The Validity of Surrogate Endpoints in Sub Groups of Metastatic Colorectal Cancer Patients Defined by Treatment Class and KRAS Status

SIMPLE SUMMARY: When evaluating new cancer therapies in clinical trials, it may take a long time to estimate their effectiveness on overall survival, an outcome typically of main interest to regulatory decision-makers. To expedite access to new therapies for patients, regulatory agencies often make their decisions based on treatment effectiveness measured on surrogate outcomes; for example looking at the impact of treatment on delaying cancer recurrence, which can be measured earlier. For such decisions to be robust, a surrogate endpoint needs to be a valid predictor of overall survival. The validation can be complex and previous research in advanced colorectal cancer has suggested that the validity of a surrogate endpoint may depend on treatment class. We have investigated this and our results indicated that the validity of surrogate endpoints is stronger within some treatment classes compared to when ignoring the treatment class. Surrogate’s validity needs careful consideration to ensure appropriate regulatory decisions. ABSTRACT: Background and Aim: Findings from the literature suggest that the validity of surrogate endpoints in metastatic colorectal cancer (mCRC) may depend on a treatments’ mechanism of action. We explore this and the impact of Kirsten rat sarcoma (KRAS) status on surrogacy patterns in mCRC. Methods: A systematic review was undertaken to identify randomized controlled trials (RCTs) for pharmacological therapies in mCRC. Bayesian meta-analytic methods for surrogate endpoint evaluation were used to evaluate surrogate relationships across all RCTs, by KRAS status and treatment class. Surrogate endpoints explored were progression free survival (PFS) as a surrogate endpoint for overall survival (OS), and tumour response (TR) as a surrogate for PFS and OS. Results: 66 RCTs were identified from the systematic review. PFS showed a strong surrogate relationship with OS across all data and in subgroups by KRAS status. The relationship appeared stronger within individual treatment classes compared to the overall analysis. The TR-PFS and TR-OS relationships were found to be weak overall but stronger within the Epidermal Growth Factor Receptor + Chemotherapy (EGFR + Chemo) treatment class; both overall and in the wild type (WT) patients for TR-PFS, but not in patients with the mutant (MT) KRAS status where data were limited. Conclusions: PFS appeared to be a good surrogate endpoint for OS. TR showed a moderate surrogate relationship with PFS and OS for the EGFR + Chemo treatment class. There was some evidence of impact of the mechanism of action on the strength of the surrogacy patterns in mCRC, but little evidence of the impact of KRAS status on the validity of surrogate endpoints.