Associations between antidiabetic medications and cerebrospinal fluid biomarkers of Alzheimer's disease

Abstract

Background

It has been hypothesized that insulin resistance is pivotal in mediating amyloid and tau dysregulations in Alzheimer's disease (AD).

Objective

To investigate the impact of different antidiabetic agents, their daily dosage intake, and treatment duration on cerebrospinal fluid (CSF) AD biomarkers among patients with type 2 diabetes.

Methods

This cross-sectional study selected patients between 50 and 80 years with diabetes and CSF AD biomarkers screened between 2017 and 2023 in the VALCODIS Cohort. CSF biomarkers were total tau (t-tau), phosphorylated tau 181 (p-tau), and amyloid-β 42 (Aβ₄₂). Analytical variables were obtained. Antidiabetic prescriptions were recorded in defined daily doses (DDD), according to the ATC/DDD 2021 system, and years of drug exposure duration before lumbar puncture. Logistic regressions were performed to establish the correlations between drug usage and AD biomarker alteration.

Results

Among patients with diabetes, Insulin consumption was associated with lower odds of abnormal Aβ₄₂ levels (OR 0.36 [95% CI 0.15, 0.76]) and tau pathology (OR 0.49 [95% CI 0.24–0.98]). Metformin was related to lower odds of pathological p-tau when diabetes was uncontrolled, acting on t-tau and t-tau/Aβ₄₂ ratio when it was concomitant with insulin, and patients had controlled diabetes. Lower odds of pathological levels of tau were observed when additional oral antidiabetic drugs were added among metformin users. iSGLT2 was associated with tau pathology.

Conclusions

The impact of antidiabetics on AD-related pathological biomarkers may depend on diabetes management.

Keywords

Alzheimer's disease antidiabetic agents diabetes control insulin metformin type 3 diabetes mellitus

Introduction

Dementia is a complex neurological condition that poses significant challenges in clinical practice and research. It is characterized by progressive cognitive decline and diminished ability to perform everyday activities due to accumulated neuropathological changes. Alzheimer's disease (AD) is a leading cause of dementia, often coexisting with cerebrovascular pathology.^1,2 Neuropathologically, specific hallmarks of AD include amyloid-β (Aβ) senile plaques and tau neurofibrillary tangles.^3,4

Type 2 diabetes (hereafter, diabetes), an age-related chronic metabolic disorder characterized by hyperglycemia, insulin resistance, or a relative lack of insulin, is a modifiable risk factor for dementia and AD.⁵ Previous research has demonstrated that pathological biological processes not necessarily unique to AD, such as inflammation and impaired insulin signaling,^2,5,6 contribute to AD pathology.⁵ Under normal conditions, insulin and insulin-like growth factor 1 (IGF-1) regulate the phosphoinositide 3-kinase/protein kinase B (PI3 K/Akt) pathway,^2,6 which modulates the insulin-degrading enzyme (IDE)—responsible for Aβ clearance—and inhibits glycogen synthase kinase-3 (GSK3β) activation, reducing tau phosphorylation.^2,6 As a result, disruptions in brain insulin levels lead to Aβ dysregulation and accumulation of phosphorylated tau (p-tau),^2,3,6 resulting in higher levels of tau pathology in the cerebrospinal fluid (CSF) in older adults with untreated diabetes than those with normoglycemia.⁷

It is plausible that antidiabetic therapies aiming to restore impaired brain insulin function may offer neurocognitive advantages to individuals at risk of AD.⁸ Antidiabetic agents can be categorized based on their mechanisms of action into insulins, drugs stimulating insulin release, insulin sensitizers, and others, such as sodium-glucose co-transporter-2 (SGLT2) inhibitors and glucosidase inhibitors. Insulins elevate insulin levels to facilitate glucose uptake into cells.⁹ In addition, medications that stimulate insulin release can be further divided into three categories. First, insulin secretagogues, encompassing sulfonylureas and repaglinide, act on pancreatic β-cell potassium channels to trigger insulin release. Second, drugs that act on incretins, specifically glucagon-like peptide-1 (GLP-1) and the dipeptidyl peptidase 4 (DPP-4) inhibitors. GLP-1 enhances insulin secretion in response to food intake and is inhibited by DPP-4, which contributes to insulin delivery by stopping GLP-1 degradation.^9–11 Third, insulin sensitizers, such as metformin and pioglitazone, enhance insulin sensitivity in target tissues and improve lipid metabolism.^9,10 However, previous studies show discrepancies in human AD pathology and antidiabetic intake, highlighting the need to understand the diabetes-AD relationship better.

In this study, we examined the associations of different antidiabetic drugs with CSF biomarker of AD, and the role of diabetes control in such associations within a Spanish memory clinic cohort, shedding light on the interplay between diabetes management and AD pathology. By elucidating these relationships, our research aims to provide valuable insights into the mechanisms underlying dementia and AD, contributing to the broader understanding of neurological disorders within the clinical neuroscience community.

Methods

Study design and participants

This cross-sectional study is based on the VALCODIS cohort, created with patients referred to the Cognitive Disorders Unit of the Hospital Universitari i Politècnic de València (Spain) due to subjective memory complaints, and aiming to contribute to the AD diagnosis and prognosis.¹² From January 2017 to December 2023, 1249 patients aged between 40 and 80 years who attended their first visit to the unit, and signed the informed consent were screened. Participants aged less than 50 years (n = 19), with unavailable CSF sampling (n = 325), or non-type 2 diabetes patients (n = 692) were excluded. As a result, 213 patients were considered in the current study (Figure 1).

Figure 1.

Flow chart diagram for study participants selection.

The study was conducted in accordance with the Declaration of Helsinki and approved by the Medicaments Research Ethics Committee at the Health Research Institute Hospital La Fe (2016/0257 and 202-705-1). Informed consent was obtained from all subjects or legal representatives involved in the study.

This study was approved by the Medicaments Research Ethics Committee at the Health Research Institute La Fe (2016/0257 and 2020-705-1), in compliance with the local regulations.

Data collection and clinical assessments

In line with the standard assessment procedure followed at the Hospital Universitari i Politècnic de València, all participants underwent a series of routine clinical examinations. These included visits with doctors, nurses, and neuropsychologists for extensive medical, physical, and cognitive assessments, CSF and blood samplings. Clinical and lifestyle information was collected retrospectively via hospital medical records between July 2021 and July 2024. Vital signs and biochemistry parameters selected were within six months before or after the lumbar puncture. In cases where multiple assessments occurred within this period, the mean value was used.

Education was dichotomized as low (no studies or primary school) versus high (high school or college). Smoking status was classified as never, former, or current smoking. Patients were classified as having hypertension if their blood pressure was ≥140/90 mmHg or if they were treated with antihypertensive drugs (Anatomical Therapeutic Chemical (ATC) code: C02-C09). Dyslipidemia was determined through medical diagnosis, lipid modifying agents (ATC code: C10) prescription, or total cholesterol levels >220 mg/dl. Antiplatelet drugs included ATC starting with B01AC. Finally, neuropsychologists assessed global cognitive functioning with the Mini-Mental State Examination (MMSE).¹³ The latest MMSE scores within a two-month window of the lumbar puncture procedure were selected (n = 199). A MMSE score ≤24 conventionally suggests the presence of cognitive impairment.¹⁴

Assessment of ad-related biomarkers

CSF samples were collected through lumbar puncture following the clinical routine guidelines in the Hospital Universitari i Politècnic de València, Spain. The determinations of CSF Aβ₄₂, t-tau, and p-tau were assessed by chemiluminescence immunoassay (Lumipulse^® G, Fujirebio, Japan) using a fully automated system.¹⁵

The established cut-offs for CSF biomarkers indicating pathological levels of specific and non-specific pathology in AD were ≤725 pg/ml for Aβ₄₂, > 485 pg/ml for t-tau, ≥ 56 pg/ml for p-tau, and >0.51 for t-tau/Aβ₄₂ ratio.¹⁶

Type 2 diabetes and use of antidiabetic agents

Patients were classified as people with diabetes if they had a medical history, were prescribed antidiabetic drugs or had glycated hemoglobin (HbA1c%) > 6.4%, following the American Diabetes Association guidelines.¹⁷ Uncontrolled diabetes was identified with HbA1c was ≥7.5%.

Antidiabetics agents

Antidiabetic exposure assessment included prescriptions based on their ATC codes (A10). They were regrouped according to their pharmacological mechanisms: insulins (ATC code A10A); insulin sensitizers such as metformin (A10BA02) and pioglitazone (A10BG03); SGLT-2 inhibitors (A10BK); and drugs that stimulate insulin release including insulin secretagogues (sulfonylureas [A10BB] and repaglinide [A10BX02]) and drugs acting on incretins, such as GLP-1 analogs (A10BJ) and DPP-4 inhibitors (A10BH). The defined daily doses (DDDs) for each drug were aggregated within their respective groups (insulins, drugs that stimulate insulin release, insulin sensitizers, or SGLT2 inhibitors).

The following parameters were considered: medication usage (user/non-user), initial prescription date, and prescribed daily dose. Prescribed daily doses were considered as DDD of 2021 since the actual adherence was unavailable. Subsequently, DDDs were determined using the ATC/DDD 2021 guidelines.¹⁸ The duration of drug exposure (drug exposure time) was calculated by subtracting the lumbar puncture date from the drug start date, expressed in years. The most prolonged duration of drug exposure was selected for analysis. Additionally, and given the non-normal distribution of the total cumulative years of antidiabetic use up to the date of the lumbar puncture, the variable “time of drug exposure (years)” was stratified into quartiles for each therapeutic group. Thus, the following categories were created: Non-users, up to 4 years of drug exposure, 4 to 8 years of exposure, and eight or more years of exposure.

Statistical analysis

Descriptive analyses compared patients with CSF AD pathological values with unimpaired values using a Chi-squared test for categorical variables and a t-test for continuous variables. AD-related CSF biomarkers were analyzed dichotomously (normal versus impaired, normal as reference level). Missing data were not considered for the analysis.

