Abstract Several dietary patterns have been associated with better health outcomes. However, the extent to which adherence to different so-called healthy diets overlaps, whether these dietary patterns are equally beneficial, and whether they are associated with similar DNA methylation profiles, remains unclear. Therefore, we investigate the overlap in adherence to ten diet quality scores, and examine the associations of these scores with both biological aging markers and DNA methylation profiles. We use data from the Rhineland Study, a large population-based cohort, and validated our findings using corresponding data from the EPIC-Potsdam cohort. Here, we show minimal overlap of participants classified in the highest quartile (top 25%) of adherence across different diet quality scores. Adherence to a healthy dietary pattern is associated with reduced epigenetic aging, although associations differed in magnitude across diet quality scores. Different dietary patterns are associated with distinct methylation profiles, which however largely converged onto the same biological pathways. Our findings suggest that general adherence to a healthy dietary pattern could promote health through similar epigenetic mechanisms, despite variations in dietary composition.

Introduction: Very long-chain saturated fatty acids (VLCSFA: C20:0, C22:0, C24:0) may influence cardiometabolic health differently from other, often detrimental, saturated fatty acids (SFA). Evidence remains inconclusive, partly because VLCSFA are metabolically derived from SFA, making it difficult to disentangle their individual effects due to potential confounding. Prior studies rarely accounted for correlations with other lipids or do not consider VLCSFA-specific lipid classes. Hypothesis and Objectives: VLCSFA have distinct cardiovascular health effects compared to other SFA with which they are correlated, and this might mask the associations of VLCSFA with disease risk. We investigated prospective associations of circulating VLCSFA across multiple plasma lipid classes with the incidence of cardiovascular disease (CVD), accounting for confounding by correlated lipids. Methods: We constructed a nested case-cohort study within the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort with 1,704 participants (547 cases of CVD). Plasma concentrations of VLCSFA were measured in 12 lipid classes. A data-driven network including VLCSFA, precursors, and downstream metabolites was used to identify correlations and adjust for them in multivariable-adjusted Cox regression models between individual lipids and disease risk. Results: C20:0 was distributed across more lipid classes than C22:0 and C24:0. Associations for C20:0 and C24:0 with CVD varied by class; however, these relationships were only observed adjusting for correlated lipids identified in the network. Free fatty acid C20:0 (hazard ratio [HR] per SD: 0.68, 95% CI: 0.50-0.91), diglyceride C20:0 (0.69, 0.57-0.84), dihydroceramide C24:0 (0.62, 0.39-0.98), and lactosylceramide C24:0 (0.62, 0.46-0.83) were inversely associated with CVD only after adjusting for direct lipid neighbors. However, dihydroceramide C20:0 (1.34, 1.05-1.70) showed a positive association. Monoglycerides and cholesteryl esters containing VLCSFA were associated to higher risk of CVD. Conclusions: VLCSFA show different metabolic roles in CVD and highlight the importance of adjusting for confounding by correlated lipids to isolate their relationships. These findings challenge the traditional view that SFA exert uniform negative effects and suggest class-specific VLCSFA profiles may improve risk prediction of cardiometabolic diseases, guiding more precise prevention strategies.

Triple-negative breast cancer (TNBC) accounts for 10–15

OBJECTIVE:To examine whether IgG N-glycosylation patterns are prospectively associated with incident diabetic nephropathy and neuropathy. RESEARCH DESIGN AND METHODS:We analyzed IgG N-glycosylation profiles in four cohorts: EPIC-Potsdam, DiaGene, GenoDiabMar, and Hoorn DCS. Among 3,263 individuals with and without type 2 diabetes at profiling, 674 incident neuropathy and 639 incident nephropathy cases occurred after diabetes diagnosis. Associations of IgG N-glycan peaks (IgG-GPs) and traits with incident outcomes were examined using Cox models and meta-analyzed across cohorts. RESULTS:Agalactosylated, asialylated, and bisected IgG-GPs were associated with higher nephropathy risk (IgG-GP3: hazard ratio [HR] 1.13 [95% CI 1.04-1.24]; IgG-GP4: HR 1.15 [95% CI 1.05-1.26]), while galactosylated and sialylated IgG-GPs were associated with lower risk (IgG-GP14: HR 0.86 [95% CI 0.78-0.94]; IgG-GP18: HR 0.85 [95% CI 0.77-0.94]) (all false discovery rate <0.05). Associations of IgG-GPs with incident neuropathy were rendered nonsignificant after multiple testing correction. CONCLUSIONS:IgG N-glycosylation may reflect immune-related mechanisms relevant to diabetic nephropathy; but the evidence between IgG-GPs and diabetic neuropathy remains inconclusive.

