Validation of Comorbidity Indexes in Reduced-Intensity Conditioning (RIC) Allogeneic Stem Cell Transplantation. the Hematopoietic Cell Transplantation Comorbidity Index Is the Best Predictor of NRM and Survival.

Introduction: Morbidity and mortality after allogeneic hematopoietic stem cell transplantation (AlloHSCT) are considerable, but historically it has been difficult to generate a reproducible and easy model for predicting mortality after the procedure. The use of RIC or nonmyeloablative conditioning regimens may allow an extension of the procedure to patients not eligible for conventional AlloHSCT. However, such patients are usually older and have more comorbidities than patients who meet the strict criteria for a conventional AlloHSCT. In an attempt to predict before transplantation the higher mortality risks attributable to patient comorbidities, researchers from Seattle have recently developed several comorbidity indexes (HCT-CI Blood, 2005 and PAM Ann Int Med, 2006) with the aim of improving non-transplant classical models, such as the Charlson Comorbidity Index (CCI). The validation of these comorbidity indexes in other institutions and in different disease and conditioning-related settings is of utmost importance for the extension of their use to clinical practice and clinical research. We thus performed a retrospective study in our institution to validate these indexes in the setting of RIC AlloHSCT recipients only.

Methods: Overall survival (OS) curves were estimated using the Kaplan-Meier method, while the cumulative incidence of non-relapse mortality (NRM) was calculated considering relapse as a competitive event. Likelihood ratio statistics from proportional hazard regression models were computed for each index for the study events over the first two years post-AlloHSCT. The c-statistic was calculated to estimate the predictive capacity of each index for the analyzed transplant outcomes as previously described (Stat Med, 1984).

Results: We retrospectively calculated for each patient the HCT-CI, PAM and CCI scores. Study subjects included 194 consecutive patients who underwent RIC-AlloHSCT in our institution up to November 2006, to allow at least 18 months follow-up in all cases. The median age was 57 years (range 17–71). According to the score, three risk groups were created for each index (low, intermediate and high risk). No significant differences were found between the three risk groups of each index regarding age, sex, underlying disease, relapse risk at transplantation and donor type (79% HLA identical sibling donors). The median patient pre-transplant comorbidity scores for CCI, HCT-CI and PAM were 0, 3.5 and 22, respectively. For predicting 2-year NRM, the HCT-CI score groups were associated with the highest predictive hazard ratios (HR) (Low risk, HR 1.0; Intermediate risk, HR, 5.01 [95% C.I., 1.53–16.4]; High risk, 6.97 [95% C.I., 2.20–22.1], p<0.01). However, we found no predictive value for NRM with the risk group categorization using the PAM nor the CCI. The better predictive capacity for NRM of the HCT-CI than PAM and CCI was confirmed with the c-statistics (c-statistics of 0.672, 0.634, 0,595, respectively). Regarding the 2-year OS, the HCT-CI scores categories were also associated with the highest predictive HR (Low risk, HR 1.0; Intermediate risk, HR 1.6 [95% C.I., 0.99–2.72], High risk, 2.7 [95% C.I., 1.66–4.28], p<0.01). In conclusion, our single-center study suggests that the HCT-CI is a good predictor of 2-year NRM and survival after a RIC AlloHSCT, and thus we encourage international efforts to validate and improve this and other predictive indexes.