Introduction and Objective: Incidence of type 1 diabetes (T1D) has been rising in U.S. children over the past two decades, while incidence patterns in young adults are less known. We estimated incidence of T1D in 2018-2023 among individuals aged <45 years by age group, race/ethnicity, and sex. Methods: The Diabetes in Children, Adolescents, and Young Adults (DiCAYA) Network includes six health system/networks (HS) and two population-level geographic sites, with >26 million individuals surveyed. Incident T1D was identified in electronic health records. Rates of new cases were calculated by age, sex, race/ethnicity and site type. Results: Between 2018 and 2023, 17,461 incident cases were identified among youth <18 years old. Incidence rates were 25.6 and 31.9 per 100,000 persons per year in HS and geographic sites, respectively, with variation by age, sex (in geographic sites) and race/ethnicity (Table). Rates were highest in non-Hispanic White and non-Hispanic Black, followed by Hispanic youth. In adults age 18-44 years, 28,310 cases occurred over the same period, with rates of 21.6 and 38.3 and per 100,000 persons per year in HS and geographic sites, respectively. Rates were higher in males and showed similar race/ethnicity patterns as youth. Conclusion: Youth-onset T1D incidence in 2018-2023 is higher than prior U.S. estimates from 2002-2018 and young adult-onset rates are similar in magnitude to youth-onset incidence. Disclosure T. Crume: None. A. Rajan: None. A. Bellatorre: None. R. Conway: None. M. Pavkov: None. I. Zaganjor: None. R. Anthopolos: None. J. Divers: None. S. Conderino: None. L. Thorpe: None. A.D. Liese: None. M. Mefford: Research Support; Current; Merck & Co., Inc. K. Reynolds: Research Support; Ended; Merck & Co., Inc. H. Shao: None. A.G. Hirsch: None. M. Rosenman: None. T.C. Ong: None. T.S. Hannon: None. B.S. Schwartz: None. E.M. Apperson: Speaker's Bureau; Current; Sanofi. D. Dabelea: None. Funding Centers for Disease Control and Prevention (U18DP006521; U18DP006512; U18DP006509; U18DP006500; U18DP006513; U18DP006506; U18DP006693; U18DP006694; U18DP006517; U18DP006518; U18DP006633)

Introduction and Objective: Diabetes prevalence among young adults in the U.S. is poorly characterized by subtype and within subgroups. We used electronic health records (EHR) to estimate diabetes prevalence by type in 2018 and 2022 among young adults in the U.S. Methods: We used data from five Diabetes in Children, Adolescents, and Young Adults Network centers: four health system centers (HS) (Table 1) and a geographic-based center in Colorado (CO), with a total of more than 30 million young adults under surveillance. We used validated computable phenotypes (CP) to identify adults ages 18 to <45 years with diabetes (type 1 [T1D], type 2 [T2D]). Prevalent cases met the CP within the index year or up to several years prior, depending on center capacity. Results: T1D prevalence in young adults in 2018 was 4.7 (4.6, 4.8) per 1000 among the HS and 4.2 (4.1, 4.3) in CO. T2D prevalence in 2018 was 22.3 (22.2, 22.5) per 1000 for the HS and 9.1 (8.8, 9.3) in CO. T1D was most prevalent among non-Hispanic (NH) White and T2D was most prevalent among NH-Black and Hispanic young adults (Table 1). In most centers, T1D was more prevalent in males versus females and decreased with age; T2D was more prevalent in females and increased with age. Conclusion: This was one of the largest EHR-based studies of diabetes prevalence in young adults in the U.S., providing sufficient sample sizes to generate estimates by subtype and subgroup. Disclosure A.G. Hirsch: None. M. Rosenman: None. J. Divers: None. R. Anthopolos: None. T. Crume: None. D. Dabelea: None. S. Kim: None. A. Rajan: None. N. Laiteerapong: Research Support; Current; Novo Nordisk. B.E. Dixon: None. K. Reynolds: Research Support; Ended; Merck & Co., Inc. M. Mefford: Research Support; Current; Merck & Co., Inc. L. Thorpe: None. B.S. Schwartz: None. Funding The DiCAYA Network (Assessing the Burden of Diabetes by Type in Children, Adolescents, and Young Adults) work was funded by the Centers for Disease Control and Prevention’s National Center for Chronic Disease Prevention and Health Promotion and the National Institute of Diabetes and Digestive and Kidney Diseases(U18DP006521 to Children’s Hospital of Pennsylvania; U18DP006512 to University of Florida; U18DP006509 to Geisinger; U18DP006500 to Indiana University and Purdue University at Indianapolis; U18DP006513 to University of South Carolina; U18DP006506 to Kaiser Foundation Hospitals; U18DP006693 and U18DP006694 to Lurie Children’s Hospital; U18DP006517 to University of Colorado Component-A; U18DP006518 to University of Colorado Component-B; and U18DP006633 to New York University Long Island School of Medicine).

