A Retrospective Validation of Three Standard Prognostic Instruments Used to Inform Decisions Regarding Autologous and Allogeneic Stem Cell Transplantation

Background The Hematopoietic Cell Transplantation (HCT) – Specific Comorbidity Index (HCT-CI) measures the risk of transplant related mortality and predicts overall survival (OS) based on co-morbidities. In a prospective validation study of the HCT-CI by the Center for International Blood and Marrow Transplant Research (CIBMTR), HCT-CI ≥ 3 was associated with lower OS in both allogeneic (allo) and autologous (auto) HCT regardless of diagnoses, age, or conditioning. In contrast to the patients (pts) analyzed in the CIBMTR study, pts transplanted at Wake Forest more often have HCT ≥3. Methods We retrospectively analyzed data from pts transplanted from Sep 2014 to Sep 2016. 2-yr OS was observed for all pts. Results from 3 tools were used to compare predicted vs observed OS: HCT-CI, Disease Risk Index (DRI), and the CIBMTR survival calculator. Results Over 2 yrs, 216 pts were transplanted – 132 auto, median age 62 (26-78) and 84 allo, median age 55.5 (16-74). 59% of pts had a HCT-CI ≥3. In both groups, predicted and observed OS were not statistically different for those with HCT-CI 0. In contrast, predicted and observed OS for HCT-CI ≥3 were significantly different (p , predicted and observed 1-yr OS for all allo pts in the dataset were no different; however, in the HCT-CI ≥3 subset, observed 1-yr OS was significantly better than predicted. Additionally, observed vs predicted 2-yr OS using the DRI was significantly different in every risk group. Regardless of the tool applied, we consistently had better observed OS than predicted OS for high risk pts. Of pts with HCT-CI ≥ 3, 80.4% had a pulmonary score of 2 or 3. Conclusions Compared to the dataset utilized by the CIBMTR for validation of the HCT-CI, scores ≥ 3, largely accounted for by poor pulmonary function, are over represented in our population. Further analysis will need to be done to determine if the regional prevalence of poor pulmonary function is contributing to a significant difference in the predicted vs observed OS in our allo pts.