EASIX in patients with acute graft-versus-host disease: a retrospective cohort analysis

Abstract

Background: Endothelial dysfunction links thrombotic microangiopathy to steroid-refractory graft-versus-host disease (GVHD) after allogeneic stem-cell transplantation. We aimed to assess if the simple formula-lactate dehydrogenase (U/L) × creatinine (mg/dL)/thrombocytes (10⁹ cells per L)-termed the Endothelial Activation and Stress Index (EASIX), might be valuable for the prediction of death in patients with acute GVHD after allogeneic stem-cell transplantation.

Methods: For this retrospective analysis, we analysed a training cohort (in Germany) and three validation cohorts (in Germany and the USA) of patients with acute GVHD who had received consecutive allogeneic stem-cell transplantation. The primary endpoint was prediction of overall survival when measured at acute GVHD onset (EASIX-GVHD). We validated the prognostic strength of EASIX-GVHD for overall survival and non-relapse mortality in the three independent cohorts by calculating the prediction error (integrated Brier score), and concordance index.

Findings: In the total cohort of patients with acute GVHD (n=311), EASIX-GVHD predicted overall survival in univariable and multivariable models (univariate analysis, hazard ratio [HR] for a one-fold increase 1·16, 95% CI 1·12-1·20, p=0·0004). However, in the subpopulation of patients with myeloablative conditioning (n=72), EASIX-GVHD did not predict overall survival, which is probably attributable to thrombocytopenia at GVHD onset (73 × 10⁹ cells per L [IQR 29·75-180·00] for myeloablative conditioning vs 160 × 10⁹ cells per L [90·0-250·5] for reduced-intensity conditioning; p<0·0001). In patients who received reduced-intensity conditioning (n=239), EASIX-GVHD was a strong predictor of overall survival (HR for a two-fold change of 1·23, 95% CI 1·13-1·34; p<0·0001) and non-relapse mortality (cause-specific HR for a two-fold change of 1·24, 1·12-1·38; p<0·0001). Model validation for prediction of overall survival and non-relapse mortality by EASIX-GVHD was successful in two independent cohorts of adult patients with reduced-intensity conditioning (n=141, n=173) and in a cohort with mainly paediatric patients (n=89).

Interpretation: In patients with reduced-intensity conditioning, EASIX-GVHD is a powerful predictor of survival after GVHD. EASIX-GVHD could be the future basis for development of risk-adapted GVHD treatment strategies.

Funding: There was no external funding source for this study.