This review synthesizes current evidence on how the spleen regulates systemic lipid metabolism through coordinated control of lipoprotein processing, cholesterol trafficking, and lipidimmune interactions. Splenic macrophages and lymphocytes contribute to ApoE and ApoAI production, enabling HDL formation, remnant clearance, and reverse cholesterol transport. Scavenger receptors such as SRA1, CD36, and SRB1 mediate the uptake and remodeling of modified lipoproteins, whereas LXR and PPARdependent pathways govern cholesterol efflux and fatty acid oxidation. Additional lipidlinked processes, including iron–lipid crosstalk, ceramide generation, sphingolipid signaling, and plateletderived lipid mediators, position the spleen as a central immunometabolic hub, although the strength of evidence varies from direct experimental demonstration to correlative and model-based inference. The aim of this study is to consolidate the mechanistic and experimental findings demonstrating that disruption of splenic lipid handling, including after splenectomy, promotes dyslipidemia and contributes to metabolic disease progression.
Routine lipid parameters may provide accessible markers of postoperative metabolic adaptation after pancreatic cancer surgery, but their early perioperative trajectories remain incompletely characterized, particularly across different resectional and non-resectional procedures. We evaluated perioperative lipid-protein remodeling and its relationship with nutritional deterioration after pancreatic resection and palliative gastrointestinal surgery. This retrospective single-center cohort included 213 adults with histopathologically confirmed pancreatic adenocarcinoma (resected specimen or biopsy) undergoing pancreaticoduodenectomy, distal pancreatectomy with splenectomy, palliative gastrojejunostomy, or exploratory laparotomy without resection. Preoperative and approximately 30-day postoperative lipid profile, albumin, total protein, and CA 19 − 9 were analyzed using paired tests, mixed-effects models, ANCOVA, and correlation analyses. At 30 days, albumin decreased by 8.65 g/L (95
Proliferative diabetic retinopathy (PDR) is an advanced complication of type 2 diabetes and an important cause of visual impairment. Routinely available metabolic measures may provide complementary information on advanced disease severity and support timely ophthalmic assessment, particularly where access to specialist services is limited. Here, this research examined whether five composite metabolic indices—triglyceride-glucose (TyG), triglyceride glucose-body mass index (TyG-BMI), atherogenic index of plasma (AIP), metabolic score for insulin resistance (METS-IR), and cholesterol, high-density lipoprotein, and glucose (CHG) index—could differentiate stage 6 PDR from stages 4/5. This three-center cross-sectional study included 484 patients with type 2 diabetes mellitus and PDR stages 4–6 between January 2021 and December 2025. The primary analysis compared stage 6 with combined stages 4/5. Multivariable logistic regression estimated odds ratios (ORs) per 1-standard deviation (SD) increase in each index after adjustment for demographic, clinical, and center-level factors. Firth bias-reduced regression was used to assess sparse-data sensitivity, while ordered-stage, quartile, and component-excluded models were employed for secondary analyses. The primary comparison included 347 patients with stage 4 or 5 PDR and 137 with stage 6 PDR. All five indices were positively associated with stage 6 PDR, with adjusted ORs per 1-SD increase of 18.75 (95
LDL-C goal achievement is suboptimal in patients undergoing coronary artery bypass grafting (CABG). We aimed to find factors associated with LDL-C goal achievement in patients under stable lipid-lowering therapy (LLT) at admission for CABG. A prospective, observational study was conducted. Stable LLT was defined as LLT for at least 4 weeks, without any change in intensity and without discontinuation of oral LLT ≥ 1 day prior to laboratory testing. For primary analysis, LLT was categorized according to treatment intensity (author defined categories (low-, moderate-, high intensity) with respect to current ESC/EAS guidelines). Achievement rate (
Per- and polyfluoroalkyl substances (PFAS) are persistent synthetic chemicals detected in populations, including children, through exposure from food, drinking water, and consumer products. Although growing evidence links PFAS exposure to altered pediatric lipid profiles, few longitudinal studies have determined whether developmental timing during early childhood is more strongly associated with lipid alterations than exposure magnitude. Serum PFAS concentrations and lipid levels, including total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TG), were measured in 212 children from the Environment and Development of Children cohort at ages 2, 4, 6, and 8 years. Generalized additive mixed models (GAMMs) characterized longitudinal dose–response relationships, tested whether associations varied by exposure age, and whether baseline concentrations at age 2 years were associated with lipid trajectories from ages 2 to 8 years. Age-specific linear regression models (lagged and concurrent) and quantile-based g-computation (q-gcomp) were used to estimate age-specific associations and joint effects of the PFAS mixture on lipid levels at age 8 years. In age-pooled GAMMs at ages 2, 4, 6, and 8 years, exposure-response relationships varied in both shape and direction across PFAS compounds for all lipid outcomes, including linear, non-linear and inverse patterns. Baseline PFAS concentrations at age 2 years were not associated with the rate of age-related change in any lipid outcome from ages 2 to 8 years. By contrast, in age-specific regression models evaluating lipid levels at age 8 years, higher HDL-C was associated with PFAS exposure at ages 6 and 8 years, and most associations remained significant after false discovery rate correction. In q-gcomp analyses, the PFAS mixture was positively associated with HDL-C at age 8 years for exposures measured at ages 6 and 8 years. In age-specific analysis, childhood PFAS exposure was associated with higher HDL-C concentrations, and the strength of this association varied by exposure age rather than corresponding directly to exposure magnitude. These findings suggest that developmental timing should be considered when evaluating the effects of childhood PFAS exposure on blood lipids, and require confirmation in larger longitudinal cohort studies.
