Whole-genome sequencing for surveillance of tuberculosis drug resistance and determination of resistance level in China

Abstract

Objectives: Phenotypic drug susceptibility testing for prediction of tuberculosis (TB) drug resistance is slow and unreliable, limiting individualized therapy and monitoring of national TB data. Our study evaluated whole-genome sequencing (WGS) for its predictive accuracy, use in TB drug-resistance surveillance and ability to quantify the effects of resistance-associated mutations on MICs of anti-TB drugs.

Methods: We used WGS to measure the susceptibility of 4880 isolates to ten anti-TB drugs; for pyrazinamide, we used BACTEC MGIT 960. We determined the accuracy of WGS by comparing the prevalence of drug resistance, measured by WGS, with the true prevalence, determined by phenotypic susceptibility testing. We used the Student-Newman-Keuls test to confirm MIC differences of mutations.

Results: Resistance to isoniazid, rifampin and ethambutol was highly accurately predicted with at least 92.92% (95% confidence interval [CI], 88.19-97.65) sensitivity, resistance to pyrazinamide with 50.52% (95% CI, 40.57-60.47) sensitivity, and resistance to six second-line drugs with 85.05% (95% CI, 80.27-89.83) to 96.01% (95% CI, 93.89-98.13) sensitivity. The rpoB S450L, katG S315T and gyrA D94G mutations always confer high-level resistance, while rpoB L430P, rpoB L452P, fabG1 C-15T and embB G406S often confer low-level resistance or sub-epidemiological cutoff (ECOFF) MIC elevation.

Conclusion: WGS can predict phenotypic susceptibility with high accuracy and could be a valuable tool for drug-resistance surveillance and allow the detection of drug-resistance level; It can be an important approach in TB drug-resistance surveillance and for determining therapeutic schemes.

Keywords: Drug-resistance; Minimum inhibitory concentration; Surveillance; Tuberculosis; Whole-genome sequencing.