Racial Disparities in Epigenetic Aging of the Right vs Left Colon

Abstract

There are well-documented racial differences in age-of-onset and laterality of colorectal cancer. Epigenetic age acceleration is postulated to be an underlying factor. However, comparative studies of side-specific colonic tissue epigenetic aging are lacking. Here, we performed DNA methylation analysis of matched right and left biopsies of normal colon from 128 individuals. Among African Americans (n = 88), the right colon showed accelerated epigenetic aging as compared with individual-matched left colon (1.51 years; 95% confidence interval [CI] = 0.62 to 2.40 years; 2-sided P = .001). In contrast, among European Americans (n = 40), the right colon shows remarkable age deceleration (1.93 years; 95% CI = 0.65 to 3.21 years; 2-sided P = .004). Further, epigenome-wide analysis of DNA methylation identifies a unique pattern of hypermethylation in African American right colon. Our study is the first to report such race and side-specific differences in epigenetic aging of normal colon, providing novel insight into the observed younger age-of-onset and relative preponderance of right-side colon neoplasia in African Americans.

Although incidence and mortality rates of colorectal cancer (CRC) in the United States have steadily declined, reductions have been strikingly slower among African Americans (AAs) (1). Increasing data show molecular, clinical, and epidemiological differences between carcinoma of right vs left colon (2). AAs are more likely to develop right-sided lesions and be diagnosed with CRC at a younger age than European Americans (EAs) (3–5). The mechanisms underlying these racial disparities remain poorly understood.

DNA methylation changes with age, and aberrant DNA methylation is a hallmark of CRC (6). Various DNA methylation-based age estimators have been developed to predict chronological age. Among these, Horvath’s method (DNA methylation age [DNAmAge]) is perhaps the most robust (7). Epigenetic age acceleration (AgeAccel), the residual derived from regressing DNAmAge on chronological age, has been associated with CRC (8). Here, we tested the hypothesis that normal colonic tissue exhibits side- and race-specific patterns of epigenetic aging and propose that these differences may help explain the observed racial disparities in risk of side-specific CRC.

The study design and analysis are detailed in the Supplementary Methods and Supplementary Table 1 (available online). Infinium MethylationEPIC BeadChip array (Illumina, San Diego, CA) was performed on matched right (ascending) and left (descending) normal colon biopsies from 128 individuals (88 AAs, 40 EAs). For analysis of differential methylation, a 2-sided moderated empirical Bayesian regression was performed (9). Bonferroni correction was set at 5%; only sites with mean methylation difference greater than 5% were considered statistically significant. Two-sided mixed-effects models and linear regression were performed for analysis of AgeAccel, and a P value of less than .05 was considered statistically significant. Two-sided Fisher exact tests were used to determine whether intra-individual concordance for age acceleration could be observed between AA and EA.

As expected, chronological age and DNAmAge in normal colon biopsies were strongly correlated. As shown in Figure 1, A, the correlation was stronger in EAs than in AAs. No statistically significant side difference in AgeAccel was observed in the study population as a whole. Stratification by race, however, revealed substantial differences in side-specific aging (Figure 1, B). In EAs, an increase was observed in AgeAccel of the left vs paired right colon (1.93 years; 95% confidence interval [CI] = 0.65 to 3.21 years; P = .004). We validated these findings in 3 independent, publicly available datasets (Supplementary Figure 1, available online) (10–12). In contrast, AA right colon showed a substantial increase in AgeAccel compared with the left colon (1.51 years; 95% CI = 0.62 to 2.40 years; P = .001). Next, we compared side-specific AgeAccel between the 2 racial groups. Left colon AgeAccel was higher in EAs than in AAs (2.74 years; 95% CI = 1.21 to 4.44 years; P = .001). A non-statistically significant increase was observed in the right colon of AAs vs EAs (0.81 years; P = .31). Subjects were further categorized by concordance and discordance of AgeAccel measures between colon locations (Table 1). Overall, a higher proportion of AAs had epigenetically older right than left colon as compared with EAs (60.2% vs 27.5%; P = .001). Consistent differences were observed when EA and AA age ranges were the same (Supplementary Figure 2, A-C, available online). We observed similar results using Epigenetic Timer of Cancer (13) (Supplementary Figure 3, available online).

Figure 1.

Race-specific differences in epigenetic aging and DNA methylation. A) DNA methylation age was found to be more strongly correlated with chronological age in European Americans (EAs) in both right and left colon. B) Age acceleration was calculated by regressing DNA methylation age on chronological age. Positive values indicate age acceleration, and negative values indicate deceleration for that individual and tissue. Inverse directions of effect for increased epigenetic age acceleration was observed in African American (AA) and EA colon. C) T-statistics generated from regression models were plotted. Negative t-statistics correspond to hypermethylation at that site in AA vs EA colon. Non-statistically significant t-statistics (pink) were centered around zero, as expected. Statistically significant t-statistics (blue) were found to be enriched for hypermethylation in AA right and left colon.

Table 1.

