Bronchoalveolar Lavage Fluid Lipoarabinomannan as a Rapid Rule-Out Tool for Smear-Negative Pulmonary Tuberculosis in a Non-HIV-Predominant, High-Burden Setting: A Prospective Study

Abstract Background In high-burden, non-HIV-predominant settings, rapid exclusion of smear-negative pulmonary tuberculosis (TB) is as clinically critical as its confirmation. Delayed diagnosis prolongs empirical anti-TB therapy, defers treatment for alternative etiologies, and compounds socioeconomic burden. Bronchoalveolar lavage fluid (BALF) lipoarabinomannan (LAM) delivers point-of-care results immediately after bronchoscopy, yet its utility as a rule-out biomarker in non-HIV populations remains poorly characterized. Methods We conducted an exploratory prospective study at a tertiary referral center in Bangkok, Thailand (February 2023–February 2024). Adults with clinical and radiographic features of pulmonary TB, negative sputum AFB smear and molecular tests, and a clinical indication for bronchoscopy were enrolled. BALF was tested using the Alere Determine™ TB-LAM Ag lateral flow assay, compared with mycobacterial culture and PCR (Xpert MTB/RIF), supplemented by histopathology when applicable. Results Of 219 participants (99.5% non-HIV; mean age 64 ± 13 years), 24 (11.0%) had confirmed TB. A negative BALF-LAM result carried an NPV of 92.2% (95% CI: 87.1–95.4%), reducing the post-test probability of TB from 11.0% to 8.4% (negative likelihood ratio: 0.74). Sensitivity was 45.8% (95% CI: 25.6–67.2%) and specificity 74.9% (95% CI: 68.2–80.7%). No clinical or radiographic variable was independently associated with false-positive results. Under a composite reference standard including clinical TB (n = 38), NPV remained 84.9% (95% CI: 78.5–89.6%). Conclusions BALF-LAM functions as a reliable, rapid rule-out adjunct for smear-negative pulmonary TB in non-HIV-predominant, high-burden settings. Its high NPV, low cost, and point-of-care availability suggest potential utility within post-bronchoscopy diagnostic algorithms, pending multi-center validation.