Comparison of Pretransplantation Prediction Models for Nonrelapse Mortality in Patients with Myelofibrosis Undergoing Allogeneic Stem Cell Transplantation

Allogeneic stem cell transplantation (alloSCT) is the only known curative treatment for myelofibrosis (MF). Risk assessment remains important for patient counseling and predicting survival outcomes for relapse and nonrelapse mortality (NRM). Outcome-prediction tools can guide decision-making. Their use in MF has relied on their extrapolation from other malignancies. The primary objective of this study was to assess the performance of the Hematopoietic cell Transplantation Comorbidity Index (HCT-CI), the augmented HCT-CI (aHCT-CI), and the Endothelial Activation and Stress Index (EASIX) in predicting NRM in patients with MF undergoing alloSCT. We retrospectively reviewed patients with MF undergoing alloSCT between 2012 and 2020 at the Mayo Clinic. Data were abstracted from the electronic medical record. EASIX score was calculated before starting conditioning therapy and analyzed based on log2- transformed values. We evaluated the log2-EASIX scores by quartiles to assess the effect of increasing values on NRM. NRM was evaluated using competing risk analyses. We used the Kaplan-Meier and log-rank methods to evaluate OS. The Fine-Gray model was used to determine risk factors for NRM. The performance of HCT-CI and aHCT-CI was compared by evaluation of model concordance given the high correlation between HCT-CI and aHCT-CI (r = .75). A total of 87 patients were evaluated. The median duration of follow-up after alloSCT was 5 years (95% confidence interval [CI], 4.4 to 6.31 years). Patients with a high HCT-CI score had significantly increased cumulative incidence of NRM at 3 years (35.5% versus 11.6%; P = .011) after alloSCT. A progressively increasing 3-year NRM was observed with increasing aHCT-CI risk category, and patients with a high or very high aHCT-CI score had significantly higher 3-year NRM compared to those with intermediate-risk or low-risk aHCT-CI scores at 3 years post-alloSCT (31.9% versus 6.52%; P = .004). An increasing log2-EASIX score quartile was not associated with 3-year NRM (19.0% versus 10.1% versus 25% versus 14.3%; P = .59), and the EASIX score was not found to be a predictor of post-transplantation NRM. A high HCT-CI was associated with significantly worse 3-year overall survival (OS) (hazard ratio [HR], 4.41; 95% CI, 1.97 to 9.87; P < .001). A high or very high aHCT-CI was significantly associated with poor 3-year OS (HR, 3.99; 95% CI, 1.56 to 10.22; P = .004). An increasing log2-EASIX score quartile group was not associated with 3-year OS (3-year OS rate, 66.7% versus 80.4% versus 64.6% versus 76.2%; P = .57). The EASIX score should not be used routinely in patients with MF. Both the HCT-CI and the aHCT-CI are accurate in predicting long-term survival outcomes in this patient population. Further studies are important to validate our findings of the role of EASIX in predicting NRM in patients with MF or other myeloproliferative neoplasms undergoing alloSCT.