Atopic dermatitis (AD) and rheumatoid arthritis (RA) are chronic inflammatory disorders often treated with Janus kinase inhibitors, including upadacitinib. However, upadacitinib’s longitudinal effects on lipid levels in Asian patients remain poorly characterized. This retrospective study included adult Korean patients with AD or RA receiving upadacitinib 15 mg daily for ≥ 12 weeks between 2020 and 2023. Serum TC, LDL, HDL, non-HDL cholesterol, and TG were assessed at baseline and during the first 24 weeks of treatment. Clinically relevant lipid threshold attainment, initiation or intensification of lipid-lowering therapy, and discontinuation of upadacitinib were also ascertained. Multivariable logistic regression was used to identify factors associated with ≥ 10
The sex-specific long-term changes occurring in body composition and their metabolic implications after gastrectomy remain unclear. This longitudinal study examined these changes in body composition and lipid profile for up to 5 years after gastrectomy to comprehensively understand long-term health and metabolic outcomes across sexes. Seventy-two patients who underwent gastrectomy for gastric neoplasms were included in this study. Body composition was assessed using dual-energy X-ray absorptiometry (DXA) and abdominal computed tomography (CT) at baseline and 1, 2–3 years, and 4–5 years postoperatively. DXA measured appendicular skeletal muscle mass (ASM), fat mass, and lean mass, while CT assessed skeletal muscle area (SMA) and density at the L3 vertebral level and visceral and subcutaneous fat area (VFA/SFA) at the umbilical level. Lipid profiles (total cholesterol [TC], high-density, and low-density lipoprotein cholesterol [HDL-C/LDL-C]) were evaluated concurrently. Multivariate regression analysis identified predictors of lipid profile changes. In total, 40 men and 32 women with a mean (± standard deviation) age of 58.6 ± 10.6 years were included. Body weight, fat mass, lean mass, and SMA decreased during the first postoperative year and remained reduced thereafter in both sexes (all P < 0.05). VFA and SFA values decreased initially and increased marginally thereafter but remained significantly lower than baseline (all P < 0.001). In men, the ASM value was maintained during the first year and increased marginally thereafter, whereas it decreased significantly by 4–5 years (P = 0.007) in women. In men, TC, HDL-C, and LDL-C values improved over 4–5 years, whereas only HDL-C levels improved in women. Reductions in fat mass and VFA were independently associated with LDL-C decrease. Significant sex-specific differences in long-term body composition and lipid profile outcomes were observed after gastrectomy. Women experienced persistent muscle loss and limited improvement in lipid profiles compared to men. These findings support the need for sex-specific postoperative care to prevent sarcopenia and optimize long-term metabolic health and quality of life.
