Background: Several DNA methylation (DNAm) algorithms have recently emerged as robust predictors of aging and adverse health outcomes in older adults, offering valuable insights into cardiovascular disease (CVD) risk stratification. However, their predictive performance for CVD varies significantly. This study aimed to systematically investigate the associations of 12 widely used DNAm algorithms with CVD and mortality risk. Methods: Data from the NHANES (National Health and Nutrition Examination Survey) 1999 to 2002 were used to assess 12 DNAm algorithms (eg, HannumAgeacc, PhenoAgeacc, GrimAgeMortacc, GrimAge2Mortacc) in relation to CVD risk and mortality. Two cohorts were analyzed: one for CVD risk (n=1230) and another for CVD mortality risk (n=1606). DNAm was measured using the Infinium Methylation EPIC BeadChip kit (Illumina). Odds ratios (ORs) and hazard ratios (HRs), along with 95% CIs per SD increase of these DNAm algorithms, were calculated. Results: Significant associations were observed for GrimAgeMortacc and GrimAge2Mortacc with coronary heart disease and heart attack, with multivariable‐adjusted ORs per SD increase ranging from 2.15 to 2.76. However, several algorithms exhibited no significant association with self‐reported prevalent CVD. For mortality risk, HannumAgeacc, PhenoAgeacc, ZhangAgeacc, GrimAgeMortacc, and GrimAge2Mortacc were significantly associated with CVD mortality. The multivariable‐adjusted HRs per SD increase were 1.19 (95% CIs, 1.05–1.34), 1.13 (95% CIs, 1.01–1.26), 1.63 (95% CI, 1.08–2.47), 1.90 (95% CIs, 1.51–2.40), and 1.87 (95% CIs, 1.51–2.32), respectively. These associations were consistent across biological sex, age (≥50 and <65 versus ≥65 years), and race and ethnicity groups. Conclusions: DNAm algorithms, particularly GrimAgeMortacc and GrimAge2Mortacc, may serve as valuable tools for CVD risk stratification and mortality risk assessment.

딩총위, et al. Korean J Community Nutr. 2022 Jun;27(3):223-244. https://doi.org/10.5720/kjcn.2022.27.3.223

Background: While DNA methylation ( DNAm )‐based algorithms of aging show promise for predicting age‐related diseases, their comparative utility for incident cardiovascular disease ( CVD ) remains underexplored. We aimed to systematically compare 13 DNAm algorithms in relation to CVD risk in older adults. Methods: This prospective cohort study included 2112 participants from the US HRS (Health and Retirement Study) with baseline DNAm data (2016) and 6‐year follow‐up. DNAm profiling was performed using the Infinium MethylationEPIC BeadChip kit ( EPIC , Illumina, San Diego, CA ). Thirteen widely used DNAm algorithms of aging were evaluated. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% CIs per 1‐SD increase in risk scores for incident CVD, including any CVD, congestive heart failure, and stroke. Results: During a median of 37 months’ follow‐up, a total of 288 CVD cases (congestive heart failure, 38; stroke, 152) were observed. Zhang_DNAmAge, methylation Pace of Aging, and DNAmGrimAgeAcc showed positive associations with any CVD risk, with HRs per SD increase of 1.25 (95% CI, 1.07–1.45), 1.16 (95% CI, 1.02–1.33), and 1.24 (95% CI, 1.05–1.47), respectively. For specific CVD subtypes, DNAmGrimAge was associated with congestive heart failure risk ( HR , 1.47 [95% CI, 1.11–1.95]), while both mPoA ( HR , 1.24 [95% CI, 1.03–1.49]) and HorvathAcc ( HR , 1.28 [95% CI , 1.07–1.55]) predicted incident stroke. Significant dose–response relationships ( P ‐trend <0.05) were observed for Zhang_DNAmAge and DNAmGrimAgeAcc across all CVD outcomes. Conclusions: Zhang_DNAmAge, methylation Pace of Aging, and DNAmGrimAgeAcc function as independent predictors of incident CVD . These epigenetic biomarkers may improve early CVD detection and risk stratification in aging populations.

