More Sensitive Clinical Endpoints

A situation that has attracted a lot of attention over the last decade is the identification of surrogates when the ultimate endpoint of interest is survival, as is frequently the case in advanced forms of cancer. Overall survival (OS) requires prolonged follow-up, except in fast progressing tumor types, and treatment effects on OS are diluted by competing causes of death, especially in elderly patient populations. In addition, the effect of a new agent on OS may be confounded by the further lines of treatment a patient is likely to receive after failing on this agent. For these reasons, progression-free survival (PFS) is often used as the primary endpoint in clinical trials, with OS being a secondary (but crucially important) endpoint. PFS is often defined as the time from randomization to an increase by more than 25% in the largest dimension of the tumor(s). The treatment effects on PFS tend to be larger than those on OS (Ciani et al., 2013). However, PFS is a controversial endpoint because some new agents (e.g., anti-angiogenic agents in advanced breast cancer) have been shown to have a marked effect on PFS but no sizable benefit on OS. The meta-analytic approach to surrogate endpoint evaluation has been extensively used to investigate PFS as a surrogate endpoint for OS, with diverging results. PFS was found to be a poor surrogate for OS in advanced breast cancer (Burzykowski et al., 2008). In contrast, PFS was a good surrogate in advanced ovarian cancer, using the dataset of Section 2.2.3 (Burzykowski et al., 2001). PFS also appeared to be a good surrogate for OS in advanced colorectal cancer treated with fluoropyrimidines, using the dataset of Section 2.2.4 (Buyse et al., 2007). However, PFS was not as good a surrogate in colorectal cancer treated with more recent therapies, perhaps because of the much larger number of lines of active therapies patients currently receive in this disease (Shi et al., 2015; Ciani et al., 2015). The value of PFS as a surrogate for OS seemed questionable in advanced non-small cell lung cancer (Laporte et al., 2013), and also in advanced cancer, using the dataset of Section 2.2.6 (Paoletti et al., 2013). All in all, PFS cannot be assumed a priori to be a good surrogate for OS in advanced solid tumors; each situation has to be assessed on its own merits (Ciani et al., 2014).

It is worth noting that in patients with less advanced tumors that can be resected surgically, disease-free survival (DFS) is a very good surrogate in every tumor type for which formal surrogacy analyses have been carried out, e.g., colon cancer (Sargent et al., 2005), operable or locally advanced head and neck cancer (Michiels et al., 2009), lung cancer (Mauguen et al., 2013), and gastric cancer, using the dataset of Section 2.2.6 (Oba et al., 2013). Taken together, these findings suggest that in early forms of cancer that are amenable to local treatment, DFS can be used as a reliable surrogate for OS. In contrast, in more advanced forms of cancer that are no longer amenable to local treatment, PFS cannot be used as a reliable surrogate for OS. These contrasting findings for DFS and PFS may seem counterintuitive (Buyse et al., 2016). If anything, one would expect PFS to be a better surrogate for OS than DFS, because the time between tumor progression and death is much shorter in advanced disease than after surgical resection of the tumor. However, from a biological standpoint, the reappearance of a tumor after a long disease-free period may be a far more consequential event than an increase in size of a measurable tumor mass.

< Prev   CONTENTS   Source   Next >