Lipid peroxidation (LPO) has traditionally been regarded as a damaging consequence of oxidative stress; however, accumulating evidence indicates that it also participates in context-dependent cellular signaling. This review summarizes the biochemical foundations of LPO, including the kinetics of free-radical chain reactions, substrate specificity, and the biological activities of key oxidation products such as 4-hydroxynonenal and malondialdehyde. Under tightly regulated conditions, LPO-derived metabolites contribute to cell-cycle regulation, immune and inflammatory signaling, and potentially physiological ferroptotic cell turnover. In contrast, excessive or persistent LPO promotes cellular injury through covalent modification of proteins and lipids, mitochondrial dysfunction, proteostasis disruption, regulated cell death, and remodeling of local tissue microenvironments. These mechanisms have been implicated in the development and progression of neurodegenerative disorders, metabolic diseases, atherosclerosis, and malignancies. In cancer, the effects of LPO are strongly influenced by concentration and cellular context: extensive lipid peroxide accumulation can promote tumor cell death, whereas sublethal LPO may activate adaptive antioxidant signaling and support proliferative and invasive phenotypes, metastatic potential, and treatment resistance in selected contexts. This context dependence may partly explain the limited and inconsistent clinical benefits of conventional broad-spectrum antioxidant approaches. Accordingly, more selective modulation of LPO-related enzymes and defense systems, including lipoxygenases and glutathione peroxidase 4, has emerged as a potential therapeutic strategy. Nanotechnology-enabled delivery platforms may further improve the spatial and temporal control of such interventions; nevertheless, most of these approaches remain at preclinical or early translational stages and continue to face challenges related to safety, pharmacokinetics, target specificity, and disease-dependent therapeutic windows. By integrating fundamental molecular mechanisms with emerging therapeutic strategies, this review provides a balanced framework for understanding the dual biological roles of LPO and for informing the future development of more precise LPO-targeted interventions.
Although familial hypercholesterolemia (FH) is common and associated with significant cardiovascular morbidity, screening programs are scarce. Furthermore, the effectiveness of nutritional advice for parents on children’s LDL cholesterol (LDL-C) levels is unclear, especially in FH versus non-FH hypercholesterolemia. In the Fr1dolin study, 367 of 14,513 children in Lower Saxony and Hamburg were positively screened for LDL hypercholesterolemia with an LDL-C > 135 mg/dl (> 3.5 mmol/l). The screening was considered positive if the first finding of hypercholesterolemia was confirmed in a second sample. These 367 children were invited to a follow-up examination including an extended lipid status, nutritional counseling and genetic analysis. 289 out of 367 screen-positive children (57.4
Although alterations in lipid metabolism were associated with neurodegenerative diseases, the association between lipids in cord blood and infants’ neurodevelopment remains poorly characterized. Neonates’ cord blood lipids were detected, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), Non-high-density lipoprotein cholesterol (Non-HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG). At one year of age, neurodevelopmental performance was assessed among infants in a birth cohort from Yunnan, China. Multivariable linear and logistic regression analyses were conducted to evaluate whether lipid concentrations measured in cord blood were related to neurodevelopmental performance during infancy. This study included 315 infants. The prevalence of neurodevelopmental delay was 13.7
Maternal obesity and GDM increase the risk of fetal overgrowth. It has been shown that maternal adiponectin, decreased with these pregnancy pathologies, stimulates ceramide synthesis regulating placental nutrient transfer and fetal growth. We hypothesized that placental sphingolipid content, particularly ceramides, would be associated with fetal overgrowth in maternal obesity and GDM. Human term placentas from control, obese and obese with GDM women were collected. Lipidomic analysis by LC–MS/MS was performed on placental homogenates and membranes to assess sphingolipid content. Placental expression of enzymes involved in sphingolipid metabolism was studied by immunoblot. One or two-way ANOVA were used to analyze statistical differences on sphingolipid content and sphingolipid-related protein expression between groups and correlation with maternal BMI and z-score were determined with Pearson’s correlation coefficient. Placental ceramides decreased with maternal obesity with and without GDM, particularly in the basal membrane (BM), and were negatively correlated with birthweight and maternal BMI. Expression of ceramide synthase, CerS, was decreased and the expression of ceramidase, ASAH1, increased and positively correlated with birthweight. Sphingolipids such as sphingomyelins (SM) were specifically decreased in the microvillous membrane (MVM) of female placentas and negatively correlated with birthweight. The expression of sphingomyelinase (nSMase) was also positively correlated with maternal BMI and birthweight. Our results suggest a role for sphingolipids, particularly ceramides, in placental membranes in regulating fetal growth in pregnancies complicated by maternal metabolic disease and highlight the importance of fetal sex in establishing placental sphingolipid content in pregnancy pathologies.
