Impact of antidiabetic therapy on cognitive function in patients with type 2 diabetes and Alzheimer's disease: A comprehensive analysis

Abstract

Background

Several hypotheses suggest that type 2 diabetes mellitus (T2DM) is a risk factor for Alzheimer's disease (AD), and antidiabetic medications may influence cognitive function in these patients.

Objective

This study aims to provide a comprehensive evaluation of the impact of single and combination antidiabetic therapies on cognitive function in patients with both AD and T2DM.

Methods

We analyzed data from the National Alzheimer's Coordinating Center (NACC), covering a 17-year period from June 2005 to August 2022. This study included 3234 patients with both AD and T2DM. After applying exclusion criteria, 964 patients were analyzed using propensity score matching and a generalized linear mixed model. Patients were categorized based on their longitudinal use of oral antidiabetic medications.

Results

Based on the 964 patients’ cohort, the study found that patients receiving metformin with sulfonylureas (RR = 1.105 [1.022, 1.195], p = 0.013) and metformin with a dipeptidyl peptidase 4 inhibitor (DPP-4i) (RR = 1.132 [1.021, 1.256], p = 0.019) experienced a significantly slower decline in MMSE scores over time when compared to patients receiving metformin with thiazolidinediones (TZD).

Conclusions

This study demonstrates that combination therapies involving metformin with sulfonylureas or DPP-4i are associated with a slower rate of cognitive decline compared to metformin with TZD in patients with AD and T2DM. These findings provide novel evidence for the long-term cognitive benefits of specific antidiabetic therapies and offer valuable insights for clinical decision-making in this dual-affected population.

Keywords

Alzheimer's disease drug repurposing mini-mental state examination type 2 diabetes

Introduction

Alzheimer's disease (AD), the most prevalent form of dementia, is characterized by progressive neural degeneration and cognitive decline.¹ Its current global trends suggest a surge to approximately 150 million cases by 2050.² Type 2 diabetes mellitus (T2DM), representing nearly 98 percent of the 530 million diabetes cases in adults aged 20 to 79,³ is one of the most common concomitant diseases in AD patients. According to a community-based research, up to 35% of AD patients have diabetes.⁴ Research indicates that patients with concurrent AD and T2DM face greater challenges in managing cognitive decline than those with a single condition,^5,6 which underscores the importance of addressing the specific challenges faced by patients who have both AD and T2DM.

Real-world evidence underscores the role of numerous oral antidiabetic agents, including metformin, dipeptidyl peptidase 4 inhibitors (DPP-4i), and sulfonylureas, in potentially mitigating the progression of AD. Long-term and high-dose metformin use was found to be associated with a reduced risk of developing AD in older adults with diabetes.⁷ Then in a cohort study, DPP-4i was found to reduce the risk of dementia more effectively than sulfonylureas in older adults with T2DM.⁸ Additionally, it has been reported that the combined use of sulfonylureas and metformin can reduce the risk of dementia by 35% over 8 years in T2DM patients.⁹ These studies indicate the potential of oral antidiabetic medications in mitigating cognitive deterioration. Conversely, limited reports suggest potential cognitive benefits with thiazolidinediones (TZD) use. In a phase III trial, rosiglitazone monotherapy showed no cognitive or global function benefits in AD patients.¹⁰ Moreover, TZD were found to be associated with a faster decline in AD patients.¹¹

Despite growing interest in the cognitive effects of antidiabetic therapies, most research has focused on single-drug treatments, leaving the impact of combination therapies largely underexplored. Since combination therapy is widely used in T2DM management, understanding its long-term effects on cognitive decline is essential for optimizing treatment strategies in patients with both AD and T2DM.

To address these gaps, this study comprehensively evaluates the impact of both single and combination antidiabetic therapies on cognitive function, measured by MMSE scores, in patients with AD and T2DM. By analyzing longitudinal data from a large cohort and incorporating a wide range of treatments, we aim to provide critical evidence that could inform more effective strategies for preserving cognitive function in this dual-affected population.

Methods

Data source

This study is a retrospective analysis utilizing data from the National Alzheimer's Coordinating Center (NACC) database, which collects and harmonizes data from 39 Alzheimer's Disease Centers (ADCs) across the United States.¹² This nationwide network contributes to one of the largest multicenter databases of carefully validated clinical and neuropathological data for AD. The NACC database, structured in a standardized manner to form a Uniform Data Set (UDS), has been continuously collecting data since 2005.^13,14 The minimum interval between assessments was 6 months, and the maximum interval was 13 months. Participants are enrolled through clinician referrals, self-referrals by patients or their family members, and active recruitment from community organizations and volunteers. Thus, the NACC database is considered a referral-based or volunteer case series. All participants and co-participants provided written informed consent.

Standard protocol approvals, registrations, and patient consents

Informed consent from the participants or their proxies was obtained through an institutional review board–approved protocol at each site for NACC data.

