Accelerating epigenetic age in right-sided colon cancer predicts a favorable outcome Authors: Qin Ye, Ping Song, Wangxiong Hu, Gangzhi Zhang Published: 2025 ABSTRACT BackgroundDNA methylation age, referred to as ‘epigenetic age, epiAge’ is an accurate biomarker for aging, including cancer. As a well established age-associated disease, colorectal cancer (CRC) is a highly heterogeneous cancer and can be divided into left-sided colon cancer (LCC), right-sided colon cancer (RCC) and rectal cancer according to the anatomical location of the tumor. However, the association of epigenetic age acceleration (EAA) with CRC tumor location, clinical characteristics, and patient outcomes remain to be clarified.MethodsA total of 548 samples were collected for epiAge estimation, including 192 samples consisting of 96 CRC tumors and 96 matched normal tissues from GSE77718 and 356 CRC tumors from The Cancer Genome Atlas (TCGA). EpiAge were computed using four different epiAge models (Horvath2013, PhenoAge, ZhangQ2019, and EpiTOC). EpiAge and EAA were compared between LCC and RCC. Log rank test was used to assess the association of EAA with LCC and RCC survival.ResultsA significantly older epiAge in the RCC than LCC was observed (P = 0.00025) but not in normal colon tissues. In addition, compared with LCC, EAA was enhanced in female and microsatellite instability-high RCC samples. Meanwhile, we found that epiAge decreased continuously from RCC stage I to stage IV. In contrast, in LCC, stage I had the most serious epiAge deceleration (stage I: − 8.9; stage II: − 1.71; stage III: − 4.91; stage IV: − 4.67). Further, EAA is significantly and negatively or positively associated with RCC and LCC mortality, respectively.ConclusionsOverall, our results suggest that EAA may be a useful biomarker for gaining a deeper understanding of CRC high heterogeneity from a biological aging perspective and will facilitate prognostic judgement and targeted treatments.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13148-025-02025-5. Background Colorectal cancer (CRC) is the third most common cancer in the world, with around two million new cases worldwide in 2022 [1]. As a highly heterogeneous disease, the five-year survival rate of CRC and treatment varies drastically at diagnosis. The risk of CRC is closely correlated with age because most CRCs arise from adenomatous polyps or serrated polyps that are progressive accumulation of genetic and epigenetic alterations in colon epithelial cells with many years. Thus, DNA methylation status alterations in the colon mucosa could serve as epigenetic markers for colon adenoma and/or CRC risk [2]. DNA methylation changes with age, and various DNA methylation-based age estimators have been developed to predict chronological age with a high consistency [3]. Brown et al. [4] concluded that the odds of harboring a preneoplastic polyps (PNP) are 16% higher with one year of accelerated aging in persons under age 50. Devall et al. [5] further found racial disparities in epigenetic aging of the right compared with left normal colon. In 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; two-sided P = 0.001). 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; two-sided P = 0.004). Additionally, Zheng et al. [6] demonstrated that epigenetic age acceleration is significantly associated with consensus molecular subtypes (CMS). Epigenetic age acceleration for CMS1, CMS3, and CMS4 were 23.90 years (P = 5.55E-11; 95% CI 17.10 ~ 30.69), 9.16 years (P = 5.84E-03; 95% CI 2.68 ~ 15.65), and 6.05 years (P = 2.69E-02; 95% CI: 0.70 ~ 11.41) when compared with CMS2, respectively. We previously found left- and right-sided colon cancer (LCC and RCC) differred greatly using multi-omics data and RCC had an unfavorable prognosis [7]. Nevertheless, epigenetic age acceleration (EAA) difference remains unexplored between LCC and RCC. In the present study, using four well-established tissue-available epigenetic clocks Horvath [8], PhenoAge [9], ZhangQ [10], and EpiTOC [11], we estimated the biological tissue age of the normal colon and CRC. EAA were defined by comparing the estimated biological age with the individual’s chronological age, to assess whether accelerated or dysfunctional aging in