Epidemiology and Molecular-Pathologic Characteristics of CpG Island Methylator Phenotype (CIMP) in Colorectal Cancer.

Background: CpG island methylator phenotype (CIMP) forms a distinct epigenetic phenotype in colorectal cancer (CRC). Though associated with distinct clinicopathologic characteristics, limited evidence exists of the association of CIMP with patient's reported lifestyle factors and tumor molecular characteristics. We assessed the associations of these characteristics in a pooled analysis of CRC patients.

Patients And Methods: We pooled data from 3 CRC patient cohorts: Assessment of Targeted Therapies Against Colorectal Cancer (ATTACC), biomarker-based protocol (Integromics), and The Cancer Genome Atlas (TCGA). CIMP was measured using the classical 6-gene methylated-in-tumor (MINT) marker panel (MINT1, MINT2, MINT31, p14, p16, and MLH1) in ATTACC and genome-wide human methylation arrays in Integromics and TCGA, respectively. CIMP-High (CIMP-H) was defined as ≥ 3 of 6 methylated markers in ATTACC. In TCGA and Integromics, CIMP-H group was defined on the basis of clusters of methylation profiles and high levels of methylation in tumor samples. Baseline comparisons of characteristics across CIMP groups (CIMP-H vs. CIMP-0) were performed by Student t test or chi-square test for continuous or categorical variables, respectively. Further logistic regression analyses were performed to compute the odds ratio (OR) of these associations.

Results: Pooled prevalence of CIMP-H was 22% across 3 data sets. CIMP-H CRC tumors were associated with older age at diagnosis (OR, 1.02; 95% confidence interval [CI], 1.01, 1.03), microsatellite instability-high (MSI-H) status (OR, 9.15; 95% CI, 4.45, 18.81), BRAF mutation (OR, 7.70; 95% CI, 4.98, 11.87), right-sided tumor location (OR, 2.40; 95% CI, 1.78, 3.22), poor differentiation (OR, 2.94; 95% CI, 1.95, 4.45), and mucinous histology (OR, 2.47; 95% CI, 1.77, 3.47), as reported previously in the literature. CIMP-H tumors were also found to be associated with self-reported history of alcohol consumption (OR, ever vs. never, 1.58; 95% CI, 1.07, 2.34). Pathologically, CIMP-H tumors were associated with the presence of intraepithelial lymphocytes (OR, 3.31; 95% CI, 1.41, 7.80) among patients in the Integromics cohort.

Conclusion: CIMP-H tumors were associated with history of alcohol consumption and presence of intraepithelial lymphocytes. In addition, we confirmed the previously known association of CIMP with age, MSI-H status, BRAF mutation, sidedness, and mucinous histology. Molecular pathologic epidemiology associations help us explore the underlying association of lifestyle and clinical factors with molecular subsets like CIMP and help guide cancer prevention and treatment strategies.