Introduction and Objective: Little is known about the impact of the transient impact of physiological insulin resistance on insulin secretion dynamics during puberty in healthy youth. We examined measures of glycemia and insulin dynamics across puberty in healthy youth. Methods: A total of 337 youth ages 9-15 years (Tanner stages I through 4/5) participating in the Colorado Healthy Start Study completed a 3-hour oral glucose tolerance test (OGTT) with multiple measurements of glucose and C-peptide, as well as Tanner stage and body mass index (BMI) assessments. We modeled insulin secretion rates (ISR), incremental areas under the curve (iAUC) for C-peptide and glucose, and C-peptide clearance. Additional markers of insulin sensitivity included HOMA-IR, QUICKI, 1/insulin. Linear and ordinal logistic regressions were used to test for associations between Tanner stage and glucose-insulin outcomes, before and after adjustment for age and sex. Results: After adjustment, higher Tanner stage was associated with lower insulin sensitivity (HOMA-IR, QUICKI, 1/insulin), lower insulin clearance, higher insulin secretion rate, and higher iAUC for C-peptide (Table). There were no significant differences in fasting and 2-h glucose, glucose iAUC, or HbA1c across Tanner Stages. Conclusion: Puberty is associated with significant changes in insulin dynamics but does not appear to affect glycemia in healthy adolescents. Disclosure J. Giles: None. C.C. Cohen: None. D. Glueck: None. E.K. Englund: None. D. Dabelea: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (5R01DK133235-03)

Introduction: Increasing arterial stiffness and blood pressure (BP) are thought to contribute to cardiovascular disease risk in youth-onset type 1 (T1D) and type 2 (T2D) diabetes. Prior work in generally healthy youth found that arterial stiffness preceded changes in BP. Whether this temporal relationship exists in a youth-onset diabetes population, who have higher arterial stiffness and BP compared to healthy controls, is unknown. Further, which arterial stiffening mechanisms are most involved (load-dependent stiffening due to elevated BP or structural stiffening due to arterial wall thickening and remodeling) is unknown. Hypothesis: Our objective is to investigate the longitudinal bidirectional relationships between arterial stiffness (measured by carotid-femoral pulse wave velocity [PWV CF ]), and BP in youth-onset T1D or T2D. Additionally, we will use a recently-published modeling strategy to uncouple structural stiffness (PWV Structural ) and load-bearing stiffness (PWV Load ) components from measured PWV CF . We hypothesize that baseline PWV measures will predict future BP, as in generally healthy youth. Methods: Participants from the SEARCH for Diabetes in Youth Study (T1D, n=566; T2D, n=93; baseline age, 18.0 ± 4.6 years) had clinical, BP, and PWV CF assessed at two timepoints (mean follow-up time: 4.5 ± 1.1 years). Using a modeling approach which incorporates arterial mechanics, PWV CF , and BP, we calculated the PWV Structural and PWV Load components. We constructed separate multivariable linear regression models to quantify associations of (A) baseline PWV measures with follow-up BP, and (B) baseline BP with follow-up PWV. Results: In fully-adjusted models ( Image 1 ), a 1-SD higher baseline PWV Load was associated with a 4.03 (95%CI: 2.68-5.37) mmHg higher SBP at follow-up, but baseline PWV CF (ß: 1.15, 95%CI: -0.28-2.57) and PWV Structural (ß: 0.45, 95%CI: -0.95-1.85) were not associated. In the reverse analysis ( Image 2 ), a 10 mmHg higher SBP at baseline was associated with 0.08 (95%CI: 0.02-0.13) SD higher PWV CF , 0.32 (95%CI: 0.24-0.41) SD higher PWV Load , and 0.05 (95%CI: -0.01-0.11) SD higher PWV Structural . Bidirectional associations were largely consistent for DBP. Conclusions: In adolescents and young adults with youth-onset diabetes, BP may temporally precede changes in arterial stiffness, with load-dependent stiffening playing a larger role. This suggests BP control as a primary target for reducing arterial stiffness and CVD risk in youth-onset diabetes.

