Sodium-glucose cotransporter-2 inhibitors use and risk of dementia in different ethnoracial groups with type 2 diabetes

Dear Editor,

The link between dementia and glucose-lowering drugs has been widely studied. Tang et al. reported that new users of sodium-glucose cotransporter-2 (SGLT2) inhibitors had a lower risk of dementia compared to sulfonylurea users.¹ Other glucose-lowering drugs, like metformin and glucagon-like peptide-1 (GLP-1) receptor agonists (GLP-1RAs), are also associated with reduced dementia risk and slower cognitive decline in type 2 diabetes patients. Previous studies used various glucose-lowering drugs as active comparators to assess SGLT-2 inhibitors' effects on dementia.^{2, 3} However, studies have shown significant ethnoracial disparities in dementia, with Black individuals (compared to White individuals) and Hispanic individuals (compared to non-Hispanic, predominantly White individuals) experiencing higher prevalence and incidence rates of dementia.^{4, 5} Moreover, both Black and Hispanic populations are at a greater risk of developing type 2 diabetes, which is itself a known risk factor for dementia. Ensuring that these at-risk ethnoracial groups can access and benefit from targeted interventions is therefore essential for promoting brain health equity. To address the knowledge gap, we hereby conducted an analysis using the TriNetX research network, which has not been fully explored in previous studies.^{6, 7}

We utilized the TriNetX research network, specifically the US Collaborative Network subset, which includes 66 US institutions and is updated monthly, and applied in various epidemiological studies.^{8, 9} It offers access to anonymized electronic medical records and claims data, including patient information from healthcare organizations (HCOs) related to disease diagnoses, medication prescriptions, medical procedures, and laboratory results. Participants were aged 50 years and older, diagnosed with diabetes mellitus between January 1, 2018, and December 31, 2021, and had more than two visit records. The control group included diabetes patients using sulfonylureas, metformin, dipeptidyl peptidase-4 (DPP4) inhibitors, and GLP-1 RAs. Exclusions were made for those under 50; those with prior dementia, Parkinson's disease, or Alzheimer's disease; or those who were deceased before the index date. The case group consisted of SGLT-2 users. Individuals who had ever used the alternate drug were excluded when comparing the two drug groups. Before matching and exclusion of prior glucose-lowering drug use, there were 279,122 SGLT-2 inhibitor users, 512,944 sulfonylurea users, 1,301,591 metformin users, 319,098 DPP4 inhibitor users, and 306,499 GLP-1 RA users. A 1:1 propensity score matching was performed in all analyses. 1:1 ratio. The primary outcome was new-onset dementia, with participants monitored from 3 months post-index date until dementia occurrence or their last visit, whichever came first. To assess potential ethnoracial disparities, we conducted stratification analyses based on ethnoracial subgroups, including White, Black, Hispanic, and Asian subgroups. All analyses were conducted using the TriNetX analytical platform, and the research was exempt from Institutional Review Board (IRB) approval.

Our analysis indicates that SGLT2 inhibitor users had a lower hazard ratio (HR) of dementia compared to those using sulfonylureas (HR: 0.81, 95% confidence interval [CI]: 0.76 to 0.86), metformin (HR: 0.88, 95% CI: 0.80 to 0.97), and DPP4 inhibitors (HR: 0.70, 95% CI: 0.66 to 0.74), but increased risk compared to GLP1-RA users (HR: 1.12, 95% CI: 1.05 to 1.20). In Whites, SGLT2 inhibitors were associated with a lower risk compared to sulfonylureas and DPP4 inhibitors, but with a slightly higher risk compared to GLP1-RA, with no significant difference compared to metformin. In Blacks, SGLT2 inhibitors had a lower risk compared to sulfonylureas and DPP4 inhibitors, but no significant difference compared to metformin or GLP1-RA. Among Hispanics, SGLT2 inhibitors were associated with a lower risk compared to sulfonylureas, DPP4 inhibitors, and GLP1-RA, but showed no significant difference with metformin. In Asians, no statistically significant differences were observed (Figure 1).

Our findings suggest that SGLT2 inhibitors may reduce dementia risk compared to other blood sugar medications, with benefits extending across different ethnoracial groups. However, we did not test the interaction between ethnoracial group and treatment terms, so we cannot conclude if these differences are significant. Additionally, the impact of changing drug use was not fully assessed. In line with Tang et al.,¹ we report that Hispanic SGLT2 users present a significantly lower dementia risk compared to sulfonylurea and GLP1-RA users. Moreover, the lack of significance in the Asian group may be due to the small sample size, highlighting the need for larger studies on Asian Americans. Notably, in the current study, SGLT2 users had a higher dementia risk compared to GLP1-RA users, especially in Whites, whereas this trend was reversed in Hispanics. The association may be influenced by unmeasured factors like medication dosage, warranting further head-to-head studies of SGLT2 versus GLP1-RA for dementia risk.

All the authors were involved in drafting or revising the article and approved of the submitted version. Study conception and design: Shih-Chi Yang, Wen-Chieh Liao, Yi-Sheng Jhang, Hui-Chin Chang, Shiu-Jau Chen, Yu-Jung Su, Shuo-Yan Gau. Data acquisition: Hui-Chin Chang and Shuo-Yan Gau. Data analysis and demonstration: Shih-Chi Yang, Hui-Chin Chang, Shiu-Jau Chen, and Shuo-Yan Gau. Original draft preparation: Shih-Chi Yang, Wen-Chieh Liao, Yi-Sheng Jhang, Hui-Chin Chang, Shiu-Jau Chen, Yu-Jung Su, Shuo-Yan Gau.

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.

Data in this study were retrieved from TriNetX Research Network. All data available in the database were administrated by the TriNetX platform. Detailed information can be retrieved at the official website of the research network (https://trinetx.com).

The TriNetX database was previously approved by the Western Institutional Review Board (Western IRB). The subsequent determination regarding the de-identification process attested on December 2020 replaced the need for Western IRB approval in TriNetX studies. Ethical approval was not required for the study involving humans in accordance with local legislation and institutional requirements. Written informed consent to participate in this study was not required from the participants or the participants' legal guardians/next of kin in accordance with national legislation and institutional requirements. The data were reviewed as a secondary analysis of existing data, which did not involve intervention or interaction with human subjects.