The human metabolome reflects complex metabolic states affected by genetic and environmental factors. However, metabolites associated with type 2 diabetes (T2D) risk and their determinants remain insufficiently characterized. Here we integrated blood metabolomic, genomic and lifestyle data from up to 23,634 initially T2D-free participants from ten cohorts. Of 469 metabolites examined, 235 were associated with incident T2D during up to 26 years of follow-up, including 67 associations not previously reported across bile acid, lipid, carnitine, urea cycle and arginine/proline, glycine and histidine pathways. Further genetic analyses linked these metabolites to signaling pathways and clinical traits central to T2D pathophysiology, including insulin resistance, glucose/insulin response, ectopic fat deposition, energy/lipid regulation and liver function. Lifestyle factors-particularly physical activity, obesity and diet-explained greater variations in T2D-associated versus non-associated metabolites, with specific metabolites revealed as potential mediators. Finally, a 44-metabolite signature improved T2D risk prediction beyond conventional factors. These findings provide a foundation for understanding T2D mechanisms and may inform precision prevention targeting specific metabolic pathways.

Importance:Weight gain is common during menopause, and healthy dietary patterns are key to its management. However, effectiveness of different diets for weight management during this period remains unclear. Objective:To examine and compare associations of multiple dietary patterns with weight gain and obesity risk in the years surrounding menopause. Design, Setting, and Participants:This prospective, population-based cohort study included women observed over a 12-year period surrounding menopause in the Nurses' Health Study II (1989-2019). Data analysis was performed between November 2024 and May 2025. Exposures:Diet was assessed every 4 years using validated food frequency questionnaires. Dietary scores included the plant-based diet index (PDI), healthy PDI, unhealthy PDI, Mediterranean diet, Dietary Approaches to Stop Hypertension, Planetary Health Diet Index (PHDI), low-carbohydrate diet (LCD), healthy LCD, unhealthy LCD, empirical dietary inflammatory pattern, empirical dietary index for hyperinsulinemia (EDIH), and ultraprocessed food intake. Main Outcomes and Measures:The outcomes were annual changes in self-reported body weight (kilograms per year) and incident obesity. Generalized estimating equations were used to estimate annual weight change across dietary patterns. Cox proportional hazards models were used to estimate risk of obesity across dietary patterns. Results:Among 38 283 women (mean [SD] age, 45.6 [3.0] years), the mean (SD) weight gain was 0.80 (1.00) kg per year. During 340 122 person-years of follow-up, 5214 women developed obesity. After adjusting for age, race and ethnicity, marital status, income, postmenopausal hormone therapy use, parity, smoking, alcohol, energy intake, physical activity, and baseline body mass index, the reverse EDIH (quintile 5 vs 1) was associated with the largest reduction in weight gain (mean, -0.28 kg/y; 95% CI, -0.30 to -0.26 kg/y). For incident obesity, the lowest risk was observed for the PHDI (hazard ratio, 0.46; 95% CI, 0.42 to 0.51) and reverse EDIH (hazard ratio, 0.51; 95% CI, 0.46 to 0.56). EDIH showed the largest positive correlations with red or processed meats, sodium, and French fries. PHDI showed the largest positive correlations with nuts, unsaturated fats, whole grain carbohydrates, and vegetable protein. Conclusions and Relevance:In this prospective cohort study of women during menopause, adopting low-insulinemic and planetary health diets, low in red and processed meats, sodium, potatoes, and French fries and rich in nuts, legumes, fruits, vegetables, and whole grains, was associated with optimized weight management.

