Discordance of the Repeat GeneXpert MTB/RIF test for Rifampicin Resistance Detection among Patients Initiating MDR-TB Treatment in Uganda

We recruited all patients with Xpert Rifampicin-Resistant (RR)-TB detected, referred to the MDR-TB ward at Mulago National Referral Hospital, Kampala, Uganda for MDR-TB treatment initiation between September 2017 and October 2019. Using baseline samples collected for patients screened for STREAM 2 trial, smear microscopy, repeat Xpert test and first-line MTBDRplus assay were done. Culture-based drug-susceptibility testing was done on discordant samples. We analysed for factors associated with discordant results and false RR as determined as no RR detected by at least two of the methods used (reference standard). Of 126/130 patients who had results of repeat Xpert, 97 (77.0%) had M. tuberculosis detected of which 81 (83.5%) had RR-detected, 1 (1.0%) indeterminate. A total of 10/96 (10.4%) patients were rifampicin susceptible by at least two of the methods. Having false Xpert RR was associated with low bacillary load measured by high cycle threshold (Ct) value i.e. low (Ct 22–28) and very low (Ct > 28) of the initial Xpert test 0.09 (0.05; 0.01–1.08) and 0.02 (0.01; 0.01–0.35) respectively. Our results show that a repeat Xpert test on another sputum sample for patients with initial low M. tuberculosis detected RR-detected, would exclude 10% of the TB patients from unnecessary MDR-TB treatment initiation.