Introduction and Objective: Type 2 diabetes (T2D) incidence has been increasing among youth. We estimated T2D incidence among youth (10-17 years) and young adults (18-44 years) using multiple networks of electronic health records (EHRs). Methods: The DiCAYA network includes six health systems/networks (HS) and two geographic centers in the United States. Incident T2D was identified in HS and geographic center EHRs leveraging a common data model and using first-time ICD-10 diagnosis codes in 2018-2023. Rates were calculated by age, sex, race/ethnicity, and site type. Results: Among youth aged 10-17, T2D incidence rates were 29.2 and 39.8 per 100,000 persons per year for geographic centers and HS, respectively (Table). Rates increased with age, were higher among females, and were highest among non-Hispanic Black and Hispanic youth. Among young adults aged 18-44, T2D incidence rates were 195.0 and 326.8 per 100,000 persons per year for geographic centers and HS, respectively. Patterns by age and race/ethnicity were similar to those in youth. Sex-specific patterns differed, with female rates higher in geographic centers but male rates higher in HS. Conclusion: T2D incidence rates exhibited similar patterns by age and race/ethnicity in youth and young adults. While the magnitude of rates varied by type of EHR network, findings are consistent with prior evidence of high T2D incidence in U.S. youth. Disclosure M. Mefford: Research Support; Current; Merck & Co., Inc. A. Rajan: None. T. Crume: None. D. Dabelea: None. A.G. Hirsch: None. H.L. Kirchner: None. M.E. Wandai: None. B.E. Dixon: None. T.S. Hannon: None. L.K. Billings: Advisory Panel; Current; Novo Nordisk, Lilly, Sanofi, Amgen Inc., Bayer AG. H.S. Gordon: None. J. Divers: None. L. Thorpe: None. R. Anthopolos: None. D.C. Lee: None. A.D. Liese: None. C. Rudisill: None. I. Zaganjor: None. M. Pavkov: None. K. Reynolds: Research Support; Ended; Merck & Co., Inc. Funding This work was supported by the Centers for Disease Control and Prevention’s National Center for Chronic Disease Prevention and Health Promotion and Prevention and the National Institute for Diabetes and Digestive and Kidney Diseases (grants U18DP006521 to Children’s Hospital of Pennsylvania; U18DP006512 to University of Florida; U18DP006509 to Geisinger; U18DP006500 to Indiana University and Purdue University at Indianapolis; U18DP006513 to University of South Carolina; U18DP006506 to Kaiser Foundation Hospitals; U18DP006693 and U18DP006694 to Lurie Children’s Hospital; U18DP006517 to University of Colorado Component-A; U18DP006518 to University of Colorado Component-B; and U18DP006510 to New York University Long Island School of Medicine).