Metabolomic indices summarizing diet-related metabolic responses are instrumental for examining and replicating diet–disease associations. Here we aim to identify metabolomic signatures characterizing the amounts and types of dietary carbohydrate and assess their associations with type 2 diabetes (T2D) risk. Nutritional metabolomics indices were developed using data from 1,196 healthy participants in the Lifestyle Validation Study with 7-day diet records (7DDRs). Elastic net regression within cross-validation was used to derive metabolomic indices of total carbohydrates and primary food sources. Replication was conducted using feeding menu data among 153 women from the Nutrition and Physical Activity Assessment Study. Associations with incident T2D were examined using multivariable Cox regression in 11,454 participants from the Nurses’ Health Study, Nurses’ Health Study II and Health Professionals Follow-up Study. Metabolites positively associated with total carbohydrates and added sugars mainly included glycerolipids (diacylglycerols and triglycerides), whereas glycerophospholipids (phosphatidylethanolamines and phosphatidylcholines) were inversely associated. Whole grains were linked to betaine, 3-indolepropionic acid (IPA) and hippuric acid; vegetables and legumes to IPA, N-acetylornithine and pipecolic acid; and fruits to proline-betaine and IPA. Identified metabolomic signatures showed significant correlations with a 7-day diet record-assessed diet in the Lifestyle Validation Study (Pearson r 0.33–0.65). In the Nutrition and Physical Activity Assessment Study, the metabolomic index of total carbohydrates was also significantly correlated with intake (r = 0.40). Signatures for total carbohydrates, added sugars, refined grains and potatoes were associated with higher T2D risk (HR per s.d. (95

Background: In aging societies, improving health-related quality of life (HRQoL) is vital, as it not only predicts cardiovascular disease and mortality but also reflects overall wellbeing. Because HRQoL often declines across menopause, a pivotal stage of women’s aging, we prospectively examined and compared 11 dietary patterns with HRQoL change during the transition. Methods: We analyzed 23,129 women from the Nurses’ Health Study (NHS) and NHSII who entered menopause between the first diet and the last of HRQoL assessment (NHS: 1984–2000; NHSII: 1991–2001). HRQoL was measured with the Medical Outcomes Study 36-Item Short Form Health Survey, yielding eight domains summarized as Physical (PCS) and Mental (MCS) Component Scores. Diet was assessed with validated food-frequency questionnaires. We calculated cumulative average scores for 11 dietary pattern indices across assessments prior to outcome assessment and categorized them into quintiles. Generalized linear models with repeated measures estimated mean 4-year changes in HRQoL. Results: During the follow-up, mean (SD) PCS declined [−1.30 (7.41) points/4-years] and MCS improved [1.58 (8.26)]. Healthier dietary patterns [Plant-based Diet Index (PDI), healthy PDI (hPDI), Mediterranean Diet (MedDiet), Dietary Approaches to Stop Hypertension (DASH), Mediterranean–DASH Intervention for Neurodegenerative Delay (MIND), Alternative Healthy Eating Index(AHEI)-2010, Planetary Health Diet Index (PHDI)] were associated with less declines in PCS scores and unhealthy patterns [unhealthy PDI (uPDI), empirical dietary inflammation pattern (EDIP), empirical dietary index for hyperinsulinemia (EDIH), ultra-processed foods (UPF)] were associated with more PCS declines, with the largest magnitudes [Δ, Q5 vs. Q1 (95% CI)] observed for MIND [0.56 (0.38, 0.75)]. Healthier dietary patterns (hPDI, MedDiet, DASH, MIND, AHEI) were associated with greater MCS improvement, and unhealthy patterns (uPDI and UPF) were associated with less MCS improvement, showing the largest magnitudes for DASH [0.46 (0.27, 0.65)]. Associations were similar across eight domains. More vegetables, fruits, fish aligned with better HRQoL change; more fast/fried foods, sweetened drinks or snacks, and red and processed meat aligned with worse. Conclusions: Plant-forward, neuroprotective and blood pressure–lowering patterns, including healthy animal based and less processed foods may optimize physical and mental wellbeing across menopause.

