Discordance across Phenotypic and Molecular Methods for Drug Susceptibility Testing of Drug-Resistant Mycobacterium tuberculosis Isolates in a Low TB Incidence Country

With increasing incidence of multidrug-resistant tuberculosis (MDR-TB), accurate drug susceptibility testing (DST) of Mycobacterium tuberculosis to first-line drugs has become crucial for proper patient management. We evaluated concordance of DST results for 70 M. tuberculosis isolates across two phenotypic and two molecular methods: BACTEC 460TB, MGIT 960 system, GenoType MTBDRplus and DNA sequencing of gene segments most commonly implicated in conferring resistance to anti-TB drugs. Most (84%) M. tuberculosis isolates were multidrug-resistant. Twenty-four isolates yielded discrepant DST results. For rifampicin, isoniazid and streptomycin, 96%, 97% and 93% of isolates, respectively, were susceptible or resistant by all four methods, whereas for ethambutol, this agreement was observed for only 76% of isolates (P<0.05 for rifampicin or isoniazid or streptomycin versus ethambutol). Occurrence of rare mutations in three isolates that confer low-level resistance caused lower agreement for rifampicin among the four methods (kappa coefficient (κ) range, 0.84 to 0.95). For isoniazid, there was perfect agreement among phenotypic methods and molecular methods (κ, 1.00) but lower agreement between phenotypic and molecular methods. Three isolates were detected as polydrug-resistant by MGIT 960 system but as multidrug-resistant by DNA sequence-based method. The agreement was higher for streptomycin among the two phenotypic methods (κ, 0.97) while targeted sequencing yielded lower agreement (κ range, 0.86 to 0.89). The discrepancy for ethambutol resulted largely due to lower concordance of MGIT 960 results (κ range, 0.53 to 0.64). The MGIT 960 system is an accurate method for DST of M. tuberculosis against isoniazid and streptomycin while the results of rifampicin susceptibility should be complemented with DNA sequencing-based method when the suspicion for resistance is high. The possibility of false susceptibility to ethambutol with MGIT 960 system suggests that molecular or other phenotypic methods may be more useful when accurate ethambutol susceptibility results are warranted.