多族裔人群中胰腺癌风险的前瞻性全表观基因组关联研究

Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal cancer, with known differences in incidence across populations in the US. Epigenome-wide association studies (EWAS) have identified DNA methylation markers of PDAC risk, but have been mostly performed in European ancestry populations with small sample sizes and post-diagnostic samples. Prospective EWAS in diverse populations may reveal novel biomarkers and provide insights into racial/ethnic differences in PDAC biology. Methods: Using a nested case-control design, we conducted an EWAS of PDAC risk in the Multiethnic Cohort, a large prospective cohort of African American, Japanese American, Latino, Native Hawaiian, and White residents of Los Angeles and Hawaii. Incident PDAC cases (N=689) were matched to controls (N=806) on age, sex, race/ethnicity, year of blood draw, and study site. DNA methylation was measured in pre-diagnostic blood using the Illumina EPICv2 array (>930,000 CpG sites). Associations between DNA methylation at each CpG site and PDAC risk was assessed using conditional logistic regression, adjusting for matching factors, diabetes, family history of PDAC, smoking, BMI, and DNA methylation-predicted lymphocyte cell type proportions. Differentially methylated region (DMR) and pathway analyses were further conducted. The most significant results were examined in a prior EWAS of PDAC in the Nurses' Health Study, Physicians' Health Study and Health Professionals Follow-up Study (N=393 cases/431 controls). Results: We identified 105 CpG sites significantly associated with PDAC risk (p<9×10⁻⁸). The top CpG sites were annotated to genes related to tumor suppression, metabolism, and inflammation (e.g., TRAK1 , INPP5A , ACOX3 , WBP1L , RASSF8-AS1) . Most of the top CpG sites exhibited a positive association between DNA methylation levels and PDAC risk (odds ratios [OR] per 0.01 increase in methylation beta value 1.03-2.33). Heterogeneity across race/ethnicity was observed for one site in ZNF713 , which was associated with PDAC in Native Hawaiians only (OR 1.65, 95% CI 1.52-1.80; p-heterogeneity<6.6x10 -25 ). DMR analysis revealed 738 regions associated with PDAC (FDR-adjusted p<0.05), with the top regions annotated to genes related to genetic regulation and cancer progression (e.g. PRMT7 , NDUFC1 , NAA15 , TFDP1). KEGG and GO pathway analyses identified 18 and 388 significantly enriched pathways, respectively (p < 0.05); the top pathways were DNA replication and catalytic complex. In the validation analyses, none of the 105 sites reached significance (p<0.01), but 26 showed the consistent direction of association. Conclusions: In this multiethnic population, differential methylation in genes related to cancer, inflammation, and metabolism were associated with PDAC risk. Further studies integrating larger discovery and validation data are warranted to elucidate underlying mechanisms across populations.