Semaglutide and cardiovascular risk in type 1 diabetes: A predictive modelling analysis in a double‐blind, randomised, placebo‐controlled cross‐over trial

Semaglutide, a glucagon-like peptide-1 (GLP-1) receptor agonist, is well established for glucose-lowering in type 2 diabetes (T2D) and for producing sustained weight loss in people with overweight or obesity.1 It also demonstrates cardiovascular protection in high-risk populations, including those with T2D and/or obesity.2, 3 Although not approved as an adjunct to insulin in type 1 diabetes (T1D), small trials suggest semaglutide improves HbA1c, time-in-range, weight, and insulin requirements without increasing hypoglycaemia or diabetic ketoacidosis risk.4, 5 Its cardiovascular effects in T1D, however, remain unknown. This evidence gap is particularly relevant given the two- to eightfold higher risk of cardiovascular disease (CVD) in people with T1D compared with those without diabetes.6 Glycaemic target attainment remains suboptimal despite advanced insulin technologies,7 and rising rates of overweight and obesity further increase cardiometabolic risk. Although off-label semaglutide use in T1D is increasing, high-quality evidence to guide its safe and effective use is limited. In the absence of cardiovascular outcome trials, T1D-validated risk prediction models (e.g., the Steno Type 1 Risk Engine ST1RE8 and the Scottish Diabetes Research Network SDRN9) may offer insight into potential cardiovascular benefits of semaglutide. Accordingly, we evaluated the impact of semaglutide on predicted 10-year CVD risk in adults with T1D, hypothesising a reduction in estimated risk. This secondary analysis used participant-level data from a previously reported 32-week, double-blind, randomised, placebo-controlled crossover trial (NCT05205928) evaluating once-weekly subcutaneous semaglutide in adults with T1D during automated insulin delivery.4 Each 15-week treatment arm included an 11-week drug titration phase (0.25 mg weekly for 4 weeks, 0.5 mg weekly for 4 weeks, then up to 1.0 mg weekly thereafter) of semaglutide or matching placebo while participants used their own insulin pump and continuous glucose monitor, followed by 4 weeks of research-based closed-loop insulin delivery. A 2-week washout period separated the two arms, for a total study duration of 32 weeks. The primary aim of the trial was to compare the percentage of time spent in the target glucose range (3.9–10.0 mmol/L) during the final 4 weeks of each intervention. Of the 28 randomised participants, 24 completed both treatment arms. Two participants with established CVD were excluded from the present analysis, yielding a final cohort of 22 participants. The 10-year risk of a first fatal or nonfatal cardiovascular event was estimated at baseline and post-treatment using two validated T1D risk prediction models: ST1RE8 and SDRN.9 ST1RE incorporates sex, albuminuria status, physical activity, and smoking status as categorical variables, and age, diabetes duration, HbA1c, systolic blood pressure, low-density lipoprotein (LDL) cholesterol, and estimated glomerular filtration rate (eGFR) as continuous variables. SDRN includes sex, albuminuria status, diabetic retinopathy, smoking status, treatment for hypertension or dyslipidaemia, and atrial fibrillation as categorical variables, and age, diabetes duration, HbA1c, mean 3-year HbA1c (approximated from screening values), body-mass index (BMI), height, weight, systolic blood pressure, total cholesterol-to-high-density lipoprotein (HDL) ratio, and eGFR as continuous variables. Multiple imputation was used for handling missing data. Percentage changes in predicted cardiovascular risk and absolute changes in key modifiable parameters included in the risk models were compared between arms using paired-sample t-tests. All analyses were conducted in R (v4.3.2), with statistical significance defined as p < 0.05. Among the 22 participants, 59% were female, with a mean (± standard deviation) age of 45 ± 13 years, diabetes duration 28 ± 13 years, and BMI of 32.5 ± 5.6 kg/m2. Median interquartile range baseline 10-year predicted CVD risk was ~12% (ST1RE: 12.4% 8.5, 20.8, SDRN: 11.5% 4.3, 18.5). Within-group and placebo-adjusted