Handgrip strength is an important marker of health status and is influenced by modifiable risk factors, including dietary intake. Trace elements (TE) copper, iron, iodine, manganese, selenium, and zinc are involved in physiological processes relevant to muscle function. However, the individual and interactive effects of these TE on handgrip strength have not been investigated in community-dwelling non-diabetic older adults. We performed a cross-sectional analysis within the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort, comprising 1432 participants (median age 67 years, 59

Abstract The human lipidome comprises numerous complex lipids, dysregulation of which can contribute to the pathogenesis of a wide range of diseases. Despite the high heritability of parts of the lipidome, the genetic architecture of many circulating lipid species and their structure remains mostly unknown. Thus, we perform genome-wide association studies on 970 lipid species and 267 fatty acid composite measures using samples from the population-based Rhineland Study (n = 6096). We validate our findings using corresponding data from two other independent cohorts, including FinnGen (n = 7266) and EPIC-Potsdam (n = 1188). Out of 217 lead genomic loci, we find 136 to be novel, such as FDFT1. Using mendelian randomization and individual-level gene expression data, we identify 43 possible causal associations between candidate genes and corresponding lipid species, including FDFT1 – diacylglycerol (16:0/18:0). Our findings provide new insights into the intricate genetic underpinnings of lipid metabolism, which may facilitate risk stratification and discovery of new therapeutic targets.

Supplementary table S2: Associations of physical activity with the metabolites analysed in the EPIC study.

Prediction of incident macrovascular events (iMEs) in individuals with type 2 diabetes (T2D) remains suboptimal. We aim to discover blood-based epigenetic biomarkers predicting iMEs in 752 newly diagnosed individuals with T2D, among whom 102 developed iMEs during follow-up. 461 DNA methylation sites, e.g., near ARID3A, GATA5, HDAC4, IRS2, and TMEM51, associate with iMEs. Using cross-validation, a methylation risk score (MRS) containing 87 sites predicts iMEs with an area under the curve (AUC) of 0.81 and an AUC of 0.84 for the combination of MRS and clinical risk factors, better than SCORE2-Diabetes (Systematic Coronary Risk Evaluation 2-Diabetes), UKPDS (United Kingdom Prospective Diabetes Study), Framingham, and polygenic risk scores (AUCs = 0.54-0.62). This epigenetic biomarker has a negative predictive value of 95.9% and improves the classification of iMEs with continuous net reclassification improvement (NRI) showing 90.2% improvement versus clinical factors. Atherosclerotic versus non-atherosclerotic aortas show 78 differentially methylated sites. We validate 32 sites in EPIC-Potsdam and 43 in OPTIMED cohorts, including an MRS (AUC = 0.80). Together, blood-based epigenetic biomarkers predict iMEs better than clinical risk factors, supporting its future clinical use.

Supplementary methods S1: EPIC physical activity questions

Several dietary patterns are suggested to benefit health, potentially through DNA methylation changes. However, to what extent adherence to so-called healthy diets overlaps, whether these dietary patterns are equally beneficial, and whether they affect health outcomes through the same molecular mechanisms, remains unclear. Therefore, we investigated the overlap in adherence to ten diet quality scores, and examined the associations of these scores with both biological aging markers and DNA methylation profiles. We used data from the Rhineland Study, a large population-based cohort, and validated our findings using corresponding data from the independent EPIC-Potsdam cohort. Interestingly, we found minimal overlap of participants in the top 25% of adherence across different diet quality scores. Adherence to a healthy dietary pattern was associated with reduced epigenetic age acceleration regardless of the specific dietary pattern, except for the EAT-Lancet diet. Different dietary patterns were associated with distinct methylation profiles, which however largely converged onto the same biological pathways. Our research thus indicates that general adherence to a healthy dietary pattern promotes health through similar epigenetic mechanisms, despite variations in dietary composition. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement The Rhineland Study is funded by the German Center for Neurodegenerative Diseases (DZNE). This work was supported through the Federal Ministry of Education and Research under the Diet Body Brain Competence Cluster in Nutrition Research (grant numbers 01EA1410C and 01EA1809C) and in the framework PreBeDem - Mit Pravention und Behandlung gegen Demenz (grant number 01KX2230), the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy (EXC 2151 390873048) and through SFB1454, project number 432325352, and the Helmholtz Association under the 2023 and 2024 Innovation Pool. DL is partly supported by a grant from the Alzheimer's Association (24AARFD-1192360). NAA is partly supported by a European Research Council Starting Grant (Number: 101041677). The work in EPIC-Potsdam was supported by a grant from the German Federal Ministry of Education and Research and the State of Brandenburg to the German Center for Diabetes Research (DZD; 82DZD00302 and 82DZD03D03). ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. The ethics committee of the University of Bonn's Medical Faculty approved the study, which is being conducted in accordance with the principles of the Declaration of Helsinki. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The data used in this manuscript is not publicly available due to data protection regulations. Access to the Rhineland Study data can be provided to scientists in accordance with the study's Data Use and Access Policy. Requests for additional information and/or access to the datasets can be sent to RS-DUAC{at}dzne.de. Information on data access and contact details for the EPIC-Potsdam study can be obtained at https://www.dife.de/en/research/cooperations/epic-study/. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

