Detecting Breast Cancer Risk Years Before Diagnosis via cfDNA Methylation

Pre-Diagnosis Plasma Analysis Reveals Early Regulatory Changes

Investigators analyzed genome-wide cell-free DNA (cfDNA) methylation patterns using prospectively collected plasma samples from individuals who were cancer-free at the time of blood collection but later developed breast or prostate cancer. The research team, which included Nicholas Cheng, Tom W. Ouellette, and Kimberly Skead, utilized biospecimens from the Ontario Health Study. This prospective population cohort collected biological samples and health information from more than 40,000 participants. By linking the cohort with the Canadian Cancer Registry, researchers identified individuals who developed breast or prostate cancer anywhere from a few weeks to up to nine years after plasma collection. Cancer-free controls were matched based on age, sample timing, smoking history, and alcohol consumption. The final analysis included 491 plasma samples: 171 incident breast cancer cases, 93 incident prostate cancer cases, and 227 cancer-free controls. Characterization of genome-wide plasma cfDNA methylation was performed using cell-free methylated DNA immunoprecipitation sequencing, or cfMeDIP-seq. Because 67.8% of the incident breast cancers were diagnosed at stage I and almost 89% of participants in both the case and control groups had undergone a mammogram prior to donating blood, the breast cancer cohort was heavily skewed toward screen-detected early-stage disease.

Enhancer Methylation Patterns and Biological Pathways

Rather than focusing exclusively on individual cancer-associated genes, researchers examined methylation patterns across the genome. Differentially methylated regions identified prior to diagnosis mapped frequently to promoters, enhancers, silencers, and repetitive genomic elements, with approximately 37.5% to 51.2% of the leading regions located in those regulatory areas. For breast cancer, the most informative signals emerged from hypermethylated enhancer regions. Using a discovery cohort made up of 99 controls and 99 breast cancer cases, researchers built a penalized logistic-regression classifier that focused on the top 90 hypermethylated enhancer regions. Across diverse ages, pre-diagnosis intervals of up to five years, and various breast cancer subtypes, that model attained a C-index of 0.61 and a cross-validated AUROC of 0.62. However, performance dropped in an independent held-out cohort comprising 72 cases and 44 controls. The biological pathways linked to these differentially methylated regions involved DNA repair, hypoxia, P53 signaling, hormonal regulation, cell growth, and immune-related processes. According to the study authors, these findings indicate that pre-diagnosis cfDNA contains information originating not only from emerging tumor cells but also from systemic host and immune changes accompanying cancer development.

Limitations for Clinical Translation

While the study demonstrates that molecular changes associated with breast cancer are present in blood years before a tumor becomes clinically apparent, the predictive signals distinguish future breast cancer cases from controls only modestly. Early-stage breast cancer remains one of the most challenging settings for blood-based cancer detection because small, localized tumors typically release minimal tumor-derived DNA into the circulation. Ultimately, the researchers determined that cfDNA methylation acts as a potential adjunct for risk stratification rather than substituting for mammograms or serving as an independent early detection assay.