DNA Methylation Signatures Distinguish Breast Cancer Histological Subtypes and Predict Hormone Receptor Status: A TCGA-Based Integrative Analysis

Abstract

Objective: DNA methylation alterations contribute to breast cancer heterogeneity, yet their utility for distinguishing histological subtypes remains underexplored. This study aimed to identify genome-wide methylation signatures across breast cancer subtypes and develop compact probe panels for molecular classification. Materials and Methods: HumanMethylation450 (HM450) data from 99 samples (40 infiltrating ductal carcinoma (IDC), 40 infiltrating lobular carcinoma (ILC), 14 metaplastic, 5 medullary) and 1,098 clinical samples from The Cancer Genome Atlas (TCGA) breast invasive carcinoma (BRCA) were analyzed. Differentially methylated positions (DMPs) were identified using limma with Benjamini-Hochberg false discovery rate correction. Support vector machine classifiers with minimal probe panels were developed for hormone receptor endpoints. Results: Of the 404,059 quality-filtered CpG probes, 88,909 (22.0%) were differentially methylated across subtypes (False discovery rate (FDR)<0.05). The ILC-metaplastic comparison yielded the largest divergence (58,661 DMPs). Classifiers achieved area under the curve (AUC) values of 0.935, 0.934, and 0.866 for estrogen receptor (ER), triple-negative breast cancer (TNBC), and progesterone receptor (PR) status, respectively, using 8-bit polymerase chain reaction (PCR). Fifteen CpG probes in internal cross-validation. Conclusion: Histological subtypes harbor distinct methylation landscapes dominated by global hypomethylation in metaplastic carcinoma. Compact CpG probe panels predict hormone receptor status with high accuracy in internal cross-validation, suggesting the potential of methylation-based diagnostic assays as a complementary approach to immunohistochemistry (IHC), pending external validation.