BACKGROUND:Very long-chain saturated fatty acids (VLCSFA) may influence cardiometabolic health differently from other, often detrimental, saturated fatty acids (SFA). Evidence remains inconclusive, partly because VLCSFA are metabolically derived from SFA, making it difficult to disentangle their individual effects due to potential confounding of correlated lipids. Prior studies rarely accounted for correlations with other lipids or do not consider VLCSFA-specific lipid classes. We investigated prospective associations of circulating VLCSFA (C20:0, C22:0, C24:0) across multiple plasma lipid classes with type 2 diabetes (T2D) and cardiovascular disease (CVD), accounting for confounding by correlated lipids. METHODS:We constructed two nested case-cohort studies within the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort: 1911 in the T2D case-cohort (774 cases); 1704 in the CVD case-cohort (547 cases). Plasma concentrations of VLCSFA were measured in 12 lipid classes. A data-driven network including SFA across all lipid classes was used to identify precursors and downstream lipid metabolites for each lipid class of VLCSFA. The correlated lipids were gradually incorporated in multivariable-adjusted Cox regression models between individual lipids and disease risk. RESULTS:C20:0 was distributed across more lipid classes than C22:0 and C24:0. After including all correlated precursors and downstream lipid metabolites in the model, we observed that higher C22:0 levels were linked to higher T2D risk, while associations for C20:0 and C24:0 varied by class. Ceramides C20:0 (hazard ratio [HR] per SD: 0.52, 95% CI 0.35-0.79) and C24:0 (0.46, 0.27-0.79) were inversely associated with T2D, whereas dihydroceramides C20:0 (1.36, 1.07-1.72) and sphingomyelin C24:0 (1.61, 1.15-2.26) showed positive associations. Monoglycerides and cholesteryl esters containing VLCSFA were associated to higher risk of both outcomes. Most of these relationships were not observed when the confounding or mediation by correlated lipids was not considered. CONCLUSIONS:VLCSFA show different metabolic roles in cardiometabolic diseases and highlight the importance of adjusting for confounding by correlated lipids. These findings challenge the traditional view that SFA exert uniform negative effects and suggest class-specific VLCSFA profiles may improve risk prediction of cardiometabolic diseases, guiding more precise prevention strategies.

AIMS:To examine whether zinc (Zn) and copper (Cu) status influence the association of estimated delta-5 desaturase (D5D), delta-6 desaturase (D6D), and stearoyl-CoA desaturase-1 (SCD1) activities with type 2 diabetes (T2D) risk. METHODS:We used a nested case-cohort design within the EPIC-Potsdam Study (n = 1979; 447 incident T2D cases). Desaturase activities were estimated using erythrocyte fatty acids (FA): D5D (20:4n-6/20:3n-6), D6D (18:3n-6/18:2n-6), and SCD1 (16:1/16:0 [SCD1-16], 18:1/18:0 [SCD1-18]). We evaluated associations between desaturases and serum Zn or Cu, assessed interactions between serum Zn or Cu and desaturase activities in Cox regression models for T2D risk, and examined modification by Zn transporter SLC30A8 genetic variant and metal-related polygenic risk scores. RESULTS:Higher serum Zn was significantly associated with lower SCD1-18 activity (β per 1 SD = -0.09). Zn status showed a non-linear modifying effect on the D5D-T2D relationship (p-interaction = 0.03), though an inverse D5D association was observable consistently across Zn levels. Serum Cu was positively associated with SCD1-16 (β = 0.13) and SCD1-18 (β = 0.08) and negatively associated with D5D activity (β = -0.13). Stronger inverse associations of higher D5D activity with T2D risk were observed at low Cu levels (HR 0.69, 95% CI 0.58-0.81) versus higher levels (HR 0.95, 95% CI 0.80-1.13) (p-interaction = 0.009). The SLC30A8 variant rs13266634 significantly modified the D5D-T2D association. Furthermore, the inverse association of D5D with T2D was stronger among participants with a higher Cu genetic score. CONCLUSIONS:Zn and Cu status modified the relationship between FA desaturases and T2D risk. This was supported by serum Zn and Cu levels and by genetic variation related to their transport and homeostasis.

