OBJECTIVES: To explore the validity of surrogate endpoints in subgroups of patients with advanced colorectal cancer (aCRC), defined by the KRAS biomarker status.

METHODS: A systematic review was undertaken to identify all randomised controlled trials (RCTs) of any pharmacological therapies for aCRC that were published between 2003 and 3^rd April 2020 (PROSPERO registration CRD42020167075). Trials included were those reporting treatment effect on at least two of the three outcomes: tumour response (TR), progression-free survival (PFS) and overall survival (OS) by KRAS status. were performed for each pair of outcomes (TR as surrogate endpoint for PFS, TR for OS and PFS for OS) for KRAS wild-type (WT), KRAS mutant-type (MT) and both KRAS WT and MT subgroups combined.

RESULTS: In preliminary results 19 RCTs were identified that reported treatment effects by KRAS status. 8 of the RCTs reported outcomes for both WT and MT KRAS patients (16 groups of patients), 10 recruited only KRAS WT patients, and 1 recruited only KRAS MT patients, resulting . The surrogate relationship (quantified by the between-studies correlation) between the treatment effects on TR and PFS for KRAS WT and MT combined was -0.69 (95% CrI: -0.92, -0.24) from 22 trials-subgroups, and for KRAS WT patients -0.66 (95% CrI: -0.95, 0.02) from 14 trials. The correlation was much lower for TR and PFS in the KRAS MT subgroup (-0.22 (95% CrI: -0.87, 0.76) from 8 trials) possibly due to fewer data available. There were no differences in surrogacy patterns between the biomarker groups for the PFS/OS and TR/OS pairs of outcomes.

CONCLUSIONS: There is some evidence that the surrogate relationship between treatment effects on TR and PFS may depend on the KRAS biomarker status. Results based on the data from the full systematic review will be presented.