2032. Predictors of 6-Week Mortality in Patients with Positive Bronchoalveolar Lavage (BAL) Galactomannan (GM)

Abstract

BackgroundBAL-GM is a mycologic criterion for diagnosis of probable invasive aspergillosis (IA). However, in a contemporary cohort of consecutive patients with BAL-GM measured as part of their workup for potential IA, we previously showed that 42% of positive (≥0.5) BAL-GM values can be falsely positive; positive predictive value was increased by using higher cutoffs and in patient groups with high pre-test probability for IA. In this study from the same cohort, we analyze the prognostic value of BAL-GM and identify predictors of 6-week mortality, the main outcome in most studies of mold-active antifungal drugs.MethodsWe reviewed clinical and microbiologic data of patients who had ≥1 positive BAL-GM (≥0.5), at Brigham and Women’s Hospital (November 2009–March 2016). We applied EORTC/MSG invasive mold infection (IMI) definitions to classify cases as possible, probable or proven IMI, excluding BAL-GM result as mycologic criterion, and used Cox regression to identify factors associated with 6-week all-cause mortality.ResultsWe studied 134 patients (median age 58 years, 49% women, 55% with hematologic malignancy, 10% solid-organ and 34% hematopoetic stem-cell transplant recipients). APACHE II score, liver disease, acute kidney injury, and shock were independently associated with higher 6-week mortality. ICU stay, mechanical ventilation, corticosteroids, hypertension, EORTC/MSG category, serum-GM and antifungal treatment were associated with higher mortality in univariate, but not multivariate analyses. BAL-GM value was independently associated with 6-week mortality (adjusted HR 1.24(continuous variable), 95% CI 1.1–1.39, P < 0.001). The association of BAL GM strata with 6-week crude mortality was significant in patients with possible, probable or proven IMI, but not in those without IMI (Figure 1). ConclusionHigher BAL-GM values were an independent predictor of 6-week mortality, having prognostic value in patients with possible, probable or proven IMI, but not in patients who did not meet other criteria for IMI. We propose critical reassessment of BAL-GM cutoff values in different patient populations.Figure 1.Kaplan–Meier (KM) curves for different cutoffs of BAL GM.Disclosures All authors: No reported disclosures.