5-hydroxymethylated biomarkers in cell-free DNA predict occult colorectal cancer up to 36 months prior to diagnosis in the PLCO Cancer Screening Trial

Purpose: Using the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial samples, we identified cell-free DNA (cfDNA) biomarker candidate genes bearing the epigenetic mark 5-hydroxymethylcytosine (5hmC) that detect occult colorectal cancer (CRC) up to 36 mo prior to clinical diagnosis. Methods: PLCO study subjects were matched by age, race, and sex as cases (n = 201, diagnosed with CRC within 36 mo of blood collection) and controls (n = 402, no cancer diagnosis on follow-up, average 16.3 years after entering the study). Archived plasma samples (300 µL per study subject) were obtained from the National Cancer Institute (NCI), and we employed the sensitive 5hmC-Seal chemical labeling approach on 3 - 8 ng of extracted cfDNA. Following next-generation sequencing (NGS) and genome-wide mapping of 5hmC, we then conducted association studies and machine-learning modeling to analyze the genome-wide 5hmC profiles within training and validation groups that were randomly selected at a 2:1 ratio. Results: Robust genome-wide 5hmC profiles were successfully obtained from these decades-old samples. Association analyses using the Cox proportional hazards models suggested several epigenetic pathways relevant to CRC development distinguishing cases from controls. A weighted Cox model, comprised of 32-associated gene bodies, showed predictive detection value for CRC as early as 24-36 mo prior to overt tumor diagnosis. Furthermore, a trend for increased predictive power was observed for blood samples collected closer to CRC diagnosis. Notably, the 5hmC-based predictive model showed comparable performance regardless of sex and self-reported race/ethnicity, and significantly outperformed risk factors such as age and obesity assessed as BMI (body mass index). Conclusion: An assay and machine learning modeling of 5hmC epigenetic signals on cfDNA revealed candidate biomarkers and a scoring algorithm with the potential to predict CRC occurrence despite the absence of clinical symptoms or the availability of effective predictors. Developing a minimally-invasive clinical assay that detects 5hmC-modified biomarkers holds promise for improving early CRC detection and ultimately patient outcomes. Future investigations to expand this strategy to prospectively collected samples are warranted. Overall design: To develop our biomarkers and prediction algorithm, we used samples obtained from the PLCO Cancer Screening Trial. Samples were taken from the intervention arm of the colorectal cancer study. "Cases" were defined as individuals who received a colorectal cancer diagnosis within 36 months of sample collection. "Controls" were defined as individuals who remained cancer-free until lost to follow-up. Cases were matched to Controls in a 1:2 case:control ratio and matched by age, sex, and race. Samples were provided with a masked identifier that was later unmasked to the our study team. NCI's Cancer Data Access System (CDAS) only permits that we provide the PLCO ID with the sequencing data; if GEO users wish to analyze the data, they will need to request information about the samples' case vs. control status and subject clinicodemographic information from CDAS. Must contact NCI CDAS to unmask clinicopathologic characteristics