changes in predicted cardiovascular risk and key model parameters are presented in Figure 1. After 15 weeks of semaglutide, the placebo-adjusted change in predicted 10-year CVD risk did not differ significantly using either the ST1RE model (p = 0.07; Figure 1A) or the SDRN model (p = 0.78; Figure 1B). Semaglutide reduced body weight (p < 0.01; Figure 1C), HbA1c (p < 0.01; Figure 1D), HDL cholesterol (p = 0.01; Figure 1E), but did not significantly change LDL cholesterol (p = 0.10; Figure 1F), eGFR (p = 0.56; Figure 1G), or urinary albumin-to-creatinine ratio (UACR; p = 0.63; Figure 1H). Albuminuria status remained stable in most participants (82% none; 14% persistent), with one participant (5%) developing transient microalbuminuria on semaglutide. Excluding this individual, the placebo-adjusted changes in predicted risk were −6.2% (95% CI: −9.4, −3.1; p < 0.01) with ST1RE and −0.8% (95% CI: −7.9, 6.3; p = 0.82) with SDRN. In this secondary analysis, semaglutide did not significantly reduce predicted 10-year CVD risk in adults with T1D. The modest risk reductions observed with the ST1RE model, however, suggest a possible trend toward long-term cardiovascular benefit. Notably, the two validated models yielded different effect estimates, in contrast to previous reports of concordance.10, 11 This discrepancy may reflect the small sample size, structural differences between the models (such as the inclusion of lipid parameters in SDRN but not ST1RE), and differing weightings of HbA1c, weight, UACR and cholesterol. In our study, semaglutide reduced body weight and HbA1c, and modestly lowered HDL cholesterol, while LDL cholesterol, eGFR and UACR remained unchanged—factors that may have contributed to variations between model outputs. Additionally, one participant developed transient microalbuminuria during semaglutide treatment. Excluding this individual revealed a modest reduction in predicted 10-year cardiovascular risk; however, this finding should be interpreted cautiously given the small sample size and the conceptual limitations of excluding a participant whose albuminuria would theoretically increase cardiovascular risk. In patients with T2D and high cardiovascular risk, semaglutide confers robust cardiovascular protection.2, 3 A 2025 meta-analysis of over 19 000 patients with T2D reported an 18% reduction in major adverse cardiovascular events, a 19% reduction in cardiovascular death, and a 26% reduction in coronary revascularisation.12 Whether these benefits translate to T1D remains unknown. Although cardiovascular outcome trials in T1D are not planned, ongoing tirzepatide studies (e.g., SURPASS-T1D-1/2) will provide complementary evidence on GLP-1-based cardiometabolic effects. This study has several strengths. The double-blind, randomised, placebo-controlled crossover design allowed each participant to serve as their own control, reducing variability and increasing statistical efficiency. Two validated T1D-specific cardiovascular risk models (ST1RE and SDRN) provided complementary insights into predicted 10-year risk, enhancing robustness. High participant retention further supports reliability, strengthening internal validity and hypothesis generation. The study has several limitations, including small sample size, short follow-up, and a crossover design, which may limit detection of long-term cardiovascular risk changes and introduce carry-over effects. With semaglutide titrated to 1.0 mg weekly, a submaximal dose relative to higher regimens, and participants having low-to-moderate baseline 10-year CVD risk (~12%), the potential for observable risk reduction was inherently limited. Interpretation of predicted cardiovascular effects may also be limited by potential carry-over effects on HbA1c, as the 2-week washout in this cross-over design is short relative to the ~120-day lifespan of erythrocytes. Alternative glycaemic measures, such as fructosamine or continuous glucose monitoring-derived estimated HbA1c, could partially address these limitations but would require additional assumptions. These considerations should therefore be kept in mind when interpreting findings from risk prediction models. Although exploratory, our findings provide an important