Utility of Xpert MTB/RIF using bronchoalveolar lavage samples for the diagnosis of pulmonary tuberculosis in negative or sputum-scarce adults: a systematic review and meta-analysis

Abstract

Introduction: Early diagnosis of pulmonary tuberculosis (PTB) is critical to reducing morbimortality in adults. In cases of negative-sputum or limited sputum production, bronchoalveolar lavage (BAL) samples offer an alternative for diagnosis. Culture, the gold standard, requires a high bacterial load and is time-consuming. Xpert MTB/RIF provides faster results with a higher cost. Previous systematic reviews present substantial limitations. Therefore, the diagnostic utility of Xpert MTB/RIF using BAL samples in adults with negative or scant sputum for PTB needs to be reassessed. Methods: A systematic search of MEDLINE, Embase, LILACS and Web of Science was conducted without language or publication date restriction. Cross-sectional diagnostic studies of adults with sputum negative or sputum-scarce presumptive PTB, who underwent bronchoscopy to obtain samples for Xpert MTB/RIF and culture, were included. Screening and data extraction were performed independently. A bivariate hierarchical random-effects model estimated sensitivity, specificity, likelihood ratios, diagnostic odds ratios, and generated an SROC curve with 95% confidence intervals. Meta-analysis was performed using Metadisc 2.0. Heterogeneity was assessed using I 2 and Cochrane thresholds. Sensitivity analyses were performed based on study design, population differences, country, culture method and risk of bias. Methodological quality was assessed using the QUADAS-2 tool. Publication bias was investigated using the funnel plot. The certainty of evidence was evaluated using the GRADE approach. Results: Published between 2013 and 2025, 13 studies met the inclusion criteria, encompassing 1,807 participants (591 with TB and 1,216 without; disease prevalence ~32.7%). The pooled sensitivity and specificity of Xpert MTB/RIF on BAL samples for PTB in adults who were smear-negative or sputum-scarce were 92.0% (95% CI: 85.8–95.7) and 94.3% (95% CI: 87.4–97.5), respectively. The diagnostic odds ratio was 188.66 (95% CI: 71.74 - 496.13), with a positive likelihood ratio of 16.0 (95% CI: 7.15 - 35.82) and a negative likelihood ratio of 0.085 (95% CI: 0.04 - 0.15). Heterogeneity was moderate (I² = 69%). Sensitivity analyses based on study design, culture method, HIV prevalence, country type, and risk of bias showed consistent results. Although funnel plot inspection suggested possible publication bias, Deeks’ test was not statistically significant (p = 0.0595). Overall, the certainty of evidence was rated moderate, mainly due to risk of bias in several studies. Conclusion: Our meta-analysis demonstrates that Xpert MTB/RIF on BAL exhibits high diagnostic accuracy for detecting PTB in adults who are sputum-scarce or sputum-negative, with an overall moderate quality of evidence.