Subject selection

Among the 46,513 patients registered in NACC from June 2005 to August 2022, our study included 3234 patients with AD and T2DM. The diagnosis of AD was determined using the NACCALZD variable, which is based on the criteria established by the National Institute on Aging-Alzheimer's Association (NIA-AA).¹⁵ The presence of diabetes was identified using the clinician-assessed variables DIABETES and DIABTYPE, which was determined based on a clinical evaluation of each patient's medical history. Subjects with epilepsy, traumatic brain injury, vascular dementia, Parkinson's disease, Lewy body dementia, type 1 diabetes, gestational diabetes, diabetes insipidus, or latent autoimmune diabetes were excluded. Patients with only one visit were excluded. Patients with MMSE scores outside the valid range of 0–30 were excluded. Specifically, we retained only observations with MMSE scores coded from 0 to 30 and excluded those with special codes indicating missing or incomplete assessments, as defined by the NACC. To avoid floor effects, patients whose baseline MMSE < 10 were also excluded.¹⁶

Drug exposures and cognitive outcomes

Data regarding diabetes treatment were obtained from the UDS Medication Form.¹⁴

In this study, the ‘combination antidiabetic therapy’ group comprises patients who exclusively used two specific antidiabetic drugs simultaneously (e.g., metformin and sulfonylureas) throughout the study period. This combination therapy period begins with the initiation of the two-drug regimen and requires a consistent drug use record for this combination. Conversely, the ‘Single therapy’ group includes patients treated with only one oral antidiabetic medication throughout the entire study duration, without any use of combination therapy.

The outcome of interest was cognitive function, assessed by the MMSE¹⁷ (Score range from 0 to 30. Higher score indicates better cognitive performance) due to its sensitivity and specificity with AD. Most participants underwent annual MMSE assessments, with a follow-up duration ranging from 1 to 13 years.

Statistical analysis

For the baseline subject characteristics, continuous variables were reported in mean (standard deviation) and categorical variables were reported in counts (proportions).

Propensity score matching (PSM) was employed to compare antidiabetic medication users and non-users within the cohort of patients with diabetes and dementia. We utilized a 1:1 to 1:4 nearest-neighbor matching, adhering to a 0.15 standardized mean difference (SMD) threshold.¹⁸ The variables included in the PSM were baseline MMSE scores, sex, depression, age, race, apolipoprotein E ε4 status, hypercholesterolemia, hypertension, stroke, vitamin B12 deficiency, cardiovascular disease (defined as the union set of ‘Heart attack/cardiac arrest’, ‘Atrial fibrillation’, ‘Angioplasty/endarterectomy/stent’, ‘Cardiac bypass procedure’, ‘Pacemaker and/or defibrillator’, ‘Pacemaker’, ‘Congestive heart failure’, ‘Angina’, ‘Heart valve replacement or repair’ and ‘Other cardiovascular disease’ in NACC database¹¹), body mass index, AD medications, and the concomitant use of other oral and injectable hypoglycemic medications, including insulin and incretins. In this study, the R ‘matchit’ package¹⁹ was utilized for PSM to ensure balanced comparison groups.

Group comparisons were conducted using t-tests to ensure no significant differences in baseline MMSE between matched comparison groups.

We employed the generalized linear mixed model (GLMM) to analyze the relationship between groups, accommodating for the non-normally distributed outcome and mixed effects in the data. Within the GLMM, the decision to use a Poisson or negative binomial model was determined based on the assessment of overdispersion. The Poisson model was used if there was no overdispersion on the MMSE, while the negative binomial model was used when MMSE overdispersion was present. The outcomes of the GLMM analyses are presented as rate ratios (RR) with 95% confidence intervals (CI), providing a measure of the relative rate of change in the outcome variable. All covariates not balanced in the PSM were included in the GLMM model to adjust for potential confounding effects. GLMM was performed using the R package glmmTMB.²⁰ The data cleaning process was conducted using SAS software (version 9.4 SAS Institute Inc., Cary, NC, USA), and all further analyses were performed in R (version 4.2.0, R Foundation for Statistical Computing, Vienna, Austria).

Data availability

Data for this study was from NACC and are available [https://naccdata.org] with the permission of NACC.

Results

Subject characteristics

In the study period between June 2005 and August 2022, a cohort of 964 patients (2811 visits) with concurrently diagnosis of AD and T2DM was analyzed, as detailed in Supplemental Figure 1 in the supporting information. From the baseline characteristics table (shown as Table 1), the average age of the cohort was 76.02 years, with females constituting 48% (467 patients) of the population. The average body mass index (BMI) was noted at 28.63 kg/m², and the average educational attainment was 13.83 years. At baseline, participants displayed an average MMSE score of 23.93. Among these participants, 69 patients (9%) were carriers of the APOE ε4 allele, and 405 patients (42%) were reported to be using AD medication. Regarding the use of injectable hypoglycemic medications, 153 patients (16%) were on insulin therapy at baseline. Other baseline characteristics of patients in different treatment groups are summarized in Table 1.

Table 1.

Baseline subject characteristics.

	Patients with AD and T2DM (n = 964)
Baseline demographics
Age (y)	76.02 (8.21)
Gender: Female	467 (48%)
Race	639 (66%) White 228 (24%) Black or African American 96 (10%) Others
BMI (kg/m²)	28.63 (5.24)
Education (y)	13.83 (3.97)
AD-related measures at baseline
Baseline MMSE	23.93 (4.78)
APOE ε4 carrier	69 (8%)
AD medication use	405 (42%)
Comorbidities at baseline
Hypercholesterolemia	730 (76%)
Hypertension	730 (76%)
Depression	223 (23%)
Stroke history	55 (6%)
Vitamin B12 deficiency	67 (7%)
Cardiovascular disease	82 (9%)
Injectable hypoglycemic medication at baseline
Insulin	153 (16%)

AD: Alzheimer's disease; APOE: apolipoprotein E; MMSE: Mini-Mental State Examination.