Introduction and Objective: PA, a lifestyle target for preventing and treating T2D, may lower risk of CI. We examined longitudinal trajectories of PA and their association with CI in the Diabetes Prevention Program Outcomes Study (DPPOS). Methods: Participants (n=1,560) at high T2D risk randomized to lifestyle, metformin, or placebo (1996-1999) had ~3y of active intervention (DPP) and ~21y of unmasked follow-up (DPPOS). CI status was determined ~25y post-randomization using NACC-UDS (v3) criteria. PA (metabolic equivalent [MET] hours/week) was assessed using the Modifiable Activity Questionnaire at up to 15 visits. Linear mixed-effects models summarized PA during the study period (mean and slope). Multinomial logistic regression assessed associations between PA parameters and CI risk (Dementia, Mild CI [MCI] vs no CI). Results: At cognitive assessment, participants were ~74 years (IQR 68-80); 72% women, and 71% had T2D. Higher mean PA was associated with 19% decreased odds of dementia per additional 10 MET-h/wk (Table, global p=0.023). PA slope was not associated with CI. No significant heterogeneity was observed by diabetes status at time of assessment. Conclusion: Higher mean PA over 21 years was associated with decreased risk of dementia. In the context of prediabetes and T2D, these findings underscore PA throughout mid- to late-life as a critical modifiable protective factor for dementia. Disclosure A.H. Tjaden: None. J.P. Crandall: None. H.P. Hazuda: None. E.M. Venditti: None. Q. Pan: None. B. Laferrère: Consultant; Ended; Azurity. E.S. Strotmeyer: Research Support; Current; Amgen Inc. O. Carmichael: Research Support; Current; Eli Lilly and Company. D. Dabelea: None. E. Groessl: None. W. Knowler: None. N. Nadkarni: None. J.M. Noble: None. B. Rockette-Wagner: None. J.A. Luchsinger: Consultant; Current; Merck KGaA, Novo Nordisk. Other - Merck KGaA provides investigational product including matching placebo free of charge for an NIH funded clinical trial.; Current; Merck KGaA. P. Palta: None. D. Research Group: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) U01 DK048489, National Institute on Aging of the NIH 5 U19 AG078558.

Introduction and Objective: Whether metformin affects dementia risk is unclear. We examined the association of randomized metformin (MET) vs placebo (PBO) and Intensive Lifestyle (ILS) interventions in the Diabetes Prevention Program (DPP) with cognitive impairment (CI) in the DPP Outcomes Study (DPPOS). Methods: Participants (n=1483, mean age 74 ± 8.4) were evaluated for prevalent CI in 2022-2024. Randomization to MET, ILS or PBO (1996-1999) was followed by masked MET (1996-2002) and then open label MET in the original MET group until 2022. CI syndromes, using the National Alzheimer’s Coordinating Center Uniform Dataset version 3 forms, included (in order of severity) no CI, CI-not-Mild Cognitive Impairment (MCI), non-amnestic MCI (naMCI), amnestic MCI (aMCI), dementia. We examined the association of randomization to MET with CI syndromes using multinomial logistic regression. Results: MET exposure was 15.5 ± 7.7 person years in the MET group and after developing type 2 diabetes out-of-study MET was 4.5 ± 5.1 person years in PBO and 3.8 ± 4.8 person years in ILS groups. Randomization to MET was associated with 59-62% and 60% lower odds of dementia compared with PBO and ILS, respectively (Table). Conclusion: Long-term exposure to metformin may prevent dementia among persons with pre-diabetes and type 2 diabetes. Longer follow-up with more dementia cases is needed to confirm our finding. Disclosure P. Wander: None. L. Doherty: None. Q. Pan: None. O. Carmichael: Research Support; Current; Eli Lilly and Company. D. Dabelea: None. R.S. Turner: Research Support; Current; Lilly, Eisai Inc. Speaker's Bureau; Current; Lilly. Consultant; Current; Re:Cognition Health. S. Kuo: None. M. Munshi: Advisory Panel; Current; Abbott Diabetes. Research Support; Ended; Dexcom, Inc. A. Wallia: Research Support; Current; UnitedHealth Group. J.M. Noble: None. V. Shah: None. N. Nadkarni: None. S. Mudaliar: None. M. Temprosa: None. W. Knowler: None. J.A. Luchsinger: Consultant; Current; Merck KGaA, Novo Nordisk. Other - Merck KGaA provides investigational product including matching placebo free of charge for an NIH funded clinical trial.; Current; Merck KGaA. D. Research Group: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) (U01 DK048489), National Institute on Aging of the NIH (5 U19 AG078558).

