Abstract
Objective:
To examine associations of serum insulin and related measures with neuropathology and cognition in older persons.
Methods:
We studied 192 older persons (96 with diabetes and 96 without, matched by sex and balanced by age-at-death, education, and postmortem interval) from a community-based, clinical-pathologic study of aging, with annual evaluations including neuropsychological testing (summarized into global cognition and five cognitive domains) and postmortem autopsy. We assessed serum insulin, glucose, leptin, adiponectin, hemoglobin A1C, advanced glycation-end products (AGEs), and receptors for AGEs (RAGE) and calculated Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) and adiponectin-leptin ratio. Using adjusted regression analyses, we examined the associations of serum measures with neuropathology of cerebrovascular disease and Alzheimer’s disease (AD), and with the level of cognition proximate-to-death.
Results:
Higher HOMA-IR was associated with the presence of brain infarcts and specifically microinfarcts, and higher HOMA-IR and leptin were each associated with subcortical infarcts. Further, higher leptin levels and lower adiponectin-leptin ratios were associated with the presence of moderate to severe atherosclerosis. Serum insulin and related measures were not associated with the level of AD pathology, as assessed by global, as well as amyloid burden or tau tangle density scores. Regarding cognitive outcomes, higher insulin and leptin levels, and lower adiponectin and RAGE levels respectively, were each associated with lower levels of global cognition.
Interpretation:
Peripheral insulin resistance indicated by HOMA-IR and related serum measures were associated with a greater burden of cerebrovascular neuropathology as well as with lower cognition.
Introduction
Type 2 diabetes mellitus (T2DM) is a complex disease, which is often caused by obesity or the accumulation of an excessive amount of body fat1 and with lifestyle and other factors. Among a multitude of complications, T2DM has been increasingly recognized over the past two decades as an important risk factor for cognitive impairment and dementia attributed to Alzheimer’s disease (AD) in older individuals.2–4 Nevertheless, the pathophysiological mechanisms underlying the link between T2DM and dementia remain uncertain and remain somewhat ambiguous on postmortem histoanatomical assessments. Individuals with T2DM are known to be susceptible to cerebral vascular diseases including stroke,5 which itself is a major cause of dementia (e.g., vascular contributions to cognitive impairment and dementia [VCID]).6 This mechanism is further supported by postmortem neuropathologic studies showing that diabetes is associated with a higher burden of cerebrovascular disease pathology such as cerebral infarctions, arteriolosclerosis, and atherosclerosis.7–9 By contrast, the results of previous studies examining the association between T2DM and AD neuropathology are mixed.7–10 This suggests that the mechanistic links between T2DM and dementia are complex and warrant further investigation. A core biological feature of both T2DM and obesity is insulin resistance, which has gained increasing attention from researchers studying the relationship between T2DM and AD in recent years.11 Defined as impairment of biological response (e.g., lowering of plasma glucose) to a given concentration of insulin,12 insulin resistance in clinical and population studies is often measured using the Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) based on fasting glucose and insulin concentrations.13 Previous studies have found that, in cognitively normal older adults, peripheral insulin resistance is associated with poorer performance on neurocognitive tests as well as abnormalities in AD biomarkers, including increases in cerebrospinal fluid phosphorylated tau protein level and lower global cerebral glucose metabolism on positron emission tomography (PET) scans.14, 15
Furthermore, accumulating evidence suggests that a major mechanism to induce peripheral insulin resistance involves an imbalance of pro- and anti-inflammatory adipokines (e.g., leptin and adiponectin, respectively) expressed by adipose tissue,16, 17 which could also account for the association between T2DM and dementia potentially via hormonal mechanisms. For example, a recent study found that leptin, but not adiponectin, was associated with cognitive impairment in older adults,18 whereas another revealed that a low adiponectin-leptin ratio was significantly associated with higher odds of AD diagnosis after adjustment of multiple demographic and clinical variables.19 Further, some studies suggest that adipokines implicated in glucose metabolism including adiponectin, may be particularly important in cognitive decline of older women.20, 21 Yet another mechanism associated with insulin resistance involves the accelerated formation of advanced glycation end products (AGEs),22 nonenzymatically glycated proteins or lipids that induce oxidative stress and inflammation through the interaction with receptors for AGEs (RAGE).23, 24 Indeed, emerging evidence suggests that serum levels of AGEs and RAGE were not only highly correlated with their levels in the brain,25 but also associated with cognitive decline and dementia.26–28 Nevertheless, rigorous studies investigating the above questions are scarce, and whether serum insulin and related measures such as adipokines and AGEs in older adults are associated with postmortem neuropathology remains largely unknown.
In this study, we examined the associations between serum insulin and related measures including glucose, leptin, adiponectin, hemoglobin A1C (A1C), AGEs, RAGE, HOMA-IR, and adiponectin-leptin ratio with neuropathology of cerebrovascular disease (brain infarcts and atherosclerosis, arteriolosclerosis, amyloid angiopathy) and AD (including amyloid and tau tangles), and summary measures of cognition (global and cognitive domain measures). Data from 192 deceased and autopsied older adults with or without diabetes, who participated in the Rush Memory and Aging Project (MAP), a community-based, clinical-pathologic study of aging were analyzed. We tested the hypothesis that a higher level of insulin resistance is associated with a greater burden of cerebrovascular disease and AD neuropathology. We also examined the associations between insulin and related measures and levels of cognition proximate-to-death.
Methods
Participants
All participants for this study were enrolled in the Rush Memory and Aging Project (MAP), an ongoing, prospective, community-based clinical-pathologic cohort study of aging with an emphasis on cognitive decline and risk of Alzheimer’s disease.29 This study received approval from the Institutional Review Board of Rush University Medical Center. Participants signed an informed consent to undergo yearly testing and an anatomical gift act to donate the brain, the spinal cord, and selected nerve and muscles at the time of death.29 MAP began enrolling residents of retirement communities in 1997. At the time of analyses, 2184 people enrolled and 1021 died over the course of the study. Of 980 persons eligible for autopsy, 840 underwent an autopsy (86% autopsy rate). Of those autopsied, 819 had neuropathologic data available at the time of the analyses for this study.
