Whole genome sequencing in drug susceptibility testing of Mycobacterium tuberculosis in routine practice in Lyon, France.

Abstract

Rapid and correct determination of Mycobacterium tuberculosis (MTB) drug-susceptibility is a challenge for tuberculosis management. Phenotypic drug-susceptibility testing (DST) remains the reference method but is time consuming. Herein, we compared the genotypic prediction of the first-line drug susceptibility profile obtained by whole-genome sequencing (WGS) to that obtained by phenotypic DST and by line-probe assay (LPA).All MTB strains isolated from patients during routine practice at the mycobacteria laboratory of the Lyon University Hospital, France, between November 2016 and July 2019 were included (n=274). These were tested for the first-line drugs using phenotypic DST (mycobacteria-growth-indicator-tube assay), and for genotypic prediction of the susceptibility profile with LPA and WGS.Considering phenotypic DST as reference, WGS predicted resistance to rifampicin, isoniazid, ethambutol, and pyrazinamide with sensitivities of 100%, 100%, 100% and 93.8%, respectively, and susceptibility to these drugs with specificities of 99.6%, 100%, 98.5% and 100%, respectively. The performance of LPA was poorer; sensitivity was 83.3% for rifampicin and 85.7% for isoniazid susceptibility. Five isolates were classified as susceptible according to phenotypic DST (1 for rifampicin, 4 for ethambutol), while WGS detected resistance mutations in rpoB and embB genes.WGS, used under appropriate conditions of quality control, has good performance to predict the resistance profile for the 4 first-line drugs, and corrects the phenotypic DST. This study highlights the need for future guidelines recommending WGS as the initial tool in routine practice in areas where the prevalence of TB and drug-resistant MTB are low.