Logistic regression models were used to estimate the odds ratios (ORs) with 95% confidence intervals (95% CIs) of the consumption of each antidiabetic agent drug group with impaired AD-related CSF biomarkers. Specifically, we investigated the associations of individual AD-related biomarkers with (1) antidiabetic medication usage (users versus non-users) [Models 1 and 4; Tables 2, 4, and Supplemental Table 1]; (2) DDDs consumption as continuous exposure [Models 2 and 5; Tables 2, 4, and Supplemental Table 1]; and, (3) drug exposure time (in years) as a continuous variable [Models 3 and 6; Tables 2, 4, and Supplemental Table 1], for each of the four groups of antidiabetic agents (insulins, drugs that stimulate insulin release, insulin sensitizers, or SGLT2 inhibitors). DDDs and time of drug exposure (years) were examined in separate models due to their strong correlation (correlation between insulins DDD and time = 0.98; p > 0.001; the correlation between metformin DDDs and time = 0.72; p > 0.001).

Additional logistic regressions were used to analyze whether the biomarkers’ alteration presence was associated with the years of drug exposure before the biomarker's lumbar puncture (Table 3). Moreover, we analyzed the categorized variable of time of years exposure through the Kruskal-Wallis rank sum test corrected by the Bonferroni method.

All models were adjusted for age and sex (female sex as reference). Additionally, specific models were also adjusted for educational level (high academic level as reference) and cardiovascular comorbidities, such as hypertension, dyslipidemia, antiplatelet drug intake, and smoking habits. Reference levels were: not having those comorbidities, not taking antiplatelet drugs, and not smoking. The affected models were the following: Models 4–6 from Table 2, Models 2 and 4 from Table 5, Supplemental Table 1, and Model 2 from Supplemental Table 2 and Supplemental Table 3.

Since metformin is the most frequently used insulin sensitizer, and all patients in the insulin sensitizers group were taking it, we performed a stratified analysis by selecting just metformin consumers. Then, we assessed the odds of having altered biomarkers when consuming concomitant metformin and insulin (Supplemental Table 1), and the odds of having altered biomarkers according to metformin combination (Supplemental Table 2).

Lastly, we assessed the influence of diabetes control in a stratified analysis by considering only patients with HbA1c% records. As a result, we tested the association between antidiabetics among patients with controlled diabetes and among patients with uncontrolled diabetes (Tables 4, 5 and Supplemental Table 3).

All tests were two-tailed, and a p < 0.05 indicated statistical significance. Since this study is exploratory rather than confirmatory, it aims to identify new working hypotheses regarding the relationship between antidiabetic agents, diabetes control, and dementia. Therefore, we did not adjust the p-values for multiple comparisons due to the descriptive nature of this study, following existing recommendations.^19,20 Data analyses were performed using R (version 4.3.0).²¹

We used the G*Power program²² to determine the minimum sample size for the study since our hypothesis was based on comparing the proportion of patients with pathological CSF AD biomarker levels according to their antidiabetic medication.

The calculation was based on a Chi-square test to compare contingency tables, assuming a medium effect size and setting a significance level (Alpha) of 0.05. These parameters required at least 88 participants to achieve a minimum power of 0.8. Ultimately, the study included 213 participants and a post hoc calculation under the same conditions showed a power of 0.99.

Results

Characteristics of the study participants

Table 1 displays sociodemographic and clinical characteristics of the study participants. Of the 213 patients, 61 (31.5%) had amyloid alteration, 80 (37.6%) had impaired t-tau levels, 106 (49.8%) had pathological p-tau levels, and 88 (41.3%) had an abnormal t-tau/Aβ₄₂ ratio. APOE ε4 carriers were likelier to have altered Aβ₄₂ levels and t-tau/Aβ₄₂ ratio (49.1 and 39.1%).

Table 1.

Description of patients with diabetes, according to their cerebrospinal fluid Alzheimer's disease biomarkers*.