Clinical Evaluations
Participants underwent annual clinical evaluations that included a medical history, physical examination, and neuropsychological testing. Nineteen cognitive tests were grouped to form composite measures of global cognition and five cognitive domains including episodic memory, working memory, semantic memory, perceptual speed, and visuospatial ability30 Raw scores of individual tests were first converted to z scores, using the baseline mean and standard deviation from the entire cohort, and then z scores for all tests in each domain were averaged to create a composite score, as previously described.30 Presence of diabetes was determined based on medical history, visually inspected antidiabetic medications, or both, and most, if not all participants, had type 2 diabetes.31
Serum Measures
Participants also underwent annual blood draws in the community using standardized procedures, but being in a fasting state was optional for participants. Serum biomarkers related to glycemia, insulin resistance and adiposity were measured proximate to death. Measurements were collected in duplicate by enzyme-linked immunosorbent assays (ELISAs) using the following kits according to the manufacturers’ protocols: insulin (Crystal Chem, catalog # 90095), leptin (Crystal Chem, catalog # 80968), adiponectin (Crystal Chem, catalog #80571), AGE (MyBiosource, catalog #MBS267540), and RAGE (Biotechne, catalog #DRG00). Glucose and A1C were measured annually starting in 2007 at a commercial laboratory (Quest Diagnostics, Inc, Wood Dale, IL), with certification by the National Glycohemoglobin Standardization Program (NGSP), as previously published.32
Given that blood specimens were collected annually, we needed to determine how to systematically select specimens from a particular time point for research for this study. For participants with two or more aliquots of serum for a given visit (as we did not want to use the “last” aliquot available for a given visit), we selected the serum that was collected as proximate to death as possible, but within 4 years or less prior to death (excluding the last visit). More details on the selection are provided under the section on the selection of cases. The time interval from the serum sample used until death had a median of 1.9 years (IQR: 1.5 – 2.5 years), and a mean of 2.0 years (SD= 0.7).
For analytic purposes, the serum measures were classified into three categories. The first category was insulin and directly related serum measures, which included insulin, glucose, leptin, and adiponectin. The second category was glycation products, which included A1C, AGEs and RAGE. The third category was calculated measures, which included adiponectin-leptin ratio, a measure of risk of dementia and cardiovascular disease,19, 33–35 and the HOMA-IR, a measure of peripheral insulin resistance.13, 36 HOMA-IR was calculated using the following formula: HOMA-IR = insulin (mU/L) × glucose (mg/dL) / 405.
Pathologic Data
Systematic neuropathologic evaluations were conducted on each autopsied brain (median postmortem interval = 6.9 hours [inter-quartile range: 6.0–9.0 hours] in the parent MAP study), blinded to clinical data, as previously described,29 and common age-related neuropathologies were examined. A standardized global AD pathology measure using a modified silver stain summarized counts of neuritic plaques, diffuse plaques, and neuronal neurofibrillary tangles.37 Furthermore, immunohistochemical measures of global Aβ burden and tau tangle density were also available, including as a summary measure across different brain regions.38 Given that some AD pathology measures were not normally distributed, we transformed these data using the square root.
Neuropathologic also included systematic assessment for cerebrovascular disease, as previously published and described in detail elsewhere.39 In summary, gross (macroscopic) infarcts were identified on gross examination and confirmed on microscopic examination. Gross infarcts were classified by number, volume (in mm2), location, and age (chronic, subacute, acute). Microinfarcts, defined as not identified by the naked eye, but on histologic examination only, were recorded including their location and age.40 For this study, only chronic infarcts were considered, and categorical variables with two levels (present, as one or more infarcts; absent, the reference group) were used for documentation. Cerebral vessel pathologies including atherosclerosis, arteriolosclerosis, amyloid angiopathy were also identified and documented, and the severity was grouped into two levels for analyses (not present or mild; moderate or severe).
Selection of Cases with and without Diabetes
An algorithm using Mahalanobis Distance was used to select a matched sample of older persons with and without diabetes by sex and balanced on age (within 5 years), education (within 4 years), and postmortem interval (within 12 hours maximum, with a median of 6.5 hours [inter-quartile range: 5.7–8.2 hours] for brain specimens used in this study). The objective was to obtain a balanced sample of participants with and without diabetes, with a minimum multivariate distance between these key variables. We included 192 autopsied participants with available specimens of brain, muscle and serum (ongoing research on peripheral specimens), and complete data on essential clinical and pathologic measures of interest to this study. The sample included 96 with diabetes and 96 without diabetes (matched 1:1).
Statistical Approach
Initial statistical analyses included a detailed examination of the distribution and the correlation structure of the variables of interest, particularly insulin and related serum measures. As previously described, these measures were classified into 3 categories: direct serum measures (insulin, glucose, leptin, and adiponectin), glycation products (A1C, AGEs, and RAGE), and calculated measures (adiponectin-leptin ratio and HOMA-IR). In our subsequent analyses, to correct for multiple comparisons within each category, we adopted Bonferroni-corrected alpha values of 0.0125, 0.0166, and 0.025, respectively. First, we compared insulin and related serum measures between persons with and without diabetes using paired t-tests. Next, we examined the associations between these serum measures and postmortem neuropathologic data (brain infarcts and cerebral vessel pathologies; AD pathology). We fit separate logistic regression models with the cerebral infarcts (present vs not) and cerebral vessel pathology measures (severity was categorized as two levels: not present or mild, vs moderate or severe) as the outcome and the serum measures as the predictor. We fit separate linear regression models with AD pathology measures as the outcome and serum measures as the predictor. All models with neuropathology as the outcomes were controlled for age at death and sex. Last, we examined the associations between the serum measures and late-life cognition assessed at the same study visit, using separate linear regression models controlled for age at death, sex, and education. In all models, age and education were centered on their means for interpretation purposes. All analyses were conducted using SAS/STAT software, version 9.4 of the SAS system for Linux (SAS Institute, Cary, NC).
Results
Participant characteristics
Demographic and clinical characteristics of the 192 participants included in the study are shown in Table 1. The mean age at death was 89.9 years, and education 14.1 years; Two thirds (66%) were women, and 97% were non-Hispanic White. The characteristics were not different between the two groups of participants with and without diabetes. Among the 96 participants with diabetes, 63 (66%) used an oral hypoglycemic drug (including 63 on sulfonylureas, 38 on metformin, and 3 on meglitinides), and 22 (23%) used insulin.
TABLE 1.
