Prognostic Models for Predicting Outcomes after Autologous Stem Cell Transplantation (ASCT) in Patients with Relapsed or Refractory Hodgkin’s Disease: How Predictive Are They?.

Multiple prognostic models for patients with advanced Hodgkin’s disease have been developed in an effort to identify high-risk individuals for ASCT and predict outcomes of high dose therapy and transplant. The utility of four of these models was examined in a cohort of patients transplanted in our program between 1993 and 2005. A total of 113 patients with relapsed or refractory Hodgkin’s disease received ASCT. Forty-five patients received a conditioning regimen of busulfan, melphalan, and thiotepa (BuMelTT) whereas 68 patients received other standard conditioning regimens (SCR) including Cy/TBI/VP (23), CBV (39), and other (6). Median age is 34.5 years for BuMelTT and 34.0 years for SCR patients. Ninety-eight percent of patients receiving BuMelTT had at least two or more prior regimens compared with 86% SCR patients. Eighty-seven percent of BuMelTT and 93% of SCR patients had chemosensitive disease prior to transplant. Median followup is 113 weeks for BuMelTT patients versus 201 weeks for SCR patients. To date 37% of the BuMelTT and 56% of SCR patients have relapsed (p<0.05) with a median time to relapse of 30 and 36 weeks (p=NS) respectively. Statistically significant factors differentiating the BuMelTT and SCR groups included median followup post-ASCT, total percent relapse after transplant, death after transplant, relapse in prior radiation field, and stage at salvage therapy pre-ASCT. Median OS, EFS, and RFS have not yet been reached for BuMelTT patients compared with 439 (p=0.03), 67 (p=NS), and 74 (p=NS) weeks for the SCR patients. TRM was five percent in both groups at five years. We investigated the predictive value of four prognostic models on the entire group, BuMelTT and SCR cohorts. Models evaluated included Roswell Park (RPCI), Memorial Sloan-Kettering, German Hodgkin’s Study Group, and Southwestern Oncology Group (SWOG). Each of these models has identified three negative prognostic indicators on the basis of multivariate analysis. Patients who had 0 or 1 factor were considered good risk and those that had 2 or 3 factors poor risk. Only the SWOG model (>2 prior regimens, relapse in previous radiation field, extranodal disease) was predictive for OS for the entire group (p<0.01) and for the BuMelTT (p=0.05) and SCR (p<0.01) cohorts. The SWOG model was also able to differentiate EFS between good and poor risk patients for the entire group (p=0.02) and SCR group (p=0.05) but not for the BuMelTT cohort. None of the models could differentiate between good and poor risk patients with regard to RFS. When good and poor risk were defined as 0 factors versus 1 to 3 factors, the RPCI model (chemotherapy-resistant disease, poor performance status, and 3 or more chemotherapy regimens prior to transplant) was predictive of OS (p=0.02) and EFS (p=0.02) but only for the BuMelTT cohort. These results suggest that BuMelTT may improve long-term outcomes in patients with advanced Hodgkin’s disease. The prognostic models evaluated had limited utility in our patient cohort with only the SWOG model being predictive of outcome. Large multi-institutional studies are needed to improve prognostic models for transplantation in advanced Hodgkin’s disease.