			Impaired Aβ₄₂ (n = 67)	Regular Aβ₄₂ (n = 146)	p	Impaired t-tau (n = 80)	Regular t-tau (n = 133)	p	Impaired p-tau (n = 106)	Regular p-tau (n = 106)	p	Impaired t-tau/Aβ₄₂ (n = 88)	Regular t-tau/Aβ₄₂ (n = 125)	p
Age			70.13 ± 4.89	68.35 ± 6.05	0.02	71.12 ± 4.91	67.58 ± 5.85	<0.05	70.30 ± 4.37	65.45 ± 6.12	<0.05	71.42 ± 4.54	67.14 ± 5.89	<0.05
Sex (Female)			36 (53.7%)	61 (41.8%)	0.1	39 (48.8%)	58 (43.6%)	0.5	53 (50.0%)	43 (40.6%)	0.2	48 (54.5%)	49 (39.2%)	0.02
Educational level
	Low (no/primary school)		43 (66.2%)	83 (61.0%)	0.5	57 (73.1%)	69 (56.1%)	0.01	71 (68.3%)	54 (56.3%)	0.2	65 (76.5%)	61 (52.5%)	<0.01
	High (high school/college)		22 (33.8%)	53 (39.0%)	0.5	21 (26.9%)	54 (43.9%)	0.01	33 (31.7%)	42 (43.7%)	0.2	20 (23.5%)	55 (47.4%)	<0.01
Clinical features
Number of concomitant medications			8.8 ± 3.7	9.3 ± 4.2	0.4	8.9 ± 4.0	9.3 ± 4.1	0.5	8.63 ± 9.9	9.7 ± 4.2	0.05	8.6 ± 3.6	9.5 ± 4.4	0.1
MMSE, score			22.9 ± 4.6	24.7 ± 4.2	0.01	22.9 ± 4.6	24.9 ± 4.1	<0.01	23.3 ± 4.7	25.0 ± 3.9	<0.01	21.3 ± 4.5	25.4 ± 3.9	<0.01
	Cognitive impairment (MMSE <24)		35 (55.6%)	53 (39.0%)	0.03	45 (58.4%)	43 (35.3%)	<0.01	54 (52.9%)	33 (34.4%)	0.02	54 (64.3%)	34 (29.6%)	<0.01
APOEe4 Carriers			28 (49.1%)	18 (16.2%)	<0.01	23 (32.9%)	23 (23.5%)	0.2	28 (33.7%)	18 (21.2%)	0.07	28 (39.4%)	18 (18.6%)	<0.01
CSF AD Biomarkers
	Aβ₄₂ (pg/dl)		553.9 ± 113.2	1138.5 ± 418.8	<0.01	850.5 ± 466.5	1017.3 ± 420.9	<0.01	870.7 ± 464.9	1044.2 ± 406.3	<0.01	1164.0 ± 455.7	657.2 ± 184.1	<0.01
	Pathologic Aβ₄₂ (≤ 725 pg/ml)		-	-	-	40 (50%)	27 (20.3%)	<0.01	47 (44.3%)	19 (17.9%)	<0.01	54 (61.4%)	13 (10.4%)	<0.01
	t-tau (pg/dl)		574.5 ± 308.5	408.9 ± 272.4	<0.01	761.9 ± 264.2	279.9 ± 94.7	<0.01	666.6 ± 273.0	253.6 ± 115.5	<0.01	710.9 ± 282.6	285.0 ± 126.6	<0.01
	Pathologic t-tau (> 485 pg/ml)		40 (59.7%)	40 (27.4%)	<0.01	-	-	-	74 (69.8%)	5 (4.7%)	<0.01	67 (76.1%)	13 (10.4%)	<0.01
	p-tau (pg/dl)		95 ± 54.9	61.9 ± 44.6	<0.01	119.5 ± 51.8	43.6 ± 16.7	<0.01	108 ± 48.3	35.8 ± 10.1	<0.01	114.4 ± 51.8	42.4 ± 16.6	<0.01
	Pathologic p-tau (≥56 pg/ml)		47 (71.2%)	59 (40.4%)	<0.01	74 (93.7%)	32 (24.1%)	<0.01	-	-	-	70 (90.8%)	27 (21.6%)	<0.01
	Ratio t-tau/Aβ₄₂		1.1 ± 0.7	0.4 ± 0.3	<0.01	1.1 ± 0.7	0.3 ± 0.2	<0.01	0.9 ± 0.6	0.3 ± 0.1	<0.01	1.1 ± 0.6	0.3 ± 0.1	<0.01
	Pathologic t-tau/Aβ₄₂ (> 0.51)		54 (80.6%)	34 (23.3%)	<0.01	67 (83.8%)	21 (15.8%)	<0.01	79 (74.5%)	8 (7.5%)	<0.01	-	-	-
Cardiovascular risk factors
Hypertension			46 (68.7%)	102 (69.9%)	0.8	55 (68.8%)	93 (69.9%)	0.8	71 (67.0%)	76 (71.7%)	0.4	60 (68.2%)	88 (70.4%)	0.7
	Systolic Blood Pressure (mmHg)		137.2 ± 15.4	133.9 ± 14.1	0.2	135.9 ± 14.7	134.4 ± 14.5	0.6	135.0 ± 13.5	135.1 ± 16.0	1.0	135.8 ± 14.4	134.5 ± 14.8	0.6
	Diastolic Blood Pressure (mmHg)		73.6 ± 8.3	76.9 ± 8.5	0.04	73.4 ± 6.4	77.5 ± 9.5	<0.01	73.6 ± 6.6	78.8 ± 9.9	<0.01	73.6 ± 7.0	77.5 ± 9.3	<0.01
Dyslipidemia			52 (77.6%)	117 (80.1%)	0.7	60 (75.0%)	109 (82.0%)	0.2	84 (79.2%)	84 (79.2%)	1	69 (78.4%)	100 (80%)	0.8
	Total cholesterol (mg/dl)		166.4 ± 33.2	162.1 ± 38.9	0.5	159.3 ± 35.6	166.1 ± 38.0	0.2	159.8 ± 32.1	167.2 ± 41.5	0.2	163.0 ± 31.7	163.8 ± 40.6	0.9
Antiplatelet drugs intake			20 (29,9)	48 (32.9)	0.7	29 (36.2)	39 (29.3)	0.3	33 (31.1)	34 (32.1)	0.9	28 (31.8)	40 (32.0)	0.9
Smoking habit	Yes		12 (19.4%)	29 (23.8%)	0.7	11 (15.1%)	30 (27.0%)	<0.01	16 (17.6%)	25 (27.2%)	0.2	11 (13.9%)	30 (28.6%)	0.01
	No		39 (62.9%)	75 (61.5%)		42 (57.5%)	72 (64.9%)		56 (61.5%)	57 (62.0%)		50 (63.3%)	64 (61.0%)
	Former		11 (17.7%)	18 (14.7%)		20 (27.4%)	9 (8.1%)		19 (20.9%)	10 (10.8%)		18 (22.8%)	11 (10.5%)
Diabetes-related features
Glucose (mg/dl)			117.1 ± 25.4	131.7 ± 45.3	<0.01	129.5 ± 38.8	125.3 ± 41.5	0.5	128.1 ± 39.6	125.7 ± 41.5	0.7	121.9 ± 28.4	130.4 ± 46.9	0.1
HbA1c (%)			6.7 ± 1.1	6.9 ± 1.3	0.2	6.8 ± 1.1	6.8 ± 1.3	0.7	6.9 ± 1.1	6.7 ± 1.3	0.3	6.7 ± 1.0	6.9 ± 1.4	1.2
	HbA1c ≤ 7.5%		12 (22.6%)	27 (24.1%)	0.8	17 (25.4%)	22 (22.4%)	0.7	26 (29.5%)	13 (16.9%)	0.06	17 (23.0%)	22 (24.2%)	0.8
Antidiabetic drugs
Number of antidiabetics			2.0 ± 1.1	2.1 ± 1.1	0.6	1.9 ± 1.1	2.1 ± 1.1	0.2	1.9 ± 1.0	2.2 ± 1.1	0.09	1.9 ± 1.0	2.1 ± 1.1	0.2
	One		28 (41.8%)	56 (38.4%)	0.6	33 (41.2%)	51 (38.3%)	0.7	43 (40.6%)	40 (37.7%)	0.7	36 (40.9%)	48 (38.4%)	0.7
	Two or more		39 (58.2%)	90 (61.6%)	0.6	47 (58.8%)	82 (61.7%)	0.7	63 (59.4%)	66 (62.3%)	0.7	52 (59.1%)	77 (61.6%)	0.7
Type of antidiabetic drugs
Insulins			9 (13.4%)	43 (29.5%)	0.01	15 (18.8%)	37 (27.8%)	0.1	24 (22.6%)	28 (26.4%)	0.5	16 (18.2%)	36 (28.8%)	0.07
Insulin and oral antidiabetic			8 (11.9%)	34 (23.3%)	0.05	12 (15.9%)	30 (22.6%)	0.2	17 (16.0%)	25 (23.6%)	0.2	15 (17.05)	27 (21.6%)	0.4
Oral antidiabetics			66 (98.5%)	137 (93.8%)	0.1	77 (96.2%)	126 (94.7%)	0.6	99 (93.4%)	103 (97.2%)	0.2	87 (98.9%)	116 (92.8%)	0.04
	Oral antidiabetics excluding insulins		58 (86.6%)	103 (70.5%)	0.01	65 (81.2%)	96 (72.2%)	0.1	82 (77.4%)	78 (73.6%)	0.5	72 (81.8%)	89 (71.2%)	0.07
Drugs that stimulate insulin release			41 (61.2%)	81 (55.5%)	0.4	49 (61.3%)	73 (54.9%)	0.4	65 (61.3%)	57 (53.8%)	0.3	57 (64.8%)	65 (52.0%)	0.06
	Insulin secretagogues		11 (16.4%)	20 (13.7%)	0.6	12 (15.0%)	19 (14.3%)	0.9	18 (17.0%)	13 (12.3%)	0.3	15 (17.0%)	16 (12.8%)	0.4
		Sulfonylureas	5 (7.5%)	6 (4.1%)	0.3	5 (6.2%)	6 (4.5%)	0.6	8 (7.5%)	3 (2.8%)	0.1	7 (8.0%)	4 (3.2%)	0.1
		Repaglinide	6 (9.0%)	14 (9.6%)	0.9	7 (8.8%)	13 (9.8%)	0.8	10 (9.4%)	10 (9.4%)	1	8 (9.1%)	12 (9.6%)	0.9
	Acting on incretins		34 (50.7%)	72 (49.3%)	0.8	41 (51.2%)	65 (48.9%)	0.7	53 (50.0%)	53 (50.0%)	1	47 (53.4%)	59 (47.2%)	0.4
		GLP-1	6 (9.0%)	16 (11.0%)	0.6	3 (3.8%)	19 (14.3%)	0.01	4 (3.8%)	18 (17.0%)	0.002	4 (4.5%)	18 (14.4%)	0.02
		iDPP-4	28 (41.8)	56 (38.4)	0.6	38 (47.5%)	46 (34.6)	0.06	49 (46.2%)	35 (33.0%)	0.05	43 (48.9%)	41 (32.8%)	0.02
Insulin sensitizer			52 (77.6%)	110 (75.3%)	0.7	60 (75.0%)	102 (76.7%)	0.8	75 (70.8%)	86 (81.1%)	0.08	67 (76.1%)	95 (76.0%)	1.0
	Biguanides		52 (77.6%)	110 (75.3%)	0.7	60 (75.0%)	102 (76.7%)	0.8	75 (70.8%)	86 (81.1%)	0.08	67 (76.1%)	95 (76.0%)	1.0
		Metformin, plain	22 (32.8%)	39 (26.7%)	0.5	25 (31.2%)	36 (27.1%)	0.6	29 (27.4%)	31 (29.2%)	0.7	27 (30.7%)	34 (27.2%)	0.7
		Metformin + insulin	1 (1.5%)	3 (2.1%)		2 (2.5%)	2 (1.5%)		2 (2.3%)	2 (1.6%)		2 (2.3%)	2 (1.6%)
		Metformin + oral antidiabetic	10 (14.9%)	18 (12.3%)		10 (12.5%)	18 (13.5%)		14 (15.9%)	14 (11.2%)		14 (15.9%)	14 (11.2%)
		Metformin + insulin + additional oral antidiabetic	6 (9.0%)	26 (17.8%)		10 (12.5%)	18 (13.5%)		11 (12.5%)	21 (16.8%)		11 (12.5%)	11 (16.8%)
	Pioglitazone		4 (6.0%)	1 (0.7%)	0.01	3 (3.8%)	2 (1.5%)	0.3	5 (4.7%)	0 (0.0%)	0.02	3 (3.4%)	2 (1.6%)	0.4
SGLT2 inhibitor			19 (28.4%)	38 (26%)	0.7	15 (18.8%)	42 (31.6)	0.04	19 (17.9%)	38 (35.8%)	0.003	16 (18.2%)	41 (32.8%)	0.01
Alpha glucosidase inhibitors			2 (3.0%)	0 (0.0%)	0.03	2 (2.5%)	0 (0.0%)	0.06	2 (1.9%)	0 (0.0%)	0.1	2 (2.3%)	0 (0.0%)	0.09

Patients with type 2 diabetes (n = 213). Data are presented as mean ± standard deviation for continuous variables or number (proportion, %) for categorical variables. Means were compared with t-test and proportions with Chi-squared test. Total sample size: 213 patients. Missing data (n): APOE status: Diastolic Blood Pressure = 92; Educational level = 12; HbA1c = 46; MMSE = 14; Systolic Blood Pressure = 92; Smoking habit = 29; Total cholesterol = 43; Total glucose = 46. GLP-1: glucagon-like peptide-1; DPP4: dipeptidyl peptidase 4; MMSE: Mini-Mental State Examination; SGLT2: sodium-glucose cotransporter-2. *The identification of AD patients was based on CSF Aβ₄₂ (cut-off ≤725 pg/ml).

Patients with pathological biomarker levels were older and with worse MMSE punctuations than those with regular biomarker levels, showing a worse cognitive function than those patients without pathological AD biomarkers. In addition, there was a higher prevalence of patients with low educational level among patients with impaired t-tau and t-tau/Aβ₄₂ ratio.

Vascular risk factors, such as dyslipidemia, antiplatelet drug intake, and hypertension, were distributed similarly in all groups. Nevertheless, diastolic blood pressure levels were higher among patients with altered biomarkers. As for smoking habits, we observed that there were more smoker patients on the normal t-tau and t-tau/Aβ₄₂ ratio groups.

All patients were taking at least one antidiabetic drug. Specifically, 84 (39.4%) were treated with one antidiabetic agent and 129 (60.6%) with two or more. The use of insulin sensitizers, such as metformin, was the most frequent (76.1%), followed by drugs that stimulate insulin release (57.3%)- (insulin secretagogues (14 .5%), and drugs acting on incretins (49.8%)- SGLT2 inhibitors (26.8%), and insulins (24.4%), as shown in Figure 2.

Figure 2.

Use of antidiabetic medications among patients with diabetes in the cohort (A) and percentage of patients taking each drug plain or in combination with another antidiabetic agent (B). N = 213 patients.

Lower glucose levels were more common among people with diabetes and amyloid pathology. Patients with altered t-tau/Aβ₄₂ ratio consumed more oral antidiabetics and DPP-4 inhibitors than those without impaired levels. In contrast, patients with impaired levels of Aβ₄₂ were more likely to consume oral antidiabetics without additional insulin treatment. Finally, iSGLT2 and GLP-1 were more prescribed among patients without tau pathology (Table 1).