Characteristics of Participants
| Characteristic | Total n= 192 |
With Diabetes n= 96 |
Without Diabetes n= 96 |
|||
|---|---|---|---|---|---|---|
| Demographic | ||||||
| Age at death, years (SD) | 89.9 | (5.7) | 89.5 | (6.0) | 90.3 | (5.4) |
| Women, n (%) | 126 | (66%) | 63 | (66%) | 63 | (66%) |
| White, Non-Hispanic, n (%) | 187 | (97%) | 91 | (95%) | 96 | (100%) |
| Education, years (SD) | 14.1 | (2.9) | 14.0 | (3.0) | 14.2 | (2.7) |
| Clinical | ||||||
| Stroke, n (%) | 41 | (22%) | 18 | (19%) | 23 | (24%) |
| History of hypertension, n (%) | 140 | (73%) | 80 | (83%) | 60 | (63%) |
| Systolic blood pressure, mmHg (SD) | 127.7 | (23.1) | 129.4 | (22.8) | 126.0 | (23.4) |
| Diastolic blood pressure, mmHg (SD) | 71.2 | (11.5) | 72.1 | (11.2) | 70.3 | (11.7) |
| Body mass index, kg/m2 (SD) | 26.2 | (4.9) | 27.4 | (4.9) | 25.1 | (4.6) |
Serum insulin and related measures
Table 2 shows the data on serum insulin and related measures, as well as the pairwise difference in the serum measures between persons with and without diabetes. There was no difference in levels of insulin, leptin, or adiponectin. As expected, persons with diabetes had higher levels of glucose and A1C (p < 0.001) than those without diabetes. In this dataset, we found that persons with diabetes had lower AGEs levels than those without diabetes (p = 0.007). Those with diabetes showed on average higher HOMA-IR (p = 0.002) and lower adiponectin-leptin ratio (p = 0.023) than those without, suggesting peripheral insulin resistance and adipose tissue dysfunction among persons with diabetes.
TABLE 2.
Insulin and Related Measures in Serum
| Median (interquartile range) | t statistic (p value) |
|||
|---|---|---|---|---|
| Total n= 192 |
With Diabetes n= 96 |
Without Diabetes n= 96 |
Pairwise Difference |
|
| Direct serum measures | ||||
| Insulin (mU/L) | 20.46 (9.96 – 39.58) | 26.09 (13.07 – 41.20) | 17.54 (8.72 – 36.16) | 1.36 (0.176) |
| Glucose (mg/dL) | 110.00 (89.50 – 148.00) | 137.50 (105.00 −189.50) | 96.50 (86.00 – 111.50) | 6.90 (<0.001) |
| Leptin (ng/mL) | 18.96 (8.45 – 50.85) | 25.11 (13.12 – 58.74) | 13.17 (7.31 – 35.75) | 1.71 (0.091) |
| Adiponectin (μg/mL) | 12.75 (7.41 – 21.71) | 11.25 (6.92 – 18.57) | 15.47 (8.25 – 22.93) | −2.15 (0.034) |
| Glycation products | ||||
| A1C * (%) | 6.00 (5.65 – 6.40) | 6.40 (6.00 – 7.00) | 5.75 (5.50 – 6.00) | 8.24 (<0.001) |
| AGEs ** (ng/mL) | 6.90 (3.78 – 10.56) | 5.95 (3.30 – 9.83) | 9.07 (4.77 – 11.58) | −2.75 (0.007) |
| RAGE ***(pg/mL) | 2024.97 (1463.97 – 2788.18) | 1101.36 (1302.54 – 2788.18) | 1990.35 (1549.46 – 2847.66) | −0.44 (0.662) |
| Calculated measures | ||||
| HOMA-IR | 6.05 (2.23 – 13.33) | 8.93 (3.54 – 17.80) | 4.23 (1.82 – 9.60) | 3.26 (0.002) |
| Adiponectin/ leptin | 0.56 (0.20 – 1.94) | 0.41 (0.20 – 1.01) | 0.93 (0.21 – 3.29) | −2.30 (0.023) |
A1C: Hemoglobin A1C; data available in 152/192 persons (76/96 with diabetes and 76/96 without diabetes)
AGEs: Advanced glycation-end products
RAGE: Receptor for advanced glycation-end products
Bold values denote statistical significance considering a Bonferroni adjusted alpha of 0.0125 for direct serum measures, 0.0166 for glycation products, and 0.025 for calculated measures.
Associations of the serum insulin and related measures with brain infarcts
Given that diabetes is associated with cerebrovascular disease and cerebrovascular disease is a common cause of cognitive dysfunction, we studied the associations of serum insulin and related measures with the presence of brain infarcts as the outcome, using separate age and sex adjusted logistic regression models. As shown in Table 3, we found that a higher HOMA-IR level was associated with an increased odds of having a brain infarct of any size or location (p = 0.01). Secondary analyses were done to examine infarcts by size (gross and microinfarcts, separately) and location (cortical and subcortical, separately). A higher level of HOMA-IR was associated with an increased odds of having a microinfarct. Serum insulin and related measures were not associated with presence of gross infarcts or cortical infarcts. A higher level of leptin and HOMA-IR were associated with an increased odds of having a subcortical infarct.
TABLE 3.
