Meeting Brochure and registration form      SMDM Homepage

Tuesday, October 23, 2007 - 8:30 AM
E-1

EVALUATION OF PROGRESSION-FREE SURVIVAL AS A SURROGATE ENDPOINT IN ADVANCED BREAST CANCER

Rebecca Miksad, MD1, Vera Zietemann, MPH, PhD2, Raffaella Matteucci Gothe2, Ruth Schwarzer2, Annette Conrads-Frank2, Petra Schnell-Inderst2, Bjorn Stollenwerk2, Petra Plieschnegger2, Eva Esteban2, and Uwe Siebert3. (1) Institute for Technology Assessment and Beth Israel Deaconess Medical Center, Boston, MA, (2) University for Health Sciences, Medical Informatics and Technology, Hall, Austria, (3) Institute for Technology Assessment, Boston, MA and University for Health Sciences, Medical Informatics and Technology, Hall, Austria

Purpose: As a potential surrogate endpoint, progression free survival (PFS) allows efficacy assessment when overall survival (OS) data are not available. However, the association between PFS and OS has not been rigorously evaluated in metastatic breast cancer (MBC) for anthracycline-based (A) or taxane-based (T) chemotherapy. We assessed the association between PFS and OS using meta-analytic approaches.

Methods: After performing a literature review of randomized, controlled A and T trials for MBC, progression-based endpoints were classified by prospective definitions. Assuming a constant rate, treatment effects were derived as hazard ratios (HR) for PFS (HRPFS) and OS (HROS). Overall agreement between HRs was assessed with kappa statistics. Fixed effects models to predict HROS based on HRPFS were developed and internally validated. We performed sensitivity and subgroup analyses for the constant rate assumption, PFS definition, year of last patient recruitment, and line of treatment.

Results: 15 A trials and 16 T trial fulfilled the inclusions criteria, allowing 17 A comparisons (n=4155) and 17 T comparisons (n=5509). Progression-based endpoints in 40% (T) and 46% (A) of trials were re-categorized by prospective definitions. The HROS and HRPFS agreed in a negative direction in 25% (A) to 50% (T) and in a positive direction in 62.5% (A) to 50%(T) (kappa=0.71, p=0.0029 (A) and kappa=0.75, p=0.0028 (T)). In the fixed effects models, HRPFS was a significant predictor of HROS for A (p=0.0019) and T (p=0.012). Explained variance (R2) was 0.35 (T) and 0.49 (A). In cross validation, 97% of 95% prediction intervals crossed the equivalence line, with 82% (A) and 76% (T) of predicted HROS point estimates agreeing in the direction of the observed HROS. Results were robust in sensitivity and subgroup analysis.

Conclusions: Based on this meta-analysis of A and T trials for MBC, the treatment effect on PFS is significantly associated with the treatment effect on OS. However, predictions based on PFS are surrounded with uncertainty. Half (A) to one third (T) of the variance in the treatment effect on OS is explained by the variance in the treatment effect on PFS. Using limited data, heterogeneity of results could not be explained by the constant rate assumption, PFS definition, year of last patient recruitment or line of therapy. Adherence standardized definitions may increase the reliability and validity of PFS data.


See more of Concurrent Abstracts E: Cost-effectiveness Analysis: Theory
See more of The 29th Annual Meeting of the Society for Medical Decision Making (October 20-24, 2007)