Background: The type of cooking oil used is an under recognized dietary factor influencing cardiometabolic disease in India. Little is known about how commonly used oils relate to cardiometabolic risk in Asian Indians. Methods: We analyzed data from 18,090 adults (mean age=42y, 51% female) in the nationally representative ICMR-INDIAB study. Diet was assessed by a validated food frequency questionnaire where participants reported their top 3 cooking oils. Dyslipidemia was defined as high LDL-C (≥130 mg/dl), high triglycerides (≥150 mg/dl), and low HDL-C (men <40mg/dl, women <50 mg/dl) per NCEP ATP III guidelines. Logistic regression using sampling weights evaluated associations between main cooking oil, dyslipidemia, general (BMI≥25 kg/m 2 ) and abdominal obesity (waist >90cm men or >80cm women), adjusting for sociodemographic, lifestyle, and dietary factors. Results: Cooking oil preferences showed distinct regional patterns: mustard oil was predominant in North, East, Northeast, and Central regions; peanut (groundnut) oil in the West; palm, sunflower, and peanut oils in South, coconut oil in the state of Kerala (South); and soybean oil in states of Maharashtra (West), Madhya Pradesh (Central), and Mizoram (Northeast). Across India, users of coconut oil [OR=2.89 (2.03-4.11)], butter/ghee [OR=3.45 (1.48-8.03)] sunflower [OR=1.49 (1.12-1.97)], soybean [OR=1.81 (1.32-2.48)], or palm oil [OR=1.77 (1.27-2.46)] had a significantly higher odds of high LDL-C than those using peanut oil. Conversely, compared to peanut oil users, users of mustard oil [OR=0.68 (0.55-0.86)], sunflower oil [OR=0.80 (0.64-0.99)], and soybean oil [OR=0.72 (0.57-0.91)] had lower odds of low HDL-C. The likelihood of general obesity was higher among those using coconut oil [OR=1.45 (1.08-1.96)], sunflower oil [OR=1.45 (1.17-1.80)], or butter/ghee [OR=2.34 (1.12-4.89)] when compared to peanut oil users. No associations were observed with abdominal obesity. When examining fatty acid subtypes from cooking oils, each 2%E from saturated fat was associated with 6% higher odds of high LDL-C and abdominal obesity, while each 2%E from PUFA was linked to 3% lower odds of abdominal obesity. Conclusions: In this nationally representative sample of Indian adults, cooking oil choice was significantly associated with cardiometabolic risk. Promoting healthier oils, such as peanut and mustard oil, via dietary guidelines and the Public Distribution System, may help reduce dyslipidemia and obesity in India.

BACKGROUND:Numerous carbohydrate quality metrics (CQMs) have been suggested, yet the optimal one(s) associated with the lowest type 2 diabetes (T2D) risk remains unknown. OBJECTIVES:We aimed to systematically compare 23 CQMs with T2D risk, identify the 5 strongest associations, propose an alternate Carbohydrate Quality Index (aCQI), and compare it with the existing CQI regarding T2D risk and cardiometabolic biomarkers. METHODS:We included participants of 3 prospective cohort studies [Nurses' Health Study I (1984-2020) and II (1991-2019), and Health Professionals Follow-up Study (1986-2020)], who were free of cancer, diabetes, and cardiovascular disease. Our primary outcome was incident T2D. We examined 13 plasma biomarkers in relation to CQIs among a subset. RESULTS:During 5,628,955 person-years of follow-up among 213,704 adults, 22,351 cases of incident T2D were ascertained. In multivariable-adjusted models, comparing Q5 to Q1, intakes of cereal fiber [relative risk (RR): 0.77 (0.74-0.81)], whole-fruit carbohydrates [RR: 0.80 (0.76-0.84)], glycemic index [RR: 1.20 (1.14-1.26)], sugar from sugar-sweetened beverages [RR: 1.22 (1.17-1.28)], and whole-grain carbohydrates [RR: 0.86 (0.82-0.91)] had the strongest associations with T2D risk. The aCQI [RR: 0.71 (0.68-0.75)], comprising these variables, had a larger magnitude of association with T2D risk than the original CQI [RR: 0.82 (0.79-0.87)], which included total fiber intake, glycemic index, the ratios of whole to total grains, and solid to total carbohydrates. The aCQI had significant associations with a larger percentage of differences in cardiometabolic biomarker concentrations, such as C-peptide, leptin, and LDL cholesterol, than the CQI (all P-trend ≤ 0.001). CONCLUSIONS:The novel aCQI, comprised carbohydrates from whole fruits, whole grains, sugar-sweetened beverages, cereal fiber, and glycemic index, was more strongly associated with risk of T2D and cardiometabolic biomarkers than its individual components or the existing CQI, necessitating further research.