Relationship between oral antidiabetic drug and MMSE score throughout the time

In patients with AD and T2DM, significant differences in MMSE score change over time were observed in specific comparisons between combination therapies. The results from the GLMM comparing different antidiabetic medications are shown in Tables 2 and 3. Table 2 shows the comparison between combination and monotherapy antidiabetic medication, while Table 3 includes all the comparison pairs among combination antidiabetic medication.

Table 2.

Comparisons between combination antidiabetic medication and single antidiabetic medication on annual point change in MMSE scores.

Drug versus drug
	RR (95% CI)	Z	p
Sulfonylureas versus Metformin × time (125 user pairs)	0.979 [0.938, 1.021]	−1.00	0.319
TZD versus Metformin × time (21 user pair)	0.928 [0.792, 1.086]	−0.932	0.351
TZD versus Sulfonylureas × time (21 users versus 42 users)	0.949 [0.848, 1.062]	−0.91	0.362
Metformin + DPP-4i versus Metformin × time (24 users versus 48 users)	1.024 [0.939, 1.116]	0.53	0.596
Metformin versus Metformin + TZD × time (78 users versus 26 users)	1.042 [0.949, 1.145]	0.87	0.385
Sulfonylureas + TZD versus Metformin × time (21 users versus 63 users)	0.979 [0.896, 1.070]	−0.47	0.635
Metformin + sulfonylureas versus Metformin × time (120 user pairs)	0.988 [0.948, 1.031]	−0.54	0.588
Metformin + DPP-4i versus Sulfonylureas × time (26 users versus 52 users)	1.056 [0.952, 1.171]	1.029	0.303
Metformin + TZD versus Sulfonylureas × time (26 user pairs)	0.935 [0.838, 1.043]	−1.21	0.227
Sulfonylureas + TZD versus Sulfonylureas × time (21 users versus 42 users)	1.006 [0.899, 1.126]	0.106	0.916
Metformin + Sulfonylureas versus Sulfonylureas × time (118 user pairs)	1.021 [0.977, 1.068]	0.937	0.349
TZD versus Metformin + TZD × time (21 user pairs)	1.016 [0.885, 1.167]	0.228	0.820
TZD versus Metformin + sulfonylureas (21 users versus 42 users)	1.008 [0.861, 1.181]	0.099	0.921
TZD versus Metformin + DPP-4i × time (21 user pairs)	0.916 [0.804, 1.044]	−1.312	0.190
TZD versus Sulfonylureas + TZD × time (21 user pairs)	0.989 [0.863, 1.133]	−0.158	0.875

RR: rate ratio.

Table 3.

Comparisons within combination antidiabetic medication on annual point change in MMSE scores.

Drug versus drug
	RR (95% CI)	Z	p
Sulfonylureas + TZD versus Metformin + sulfonylureas × time (21 user pairs)	0.999 [0.878, 1.138]	−0.008	0.993
Metformin + DPP-4i versus Metformin + sulfonylureas × time (26 users versus 52 users)	1.045 [0.960, 1.138]	1.02	0.307
Metformin + sulfonylureas versus Metformin + TZD × time (105 users versus 26 users)	1.105 [1.022, 1.195]	2.50	0.013
Sulfonylureas + TZD versus Metformin + TZD × time (21 user pairs)	1.042 [0.930, 1.168]	0.714	0.475
Metformin + DPP-4i versus Metformin + TZD × time (26 users versus 25 users)	1.132 [1.021, 1.256]	2.347	0.019
Sulfonylureas + TZD versus Metformin + DPP-4i × time (21 user pairs)	0.935 [0.842, 1.038]	−1.267	0.205

RR: rate ratio.

In comparisons between monotherapy and combination therapy (shown in Table 2), no significant differences were observed. Specifically, no significant differences were observed between metformin combined with TZD and sulfonylureas monotherapy (RR = 0.935 [0.838, 1.043], p = 0.227). Additionally, metformin combined with sulfonylureas did not differ significantly from sulfonylureas monotherapy (RR = 1.021 [0.977, 1.068], p = 0.349).

In comparisons among combination therapies (shown in Table 3), metformin combined with sulfonylureas was associated with a significantly slower decline in MMSE scores when compared to metformin with TZD (RR = 1.105 [1.022, 1.195], p = 0.013). Similarly, metformin with DPP-4i demonstrated a significant association with slower MMSE decline compared to metformin with TZD (RR = 1.132 [1.021, 1.256], p = 0.019). However, no significant differences were observed between metformin combined with sulfonylureas and sulfonylureas combined with TZD (RR = 0.999 [0.878, 1.138], p = 0.993) or metformin with DPP-4i (RR = 1.045 [0.960, 1.138], p = 0.307). Baseline characteristics before and after PSM for the comparison groups with significant differences are presented in Supplemental Table 3.

To examine the robustness of our findings, we conducted a sensitivity analysis excluding participants who received insulin therapy. The results remained consistent with the primary analysis: both metformin + sulfonylureas versus metformin + TZD and metformin + DPP-4i versus metformin + TZD still showed significant difference on MMSE decline. Groups without significant differences in the primary analysis also remained non-significant (Supplemental Table 4).

