CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: A screening study for identification of anticancer effects of prescription drugs Authors:  Klaus Kaae Andersen - Danish Cancer Society Research Center (Denmark) [presenting]
Abstract: Cancer incidence is increasing on a global scale and cancer mortality is still high for many cancer types. A promising approach to meet the need for improved cancer therapy is repurposing of drugs initially marketed for non-cancer indications. We present preliminary results from an ongoing project where the aim is to search for new potential drug candidates for repurposing, by mounting a systematic screening for prescription drugs that might be associated with improved prognosis among cancer patients. We utilize Danish nationwide registries with continuous long-term data on drug prescriptions, cancer diagnoses, and mortality outcomes. We present an approach to set up the search in an automated way with procedures for ensuring reproducibility of results. Specifically, we use separate Cox models for each drug/cancer combination to estimate harm/benefit of the drug on cancer-specific mortality, while applying propensity scores for confounder adjustment. We adjust for multiple testing of main effects, and in evaluation of effect modification by treatment, relevant effects are filtered by $p$-values of corresponding main effects. All potential signals are evaluated for meaningful dose-response relationships along with calculation of prognostic indices, $C$-indices and Kaplan-Meier curves for risk groups. External validation is conducted in independent cohorts.