Rifampicin monoresistant tuberculosis: Exploring the interplay with HIV and mortality: An individual participant data meta-analysis

Multidrug-resistant or rifampicin-resistant tuberculosis (MDR/RR-TB) poses a significant global public health challenge. While most people with RR-TB have MDR-TB (resistance to isoniazid and rifampicin), rifampicin mono-resistant tuberculosis (RmR-TB, rifampicin resistance, isoniazid susceptibility) is prevalent in certain regions. We aimed to compare mortality between patients with RmR-TB and MDR-TB and assess potential effect modification by HIV status during treatment. We conducted an individual patient data meta-analysis involving 16,651 individuals with confirmed drug-resistant tuberculosis, all of whom had drug susceptibility test results for at least rifampicin, isoniazid, and fluoroquinolones and had initiated TB treatment. Mixed-effects logistic regression was used to estimate mortality during two-year treatment, adjusting for age, sex, history of TB, disease site, and HIV status. Stratified analyses explored effect modification by HIV status. Among 16,568 patients, 2878 (17.4%) were classified as having RmR-TB, 11,236 (67.8%) as MDR-TB, 2384 (14.4%) as pre-XDR-TB, and 70 (0.4%) as XDR-TB. During the first four months of treatment, the cumulative incidence of mortality was higher among patients with RmR-TB compared to those with MDR-TB, pre-XDR-TB and XDR-TB. Patients with RmR-TB had 14% higher odds of mortality compared to patients with MDR-TB (32.2% vs 23.7%; aOR 1.14; 95% CI: 1.02-1.27), but a lower mortality compared to patients with pre-XDR or XDR-TB. Although HIV status did not significantly modify the effect (p-value interaction = 0.362), the odds of mortality was only higher among RmR-TB patients with HIV co-infection (aOR 1.17; 95% CI:1.03-1.33). Using a large individual patient data meta-analysis, this study confirmed the counterintuitive finding of prior smaller studies that the odds of mortality are higher among patients with RmR-TB compared to MDR-TB. While HIV co-infection was not an effect modifier, the higher odds of mortality were only observed among RmR TB patients living with HIV.