Association of Serum Insulin and Related Measures with Brain Infarcts*
| Predictors | OR, Estimate (SE, p) | ||||
|---|---|---|---|---|---|
| Any infarct | Size of infarct | Location of infarct | |||
| Gross infarcts | Micro-infarcts | Cortical infarcts | Subcortical infarcts | ||
| Direct serum measures | |||||
| Insulin | 1.011, 0.011 (0.005, 0.033) | 1.007, 0.007 (0.005, 0.151) | 1.009, 0.009 (0.005, 0.082) | 1.005, 0.005 (0.005, 0.380) | 1.013, 0.013 (0.005,0.015) |
| Glucose | 1.003, 0.003 (0.003, 0.305) | 1.002, 0.002 (0.003, 0.468) | 1.006, 0.005 (0.003, 0.059) | 1.002, 0.002 (0.003, 0.610) | 1.003, 0.003 (0.003, 0.224) |
| Leptin | 1.009, 0.009 (0.004, 0.033) | 1.009, 0.009 (0.004, 0.033) | 1.004, 0.004 (0.004, 0.348) | 1.001, 0.001 (0.004, 0.760) | 1.011, 0.011 (0.004, 0.011) |
| Adiponectin | 0.982, −0.018 (0.013, 0.175) | 0.978, −0.022 (0.014, 0.119) | 0.983, −0.017 (0.015, 0.261) | 0.988, −0.012 (0.015, 0.418) | 0.984, −0.016 (0.014, 0.241) |
| Glycation products | |||||
| A1C | 1.702, 0.532 (0.260, 0.040) | 1.683, 0.520 (0.242, 0.032) | 1.597, 0.468 (0.243, 0.054) | 1.481, 0.392 (0.239, 0.101) | 1.528, 0.424 (0.237, 0.074) |
| AGEs* | 1.014, 0.014 (0.022, 0.519) | 1.044, 0.043 (0.023, 0.056) | 0.982, −0.018 (0.025, 0.460) | 1.007, 0.007 (0.023, 0.770) | 1.026, 0.026 (0.022, 0.235) |
| RAGE** | 1.000, 0.000 (0.000, 0.037) | 1.000, 0.000 (0.000, 0.710) | 1.000, 0.000 (0.000, 0.087) | 1.000, 0.000 (0.000, 0.603) | 1.000, 0.000 (0.000, 0.248) |
| Calculated measures | |||||
| HOMA-IR | 1.035, 0.0346 (0.013, 0.010) | 1.017, 0.017 (0.012, 0.141) | 1.030, 0.029 (0.012, 0.018) | 1.008, 0.008 (0.012, 0.533) | 1.035, 0.034 (0.013, 0.007) |
| Adiponectin/ leptin | 0.944, −0.058 (0.040, 0.150) | 0.949, −0.053 (0.042, 0.213) | 0.932, −0.071 (0.054, 0.191) | 0.939, −0.063 (0.051, 0.210) | 0.943, −0.059 (0.045, 0.195) |
Separate age and sex adjusted logistic regression models, for each two-level infarct outcome measure (present; absent).
Bold values denote statistical significance considering a Bonferroni adjusted alpha of 0.0125 for direct serum measures, 0.0166 for glycation products, and 0.025 for calculated measures.
When we repeated these models after adding a term for the interaction between diabetes and each serum measure separately, no interaction was found (data not shown). This suggests that having diabetes did not affect the association of serum measures with brain infarcts.
Associations of the serum insulin and related measures with cerebral vessel pathologies
In order to better understand the relation of serum measures with cerebrovascular pathology, we next tested the associations of serum insulin and related measures with the severity of cerebral vessel pathologies. We found that higher leptin levels and lower adiponectin-leptin ratios were associated with an increased odds of having moderate or severe atherosclerosis in the brain (Table 4). There was no other association of serum measures with atherosclerosis, arteriolosclerosis, or amyloid angiopathy.
TABLE 4.
Association of Serum Insulin and Related Measures with Cerebral Vessel Pathologies*
| Predictors | OR, Estimate (SE, p) | ||
|---|---|---|---|
| Atherosclerosis | Arteriolosclerosis | Amyloid angiopathy | |
| Direct serum measures | |||
| Insulin | 0.996, −0.004 (0.006, 0.489) | 0.992, −0.008 (0.006, 0.169) | 1.001, 0.001 (0.005, 0.926) |
| Glucose | 1.005, 0.005 (0.003, 0.072) | 1.002, 0.002 (0.003, 0.433) | 1.001, 0.001 (0.003, 0.656) |
| Leptin | 1.011, 0.011 (0.004, 0.009) | 0.996, −0.004 (0.004, 0.323) | 1.005, 0.005 (0.004, 0.234) |
| Adiponectin | 0.984, −0.016 (0.016, 0.314) | 1.026, 0.025 (0.014, 0.066) | 0.972, −0.028 (0.016, 0.077) |
| Glycation products | |||
| A1C | 1.412, 0.345 (0.257, 0.179) | 0.825, −0.193 (0.270, 0.476) | 1.035, 0.034 (0.243, 0.888) |
| AGEs* | 1.006, 0.006 (0.023, 0.794) | 1.018, 0.018 (0.022, 0.412) | 0.994, −0.006 (0.023, 0.804) |
| RAGE** | 1.000, −0.000 (0.000, 0.379) | 1.000, 0.000 (0.000, 0.672) | 1.000, −0.000 (0.000, 0.962) |
| Calculated measures | |||
| HOMA-IR | 1.001, 0.001 (0.013, 0.941) | 0.987, −0.013 (0.014, 0.341) | 1.009, 0.009 (0.012, 0.470) |
| Adiponectin/leptin | 0.801, −0.222 (0.096, 0.021) | 0.988, −0.012 (0.024, 0.617) | 0.976, −0.025 (0.033, 0.455) |
Separate age and sex adjusted logistic regression models, for each two-level cerebral vessel pathology outcome measure (not present or mild; moderate or severe).
Bold values denote statistical significance considering a Bonferroni adjusted alpha of 0.0125 for direct serum measures, 0.0166 for glycation products, and 0.025 for calculated measures.
Further, there was no interaction effect between diabetes and the serum measures (data not shown), suggesting that having diabetes did not affect the association of serum measures with cerebral vessel pathologies.
Associations of serum insulin and related measures with AD pathology
We next examined the outcome of neurodegenerative pathology, specifically AD pathology assessed using a global AD score, as well as amyloid burden and tau tangle density pathologies (Table 5). Separate linear regression models adjusted for age-at-death and sex, included the predictors for serum insulin and related measures. Results were that none of the serum measures were significant predictors of AD pathology.
TABLE 5.
