Association between progression-free survival and overall survival in mantle cell lymphoma: A trial-level surrogate endpoint analysis

Abstract Introduction: Survival outcomes of patients with mantle cell lymphoma (MCL) have substantially improved with the advent of novel therapies, including CD20 monoclonal antibodies, Bruton tyrosine kinase inhibitors, and CAR T-cell therapy. Given these advances, earlier endpoints in clinical trials such as progression-free survival (PFS) are frequently employed as a surrogate endpoint for overall survival (OS) to accelerate regulatory approval for new therapies and limit the costs of drug evaluation. No prior analysis, however, has systematically evaluated if PFS is a valid surrogate for OS in MCL trials. This study provides the first such analysis, assessing the strength of PFS as a surrogate endpoint for OS in MCL clinical trials. Methods: A systematic search of the database at ClinicalTrials.gov was conducted to identify all phase III randomized clinical trials in MCL from database inception through June 2025. Trials were included for analysis if they reported hazard ratios (HRs) for both PFS and OS. To assess the association of HRs between PFS and OS, weighted linear regression analysis was performed using sample size as the weighting factor. For noninferiority trials, reciprocal HRs were used to maintain consistency in the definitions of reference groups (standard of care) and comparator groups (experimental therapy). All HRs were log-transformed to facilitate linear modeling. The coefficient of determination (R²) was used to evaluate the strength of the association, with thresholds defined a priori: R² > 0.80 was considered a strong association, R² between 0.60 and 0.80 a moderate association, and R² < 0.60 a weak association. Results: The initial search yielded 89 phase III trials. Of these, 15 trials comprising 18 treatment comparisons reported HRs for PFS and OS and were included in the final analysis. The included trials involved a total of 5,630 participants. Twelve of the comparisons (66.7%) were in the first-line setting and 6 (33.3%) were in the relapsed/refractory setting. Regarding phase of treatment, 5 treatment comparisons (27.8%) exclusively assessed induction therapy, 4 (22.2%) focused solely on maintenance therapy, 8 (44.4%) assessed therapies that involved both induction and maintenance phases, and 1 (5.6%) evaluated consolidation therapy alone. Of the 16 treatment comparisons that provided confidence intervals for both PFS and OS HRs, 7 (43.8%; 2,095 participants) demonstrated concordance in statistical significance between PFS and OS. The remaining 9 comparisons (56.3%; 2,718 participants) all showed a statistically significant benefit in PFS without an accompanying benefit in OS. The overall weighted correlation coefficientbetween log HRs for PFS and OS was 0.87, with a corresponding R2 of 0.75 (p<0.001). This indicates that 75% of the variance in OS could be explained by changes in PFS, classifying the association between PFS and OS as moderate in strength according to predefined thresholds. In stratified analyses, treatment comparisons in the first-line setting (R2=0.81, p<0.001) and the relapsed/refractory setting (R2=0.79, p=0.04) demonstrated a similar strength of surrogacy of PFS for OS. Comparisons evaluating therapies spanning from induction to maintenance phases of treatment showed a moderate association between PFS and OS HRs (R2=0.73, p=0.01). No significant association was observed in comparisons exclusively assessing induction therapy (R2=0.32, p=0.32) and maintenance therapy (R2=0.83, p=0.09). Discussion: In this first surrogacy analysis of PFS in MCL, the findings suggest that PFS is a moderate surrogate for OS in MCL clinical trials. While the continued use of PFS as a surrogate endpoint appears justified, caution is warranted, especially when treatment effect sizes are modest.