Diet plays a crucial role in health, with low-carbohydrate diets often proposed to exert metabolic benefits. We aim to investigate metabolomic adaptations in 164 adults with overweight or obesity who were randomly assigned to high- (n = 54), moderate- (n = 53), or low-carbohydrate (n = 57) diets during a 20-week weight-loss maintenance phase of the Framingham State Food Study [(FS)2], a controlled, parallel feeding trial (ClinicalTrials.gov: NCT02068885). We measure fasting plasma metabolites by liquid chromatography-tandem mass spectrometry using samples from 147 participants who completed the study (n = 45, 48, and 54 in the high-, moderate-, and low-carbohydrate diet groups, respectively). Significant associations (False Discovery Rate<0.05) are identified between carbohydrate-to-fat ratio (CFR) and diet-induced changes in 148 of 479 metabolites at 20 weeks, with nearly all showing consistent trends at 10 and 20 weeks. Phosphatidylcholines plasmanyls/plasmalogens, phosphatidylethanolamines plasmanyls/plasmalogens, and sphingomyelins generally decrease with higher CFR, whereas lysophosphatidylcholines, lysophosphatidylethanolamines, and triglycerides generally increase. Our findings are largely reproducible in an independent feeding trial involving diets with similar CFR (Popular Diets Study, ClinicalTrials.gov: NCT00315354). Eleven triglyceride species (≤3 double bonds), linked to type 2 diabetes risk, increase with higher CFR. Our findings demonstrate metabolomic changes caused by varying CFR dietary patterns, offering potential insights into mechanisms that could guide targeted dietary intervention strategies. Increasing dietary carbohydrate-to-fat ratio in a randomized controlled feeding study altered circulating small molecules (metabolites), including ones associated with diabetes risk, underscoring important metabolic effects of dietary composition.

Background: The Planetary Health Diet Index (PHDI) quantifies adherence to the planetary health diet, aiming to promote human health and environmental sustainability. While self-reported PHDI is linked to lower cardiometabolic risk, the underlying biological pathways remain unclear. We aimed to identify a metabolomic signature of PHDI and evaluate its association with cardiovascular disease (CVD) and mortality. Methods: In this prospective cohort study, we included 11,190 adults from three US cohorts (Nurses’ Health Study [NHS], NHS II, and Health Professionals Follow-up Study) with up to 30 years of follow-up. Self-reported PHDI and covariates, including body-mass index (BMI), sociodemographic, and lifestyle factors, were derived from the two questionnaires closest to blood draw. Plasma metabolites were profiled using liquid chromatography-tandem mass spectrometry. We used elastic net regression to identify a metabolomic signature of the PHDI. Associations with incident CVD (1,861 events) and mortality (3,788 deaths) were assessed using Cox proportional hazards models. Findings were replicated in the Women’s Health Initiative (WHI; n=1,711). Findings: We identified an 84-metabolite signature of PHDI (Pearson r=0·46), characterised by higher long-chain highly unsaturated lipids and plant-related amino acids and lower carnitines and saturated lipid species, reflecting favourable lipid remodelling and oxidative stress modulation. Higher metabolomic PHDI was associated with lower risks of CVD (hazard ratio [HR] per 1-SD increment 0·95 [95% CI 0·90-1·00]) and total mortality (0·93 [0·89-0·96]), independently of self-reported PHDI, BMI, and other conventional risk factors. The signature was replicated in WHI for CVD (0·92 [0·84-1·00]). Interpretation: We identified and validated a metabolomic signature of PHDI that captures biological pathways linking adherence to a sustainable dietary pattern with cardiovascular health. These findings provide mechanistic insight into the potential cardiometabolic benefits of the planetary health diet and identify metabolic pathways that could inform future prevention strategies.

