The Serum Galactomannan Index Predicts Mortality in Allogeneic Hematopoietic Stem Cell Transplant Recipients with Probable Pulmonary Invasive Aspergillosis

Background: Invasive pulmonary aspergillosis (IPA) remains a significant cause of infectious morbidity and mortality in hematopoietic cell transplant (HCT) recipients. The serum and BAL galactomannan (GM) index are widely used mycologic criteria for the diagnosis of IPA; however, their prognostic value at the time of diagnosis is poorly defined. It remains unclear whether IPA patients with a positive serum or BAL GM index have different clinical outcomes than IPA patients with a negative GM index in the corresponding compartment, and whether higher GM index cut-off levels correlate with poor outcome. In this study, we examined the prognostic value of quantitative aspects of both the serum and BAL GM index at the time of IPA diagnosis.

Methods: We retrospectively analyzed a cohort of 100 adult patients with probable IPA (per 2008 revised EORTC criteria) diagnosed in 2004-2010 within 100 days after a first allogeneic HCT. All subjects had a chest CT, a serum GM, and a BAL GM index performed at the time of IPA diagnosis. We evaluated the impact of different GM cut-off levels in serum and BAL on mortality. Kaplan Meier curves were used to estimate survival, and Cox proportional hazards models were used to evaluate univariate and adjusted hazards ratios for 180 day all-cause mortality associated with different serum and BAL GM index cutoff values.

Results: A diagnosis of probable IPA was made by a positive serum GM index (≥ 0.5) alone in 32 patients and a positive BAL GM index (≥0.5) alone in 47 patients. In 21 patients, the serum and BAL GM indices were both positive. Overall mortality in all patients was 52% at 180 days. Patients with a positive serum GM index at the time of IPA diagnosis had an increased mortality (60.4%; n = 53) compared to patients with a negative serum GM index (42.6%; n = 47). In contrast, a positive BAL GM index had no effect on mortality (Figure 1A,B). In addition, the magnitude of the serum GM index was associated with enhanced mortality. When compared to a serum GM index of ≤ 0.5, increasing values of serum GM were associated with an increased HR of 180 day mortality (serum GM ≥ 1: HR = 2.18 (1.15-4.15); serum GM ≥ 1.5: HR = 2.97 (1.53-5.81); serum GM ≥ 2: HR = 3.68 (1.63-8.30)). This result was confirmed in a multivariate analysis that adjusted for acute GVHD, sex, underlying hematologic disease severity, and elevated creatinine. No association was seen with increasing magnitudes of BAL GM indices.

Conclusions: This study shows an increased HR for mortality with increasing cutoffs of serum GM indices, with no corresponding increase in HR with increasing BAL GM indices. We propose that the initial serum GM index value at the time of IPA diagnosis represents both an important prognostic indicator and a valuable covariate in future analyses on outcomes in HCT recipients.

Figure 1: Kaplan Meier survival curve of different magnitudes of serum GM indices (A), and BAL GM indices (B).