Progression-free survival as a surrogate for overall survival in first-line chemotherapy for advanced pancreatic cancer

Abstract

Background: Overall survival (OS), as the primary end-point in first-line chemotherapy trials, requires a prolonged follow-up time and may be confounded by subsequent regimens. This study aimed to evaluate the correlation between OS and surrogate end-points (progression-free survival [PFS], response rate and disease control rate), and to identify a potential surrogate for OS in advanced pancreatic cancer.

Methods: Based on an electronic search, we identified randomized controlled phase II and III trials of first-line chemotherapy for advanced pancreatic cancer. Correlation analyses were performed between surrogate end-points and OS, and between improvements in surrogates and those in OS.

Results: Fifty trials (II/II-III/III, 17/2/31) with 111 treatment arms were identified, and 15,906 patients were analysed. PFS was most strongly correlated with OS (correlation coefficient, 0.76). Weighted linear regression models revealed the greatest determinant coefficient of 0.84 between the hazard ratio (HR) of the experimental arms compared with the control arms of PFS and that of OS. The approximate equation was log HROS = 0.01 + 0.77 × log HRPFS, indicating that risk reduction of OS via chemotherapy would translate into a 77% risk reduction of PFS. The surrogacy of PFS for OS was robust throughout our subgroup analyses: e.g., biologic versus non-biologic regimens, locally advanced versus metastatic disease.

Conclusions: The surrogacy of PFS for OS in pancreatic cancer was validated. Therefore, the use of PFS as the primary end-point in clinical trials could facilitate the early introduction of new effective chemotherapy regimens into clinical practice.

Keywords: Drug therapy; End-point determination; Pancreatic neoplasms; Survival; Treatment outcome.