IDENTIFYING CANDIDATE SURROGATE ENDPOINTS FOR OVERALL SURVIVAL IN RELAPSED/REFRACTORY DIFFUSE LARGE B-CELL LYMPHOMA: ESTABLISHING INDIVIDUAL-LEVEL SURROGACY
Author(s)
Sneha Rai, MSc, Statistics, Mohd Kashif Siddiqui, MBA, MPharm, Alosh Greeny, MPharm, Shivani Mathur, MPharm, Sumaiya Qasim, MPharm, Rajdeep Singh, MS, Suraj Kumar, MPharm, Karan Karan, MS, Jatin Gupta, MPharm.
EBM Health Consultants, Delhi, India.
EBM Health Consultants, Delhi, India.
OBJECTIVES: Overall survival (OS) remains the most clinically meaningful endpoint in relapsed/refractory diffuse large B-cell lymphoma (R/R DLBCL). However, its use may be limited by prolonged follow-up requirements, treatment crossover, and subsequent lines of therapy. Earlier endpoints, including response rate, time-to-progression (TTP), event-free survival (EFS), and progression-free survival (PFS), may serve as candidate surrogate endpoints (CSE) for OS if adequately validated. This study aimed to identify CSE for OS in patients with R/R DLBCL.
METHODS: An existing SLR was used to identify studies of patients with R/R DLBCL. Eligible studies were required to report OS and at least one CSE. Studies with follow-up shorter than 48 months were excluded to ensure adequate maturity of survival outcomes. CSEs included binary response outcomes and time-to-event (TTE) outcomes, including TTP, EFS, and PFS. In line with NICE DSU TSD20 principles, CSEs were first screened using correlation-based analyses. Outcomes with correlation coefficients >0.70 were considered to show strong aggregate association with OS and were taken forward for further assessment. Individual-level association between TTE endpoints and OS was assessed using Kendall’s tau from bivariate copula models. Five candidate marginal distributions were evaluated across four copula families. A Kendall’s tau of ≥0.70 was considered indicative of strong individual-level association.
RESULTS: A total of 95 studies contributed to this analysis. PFS demonstrated strongest association with OS, with correlation coefficient ranging from 0.72-0.83 outperforming all other CSEs. Gaussian copula with Weibull marginal distribution provided the best fit, yielding a Kendall’s tau of 0.75. Across the four copula families, Kendall’s tau ranged from 0.69-0.75, suggesting that association between PFS and OS was not sensitive to copula specification.
CONCLUSIONS: The findings support PFS as a promising CSE for OS in R/R DLBCL. However, trial-level validation is required before PFS can be considered a reliable substitute for OS in comparative treatment-effect estimation.
METHODS: An existing SLR was used to identify studies of patients with R/R DLBCL. Eligible studies were required to report OS and at least one CSE. Studies with follow-up shorter than 48 months were excluded to ensure adequate maturity of survival outcomes. CSEs included binary response outcomes and time-to-event (TTE) outcomes, including TTP, EFS, and PFS. In line with NICE DSU TSD20 principles, CSEs were first screened using correlation-based analyses. Outcomes with correlation coefficients >0.70 were considered to show strong aggregate association with OS and were taken forward for further assessment. Individual-level association between TTE endpoints and OS was assessed using Kendall’s tau from bivariate copula models. Five candidate marginal distributions were evaluated across four copula families. A Kendall’s tau of ≥0.70 was considered indicative of strong individual-level association.
RESULTS: A total of 95 studies contributed to this analysis. PFS demonstrated strongest association with OS, with correlation coefficient ranging from 0.72-0.83 outperforming all other CSEs. Gaussian copula with Weibull marginal distribution provided the best fit, yielding a Kendall’s tau of 0.75. Across the four copula families, Kendall’s tau ranged from 0.69-0.75, suggesting that association between PFS and OS was not sensitive to copula specification.
CONCLUSIONS: The findings support PFS as a promising CSE for OS in R/R DLBCL. However, trial-level validation is required before PFS can be considered a reliable substitute for OS in comparative treatment-effect estimation.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
CO209
Topic
Clinical Outcomes, Health Technology Assessment, Methodological & Statistical Research
Topic Subcategory
Relating Intermediate to Long-term Outcomes
Disease
Oncology