[A study on inoculum density and reproducibility of drug susceptibility testing by BACTEC MGIT 960]

The BACTEC MGIT 960 drug susceptibility system (MGIT AST) has been recently introduced in Japan. The issue of discordant MGIT results compared with the conventionally used Ogawa method has been raised. It has been speculated that discordant results might be due to MGIT inoculum density since there is no standardization step other than dilution of growth for tubes beyond 2 days after MGIT turns out to be positive. In this study, we examined the reproducibility of the MGIT AST system.

Nineteen sputum specimens from drug-resistant and susceptible pulmonary tuberculosis patients were processed with CCE pretreatment reagent (Japan BCG), inoculated into 3 MGIT tubes, and loaded into the MGIT 960. Inocula for MGIT AST were prepared 1, 3, and 5 days after MGIT tubes became positive. Cultures on day 3 and 5 were diluted 1: 5 with saline. Ten-fold dilutions from each positive culture were plated on Middlebrook 7H11 agar plates for CFU determination. MGIT AST results were compared with those of the conventional proportion method on Ogawa egg and Vite-spectrum (Kyokuto), or Pyrazinamidase (Pzase) assay and Kyokuto PZA test.

A total of 15 specimens were culture positive in all 3 tubes. Four of 19 cases were removed from the analysis because of negative cultures in one or more tubes. Three of 4 culture negative cases were MDR-TB. Colony counting showed the mean CFU/ml of inocula prepared from tubes 1, 3, and 5 days after MGIT tube became positive were 3.6 x 10(6), 1.6 x 10(6), 3.1 x 10(6), respectively. There was no significant difference although the CFU range was wide (8 x 10(4)-2 x 10(7)). MGIT AST results were consistent among 3 inocula. Moreover, overall concordance rates between MGIT AST and the conventional methods were over 90% for 5 first-line antituberculosis drugs. These results indicate that the BACTEC MGIT 960 system is very useful for rapid diagnosis of drug resistant tuberculosis.