Bronchoalveolar lavage fluid cellular fractions reflect individual histopathological features in interstitial pneumonia

Bronchoalveolar lavage fluid (BALF) cellular analysis is an adjunctive diagnostic tool in interstitial pneumonia (IP), yet its relationship with individual histopathological features remains unclear. We retrospectively compared BALF cellular fractions with 24 histopathological features in 67 IP patients who underwent BALF and surgical lung biopsy at a single center (2005-2020). False discovery rate correction was applied (q < 0.05), followed by multivariable logistic regression and Kaplan-Meier analysis. BALF lymphocyte proportion was negatively associated with fibroblastic foci (13.0% vs 29.0%, q = 0.002) and positively with bronchiolitis (q = 0.029), intraluminal organization (q = 0.002), and epithelioid cell granulomas (q = 0.012). On multivariable analysis, fibroblastic foci were independently associated with a lower likelihood of BALF lymphocytosis (odds ratio 0.23, p = 0.018), and this association persisted after exclusion of patients with a usual interstitial pneumonia (UIP) or probable UIP pattern and of patients with idiopathic pulmonary fibrosis. Intraluminal lymphocyte proliferation was also associated with BALF lymphocytosis, although this estimate was based on few patients. Patients with fibroblastic foci had poorer survival (p = 0.003), whereas survival did not differ when patients were stratified by BALF lymphocyte proportion. ROC analysis identified a BALF lymphocyte cutoff of 18.5% for the presence of fibroblastic foci (AUC 0.76), and adding BALF lymphocyte proportion to the high-resolution computed tomography pattern improved discrimination of fibroblastic foci (AUC 0.847 versus 0.686). Neutrophil and eosinophil proportions showed no significant associations after correction. BALF lymphocyte proportion is associated with several histopathological features in IP, most strongly with fibroblastic foci, and may provide complementary information where surgical lung biopsy is not feasible.