Correlation between Progression-Free Survival and Overall Survival in Patients with Relapsed Hodgkin Lymphoma: An Analysis of Individual Patient Data from Randomized GHSG Trials

Background: Progression-free survival (PFS) and overall survival (OS) are predominant measures of treatment efficacy in Hodgkin lymphoma (HL). Despite preference for OS from many regulatory authorities, PFS is most relevant to patients and frequently serves as primary endpoint in clinical trials. Recently we found a strong correlation between treatment effects on PFS and OS in the first-line setting (Bröckelmann PJ et al. EHA23/ICML23). However, their relationship in the relapsed setting remains unclear. We aimed to evaluate the correlation of PFS with OS after treatment of relapsed HL. Methods: We evaluated individual patient data of 375 patients with relapsed HL treated within the prospective randomized GHSG trials HDR1 (N= 141) and HDR2 (N= 234) between 02/1993 - 06/2007. Briefly, both trials included patients of 18 to 60 years of age with histologically confirmed relapsed HL after ≥1 prior line of treatment, who had not previously received high-dose chemotherapy and autologous stem-cell transplantation (ASCT). In HDR1 patients received two cycles of Dexa-BEAM (dexamethasone, carmustine, etoposide, cytarabine, and melphalan) and either two further courses of Dexa-BEAM or high-dose BEAM and ASCT. In HDR2, patients were assigned to two cycles of dexamethasone, cytarabine, and cisplatin (DHAP), and thereafter randomly assigned to either BEAM followed by ASCT or sequential cyclophosphamide, methotrexate, and etoposide in high-doses before BEAM followed by ASCT. All patients who received at least one dose of salvage chemotherapy within the respective trial were included in the analysis. We correlated treatment effects and risk factor effects on PFS and OS with the Wei-Lin-Weissfeld (WLW) method and adjusted the analysis for age and sex. Additionally, we applied copula models to correlate the time-to-event variables PFS and OS themselves using their marginal distributions. Results: With a median follow-up of 4.5 years, at least one PFS event, defined as relapse or progression of relapsed HL or death for any reason, was recorded in 146 of 375 patients (38.9%). At least one OS event, defined as death for any reason, was documented in 111 of 375 patients (29.6%). The hazard ratios for treatment effects on PFS and OS were 0.61 (95% confidence interval, CI= 0.38-0.98) and 1.01 (95% CI= 0.63-1.63) for HDR1 and 1.1 (95% CI= 0.69-1.75) and 0.8 (95% CI= 0.44-1.47) for HDR2, respectively. The WLW analysis revealed high correlations of treatment effects on PFS and OS at patient level within the trials. Pearson correlation r of the treatment effects on PFS and OS were r= 0.72 (95% confidence interval CI= 0.66-0.76, P< 0.001) in all 375 patients analyzed, r= 0.72 (95% CI= 0.63-0.79, P<0.001) in the HDR1 trial and r= 0.74 (95% CI= 0.67-0.79, P< 0.001) in the HDR2 trial. The analysis with different risk factors and copula models revealed similarly high correlations, confirming a strong relation between PFS and OS in relapsed HL patients. Conclusions: PFS and OS as well as treatment effects and prognostic effects on PFS and OS are highly correlated in patients treated for relapsed HL. Confirming results in first-line treatment, PFS thereby constitutes a highly relevant endpoint also in the setting of relapsed HL.