Associations of Serum Insulin and Related Measures with AD Neuropathology*
| Predictors | Estimate (SE), p value | ||
|---|---|---|---|
| Global AD score | Amyloid burden** | Tau tangle density** | |
| Direct serum measures | |||
| Insulin | 0.002 (0.002,0.143) | 0.003 (0.003,0.344) | 0.003 (0.003,0.375) |
| Glucose | −0.001 (0.001,0.136) | −0.003 (0.002,0.043) | −0.003 (0.002,0.091) |
| Leptin | 0.001 (0.001,0.416) | 0.002 (0.002,0.405) | 0.002 (0.002,0.427) |
| Adiponectin | −0.007 (0.004,0.083) | −0.012 (0.008,0.158) | −0.014 (0.008,0.080) |
| Glycation products | |||
| A1C | −0.026 (0.074,0.727) | −0.010 (0.1409,0.480) | −0.152 (0.136,0.265) |
| AGEs | −0.011 (0.007,0.107) | −0.015 (0.0131,0.256) | −0.004 (0.013,0.781) |
| RAGE | −0.000 (0.000,0.631) | 0.000 (0.000,0.661) | −0.000 (0.000,0.066) |
| Calculated measures | |||
| HOMA-IR | 0.001 (0.004,0.752) | −0.001 (0.007,0.845) | 0.000 (0.007,0.964) |
| Adiponectin/leptin | −0.007 (0.005,0.186) | −0.014 (0.011,0.201) | −0.014 (0.010,0.177) |
Separate linear regression models (one per row) adjusted for age-at-death and sex.
Note that amyloid burden and tau tangle density were transformed using the square root in these models.
Bold values denote statistical significance considering a Bonferroni adjusted alpha of 0.0125 for serum measures, 0.0166 for glycation products, and 0.025 for calculated measures.
Since the association of the serum measures with AD pathology may vary by diabetes status, we next conducted analyses in which we added an interaction term between diabetes and the serum measures. The interaction terms were not significant (data not shown, all p values >0.05), except for the interaction of diabetes by HOMA-IR, for which there was an association with lower levels of tau tangle density (estimate = −0.039, SE = 0.015, p = 0.010). To better understand this interaction, we performed a post hoc stratified analysis by diabetes status. Higher HOMA-IR levels were associated with higher levels of tau tangle density (estimate = 0.029, SE = 0.014, p = 0.045) in participants without diabetes. However, this association was not significant in those with diabetes (estimate= −0.009, SE= 0.007, p = 0.216).
Associations of serum insulin and related measures with cognition
To test whether insulin measures are associated with cognition, we conducted analyses using summary measures of cognition proximate to death as the outcomes (global score and five cognitive domains), using a series of linear regression models adjusting for age at serum visit, sex, and education. As shown in Table 6, we found that higher insulin and leptin levels, and lower adiponectin and RAGE levels were associated with lower levels of global cognition.
TABLE 6.
Association of Serum Insulin and Related Measures with Global Cognition and Five Cognitive Domains*
| Predictors | Estimate (SE, p value) | |||||
|---|---|---|---|---|---|---|
| Global Cognition | Episodic Memory | Semantic Memory | Working Memory | Perceptual Speed | Visuospatial Skills | |
| Direct serum measures | ||||||
| Insulin | −0.006 (0.002,0.004) | −0.007 (0.003,0.020) | −0.008 (0.002,0.000) | −0.003 (0.002,0.218) | −0.004 (0.003,0.141) | 0.000 (0.003,0.959) |
| Glucose | 0.001 (0.001,0.390) | 0.002 (0.002,0.246) | −0.000 (0.001,0.958) | −0.001 (0.001,0.547) | 0.002 (0.002,0.125) | 0.002 (0.002,0.195) |
| Leptin | −0.005 (0.002,0.004) | −0.006 (0.002,0.014) | −0.006 (0.002,0.001) | −0.005 (0.002,0.012) | −0.003 (0.002,0.079) | −0.004 (0.002,0.066) |
| Adiponectin | 0.014 (0.006,0.011) | 0.012 (0.007,0.105) | 0.014 (0.006,0.016) | 0.021 (0.006,0.000) | 0.012 (0.006,0.036) | 0.011 (0.006,0.091) |
| Glycation products | ||||||
| A1C | −0.033 (0.010,0.742) | 0.069 (0.136,0.615) | −0.100 (0.107,0.356) | −0.121 (0.107,0.257) | 0.005 (0.116,0.968) | −0.057 (0.121,0.639) |
| AGEs | −0.008 (0.009,0.388) | −0.008 (0.012,0.499) | −0.001 (0.010,0.905) | −0.011 (0.010,0.287) | −0.002 (0.011,0.834) | −0.021 (0.012,0.080) |
| RAGE | 0.000 (0.000,0.014) | 0.000 (0.000,0.017) | 0.000 (0.000,0.465) | 0.000 (0.000,0.007) | 0.000 (0.000,0.061) | 0.000 (0.000,0.767) |
| Calculated measures | ||||||
| HOMA -IR | −0.009 (0.005,0.084) | −0.009 (0.006,0.168) | −0.012 (0.005,0.018) | −0.007 (0.006,0.193) | −0.001 (0.006,0.938) | 0.007 (0.007,0.309) |
| Adiponectin/ Leptin | 0.013 (0.007,0.076) | 0.007 (0.009,0.464) | 0.013 (0.007,0.066) | 0.017 (0.008,0.027) | 0.017 (0.007,0.014) | 0.002 (0.008,0.772) |
Separate linear regression models adjusting for age at serum visit, sex, and education.
Bold values denote statistical significance considering a Bonferroni adjusted alpha of 0.0125 for direct serum measures, 0.0166 for glycation products, and 0.025 for calculated measures.
Cognitive data were from the same evaluation cycle that the serum sample was collected.
In secondary analyses, we examined the associations of the serum measures with individual cognitive domains (Table 6). We found that higher insulin, leptin, and HOMA-IR levels were associated with lower levels of semantic memory function. Higher leptin, and lower adiponectin and RAGE levels, were associated with lower levels of working memory. Lower adiponectin-leptin ratios were associated with lower levels of perceptual speed. We did not find any other association of serum measures with episodic memory, semantic memory, working memory, perceptual speed, and visuospatial skills.
We found an interaction between diabetes status and leptin levels in models with global cognition (estimate = 0.009, SE = 0.003, p = 0.005), working memory (estimate = 0.009, SE = 0.004, p = 0.012), and visuospatial skills (estimate = 0.011, SE = 0.004, p = 0.005) as the outcomes. Stratified analyses showed that in participants without diabetes, higher leptin was associated with lower global cognition (estimate = −0.009, SE = 0.003, p < 0.001), lower working memory (estimate = −0.008, SE = 0.002, p = 0.001), and lower visuospatial skills (estimate = −0.007, SE = 0.003, p = 0.004). There was no association of serum measures with cognition in participants with diabetes.