the right colon is associated with a poor outcome. Materials and methods Cohorts information The DNA methylation data (Illumina Infinium HumanMethylation450) of 548 samples for this study were downloaded from two public repositories: NCBI GEO (National Center for Biotechnology Information Gene Expression Omnibus) and TCGA (The Cancer Genome Atlas), including 356 CRC samples from TCGA and 192 samples from 96 matched tumor-normal tissue pairs of CRC patients (GSE77718, 96 tumors and 96 normal tissues). Batch correction of TCGA and GSE77718 datasets was not performed because epiAge and EAA calculation was analyzed entirely separately and no intercomparison between them. The basic CRC clinical information containing anatomical location was also downloaded from TCGA and GEO, respectively. Epigenetic age (epiAge) and epigenetic age acceleration (EAA) The beta value matrix was normalized with the BMIQ method before epiAge estimation [12]. To obtain a reliable result, we used four epigenetic clock models that are suitable for tissue samples and have high accuracy of epiAge prediction as blood samples include Horvath [8], PhenoAge [9], ZhangQ [10], and EpiTOC/LuA2023p2 [11] to calculate average epiAge and EAA. They use different beta values of CpG loci (Horvath, 353 CpGs,PhenoAge, 513 CpGs,ZhangQ2019, 514 CpGs; EpiTOC, 816 CpGs) to calculate epiAge. Then, EAA is estimated as the residual of regression of epiAge on chronological age [13]. The average epiAge and EAA derived from the four models were used for further analysis. Somatic mutation data for TCGA CRCs CRC somatic mutational profiles were downloaded from TCGA. Silent mutations, RNA mutations, and any mutation located within the intron, flanking sequence, 5’ untranslated region (UTR), or 3’UTR were discarded. Then, the epiAge and EAA of each patient was added to the tumor mutational burden (TMB) via a unique sample ID. Immune and stromal scores estimation The level 3 mRNA RNASeqV2 datasets for the CRC samples were downloaded from the TCGA. The stromal and immune scores were calculated based on the transcriptome data by the estimate package in R [14]. Statistical analysis Survival differences between different groups were tested by the Kaplan–Meier method and analyzed with the log-rank test with functions survfit and survdiff in the survival package for R [15]. Cox univariate model was performed with function coxph in the R package survival. Differences in values between groups were analyzed for significance using Wilcoxon signed-rank test unless otherwise stated. A P value < 0.05 was considered significant. All statistical analyses were performed using R (Version: 4.1.1, 64-bit). Results EpiAge is older in RCC but not in right normal colon mucosa To study epigenetic aging differences in the cancer and adjacent normal colon between left and right anatomical position, we divided the samples into left colon mucosa (LCM), right colon mucosa (RCM), LCC and RCC (Table S1). Interestingly, we found a significantly older epigenetic age (epiAge) in the RCC (median: 74.06) than LCC (GSE77718, median: 70.33, P = 0.00025). However, this significant variation is largely lost in matched normal colon tissues (GSE77718, P = 0.73, Fig. 1A). In addition, the epiAge of LCC remained basically unchanged from colon mucosa to cancer (GSE77718, Fig. 1A). These results indicate that the pattern of DNA methylation observed in normal colonic tissues is specifically altered in RCC.Fig. 1Distribution of epiAge and epiAge acceleration in LCC and RCC. A The distribution of epiAge in LCC, RCC, and their matched normal colon tissues in GSE77718 cohort. B The distribution of epiAge and EAA in LCC and RCC in TCGA cohort Distribution of epiAge and epiAge acceleration in LCC and RCC. A The distribution of epiAge in LCC, RCC, and their matched normal colon tissues in GSE77718 cohort. B The distribution of epiAge and EAA in LCC and RCC in TCGA cohort Epigenetic age acceleration is associated with RCC Given that RCC tend to be associated with older epiAge, whether epiAge was accelerated in RCC needs further determination. Results showed that EAA led to an older epiAge in TCGA RCC (median 77.92 vs. 71.44, P = 6.5e-6, Fig. 1B), which was consistent with the findings in