Abstract
The GenoType MTBDRsl test and DNA sequencing were used to rapidly detect second-line drug- and ethambutol (EMB)-resistant Mycobacterium tuberculosis. The ability of these two assays to detect the presence of mutations associated with resistance to fluoroquinolones (FLQ), aminoglycosides/cyclic peptide (AG/CP), and EMB in the gyrA, rrs, and embB genes (for the GenoType MTBDRsl test) and gyrA, gyrB, rrs, eis, embC, embA, embB, and embR genes (for DNA sequencing) was compared to that of conventional agar proportion drug susceptibility testing (DST). We evaluated 234 multidrug-resistant (MDR) M. tuberculosis isolates. The two molecular methods had high levels of specificity (95.8 to 100%). The sensitivities for FLQ resistance detection for both methods were 85.1%. For AG (kanamycin [KM] and amikacin [AM]) and CP (capreomycin CAP]), the sensitivities of resistance detection using the GenoType MTBDRsl test were 43.2%, 84.2%, and 71.4%, respectively, while with the inclusion of an extra gene, eis, in sequencing, the sensitivity reached 70.3% for detection of KM resistance. The sensitivities of EMB resistance detection were 56.2% and 90.7% with the GenoType MTBDRsl test and sequencing, respectively. We found that the GenoType MTBDRsl test can rapidly detect resistance to FLQ, CAP, and AM. The accuracy of the GenoType MTBDRsl test for the detection of FLQ and AM resistance was comparable to that of conventional DST; however, the test was less accurate for the detection of KM and EMB resistance and demonstrated a poor predictive value for CAP resistance. We recommend including new alleles consisting of the eis promoter and embB genes in molecular analysis. However, conventional DST is necessary to rule out false-negative results from molecular assays.