Discussion
In this clinical-pathologic study of nearly 200 older persons with and without diabetes, we found that serum insulin and related measures were associated with cerebrovascular pathology. Higher HOMA-IR was associated with the presence of infarcts, especially microinfarcts. Higher HOMA-IR and leptin were each associated with subcortical infarcts specifically. Examining vessel pathology outcomes, a higher leptin and lower adiponectin-leptin ratio were associated with more severe brain atherosclerosis. In additional analyses with neurodegenerative pathology outcomes, we did not find that serum insulin or related measures were associated with postmortem AD pathology. However, in secondary analyses, insulin resistance indicated by higher HOMA-IR, was associated with higher levels of tau tangle density in persons without diabetes. Finally, regarding cognitive outcomes, we found that higher leptin and insulin levels, and lower adiponectin and RAGE levels were associated with lower levels of global cognition. Taken together, our findings suggest that insulin resistance and adipose dysfunction in older persons with or without diabetes, are associated with cerebrovascular but not AD neuropathology, and with lower cognitive function.
Prior research has explored the relation of peripheral (most often blood) insulin and related measures to brain structure and function, but we are not aware of data derived from direct examination of human tissue as we have presented here with postmortem brain. Indeed, a number of studies have leveraged neuroimaging to examine blood markers such as insulin resistance by HOMA-IR and others, with brain MRI or PET markers of cerebrovascular disease or neurodegeneration, with mixed findings.19, 41–48 Brain MRI alterations, including those indicative of cerebral small vessel disease such as worse white matter hyperintensities (WMH) and brain infarcts, have been found to be associated with insulin resistance and related markers.45 While not all studies find such an association,44 intranasal insulin therapy may decrease WMH and improve cognition and cerebrospinal fluid (CSF) biomarkers of AD.46 And in recent developments, enhanced neuronal insulin signaling assessed from plasma extracellular vesicles, was associated with lower WMH and higher temporal lobe volume.47 Yet other studies show that peripheral insulin resistance is associated with less cortical gray matter volume in brain.48 Using another technology with brain PET imaging, the HOMA-IR was found to be associated with increased uptake of radiotracers targeting fibrillar beta-amyloid in older cognitively-normal individuals who were APOEɛ4/4 homozygote,41 but in another study without taking APOE into account, was not associated with increased uptake of radiotracers targeting tau.42 In our study, we did not find an association of serum markers with amyloid accumulation. Our A1C finding is consistent with a previous PET study which showed that elevated A1C was not associated with amyloid-β accumulation,49 but contrary to another study in which elevated fasting glucose was associated with faster amyloid-β accumulation in the Pittsburgh Compound-positive group.50 Future research is needed to further elucidate these disparate associations.
Our neuropathologic study extends knowledge on peripheral insulin resistance and related markers and brain in several ways. First, we analyzed postmortem tissue with pathologically proven cerebrovascular disease and AD pathology, rather than relying on in-vivo biomarkers of pathology such as by neuroimaging or CSF. Second, data allowed for a detailed evaluation of a range of neuropathologies, within both categories of cerebrovascular disease and AD pathology, and a range of blood insulin and related measures. For instance, while silent brain infarcts may be involved,45 we found more specifically that more peripheral insulin resistance (higher HOMA-IR) as well as higher leptin, were associated with subcortical infarcts. Further, we showed that higher leptin and lower adiponectin-leptin ratio were associated with the underling vessel pathology of atherosclerosis. Separately, we did not find clear evidence for a relation to AD pathology, but our secondary analyses suggest that higher HOMA-IR may also be associated with tau tangles in persons without diabetes. Perhaps, cerebrovascular processes such as brain hypoperfusion of watershed zones may play a role in increasing tau pathology.51, 52 Results of previous studies examining the association between T2DM and AD neuropathology are mixed.7–10 Several factors may contribute to the mixed results, including that different tools are used to analyze AD neuropathology and at different time points in the disease process (e.g., PET imaging, postmortem tissue, etc.), autopsy material are often derived from symptomatic/later stage AD patients, and analyses may not control for T2DM medications (with normalized blood glucose and serum insulin metrics). Thus, relationships between T2DM and AD neuropathology may be confounded. These relationships are complex, and support the idea that Aβ/tau are not the only driver of the association between T2DM and AD, and other mechanistic links need to be better understood. In summary, our data provide additional insights into the mechanisms linking insulin resistance to brain disease and dysfunction. But additional research will need to further elucidate these and other mechanisms involved in cognitive impairment and dementia, including the role of insulin resistance in brain itself.53–56
More broadly, several published studies examined brain function outcomes of cognition, in addition to in-vivo biomarkers. In a large sample of the UK Biobank without diabetes, A1C variability was associated with lower MRI hippocampal volume and increased dementia risk.43 In another recent but much smaller study, leptin was associated with lower MRI hippocampal volumes in those with obesity, and both leptin and a leptin-adiponectin ratio were inversely associated with cognition.19 Yet another recent study of more than 1,000 older persons, leptin was associated with CSF Aβ (inversely), as well as with AD confirmed by CSF biomarkers, but there was no relation to slope of cognitive decline on the Mini-Mental State Examination scores.57 While we did not examine for decline in cognition, we found that higher leptin and insulin levels, and lower adiponectin and RAGE levels were associated with lower levels of global cognition. Further, in additional analyses, we found relations of insulin resistance (insulin and HOMA-IR levels) and adipokines with lower function in some (memory) but not other cognitive domains. Thus, our findings suggest that insulin resistance and adipose dysfunction are associated with lower cognitive function, the prototypical clinical expression of common neuropathology in aging, cerebrovascular disease and AD.
Our study has several strengths. We assessed insulin and related measures including adipokines, AGEs, and RAGE, using robust ELISA techniques. We enhanced our serum panel with calculated measures including the widely used HOMA-IR for insulin resistance, and the adiponectin-leptin ratio which has been found to be associated with AD diagnosis. Further, we leveraged antemortem blood samples and postmortem brain tissue data from the same persons, to quantify measures of peripheral insulin resistance and related measures, and to relate these peripheral measures to central (brain) neuropathology. We also had detailed neuropathology, collected blinded to all clinical data, spanning cerebrovascular disease including different data on infarcts and three cerebral vessel pathologies, and neurodegeneration (with global AD and specific AD pathologies). Lastly, MAP participants undergo annual neuropsychological testing and cognitive evaluations (among other assessments), from which we constructed summary measures of global cognition and five different cognitive domains, from the same time point as the collection of the serum measures. This approach allows to minimize floor and ceiling effects with the cognitive data, and to elucidate cognitive systems potentially involved in insulin signaling pathways.
