Introduction

Type 2 diabetes and obesity are growing global health concerns. Between 2017 and 2020, 62.8% of US adults with diabetes also had obesity (body mass index [BMI] ≥30), which is approximately 17 million individuals.1 These metabolic disorders, characterized by insulin resistance, oxidative stress, and chronic inflammation, contribute to neurodegeneration.2 The 2024 Lancet Commission on Dementia identified diabetes (relative risk, 1.7) and obesity (relative risk, 1.3) as modifiable dementia risk factors,3 with their population-attributable fractions increasing over time.4 Patients with both conditions face an elevated risk of neurodegenerative and cerebrovascular diseases, including dementia, Parkinson disease, and stroke.5

Large cohort studies investigating the neuroprotective properties of antidiabetic drugs have suggested that glucagon-like peptide 1 receptor agonists (GLP-1RAs) are among the most promising antidiabetic drugs in reducing dementia risk.6-8 First approved for type 2 diabetes treatment in 2005, GLP-1RAs have transformed diabetes and obesity management by improving glycemic control, weight loss, and cardiovascular outcomes.9,10 Semaglutide and tirzepatide are newer-generation agents with superior metabolic benefits compared with earlier GLP-1RAs, such as exenatide, liraglutide, lixisenatide, and dulaglutide.11 Semaglutide, a long-acting GLP-1RA with high receptor affinity, was approved by the US Food and Drug Administration for type 2 diabetes as Ozempic (Eli Lilly) in 2017 and Rybelsus (Novo Nordisk) in 2019 and for weight management as Wegovy (Novo Nordisk) in 2021.12 Tirzepatide, a dual GLP-1 and glucose-dependent insulinotropic polypeptide receptor agonist, was approved for type 2 diabetes as Mounjaro (Eli Lilly) in 2022 and for obesity as Zepbound (Eli Lilly) in 2023. Tirzepatide enhances incretin signaling and insulin secretion and promotes greater weight loss than semaglutide.13-16 Beyond metabolic benefits, semaglutide has shown promise in cardiovascular disease17 and alcohol and substance use disorder treatment18-20 and neurocognitive protection.21 Although tirzepatide’s neuroprotective outcomes remain clinically unconfirmed, it has shown favorable cardiometabolic and neuroprotective outcomes in cellular models.22,23

Preclinical studies have shown that GLP-1RAs exert neurocognitive protection by reducing amyloid-β deposition, tau hyperphosphorylation, neuroinflammation, and oxidative stress, while also improving cognitive function in murine Alzheimer disease24,25 and human brain organoid models.26 Clinical studies also have suggested promising neuroprotection of GLP-1RAs,27-29 as evidenced by post hoc analyses of 3 global randomized clinical trials, including LEADER for liraglutide27 and SUSTAIN 628 and PIONEER 6 for semaglutide,29reporting a hazard ratio (HR) for dementia of 0.47 (95% CI, 0.25-0.86).30 Additionally, a nationwide Danish registry–based study also found a lower risk of dementia among GLP-1RA users (HR, 0.89; 95% CI, 0.86-0.93).30 Currently, 2 ongoing global phase 3 clinical trials (EVOKE and EVOKE+) are investigating semaglutide’s neuroprotective efficacy in early-stage symptomatic Alzheimer disease.31 Regarding Parkinson disease, earlier GLP-1RAs, such as liraglutide and lixisenatide, have been shown to protect dopaminergic neurons and reduce motor disability in animal models.32 In addition, lixisenatide slowed motor decline in a phase 2 randomized clinical trial in patients with early Parkinson disease.33 These findings support further investigation into the efficacy of newer GLP-1RAs in ameliorating neurodegenerative diseases.

The vascular complications of diabetes also significantly increase stroke risk, as patients have a twofold higher incidence of stroke compared with individuals without diabetes34 and a 60% increased risk of stroke recurrence.35 For cerebrovascular protection in ischemic stroke models, GLP-1RAs have been shown to reduce infarct size and improve vascular function.36 A large US cohort study found that semaglutide was associated with a reduced risk of ischemic stroke in patients with type 2 diabetes, especially in those with cardiovascular disease.37 Similarly, another Danish study comparing GLP-1RAs and dipeptidyl peptidase 4 (DPP-4) inhibitors in patients with type 2 diabetes found that GLP-1RA users had a lower risk of severe stroke (adjusted risk ratio, 0.49; 95% CI, 0.24-1.00) and reduced mortality.38

Semaglutide and tirzepatide have longer half-lives and enhanced pharmacokinetics, with emerging evidence suggesting greater neuroprotective potential than earlier GLP-1RAs. However, large-scale clinical studies evaluating their association with neurodegenerative and cerebrovascular outcomes remain limited, particularly in high-risk patients with both type 2 diabetes and obesity. The aim of this study was to assess the incidence of dementia, Parkinson disease, stroke, intracerebral hemorrhage, and all-cause mortality among patients treated with semaglutide or tirzepatide vs other antidiabetic drugs using electronic health record–based data.