hypothesis-generating signal that semaglutide as add-on to insulin may favourably influence cardiovascular risk in T1D. Nonetheless, these results are preliminary. Dedicated and more adequately powered outcome trials in diverse T1D populations are warranted to definitively assess semaglutide's impact on cardiovascular outcomes in people living with T1D. All authors contributed to the collection of the data and data interpretation. Massimo Nardone performed the statistical analysis. Luxcia Kugathasan and Massimo Nardone wrote the first draft of the manuscript. David Z. I. Cherney, Vikas S. Sridhar, Massimo Nardone, and Luxcia Kugathasan were involved in the study design. Luxcia Kugathasan, Michael A. Tsoukas, Massimo Nardone, Vikas S. Sridhar, Marcel H. A. Muskiet, David Z. I. Cherney, Ahmad Haidar, and Melissa-Rosina Pasqua provided critical revision for important intellectual content and approved the final version of the manuscript for submission. David Z. I. Cherney is the guarantor of this work and, as such, takes full responsibility for the work and/or conduct of the study, had access to the data, and controlled the decision to publish. The authors have nothing to report. This study has no funding to report. The funding for the original study was supported by the Canada Research Chair, awarded to Ahmad Haidar. Massimo Nardone is supported by a Banting and Best Diabetes Centre Postdoctoral Fellowship at the University of Toronto. Luxcia Kugathasan is supported by the CIHR CANTRAIN Postdoctoral Fellowship Award. Vikas S. Sridhar is supported by the Department of Medicine Eliot Phillipson Clinician Scientist Training Program, CIHR Banting and Best Doctoral Award. Marcel H. A. Muskiet is supported by the Professor Michaela Diamant Postdoctoral Fellowship. David Z. I. Cherney is the Gabor Zellerman Chair in Nephrology Research, University of Toronto, and is also supported by a CIHR-Kidney Foundation of Canada Team Grant Award, with additional support from Breakthrough-T1D. Luxcia Kugathasan has no potential conflicts of interest relevant to this article. Michael A. Tsoukas has received speaking honoraria from CPD, Boehringer-Ingelheim, Eli Lilly, Novo Nordisk, Janssen, Sanofi, and Bayer. Massimo Nardone has no potential conflicts of interest relevant to this article. Vikas S. Sridhar has received conference and travel support from Merck Canada and Lexicon pharmaceuticals and has received advisory board fees from Novo Nordisk Canada. Marcel H. A. Muskiet serves as a speaker-consultant and advisory board member for AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Novo Nordisk, and Sanofi; all honoraria paid to his employer. David Z. I. Cherney has received honoraria from Boehringer Ingelheim-Lilly, Merck, AstraZeneca, Sanofi, Mitsubishi-Tanabe, Abbvie, Janssen, Bayer, Prometic, BMS, Maze, Gilead, CSL-Behring, Otsuka, Novartis, Youngene, and Novo Nordisk, and has received operational funding for clinical trials from Boehringer Ingelheim-Lilly, Merck, Janssen, Sanofi, AstraZeneca, CSL-Behring, and Novo Nordisk. He is the recipient of a 5-year Team Grant in Diabetes Complications from the CIHR-Kidney Foundation of Canada, with additional support from Breakthrough T1D. He is the Gabor Zellerman Chair in Nephrology Research, University of Toronto. Ahmad Haidar has acted as a consultant for Eli Lilly and has received drugs, supplies, equipment, and other in-kind support from Tandem, Adocia, Dexcom, Eli Lilly, and Ypsomed. Melissa-Rosina Pasqua has received speaker honoraria from Medtronic Diabetes, Sanofi, and Abbott. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70525. No new data were generated for this post hoc analysis. The data analysed and the protocol of the original study have been previously published. All raw data collected from the study cannot be made publicly available given limitations from the informed consent form and ethics board approval. The raw data can be shared by the Melissa-Rosina Pasqua without cost for unrestricted non-commercial purposes; however, the use will be subject to approval from the Research Ethics Board of the McGill University Health Centre. After approval, reasonable efforts will be made for the data to be shared within 3 months.