GSE77718 (Table S2). In addition, considering some other factors such as microsatelite instability-stable/microsatellite instability-high (MSS/MSI-H), gender, and stages may be associated with epiAge heterogeneity within CRCs, clarify whether epiAge tend to correlate with above indicators is very necessary. Notably, no significant gender-associated difference in epiAge or EAA was observed in TCGA LCC. In contrast, female TCGA RCC patients exhibited significantly older median epiAge (83.07 years) and positive median EAA (4.48 years) compared to male RCC patients (median epiAge: 76.87 years; median EAA: − 1.2 years; Fig. 2). Intriguingly, we found that positive EAA was significantly associated with older epiAge in TCGA RCC MSI-H samples (median 91.15, EAA: 11.11) than in MSS samples (median 76.21, EAA: − 1.2, P = 5.7e-5). Nevertheless, this trend was lost in TCGA LCC (P = 0.87). Meanwhile, we also found that the median epiAge that continuously decreased from TCGA RCC stage I to stage IV (stage I: 85.05, 7.23; stage II: 80.32, 3.29; stage III: 75.7, − 0.88; stage IV: 71.09, -4.62), and EAA shifted from acceleration to deceleration in advanced stages (Fig. 2). In contrast, in TCGA LCC, early stage I had the most pronounced epiAge deceleration (stage I: 69.99, − 8.9; stage II: 74.50, − 1.71; stage III: 68.38, − 4.91; stage IV: 74.55, − 4.67) (P = 0.025, Fig. 2).Fig. 2Clinical indicators that are associated with epiAge and EAA heterogeneity within CRC. A The sexual distribution of epiAge and EAA in TCGA LCC cohort. B The microsatellite status distribution of epiAge and EAA in TCGA LCC cohort. C The stage distribution of epiAge and EAA in TCGA LCC cohort. D The sexual distribution of epiAge and EAA in TCGA RCC cohort. E The microsatellite status distribution of epiAge and EAA in TCGA RCC cohort. F The stage distribution of epiAge and EAA in TCGA RCC cohort Clinical indicators that are associated with epiAge and EAA heterogeneity within CRC. A The sexual distribution of epiAge and EAA in TCGA LCC cohort. B The microsatellite status distribution of epiAge and EAA in TCGA LCC cohort. C The stage distribution of epiAge and EAA in TCGA LCC cohort. D The sexual distribution of epiAge and EAA in TCGA RCC cohort. E The microsatellite status distribution of epiAge and EAA in TCGA RCC cohort. F The stage distribution of epiAge and EAA in TCGA RCC cohort To disentangle the inconsistence of EAA between LCC and RCC, we explored whether methylation drift caused by clonal evolution or epigenetic remodeling under microenvironmental pressure. Results showed that the trend of the TMB of RCC decreases with advancing stages, which was consistent with the epiAge results. In LCC, however, no significant association was found. The immune/scoremal score both showed no significant differences between different stages of RCC and LCC (Fig. S1). EAA correlated with poor LCC outcomes and a favorable role in RCC Considering the epiAge deceleration and acceleration differed between LCC and RCC (Table S3), whether the epiAge deceleration and acceleration were associated with the CRC patient prognosis requires further investigation. To explore the epiAge deceleration contradiction identified between RCC and LCC, we also examined the correlation of survival status using EAA data. Indeed, in TCGA LCC, epiAge acceleration was significant worse prognosis (log rank P = 0.0086, Fig. 3A). On the contrary, epiAge acceleration conferred a favorable outcome in TCGA RCC (log rank P = 0.00046, Fig. 3B). Notably, the overall opposite trend of EAA and patient survival between TCGA LCC and RCC remained basically unchanged using the four epiAge models separately (Fig. S2). This discovery was consistent with the findings in stage trend, suggesting that epiAge acceleration had the opposite effects to LCC and RCC.Fig. 3Prognostic analysis of EAA status using TCGA dataset. A KM plot of OS status for TCGA LCC samples stratified by EAA. B KM plot of OS status for TCGA RCC samples stratified by EAA Prognostic analysis of EAA status using TCGA dataset. A KM plot of OS status for TCGA LCC samples stratified by EAA. B KM plot of OS status for TCGA RCC samples stratified by EAA Discussion RCC and LCC have long been regarded as distinct cancers