Methods

Data Source

This retrospective cohort study used data from the TriNetX platform, a global health research network that provides electronic health records for approximately 275 million individuals across 19 countries, covering a diverse range of ages, races and ethnicities, income levels, insurance groups, and clinical settings. The dataset includes diagnoses, medications, laboratory results, procedures, demographics, and clinical outcomes, enabling robust clinical evidence generation. The TriNetX platform is compliant with the Health Insurance Portability and Accountability Act and General Data Protection Regulation. The Western Institutional Review Board granted TriNetX a waiver of informed consent because this platform only aggregates counts and statistical summaries of deidentified information. The use of TriNetX in this study was approved by the Chung Shan Medical University Hospital Institutional Review Board. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (eTable 1 in Supplement 1).

This study primarily analyzed data from the TriNetX US collaborative network, which aggregates deidentified electronic health records from 67 US health care organizations. The study period spanned from December 1, 2017, to June 30, 2024 (the latest identifiable date in the executed analysis period as of December 2024), corresponding to the Food and Drug Administration approval timelines for semaglutide (December 2017) and tirzepatide (May 2022) for type 2 diabetes and obesity management.

Study Population

We enrolled adult patients aged 40 years or older with diagnoses of both type 2 diabetes and obesity (International Statistical Classification of Diseases, Tenth Revision codes E11 and E66, respectively) who were new users of semaglutide, tirzepatide, or other antidiabetic drugs between 2017 and 2024.7 Patients were included for analysis if they had at least 1 year of follow-up after the index prescription. Exclusion criteria included prior diagnosis of type 1 diabetes or neurodegenerative or cerebrovascular diseases before the index date, prior use of other GLP-1RAs (exenatide, dulaglutide, liraglutide, and lixisenatide), and a switch to semaglutide or tirzepatide within 3 months of initiating another antidiabetic treatment. The final cohort included 52 826 GLP-1RA users and 161 945 other antidiabetic drug users (Figure 1). Patients receiving semaglutide (Ozempic, Rybelsus, or Wegovy) or tirzepatide (Zepbound or Mounjaro) were assigned to the GLP-1RA group, while those treated with biguanides, sulfonylureas, α-glucosidase inhibitors, thiazolidinediones, DPP-4 inhibitors, and sodium-glucose cotransporter 2 inhibitors were assigned to the other antidiabetic drug group.

Prespecified Outcomes and Covariates

The primary outcomes were new-onset neurodegenerative diseases, including dementia (Alzheimer disease, vascular dementia, and other dementia), mild cognitive impairment, and Parkinson disease, and cerebrovascular diseases, including ischemic stroke and intracerebral hemorrhage. The secondary outcome was all-cause mortality, identified from death records in the TriNetX database during the study period. Demographic covariates included age, sex, BMI (calculated as weight in kilograms divided by height in meters squared), race (American Indian or Alaska Native, Asian, Black, Native Hawaiian or Other Pacific Islander, White, other [multiracial or not otherwise specified], or unknown) and ethnicity (Hispanic or Latino or not Hispanic or Latino), smoking status, alcohol use, chronic diseases (hypertension, hyperlipidemia, ischemic heart disease, and chronic kidney disease), glycated hemoglobin (HbA1c) level, estimated glomerular filtration rate, socioeconomic factors (education, employment, and income), and health care use. Race and ethnicity were derived from electronic health records and included because the efficacy of GLP-1RAs may differ among racial and ethnic groups. Covariates were measured within 6 months prior to each patient’s index date to reflect baseline characteristics before treatment initiation. A list of covariates and their corresponding codes is provided in eTable 2 in Supplement 1.