Antidiabetic agents and CSF biomarkers of ad pathology

We further examined the associations between specific antidiabetic agents (insulins, insulin sensitizers, stimulate insulin release, and SGLT2 inhibitors) and AD-related CSF biomarkers (Aβ₄₂, t-tau p-tau, t-tau/Aβ₄₂) regarding the influence of drug intake, dose and continued years of exposure, as showed in Table 2.

Table 2.

Associations between antidiabetic agent usage and Alzheimer's disease-related CSF biomarkers alterations in patients with type 2 diabetes.

Antidiabetic agent	Biomarker (impaired levels)	Model 1 Ors (95% CI)	p	Model 2 Ors (95% CI)	p	Model 3 Ors (95% CI)	p	Model 4 Ors (95% CI)	p	Model 5 Ors (95% CI)	p	Model 6 Ors (95% CI)	p
Insulins (n = 52)	Aβ₄₂	0.36 (0.15, 0.76)	0.01	0.45 (0.19, 0.87)	0.038	0.86 (0.74, 0.96)	0.02	0.30 (0.11, 0.72)	0.01	0.19 (0.04, 0.57)	0.01	0.81 (0.65, 0.95)	0.02
	t-tau	0.56 (0.27, 1.11)	0.11	0.52 (0.25, 0.95)	0.058	0.95 (0.86, 1.04)	0.27	0.51 (0.21, 1.19)	0.13	0.46 (0.18, 1.00)	0.08	0.96 (0.84, 1.07)	0.46
	p-tau	0.74 (0.38, 1.46)	0.39	0.66 (0.37, 1.08)	0.13	0.99 (0.92, 1.09)	0.98	0.63 (0.28, 1.43)	0.27	0.54 (0.24, 1.10)	0.11	0.96 (0.86, 1.09)	0.54
	t-tau/Aβ₄₂	0.49 (0.24, 0.98)	0.05	0.51 (0.25, 0.93)	0.046	0.90 (0.81, 0.99)	0.048	0.39 (0.16, 0.92)	0.03	0.23 (0.06, 0.63)	0.01	0.88 (0.76, 1.00)	0.06
Insulin sensitizer (n = 162)	Aβ₄₂	1.33 (0.67, 2.78)	0.42	1.44 (0.75, 2.81)	0.27	1.00 (0.95, 1.06)	0.86	1.11 (0.51, 2.46)	0.80	1.32 (0.64, 2.76)	0.45	0.99 (0.93, 1.05)	0.70
	t-tau	1.1 (0.56,2.22)	0.78	0.95 (0.49, 1.82)	0.88	1.02 (0.96, 1.08)	0.51	1.31(0.58, 3.01)	0.52	1.26 (0.59, 2.72)	0.55	1.01 (0.95, 1.08)	0.70
	p-tau	0.67 (0.33, 1.34)	0.26	0.68 (0.35, 1.31)	0.25	0.97(0.91, 1.03)	0.29	0.79 (0.35, 1.80)	0.58	0.88 (0.41, 1.88)	0.74	0.96 (0.89, 1.02)	0.24
	t-tau/Aβ₄₂	1.39 (0.69, 2.86)	0.36	1.14 (0.59, 2.22)	0.69	1.01 (0.95,1.07)	0.75	1.61 (0.71, 3.77)	0.26	1.40(0.66, 3.04)	0.39	1.00 (0.93, 1.06)	0.91
Stimulate insulin release (n = 122)	Aβ₄₂	1.2 (0.64, 2.15)	0.62	1.09 (0.69, 1.71)	0.70	1.04 (0.98, 1.11)	0.20	1.24 (0.64, 2.44)	0.53	1.03 (0.61, 1.70)	0.91	1.05 (0.98, 1.13)	0.18
	t-tau	1.05 (0.58, 1.92)	0.82	0.86 (0.54, 1.34)	0.51	1.01 (0.95, 1.08)	0.67	1.01 (0.49, 2.06)	0.98	0.59 (0.33, 1.03)	0.07	1.00 (0.92, 1.08)	0.97
	p-tau	1.07 (0.58, 1.94)	0.83	1.01 (0.65, 1.60)	0.95	1.02 (0.95, 1.09)	0.54	1.36 (0.67, 2.77)	0.39	1.06 (0.62, 1.84)	0.81	1.04 (0.96, 1.13)	0.36
	t-tau/Aβ₄₂	1.41 (0.77, 2.59)	0.27	1.05 (0.67, 1.65)	0.87	1.02 (0.96, 1.10)	0.40	1.36 (0.67, 2.76)	0.39	0.79 (0.46, 1.36)	0.41	1.03 (0.95, 1.11)	0.46
SGLT2 inhibitors (n = 57)	Aβ₄₂	1.24 (0.63, 2.40)	0.52	1.30 (0.67, 2.47)	0.42	0.95 (0.72, 1.21)	0.68	0.90 (0.43, 1.85)	0.78	1.11 (0.55, 2.20)	0.76	0.88 (0.65, 1.13)	0.35
	t-tau	0.57 (0.28, 1.13)	0.12	0.56 (0.26, 1.12)	0.12	0.82 (0.59,1.07)	0.17	0.47 (0.21, 1.03)	0.06	0.52 (0.22, 1.13)	0.11	0.81 (0.57, 1.08)	0.17
	p-tau	0.42 (0.21, 0.85)	0.01	0.53 (0.26, 1.04)	0.07	0.80 (0.60, 1.04)	0.11	0.44 (0.20, 0.94)	0.03	0.58 (0.27, 1.21)	0.15	0.83 (0.61, 1.09)	0.20
	t-tau/Aβ₄₂	0.51 (0.25, 1.03)	0.06	0.67 (0.33, 1.34)	0.27	0.80 (0.59, 1.05)	0.13	0.36 (0.16, 0.79)	0.01	0.57 (0.26, 1.21)	0.15	0.78 (0.56, 1.03)	0.10

Patients with type 2 diabetes n = 213. Models 1–3: Logistic regressions adjusted by age and sex. Exposure for Models 1 and 4 = Drug users (yes/no); Exposure for Models 2 and 5 = Drug consume in DDDs; Exposure for Models 3 and 6 = Drug exposure, in years. Models 4–6: Logistic regressions adjusted by age, sex, educational level, hypertension, dyslipidemia, smoking habit, and antiplatelet drugs intake. Reference levels = AD biomarker not pathological, female, high educational level, without hypertension and dyslipidemia, non-smoker, and not taking antiplatelet drugs or the drug of interest. Exposure for Model 4 = Drug users (yes/no); Exposure for Model 5 = Drug consume in DDDs; Exposure for Model 6 = Drug exposure, in years. Pathologic biomarker levels: Aβ₄₂< 725pg/ml; t-tau >84 pg/ml; p-tau > 349 pg/ml; t-tau/Aβ₄₂> 0.51. SGLT2: sodium-glucose cotransporter-2.

Using insulins, increasing doses and longer exposure times (reference: non-users) reduced the odds of pathological levels of Aβ₄₂. The association was statistically significant after adjusting for educational level and cardiovascular comorbidities (Table 2, Models 1–6). Insulin consumption and higher doses also reduced the odds of t-tau/Aβ₄₂ alteration.

Additionally, patients who used insulins for eight years or more tended to have a lower likelihood of having pathological Aβ₄₂ and t-tau/Aβ₄₂ ratio (OR95% = 0.21 [0.03, 0.82], p = 0.0495 for Aβ₄₂ impaired levels and OR95% = 0.3 [0.07, 0.98], p = 0.058 for t-tau/Aβ₄₂ pathological ratio), as shown in Figure 3 and Table 3. When comparing the mean ages among patients in different insulin exposure groups, no significant differences were observed. However, individuals who used insulin for over eight years started this treatment at a younger age, with a mean age of 58 ± 6 and a p < 0.001. These findings suggest that the antidiabetic effect of insulin may have a more noticeable impact during this period.

Figure 3.

Insulin consumption (in years) compared with Aβ₄₂ levels (A) and the t-tau/Aβ₄₂ ratio (B). Model A: Med [IQR] less than 4 years of insulin exposure = 868.7 [742.3; 1159.3]; From 4 to 8 years = 1139.0 [864.9; 1453.5], More than 8 years = 1663.0 [1035.0; 1887.0]; Non-insulin users = 810. 0 [638.0; 1058.0]. Model B: Med [IQR] less than 4 years of insulin exposure = 0.35 [0.26; 0.65]; From 4 to 8 years = 0.21 [0.17; 0.46], More than 8 years = 0.25 [0.21; 0.35]; Non-insulin users = 0.40 [0.23; 1.02]. AD: Alzheimer's disease; Aβ₄₂: amyloid-β 42; T-tau: total tau. Statistical tests were performed by Kruskal-Wallis rank sum test corrected by the Bonferroni method. No color is necessary for this figure.

Table 3.

Association between categorized antidiabetics years of exposure and pathologic AD biomarkers.

