First Line Therapy Evaluation Using Propensity Score Approach in Newly Diagnosed Advanced Classical Hodgkin Lymphoma Patients from Prospective Real-World Realysa Cohort and Phase 3 AHL2011 Trial

Background: AHL2011, a randomized phase III trial demonstrated that PET-CT after two cycles of BEACOPP_escalated chemotherapy safely guided treatment of patients aged 16–60 years with newly diagnosed advanced classical Hodgkin lymphoma (cHL) and allowed the use of ABVD in early responders with reduced toxicities and no impaired disease control. REALYSA study, a real-life multicentric cohort set up in France since 2018, included 6000 newly diagnosed lymphoma patients with over 1000 cHL patients. In the present study, we investigated whether real-world data could reproduce results issued from a randomized phase III clinical trial using the propensity score approach.

Methods: After stringent propensity score matching (PSM) on 10 baseline parameters (gender, age, Ann Arbor stage, ECOG Performance Status, number of extranodal sites involved, B symptoms, albumin, hemoglobin, white blood cell count and lymphocytes), we performed a PSM testing ratios 1:1 (one patient from AHL2011 trial and one patient from REALYSA cohort). All patients from AHL2011 and REALYSA included in this study had PET-CT monitoring after two cycles of BEACOPP_escalated. This approach allowed us to build well-balanced pairs as demonstrated by the absolute values of the standardize mean differences (SMDs) below 0.1 for all matching covariates except for ECOG (SMD=0.15). We also used the Inverse Probability of Treatment Weighting (IPTW, IPTW stabilized) to compare the experimental arm of the AHL2011 trial with a synthetic control arm established with real-world data from REALYSA cHL patients who respected the main inclusion criteria of AHL2011 trial and were treated following this PET-guided approach. The median follow-up was censored in the AHL2011 data in order to be comparable to the median follow-up observed in the REALYSA cohort.

Results: Among the 823 patients of AHL2011 and 942 patients of REALYSA with newly diagnosed cHL, 390 and 100 patients respectively were eligible for propensity score calculation, corresponding to AHL2011 TEP-guided therapy, aged between 16 and 60 years, no missing data for the covariates included in the PSM and at least 12 months of follow-up in the REALYSA study. After 1:1 PSM (n=98 AHL2011 and n=98 REALYSA), the overall response rate at the end of induction was 91.8% vs 89.8% (p=0.81) for patients included in AHL2011 and REALYSA, respectively. The proportion of patients with positive PET-CT after 2 cycles was similar in AHL2011 (13%) and REALYSA (14%). After a median follow-up of 13 months, the 1-year progression free survival rates were 93.8% and 87.8% (HR=0.87; 95%CI 0.4-2.0; P=0.74) in AHL2011 and REALYSA, respectively (Figure 1A). Using IPTW model including covariates from the PSM, similar PFS were obtained between experimental arm of AHL2011 and patients included in REALYSA (Figure 1B). IPTW stabilized and other IPTW models also showed similar results for PFS.

Conclusions: In this study, we demonstrated using a propensity score approach that patients with newly diagnosed advanced cHL treated according to the AHL2011 PET-guided frontline approach in routine care and included in REALYSA cohort reached similar outcomes as patients included in prospective phase III clinical trial. Beyond the relevant validation of the AHL2011 trial results in real-life, which will need longer follow up, these data could be used to design composite control arm for future clinical trials.