Outcomes for multidrug-resistant tuberculosis patients with and without resistance to fluoroquinolones and second-line injectable drugs: A meta-analysis of individual patient data

Poor treatment outcomes have been reported for tuberculosis (TB) patients harbouring strains resistant to isoniazid and rifampicin (multidrug resistance or MDR-TB), fluoroquinolones and/or second-line injectable drugs. We undertook a meta-analysis for response to treatment using individual data for MDR-TB patients whose strains had additional resistance to fluoroquinolones (MDR-TB+FQ), second-line injectables (MDR-TB+Inj) or both (extensive drug resistance; XDR-TB) including demographic and clinical details, treatment regimens, and outcomes. 26 centres provided data for 424 MDR-TB+FQ, 1129 MDR-TB+Inj, 405 XDR-TB, and 4776 other MDR-TB patients susceptible to FQ and Inj. Success was lower in MDR-TB+FQ (adjusted OR=0.6 [95%CL 0.5-0.7]) and XDR-TB patients (0.4 [0.3-0.6]) than in those with MDR-TB+Inj (0.8 [0.7-0.9]) and those with MDR-TB and no additional resistance (reference). No single drug was significantly associated with treatment success in MDR-TB+FQ and XDR-TB patients. In XDR-TB patients, success was highest if at least 6 drugs were used in the intensive phase (4.9 [1.4-16.6]) and 4 in the continuation phase (6.1[1.4-26.3]). Study results suggest that regimens of a similar duration to those recommended in MDR-TB patients but containing more drugs achieve better results in XDR-TB patients. As all data in the analysis were from observational studies, bias may be substantial and better quality evidence will be needed to guide the optimization of regimens.