Introduction and Objective: Polycystic ovary syndrome (PCOS) is the most common endocrinopathy in women, and prevalence is reportedly elevated in girls and women who have diabetes. However, little is known about risk factors associated with PCOS in women with diabetes. Methods: We examined factors at diabetes diagnosis among participants in the SEARCH for diabetes study who received a provider diagnosis of PCOS or polycystic ovaries (PCO/S) at a study visit up to 18 years after diagnosis. There were 1,432 female participants with PCO/S diagnosis information available, 1183 with type 1 diabetes (T1D) and 249 with type 2 diabetes (T2D). Results: The prevalence of provider-diagnosed PCO/S was 4.9% (n=58) in participants with T1D and 18.5% (n=46) in participants with T2D. Characteristics by diabetes type and PCO/S diagnosis are shown in the table. In logistic regression models, PCO/S diagnosis was associated with older age at diagnosis (OR 1.14, 95% CI 1.06-1.22, p=0.0003), lower parental education level (p=0.0026), and higher BMI z-score (OR 1.6, 95% CI 1.2-2.0, p=0.001). A diagnosis of PCO/S was 2.6-fold higher among girls with T2D than T1D (95% CI 1.4-5.1, p=0.005). There was no association of PCO/S diagnosis with health insurance type, race/ethnicity or family income. Conclusion: Girls diagnosed with T2D were more likely to receive a provider diagnosis of PCO/PCOS than girls diagnosed with T1D, and PCO/S was associated with older age at diagnosis, higher BMI z-score and lower levels of parental education. Disclosure J. Snell-Bergeon: None. C. Kim: None. A. Bellatorre: None. D. Dabelea: None.

Background/Objectives: We aimed to identify eating habits associated with hepatic fat fraction (HFF) and assess effect modification by an established genetic variant for fatty liver disease, PNPLA3 rs738409, among 381 general-risk adolescents. Methods: Dietary intake was assessed using the Block Kids Food Frequency Questionnaire and HFF was measured via magnetic resonance imaging (MRI) at age ~16 years. We first characterized naturally occurring dietary patterns using principal component analysis followed by reduced-rank regression with HFF as the response variable to identify a dietary pattern that is both relevant to the population and associated with HFF. Next, we investigated associations of the dietary pattern with HFF using linear regression models that accounted for maternal gestational diabetes, education, and prenatal smoking and child sex, age, Tanner stage, and BMI. Finally, we tested for a dietary pattern and PNPLA3 rs738409 interaction and stratified by genotype if P-interaction < 0.05. Results: The participants were 16.7 ± 1.2 years (range: 12.6-19.6 years). Half were female (50.4%) and 52.0% identified as non-Hispanic White. The dietary pattern of interest was composed of vegetables, fruit, nuts and seeds, oatmeal, sports bars, crackers and sandwiches, and beef, and was inversely associated with HFF (-0.48 [95% CI: -0.81, -0.16]). Stratified analyses revealed the strongest inverse association observed between the diet pattern score and HFF in the high-risk-variant (GG) group (-2.19 [-4.35, -0.03]), followed by the intermediate-risk (CG) group (-0.43 [-0.77, -0.10]), but not the low-risk (CC) group (-0.32 [-0.77, 0.13]). Conclusions: A diet high in vegetables, fruit, nuts and seeds, oatmeal, sports bars, crackers and sandwiches, and beef-potentially capturing an active, on-the-go lifestyle-is associated with lower HFF during adolescence, especially among individuals at genetic risk.

