Impact of Glucagon‐Like Peptide‐1 Receptor Agonists on the Dementia Incidence in Patients With Type 2 Diabetes Mellitus: A Population‐Based Longitudinal Cohort Study

Dementia represents a rapidly growing global health challenge, with > 50 million individuals affected worldwide as of 2016. This number may triple by 2050, owing to increasing life expectancies [1, 2, 3]. It stands as a leading cause of disability among older people. Despite new monoclonal antibody treatments aimed at slowing dementia progression, their use may be limited by severe adverse reactions. Numerous studies have indicated a link between type 2 diabetes mellitus (T2DM) and increased risk of cognitive decline and dementia [4, 5, 6]. The co‐occurrence of T2DM and dementia imposes a substantial economic burden on both affected families and healthcare systems. Consequently, mitigating the effect of dementia incidence in individuals with T2DM emerges as a critical health priority.

Previous studies have highlighted the importance of achieving optimal glycaemic control and minimising diabetic complications in reducing dementia risk [7, 8]. Moreover, studies have demonstrated that glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) contribute to favourable glycaemic control, weight loss, and renal protection [9, 10, 11, 12]. Despite these findings, the precise association between GLP‐1RAs and dementia incidence remains vague. Thus, we undertook a retrospective cohort study to elucidate the relationship between GLP‐1RAs and dementia incidence within the broader Taiwanese population. This study aimed to discern whether the utilization of GLP‐1RAs correlates with incident dementia risk in a cohort of Taiwanese patients diagnosed with T2DM.

This cohort study demonstrated that new GLP‐1RAs use compared with non‐GLP‐1RAs use was associated with lower dementia risk in people with T2DM. The result was robust in the sensitivity and subgroup analyses. Specifically, among patients with severe T2DM with complications (such as retinopathy, nephropathy, and neuropathy) and T2DM duration < 10 years, GLP‐1RAs users displayed lower rates of incident dementia than non‐GLP‐1RAs users. It appeared that patients with newly diagnosed T2DM may exhibit a more pronounced preventive effect against dementia compared with patients with existing T2DM.

Patients with T2DM have an accelerated rate of cognitive decline and a 1.6‐fold increased risk of dementia development [14, 15]. Important and shared pathological features of T2DM with cognitive decline and dementia, which are characterised by metabolic brain alterations, for example, insulin resistance, altered glucose uptake, and altered glucose utilization. These similarities in pathology are reflected in clinical studies that have demonstrated an increased dementia risk in individuals with T2DM [16, 17, 18].

Among antidiabetic drugs, glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) play an interesting role in the modulation of neuroinflammation [19, 20, 21]. GLP‐1RAs also play neurotropic and neuroprotective roles in the central nervous system [22]. In addition, GLP‐1RAs reduce Aβ aggregation/deposition and hyperphosphorylation of tau protein, oxidative stress, and neuronal apoptosis, increasing cell proliferation and neurogenesis [23]. In vitro and animal studies have shown that pretreatment with bilateral intrahippocampal injection of liraglutide improves learning and memory deficit in murine ad models [24, 25]. Two clinical studies have revealed a lower risk of all‐cause dementia among GLP‐1RAs users than among non‐GLP‐1RAs users [26, 27].

Using multiple strategies to mitigate bias after adjustments for sex, age, comorbidities, and concurrent medication, the present cohort study strengthened previous findings that GLP‐1RAs use was associated with a lower dementia risk [27, 28, 29]. This study suggests that in patients with T2DM taking GLP‐1RAs, the risk of dementia development is 10% lower than that in non‐users, which may be the possible cause for the increase in its use. Intensive risk factor modification, particularly during midlife (age 45–65 years) was considered to have the potential to delay or prevent a substantial number of dementia cases worldwide [30, 31]. Our cohort study proved that GLP‐1RAs use significantly reduced the incidence of dementia in participants aged < 65 and 65–75 years (aHR, 0.87; 95% CI, 0.77–0.98; aHR, 0.90; 95% CI, 0.83–0.99). GLP‐1RAs reduce the incidence of dementia equally in both sexes, regardless of the duration of T2DM being < 6 or 6–10 years and presence of complications such as retinopathy, neuropathy, and neuropathy. Comparatively, the incidence rate of dementia by GLP‐1RAs type and use versus non‐use of liraglutide and dulaglutide correlated with a decreased incidence of dementia.