The longitudinal trends of MMSE scores for comparison pairs with significant differences are visualized in Figures 1 and 2. Figure 1 illustrates the comparison between metformin + DPP-4i and metformin + TZD, while Figure 2 depicts metformin + sulfonylureas versus metformin + TZD. Supplemental Figures 2 and 3 present additional comparisons involving metformin monotherapy, with Supplemental Figure 2 comparing sulfonylureas versus metformin and Supplemental Figure 3 comparing sulfonylureas + TZD versus metformin.

Figure 1.

Association between ‘Metformin + TZD’ and ‘Metformin + DPP-4i’. Thick lines: total estimated association adjusted for covariates; thin lines: estimated association adjusted for covariates per individual.

Figure 2.

Association between ‘Metformin + TZD’ and ‘Metformin + Sulfonylureas’. Thick lines: total estimated association adjusted for covariates; thin lines: estimated association adjusted for covariates per individual.

Discussion

In this study, we found that different antidiabetic therapies had varying impacts on cognitive decline in patients with AD and T2DM. This finding aligns with the point of view that the two diseases may share overlapping pathophysiological mechanisms, such as insulin resistance and neuroinflammation. Michailidis et al. discussed this link and the concept of AD as “type 3 diabetes”, emphasizing the role of disrupted insulin signaling in cognitive decline.²¹ Notably, metformin combined with DPP-4i or sulfonylureas was associated with a slower decline in MMSE scores compared to metformin combined with TZD.

Given that metformin was involved in all comparisons showing significant differences, its relationship with cognitive function changes has been widely debated in other studies, but no definitive conclusion has been reached.²² In our study, metformin monotherapy showed no significant difference in slowing cognitive decline compared to either metformin combined with sulfonylureas or the combination of sulfonylureas and TZD. These findings are consistent with the study from Newby et al., which also reported no significant association between metformin use and mild cognitive impairment (MCI) compared to sulfonylureas.²³ Similarly, our previous study found that, in T2DM patients with probable AD, there was no association between metformin use and the risk of severe dementia.²⁴ All these observations showed the potential limitations of metformin monotherapy in significantly altering cognitive outcomes in patients with AD and T2DM, which highlights the importance of exploring potential synergies when metformin is combined with other antidiabetic medications. Given that combination antidiabetic therapy is a prevalent approach in managing T2DM, it is important to understand its implications, especially in the context of complex diseases where single-drug interventions may not fully address the broad spectrum of pathological changes.

Among the combination antidiabetic therapy comparisons, only the combination therapy of metformin with sulfonylureas and metformin with DPP-4i demonstrated a significant association with slower cognitive decline. We will first discuss the significant additional impact of combination antidiabetic therapies on cognitive function. Then, the potential negative impact of TZD on cognitive function will be discussed further in a subsequent paragraph.

Our study indicates that the combination of metformin with DPP-4i was associated with a slower decline of MMSE scores compared to the combination of metformin with TZD. Supporting our finding, Juraj and colleagues reported that DPP-4i users experienced a slower MMSE decline compared to non-users in diabetic dementia patients²⁵ This pattern is further reinforced by broader population studies by Chen et al., which reported an association between DPP-4 inhibitors and a reduced risk of dementia.²⁶ Mechanistically, DPP-4i may exert neuroprotective effects by increasing GLP-1 levels and reducing amyloid-beta accumulation.²⁷ GLP-1 elevation also leads to GSK3 downregulation, reversing tau hyper-phosphorylation and mitigating oxidative stress, potentially slowing AD progression.²⁸ Additionally, treatments with saxagliptin and vildagliptin have been reported to reduce levels of TNF-α and IL-1β, indicating a decrease in AD-related brain inflammation.²⁹ Activation of AMPK and GSK-3 pathways by DPP-4i may further enhance antioxidative effects and reduce amyloid-beta production.^30,31 These findings collectively suggest a potential therapeutic benefit of DPP-4i as an add-on drug for metformin in managing cognitive decline in AD and T2DM patients.

Our study revealed that the combination of metformin and sulfonylureas is associated with a slower decline in MMSE scores compared to combinations involving metformin and TZD. Our finding aligns with Zhang et al., who found cognitive benefits of sulfonylureas in T2DM patients compared to other antidiabetic drugs.³² However, conflicting evidence exists; some studies reported a higher dementia risk with sulfonylureas compared to alternative therapies, potentially due to methodological differences, reference groups, or within-class variability among sulfonylureas.^33,34 Our literature review reveals a notable gap in mechanistic in vitro studies exploring the link between sulfonylureas and cognitive function, a gap that may result from the mixed clinical evidence available. Given the conflicting findings in the literature and our relatively small sample size for non-metformin therapies, we caution against overinterpreting our results. Nonetheless, our findings highlight the need for future research that incorporates larger sample size and detailed mechanistic studies to elucidate the role of sulfonylureas in cognitive function.