BACKGROUND:Current evidence suggests higher physical activity (PA) levels are associated with a reduced risk of colorectal cancer. However, the mediating role of the circulating metabolome in this relationship remains unclear. METHODS:Targeted metabolomics data from 6,055 participants in the European Prospective Investigation into Cancer and Nutrition cohort were used to identify metabolites associated with PA and derive a metabolomic signature of PA levels. PA levels were estimated using the validated Cambridge PA index based on baseline questionnaires. Mediation analyses were conducted in a nested case-control study (1,585 cases, 1,585 controls) to examine whether individual metabolites and the metabolomic signature mediated the PA-colorectal cancer association. RESULTS:PA was inversely associated with colorectal cancer risk (OR per category change: 0.90, 95% confidence interval, 0.83-0.97; P value = 0.009). PA levels were associated with 24 circulating metabolites after FDR correction, with the strongest associations observed for phosphatidylcholine acyl-alkyl (PC ae) C34:3 (FDR-adjusted P value = 1.18 × 10-10) and lysophosphatidylcholine acyl C18:2 (FDR-adjusted P value = 1.35 × 10-6). PC ae C34:3 partially mediated the PA-colorectal cancer association (natural indirect effect: 0.991, 95% confidence interval, 0.982-0.999; P value = 0.04), explaining 7.4% of the association. No mediation effects were observed for the remaining metabolites or the overall PA metabolite signature. CONCLUSIONS:PC ae C34:3 mediates part of the PA-colorectal cancer inverse association, but further studies with improved PA measures and extended metabolomic panels are needed. IMPACT:These findings provide insights into PA-related biological mechanisms influencing colorectal cancer risk and suggest potential targets for cancer prevention interventions.

Polygenic scores (PGSs) for body mass index (BMI) may guide early prevention and targeted treatment of obesity. Using genetic data from up to 5.1 million people (4.6% African ancestry, 14.4% American ancestry, 8.4% East Asian ancestry, 71.1% European ancestry and 1.5% South Asian ancestry) from the GIANT consortium and 23andMe, Inc., we developed ancestry-specific and multi-ancestry PGSs. The multi-ancestry score explained 17.6% of BMI variation among UK Biobank participants of European ancestry. For other populations, this ranged from 16% in East Asian-Americans to 2.2% in rural Ugandans. In the ALSPAC study, children with higher PGSs showed accelerated BMI gain from age 2.5 years to adolescence, with earlier adiposity rebound. Adding the PGS to predictors available at birth nearly doubled explained variance for BMI from age 5 onward (for example, from 11% to 21% at age 8). Up to age 5, adding the PGS to early-life BMI improved prediction of BMI at age 18 (for example, from 22% to 35% at age 5). Higher PGSs were associated with greater adult weight gain. In intensive lifestyle intervention trials, individuals with higher PGSs lost modestly more weight in the first year (0.55 kg per s.d.) but were more likely to regain it. Overall, these data show that PGSs have the potential to improve obesity prediction, particularly when implemented early in life.

