Longitudinal analysis of healthy colon establishes aspirin as a suppressor of cancer-related epigenetic aging

Abstract

Background: Colon cancer (CC) is the third most common cancer worldwide, highlighting the importance of developing effective prevention strategies. Accumulating evidence supports that aspirin use reduces CC incidence. We reported previously that aspirin suppresses age-associated and CC-relevant DNA methylation (DNAm) in healthy colon. Here we addressed the aspirin's effectiveness in longitudinal cohort.

Methods: We measured genome-wide DNAm in 124 healthy normal mucosa samples taken at baseline (time point 1, t1) and after 10-years follow-up (time point 2, t2) from a longitudinal female screening cohort. We investigated the time-dependent methylation drift in aspirin users and nonusers using multivariable regression and related the modulatory effect of aspirin to colonic epigenome-aging and CC.

Results: Over time, compared to nonusers, long-term (≥ 2 years) aspirin users showed less hypermethylated CpGs (proximal: 17% vs. 87%; distal: 16% vs. 70%) and more hypomethylated CpGs (proximal: 83% vs. 13%; distal: 84% vs. 30%). Overall, users showed 2% (P = 0.02) less mean methylation levels than nonusers in proximal colon and displayed repressed methylation age (mAge). Methylation loss in users occurred at several CC-specific tumor suppressors that gained methylation in nonusers. Methylation loss in users effected genes involved in immune system and inflammation, while methylation gain in nonusers effected genes involved in metabolism.

Conclusions: This is the first longitudinal study demonstrating effectiveness of aspirin-use in suppression of age-related and CC-relevant hypermethylation in the normal colon. These findings provide a rationale for future studies to evaluate loci that may serve as markers to identify individuals that will benefit most from aspirin and hence increase its efficiency in CC prevention and therapy.

Keywords: Aging; Aspirin; Colon cancer; DNA methylation; Epigenetics.

Fig. 1

Genome-Wide DNA methylation in aspirin users and nonusers. a PCA for 10,000 most variable CpGs across 124 colon biopsies by colon location or by aspirin use (U: users, Nu: nonusers) and by time points (baseline: t1, 10 years follow-up: t2). Shown are the principal component 1 (PC1) on x- and PC2 on y-axes. b Number of unique U- and Nu-dmCpGs in proximal and distal colon; hypermethylated (hyper), hypomethylated (hypo). Benjamini-Hochberg (BH) false discovery rate adjusted P values (adj. P) < 0.05 were used as cutoff to identify significant dmCpGs. c Percentage (%) methylation change (t2 vs. t1) at U- and Nu-dmCpGs in users and nonusers for both colon locations. Maximum and median values are shown on the graph. d Methylation index (MI) for each subject, as the percentage mean methylation across all dmCpGs (Nu-dmCpGs + U-dmCpGs). Each circles represent one subject. For paired analysis, three individuals who were aspirin nonusers at t1 but became users in t2 were excluded from this analysis. Number of individual in each group is mentioned at the bottom. Shown are medians (line) and mean (black circle). P values by paired Wilcoxon signed rank test

Fig. 2

Functional analysis of dmCpGs. a Number of U- and Nu-dmCpGs at promoters, enhancers and weak-transcripts (top), promoters with and without CpG Island (CGI) (bottom). P values by fisher’s test. b Percentage (%) methylation change (t2 vs. t1) on U- and Nu-dmCpGs associated genes that are TSGs (black) or oncogenes (white) and are epigenetic regulators. c Pathways enriched in U- and Nu-dmCpGs (hyper and hypo) associated-genes found in proximal and distal colon. Venn diagram shows overlap of pathways (13 unique in users, 10 unique in nonusers, 7 common), and heatmaps illustrate the significance of enrichment (adj. P values < 0.05)

Fig. 3

Epigenetic age deceleration in aspirin users. a Percentage methylation change on blood-based Hannum’s (61CpGs that passed QC out of 71CpGs), multi-tissue-based Hovath’s (326CpGs that passed QC out of 356CpGs) and blood-based phenoAge (495CpGs that passed QC out of 513CpGs) aging biomarkers. P values by Wilcoxon rank sum test. QC: quality control. b Correlation of the three epigenetic age estimates in the normal colon with the chronological age of the individual. The raw beta values without masking probes based on a detection P value were submitted. Separate plots are shown for each time point (t1 and t2) for proximal (top) and distal (bottom) samples. Shown are the person correlation coefficient and associated P values. c Distribution of epigenetic age acceleration in aspirin users and nonusers over time

Fig. 4

Schematic diagram illustrating the suppressed methylation age (mAge) in aspirin users. (↓) reduced colon cancer (CC) risk; (↑) increased colon cancer (CC) risk