Given that metformin + TZD served as the reference group in both comparisons that showed significant differences (i.e., with metformin + sulfonylureas and metformin + DPP-4i), our findings indicate that these two combinations were associated with a relatively slower cognitive decline. While this pattern may imply a less favorable association with metformin + TZD, we caution that direct conclusions about the isolated effect of TZD cannot be drawn from this analysis, as it was not assessed independently in our analysis due to the sample size issue. This is consistent with the study from Wu et al. that reported the TZD use was associated with a faster decline in people with AD dementia.¹¹ Similarly, the ACCORD-MIND study found that rosiglitazone, a type of TZD, was linked to greater cognitive decline over 40 months in T2DM patients.³⁵ Although a randomized double-blind trial in 2006 where combined metformin and rosiglitazone therapy appeared to improve working memory.³⁶ This differs from our findings, which focused on overall cognitive function using MMSE, suggesting that the results difference may come from the different cognitive outcomes. Interestingly, Tang et al. reported that metformin with TZD was associated with lower risk of all-cause dementia.³⁷ However, this discrepancy may reflect differences in study populations, as their study included individuals without pre-existing AD, whereas our study focused on patients already diagnosed with AD and T2DM. This distinction is particularly relevant, as the role of TZD in cognitive function may differ depending on the stage of neurodegeneration or the presence of AD pathology. Supporting evidence from animal models further raises concerns about TZD's impact on cognition, as pioglitazone showed no beneficial effects on cognitive function.^38,39 This evidence, illustrating the association between TZD use and cognitive decline, clearly underscores the complexity of TZD's impact on cognitive function in the context of AD and T2DM. Due to the imbalanced covariates after propensity score matching in the single TZD group, we recommend that the findings on these comparison groups be interpreted with caution. Given the mixed results from both clinical and animal model studies, there is a need for further comprehensive research to explore the mechanisms and clinical implications of TZD therapy in AD and T2DM populations.

Beyond the comparison groups with significant differences, comparison groups with insignificant differences also offer important insights. In the comparison between ‘Metformin + DPP-4i versus Metformin + Sulfonylureas’, a consistent trend was observed that aligns with previous research. Secnik et al. reported that sulfonylureas were associated with a larger decline in MMSE compared to DPP-4i, although this difference disappeared in imputed analyses, indicating result instability.²⁵ Similarly, Zullo et al. reported no significant differential impact on cognitive decline between Sulfonylureas and DPP-4i in nursing home residents.⁴⁰ Despite the absence of statistical significance, the hazard ratios for cognitive decline and altered mental status were consistently lower for DPP-4i users compared to sulfonylureas users.⁴⁰ Although our study could not compare the effects of single drugs due to the limited sample size of single DPP-4i users, the relevance of these findings is not diminished. In addition to metformin dual therapies, we analyzed other combination therapies, such as sulfonylureas + TZD. This combination showed no significant differences in cognitive decline when compared to metformin + sulfonylureas, aligning with findings by Kim et al., who reported that compared to metformin + sulfonylureas, sulfonylureas + TZD did not reduce the incidence of AD events.⁴¹ Wu et al. also concluded that sulfonylureas and TZD were not protective against cognitive decline in cognitively normal T2DM patients.⁴² However, a prospective study by Tang et al. suggested that sulfonylureas + TZD was associated with a 15% lower risk of AD incidence when compared to metformin monotherapy.³⁷ These conflicting findings likely reflect differences in study design, databases, and population characteristics. As this combination consists of two second-line treatments, the limited sample size in our study is not unexpected. Although several studies, including ours, have explored the impact of sulfonylureas + TZD on AD progression, more research is needed to better understand its role in AD and T2DM patients.

In our study population, comorbid conditions such as hypertension and hypercholesterolemia were highly prevalent, and many participants were likely to be using antihypertensive or lipid-lowering medications. While these variables were available in the dataset, our primary analysis did not include specific medication use for these comorbidities, as our research scope focused on antidiabetic therapies. However, we did control for the presence of hypertension and hyperlipidemia through propensity score matching to reduce confounding bias. The cognitive effects of these medications remain an area of ongoing research. While some studies suggest that blood pressure variability and intensive control may impact cognition, there is no consistent support for the superiority of specific antihypertensive drug classes in preventing cognitive decline in older adults.^43,44 Similarly, although statins have been linked to short-term changes in cognitive domains through alterations in lipid and inflammatory markers, they appear to have minimal long-term effects on cognition.⁴⁵ Given these uncertainties, future studies may benefit from exploring the potential interactive effects of these medications on cognitive function in patients with AD and T2DM.

Our study has several strengths. Firstly, the NACC database we used includes cognitive test scores, comorbidities, medication history, neuroimaging, and genotypic data, ensuring a comprehensive and representative dataset for studying cognitive decline in AD patients.¹³ The data undergo centralized auditing and quality control, ensuring the reliability and validity of our research results. Then, we performed a comprehensive analysis of both single and combination antidiabetic therapies and their effects on cognitive function in patients with both AD and T2DM. While previous research has primarily focused on single-drug therapies, our study expands this scope by evaluating combination therapies, which are widely used in clinical practice but remain underexplored regarding their cognitive effects. Rigorous methodology is another key strength. When evaluating the drug exposure situation, we used as-treated analysis to accurately capture the drug use of the patients. Additionally, this study used propensity score matching which could minimize confounding bias, ensuring balanced comparisons across treatment groups. We further accounted for APOE genotype, a critical genetic factor influencing cognitive decline, enhancing the specificity of our analysis. Due to the limited prevalence of APOE ε4 carriers, subgroup analyses were not conducted. Future studies with larger and more diverse populations may explore whether APOE genotype modifies the effect of antidiabetic therapies on cognitive decline. Finally, GLMM allows us to effectively handle the non-normally distributed outcomes and manage several time-varying covariates within our repeated measure design. Each of these methodological choices contributes significantly to the reliability of our findings.