The trace elements (TEs) selenium, zinc, copper, manganese, iodine and iron are essential micronutrients that support essential metabolic functions. Imbalance in their homeostasis might contribute to the pathogenesis of major age-related chronic diseases including type 2 diabetes (T2D) and cardiovascular diseases (CVD). Emerging evidence suggests that TEs may affect health outcomes via epigenetic changes. However, few epigenome-wide association studies (EWAS) have explored TE-associated DNA methylation markers and their links to chronic disease outcomes. We conducted TE-specific exploratory EWAS using a random subcohort (n = 1030) from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort. The association between identified CpG sites and incident chronic diseases was evaluated using a case-cohort design comprising random subcohort participants and incident cases of T2D (n = 654) and CVD (n = 334). DNA methylation was measured with the MethylationEPIC BeadChip array. We used Prentice-weighted Cox proportional hazards regression models to estimate multivariable-adjusted hazard ratios (HRs) and 95

IntroductionLong-chain (LC) n-3 polyunsaturated fatty acids (PUFA) are critical nutrients in vegetarian and vegan diets due to the absence of fish and other animal products. α-Linolenic acid (ALA) is the main plant derived precursor for eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA), yet conversion efficiency is limited and influenced by several dietary and metabolic factors. Therefore, the NuEva Study aimed to investigate the impact of flaxseed oil on fatty acid profiles depending on age, sex, body mass index (BMI), dietary pattern (Western diet (omnivores), flexitarian, vegetarian, vegan), and status of relevant nutrients.MethodsThe NuEva study is a prospective non-randomized intervention with parallel diet groups (Western diet (omnivores), flexitarian, vegetarian, vegan; n = 168), which includes nutrient-optimized menu plans (12 months) combined with flaxseed oil supplementation (3 g/d ALA for 9 months). Fatty acids were analyzed at baseline and repeatedly throughout the intervention period focusing on n-6 and n-3 PUFA in plasma and erythrocytes lipids. Furthermore, potential modulators of ALA conversion (age, sex, BMI, linoleic acid, arachidonic acid, and EPA status) were investigated.ResultsIn Western diet participants, erythrocyte n-6 PUFA increased by 5.5%, mainly due to arachidonic acid. In contrast, ALA (+22.5–38.4%), EPA (+27.3–40.7%), DPA (+27.2–40.7%) and DHA (+12.8–26.0%) increased significantly across all dietary patterns. Conversion efficiency was unaffected by sex, BMI, age, linoleic acid, or arachidonic acid, but individuals with low baseline EPA showed markedly greater increases in EPA (+62.9% vs. +12.9%), DPA (+41.9% vs. +22.3%), and DHA (+27.0% vs. +7.6%) compared to subjects with higher EPA status.ConclusionIn conclusion, flaxseed oil supplementation combined with a controlled diet effectively improves n-3 LCPUFA status irrespective of habitual diet. The extent of relative improvement was primarily determined by baseline EPA concentrations.

Supplementary table S6: Associations of physical activity with the metabolites analysed in the EPIC study without BMI adjustment.

The trace elements selenium, zinc, copper, manganese, iodine, and iron are crucial for various physiological processes, including enzymatic reactions and immune responses. Dyshomeostasis of trace elements is associated with a variety of diseases including diabetes and cardiovascular diseases. It has not been clarified whether blood trace elements associate with the risk of diabetes-related vascular complications. We aimed to investigate the prospective associations between pre-diagnosis serum levels of trace elements with vascular complications in diabetes. Participants with incident diabetes and free of micro- and macrovascular disease and with pre-diagnostic serum trace element measurements from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort (n = 627) were followed for microvascular and macrovascular complications (n = 212 and n = 69, respectively, median follow-up: 12.8 years). We used Cox Proportional Hazard models to investigate the associations between baseline trace element levels (per SD difference) and the risk of developing diabetes-related vascular complications. To investigate the interactions and nonlinear associations between TEs and risk of diabetes-related complications, we applied Bayesian kernel machine regression (BKMR). In multivariable models, higher iodine levels were associated with higher risk of developing total vascular complications (HR per SD, 95

Brain insulin action plays an important role in metabolic and cognitive health, but there is no biomarker available to assess brain insulin resistance in humans. Here, we developed a machine learning framework based on blood DNA methylation profiles of participants who did not have type 2 diabetes with and without brain insulin resistance and detailed metabolic phenotyping. We identified 540 DNA methylation sites (CpGs) as classifiers of brain insulin resistance in a discovery cohort (n = 167), results that were validated in two replication cohorts (n = 33 and 24) with high accuracy (83 to 94%). All 540 CpGs were differentially methylated and annotated to 445 genes mapping to neuronal development and axonogenesis processes. Methylation patterns of 98 of 540 CpGs exhibited a strong and significant (P < 0.05) blood-brain correlation, indicating that blood cells are a reliable proxy to capture brain-specific DNA methylation changes. These blood-based epigenetic signatures could potentially serve in the future for the early detection of individuals with brain insulin resistance in a broad clinical setting.