Automated broth-based systems versus the MYCOTB plate for antimicrobial susceptibility testing of the Mycobacterium tuberculosis complex: challenges in interpretation

Abstract We examined categorical agreement between automated mycobacterial susceptibility testing methods (Mycobacterial Growth Indicator Tube [MGIT] 960 System and the VersaTREK Mycobacteria Detection and Susceptibility System) which are based on single critical concentration (CC) "breakpoints" and a commercial microbroth dilution method (Sensititre Mycobacterium tuberculosis MIC Plate [MYCOTB]) which provides an MIC value. Mycobacterium tuberculosis isolates (n=355) were tested against three first-line antimycobacterial agents (ethambutol [EMB], isoniazid [INH], rifampin [RIF]) using the MYCOTB plate and either the MGIT 960 (site 1, n=142) or VersaTREK (site 2, n=213) systems. Overall categorical agreement was 96.8%. When stratified by drug and CC-defined susceptible and resistant isolates, concordance ranged from 75% to 100%. Interpretation of MIC-based results versus established CC-based results was challenging for drugs whose CC was not represented by an exactly equivalent concentration in the manufacturer-defined dilutions on the MYCOTB plate (EMB, INH). We propose interpretations of MYCOTB plate MICs using the currently available plate configuration.