One limitation of this study is that the NACC database lacks some variables that might introduce potential bias, such as HbA1c, socioeconomic status, renal function and diabetes severity. Furthermore, as the inherent limitation of an observational dataset, we have no access to the drug exposure situation and disease duration before entry into the database and the adherence of drug use in each visit, which may introduce bias to the outcome and limit our ability to assess disease chronology. Third, the sample size for certain antidiabetic therapy groups, such as single TZD (n = 21), single DPP-4i (n = 13), and other therapies, was relatively small, which limits the efficacy of propensity score matching and reduces the generalizability of the findings from these groups. This limitation arises from our strict inclusion criteria and rigorous definitions of monotherapy and combination therapy, which were applied to ensure data quality and minimize potential confounding bias. While the smaller sample size limits the ability to draw solid conclusions for these therapy groups, our analysis benefits from the use of PSM and GLMM, which strengthen the validity of the results despite small sample size. It is also worth noting that our sample sizes are comparable to those in similar studies within the field,^25,46 reflecting the unique challenges of studying this specific population with stringent therapeutic definitions. Nonetheless, further research with larger sample sizes is still needed to confirm the trends and provide a deeper understanding of the cognitive impact of therapies like single TZD and TZD-related combination therapy. Finally, although MMSE is a widely used and validated tool, it has limitations in assessing certain cognitive domains such as executive function and visuospatial abilities. Future research may benefit from including a broader neuropsychological battery to more comprehensively evaluate cognitive function.

Although our study focused on cognitive outcomes, it is important to acknowledge that antidiabetic therapies may pose a risk of hypoglycemia in older adults with AD and T2DM.⁴⁷ Future studies are needed to evaluate both the cognitive benefits and potential adverse effects to support balanced clinical decision-making.

Conclusions

This research provides comprehensive observational evidence of both single and combination antidiabetic therapies in patients with AD and T2DM. Our findings demonstrate that, compared to the metformin and TZD combination therapy, metformin combined with DPP-4i, metformin combined with sulfonylureas were associated with a slower decline in MMSE scores. These results suggest that DPP-4i and sulfonylureas may be favorable add-on options to metformin for preserving cognitive function in this population. Clinicians managing AD patients with T2DM might consider these therapies when tailoring antidiabetic regimens, particularly in light of the potential negative impact of TZD on cognitive function. Further large-scale, longitudinal studies are needed to validate these findings and explore the underlying mechanisms.

Supplemental Material

sj-docx-1-alz-10.1177_13872877251365662 - Supplemental material for Impact of antidiabetic therapy on cognitive function in patients with type 2 diabetes and Alzheimer's disease: A comprehensive analysis

Supplemental material, sj-docx-1-alz-10.1177_13872877251365662 for Impact of antidiabetic therapy on cognitive function in patients with type 2 diabetes and Alzheimer's disease: A comprehensive analysis by Chen Jiang, Junnan Qi, Hongyi Zou, Oscar Lopez, Xiang-Qun Xie and Ying Xue in Journal of Alzheimer's Disease

Footnotes

Acknowledgments

We are grateful for the funding from the Competitive Medical Research Fund of the UPMC Health System, the University of Pittsburgh, and the National Institute on Aging of the National Institutes of Health under award number R56AG074951 and National Institute on Drug Abuse (NIDA) P30 PDA035778A. The data used in our study were from the NACC database. The NACC database is funded by NIA/NIH Grant U24 AG072122. NACC data are contributed by the NIA-funded ADRCs: P30 AG062429 (James Brewer), P30 AG066468 (Oscar Lopez), P30 AG062421 (Bradley Hyman), P30 AG066509 (Thomas Grabowski), P30 AG066514 (Mary Sano), P30 AG066530 (Helena Chui), P30 AG066507 (Marilyn Albert), P30 AG066444 (John Morris), P30 AG066518 (Jeffrey Kaye), P30 AG066512 (Thomas Wisniewski), P30 AG066462 (Scott Small), P30 AG072979 (David Wolk), P30 AG072972 (Charles DeCarli), P30 AG072976 (Andrew Saykin), P30 AG072975 (David Bennett), P30 AG072978 (Neil Kowall), P30 AG072977 (Robert Vassar), P30 AG066519 (Frank LaFerla), P30 AG062677 (Ronald Petersen), P30 AG079280 (Eric Reiman), P30 AG062422 (Gil Rabinovici), P30 AG066511 (Allan Levey), P30 AG072946 (Linda Van Eldik), P30 AG062715 (Sanjay Asthana), P30 AG072973 (Russell Swerdlow), P30 AG066506 (Todd Golde), P30 AG066508 (Stephen Strittmatter), P30 AG066515 (Victor Henderson), P30 AG072947 (Suzanne Craft), P30 AG072931 (Henry Paulson), P30 AG066546 (Sudha Seshadri), P20.

ORCID iDs

Chen Jiang

Junnan Qi

Hongyi Zou

Xiang-Qun Xie

Ethical considerations

This study was conducted using de-identified data from the National Alzheimer's Coordinating Center (NACC), which collects and maintains data in accordance with the ethical principles outlined in the Declaration of Helsinki and approved institutional review board (IRB) protocols at each participating Alzheimer's Disease Research Center (ADRC). The use of NACC data in this study was deemed exempt from additional IRB approval at our institution as it involves only secondary analysis of de-identified data.