Biomarker	Time of exposure	Insulins ORs (95% CI)	p	Stimulate insulin release ORs (95% CI)	p	Insulin sensitizers exposure ORs (95% CI)	p	iSGLT2 ORs (95% CI)	p
Aβ₄₂ (impaired)	Less than 4 years	0.5 (0.2, 1.4),	0.2	0.9 (0.4, 1.9)	0.8	1.3 (0.5, 3.2)	0.6	1.4 (0.6, 2.9)	0.4
	More than 4 but less than 8 years	0.3 (0.05, 1.2)	0.1	0.8 (0.3, 2.1)	0.7	1.2(0.4, 3.0)	0.7	0.9 (0.2, 3.5)	0.9
	8 years or more	0.21 (0.03, 0.82)	0.0495	1.3 (0.6, 3.1	0.5	1.2 (0.6, 2.6)	0.6	NA	NA
t-tau (impaired)	Less than 4 years	0.6 (0.2, 1.5)	0.3	0.6 (0.3, 1.3)	0.2	1.0 (0.4,2.5)	0.9	0.5 (0.2, 1.2)	0.2
	More than 4 but less than 8 years	0.5 (0.1, 1.7)	0.3	0.7 (0.3, 1.8)	0.5	0.9 (0.3, 2.3)	0.9	0.6 (0.1, 2.6)	0.6
	8 years or more	0.6 (0.2, 1.8)	0.4	1.3 (0.5, 3.0)	0.5	1.1 (0.5, 2.5)	0.7	NA	NA
p-tau (impaired)	Less than 4 years	0.6 (0.2, 1.6)	0.3	1.0 (0.5, 2.2)	0.9	0.7 (0.3, 1.8)	0.5	0.4 (0.2, 0.9)	0.03
	More than 4 but less than 8 years	0.6 (0.2, 2.1)	0.5	0.9 (0.3, 2.1)	0.7	0.6 (0.2, 1.6)	0.3	0.5 (0.1, 2.2)	0.4
	8 years or more	1.1 (0.3, 3.4)	0.9	1.6 (0.7, 4.0)	0.3	0.6 (0.3, 1.3)	0.2	NA	NA
t-tau/Aβ₄₂ (impaired)	Less than 4 years	0.6 (0.2, 1.6)	0.3	1.2(0.5, 2.5)	0.7	1.4 (0.6, 3.6)	0.4	0.6 (0.3, 1.4)	0.2
	More than 4 but less than 8 years	0.6 (0.2, 2.1)	0.5	0.97 (0.4, 2.4)	0.9	1.2 (0.5, 3.2)	0.7	0.5 (0.09, 2.0)	0.3
	8 years or more	0.3 (0.07, 0.976)	0.058	1.6 (0.7, 3.7)	0.3	1.3 (0.6, 2.8)	0.5	NA	NA

Patients with type 2 diabetes (n = 213). Logistic regressions adjusted by age and sex. Reference levels = AD biomarker not pathological, female and not taking the drug of interest. Model 1 = Insulin exposure; Model 2 = Stimulate insulin release exposure; Model 3 = Insulin sensitizers exposure; Model 4 = iSGLT2 exposure.

Lastly, we observed that patients with a non-pathological t-tau/Aβ₄₂ ratio tended to consume more insulins (29 versus 18%; Figure 4). Moreover, there was an increased percentage of patients taking insulins with an antidiabetic drug other than metformin among this group.

Figure 4.

Metformin and insulin intake among patients with diabetes mellitus type 2 according to pathological levels of the ratio t-tau/Aβ₄₂. (A) Non-pathological t-tau/Aβ₄₂ (<0.51); (B) Pathological t-tau/Aβ₄₂ (>0.51). M+: Metformin user; M-: Metformin non-user; I+: Insulin user; I-: Insulin non-user. Blue, Consumes insulin; Light Blue (M-I+), Consumes insulin but not metformin; Medium Blue (M + I+), Consumes insulin and metformin; Dark Red, Does not consumes insulin; Medium red (M + I-), Consumes metformin but not insulin; Light red (M-I-), Does not consumes metformin or insulin. Print it in color, if possible.

Insulin sensitizers DDDs and exposure time showed no associations with altered AD biomarkers (Table 2). No differences were detected regarding the consumption of plain metformin versus an antidiabetic agents other tan metformin or non-users. Nevertheless, a stratified analysis of patients taking metformin (n = 162) showed that patients taking metformin and insulin were less likely to have altered Aβ₄₂ levels (Supplemental Table 1) than patients taking metformin but not insulin. Additionally, we compared plain metformin users with patients using concomitant metformin with (a) other oral antidiabetics; (b) just insulin, and (c) with insulin and oral antidiabetics (triple therapy, at least). The concomitant consumption of insulin, metformin, and other antidiabetic drugs was associated with lower odds of t-tau impaired levels (OR95% = 0.4 [0.1, 0.96], p = 0.048) and with a reduction of the odds for Aβ₄₂, t-tau, and t-tau/Aβ₄₂ among patients who at least completed high school and without other cardiovascular comorbidities (Supplemental Table 2).

No statistically significant associations were found between drugs that stimulate insulin release (insulin secretagogues and drugs acting on the incretins), their DDDs, or exposure time and AD biomarkers.

Patients consuming iSGLT2, instead, were less likely to have impaired p-tau levels (OR95% = 0.4 [0.2, 0.8], p = 0.01), even adjusted by cardiovascular comorbidities (Table 2). iSGLT2 intake also showed a decreased odd of t-tau/Aβ₄₂ ratio alteration among patients with high educational levels and without added cardiovascular comorbidities.

The role of glycemic control

We tested whether the associations between the above-mentioned antidiabetic groups and AD pathological biomarkers may be attributed to effective glycemic control among the 165 patients with diabetes and available HbA1c records. Of these, 39 (23.6%) had HbA1c ≥ 7.5%, indicating uncontrolled diabetes.

The use of insulins and iSGLT2 were each associated with increased odds of uncontrolled diabetes (OR95% = 10.8 [4.7, 26.1], p < 0.001 and OR95% = 2.3 [1.0, 5.2], p = 0.04), respectively.

Stratified analysis showed that insulin intake is associated with decreased odds of irregular t-tau, p-tau, and t-tau/Aβ₄₂ among patients with controlled diabetes. In contrast, higher insulin dosages decreased the odds for impaired Aβ₄₂ levels among patients with uncontrolled diabetes (Table 4). Educational and cardiovascular comorbidities models showed the same behavior for p-tau and t-tau/Aβ₄₂ (Supplemental Table 3).

Table 4.

Impact of antidiabetics drugs on CSF AD biomarkers according to diabetes control.

Patients with controlled diabetes (n = 126)									Patients with uncontrolled diabetes (n = 39)
Antidiabetic agent	Biomarker (impaired levels)	Model 1 ORs (95% CI)	p	Model 2 ORs (95% CI)	p	Model 3 ORs (95% CI)	p	Antidiabetic agent	Biomarker (impaired levels)	Model 4 ORs (95% CI)	p	Model 5 ORs (95% CI)	p	Model 6 ORs (95% CI)	p
Insulins (n = 17)	Aβ₄₂	0.41 (0.09,1.39)	0.19	0.80 (0.26, 1.73)	0.61	0.84 (0.58, 1.04)	0.21	Insulins (n = 24)	Aβ₄₂	0.35 (0.06, 1.65)	0.20	0.19 (0.02, 0.75)	0.04	0.89 (0.73, 1.05)	0.22
	t-tau	0.16 (0.02,0.66)	0.02	0.07 (0.0009, 0.67)	0.10	0.54 (0.15, 0.88)	0.14		t-tau	1.23 (0.32, 4.77)	0.76	0.69 (0.25, 1.71)	0.44	1.05 (0.92, 1.21)	0.45
	p-tau	0.24 (0.06, 0.82)	0.03	0.27 (0.02, 0.94)	0.13	0.84 (0.65, 1.01)	0.11		p-tau	0.85 (0.16, 3.92)	0.83	0.72 (0.26,1.93)	0.51	1.15 (0.97, 1.46)	0.16
	t-tau/Aβ₄₂	0.20 (0.04,0.72)	0.02	0.64 (0.18, 1.45)	0.36	0.77 (0.51, 0.97)	0.08		t-tau/Aβ₄₂	1.15 (0.27, 5.11)	0.85	0.39 (0.10, 1.10)	0.11	0.97 (0.84, 1.12)	0.72
Metformin (n = 101)	Aβ₄₂	0.57 (0.22, 1.49)	0.24	0.81 (0.32, 1.98)	0.64	0.95 (0.88,1.03)	0.22	Metformin (n = 27)	Aβ₄₂	3.79 (0.66, 33.14)	0.17	2.61 (0.60, 14.62)	0.23	1.02 (0.89, 1.16)	0.80
	t-tau	1.11(0.42, 3.07)	0.83	0.75 (0.30, 1.86)	0.53	0.98 (0.90, 1.06)	0.57		t-tau	1.03 (0.25, 4.37)	0.97	1.03 (0.29, 3.74)	0.97	1.04 (0.92, 1.17)	0.51
	p-tau	0.81 (0.28, 2.33)	0.70	0.51 (0.19, 1.34)	0.18	0.94 (0.86, 1.02)	0.15		p-tau	0.12 (0.01, 0.80)	0.05	0.39 (0.07, 1.68)	0.22	0.94 (0.81, 1.07)	0.35
	t-tau/Aβ₄₂	0.77 (0.27, 2.15)	0.62	0.60 (0.23, 1.53)	0.29	0.96 (0.88, 1.04)	0.38		t-tau/Aβ₄₂	3.31 (0.67, 20.71)	0.16	2.37 (0.58, 11.84)	0.25	1.02 (0.90, 1.16)	0.69
Stimulate insulin release (n = 67)	Aβ₄₂	1.15 (0.53, 2.52)	0.72	1.26 (0.67, 2.37)	0.46	1.03 (0.94, 1.12)	0.46	Stimulate insulin release (n = 28)	Aβ₄₂	1.23 (0.24, 7.45)	0.81	0.73 (0.22, 2.08)	0.58	0.99 (0.80, 1.19)	0.88
	t-tau	0.96 (0.44, 2.11)	0.93	0.69 (0.35, 1.31)	0.27	1.01 (0.92, 1.11)	0.80		t-tau	0.31 (0.06, 1.29)	0.12	0.51 (0.17, 1.21)	0.15	0.97 (0.83, 1.14)	0.74
	p-tau	0.95 (0.41, 2.13)	0.90	0.86 (0.44, 1.65)	0.65	1.03 (0.93, 1.13)	0.55		p-tau	0.32 (0.04,02 1.75)	0.22	0.39 (0.11, 1.12)	0.09	1.00 (0.83, 1.22)	0.99
	t-tau/Aβ₄₂	1.29 (0.58, 2.88)	0.52	0.80 (0.41, 1.53)	0.51	1.02 (0.93, 1.12)	0.65		t-tau/Aβ₄₂	1.52 (0.33, 7.62)	0.60	1.34 (0.53, 3.52)	0.53	1.07 (0.89, 1.28)	0.47
SGLT2 inhibitors (n = 30)	Aβ₄₂	1.19 (0.47, 2.91)	0.69	1.54 (0.60, 3.80)	0.35	0.92 (0.64, 1.24)	0.63	iSGLT2 (n = 15)	Aβ₄₂	0.82 (0.15, 3.84)	0.80	1.21 (0.31, 4.48)	0.78	1.41 (0.65, 3.56)	0.41
	t-tau	0.42 (0.15, 1.10)	0.08	0.51 (0.16, 1.41)	0.22	0.81 (0.55, 1.12)	0.25		t-tau	0.47 (0.11, 1.83)	0.29	0.51 (0.13, 1.79)	0.31	0.98 (0.45, 2.09)	0.96
	p-tau	0.36 (0.13, 0.95)	0.04	0.59 (0.19, 1.65)	0.33	0.82 (0.57, 1.12)	0.24		p-tau	0.05 (0.004, 0.36)	0.009	0.15 (0.02, 0.71)	0.03	0.65 (0.26, 1.56)	0.31
	t-tau/Aβ₄₂	0.49 (0.18, 1.26)	0.15	0.67 (0.22, 1.82)	0.44	0.77 (0.52, 1.07)	0.15		t-tau/Aβ₄₂	0.35 (0.06, 1.65)	0.20	0.19 (0.02, 0.75)	0.04	0.89 (0.73, 1.05)	0.22

