Effectiveness and safety of lung resection surgery among patients with pulmonary multi-drug resistant tuberculosis

Background: The effectiveness of medical treatment for multidrug-resistant tuberculosis (MDR-TB) is still lower than the observed with drug-susceptible TB.Objectives: To estimate the effect of adjunctive pulmonary resection, either of any extent, total (pneumonectomy) or partial lung resection (lobectomy, segmentectomy, or wedge resection), on treatment success and death during therapy in patients with MDR-TB. We studied the presence of effect modification by the following variables: number of group A drugs, additional fluoroquinolone resistance, level of experience of the center, timing of surgery relative to the culture conversion date, as well as bilateral and/or cavitary disease.Methods: We conducted an individual patient data systematic review and meta-analysis of cohorts of patients with pulmonary MDR-TB with a known surgical status. Cohort entry was set at the MDR-TB treatment start date. Follow-up continued until an end-of-treatment outcome (cure, completion, failure, death, recurrence, or loss to follow-up [LTFU]). Patients exposed to surgery (cases) were propensity score-matched with subjects treated only medically (controls) that had survived at least the same amount of time as cases at the moment of surgery. Three sources of controls were used: a) Studies performed at centers where no patients were surgically treated (non-surgical studies); b) Studies performed at centers where >=1 patient underwent surgery (surgical studies), and c) The combination of both. We used mixed-effects generalized linear regression to estimate the odds ratio (OR) for treatment success (cure or completion without recurrence) relative to an unsuccessful response (failure, recurrence, or death) and for death relative to remained alive at the end of treatment. We excluded patients LTFU from these analyses. Modification of effect was assessed through stratified analyses.Results: We evaluated 6,025 patients from 41 studies; 344 of them underwent surgery (70 pneumonectomy, 259 partial lung resection, and 15 with unknown extent). Lung resection of any extent was not significantly associated with the odds of treatment success (controls from all studies: OR, 0.99; 95% confidence interval [CI], 0.63, 1.56; from surgical studies: OR, 1.09, 95% CI, 0.70 1.71; from non-surgical studies: OR, 0.35; 95% CI, 0.09, 1.40), nor with the odds of death (controls from all studies: OR, 0.83; 95% CI, 0.43, 1.61; from surgical studies: OR, 0.79, 95% CI, 0.42, 1.48; from non-surgical studies: OR, 2.50; 95% CI, 0.23, 27.54).Partial lung resection was non-significantly associated with higher odds of treatment success and lower odds of death (controls from all the studies: OR, 1.30, 95% CI, 0.74, 2.28; OR, 0.54; 95% CI, 0.22 1.29, respectively; controls from surgical studies: OR, 1.55, 95% CI, 0.89, 2.69; OR, 0.44; 95% CI, 0.19, 1.01, respectively). Conversely, total lung resection was associated with lower odds of treatment success and higher odds of death (controls from all the studies: OR 0.52; 95% CI, 0.19, 1.41; OR, 1.57, 95% CI, 0.39, 6.31, respectively; controls from non-surgical studies: OR 0.19, 95% CI, 0.04, 0.89; OR 5.33, 95% CI, 0.65, 43.71, respectively). Among patients with fluoroquinolone-resistant MDR-TB, partial lung resection was associated with a significant decrease in the odds of death (controls from all the studies: OR 0.25; 95% CI, 0.06 to 0.99) and a non-significant increase in the odds of treatment success (controls from all the studies: OR, 1.69; 95% CI, 0.77 to 3.72). We also found a non-significant increase in the odds of treatment success (OR, 1.61; 95% CI, 0.84, 3.08) and a non-significant decrease in the odds of death among patients treated with partial lung resection at more experienced centers (OR, 0.51; 95% CI, 0.2, 1.32).Conclusion: Partial lung resection might be beneficial in treating patients with MDR-TB, particularly when performed at highly experienced centers or when there is additional resistance to fluoroquinolones