Background: Metastasis is a major cause of mortality in cancer. Identifying prognostic factors that distinguish patients who will experience metastasis in the short-term and those that will be free of metastasis in the long-term is of particular interest in current medical research. The microsatellite instability (MSI)-high (MSI-H) tumor phenotype is such a differentiator in colorectal cancer, as patients with these tumors tend not to experience metastasis. Yet, patients with MSI-low (MSI-L) or microsatellite stable (MSS) tumors may still be susceptible to develop metastasis post-diagnosis, suggesting the existence of other unknown factors. Our aim in this study was to examine whether genetic variations could differentiate patients based on the risk and/or timing of metastasis in the MSI-L/MSS colorectal cancer patient subgroup.

Methods: The patient cohort consisted of 379 stage I-III colorectal cancer patients with MSS or MSI-L tumors. We performed univariable analysis on genome-wide SNP genotype data (810,622 SNPs; minor allele frequency >1%) under different genetic models. Depending on the long-term metastasis free probability estimates for the genotype categories, we applied the mixture cure model or Cox proportional hazards (PH) regression model, as appropriate. The models tested the following associations with each SNP: risk of metastasis and time-to-metastasis in mixture cure model, and time-to-metastasis in Cox PH. For SNPs reaching genome-wide significance (p<6e-8), the validity of the genetic models were verified. The SNPs whose genetic models were valid were analyzed in an appropriate multivariable model adjusting for significant baseline characteristics.

Results: After adjusting for significant baseline characteristics, specific genotypes of ten polymorphisms were significantly associated with time-to-metastasis. A common intergenic variant was the SNP most significantly associated with time-to-metastasis (HR=15.86, 95% CI=6.83-36.83; p-value=1.3e-10). This association was detected by the mixture cure model. In addition, the Cox PH model identified nine SNPs significantly associated with time-to-metastasis (p-values were between 9.6e-10 and 4.9e-08). These ten SNPs are not located in coding regions and have not been previously reported to be associated with metastasis in colorectal cancer. However, some of these SNPs are located in genomic regions that have been shown to be associated with invasive properties in select epithelial cancer cell lines.

Summary: This is the first study to investigate genetic associations with time-to-metastasis in colorectal cancer patients using such a large set of genotype data. In addition, the consideration of a detailed study design ensured this analysis is comprehensive and inclusive. Our results suggest novel associations of genetic variations with time-to-metastasis in colorectal cancer patients that may be influential in patient outcomes. These associations, once replicated in other patient cohorts, could provide prognostic biomarkers for early metastasis in colorectal cancer patients and contribute to personalized treatment strategies for this patient subgroup.