A risk prediction model for colorectal cancer using genome-wide association study-identified polymorphisms and established risk factors among Japanese: results from two independent case-control studies

Most genome-wide association studies of colorectal cancer (CRC) carried out to date have been in populations with European ancestry, and the extent to which the identified variants contribute as predictors of CRC among Japanese populations has not been clarified. We analyzed 23 genetic variants identified in previous genome-wide association studies in a derivation case-control study with 558 cases and 1116 age-matched and sex-matched controls. Six single nucleotide polymorphisms were selected for synthesis of the genetic risk score. A dose-dependent association was observed between CRC risk and genetic risk score, which is the aggregate number of alleles in six selected variants: 8q24 - rs6983267, 15q13 - rs4779584 and rs1696961, 14q22 - rs444435, 16q22 - rs9929218, and 3q26.2 - rs1093599. The c statistic for a model that included the genetic risk score and conventional risk factors was 0.7167, versus 0.7009 with the conventional risk factors only (P=0.0013). This model was evaluated in a replication study with 547 cases and 547 age-matched and sex-matched controls, and the corresponding c statistics were 0.6356 and 0.6391 with no statistical significance. When the two studies were combined, the corresponding c statistics were 0.6132 and 0.6198 (P=0.0126). We developed a risk model that incorporates a genetic risk score and established risk factors, but this model was not satisfactory in the replication study. The results in the combined study still encourage further attempts using a similar approach among individual countries.