Concordance and discordance of right vs left colon epigenetic aging

Comparison	No. of African Americans	No. of European Americans	P c
A
Right (+)a and left (-)b	22	5	.11
Right (-) and left (+)	10	10
Right (+) and left (+)	23	13
Right (-) and left (-)	33	12
B
Right > left	53	11	.001
Right ≤ left	35	29

^a(+): Showing acceleration (ie, positive).

^b(-): Showing deceleration (ie, negative).

^c P value for testing of difference between African Americans and European Americans by 2-sided Fisher exact test.

Given the anatomical sidedness specificity of AgeAccel, we performed a paired regression between colon locations of all subjects. Bacon was used to correct for test-statistic bias and inflation in this analysis (Supplementary Figure 4, available online) (14). We identified 14 547 differentially methylated positions (DMPs) between right vs left colon, and the top 100 DMPs were highly robust (Supplementary Figure 5, A and B, available online). These findings were validated through novel analysis of publicly available datasets (11,12). There were statistically significant overlaps of DMPs between our findings and the validation datasets (Supplementary Table 2, available online). The majority of DMPs (82.2%) were replicated in at least 1 validation cohort (Supplementary Table 3 and Supplementary Figure 5, C, available online).

We next investigated race-associated differences in DNA methylation by colon sidedness. We identified 456 and 558 race-associated DMPs, respectively, in right and left colon. We observed an enrichment for hypermethylated DMPs in AA right colon (69.5%; P = 3.67 x 10^-7). In contrast, only 48.7% of left colon DMPs were hypermethylated in AA (Figure 1, C). Hypermethylated DMPs in AA right colon were enriched for cytosines previously associated with a number of traits, including ancestry, CRC, and aging (Supplementary Figure 6, A, available online) as well as at enhancers of colonic mucosa (Supplementary Table 4, available online). Trait enrichments were maintained following removal of ancestry DMPs (Supplementary Figure 6, B, available online).

Here, we present the first study showing substantial colon sidedness and race-specific differences in epigenetic aging and epigenome-wide DNA methylation in matched right vs left normal colon biopsies. The majority of AAs show accelerated aging of the right vs the left colon, whereas the opposite was true for EAs. We note Horvath clock performance appeared worse in AA colon, which wasn’t reported in blood (15). Further tissue and race-specific investigation is warranted.

Given different embryonic origins of the right and left colon, it is not surprising to see substantial within-individual AgeAccel variability. We provide evidence from 3 publicly available datasets confirming accelerated aging of left vs right colon in EAs shown in our data. There is, however, a scarcity of existing data available to validate our novel findings on AA colon. The incidence of young-onset CRC has recently increased dramatically among EAs more in left than right colon, whereas right colon cancer is still most common in AAs (16). Advanced age is a well-established risk factor for CRC. Our results provide biological plausibility for the observed relative preponderance of right colon cancer and younger age of onset in AAs as compared with EAs.

To better characterize these differences, we evaluated the association of race with DNA methylation at each colon location separately. The majority of right colon DMPs were hypermethylated in AA. These sites were enriched for ancestry, aging, and DMPs previously implicated in CRC. Of DMPs, 18.4% overlapped markers of ancestry, suggesting a role for genetic variation in contributing to DNA methylation differences in AA right colon. That these DMPs were also enriched at markers of enhancers in colon mucosa implies a potential functional impact on gene expression. Validation of this analysis was limited by the extreme scarcity of data available for AA normal colon, despite the recognized racial disparities. Our study directly addresses this lack of data. Further investigation may provide clinical insight into right-sided CRC etiology, in particular, among AAs.

In summary, we present the first study highlighting race and side-specific differences in the trajectories of epigenetic aging in normal colon mucosa. Our results provide novel insight of epigenetic aging potentially underlying racial disparities in CRC. Side-specific colonic epigenetic aging may be a promising marker to guide interventions to reduce CRC burden.

Funding

The work was supported by research grants from the National Cancer Institute Cancer Disparities Specialized Programs of Research Excellence (SPORE ) Planning Grant (P20 CA233216), Case Comprehensive Cancer Center Gastrointestinal SPORE (P50 CA150964), and the National Cancer Institute (CA143237) and a pilot grant from the University of Virginia Cancer Center (P30CA044579).

Notes

Role of the funders: Respective funding bodies took no role in the study design, analysis, interpretation of data, writing of manuscript, or decision to submit manuscript to publication.

Disclosures: All authors have no conflicts to disclose.

Disclaimer: Li Li, MD, PhD, is a member of the United States Preventive Services Task Force (USPSTF). This article does not necessarily represent the views and policies of the USPSTF.

Acknowledgments: We are grateful to our funders as well as the study subjects who graciously consented to donate biopsy and clinical data information.

Author contributions: Writing original draft: MD, LL, and GC; data curation: XS, FY, GSC, JW, DJW, and LL; study design and methodology: MD, DJW, and LL; statistical analysis and data visualization: MD and XS; sample management and annotation: FY; project conceptualization, execution, supervision, and project administration: GC and LL. All authors were responsible for writing—review and editing of the final draft.

Data Availability

Raw and processed DNA methylation data have been made available for download at Gene Expression Omnibus: GSE151732.