This narrative review provides a historical perspective on how observational research on type 2 diabetes has been developed and consolidated over the last 50 years and how well-designed cohort studies will provide us with knowledge for research and practice in the future and aid guideline development. We have included data from a large number of cohorts from every continent that have been used to study the development and/or progression of type 2 diabetes, including cohorts that are general population-based, disease-based, intervention-based and registry-based. We have structured the results from the past 50 years based on the following themes: diagnosis and screening, complications, risk factors and pathophysiology. We also discuss the strengths and weaknesses of observational research when compared with other research designs. Finally, we discuss the emerging and future directions for type 2 diabetes research using cohorts, which include novel developments, such as artificial intelligence, precision health and the exposome. We conclude that cohort research has significantly advanced our understanding of type 2 diabetes and aided guideline development, and complements experimental work, such as human randomised controlled trials and animal studies. Both approaches are essential and complementary in our pursuit to provide a more comprehensive understanding of the development and progression of type 2 diabetes, and to change dogma, practice and policies for better outcomes.

Background: Obesity is a leading cause of type 2 diabetes (T2D), with particularly high prevalence in Hispanic populations residing in the USA. However, how genetic variation influences obesity-related blood metabolite levels which, in turn, contribute to T2D progression, is not well understood. Our goal was to identify and understand genetic and dietary connections between obesity and T2D in a Hispanic cohort of older adults. Materials and Methods: We conducted a genome-wide association study on 13 specific metabolites previously associated with T2D and characteristic of individuals with abdominal obesity within the Boston Puerto Rican Health Study cohort. We further examined associations of identified metabolite quantitative trait loci (mQTLs) and their interactions with targeted dietary factors on T2D prevalence and related traits. We used gene set and pathway analysis with protein–protein interaction networks to explore the molecular mechanisms underlying the metabolic connections between obesity and T2D. Results: We identified 30 single-nucleotide polymorphisms (SNPs) acting as mQTLs for these 13 metabolites. These mQTLs were located within 19 gene regions, associated with processes such as linoleic acid metabolism, alpha-linolenic acid metabolism, and glycerophospholipid biosynthesis. Although no mQTLs were directly associated with T2D or related traits, 12 demonstrated interactions with certain food groups that affect T2D risk. Moreover, gene set and pathway analysis with protein–protein interaction networks indicated that alpha-linolenic acid metabolism, lipid metabolism, and glycerophospholipid biosynthesis and metabolism among other pathways are potential connections between T2D and obesity. Conclusions: This study identifies biochemical relationships between genetic susceptibility and dietary influences, contributing to our understanding of T2D progression in Hispanic people with obesity.