Circulating triglyceride and triglyceride-rich lipoprotein (TRL) accumulation is increasingly recognized as a residual atherosclerotic risk, but their specific effects on cardiac remodeling and heart failure (HF) remain largely unexplored. Here, this issue was investigated in Gpihbp1 knockout (KO) mice, which develop severe hyperchylomicronemia due to disruption of intravascular TRL hydrolysis. Cardiac lipid metabolism and remodeling were evaluated in 10-month-old Gpihbp1 KO mice and their wild-type littermates under physiological conditions. Mice were also subjected to transverse aortic constriction (TAC) to evaluate the effects of severe hyperchylomicronemia on pressure overload-induced remodeling and HF. Furthermore, Gpihbp1 KO mice were crossed into an low-density lipoprotein receptor (Ldlr) KO background to study the effects of severe hyperchylomicronemia on hemorheology and high-fat diet (HFD)-induced cardiac pathology. Untargeted cardiac lipidomics revealed 214 differentially regulated lipid species that were specifically enriched in glycerophospholipid, fatty acid (FA) and diacylglycerol subclasses. Quantitative real-time PCR (qPCR) confirmed that FA oxidation-related genes were downregulated and that glucose utilization-related genes were upregulated. Electron microscopy revealed swollen mitochondria with fragmented cristae, and RNA-seq revealed reduced respiratory chain gene expression. These abnormalities resulted in impaired cardiac contractile performance in 10-month-old Gpihbp1 KO mice without the induction of hypertrophy or fibrosis. After TAC, Gpihbp1 KO hearts exhibited exaggerated diastolic dysfunction and more severe myocyte hypertrophy and fibrosis. In Gpihbp1/Ldlr double-knockout (dKO) mice, reduced erythrocyte deformability and increased whole blood viscosity were observed and worsened post-HFD. Additionally, HFD consumption resulted in significant diastolic impairment, cardiac hypertrophy and fibrosis in dKO mice. Severe hyperchylomicronemia due to Glycosylphosphatidylinositol-anchored high-density lipoprotein-binding protein 1 (GPIHBP1) deficiency sensitizes the heart to pathological remodeling and HF. These findings indicate a potential pathogenic contribution of hypertriglyceridemia and TRL accumulation to cardiac disease.
Carotid plaque is an imaging marker of early subclinical atherosclerosis. Adults under 50 years of age undergoing health examinations often have low short-term absolute cardiovascular risk, yet adverse lipid–metabolic profiles may already be associated with early plaque formation. This study aimed to develop and evaluate a parsimonious lipid–metabolic risk stratification model for carotid plaque risk using routinely available health-examination variables. This retrospective single-center study constructed two related cohorts from a health management center. The cross-sectional modeling cohort included 30,557 adults under 50 years of age examined between 2018 and 2023, of whom 1277 had prevalent carotid plaque, and was used for model development and internal validation. The same-center longitudinal validation cohort included 1679 baseline plaque-free participants with classifiable longitudinal plaque trajectories from repeated examinations between January 2020 and December 2024; 105 developed the primary incident carotid plaque outcome. Thirteen routine health-examination variables were considered. Variable selection and model development were performed only in the training set. The final multivariable logistic regression model included age, male sex, low-density lipoprotein cholesterol, systolic blood pressure, high-density lipoprotein cholesterol, and fasting glucose. Performance was assessed using AUROC and PR-AUC. Low-, intermediate-, and high-risk strata were defined using tertiles of predicted probability in the training set. In the internal validation set, the final multivariable logistic regression model achieved an AUROC of 0.807 and a PR-AUC of 0.168, higher than those of the age- and sex-only model. In the same-center longitudinal validation cohort, the final model achieved an AUROC of 0.795 and a PR-AUC of 0.216 for the primary incident carotid plaque outcome. The observed outcome rates increased across low-, intermediate-, and high-risk strata: 0.75
Postoperative delirium (POD) is a prevalent and serious complication in older surgical patients. The C-reactive protein-triglyceride-glucose index (CTI) is a recently developed composite marker that integrates both Insulin resistance (IR) and systemic inflammation. This study aimed to evaluate the association between CTI and POD in older surgical patients. Older surgical patients were retrospectively enrolled. The CTI was calculated and analyzed as both a continuous variable and a binary variable dichotomized at the optimal cut-off derived from receiver operating characteristic (ROC) analysis. Univariate and multivariable logistic regression models were constructed to examine the association between CTI and POD. Joint stratification by the TyG index and CRP was performed to delineate the synergistic contributions of IR and inflammation. A total of 10,125 patients were included in the final analysis. The receiver operating characteristic (ROC) curve analysis determined the optimal cut-off value for CTI to be 7.57. As a binary variable, high CTI conferred significantly elevated odds of POD compared with low CTI in fully adjusted model (OR = 1.60, 95