In the pooled randomized controlled trial analysis during a median follow‐up of 3.61 years, patients randomized to GLP‐1RAs had a lower rate of developing dementia than those randomized to placebo (HR, 0.47; 95% CI, 0.25–0.86) [27]. The results of the meta‐analysis of four studies showed that GLP‐1RAs users had a significantly lower risk of all‐cause dementia than non‐GLP‐1RAs users (RR, 0.72; 95% CI, 0.54–0.97); however, a high level of heterogeneity was found between studies (I² = 91.3%) [29].

This study has several limitations that warrant discussion. First, although exposure to GLP‐1RAs within the cohort was confirmed through claims data, the data lacked information on treatment adherence. This absence restricts our ability to fully evaluate the effect of treatment consistency on outcomes. Second, health services, preventive services, crucial laboratory data including blood sugar levels, haemoglobin A1c levels, renal and liver function markers, and biomarkers of dementia such as serum or cerebrospinal fluid neurofilament light chain, or scores of Clinical Dementia Rating‐Sum of Boxes, were not accessible within the secondary data utilised. However, because the data were population‐based, we assumed that there were no differences between the groups. Third, because this study relies on population‐based data from the Taiwan NHI programme and claims datasets, its generalisability to other countries may be limited. On the contrary, this study can be regarded as the first analysis focussing on Asian populations. Note that healthcare systems and patient populations vary across regions, potentially influencing treatment patterns and outcomes. Fourth, educational level appears to have been confirmed as potentially related to the incidence of dementia. However, due to limitations in the database, this study was unable to balance the differences between the experimental and control groups. Fifth, the mechanism by which GLP‐1RAs protect neurons in the brain remains unclear. In theory, the large‐molecule GLP‐1RAs currently available on the market cannot cross the blood‐brain barrier to directly provide neuroprotection. Recent studies suggest that GLP‐1RAs may indirectly protect neurons through improvements in glycaemic control, weight reduction, and physiological metabolism, thereby reducing the incidence of dementia. To mitigate these limitations, rigorous statistical adjustments were employed to account for potential differences between groups. However, given the inherent constraints of observational studies, further validation through randomized clinical trials is warranted to confirm the findings and establish causality.

Collectively, this study revealed that GLP‐1RAs may represent a promising new therapeutic approach to reducing the incidence of dementia, warranting further attention and in‐depth investigation of their cognitive‐protective potential in conditions such as T2DM and dementia. Careful evaluation of their risk–benefit ratio is necessary to establish an evidence‐based foundation for future clinical applications.

Hung‐Wen Cheng, Chiu‐Hsian Lee, and Gwo‐Ping Jong conceptualized the study and drafted the manuscript. Hung‐Wen Cheng and Gwo‐Ping Jong had significant input into data curation for the analysis. Hung‐Wen Cheng, Shun‐Fa Yang, Chiu‐Hsian Lee, and Gwo‐Ping Jong had significant contribution to the study methodology. Hung‐Wen Cheng, Shun‐Fa Yang, and Gwo‐Ping Jong had significant input into the investigation process. Hung‐Wen Cheng and Pei‐Lun Liao performed the formal analysis and visualized the results, and Gwo‐Ping Jong validated the analysis. All authors made significant contributions to the interpretation of the results and the review or editing of the manuscript. Hung‐Wen Cheng, Chiu‐Hsian Lee, and Gwo‐Ping Jong were responsible for project administration. Hung‐Wen Cheng acquired funding for this study. This study used data from the TBNHI Data Repository. Chiu‐Hsian Lee and Gwo‐Ping Jong are the guarantors of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

This study was approved by the ethics committee of Chung Shan Medical University Hospital (CS2‐24009). Written consent was not sought from the study participants as only de‐identified data were obtained from the TBNHI. The ethics committee waived the need for patient consent for this study.

The authors declare no conflicts of interest.

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1002/dmrr.70058↗.

We are grateful to the Health Data Science Centre, Chung Shan Medical University Hospital, for providing administrative, technical, and funding support, which has contributed to the completion of this study. This study is based, in part, on data released by the Health and Welfare Data Science Centre, Ministry of Health and Welfare.

The data used during this study are available from the corresponding authors on reasonable request.