Patients with glycated hemoglobin levels (n = 165). Logistic regressions adjusted by age and sex. Models 1–3: Patients with controlled diabetes (HbA1c% < 7.5; n = 126). Models 4–6: Patients with uncontrolled diabetes (HbA1c% ≥7.5; n = 39). Exposure for Models 1 and 3 = Drug users (yes/no); Exposure for Models 2 and 4 = Drug consume in DDDs; Exposure for Models 3 and 6 = Drug exposure, in years. Pathologic biomarker levels: Aβ₄₂< 725 pg/ml; t-tau >84 pg/ml; p-tau > 349 pg/ml; t-tau/Aβ₄₂> 0.51. SGLT2: sodium-glucose cotransporter-2.

Metformin intake is associated with a reduced odds of impaired p-tau levels among patients with uncontrolled diabetes (Table 4). Moreover, patients with controlled diabetes taking concomitant insulins and metformin had reduced odds of t-tau and t-tau/Aβ₄₂ presence. These differences were also observed among patients with controlled diabetes and no additional cardiovascular comorbidities (Table 5).

Table 5.

Impact of concomitant insulin and metformin use on CSF AD biomarkers according to diabetes control.

	Patients taking metformin and insulin with controlled diabetes (n = 15)				Patients taking metformin and insulin with uncontrolled diabetes (n = 14)
Biomarker	Model 1 ORs (95% CI)	p	Model 2 Adjusted by CVF	p	Model 3 ORs (95% CI)	p	Model 4 ORs (95% CI)	p
Aβ₄₂ (impaired)	0.5 (0.1, 1.7)	0.3	0.3 (0.04, 1.3)	0.1	1.00 (0.2, 5.7)	0.99	6.2e32 (0, Inf)	0.99
t-tau (impaired)	0.2 (0.03, 0.8)	0.04	0.1 (0.03, 1.2)	0.1	1.2 (0.3, 5.1)	0.8	2.0 (0.2, 36.7)	0.6
p-tau (impaired)	0.2 (0.04, 0.7)	0.02	0.1 (0.02, 0.7)	0.03	0.09 (0.006, 0.7)	0.047	0.04 (0.01, 6.3)	0.5
t-tau/Aβ₄₂ (impaired)	0.2 (0.05, 0.9)	0.05	0.2 (0.02, 0.8)	0.04	3.2 (0.7, 17.3)	0.1	4.8 e38 (0., Inf)	0. 99

Exposure: Concomitant use of insulin and metformin treatment. Models 1 and 2: Patients with controlled diabetes (HbA1c% < 7.5; n = 126). Models 3 and 4: Patients with uncontrolled diabetes (HbA1c% ≥7.5; n = 39). Models 1 and 3 are logistic regressions adjusted by age and sex. Models 2 and 4 are logistic regressions adjusted by age, sex, educational level, hypertension, dyslipidemia, smoking habit, and antiplatelet drugs intake. Reference levels = AD biomarker not pathological, female, high educational level, without hypertension and dyslipidemia, non-smoker, and not taking antiplatelet drugs or metformin and insulin at the same time. Pathologic biomarker levels: Aβ₄₂< 725 pg/ml; t-tau >84 pg/ml; p-tau > 349 pg/ml; t-tau/Aβ₄₂> 0.51. SGLT2: sodium-glucose cotransporter-2.

Lastly, higher doses of iSGLT2 seemed relevant to reduce the odds of altered p-tau and t-tau/Aβ₄₂ ratio among patients with uncontrolled diabetes (OR95% = 0.2 [0.02, 0.7], p = 0.03) and (OR95% = 0.2 [0.02, 0.7], p = 0.03), Table 4, whereas iSGLT2 intake reduced the odds of altered t-tau/Aβ₄₂ ratio among patients with controlled diabetes, high educational level, and no added cardiovascular comorbidities (OR95% = 0.3, [0.08, 0.9], p = 0.03) (Supplemental Table 3).

Discussion

In this cross-sectional clinical-based study, we explored the association of various antidiabetic agents, their daily dosage intake, and treatment duration with AD-related CSF biomarkers, considering the influence of effective glycemic control. Our study highlighted four key findings. Firstly, the use of insulin is not as influential in AD-related pathology as its daily dose intake and the lifespan period in which insulin therapy starts. Secondly, the modulatory effects that diabetes control may have on antidiabetics effects on CSF AD biomarkers. Thirdly, the potential beneficial effects of combined antidiabetics therapy on AD pathology. Lastly, the promising role of iSGLT2 on tauopathies.

These results suggest that, despite diabetes serving as a significant risk factor for dementia,²³ effective diabetes management, with an emphasis on glycemic control, may attenuate its detrimental effects on AD pathology due to the brain's insulin sensitivity.^1,2,6,11,24

In this sense, Plastino et al. showed that patients receiving combined antidiabetic therapy, including both an oral treatment and insulin, exhibited better cognitive outcomes than patients solely receiving oral antidiabetic agents.²⁵ Although we did not observe statistical differences when we replicated that analysis in our cohort, we observed reduced odds of pathological t-tau and t-tau/Aβ₄₂ presence among patients with controlled diabetes consuming concomitant insulin and metformin among metformin users. Similar effects were observed when additional oral antidiabetic was added to the previous ones on patients without added cardiovascular comorbidities. Unfortunately, we could not assess diabetes control influence on this last model due to our limited sample size.

Figure 5A illustrates the shared mechanisms between diabetes and AD biomarkers, highlighting the potential target for antidiabetics as disease-modifying drugs for AD. Briefly, insulin enhances Aβ clearance⁶ and involves the PI3 K/Akt pathway, reducing tau phosphorylation² (Figure 5B). While our primary analysis shows that insulin therapy may reduce the risk of amyloid accumulation, our stratified analysis reveals that this effect on amyloid is only sustained in patients with uncontrolled diabetes, shifting its action towards tau when the condition is adequately managed. It may suggest that insulin administration could help address disruptions in amyloid clearance during the early stages, subsequently acting on mechanisms like tau once insulin resistance is adequately treated, but it requires further studies.

Figure 5.

A schematic version of the proposed mechanism of action for insulin and metformin on amyloid and tau disruptions. (A) Insulin resistance impact on amyloid and tau. Patients with uncontrolled diabetes may have higher brain insulin resistance, which would lead to decreased levels of brain insulin and decreased levels of IDE, culminating in the accumulation of Aβ plaques and induction of tau phosphorylation. Conversely, effective diabetes management could slow down the accumulation of AD pathological hallmarks; (B) Insulin effects on amyloid and tau. Insulin enhances Aβ clearance and involves the PI3 K/Akt pathway, reducing tau phosphorylation. Additionally, it ameliorates microglia and astrogliosis; (C) Metformin effects on amyloid and tau. Metformin is linked to a reduced formation of Aβ plaques by inhibiting BACE1 activity and enhancing IDE expression. Metformin promotes neuronal survival by activating AMPK. As a result, it inhibits the mTOR activation and enhances the expression of PPA2, a process that is also influenced by insulin signaling. This cascade of events can ultimately reduce GSK3β and potentially reduce tau phosphorylation. Finally, a downregulation of oxidative stress, normalization of mitochondrial dysfunction, and the inhibition of inflammation by reducing the microglia overactivation or promoting their autophagy capability are other mechanisms by which metformin may influence AD pathogenesis. Please move forward to the discussion and the cited articles for further details. Created with BioRender. AMPK: AMP-activated protein kinase; Akt: Protein kinase B; APP: amyloid-β protein precursor; BACE1: β-site APP cleaving enzyme; ERK: extracellular signal-regulated kinase; GLUT1: glucose transporter 1; GLUT3: glucose transporter 3; GSK3β: glycogen synthase kinase-3; IDE: insulin-degrading enzyme; IGF: insulin-like growth factor; INRS: insulin receptor; IRS-1: insulin receptor substrate; JNK: c-Jun N-terminal kinase; MAPK: mitogen-activated protein kinase; mTOR: mammalian target of rapamycin; M1: M1 microglia subtype; M2: M2 microglia subtype; PI3K: phosphoinositide 3-kinase; PPA2: phosphatase 2A protein; RAS: rat sarcoma virus; S: serine; T: threonine; γ-secretase: gamma secretase. Print it in color.

