Background: Assessment of breast cancer (BC) risk generally relies on mammography, family history, reproductive history, and genotyping of major mutations. However, assessing the impact of environmental factors, such as lifestyle, health related behavior or external exposures, is still challenging. DNA methylation (DNAm), capturing both genetic and environmental effects, presents a promising opportunity. Previous studies have identified associations and predicted the risk of BC using DNAm in blood, however, these studies did not distinguish between genetic and environmental contributions to these DNAm sites. In this study, associations between DNAm and BC are assessed using paired twin models, which control for shared genetic and environmental effects, allowing testing for associations between DNAm and non-shared environmental exposures and behavior. Results: Pre-diagnosis blood samples of 32 monozygotic (MZ) and 76 dizygotic (DZ) female twin pairs discordant for BC were collected at the mean age of 56.0 years, with the mean age at diagnosis 66.8 and censoring 75.2 years. We identified 212 CpGs (p<6.4*10^-8) and 15 DMRs associated with BC risk across all pairs using paired Cox proportional hazard models. All but one of the BC risk associated CpGs were hypomethylated and 198/212 CpGs had their DNAm associated with BC risk independent of genetic effects. According to previous literature, at least five of the top CpGs were related to estrogen signaling. Following a comprehensive two-sample Mendelian Randomization analysis, we found evidence supporting a dual causal impact of DNAm at cg20145695 (gene body of NXN, rs480351) with increased risk for estrogen-receptor positive BC and decreased risk for estrogen-receptor negative BC. Conclusion: While causal effects of DNAm on BC risk are rare, most of the identified CpGs associated with the risk of BC appear to be independent of genetic effects. This suggests that DNAm could serve as a valuable biomarker for environmental risk factors for BC, and may offer potential benefits as a complementary tool to current risk assessment procedures.

中文翻译：

---

乳腺癌不一致的双胞胎的预诊断血液 DNA 甲基化分析表明环境风险的重要性

背景：乳腺癌 (BC) 风险的评估通常依赖于乳房 X 光检查、家族史、生殖史和主要突变的基因分型。然而，评估生活方式、健康相关行为或外部暴露等环境因素的影响仍然具有挑战性。DNA 甲基化 (DNAm) 可以捕捉遗传和环境影响，提供了一个充满希望的机会。先前的研究已经利用血液中的 DNAm 确定了相关性并预测了 BC 的风险，然而，这些研究没有区分遗传和环境对这些 DNAm 位点的贡献。在这项研究中，使用配对双胞胎模型评估 DNAm 和 BC 之间的关联，该模型控制共享的遗传和环境影响，从而可以测试 DNAm 与非共享环境暴露和行为之间的关联。结果：收集了 32 对 BC 不一致的同卵 (MZ) 和 76 对双卵 (DZ) 女性双胞胎的诊断前血样，平均年龄 56.0 岁，诊断时平均年龄 66.8 岁，审查时平均年龄 75.2 岁。我们使用配对 Cox 比例风险模型，在所有对中确定了 212 个 CpG (p<6.4*10^-8) 和 15 个与 BC 风险相关的 DMR。除 1 个与 BC 风险相关的 CpG 外，所有 CpG 均低甲基化，并且 198/212 个 CpG 的 DNAm 与 BC 风险相关，且与遗传效应无关。根据之前的文献，至少有五个顶级 CpG 与雌激素信号传导有关。经过全面的双样本孟德尔随机化分析，我们发现证据支持 cg20145695（NXN 基因体，rs480351）处的 DNAm 具有双重因果影响，增加雌激素受体阳性 BC 的风险，降低雌激素受体阴性 BC 的风险。结论：虽然 DNAm 对 BC 风险的因果影响很少见，但大多数已确定的与 BC 风险相关的 CpG 似乎与遗传效应无关。这表明 DNAm 可以作为 BC 环境风险因素的有价值的生物标志物，并且可以作为当前风险评估程序的补充工具提供潜在的好处。