Background:The global food system significantly impacts environmental and human health, contributing to substantial greenhouse gas emissions. Objective:We examined associations between a novel Planetary Health Diet Index (PHDI) that reflects adherence to the EAT-Lancet recommendations and cardiometabolic risk in a cohort of South Asians. Methods:We analyzed data from MASALA study participants with baseline (n = 891) and 5-y follow-up (n = 735) data. The PHDI comprised 15 food components and ranged from 0 to a maximum of 140, with higher scores indicating greater adherence to the PHDI. We used multivariable linear and logistic regression models to examine cross-sectional and prospective (5-y) associations between baseline PHDI and cardiometabolic risk factors, adjusting for demographic, health, and lifestyle factors and baseline values of the outcome (prospective analyses only). Results:Among MASALA study participants (47% female, mean age 55 y), the mean PHDI score was 88.8 (SD 9.47). Prospectively, higher PHDI was associated with lower percentage difference in fasting glucose (-0.29 ± 0.15 %), glycated hemoglobin (HbA1c) (-0.08 ± 0.04%), higher high-density lipoprotein (0.40 ± 0.17 mmol/L), lower body weight (-0.37 ± 0.12 kg), body mass index (BMI) (-0.08 ± 0.03 kg/m2), waist circumference (-0.49 ± 0.17 cm), and systolic blood pressure (-0.65 ± 0.30 mmHg) (P < 0.05 for all). Each 10-unit higher PHDI was associated with a 20% lower likelihood of incident type 2 diabetes (OR [95% CI]: 0.80 [0.54, 0.86]). Cross-sectionally, at baseline, 10 unit higher PHDI was associated with (β ± SE) lower percentage difference in fasting glucose (-0.45 ± 0.22 %) and HbA1c (-0.49 ± 0.22%), lower LDL (-0.015 ± 0.007 mmol/L), CRP (-5.40 ± 2.42 ug/L), higher adiponectin (4.67 ± 2.02 mg/dL), lower body weight (-0.59 ± 0.26 kg), BMI (-0.27 ± 0.11 kg/m2), waist circumference (-025 ± 0.29 cm), visceral fat (-1.37 ± 1.32 cm2), and pericardial fat (-0.58 ± 0.43 cm3) (P < 0.05 for all). Higher PHDI scores were associated with lower odds of obesity (OR [95% CI]: 0.80 [0.71, 0.92]) and overweight (0.77 [0.74, 0.85]). Conclusions:Greater adherence to a planetary healthy diet was associated with lower cardiometabolic risk factors and risk of incident type 2 diabetes.

Background: Obesity, a major risk factor for cardiovascular disease (CVD), is a complex trait with substantial heterogeneity in its etiology, comorbidity risk, and prevention strategies. Hypothesis: Individuals with early-onset obesity-related comorbidities ( vs. late-onset) carry different genetic risk loci for obesity with varied biological consequences. Methods: We examined longitudinal data from 43,567 participants in the Nurses’ Health Studies and Health Professionals Follow-Up Study, including biennial measurements of body mass index (BMI) and diagnosis of 14 obesity-related diseases such as CVD, during up to 40 years of follow-up. We estimated long-term BMI trajectory using functional principal component analysis. Genome-wide association studies were conducted for BMI trajectory in individuals with early-onset (first event <60y old) and late-onset (first event >70y old) obesity-related diseases (Fig. a), followed by transcriptome-wide studies (TWAS) and Mendelian Randomization (MR) analyses to explore genetic variants impact on gene expressions in 49 tissue types integrating GTEx data. Results: Individuals with early-onset obesity-related diseases reached their lifetime peak BMI, on average 8.5 years earlier than those with late disease onset (Fig. b). We identified 10 genetic loci for BMI trajectory (P < 5e-8) in all participants. FTO was the only locus consistently identified for BMI trajectory regardless of the timing of first comorbidity, with a significant impact on its gene expression in skeletal muscle (Fig. c). Among individuals with early-onset comorbidities, TWAS further identified 10 loci for BMI trajectory, and MR confirmed their impact on tissue-specific expressions, including SULT1A1 in visceral adipose tissue, NPIPB7 in coronary artery and metabolic organs, and APOBR in subcutaneous adipose tissue and liver (Fig. d). However, no other gene was identified for BMI trajectory in TWAS in individuals with late-onset comorbidities. Conclusions Individuals with early-onset obesity-related comorbidities appear to carry obesity risk loci that have more significant impact on tissue-specific gene expressions. Understanding genetic heterogeneity in obesity may aid personalized prevention.