A four-month trial indicated that intranasal insulin decreased cognitive decline, showing a correlation between the improvement in cognitive functions and lower t-tau/Aβ₄₂ ratio among the insulin-treated arm.²⁶ Supporting these results, we observe a dose-response reduction in the likelihood of pathological p-tau levels, and t-tau/Aβ₄₂ ratio when insulin is administrated on diabetes-controlled patients with high educational level and no additional cardiovascular comorbidities. Other studies also suggest potential benefits of this drug administration in CSF Aβ₄₂/Aβ₄₀ and t-tau/Aβ₄₂ ratios,²⁷ increased IFN-γ, and alterations in inflammatory markers linked to AD.²⁸ Nevertheless, clinical trials aiming to assess the efficacy of intranasal insulin for the treatment of mild cognitive impairment and AD dementia yielded inconclusive results.^8,29 Regarding cohort studies, an increased risk of dementia with insulin usage has been reported,²⁴ while others have shown a stabilization of the cognitive status when insulin-treated patients were compared to those taking just oral antidiabetic drugs.²⁵ While insulin therapy offers benefits in managing glycemic instability and averting complications linked to chronic hyperglycemia, it might also heighten unfavorable cognitive outcomes due to an elevated risk of secondary hypoglycemia. Therefore, we examined whether patients taking insulins and sulfonylureas (associated with higher hypoglycemia risk) reported poorer biomarker outcomes, but our findings showed no significant differences.

Insulin is prescribed to type 2 diabetes patients when it is hard to control with oral drugs. In addition, it has been described that diabetes's impact on dementia is especially relevant in later life. Our study results suggest that starting insulin treatment early in adulthood could counteract the alteration of AD-like biomarkers. Consequently, intranasal insulin administration could be a plausible solution for diabetes-free patients too, due to its enhanced blood-brain barrier penetration and its lower risk of hypoglycemia compared to systemic insulin.^8,11,26 Since insulin transport across the blood-brain barrier is saturable,⁶ enhanced insulin signaling peripherally and centrally in the brain may offer benefits for AD and diabetes, given shared underlying mechanisms. Taking together, prolonged insulin exposure may be of particular interest in individuals around the age of 58.

Some insulin-related trials have indicated variability in memory improvements based on biological sex at birth and APOE ε4 status.² Our study, limited by a small sample of patients with diabetes with APOE genotype (n = 168 participants), showed the modifying influence of the APOE ε4 allele in the association between insulin and AD biomarkers among APOE ε4 carriers (OR95% = 0.15 [0.02, 0.76], p = 0.03), but not among non-APOE ε4 carriers. Future studies need to investigate in-depth whether the effect of insulin therapy on AD biomarkers differs in other genotypes (e.g., ε2 carriers, which are notoriously protected from AD), as well as considering biological sex differences and glycemic control.

Metformin, a first-line antidiabetic drug, may help slow the rate of cognitive decline with great potential in dementia prevention.^30,31 Specifically, metformin is linked to a reduced formation of Aβ plaques^2,10 (Figure 4C) and promotes neuronal survival,^10,32 which can ultimately reduce GSK3β and potentially reduce tau phosphorylation.² In our cohort, metformin was often combined with other antidiabetic drugs. We did not observe differences in using metformin alone or in combination concerning AD biomarker alteration, which aligns with prior findings.^30,31 Nonetheless, we detected a trend between higher metformin dosages and lower p-tau pathology among patients with uncontrolled diabetes, suggesting that the pathway linking the amount of metformin to tau pathology may be influenced by glycemic control. A study performed with the MEDALZ cohort showed that long-term and high doses of metformin are associated with a lower risk of incident AD,³³ supporting a dose-response effect in prevention of dementia. In addition, research from the Alzheimer's Disease Neuroimaging Initiative cohort showed that the group with treated diabetes had more favorable AD biomarkers than untreated diabetes, with metformin the most commonly consumed antidiabetic drug.⁷ However, we could not detect such a difference in our sample because all of the patients with diabetes in our cohort were receiving antidiabetic drugs. Other studies with the same cohort have shown a benefit in AD biomarkers among AD-like mild cognitive impairment patients treated with metformin compared to those with just AD-like cognitive decline.³⁴ Moreover, metformin has been associated with better-preserved brain volumes in specific regions compared to other drugs.³⁵ When we looked for AD biomarkers differences according to metformin combination with other drugs, we observed that combined therapy with at least insulin may exert preventive effects of amyloid and tau alteration, compared with plain metformin intake. Overall, our results match previous studies linking metformin to reduced tau levels,^7,36 suggesting that long-term antidiabetic exposure might counteract diabetes damage on dementia^30,33—a hypothesis that needs further testing.

Clinical outcomes of drugs that stimulate insulin release showed no significant associations between their usage, daily dosage or duration, and AD biomarkers. Previous studies suggested potential benefits of thiazolidinediones (an insulin-sensitizing agent) in early-stage AD,¹¹ but subsequent trials failed to confirm significant benefits.^2,10 Similarly, the GLP-1 analog liraglutide has shown promise in reducing Aβ production,¹⁰ and p-tau downregulation.¹⁰ However, clinical trials have not confirmed those results,³⁷ as observed in our study. Promising results in AD transgenic mice are also observed with DPP-4 inhibitors.^10,11 As for SGLT2 inhibitors usage, it has been associated with a 42% decreased risk of dementia, hindering AD prevention when used for more than three years.¹⁰ Additionally, a recent epidemiological study shows an association between iSGLT2 consumption and lower odds of anti-AD drug intake among patients with type 2 diabetes aged between 70 and 80 years.³⁸ In our cohort, SGLT-2 inhibitors exposure seemed to decrease the odds of p-tau and t-tau/Aβ₄₂ alteration. Other antidiabetic therapies targeting Aβ reduction include amylin analogs,² although not evaluated in the present study due to small sample size and lack of patient exposure.

Strengths and limitations

Our study has several strengths that distinguish it from others in the field. It explores in-depth the major pharmaceutical diabetes treatments in relation to CSF biomarkers of AD. In addition to the usage (yes/no) of a specific antidiabetic agent, it includes information on DDD and treatment duration. The inclusion of patients who underwent lumbar punctures allowed for a direct comparison between antidiabetic exposure and the CSF AD biomarkers, which are the same outcomes that clinical trials typically assess.³⁹ This approach sets our study apart from those primarily focusing on dementia or cognitive progression as endpoints, focusing on the underlying biological pathways. The sub-analysis of glycated hemoglobin further enhances the interpretation of the findings regarding glycemic control.

Some limitations also need to be acknowledged. First, there was a reduced sample size for some promising novel antidiabetic drugs, such as GLP-1 or DPP-4 inhibitors, and certain stratified analysis. Nonetheless, significant associations were observed between metformin and insulin with AD biomarkers. Secondly, data on the diabetes duration, history of antidiabetic therapy before recruitment, and history of hypoglycemic events are unavailable in our sample, which might have led to an underestimation of the observed associations. Thirdly, our study did not assess the genetic predisposition to AD and diabetes using, for example, polygenic risk scores. It would have helped to distinguish which patients might benefit the most from antidiabetic intake. Future studies should consider that individuals using insulin for diabetes management may have had the condition longer and possess diabetes and AD-related comorbid risk factors, which would also influence the effect of antidiabetic drugs on brain health as well as raise multi-drug interactions. Fourth, patients with pathological AD biomarkers were older than patients with unimpaired biomarkers. Although we adjusted all the analysis by age and sex for a better approach, diabetes presence is also linked to age and this needs to be considered in all interpretations. Finally, it is important to recognize that various factors, including adherence to treatment, cognitive status, psychological well-being, and socioeconomic status, may influence diabetes treatment status and must be considered in further studies. The actual effect of diabetes treatment on AD pathology can only be determined via randomized clinical trials. However, observational studies like the current study can provide valuable insights into the relationship between treatment status and AD-related pathological changes.

In summary, our study highlighted distinct associations between various antidiabetic agents and AD biomarkers. Notably, insulins, metformin and iSGLT2 stand out with the most impact. Consequently, it could be prudent to consider the use and dosage of such medications in patients with diabetes and AD pathology. Some antidiabetic medications show potential for enhancing diabetes control and may influence the progression of AD, though responses can vary among individuals. Still, it is essential to note that our findings may not be attributable to a single drug but rather influenced by a combination of medications and diabetes comorbidities such as hypertension and overweight/obesity. Further studies are needed to deepen the understanding of the interactions between peripheral insulin consumption, brain insulin, and other neurodegenerative and cerebrovascular pathologies, ultimately culminating in cognitive disorders.

Supplemental Material

sj-docx-1-alz-10.1177_13872877241304995 - Supplemental material for Associations between antidiabetic medications and cerebrospinal fluid biomarkers of Alzheimer's disease

Supplemental material, sj-docx-1-alz-10.1177_13872877241304995 for Associations between antidiabetic medications and cerebrospinal fluid biomarkers of Alzheimer's disease by Gemma García-Lluch, Anna Marseglia, Lucrecia Moreno Royo, Juan Pardo Albiach, Mar Garcia-Zamora, Miquel Baquero, Carmen Peña-Bautista, Lourdes Álvarez, Eric Westman and Consuelo Cháfer-Pericás in Journal of Alzheimer's Disease

Footnotes

Acknowledgments

The authors are grateful to Cathedra DeCo Micof-UCH, to GINEA's team and to all the participants and caregivers of the study participants.