Importance Studying how to prevent or delay not just 1 disease but multiple chronic conditions is of great importance for public health; however, few interventions have demonstrated success during long-term follow-up. Objective To examine the association of lifestyle or metformin compared with placebo on long-term multimorbidity in adults with prediabetes. Design, Setting, and Participants Observational follow-up cohort study of a randomized clinical trial conducted at 27 sites in the United States from June 1, 1996, to December 31, 2021. From June 1, 1996, through May 28, 1999, 3234 adults at high risk of diabetes enrolled in the 3-year Diabetes Prevention Program (DPP). They were subsequently enrolled in the DPP Outcomes Study (DPPOS). Of this cohort, Centers for Medicare & Medicaid Services (CMS) morbidity data were available through 2021 for 1173 participants who provided consent. Data were analyzed from June 5, 2024, to November 7, 2025. Exposures Participants in DPP were randomly assigned to intensive lifestyle intervention, metformin, or placebo. During DPPOS, medications were unmasked with discontinuation of placebo; metformin was continued. Group booster classes were offered to the lifestyle group semiannually and all participants were offered lifestyle classes quarterly until 2014. Main Outcomes and Measures The primary outcome was multimorbidity (presence of ≥2 of 15 prevalent conditions, defined in CMS’ Chronic Condition Data Warehouse and adapted for Medicare Advantage encounters). Cox proportional hazard models were applied to estimate associations between randomized treatment groups and time to development of outcomes. Results Of the 1173 participants (median age, 74 years [IQR, 70-80]; 795 [68%] were female), 997 (85%) experienced greater than or equal to 2 conditions (median, 5 [IQR, 3-7]) by the end of follow-up (316 of 385 [82%], 327 of 385 [85%], and 350 of 403 [87%], respectively, among lifestyle, metformin, and placebo groups). The risk of multimorbidity was lower among lifestyle compared with placebo participants (hazard ratio [HR], 0.79; 95% CI, 0.68-0.93) after adjustment for relevant covariates. There was no difference between participants in the metformin and placebo groups (HR, 0.91; 95% CI, 0.78-1.07). These relationships persisted when diabetes was excluded from the multimorbidity definition. When restricted to dyads of the costliest conditions, the association with lifestyle vs placebo yielded an HR of 0.57 (95% CI, 0.38-0.85). Conclusions and Relevance Among adults with prediabetes at baseline, lifestyle intervention, but not metformin, was associated with a lower burden of multimorbidity. Lifestyle programs may persistently lower the development of chronic conditions. Trial Registration ClinicalTrials.gov Identifier: DPP, NCT00004992 ; DPPOS, NCT00038727

BACKGROUND/OBJECTIVES:Prenatal visceral (VAT) and total (TAT) adipose tissue are associated with gestational diabetes and preeclampsia risk, yet cannot be routinely assessed in clinical or epidemiologic settings. Although anthropometric measures reflect adiposity in non-pregnant individuals, their utility in early pregnancy remains unclear. We examined associations between anthropometrics and TAT and VAT in the early second trimester and evaluated differences by prepregnancy BMI. SUBJECTS/METHODS:Among pregnant women (n = 61) in the Mother and Infant NuTrition (MINT) cohort, TAT and VAT were assessed via whole-body MRI at 15-weeks gestation. Anthropometrics included waist (WC), hip (HC), mid-upper arm (MUAC), calf (CC), thigh (TC), and skinfolds (SF) for iliac crest, subscapular, thigh, and triceps. Linear regression evaluated predictors of TAT and VAT by BMI category. RESULTS:TAT demonstrated strong positive correlations with BMI, WC, HC, MUAC, and TC (r = 0.76-0.92) while VAT demonstrated moderate positive correlations with BMI, MUAC, HC (r = 0.48-0.58), and strongest with WC (r = 0.60). The healthy BMI (n = 36) TAT prediction model included MUAC, HC, and iliac SF (R2 = 0.86), and VAT model included MUAC and iliac SF (R2 = 0.60). The overweight and obesity (n = 25) TAT prediction model (R2 = 0.93) included WC and HC and the VAT model (R2 = 0.79) included HC and subscapular SF. CONCLUSIONS:Anthropometric measures associated with prenatal adiposity differed by prepregnancy BMI. The TAT prediction model showed an acceptable R2 and included HC, MUAC, and iliac SF for healthy BMI, and WC and HC for overweight or obesity. Incorporating comprehensive anthropometrics into prenatal research and clinical care may improve risk identification and understanding of prenatal adiposity.

Genetic risk scores (GRSs) for type 1 diabetes (T1D) may assist T1D classification and prediction but are often developed from European populations. To improve health outcomes, it is important to understand the performance and utility of GRSs in diverse ancestry populations; we therefore assessed the capacity of three previously published T1D GRSs at differentiating people with and without type 1 diabetes in African (with/without T1D=194/235), European (n=1109/125), and Hispanic (266/170) ancestry populations in the USA, and from Cameroon and Uganda (n=144/5001).