Background:Replacing added sugars with nonnutritive sweeteners, such as sucralose, may help reduce weight gain in adults over time. Because sucralose is primarily excreted in the stool, its consumption could lead to changes in the gut microbiome. Objectives:We aimed to explore whether replacing sucrose used in beverages with small quantities of sucralose led to gut microbiome changes among Asian Indian adults with type 2 diabetes (T2D) or overweight/obesity (BMI ≥23 kg/m2) without T2D. Methods:In 2 analogous substudies nested within two 12-wk, open-label parallel-arm randomized controlled trials, adults with T2D (n = 49) or overweight/obesity and no T2D (n = 48) were instructed to replace sucrose in their daily coffee and tea with sucralose or to continue their use of sucrose. We examined changes in gut microbiome community structure and taxonomic composition profiled using 16S rRNA sequencing in stool samples collected before and after the 12-wk interventions. The false discovery rate was controlled using the Benjamini-Hochberg method (q < 0.20). Results:Compared with the control group, the sucralose intervention decreased α diversity (Shannon index: P = 0.02; Simpson index: P = 0.03) and increased β diversity (P = 0.001) in gut microbiome communities of adults with T2D, but not among adults with overweight/obesity (all between-group P > 0.05). Among 185 genera tested in the T2D trial, compared with the control, relative abundances of 14 primarily sugar-fermenting or short-chain fatty-acid-producing Firmicutes bacteria in the Lachnospiracae family were reduced, whereas Enterococcus and Pediococcus increased during the intervention (q < 0.20). In contrast, adults with overweight/obesity and no T2D showed no similar changes. Conclusions:Replacing daily sucrose added to coffee and tea with sucralose resulted in changes in gut microbiome community structure and taxonomic composition among Asian Indian adults with T2D, but not those with overweight/obesity and no T2D. Further studies are needed to understand potential health implications and the underlying drivers of these gut microbiome changes.Clinical Trial Register No. (India Trial Register): CTRI/2021/04/032686, CTRI/2021/04/032809.

Objective Metabolomic indices that summarize metabolomic responses to diet may be instrumental for the examination and establishment of associations between diet and diseases. We aim to identify metabolomic signatures of the amounts and types of carbohydrate intake and examine their association with type 2 diabetes (T2D) risk. Research design and methods The discovery phase utilized data from the Lifestyle Validation Studies, which included 1,196 participants with plasma metabolomics and primary food sources assessed using 7-day dietary records, which included intakes of carbohydrate, added sugar, whole grain, refined grain, vegetable, fruit, potato, and legume. Elastic net regression within a cross-validation framework was used to calculate metabolite profiles correlated with carbohydrate intake. The association of these profiles with incident T2D was investigated using multivariable Cox regression in 11,454 participants from the Nurses' Health Study (NHS), NHS II, and Health Professionals Follow-up Study. Results Metabolites positively associated with total carbohydrate and added sugar intake mainly included glycerolipids, while glycerophospholipids were negatively associated with these variables. Whole grain consumption was positively associated with betaine, 3-indolepropionic acid (IPA), and hippuric acid. For vegetables and legumes, IPA, N-acetylornithine, and pipecolic acid were among the top positive metabolites. Fruit was positively associated with proline-betaine and IPA. Identified metabolomic signatures showed significant correlations with carbohydrate intake (Pearson r ranging 0.56-0.80 with true intake assessed using the triad method). Signatures for total carbohydrate, added sugar, refined grain, and potato were positively associated with T2D incidence [HR per SD: 1.07 (1.06-1.12), 1.07 (1.02-1.12), 1.12 (1.07-1.18), and 1.36 (1.29-1.44), respectively]. Conversely, signatures for whole grain, vegetable, fruit, and legume were inversely associated with T2D [HR per SD: 0.73 (0.70-0.77), 0.95 (0.90-0.99), 0.88 (0.83-0.92), and 0.93 (0.88-0.97), respectively]. Conclusions This study identified a panel of plasma metabolites related to the total and specific types of carbohydrate consumption. The metabolomic signatures of carbohydrate intake from different dietary sources were differentially associated with the risk of T2D. These findings corroborated the observations of carbohydrate intake assessed using questionnaires or other recall-based instruments.