Sheetal Hardikar

Background: The past decade has seen the development of epigenetic models of aging that accurately estimate chronological age and predict disease incidence and mortality. These estimates are modulated by lifestyle and environmental factors linked to carcinogenesis, but to date this has primarily been studied in blood. Methods: We examined epigenetic aging in normal colonic tissue (n = 96), adjacent mucosa (n = 245), and tumors (n = 208), using models trained on age (Horvath, Hannum, Zhang), mortality (PhenoAge, GrimAge), aging rate (DunedinPACE), cellular mitotic history (EpiTOC, epiTOC2, miAGe), and telomere length (DNAmTL). Results: The Horvath model was the most accurate estimator of chronological age in normal colonic mucosa, with high correlation (r > 0.70) between the Horvath, Hannum, Zhang, PhenoAge, and GrimAge models as well as between mitotic clocks (r > 0.94). All models showed similar performance in normal tissue and adjacent mucosa but substantially more variation in estimates in tumors. Statistically significant differences in age acceleration were present between normal and adjacent mucosa by 6 models (Hannum, Zhang, PhenoAge, EpiTOC, epiTOC2, and miAge), while tumors showed highly statistically significant differences by all models. Age acceleration differed by region of the colon, with varying patterns by model type. Physical activity (PhenoAge), smoking history (GrimAge), and alcohol consumption (Horvath, mitotic clocks) were associated with epigenetic aging in adjacent mucosa, while smoking history, smoking intensity, and alcohol consumption were associated with DNAmTL in tumors. Conclusions: Our study reveals an impact of tissue type, region, and lifestyle factors on epigenetic aging and highlights heterogeneity between models and the need for careful consideration within study design.

Supplementary Table S1 shows the distribution and frequency of cancer types among older cancer survivors included in the study population.

Supplementary Table S2 displays the prevalence of individual multimorbidity conditions overall and stratified by AL status.

Supplementary Table S3 presents associations of primary analysis covariates with overall survival among older U.S. adults. HRs and 95% CI are presented in relation to AL and multimorbidity.

Supplementary Table S6 displays sensitivity analyses for cancer types of breast, prostate, colorectal, and non-melanoma skin cancer, and survival outcomes among older U.S. adults. HRs and 95% CI are presented for overall, and cancer-specific survival.

Supplementary Table S7 displays the associations between high AL, and multimorbidity. ORs and 95% CI are shown for (a) multimorbidity and (b) individual multimorbid conditions.

Supplementary Table S5 presents sensitivity analyses examining AL and multimorbidity in relation to overall survival among older U.S. adults. HRs and 95% CI are presented for: (a) continuous allostatic load; (b) individual allostatic load biomarkers; (c) continuous multimorbidity; and (d) individual chronic conditions.

Supplementary Table S4 presents complete case analyses evaluating associations of AL and multimorbidity with overall and cancer-specific survival.

Background:Both allostatic load (AL), a measure of cumulative physiologic stress, and multimorbidity, the presence of multiple chronic conditions, have been linked to poorer survival in older adults. However, associations remain understudied.Methods:Using the nationally representative National Health and Nutrition Examination Survey data from 1999 to 2018 with the 2019 Linked Mortality File, weighted Cox regression models estimated adjusted hazard ratios (HR) and 95% confidence intervals (CI) for the associations of AL and multimorbidity with overall survival (OS) and cancer-specific survival. AL consisted of nine biomarkers, assigning one point for each exceeding predefined clinically relevant cutoffs, for a total AL score (range: 0–9; high AL ≥3). Multimorbidity was defined as having ≥2 of 11 chronic conditions.Results:Participants (N = 3,484 adults ≥60 years with self-reported cancer) were on average 73 years of age, 46% female, 71% non-Hispanic White, 59% had high AL, and 64% had multimorbidity. Overall, 46% of participants died of any cause, including 484 cancer-specific deaths. High AL and multimorbidity were both associated with poorer OS [HR (95% CI) = 1.33 (1.15–1.54) and 1.40 (1.21–1.62), respectively] and moderately for cancer-specific survival. Results were similar for sex-specific AL cutoffs, incorporating medication use, excluding individuals <65 years, and mutually adjusting for AL and multimorbidity.Conclusions:Among older adults with cancer, high AL and multimorbidity are associated with poorer OS. Interventions that improve chronic condition management may reduce AL and multimorbidity and improve survival for older cancer survivors.Impact:Assessing AL and multimorbidity could help identify high-risk older cancer survivors and inform survivorship care.

Supplemental Table 5. Association between metabolic phenotypes defined by high sensitivity C-Reactive Protein (hs-CRP) ≥ 10 mg/L with obesity-related cancer risk in the Women’s Health Initiative cohort (N= 21,245).