The age-standardized mortality rate of colorectal cancer (CRC) has gradually decreased in high-income countries owing to increased uptake of CRC screening, while the incidence of early-onset CRC to be diagnosed to younger individuals than 50 years, has increased worldwide [1], [2], [3]. The CRC screening modalities differ by country; however, the approach is generally “One-Size-Fits-All” using age as the major criterion to start screening [4]. Taken together with the accumulated evidence that CRC is attributed to environmental and genetic risk factors [5], a precision approach to stratify individuals' risk utilizing information on their lifestyle and the human genome and to intervene according to their risk is more promising than the traditional approach.

Risk prediction models that could estimate the absolute risk of future disease and identify high-risk individuals for certain diseases have been extensively proposed. Well-validated CRC risk prediction models exist in Western countries [6], [7], [8]; however, these models have only utilized information on lifestyle and medical history, excluding individual genetic risk. Additionally, controversies regarding the use of genetic risk score (GRS), which aggregates the genetic effects of disease-associated common variants with genome-wide significance, in the CRC prediction model to improve predictive performance exist [9], [10], [11], [12], [13], [14].

Recently, genome-wide polygenetic risk scores (PRSs) that include thousands to millions of common variants, not limited to genome-wide significant susceptibility variants, have demonstrated improved predictive performance for many common diseases [15], [16], [17]. Genome-wide PRSs have predominantly been studied in Europeans and have shown poor predictive performance in other populations owing to differences in genetic architecture between populations [18], [19]. However, few studies have evaluated the CRC risk prediction model combining genome-wide PRS with lifestyle factors in Asian populations. Considering that the associations of lifestyle factors with CRC risk were not clear in Asian women, CRC risk prediction model incorporating a genome-wide PRS is expected to perform better than previous models with only lifestyle factors in Asian populations. Therefore, the development of genome-wide PRS for CRC and the evaluation of the CRC risk prediction model integrating genome-wide PRS in Asians are needed.

Hence, we developed 31 genome-wide PRSs for CRC using two methods and evaluated CRC risk prediction models combining lifestyle factors with genome-wide PRS in Japanese men and women.