistance mechanisms. 87.7% (64/73) of inconsistent pDST-S/gDST-R results concerned ETO, EMB and STR. Based on the verified pDST results, the WHO catalogue-based gDST method has an average sensitivity of 96.8% and specificity of 98.1%.

CONCLUSION

Our results confirm that the WHO catalogue-based gDST method is a reliable alternative to pDST. Unknown drug resistance mechanisms were not the main cause of pDST-R/gDST-S; novel determinants, other than DRMs, affect drug susceptibility and may explain pDST-S/gDST-R discrepancies.