Objectives: To determine the nutrient intake according to patterns of metabolic risk factors, and examine the relationship between patterns of metabolic risk factors and incidence of cardiocerebrovascular disease or fatty liver in Korean adults.Methods: Data from the Korean Genome and Epidemiology Study were used to analyze 7,019 adults aged 50-64 years old. Depending on patterns of metabolic risk factors, participants were classified into four categories: ‘normal, NOR’, ‘obesity risk factors, ORF’ , ‘vascular risk factors, VRF’, ‘cluster risk factors, CRF’. Cox regression analysis was used to evaluate the incidence of cardiocerebrovascular disease or fatty liver.Results: The mean adequacy ratio (MAR) of VRF pattern women was significantly lower than that of the normal women. In Model 3, the incidence of cardiovascular disease or fatty liver decreased by 64.4% (hazard ratio (HR), 0.36; 95% confidence interval (CI), 0.15-0.84) in MAR Q3 women compared to MAR Q1 women. In Model 2 adjusted for age and energy intake, the incidence of cardiovascular disease or fatty liver increased by 87.6% (HR, 1.88; 95% CI, 1.01-3.48) in CRF pattern women compared to NOR women, but there was no significant difference in Model 3.Conclusions: The study showed that among women, VRF patterns were associated with poor nutrient intake. There was an association between nutrient intake levels and metabolic risk factor patterns in women. Therefore, these findings can be used a basis to emphasize the need to establish nutrition management strategies tailored to individual’s metabolic risk status, for the prevention and management of metabolic syndrome complications.

Background: Chronic liver disease (CLD) represents a substantial global health challenge, with significant morbidity and mortality worldwide. Biological aging may be involved in CLD-related outcomes, but the associations of diverse DNA methylation (DNAm) aging algorithms with CLD phenotypes and long-term mortality remain incompletely characterized. Methods: Using a US nationally representative cohort, we analyzed 12 DNAm aging algorithms in 2522 adults aged ≥50. CLD was classified into viral, alcohol-related, metabolic syndrome (MetS)-related liver disease, or uncharacterized groups. Advanced fibrosis was defined by AST-to-platelet ratio index ≥0.7. Associations of algorithms with CLD and all-cause mortality (followed through 2019) were assessed using multivariable-adjusted regression and logistic regression models and Cox models, accounting for complex sampling. Results: Among participants with CLD, DNAm aging algorithms showed only nominal associations with APRI-defined advanced fibrosis. GrimAgeMortAcc, GrimAge2MortAcc, HannumAgeAcc, PhenoAgeAcc, DunedinPoAm, and HorvathTelo were significantly associated with all-cause mortality. GrimAge-based measures and PhenoAgeAcc showed the strongest associations with all-cause mortality across CLD-related phenotypes (HRs ranged from 1.31 to 1.82). HorvathTelo was inversely associated with mortality risk (HR = 0.71, 95% CI: 0.62–0.82). Conclusions: DNAm aging algorithms, particularly GrimAge-based measures, are strongly associated with long-term all-cause mortality among individuals with CLD-related phenotypes. Although associations with fibrosis-related outcomes varied according to fibrosis definitions, DNAm aging algorithms may provide additional biological information for mortality risk stratification among individuals with CLD-related phenotypes.

Cholesterol metabolism is critical in cancer biology, yet its role in hepatocellular carcinoma (HCC) remains poorly understood. We investigated the expression and prognostic significance of cholesterol homeostasis genes in HCC to identify potential biomarkers for risk stratification. Using data from The Cancer Genome Atlas, we identified differentially expressed cholesterol-related genes (logFC ≥ 1.5, false discovery rate < 0.05), intersecting them with genes annotated in the Molecular Signatures Database. Prognostic biomarkers were evaluated using univariate and multivariate Cox regression. A predictive model was constructed to assess survival outcomes. Cholesterol homeostasis genes, particularly fatty acid-binding protein 5 (FABP5) and alcohol dehydrogenase 4 (ADH4), are key prognostic biomarkers in HCC. The proposed model provides a novel approach for predicting outcomes and may inform therapeutic strategies. Further studies are needed to validate these findings and explore their clinical utility. Nine cholesterol homeostasis genes were differentially expressed, including 6 upregulated (choline kinase alpha, FABP5, fatty acid synthase, farnesyl diphosphate synthase, mevalonate diphosphate decarboxylase, and squalene epoxidase) and 3 downregulated (ADH4, activating transcription factor 5, and arginine vasopressin receptor 1A). FABP5 and ADH4 emerged as independent prognostic factors: high FABP5 expression was associated with poor prognosis, while high ADH4 expression predicted improved survival. A prognostic model integrating FABP5 and ADH4 and clinical characteristics achieved a concordance index of 0.683, effectively stratifying patients into high-risk and low-risk groups (χ ² = 20.60, P < .001) with median survival times of 1135 and 2532 days, respectively.