OBJECTIVE:Exposure to per- and polyfluoroalkyl substances (PFAS) may increase the risk of gestational diabetes mellitus (GDM), with adverse consequences for pregnant women and their offspring. However, epidemiologic studies have shown inconsistent results. We addressed this question in a large, pooled sample of U.S. women. RESEARCH DESIGN AND METHODS:Participants (n = 5,229) from 16 cohorts had singleton pregnancies. PFAS were quantified in a single plasma or serum sample during pregnancy (1999-2021); six PFAS detected in ≥60% of participants were analyzed. The primary outcome was GDM diagnosis based on self-report or medical record documentation. The secondary outcome, among 1,213 participants, was fasting glucose. We estimated associations between each PFAS and GDM using generalized estimating equations models with Poisson distribution and robust variance, and estimated associations between each PFAS and fasting glucose using generalized estimating equations models for linear regression. Effect modification by prepregnancy BMI or race and ethnicity was evaluated via interaction terms and stratification. We quantified the combined effect of the PFAS mixture using quantile-based g-computation. RESULTS:Associations between individual PFAS and GDM were null or weakly inverse; the association with the six-PFAS mixture was negative (prevalence ratio [95% CI] per quartile increase: 0.75 [0.58, 0.96]). Certain PFAS were more strongly negatively associated with GDM among participants with BMI <25 kg/m2. Associations between PFAS and fasting glucose were largely null, although both positive and negative associations were observed in specific race and ethnicity strata. CONCLUSIONS:In a large, pooled sample of U.S. pregnant women, greater concentrations of PFAS were not associated with higher prevalence of GDM.

BACKGROUND:Sugar-sweetened beverages (SSBs) have been linked to obesity and metabolic dysfunction in youth; whereas the effects of 100% fruit juice, which is also high in free sugar, remain debated. OBJECTIVE:To compare metabolomic profiles of SSB versus 100% fruit juice intake across childhood and adolescence. METHODS:Data were from 593 youth in the EPOCH cohort collected during childhood (6-14 years) and adolescence (12-19 years). A food frequency questionnaire assessed total-energy-adjusted SSB and fruit juice intakes over the past week and fasting serum samples were assayed by untargeted metabolomics profiling. Least Absolute Shrinkage and Selection Operator (LASSO) regression with bootstrapping identified metabolites associated with each beverage at each visit. Metabolites consistently selected across visits were retained and compared. RESULTS:Mean (SD) SSB intake was 6.9 (4.9) and 5.2 (3.7) servings/week in childhood and adolescence, respectively, and mean fruit juice intake was 2.4 (2.6) and 1.7 (2.0) servings/week. Of 767 metabolites, 40 were associated with SSB intake and 39 were associated with fruit juice intake at both visits. Only three metabolites were associated with SSB and fruit juice intake (argininate, cortisone, quinate); all others were unique. Both beverages were linked to metabolites of amino acid, xenobiotic, lipid, and carbohydrate metabolism (10, 11, 8 and 1 metabolites, respectively, for SSB; 7, 8, 8 and 3 metabolites, respectively, for juice intake). CONCLUSIONS:SSB and 100% fruit juice intake are associated with largely distinct metabolomic profiles, suggesting some degree of beverage-specific metabolic effects in youth. Future studies are needed to replicate these findings, explore the role of common endogenous pathways and characterise mechanisms.