ORCID iDs

Miquel Baquero

Carmen Peña-Bautista

Consuelo Cháfer-Pericás

Author contributions

Gemma García-Lluch (Data curation; Formal analysis; Methodology; Writing – original draft); Anna Marseglia (Data curation; Formal analysis; Methodology; Writing – original draft); Lucrecia Moreno Royo (Conceptualization; Funding acquisition; Investigation; Methodology; Supervision; Writing – review & editing); Juan Pardo Albiach (Formal analysis; Investigation; Supervision; Writing – review & editing); Mar Garcia-Zamora (Formal analysis; Methodology; Writing – original draft); Miquel Baquero (Conceptualization; Investigation; Supervision; Writing – review & editing); Carmen Peña-Bautista (Data curation; Investigation; Methodology; Writing – review & editing); Lourdes Álvarez (Data curation; Investigation; Methodology); Eric Westman (Investigation; Writing – review & editing); Consuelo Chafer-Pericas (Conceptualization; Formal analysis; Funding acquisition; Investigation; Supervision; 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 funded by Cathedra DeCo Micof-UCH and by Instituto de Salud Carlos III through the project PI22/00594 (Co-funded by European Union). GGL was further supported by an internship grant “VII Convocatorias de Ayudas Movilidad CEINDO 22- 23”AM was supported by the Strategic research area in Neuroscience at Karolinska Institutet (StratNeuro), Center for Innovative Medicine (CIMED) [FoUI-988254], Gamla Tjännarinor Foundation [2022-01312], Loo och Hans Ostermans foundation [FS-2023:0005], Foundation for Geriatric Diseases at Karolinska Institutet [FS-2023:0007]. CPB received support by a predoctoral “PFIS” grant FI20/00022 from the ISCIII. CC-P was supported by a postdoctoral “Miguel Servet” grant CPII21/00006 and a FIS PI22/00594 project from the Instituto de Salud Carlos III (ISCIII). A.M. received grants from the Center for Innovative Medicine (CIMED) (FoUI-988254); the Swedish Research Council for Health, Working Life and Wellfare (FORTE, 2024-00210); the Foundation for Geriatric Diseases at Karolinska Institutet (2024-02114, 2023-01598, 2022-01268); the Loo och Hans Ostermans stiftelsen (2024-02166, 2023-01645, 2022-01255); Gamla Tjänarinnor stiftelse (2023-085, 2020-00946); Demensfonden, the Strategic Research Programme in Neuroscience at KI (StratNeuro); and Edith Fernström Foundation.

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

Restrictions apply to the availability of all data generated and analyzed during this study to preserve patient confidentiality and because they were used under license. The corresponding author, Consuelo Cháfer Pericás (Hospital Universitari i Politècnic de València), will on request detail the restrictions and any conditions under which access to some data may be provided. For further information on the data and how to request access, email Dr Consuelo Cháfer Pericás (m.consuelo.chafer@uv.es).

Supplemental material

for this article is available online.

References

Kim

Arvanitakis

. Insulin resistance, cognition, and Alzheimer disease. Obesity (Silver Spring) 2023; 31: 1486–1498.

Mantik

KEK

Kim

, et al. Repositioning of anti-diabetic drugs against dementia: insight from molecular perspectives to clinical trials. Int J Mol Sci 2023; 24: 11450.

Rojas

Chávez-Castillo

Bautista

, et al. Alzheimer’s disease and type 2 diabetes mellitus: pathophysiologic and pharmacotherapeutics links. World J Diabetes 2021; 12: 745–766.

Jack

CRJ

Bennett

Blennow

, et al. NIA-AA Research framework: toward a biological definition of Alzheimer’s disease. Alzheimers Dement 2018; 14: 535–562.

Badji

Youwakim

Cooper

, et al. Vascular cognitive impairment – past, present, and future challenges. Ageing Res Rev 2023; 90: 102042.

Rhea

Leclerc

Yassine

, et al. State of the science on brain insulin resistance and cognitive decline due to Alzheimer’s disease. Aging Dis 2024; 15: 1688–1725.

McIntosh

Nation

. Importance of treatment status in links between type 2 diabetes and Alzheimer’s disease. Diabetes Care 2019; 42: 972–979.

Craft

Raman

Chow

, et al. Safety, efficacy, and feasibility of intranasal insulin for the treatment of mild cognitive impairment and Alzheimer disease dementia: a randomized clinical trial. JAMA Neurol 2020; 77: 1099–1109.

Powers

D’Alessio

. Capítulo 47: Páncreas endocrino y farmacoterapia de la diabetes mellitus y la hipoglucemia. In: Brunton

Chabner

Knollmann

(eds) Goodman & Gilman: Las Bases Farmacológicas de la Terapéutic. 13 ed. Ciudad de México, México: McGraw-Hill Interamericana Editores, S.A., 2019, pp.863–886.

10.

Huang

Cui

, et al. Clinical antidiabetic medication used in Alzheimer’s disease: from basic discovery to therapeutics development. Front Aging Neurosci 2023; 15: 1122300.

11.

Michailidis

Tata

Moraitou

, et al. Antidiabetic drugs in the treatment of Alzheimer’s disease. Int J Mol Sci 2022; 23: 4641.

12.

Baquero

Ferré-González

Álvarez-Sánchez

, et al. Insights from a 7-year dementia cohort (VALCODIS): apoE genotype evaluation. J Clin Med 2024; 13: 4735.

13.

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.

14.

Chapman

Bing-Canar

Alosco

, et al. Mini mental state examination and logical memory scores for entry into Alzheimer’s disease trials. Alzheimers Res Ther 2016; 8: 9.

15.

Peña-Bautista

Álvarez-Sánchez

Pascual

, et al. Clinical usefulness of cerebrospinal fluid biomarkers in Alzheimer’s disease. Eur J Clin Invest 2023; 53: e13910.

16.

Peña-Bautista

Álvarez-Sánchez

Ferrer

, et al. Lipid peroxidation assessment in preclinical Alzheimer disease diagnosis. Antioxidants (Basel) 2021; 10: 1043.

17.

American Diabetes Association. Understanding diabetes diagnosis. https://diabetes.org/about-diabetes/diagnosis (accessed 15 October 2023).

18.

WHOCC – Guidelines, https://www.whocc.no/atc_ddd_index_and_guidelines/guidelines/ (accessed 14 October 2023).

19.

Rosendale

Wood

Leung

, et al. Differences in cardiovascular health at the intersection of race, ethnicity, and sexual identity. JAMA Netw Open 2024; 7: e249060.

20.

Bender

Lange

. Adjusting for multiple testing–when and how? J Clin Epidemiol 2001; 54: 343–349.

21.

R-studio, https://posit.co/ (accessed 21 December 2022).

22.

Faul

Erdfelder

Buchner

, et al. Statistical power analyses using G*power 3.1: tests for correlation and regression analyses. Behav Res Methods 2009; 41: 1149–1160.

23.

Livingston

Huntley

Sommerlad

, et al. Dementia prevention, intervention, and care: 2020 report of the lancet commission. Lancet 2020; 396: 413–446.

24.

Weinstein

Davis-Plourde

Conner

, et al. Association of metformin, sulfonylurea and insulin use with brain structure and function and risk of dementia and Alzheimer’s disease: pooled analysis from 5 cohorts. PLoS One 2019; 14: e0212293.

25.

Plastino

Fava

Pirritano

, et al. Effects of insulinic therapy on cognitive impairment in patients with Alzheimer disease and diabetes mellitus type-2. J Neurol Sci 2010; 288: 112–116.

26.

Craft

Baker

Montine

, et al. Intranasal insulin therapy for Alzheimer disease and amnestic mild cognitive impairment: a pilot clinical trial. Arch Neurol 2012; 69: 29–38.

27.

Kellar

Lockhart

Aisen

, et al. Intranasal insulin reduces white matter hyperintensity progression in association with improvements in cognition and CSF biomarker profiles in mild cognitive impairment and Alzheimer’s disease. J Prev Alzheimers Dis 2021; 8: 240–248.

28.

Kellar

Lockhart

, et al. Intranasal insulin modulates cerebrospinal fluid markers of neuroinflammation in mild cognitive impairment and Alzheimer’s disease: a randomized trial. Sci Rep 2022; 12: 1346.

29.

Rosenbloom

Barclay

Kashyap

, et al. A phase II, single-center, randomized, double-blind, placebo-controlled study of the safety and therapeutic efficacy of intranasal glulisine in amnestic mild cognitive impairment and probable mild Alzheimer’s disease. Drugs Aging 2021; 38: 407–415.

30.

Sood

Capuano

Wilson

, et al. Metformin, age-related cognitive decline, and brain pathology. Neurobiol Aging 2024; 133: 99–106.

31.

Tang

Guo

Shaaban

, et al. Heterogeneous treatment effects of metformin on risk of dementia in patients with type 2 diabetes: a longitudinal observational study. Alzheimers Dement 2024; 20: 975–985.

32.

Zhao

Fan

Zhang

, et al. Metformin attenuates tau pathology in tau-seeded PS19 mice. Neurotherapeutics 2023; 20: 452–463.

33.

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.

34.

Pomilio

González Pérez

Calandri

, et al. Diabetic patients treated with metformin during early stages of Alzheimer’s disease show a better integral performance: data from ADNI study. Geroscience 2022; 44: 1791–1805.

35.

Nabizadeh

Kankam

Balabandian

, et al. Metformin use and brain atrophy in nondemented elderly individuals with diabetes. Exp Gerontol 2022; 166: 111890.

36.

Chen

Zhao

Fan

, et al. Metformin attenuates plaque-associated tau pathology and reduces amyloid- β burden in APP/PS1 mice. Alzheimers Res Ther 2021; 13: 40.

37.

Gejl

Gjedde

Egefjord

, et al. In Alzheimer's disease, 6-month treatment with GLP-1 analog prevents decline of brain glucose metabolism: randomized, clinical trial. Front Aging Neurosci 2016; 8: 108.

38.

Zamora

García-Lluch

Moreno

, et al. Assessment of sodium-glucose cotransporter 2 inhibitors (SGLT2i) and other antidiabetic agents in Alzheimer’s disease: a population-based study. Pharmacol Res 2024; 206: 107295.

39.

Cummings

Zhou

Lee

, et al. Alzheimer’s disease drug development pipeline: 2023. Alzheimers Dement (N Y) 2023; 9: e12385.

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

0.05 MB