based on their different embryology, epidemiology, pathology, prognosis, and clinical responses to chemotherapy [16–18]. RCC originates from the midgut, including the cecum, ascending colon, and hepatic flexure. In contrast, LCC originates from the hindgut, including the splenic flexure, descending colon, and sigmoid colon. RCC is more likely to be poorly differentiated and exhibit mucinous or signet ring cell features, while LCC is more likely to be well-differentiated and exhibit adenocarcinoma features [19, 20]. In addition, RCC patients are predominantly female and older, and present with a more advanced tumor stage, larger tumor size, and more poorly differentiated tumor cells than those with LCC [21]. In this study, we also found older female epiAge and EAA were enhanced in RCCs (Fig. 2D). Though most previous studies have showed superior OS (overall survival) and lower recurrence of LCC than RCC [21, 22], some studies also have different OS rate between stage I, II, and III RCC and LCC. Warschkowet al. [23] concluded that early-stage RCC had a better prognosis than LCC through interrogating 91,416 patients. Weiss et al. [24] reported that RCC had longer OS than LCC in stage II but worse OS in stage III in 53,801 patients over 65 years of age. Therefore, there must be some underlying risk factors that are responsible for the inconsistent prognosis between RCC and LCC. Our previous study and other independent studies have found RCC has higher incidences of KRAS mutation than LCC [7, 25]. Additionally, a higher proportion of stage II RCC are microsatellite instability (MSI) positive and associated with a decrease in all‐cause mortality compared to LCC because MSI is a favorable independent predictor for colon cancer survival [22, 26]. This may partially explain the EAA correlated with a better outcome of RCC found in this study (Fig. 3B). However, MSS RCC with accelerated EAA also has a better prognosis trend though the P value is marginally significant 0.051, suggesting that MSS RCC with decreased EAA has some specific oncogene or pathways activation or pathological characteristics but warrant further investigation. Surprisingly, EAA is associated with a poorer outcome of LCCs, and the molecular characteristics that are associated with EAA-derived outcome contradiction between RCC and LCC needs further clarification. Several DNA methylation-based epiclocks have been developed that can reflect the chronological age very well [27]. In breast cancer, patients with decreased epiAge are associated with poor prognosis after adjusting for major clinical variables, including tumor stage and estrogen receptor status [28]. Wang et al. [2] found that individuals at high CRC risk have deceleration of PhenoAge in their normal colons. In this study, however, we only found older epiAge in RCC than in LCC, and this may be ascribed to different sample cohorts and statistic analysis. Individuals with a high risk for RCC have epiAge deceleration measured by the epigenetic clock, which may reflect a dysfunctional epigenetic aging process. Some limitations are present in this study including the retrospective nature of the study and the heterogeneity of public datasets, which can introduce potential cohort bias. Furthermore, the absence of an independent, prospective validation cohort limits the generalizability and clinical applicability of the findings. Additionally, the statistical power for subgroup analyses may be weakened due to limited sample size. Finally, the molecular mechanism associated with the observed findings needs further investigation. Conclusions Collectively, we have investigated four established epigenetic clocks and their associations with the colon cancer heterogeneity. Our results revealed that (1) RCC had older epiAge than LCC, (2) RCC MSI-H samples tend to age acceleration and older epiAge, and (3) RCC patients have epiAge deceleration predict a poor outcome. Our results suggest the huge difference in epigenetic aging and deceleration between LCC and RCC, but validation of our results in large prospective cohort studies is warranted. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 5. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Supplementary Material 5.