Cardiometabolic multimorbidity (CMM) is increasingly common and carries substantial clinical and public-health burden. The cholesterol-high-density lipoprotein-glucose (CHG) index has been linked to adverse cardiometabolic outcomes, but prospective evidence comparing baseline and cumulative exposure and testing external reproducibility under design-appropriate survey methods remains limited. We analyzed a baseline CHG cohort (n = 7,077) and a cumulative-CHG subcohort (n = 4,881) from the China Health and Retirement Longitudinal Study (CHARLS), and conducted a supportive external replication analysis in the National Health and Nutrition Examination Survey (NHANES; 1999–2018; n = 23,797). Cox models were used in CHARLS and fasting-subsample survey-weighted logistic models in NHANES. CHG and cumulative CHG were modeled per 1-standard deviation (SD) increment and by tertiles. Restricted cubic spline and threshold analyses characterized dose–response patterns. Sensitivity analyses included a stricter disease-free cumulative CHARLS cohort and additional hypertension adjustment in NHANES. Higher baseline CHG in CHARLS was associated with greater incident CMM risk (adjusted hazard ratio [HR] per 1-SD increase, 1.55; 95
Concomitant with the escalating global burden of metabolic disorders, acute pancreatitis with hypertriglyceridemia (AP-HTG) has become highly prevalent. However, the initial triglyceride elevation does not strictly correlate with adverse clinical outcomes. Therefore, we aimed to identify novel dynamic subphenotypes reflecting the intrinsic pathophysiological state and develop an early prediction model. Clinicopathological data were collected from MIMIC-IV (n = 344, development), eICU-CRD (n = 173, external validation), and an independent local cohort (n = 319, pragmatic validation). Group-based multi-trajectory modeling (GBMTM) of 7-day white blood cell and calcium trajectories was utilized to identify novel subphenotypes. After evaluating multiple machine-learning algorithms, an optimal early prediction model was established using Boruta, LASSO, and logistic regression based on admission-day indicators. Four novel dynamic subphenotypes were identified, with the Severe-Hyperinflammatory (C3) and Lipotoxic-Hypocalcemic (C4) subphenotypes carrying substantially increased independent risks for persistent organ failure and multiple organ dysfunction syndrome. Furthermore, an early logistic regression prediction model comprising white blood cells, calcium, lymphocytes, albumin, platelets, and hematocrit was developed. This model achieved high macro-average AUCs in the development (0.924) and external validation (0.933) cohorts. Pragmatic validation in the local cohort further confirmed its utility in effectively identifying high-risk subphenotypes. To enhance interpretability and clinical utility, SHAP analysis was applied, and the model was deployed as a free web-based calculator and a nomogram specifically designed for the C4 subphenotype. This study identified four novel dynamic AP-HTG subphenotypes and developed a multicenter-validated early prediction model using admission-day indicators. It facilitates early risk stratification and individualized clinical decision-making.
Cardiovascular disease (CVD) is the leading cause of death and disability among middle-aged and older adults. Metabolic burden and frailty frequently co-occur in this population. However, few quantitative indicators capture their combined contribution to cardiovascular risk. We examined the association between a composite indicator integrating the cholesterol–HDL–glucose (CHG) index and the frailty index (FI) and the risk of incident CVD. Using baseline data from 2011 and follow-up data from the China Health and Retirement Longitudinal Study (CHARLS), we included participants aged ≥ 45 years without baseline CVD. CHG-FI was constructed by integrating CHG and FI in a multiplicative form; FI was derived using a cumulative deficit model. Cox proportional hazards models were used to estimate HRs and 95
Emerging evidence suggests that novel lipid indices may enhance cardiovascular risk prediction beyond conventional measures; however, their associations with dietary factors such as nut and seed consumption remain understudied. This cross-sectional study investigated associations between nut and seed intake and novel lipid indices among Iranian adults. Using baseline data from 2514 participants in the Isfahan Cohort Study 2, nut and seed intake were assessed using a validated food frequency questionnaire. Novel lipid indices, including Atherogenic Index of Plasma (AIP), Castelli’s Risk Indices (CRI-I/II), Atherogenic Coefficient (AC), Atherogenic Combined Index (ACI), Remnant Lipoprotein Cholesterol (RLPC), Triglycerides to High-density lipoprotein Cholesterol ratio (TG/HDL-C), Lipoprotein Combined Index (LCI), and Non-High-density lipoprotein Cholesterol (NHC) were calculated from fasting blood samples. Multivariable adjusted linear and logistic regression models evaluated relationships across nut and seed intake quartiles. Participants in the highest quartile of nut and seed intake had significantly lower odds of elevated AIP (odds ratio (OR) = 0.58, 95
