1227Colorectal cancer risk prediction models incorporating lifestyle and biomarker data: Results from the EPIC cohort. (2nd September 2021)

• Record Type: Journal Article
• Authors: Aleksandrova, Krasimira; Reichmann, Robin; Jenab, Mazda; Rinaldi, Sabina; Kaaks, Rudolf; de-Mesquita, Bas Bueno-; Riboli, Elio; Gunter, Marc
• Is Part Of: International journal of epidemiology. Volume 50(2021)Supplement 1
• Publication Date: 2021-09-02
• Subjects: Epidemiology -- Periodicals; 614.4
• DOI: 10.1093/ije/dyab168.027
• Languages: English
• ISSNs: 0300-5771

Abstract

Abstract: Background: Colorectal cancer represents a major public health concern and there is a worrying tendency of increasing incidence rates among younger people in the last decades. Risk stratification of high-risk individuals may aid targeted disease prevention. We therefore aimed to evaluate the predictive value of a wide range of lifestyle and biomarker variables using data within the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Methods: A range of lifestyle, anthropometric and dietary variables in 329, 885 participants in the EPIC cohort were evaluated as potential predictors for risk of colorectal cancer over 10 years. Biomarker measurements of 41 parameters were available for 1, 320 CRC cases and 1, 320 controls selected using incidence density matching. Best sets of predictors were selected using elastic net regularization with bootstrapping. Random survival forest was applied as a novel technique to validate the set of selected predictors taking variable interactions into account. Results: The results suggested a set of lifestyle factors including age, waist circumference, height, smoking, alcohol consumption, physical activity, vegetables, dairy products, processed meat, and sugar and confectionary that showed good discrimination (Harrell's C-index: 0.710) and excellent calibration. The analyses further revealed a set of biomarkers that increased the predictive performance beyond age, sex and lifestyle factors. Conclusions: Risk prediction models based on lifestyle and biomarker data may prove useful in the identification of individuals at high risk for colorectal cancer. Key messages: Risk prediction models incorporating lifestyle and biomarker data could contribute to developing strategies for targeted colorectal cancer prevention.