Abstract
BACKGROUND: Cellular patterns in bronchoalveolar lavage fluid (BALF) are used to distinguish or rule out particular diseases in patients with acute respiratory failure (ARF). However, whether BALF cellular patterns can predict mortality or not is unknown. We test the hypothesis that BALF cellular patterns have predictive value for mortality in patients with ARF. METHODS: This was a retrospective single-center observational study conducted in a Japanese University Hospital. Consecutive patients (n = 78) with both pulmonary infiltrates and ARF who were examined by bronchoalveolar lavage (BAL) between April 2015 and May 2018 with at least 1 year of follow-up were analyzed. Primary analysis was receiver operating characteristic curve-area under the curve (ROC-AUC) analysis for 1-year mortality. RESULTS: Among the final sample size of 78 patients, survivors (n = 56) had significantly increased lymphocyte and eosinophil counts and decreased neutrophil counts in BALF compared with non-survivors (n = 22). Among the fractions, lymphocyte count was the most significantly different (30 [12-50] vs. 7.0 [2.9-13]%, P <0.0001). In the ROC curve analysis of the association of BALF lymphocytes with 1-year mortality, the AUC was 0.787 (P <0.0001, cut-off value [Youden index] 19.0%). Furthermore, ≥20% BALF lymphocytes were significantly associated with increased survival with adjustment for baseline imbalances (1-year adjusted hazard ratio, 0.0929; 95% confidence interval, 0.0147-0.323, P <0.0001; 90-day P =0.0012). Increased survival was significantly associated with ≥20% BALF lymphocytes in both interstitial lung disease (ILD) and non-ILD subgroups (P =0.0052 and P =0.0033, respectively). In secondary outcome analysis, patients with ≥20% BALF lymphocytes had significantly increased ventilator-free days, which represents less respiratory dysfunction than those with <20% BALF lymphocytes. CONCLUSIONS: In the patients with ARF, ≥20% lymphocytes in BALF was associated with significantly less ventilatory support, lower mortality at both 90-day and 1-year follow-ups.