DNA Methylation Biomarkers Improve Colorectal Cancer Survival Prediction

DNA methylation-derived biomarkers called Protein EpiScores may improve the accuracy of disease-free and overall survival prediction in patients with colorectal cancer (CRC), compared with traditional clinical risk factors alone, suggest results of a prospective study. The findings were published in Clinical Epigenetics by researchers from Moffitt Cancer Center, Tampa, Florida.

The present study involved 136 patients with newly diagnosed CRC from the prospective ColoCare Study. For each patient, the investigators recorded 107 Protein EpiScores from pretreatment whole blood samples. Disease-free and overall survival were monitored over a median follow-up of 7.3 years and as long as 13.8 years. During follow-up, 26% of patients experienced disease recurrence, and 35% died.

The investigators compared the predictive power of the Protein EpiScores vs traditional clinical risk factors for disease-free and overall survival. The standard factors included tumor stage, age at cancer diagnosis, sex, body mass index, race, and tumor location.

After adjusting for confounding variables, the HCII, VEGFA, CCL17, and LGALS3BP Protein EpiScores were each independently associated with worse disease-free survival, with hazard ratios ranging from 1.62 to 1.71. Adding these scores to the clinical model improved the concordance index (C-index) from 0.64 to 0.70.

The LGALS3BP Protein EpiScore was also independently linked to overall survival, with a hazard ratio of 1.80. Adding this score to the model raised the C-index from 0.70 to 0.75.

The HCII, LGALS3BP, MMP12, and VEGFA Protein EpiScores were tied to both disease-free and overall survival with hazard ratios above 1.50.

The 6-point improvement for recurrence (C-index 0.64 to 0.70) resulted in 34% of patients being reclassified into more accurate risk categories. In cancer care, even incremental gains matter if they prevent undertreating high-risk patients or overtreating low-risk ones.

Protein EpiScores were developed based on previous work suggesting that DNA methylation profiles may improve disease prediction based on circulating proteins (eg, C-reactive protein) and physiologic traits (eg, smoking status) beyond directly measuring those same variables. "Protein EpiScores may therefore represent a complementary class of biomarker to direct measurements," the investigators wrote.

Previous studies have evaluated epigenetic clocks, which are derived from DNA methylation profiles, as markers for CRC risk. However, these clocks cannot pinpoint specific biological drivers of cancer progression. Although Protein EpiScores have helped uncover biological processes driving various conditions such as cardiovascular disease and cancer, this is the first study to evaluate them specifically in the context of cancer survival.

The present findings point to key biological pathways, such as those involving immune suppression and coagulation, which may be driving worse outcomes in patients with CRC. "The immediate value of our findings is highlighting biological pathways like immune suppression and coagulation as drivers of poor outcomes," a senior author from Moffitt Cancer Center said. This information can guide basic scientists and mechanistic studies to identify potential therapeutic targets.

Although Protein EpiScores require further validation before they are ready for clinical use, researchers indicated that if the findings are validated in other epidemiologic settings, these Protein EpiScores could eventually complement existing risk tools, but clinical implementation is realistically several years away. "We see these current findings more as a research tool that requires validation in larger cohorts before clinical use," a senior author stated.

Future studies may also need to recruit a more diverse patient population, given the present cohort was 93% White.