Statistical Analysis

The study maintained an active comparator, new-user design, with propensity score matching that adjusted for baseline use of other antidiabetic drugs to minimize confounding by cotreatment.39 The TriNetX built-in function was used to generate propensity score–matched groups based on demographic and clinical variables using 1:1 greedy nearest neighbor matching without replacement and a caliper of 0.1 SDs. Missing baseline data were addressed by stratified matching, categorizing variables such as HbA1c into clinically meaningful strata. Due to the deidentified and dynamic nature of the database, direct assessment of loss to follow-up was not feasible. The standardized mean difference was used to evaluate the balance of baseline characteristics between matched groups, with values less than 0.100 indicating a well-balanced comparison.40 Patients in each group were matched for age, sex, ethnicity, race, BMI, HbA1c, estimated glomerular filtration rate, baseline comorbidities, socioeconomic status, and health care use. The proportional hazards assumption was assessed using Schoenfeld residual plots and statistical tests, with all P > .05 indicating no significant violations among the tested covariates. Cox proportional hazards models were applied to estimate the HRs and 95% CIs for neurodegenerative and cerebrovascular outcomes. Cumulative probabilities were assessed using Kaplan-Meier analysis, and group differences were evaluated using the log-rank test. Subgroup analyses were performed based on sex (male vs female), age (40-60 vs ≥60 years), race (Black vs White), BMI (30-40 vs ≥40), and GLP-1RA agent (semaglutide or tirzepatide vs other antidiabetic drugs). Sensitivity analyses compared GLP-1RAs with DPP-4 inhibitors and were replicated in the TriNetX global network for comparison, which includes regions such as the US, Europe, Asia-Pacific, and Latin America, to assess the robustness of our findings. All analyses were performed using the survival, version 3.2-3 package in R, version 4.0.2 (R Foundation for Statistical Computing) on the TriNetX Analytics Platform. Statistical significance was defined as a 2-sided P < .05.

Results

Primary Outcomes

Among the 1 689 828 adults aged 40 years or older with type 2 diabetes and obesity enrolled in TriNetX, 60 860 were included, with 30 430 included in both the GLP-1RA (mean [SD] age, 57.9 [9.9] years; 50.2% female, 42.8% male, and 7.0% unknown sex; 0.4% of American Indian or Alaska Native, 3.7% Asian, 17.9% Black, 1.3% Native Hawaiian or Other Pacific Islander; 55.6% White, 5.4% other, and 15.7% unknown race; 8.4% of Hispanic or Latino, 63.1% of non–Hispanic or Latino, and 28.5% unknown ethnicity) and other antidiabetic drug (mean [SD] age, 58.0 [10.8] years; 51.4% female, 43.0% male, and 5.6% unknown sex; 0.4% of American Indian or Alaska Native, 3.7% Asian, 17.5% Black, 1.5% Native Hawaiian or Other Pacific Islander; 56.1% White, 5.3% other, and 15.5% unknown race; 8.2% of Hispanic or Latino, 61.8% non–Hispanic or Latino, and 30.0% unknown ethnicity) groups after propensity score matching (Table 1). Over a mean (SD) follow-up of 1.77 (1.35) years, GLP-1RA users (semaglutide and tirzepatide) had a significantly lower risk of dementia compared with other antidiabetic drug users (HR, 0.63; 95% CI, 0.50-0.81) (Table 2). The analysis of dementia subtypes revealed a significant association between GLP-1RA use and reduced risk of other dementia (HR, 0.69; 95% CI, 0.56-0.84) but not with Alzheimer disease or vascular dementia (eTable 3 in Supplement 1). These findings of dementia subtypes may reflect small numbers of patients in these subgroups and underscore the need for larger-scale studies to confirm these associations. No statistically significant reduction was observed in the comparison of mild cognitive impairment or Parkinson disease. Regarding cerebrovascular outcomes, GLP-1RA use was associated with a lower risk of stroke (HR, 0.81; 95% CI, 0.70-0.93) but not intracerebral hemorrhage. The Kaplan-Meier curves for the cumulative probability of neurodegenerative and cerebrovascular diseases (Figure 2) showed a clear divergence between GLP-1RA (7.19%; 95% CI, 6.12%-8.44%) and other antidiabetic drug users (9.96%; 95% CI, 8.29%-11.93%) (log-rank P < .001). Notably, this divergence of Kaplan-Meier curves gradually increased after 1 year of follow-up, suggesting a possible time-varying treatment effect.

All-Cause Mortality

The secondary outcome of all-cause mortality was significantly lower among patients treated with GLP-1RAs (HR, 0.70; 95% CI, 0.63-0.78) compared with those treated with other antidiabetic drugs. To address death as a potential competing risk for neurodegenerative and cerebrovascular outcomes, we conducted supplementary analyses using composite end points, which yielded consistent results (eTable 4 in Supplement 1).