Author contributions

Chen Jiang: Formal analysis; Visualization; Writing – original draft.

Junnan Qi: Formal analysis; Writing – review & editing.

Hongyi Zou: Formal analysis; Writing – review & editing.

Oscar Lopez: Conceptualization; Writing – review & editing.

Xiangqun Xie: Conceptualization; Funding acquisition; Writing – review & editing.

Ying Xue: Conceptualization; Data curation; Funding acquisition; Project administration; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Competitive Medical Research Fund of the UPMC Health System, the University of Pittsburgh, the National Institute on Aging of the National Institutes of Health under award number (grant number: R56AG074951), and National Institute on Drug Abuse (NIDA) P30 PDA035778A.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The data used in this study were obtained from the National Alzheimer's Coordinating Center (NACC). Access to NACC data is available upon request through an approved data use agreement. Further details on data access and application procedures can be found at .

Supplemental material

Supplemental material for this article is available online.

References

Blennow

de Leon

Zetterberg

. Alzheimer's disease. Lancet 2006; 368: 387–403.

GBD 2019 Dementia Forecasting Collaborators. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the global burden of disease study 2019. Lancet Public Health 2022; 7: e105–e125.

GBD 2021 Diabetes Collaborators. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the global burden of disease study 2021. Lancet 2023; 402: 203–234.

Janson

Laedtke

Parisi

, et al. Increased risk of type 2 diabetes in Alzheimer disease. Diabetes 2004; 53: 474–481.

Muñoz-Jiménez

Zaarkti

García-Arnés

, et al. Antidiabetic drugs in Alzheimer's disease and mild cognitive impairment: a systematic review. Dement Geriatr Cogn Disord 2020; 49: 423–434.

Tan

, et al. Evidence-based prevention of Alzheimer's disease: systematic review and meta-analysis of 243 observational prospective studies and 153 randomised controlled trials. J Neurol Neurosurg Psychiatry 2020; 91: 1201–1209.

Sluggett

Koponen

Bell

, et al. Metformin and risk of Alzheimer's disease among community-dwelling people with diabetes: a national case-control study. J Clin Endocrinol Metab 2020; 105: dgz234.

Kim

Jeon

Kim

, et al. Risk of dementia in older patients with type 2 diabetes on dipeptidyl-peptidase IV inhibitors versus sulfonylureas: a real-world population-based cohort study. J Clin Med 2018; 8: 28.

Hsu

Wahlqvist

Lee

, et al. Incidence of dementia is increased in type 2 diabetes and reduced by the use of sulfonylureas and metformin. J Alzheimers Dis 2011; 24: 485–493.

10.

Gold

Alderton

Zvartau-Hind

, et al. Rosiglitazone monotherapy in mild-to-moderate Alzheimer's disease: results from a randomized, double-blind, placebo-controlled phase III study. Dement Geriatr Cogn Disord 2010; 30: 131–146.

11.

Ouk

Wong

, et al. Relationships between memory decline and the use of metformin or DPP4 inhibitors in people with type 2 diabetes with normal cognition or Alzheimer's disease, and the role APOE carrier status. Alzheimers Dement 2020; 16: 1663–1673.

12.

Besser

Kukull

Teylan

, et al. The revised national Alzheimer's coordinating center's neuropathology form-available data and new analyses. J Neuropathol Exp Neurol 2018; 77: 717–726.

13.

Besser

Kukull

Knopman

, et al. Version 3 of the national Alzheimer's coordinating center's uniform data set. Alzheimer Dis Assoc Disord 2018; 32: 351–358.

14.

Morris

Weintraub

Chui

, et al. The uniform data set (UDS): clinical and cognitive variables and descriptive data from Alzheimer disease centers. Alzheimer Dis Assoc Disord 2006; 20: 210–216.

15.

McKhann

Knopman

Chertkow

, et al. The diagnosis of dementia due to Alzheimer's disease: recommendations from the national institute on aging-Alzheimer's association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 2011; 7: 263–269.

16.

Ravona-Springer

Luo

Schmeidler

, et al. The association of age with rate of cognitive decline in elderly individuals residing in supporting care facilities. Alzheimer Dis Assoc Disord 2011; 25: 312–316.

17.

Folstein

McHugh

. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189–198.

18.

Stuart

Lee

Leacy

. Prognostic score-based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research. J Clin Epidemiol 2013; 66: S84–S90.e81.

19.

Imai

King

, et al. Matchit: nonparametric preprocessing for parametric causal inference. J Stat Softw 2011; 42: 1–28.

20.

Brooks

Kristensen

van Benthem

, et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J 2017; 9: 378–400.

21.

Michailidis

Moraitou

Tata

, et al. Alzheimer's disease as type 3 diabetes: common pathophysiological mechanisms between Alzheimer's disease and type 2 diabetes. Int J Mol Sci 2022; 23: 2687.

22.

Luo

Ning

, et al. Association between metformin and Alzheimer's disease: a systematic review and meta-analysis of clinical observational studies. J Alzheimers Dis 2022; 88: 1311–1323.

23.

