Retrospective Cohort Analysis for Identification of Discordant Rifampicin-resistant Xpert MTB/RIF Assay Results in South Kivu, Eastern Democratic Republic of the Congo, a High Burden Tuberculosis Setting

BACKGROUND: The Xpert assay has revolutionized the rapid detection of resistance to rifampicin. However, Xpert has its pitfalls. We explored potential determinants of false-positive rifampicin resistance when using Xpert, aiming to refine the precision of tuberculosis diagnostics and subsequently contribute to better patient outcomes. METHODS: This is a retrospective cross-sectional analysis of archived Xpert files from the South Kivu province, used to diagnose Mycobacterium tuberculosis (MTB) between 2013 and 2018. Xpert cycle threshold was extracted for each molecular beacon probe, and ΔCt was calculated. We used the MTBDRplus line-probe assay, which covers the same 81-bp RRDR, as a reference test. RESULTS: Of 1900 samples positive for MTB, 220 (11.2%) were rifampicin resistant. Of the 141 patients' sputum samples that had results for both Xpert and MTBDRplus, 45 (31.9%) showed discordant results with Xpert, indicating rifampicin resistance while MTBDRplus indicated rifampicin susceptibility, suggesting false-positive rifampicin resistance detection by Xpert, predominantly in samples with very low (Ct > 28, odds ratio [OR] = 2.23, 95% CI: 1.30-3.82) or low (Ct 22-28, OR = 1.81, 95% CI: 1.21-2.71) bacterial loads. Probe E was the most frequently missed probe, followed by multiple probe dropouts or absence of probe binding (OR = 1.5, 95% CI: .731-3.076). CONCLUSIONS: Our findings indicate that low and very low MTB bacterial loads in sputum are strongly associated with discordant rifampicin resistance results when using Xpert. Further research into underlying mechanisms is needed to establish causality definitively.