Dynamic EASIX scores closely predict nonrelapse mortality after allogeneic hematopoietic cell transplantation

Abstract

Abstract Endothelial activation and stress index (EASIX) predicts nonrelapse mortality (NRM) when assessed before hematopoietic cell transplantation (HCT). We sought to determine whether changes in EASIX after HCT may be an informative marker of NRM. We evaluated 509 adults who underwent reduced intensity, unmodified (N = 149, 29%), or myeloablative ex vivo CD34+-selected allogeneic HCT (allo-HCT) (N = 306, 71%) between 2008 and 2016. Patients who underwent unmodified allo-HCT received tacrolimus-based graft-versus-host disease (GVHD) prophylaxis, whereas CD34+-selected patients received no planned immunosuppression. EASIX (lactate dehydrogenase × creatinine/platelet count) was calculated continuously until 1-year after HCT. Log transformation using base 2 (log2) was applied to all EASIX variables to reduce skew. In total, 360 patients (71%) received CD34+-selected and 149 (29%) unmodified allo-HCT. Among all patients, EASIX scores increased rapidly, peaked at day +8, then declined rapidly until day +33. Thereafter, scores declined gradually but remained above the pre-HCT baseline. In unmodified HCT, scores appeared higher over time than in CD34+-selected patients. EASIX discrimination of NRM was highest around day +180 (concordance index = 0.85) in both platforms, but the prognostic impact of EASIX across time points differed between the 2 platforms. Mean EASIX scores were higher in men (mean log2 +0.52) and in patients who developed grade 2 to 4 GVHD (+0.81) and lower in patients who received matched vs mismatched donors (−0.81, all P < .01). EASIX scores are dynamic and variably concordant with NRM when analyzed longitudinally, and patterns differ between HCT platforms. Compared to pre-HCT evaluation, post-HCT EASIX scores may better predict risk of NRM as patients acquire additional endothelial injury and toxicities.