Newby

Linden

Fernandes

, et al. Comparative effect of metformin versus sulfonylureas with dementia and Parkinson's disease risk in US patients over 50 with type 2 diabetes mellitus. BMJ Open Diabetes Res Care 2022; 10: e003036.

24.

Xue

Xie

. The association between metformin use and risk of developing severe dementia among AD patients with type 2 diabetes. Biomedicines 2023; 11: 2935.

25.

Secnik

Schwertner

, et al. The association of antidiabetic medications and Mini-mental state examination scores in patients with diabetes and dementia. Alzheimers Res Ther 2021; 13: 197.

26.

Chen

Chung

, et al. Association between the use of dipeptidyl peptidase 4 inhibitors and the risk of dementia among patients with type 2 diabetes in Taiwan. J Clin Med 2020; 9: 660.

27.

Perry

Lahiri

Sambamurti

, et al. Glucagon-like peptide-1 decreases endogenous amyloid-beta peptide (Abeta) levels and protects hippocampal neurons from death induced by Abeta and iron. J Neurosci Res 2003; 72: 603–612.

28.

Chen

, et al. Glucagon-like peptide-1 regulates mitochondrial biogenesis and tau phosphorylation against advanced glycation end product-induced neuronal insult: studies in vivo and in vitro. Neuroscience 2015; 300: 75–84.

29.

Kosaraju

Gali

Khatwal

, et al. Saxagliptin: a dipeptidyl peptidase-4 inhibitor ameliorates streptozotocin induced Alzheimer's disease. Neuropharmacology 2013; 72: 291–300.

30.

Kornelius

Lin

Chang

, et al. DPP-4 inhibitor linagliptin attenuates Aβ-induced cytotoxicity through activation of AMPK in neuronal cells. CNS Neurosci Ther 2015; 21: 549–557.

31.

Pearce

Arch

Clapham

, et al. Development of glucose intolerance in male transgenic mice overexpressing human glycogen synthase kinase-3beta on a muscle-specific promoter. Metabolism 2004; 53: 1322–1330.

32.

Zhang

Liu

, et al. Metformin therapy and cognitive dysfunction in patients with type 2 diabetes: a meta-analysis and systematic review. Medicine (Baltimore) 2020; 99: e19378.

33.

Iskander

Wang

, et al. Association of sulfonylureas with the risk of dementia: a population-based cohort study. J Am Geriatr Soc 2023; 71: 3059–3070.

34.

Orkaby

Cho

Cormack

, et al. Metformin vs sulfonylurea use and risk of dementia in US veterans aged ≥65 years with diabetes. Neurology 2017; 89: 1877–1885.

35.

Seaquist

Miller

Fonseca

, et al. Effect of thiazolidinediones and insulin on cognitive outcomes in ACCORD-MIND. J Diabetes Complications 2013; 27: 485–491.

36.

Ryan

Freed

Rood

, et al. Improving metabolic control leads to better working memory in adults with type 2 diabetes. Diabetes Care 2006; 29: 345–351.

37.

Tang

Brinton

Chen

, et al. Use of oral diabetes medications and the risk of incident dementia in US veterans aged ≥60 years with type 2 diabetes. BMJ Open Diabetes Res Care 2022; 10: e002894.

38.

Galea

Feinstein

Lacombe

. Pioglitazone does not increase cerebral glucose utilisation in a murine model of Alzheimer's disease and decreases it in wild-type mice. Diabetologia 2006; 49: 2153–2161.

39.

Masciopinto

Di Pietro

Corona

, et al. Effects of long-term treatment with pioglitazone on cognition and glucose metabolism of PS1-KI, 3xTg-AD, and wild-type mice. Cell Death Dis 2012; 3: e448.

40.

Zullo

Duprey

Smith

, et al. Effects of dipeptidyl peptidase-4 inhibitors and sulphonylureas on cognitive and physical function in nursing home residents. Diabetes Obes Metab 2022; 24: 247–256.

41.

Kim

Noh

Han

, et al. The association between second-line oral antihyperglycemic medication on types of dementia in type 2 diabetes: a nationwide real-world longitudinal study. J Alzheimers Dis 2021; 81: 1263–1272.

42.

C-Y

Shapiro

Ouk

, et al. Glucose-lowering drugs, cognition, and dementia: the clinical evidence. Neurosci Biobehav Rev 2022; 137: 104654.

43.

Yang

Luo

, et al. Effects of antihypertensive drugs on cognitive function in elderly patients with hypertension: a review. Aging Dis 2021; 12: 841–851.

44.

Wang

, et al. Synergism of antihypertensives and cholinesterase inhibitors in Alzheimer's disease. Alzheimers Dement (N Y) 2018; 4: 542–555.

45.

Gentreau

Rukh

Miguet

, et al. The effects of statins on cognitive performance are mediated by low-density lipoprotein, C-reactive protein, and blood glucose concentrations. J Gerontol A Biol Sci Med Sci 2023; 78: 1964–1972.

46.

Samaras

Makkar

Crawford

, et al. Metformin use is associated with slowed cognitive decline and reduced incident dementia in older adults with type 2 diabetes: the Sydney memory and ageing study. Diabetes Care 2020; 43: 2691–2701.

47.

Sheen

Sheu

. Association between hypoglycemia and dementia in patients with type 2 diabetes. Diabetes Res Clin Pract 2016; 116: 279–287.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.26 MB