Abstract
The study sought to determine the rate of discordant results between genotypic and phenotypic tests for the diagnosis of drug-resistant tuberculosis (DR-TB). Sputum samples and cultured isolates from suspected DR-TB patients were, respectively, analyzed for Mycobacterium tuberculosis by Xpert(®) MTB/RIF (Cepheid, Sunnyvale, CA, USA) and line probe assays (LPA) (Hain, Nehren, Germany). Discrepant rifampicin (RMP)-resistant results were confirmed using BACTEC MGIT960 (BD, New York, NY, USA). Of the 224 RMP-resistant results obtained by Xpert MTB/RIF, 5.4% were susceptible to RMP by LPA. MGIT960 showed a 75% agreement with LPA. The discrepancy was attributed to either heteroresistance or DNA contamination during LPA testing in 58.3% of cases. In 25% of the samples showing agreement in RMP resistance between Xpert MTB/RIF and MGIT960, the discrepancy was attributed to laboratory errors causing false RMP susceptible results with LPA. In 16.7% of the cases, the discrepancy was attributed to false RMP susceptible results with Xpert MTB/RIF. Out of the 224 isolates, susceptibility to isoniazid (INH) by LPA was performed in 73.7% RMP-resistant isolates, of which, 80.6% were resistant. All RMP-resistant isolates by Xpert MTB/RIF were confirmed in 98.5% by LPA if TB isolates were resistant to INH, but were only confirmed in 81.3% if TB isolates were susceptible to INH (p < 0.001). In conclusion, laboratory errors should be considered when investigating discordant results.