Subgroup and Sensitivity Analyses

Subgroup analyses (Table 3) showed associations between GLP-1RA use and neuroprotection across different subgroups, including adults aged 60 years or older (HR, 0.85; 95% CI, 0.74-0.99) and women (HR, 0.85; 95% CI, 0.74-0.97). When stratified by Black and White race, a significant association was observed among White patients (HR, 0.86; 95% CI, 0.76-0.98). When stratified by obesity categories, patients with a BMI of 30 to 40 (HR, 0.82; 95% CI, 0.72-0.93) showed a significant association, whereas no association was observed in those with a BMI of 40 or higher. Further comparison of semaglutide and tirzepatide vs other antidiabetic drugs revealed a significant reduction in dementia risk with semaglutide (HR, 0.63; 95% CI, 0.50-0.79) and stroke risk with tirzepatide (HR, 0.69; 95% CI, 0.48-0.98) (eTable 5 in Supplement 1). Tirzepatide was also significantly associated with lower all-cause mortality (HR, 0.48; 95% CI, 0.35-0.66) compared with other antidiabetic drugs.

In sensitivity analyses using the TriNetX global network (eTable 6 in Supplement 1), the results remained consistent with those from the TriNetX US network, with beneficial associations of GLP-1RAs with neurodegenerative and cerebrovascular diseases and all-cause mortality. Finally, we conducted a sensitivity analysis comparing new users of semaglutide or tirzepatide vs DPP-4 inhibitors in the TriNetX US network (eTable 7 in Supplement 1), and the results remained consistent with the main analyses, further reinforcing the robustness of our findings.

Discussion

This cohort study represents one of the most recent clinical database–driven analyses to investigate the neuroprotective and cerebrovascular associations of newer GLP-1RAs, semaglutide and tirzepatide, in patients with both type 2 diabetes and obesity. Our findings show that initiating GLP-1RA treatment was associated with a reduced incidence of dementia, stroke, and all-cause mortality in patients with type 2 diabetes and obesity, with more pronounced associations observed in older adults; women, who are known to have higher baseline risks of dementia41-43; and individuals with a BMI of 30 to 40, possibly due to their more modifiable metabolic status. These subgroup trends warrant further validation in randomized clinical trials. Given tirzepatide’s recent approval in 2022, its shorter follow-up duration may have limited the observed outcomes in this cohort, which may have been driven predominantly by semaglutide. Future research should further elucidate the distinct neuroprotective mechanisms and long-term cognitive benefits of both agents.31,44,45

Unlike prior cohort studies focusing solely on type 2 diabetes46 or obesity,9 our large-scale analysis specifically targeted high-risk patients with both diabetes and obesity treated with the newer GLP-1RAs semaglutide and tirzepatide. Epidemiologic studies suggested that the co-occurrence of type 2 diabetes and obesity may increase dementia risk through additive effects.47 Obesity-induced systemic inflammation and oxidative stress, when compounded by insulin resistance and mitochondrial dysfunction in type 2 diabetes, may accelerate cognitive decline and dementia development.5 Our findings align with previous studies supporting the neuroprotective potential of GLP-1RAs in dementia30,46 but not in mild cognitive impairment,6 suggesting that the neuroprotection offered by GLP-1RAs may become more evident during later stages of cognitive decline. Notably, GLP-1RA use was associated with a lower risk of dementia (HR, 0.63; 95% CI, 0.50-0.81) in patients with type 2 diabetes and obesity, particularly in those aged 60 years or older and in women. Prior cohort studies have consistently reported higher dementia risks in older adults and women,41-43 suggesting that the initiation of GLP-1RAs in these populations might therefore confer enhanced neuroprotection. Moreover, longitudinal studies have shown that earlier onset of diabetes is associated with a higher risk of subsequent dementia.48,49 Therefore, dementia prevention strategies should begin in midlife and include multifactorial interventions, such as promoting physical activity, supporting smoking cessation, and managing modifiable risk factors like diabetes and obesity, potentially through the use of GLP-1RAs.3,4

Clinical studies evaluating the association of semaglutide and tirzepatide with Parkinson disease remain limited.9 We did not observe a significant protection of semaglutide or tirzepatide against Parkinson disease in patients with type 2 diabetes and obesity, contrasting with prior animal model data showing that the earlier GLP-1RAs enhanced dopaminergic neuron survival.44 This discrepancy may reflect differences in the neuroprotective mechanisms of semaglutide and tirzepatide vs earlier GLP-1RAs, such as exenatide, liraglutide, and lixisenatide, which have demonstrated favorable clinical benefits in early-stage Parkinson disease.33,50 Tirzepatide, in particular, has been reported to exhibit slower brain uptake in murine models,51 and further research is needed to determine whether variations in receptor selectivity, pharmacokinetics, or blood-brain barrier permeability contribute to these divergent outcomes of GLP-1RAs in Parkinson disease.