There are several noteworthy limitations to this study. First, the participants in our study were on average 90-year-old at death, predominantly white and highly educated, and may not be representative of the general population. Further, these and other factors introduce selection biases for resilience to disease. Second, the mean A1C of the non-diabetes group was 5.75%, suggesting that some individuals in this group had impaired glucose tolerance. While including such individuals in the control group would decrease the likelihood of finding a relation of insulin and related measures with outcomes, our study nonetheless found that peripheral insulin resistance (HOMA-IR) was associated with cerebrovascular neuropathology and cognition. Third, most blood collection was done in a non-fasting state to avoid potential hypoglycemia in participants who were from a community-based study of older and often frail persons who volunteered for research. A concerted effort is made in the parent MAP study to minimize participant burden and risk for complications (e.g., hypoglycemic episodes). Yet, published research has found that even non-fasting glucose levels are related to cognitive function.58 In addition, blood measures were collected at a single time point and with a variable time interval before death, with uncertain effects on the relation to outcomes of neuropathology. Fourth, we did not have a large enough sample size to examine the effects of diabetes medications on the relation of serum measures to neuropathology and cognitive outcomes. A prior study suggests insulin use is associated with a greater decline in global cognitive function possibly due to a greater risk of hypoglycemia.59 In that paper, there was no association between sulfonylurea or metformin use and cognitive outcomes. Yet, our recently published study shows that metformin use is associated with a slower decline in global cognitive function, as well as specifically with slower declines in episodic memory and semantic memory, and with a lower odds of cerebral vessel pathologies such as atherosclerosis60. Similarly, while previous studies suggest that serum leptin is associated with total body fat mass61 and that higher levels of body fat mass are positively associated with better cognitive performance,62 too few of our 192 participants had body fat data to conduct analyses. Future research is needed to elucidate the association between body fat composition and neuropathology. Further analyses of hormonal and other variables among men versus women, and with larger samples, are also needed to help clarify these and other questions, and unexpected results such as serum AGEs being higher among the non-diabetes group compared to the diabetes group (possibly due to impaired glucose tolerance in some individuals without diabetes, blood collection in non-fasting state, or other factors). Fifth, effect sizes, especially for cognitive outcomes, were small and the clinical significance remains uncertain. Nonetheless, despite correction for multiple comparisons and the small group of about 200 persons in the study, we found significant relationships with cerebrovascular disease and cognition. Last, analyses were cross-sectional and cannot determine causality.
Acknowledgements
The authors sincerely thank all participants in the Rush Memory and Aging Project for their altruism and commitment to this ongoing study since 1997. The authors also thank all Rush Alzheimer’s Disease Center staff and faculty, in particular Traci Colvin for study coordination, Ryan Johnson for laboratory management, John Gibbons for data management, and Donna Esbjornson for statistical programming.
This work was funded by the National Institute on Aging, and National Institute of Neurological Disorders and Stroke grants: P30AG10161, P30AG072975, R01AG017917, R01NS084965, RF1AG059621, and RF1AG074549.
Footnotes
Potential Conflicts of Interest
The authors have no conflicts of interest to declare.
Data Availability
All data supporting the conclusions of this article can be requested for research purposes via the Rush Alzheimer’s Disease Center Research Resource Sharing Hub at radc.rush.edu.
References
- 1.Klein S, Gastaldelli A, Yki-Jarvinen H, Scherer PE. Why does obesity cause diabetes? Cell Metab 2022;34(1):11–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Xue M, Xu W, Ou Y, et al. Diabetes mellitus and risks of cognitive impairment and dementia: A systematic review and meta-analysis of 144 prospective studies. Ageing Res Rev 2019;55:100944. [DOI] [PubMed] [Google Scholar]
- 3.Gudala K, Bansal D, Schifano F, Bhansali A. Diabetes mellitus and risk of dementia: A meta-analysis of prospective observational studies. J Diabetes Investig 2013;4(6):640–650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Arvanitakis Z, Wilson RS, Bennett DA. Diabetes mellitus, dementia, and cognitive function in older persons. J Nutr Health Aging 2006;10(4):287–291. [PubMed] [Google Scholar]
- 5.Chen R, Ovbiagele B, Feng W. Diabetes and stroke: Epidemiology, pathophysiology, pharmaceuticals and outcomes. Am J Med Sci 2016;351(4):380–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zheng B, Su B, Udeh-Momoh C, et al. Associations of cardiovascular and non-cardiovascular comorbidities with dementia risk in patients with diabetes: Results from a large UK cohort study. J Prev Alzheimers Dis 2022;9(1):86–91. [DOI] [PubMed] [Google Scholar]
- 7.Ahtiluoto S, Polvikoski T, Peltonen M, et al. Diabetes, Alzheimer disease, and vascular dementia: A population-based neuropathologic study. Neurology 2010;75(13):1195–1202. [DOI] [PubMed] [Google Scholar]
- 8.Abner EL, Nelson PT, Kryscio RJ, et al. Diabetes is associated with cerebrovascular but not Alzheimer’s disease neuropathology. Alzheimers Dement 2016;12(8):882–889. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Arvanitakis Z, Schneider JA, Wilson RS, et al. Diabetes is related to cerebral infarction but not to AD pathology in older persons. Neurology 2006;67(11):1960–1965. [DOI] [PubMed] [Google Scholar]
- 10.Heitner J, Dickson D. Diabetics do not have increased Alzheimer-type pathology compared with age-matched control subjects. A retrospective postmortem immunocytochemical and histofluorescent study. Neurology 1997;49(5):1306–1311. [DOI] [PubMed] [Google Scholar]
- 11.Kahn BB, Flier JS. Obesity and insulin resistance. J Clin Invest 2000;106(4):473–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Moller DE, Flier JS. Insulin resistance--mechanisms, syndromes, and implications. N Engl J Med 1991;325(13):938–948. [DOI] [PubMed] [Google Scholar]
