Correlation between alternative endpoints and overall survival in metastatic colorectal cancer patients eligible to a maintenance strategy

Background: In this study, our first aim was to assess the relation between different intermediate criteria and overall survival (OS) in patients treated for a metastatic colorectal cancer (mCRC) receiving a first-line chemotherapy associated with bevacizumab. Methods: We collected retrospective data from patients with mCRC treated with 1st-line chemotherapy (generally FOLFIRI-FOLFOX regimen) plus bevacizumab between January 2006 and December 2012. We assessed the following time to event endpoints: OS; progression free survival (PFS); duration of disease control (DDC), the sum of the periods in which the disease did not progress, and the time to failure of strategy (TFS), the whole period before the introduction of second-line treatment. Linear correlation and linear regression models were used to study relations between OS and TFS, and OS and DDC, respectively. Prentice criteria for surrogacy were investigated. Results: We included 216 patients from 6 centers. 91 (42%) patients received a maintenance strategy. With a median follow-up of 57.6 (43.6-94.0) months, median OS was 24.5 (21.3-29.7) months, median PFS was 8.9 (8.4-9.7) months, median DDC was 11.0 (9.8-12.4) months, and median TFS was 11.1 (10.0-13.0) months. Pearson coefficient (coeff ) was 0.79 (CI 95% 0.73-0.83) and determination coeff was 0.62, suggesting of a satisfactory correlation between OS and DDC. Similarly, the correlation between OS and TFS was rather satisfactory with a Pearson coeff at 0.79 (CI 95% 0.73-0.84) and a determination coeff at 0.63. Linear regression analysis showed a significant association between OS and DDC, and between OS and TFS, respectively. These relationships can be modeled by the formulas: cube root (SG) = 0.9547 + 0,8286 cubic root (DDC) and cubic root (SG)= 0.9655 +0,8186 cubic root (TFS). Considering resection of metastases as the treatment, Prentice criteria were verified for both TFS and DDC. Conclusions: DDC and TFS were correlated to OS and Prentice criteria were validated for both endpoints. This makes them relevant as intermediate criteria, in the setting of patients with mCRC treated with 1st-line bevacizumab-based regimen.