**Background:** Ischemic heart disease (IHD) is the leading cause of death globally, with approximately 2 million deaths yearly in Europe. While controlled feeding trials have shown that high intakes of saturated fatty acids raise LDL cholesterol, evidence on the association between meat consumption and IHD risk has been conflicting. The authors hypothesized that interindividual variation in metabolic responses to the same diet may help explain these equivocal results. This study aimed to identify metabolomic signatures characterizing consumption of unprocessed red meat and processed meat and to assess whether such signatures are associated with IHD risk.
**Methods:** The study included 92,246 participants from the UK Biobank (mean age 56.1 years, 55.1% women) after excluding those with missing data, outliers, or pre-existing cardiovascular disease. Meat consumption was assessed using a touchscreen food frequency questionnaire with frequency categories ranging from never to once or more per day. Unprocessed red meat included beef, lamb/mutton, and pork; processed meat included bacon, ham, sausages, meat pies, kebabs, burgers, and nuggets. Plasma metabolome was profiled using high-throughput nuclear magnetic resonance spectroscopy (Nightingale Health), quantifying 167 metabolites including 140 lipids and lipoproteins, 18 organic acids, 2 apolipoproteins, and 7 fluid balance, inflammation, and glycolysis-related metabolites. Elastic net regularized regressions were used to construct metabolomic signatures in training and test datasets (1:1 split). Cox proportional hazards models were adjusted for age, sex, ethnicity, assessment center, BMI, blood pressure, education, smoking, alcohol, physical activity, Townsend Deprivation index, fruit/vegetable intake, family history of CVD, diabetes, hyperlipidemia, and cancer. Genome-wide association studies and 1-sample Mendelian randomization were performed for the metabolomic signatures.
**Key Results:** During a median follow-up of 8.74 years, 3,059 incident IHD events were documented. The metabolomic signature for unprocessed red meat consisted of 157 metabolites (Spearman r=0.223) and for processed meat consisted of 142 metabolites (Spearman r=0.329). In fully adjusted models, the red meat metabolomic signature was associated with incident IHD (HR per SD increment=1.11, 95% CI 1.06-1.16, P<0.001), and this association persisted after further adjustment for red meat consumption (HR=1.10, 95% CI 1.05-1.15). The processed meat metabolomic signature showed a stronger association (HR per SD=1.16, 95% CI 1.11-1.21, P<0.001), also independent of processed meat intake (HR=1.16, 95% CI 1.11-1.22). Self-reported red meat consumption was modestly associated with IHD (HR per serving/week=1.04, 95% CI 1.01-1.08, P=0.022), while processed meat consumption showed no significant association after full adjustment (HR=1.00, 95% CI 0.95-1.03, P=0.491). Mediation analysis indicated that 26.9% (95% CI 6.4-47.4) of the red meat-IHD association was mediated by the metabolomic signature. Individuals in the highest tertile of the red meat signature had a 55% higher IHD risk (95% CI 1.40-1.71) versus the lowest tertile; for processed meat, the highest tertile had an 81% higher risk (95% CI 1.62-2.03). Genome-wide association studies identified 45 loci for the red meat signature and 4 loci for the processed meat signature, involved in lipid and lipoprotein metabolism. Mendelian randomization showed participants in the highest quintile of genetically predicted red meat signature had HR 1.38 (95% CI 1.00-1.90) and for processed meat signature HR 1.64 (95% CI 1.06-2.53) compared with the lowest quintile.
**Clinical Implications:** This study demonstrates that metabolomic signatures can objectively capture individual metabolic responses to meat consumption and predict IHD risk beyond self-reported dietary intake. The signatures remained significant predictors after adjusting for traditional risk factors and meat consumption itself, suggesting they reflect biologically relevant pathways including lipid metabolism, inflammation, and amino acid metabolism. These findings support the potential use of metabolomic profiling for personalized dietary assessment and risk stratification. The identification of genetic loci related to lipid metabolism (e.g., PNPLA3, LDLR, LPA, APOA5) provides mechanistic insights into how meat consumption may influence cardiovascular health. Limitations include the use of a limited set of 167 named metabolites, reliance on FFQ data, and the observational nature of the study despite MR analyses.