Introduction and Objective: Hyperglycemia leads to hypomethylation of the thioredoxin-interacting protein gene TXNIP, causing its overexpression, oxidative stress, and inflammation. DCCT has demonstrated an association between higher A1c, TXNIP hypomethylation, and microvascular complications. In presymptomatic T1D, TXNIP overexpression in beta cells activates apoptosis and accelerates progression to insulin dependency; however, the role of dysglycemia and methylation is unknown. We explored the relationship between A1c and TXNIP methylation in children with presymptomatic T1D prospectively followed by the DAISY and TEDDY studies. Methods: Multivariable linear regression modeled the association between methylation measured with the 450K or EPIC array and A1c, prior to A1c = 6.5% or T1D diagnosis. Linear mixed effects models determined changes in methylation with age in TEDDY (subjects = 204, samples = 1,407) and DAISY (subjects = 420, samples = 932). We evaluated changes in gene expression due to methylation in TEDDY children via targeted expression quantitative trait methylation analysis. Results: Fifteen of 23 candidate CpGs (65%) reported by DCCT as associated with A1c had directionally concordant A1c-methylation effects in DAISY at pre-clinical levels of A1c (range: 4.7-6.4%). Methylation at cg19266329, near TXNIP, was inversely associated with A1c (est = -0.16; P = 0.01) in DAISY and decreased with age in both DAISY (est = -3.0; P = 0.003) and TEDDY (est = -8.4; P = 1.87e-16). Lower methylation at cg19266329 was associated with higher expression of TXNIP (est = -0.22; P = 0.0003). Conclusion: Youth with islet autoimmunity had dysglycemia-driven methylation patterns similar to those observed in adults with established T1D. Higher A1c resulted in hypomethylation and overexpression of TXNIP, currently a therapeutic target for stage 3 T1D prevention trials using TXNIP inhibitors, e.g., verapamil or SRI-37330 hydrochloride. TXNIP methylation may be an alternative therapeutic target for the prevention of stage 3 T1D and diabetes complications. Disclosure S.E. Ridoux: None. S.D. Slack: None. K. Hohsfield: None. D. Dabelea: None. M. Rewers: Consultant; Current; Sanofi. Advisory Panel; Current; Vertex Pharmaceuticals Incorporated. J. Norris: None. C. Kim: None. R.K. Johnson: None. Funding NIDDK (R01DK032493), NIDDK (R01DK104351), Foundation for Women's Health

Importance:Metformin may influence risk of dementia, with prior conflicting observations of protection or harm. Objective:To determine the association of randomization to metformin vs. placebo or intensive lifestyle intervention (ILS) in the Diabetes Prevention Program (DPP) with cognitive outcomes (cognitive impairment syndromes and trajectories of cognitive test performance) during the DPP Outcomes Study (DPPOS). Design Setting & Participants:Prospective long-term follow-up of DPP/DPPOS participants at 27 U.S. centers among adults who were at high risk for type 2 diabetes (T2D) at baseline. Exposures:Randomization to metformin, placebo, or ILS (1996-1999) for 3.2 years followed by open-label metformin in the original randomized metformin group until 2021. Main Outcomes & Measures:Cognitive impairment syndromes were adjudicated in 2022- 2024 in 1,483 participants (median age 74 [IQR 68, 80]) using the National Alzheimer's Coordinating Center Uniform Dataset version 3. Cognitive performance in executive and memory domains was ascertained with repeated cognitive tests between 2009 and 2024. Multinomial logistic regression and mixed-effects models were fit to examine associations of randomization to metformin with cognitive outcomes. Results:Total metformin exposure (mean ± SD) was 15.5 ±7.7 years/person in the metformin group. Persons in the placebo and ILS groups received out-of-study metformin usually after developing diabetes with mean metformin total exposure of 4.5 ±5.1 and 3.8 ±4.8 years/person in the placebo and ILS groups, respectively. Overall, the frequency distributions of the cognitive syndromes did not differ significantly by treatment group; however, randomization to metformin was associated with a 60% (OR 0.40 [95%CI 0.17, 0.97]) and 62% (OR 0.38 [95%CI 0.16, 0.89]) lower odds of dementia compared with placebo and ILS, respectively, after adjustment for demographics, education, income, and APOE-ε4 genotype. Randomization to metformin was also associated with significantly better memory performance over time (β=0.58; 95%CI: 0.09, 1.1; p=0.02; Cohen's d=0.1). Conclusions and Relevance:Long-term metformin treatment is associated with a reduced risk of dementia and better memory performance among persons with pre-diabetes or T2D. Estimates were imprecise due to a limited number of dementia cases. Longer follow-up with more dementia cases is needed to confirm our findings. KEY POINTS:Question: Is chronic metformin treatment related to the risk of dementia and cognitive impairment?Findings: Randomization to metformin in the Diabetes Prevention Program was associated with a lower risk of dementia in the Diabetes Prevention Program Outcomes Study compared with the randomization to placebo or randomization to intensive lifestyle intervention, but the overall distribution of cognitive impairment syndromes did not differ significantly by treatment group. Randomization to metformin was also related to modestly better longitudinal performance in a memory test.Meaning: Chronic metformin treatment may decrease the risk of dementia.