Prediction of Non-Relapse Mortality with Dynamic Easix Scores after Allogeneic Hematopoietic Cell Transplantation

Abstract

Introduction Endothelial Activation and Stress Index (EASIX), a surrogate of endothelial dysfunction, is a powerful tool for predicting non-relapse mortality (NRM) when evaluated prior to allogeneic hematopoietic cell transplantation (allo-HCT) and at specific landmarks post-HCT. Objectives We sought to determine if changes in EASIX over time may also be an informative marker of NRM. Methods We evaluated 509 adults who underwent RIC/NMA unmodified or MAC ex-vivo CD34+-selected allo-HCT between 2008-2016. Unmodified allo-HCT patients (pts) received a GVHD prophylaxis of sirolimus/tacrolimus and low-dose methotrexate. CD34+-selected pts received no planned immunosuppression. EASIX scores (LDH x creatinine/platelet count) were calculated continuously until 1-year post-HCT. A log transformation using base 2 (log2) was applied to all EASIX variables to reduce skew. A 1-unit increase in log2 EASIX is associated with a doubling of EASIX on the original scale. Disease relapse or death were considered competing risks for NRM. Results 360 pts (71%) received a CD34+-selected and 149 (29%) an unmodified allo-HCT. HLA matched donors were used in 435 (86%) and mismatched in 74 (14%) pts. HCT-CI was 0 in 24% of pts, 1-2 in 32%, and ≥ 3 in 44%. Median age was 56 (range 19-78) and 59% were males. Most unmodified allo-HCTs were done for non-Hodgkin lymphoma (69%), and most CD34+-selected allo-HCTs were done for acute leukemia (61%). Among all patients, EASIX scores increased rapidly, peaked at day +8, then declined until day +40. Thereafter, scores decreased but remained above pre-HCT baseline for the first year (Figure 1a). In unmodified pts, EASIX scores appeared higher over time than in CD34+-selected pts, whose scores rapidly decreased by 2 months post-HCT (Figure 1b). EASIX's ability to discriminate NRM in the subsequent 180 days was similar when analyzed as a categorical variable at landmark timepoints and as change from pre-HCT. EASIX discrimination of NRM was highest between day +180 to +210 (concordance index=0.85) (Figure 2). At this landmark, pts with log EASIX >2 died of infection (14%), GVHD (29%), relapse (33%), and other causes (25%) in the subsequent 180 days. Mean EASIX scores at this landmark were higher in men (mean log2 +0.52) and in pts who developed grade 2-4 GVHD by day +180 (+0.81), lower in pts who received matched unrelated versus mismatched donors (-0.81, all P Conclusion EASIX scores are dynamic and variably concordant with NRM when analyzed longitudinally post-HCT, and patterns differ between allo-HCT platforms. Compared to pre-HCT evaluation, dynamic EASIX scores may better predict risk of NRM over time as pts acquire additional endothelial injury and toxicities along the allo-HCT journey.