Endpoint Surrogacy in Chronic Lymphocytic Leukemia: A Pooled Analysis of the German CLL Study Group

Abstract

Introduction Overall survival (OS) is generally considered the most patient-relevant endpoint in oncology trials, however, its implementation can entail several challenges in the context of chronic lymphocytic leukemia (CLL): 1) The competing survival risk due to comorbidities in the mostly elderly CLL patient population; 2) the high efficacy of targeted therapies warrant longer observation periods or extensive patient recruitment, which can be operationally challenging; 3) the interference with increasingly potent relapse therapies, which can neutralize differences of efficacy of first line regimen. Hence, surrogate endpoints, such as progression-free survival (PFS), overall response or complete response rate (ORR, CR) or minimal residual disease (MRD), are commonly used in CLL research. However, data supporting a correlation between surrogate endpoints and OS in the treatment for CLL is scarce. Here, we present a large, systematic analysis of over 4000 patients (pts) treated in studies of the German CLL Study Group (GCLLSG) to determine the adequacy of surrogate endpoints in clinical trials for pts with CLL. Methods We analysed twelve phase II and phase III trials of the GCLLSG spanning from 1999 to 2022. Only pts receiving first-line therapy were included in the analysis. To determine correlation, the Kaplan-Meier survival function was estimated for both OS and PFS for each study separately. From these functions, the probability of survival (OS and PFS) was estimated at different time points. These pairs of time points were then correlated across the studies, with each pair of probabilities weighted by the number of pts in each corresponding study using Spearman's correlation coefficients. Additionally, we compared OS according to MRD and iwCLL response status at end-of-treatment (for fixed-duration regimens) or end-of-induction-treatment (for MRD-guided treatments) (EO[I]T) using Kaplan-Meier functions calculated from the time point of MRD assessment in peripheral blood. EO(I)T ranged from month 9 to month 15 after treatment initiation, depending on the study protocol. Results The full analysis set comprised 4237 pts. The median observation time was 67 months. Median age was 64 years (range 27 - 90). 45.1% of pts were older than 65 years. 3159 pts (74.6%) had received chemo/chemoimmunotherapy (C/CIT), whereas 1078 (25.4%) were treated with targeted therapies. There were 2114 progressive disease (PD) events and 1211 deaths captured within the trial data. The Spearman's Rho of the estimated survival probabilities for the time points 12/24, 24/36, 36/48 and 60/72 months (PFS/OS) was 0.95, 0.96, 0.94 and 0.92, respectively ( Panel A). Focussing on pts with confirmed OS event (n=1211), the Spearman's Rho between PFS and OS in months was 0.66 for pts receiving chemo/chemoimmunotherapy and 0.88 for pts receiving targeted treatment. ORR and CR rate was 89.8% and 31.5%, respectively. Pts with a CR had a significantly longer PFS (HR 0.38, 95% CI 0.34-0.42, p <0.001) and OS (HR 0.39, 95% CI 0.33- 0.46, p <0.001) compared to pts with non-CR. MRD measurements at the EO(I)T were available for 2521 pts (59.5%), of which 39% were treated with targeted agents. Rates of undetectable (<10 -4), intermediate (≥10 -4 and <10 -2) and high MRD (≥10 -2) were 59.1%, 21.7%, 19.3%, respectively, across treatment groups. The median OS for undetectable and intermediate MRD was not reached and 60.7 months for high MRD status ( Panel B). The estimated 60-month OS was 84.6%, 71.3% and 51.1%, respectively. Accordingly, median PFS for uMRD was 61.8 months and 60-month PFS was 52.6%. In pts ≤65 years of age, median OS was not reached with undetectable or intermediate MRD and 70.5 months in pts with high MRD status. In pts >65 years, median OS was not reached for uMRD and was 75.2 and 58.8 months, respectively, for pts with intermediate or high MRD. The corresponding hazard ratios for high versus uMRD status at EO(I)T were 3.33 (95% CI 2.57 - 4.3) in older and 5.89 (4.15 - 8.35) in younger pts. Conclusion In this large analysis of 12 prospective CLL trials, we found a robust correlation between PFS and OS, thereby supporting the use of PFS as a surrogate endpoint in clinical studies. More importantly, the MRD status at EO(I)T was associated with OS across all treatment modalities, demonstrating its utility in identifying pts with potentially dismal outcomes and providing a rationale for MRD-guided treatment strategies.