Sepsis-induced cardiomyopathy (SICM) is a common and severe complication of sepsis that contributes substantially to circulatory instability, organ dysfunction, and adverse clinical outcomes. Mitochondrial dysfunction and metabolic reprogramming have emerged as central mechanisms underlying its pathogenesis. Because the adult myocardium depends heavily on mitochondrial lipid metabolism for continuous energy production, disruption of lipid metabolic homeostasis may critically impair myocardial bioenergetics and stress adaptation during sepsis. Current evidence indicates that SICM is accompanied by extensive abnormalities in mitochondrial lipid metabolism, including impaired fatty acid uptake and oxidation, pathological cardiolipin remodeling, excessive lipid peroxidation, ferroptosis, disrupted lipid droplet–mitochondria interactions, and defective mitochondrial dynamics and quality control. These alterations interact with inflammatory signaling, redox imbalance, and metabolism-associated post-translational modifications, collectively promoting adenosine triphosphate (ATP) depletion, lipotoxicity, oxidative injury, and cardiac dysfunction. This review summarizes the physiological organization of mitochondrial lipid metabolism in the healthy myocardium and systematically examines the mechanisms responsible for its dysregulation in SICM. Particular emphasis is placed on cardiolipin remodeling and mitochondrial membrane lipid homeostasis as potential links among impaired substrate oxidation, respiratory chain instability, oxidative stress amplification, and myocardial injury. Emerging therapeutic strategies aimed at restoring metabolic flexibility, preserving mitochondrial membrane integrity, limiting lipid peroxidation, and improving mitochondrial quality control are also evaluated. A better understanding of these mechanisms may support earlier recognition of metabolically distinct SICM phenotypes, improve the timing and selection of targeted interventions, and facilitate the development of more precise approaches to reducing sepsis-related cardiac injury and improving patient outcomes.
The triglyceride-glucose (TyG) index is a widely used surrogate of insulin resistance, yet it captures only a narrow slice of the broader cardiometabolic dysregulation associated with metabolic dysfunction. Whether individuals with equivalent TyG values harbour meaningfully distinct multisystem metabolic phenotypes with divergent clinical implications remains poorly understood. We aimed to identify reproducible TyG-biomarker discordance profiles and characterize their associations with prevalent cardiometabolic disease and incident outcomes. Using data from 496,724 UK Biobank participants, we quantified discordance between observed and TyG-predicted values across eight routine biomarkers spanning adiposity, haemodynamics, renal function, inflammation, and lipid metabolism. Standardized residuals were projected using Uniform Manifold Approximation and Projection (UMAP) and clustered via a two-stage graph-based procedure. Soft cluster membership probabilities were derived using a Gaussian mixture model. Profile-specific associations with six prevalent cardiometabolic conditions, three medication classes, and incident major adverse cardiovascular events and type 2 diabetes over ten years of follow-up were estimated using logistic regression and Cox proportional hazards models, with additive log-ratio transformed probabilities as predictors. Plasma NMR metabolomic profiles were additionally characterized for each discordance profile using the same regression framework with Benjamini–Hochberg false discovery rate correction. Six reproducible discordance profiles were identified per sex alongside a baseline concordant profile, each characterized by systematic deviation in a distinct biological domain: discordant hypertensive, discordant renal, discordant high HDL-lipid, discordant obesity, discordant pro-atherogenic lipid, and discordant inflammatory. Profiles showed divergent associations with cardiometabolic outcomes. The discordant obesity profile was the strongest positive predictor of incident type 2 diabetes in both sexes. The discordant pro-atherogenic lipid profile was consistently inversely associated with type 2 diabetes yet positively associated with incident major adverse cardiovascular events in males, with an inverse association observed in females. Sex-specific reversals in the direction of association were observed for the discordant hypertensive profile across multiple outcomes. Metabolomic profiling corroborated the biological distinctiveness of each profile, revealing profile-specific lipoprotein subclass, inflammatory glycoprotein, and fatty acid signatures that were broadly consistent with their divergent clinical risk associations and persisted across sexes. Individuals with equivalent TyG levels harbour substantially heterogeneous multisystem metabolic phenotypes with divergent cardiometabolic risk profiles. TyG-biomarker discordance profiling provides a complementary framework for metabolic risk stratification that captures clinically meaningful heterogeneity not reflected by TyG level alone, with independent metabolomic signatures confirming the biological distinctiveness of each profile.