Our study also provides electronic health record–based evidence supporting the association of GLP-1RAs with a reduced risk of ischemic stroke, particularly with tirzepatide. These results align with the PIONEER 6 and SUSTAIN 6 trials,36 which demonstrated stroke risk reduction with semaglutide, and may be attributed to improvements in endothelial function and atherosclerotic regulation.36 Beyond its greater weight loss benefits compared with semaglutide, tirzepatide’s dual GLP-1 and glucose-dependent insulinotropic polypeptide receptor agonism may further enhance incretin responses and insulin secretion, potentially leading to a substantially lower risk of cerebral infarction than semaglutide.52 However, the risk of intracerebral hemorrhage was not significantly altered, indicating that GLP-1RAs may predominantly impact small vessel ischemia rather than hemorrhagic pathology.

One notable finding of this study was the reduction in all-cause mortality among GLP-1RA users (HR, 0.70; 95% CI, 0.63-0.78) compared with users of other antidiabetic drugs. This mortality benefit is consistent with previous studies, suggesting that GLP-1RAs may not only improve cardiometabolic health but also enhance overall survival.12,53,54 Potential contributors to reduced mortality include metabolic modulation, decreased cardiovascular and stroke events, kidney protection, anti-inflammation, and immunomodulation.12,54 Additionally, semaglutide and tirzepatide modulate dopamine-mediated reward pathways, reducing food cravings and substance addiction, offering potential benefits in lowering substance abuse–related mortality.55 Accordingly, GLP-1RAs may help alleviate the overall disease burden of obesity and diabetes, providing cardiometabolic and neurocognitive benefits that improve health and life expectancy.

Strengths and Limitations

This study has several strengths. First, we adopted an active comparator, new-user design with extensive propensity score matching to minimize potential indication bias.56 The use of diagnosis codes and disease-specific drug exposures increased the specificity of the compared outcomes. Additionally, the subgroup outcomes and sensitivity analyses added relevant internal validity to this study. Compared with previous cohort studies, our study specifically focused on a high-risk population with both type 2 diabetes and obesity and included a relatively younger cohort (aged ≥40 years), thereby expanding the potential beneficiary group.

Our study also had several limitations. First, as an observational study, residual confounding from unmeasured factors, such as frailty or functional status, cannot be excluded, potentially introducing healthy user or selection bias. These limitations warrant cautious interpretation of the observed associations despite extensive propensity score matching and adjustment for key clinical variables, including HbA1c. Second, the TriNetX database lacks biomarker data, genetic profiles, and neuroimaging assessments, limiting further mechanistic interpretations. Third, medication exposure was inferred from prescriptions without confirmation of actual adherence or accurate drug doses. Finally, we used Kaplan-Meier method and Cox proportional hazards models without accounting for death as a competing risk because the TriNetX platform does not support traditional competing-risks analysis with death as a censoring event. To partially address this limitation, we analyzed all-cause mortality as a secondary outcome and conducted supplementary analyses using composite end points. Future randomized clinical trials in high-risk populations with type 2 diabetes and obesity are warranted to validate our findings and establish causality. Such studies should compare GLP-1RAs with specific antidiabetic drugs without cotreatment, incorporate longitudinal cognitive and neuroimaging assessments, and evaluate baseline frailty, functional status, and outcome severity scales.

Conclusions

In this cohort study, we found electronic health record–based evidence that the use of GLP-1RAs, particularly semaglutide and tirzepatide, is associated with a lower incidence of dementia, stroke, and all-cause mortality in patients with both type 2 diabetes and obesity, suggesting potential neuroprotective and cerebrovascular benefits. These findings highlight the possible role of GLP-1RAs in mitigating neurodegenerative and cerebrovascular risks in this high-risk population.

Ler artigo completo

Neurodegeneration and Stroke After Semaglutide and Tirzepatide in Patients With Diabetes and Obesity

Neurodegeneration and Stroke After Semaglutide and Tirzepatide in Patients With Diabetes and Obesity

Bem-vindo(a) novamenteao Médico Exponencial!

Bem-vindo(a) novamente
ao Médico Exponencial!