- 13.Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care 2004;27(6):1487–1495. [DOI] [PubMed] [Google Scholar]
- 14.Willette AA, Bendlin BB, Starks EJ, et al. Association of insulin resistance with cerebral glucose uptake in late middle-aged adults at risk for Alzheimer disease. JAMA Neurol 2015;72(9):1013–1020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Laws SM, Gaskin S, Woodfield A, et al. Insulin resistance is associated with reductions in specific cognitive domains and increases in CSF tau in cognitively normal adults. Sci Rep 2017;7(1):9766–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Knights AJ, Funnell AP, Pearson RC, et al. Adipokines and insulin action: A sensitive issue. Adipocyte 2014;3(2):88–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kwon H, Pessin JE. Adipokines mediate inflammation and insulin resistance. Front Endocrinol (Lausanne) 2013;4:71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Feinkohl I, Janke J, Slooter AJC, et al. Plasma leptin, but not adiponectin, is associated with cognitive impairment in older adults. Psychoneuroendocrinology 2020;120:104783. [DOI] [PubMed] [Google Scholar]
- 19.Ha J, Kwak S, Kim KY, et al. Relationship between adipokines, cognition, and brain structures in old age depending on obesity. J Gerontol A Biol Sci Med Sci 2023;78(1):120–128. [DOI] [PubMed] [Google Scholar]
- 20.Letra L, Matafome P, Rodrigues T, et al. Association between adipokines and biomarkers of Alzheimer’s disease: A cross-sectional study. J Alzheimers Dis 2019;67(2):725–735. [DOI] [PubMed] [Google Scholar]
- 21.van Himbergen TM, Beiser AS, Ai M, et al. Biomarkers for insulin resistance and inflammation and the risk for all-cause dementia and Alzheimer disease: Results from the Framingham Heart Study. Arch Neurol 2012;69(5):594–600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Unoki H, Yamagishi S. Advanced glycation end products and insulin resistance. Curr Pharm Des 2008;14(10):987–989. [DOI] [PubMed] [Google Scholar]
- 23.Singh R, Barden A, Mori T, Beilin L. Advanced glycation end-products: A review. Diabetologia 2001;44(2):129–146. [DOI] [PubMed] [Google Scholar]
- 24.Unoki H, Bujo H, Yamagishi S, et al. Advanced glycation end products attenuate cellular insulin sensitivity by increasing the generation of intracellular reactive oxygen species in adipocytes. Diabetes Res Clin Pract 2007;76(2):236–244. [DOI] [PubMed] [Google Scholar]
- 25.Beeri MS, Uribarri J, Cai W, et al. Human brain and serum advanced glycation end products are highly correlated: Preliminary results of their role in Alzheimer disease and type 2 diabetes. Endocr Pract 2020;26(5):576–577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chen J, Mooldijk SS, Licher S, et al. Assessment of advanced glycation end products and receptors and the risk of dementia. JAMA Netw Open 2021;4(1):e2033012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lotan R, Ganmore I, Livny A, et al. Effect of advanced glycation end products on cognition in older adults with type 2 diabetes: Results from a pilot clinical trial. J Alzheimers Dis 2021;82(4):1785–1795. [DOI] [PubMed] [Google Scholar]
- 28.West RK, Moshier E, Lubitz I, et al. Dietary advanced glycation end products are associated with decline in memory in young elderly. Mech Ageing Dev 2014;140:10–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Bennett DA, Buchman AS, Boyle PA, et al. Religious Orders Study and Rush Memory and Aging Project. J Alzheimers Dis 2018;64(s1):S161–S189. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wilson RS, Beckett LA, Barnes LL, et al. Individual differences in rates of change in cognitive abilities of older persons. Psychol Aging 2002;17(2):179–193. [PubMed] [Google Scholar]
- 31.Arvanitakis Z, Wilson RS, Bienias JL, et al. Diabetes mellitus and risk of Alzheimer disease and decline in cognitive function. Arch Neurol 2004;61(5):661–666. [DOI] [PubMed] [Google Scholar]
- 32.Pruzin JJ, Schneider JA, Capuano AW, et al. Diabetes, hemoglobin A1C, and regional Alzheimer disease and infarct pathology. Alzheimer Dis Assoc Disord 2017;31(1):41–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Lopez-Vilaret KM, Cantero JL, Fernandez-Alvarez M, et al. Impaired glucose metabolism reduces the neuroprotective action of adipocytokines in cognitively normal older adults with insulin resistance. Aging (Albany NY) 2021;13(21):23936–23952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Kaze AD, Musani SK, Bidulescu A, et al. Plasma adipokines and glycaemic progression among african americans: Findings from the jackson heart study. Diabet Med 2021;38(5):e14465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Senkus KE, Crowe-White KM, Bolland AC, et al. Changes in adiponectin:leptin ratio among older adults with obesity following a 12-month exercise and diet intervention. Nutr Diabetes 2022;12(1):30–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Woodfield A, Porter T, Gilani I, et al. Insulin resistance, cognition and Alzheimer’s disease biomarkers: Evidence that CSF Abeta42 moderates the association between insulin resistance and increased CSF tau levels. Neurobiol Aging 2022;114:38–48. [DOI] [PubMed] [Google Scholar]
- 37.Schneider JA, Boyle PA, Arvanitakis Z, et al. Subcortical infarcts, Alzheimer’s disease pathology, and memory function in older persons. Ann Neurol 2007;62(1):59–66. [DOI] [PubMed] [Google Scholar]
- 38.Arvanitakis Z, Schneider JA, Wilson RS, et al. Diabetes is related to cerebral infarction but not to AD pathology in older persons. Neurology 2006;67(11):1960–1965. [DOI] [PubMed] [Google Scholar]
- 39.Arvanitakis Z, Capuano AW, Leurgans SE, et al. Relation of cerebral vessel disease to Alzheimer’s disease dementia and cognitive function in elderly people: A cross-sectional study. Lancet Neurol 2016;15(9):934–943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Arvanitakis Z, Leurgans SE, Barnes LL, et al. Microinfarct pathology, dementia, and cognitive systems. Stroke 2011;42(3):722–727. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ekblad LL, Tuisku J, Koivumaki M, et al. Insulin resistance and body mass index are associated with TSPO PET in cognitively unimpaired elderly. J Cereb Blood Flow Metab 2023;43(9):1588–1600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ennis GE, Betthauser TJ, Koscik RL, et al. The relationship of insulin resistance and diabetes to tau PET SUVR in middle-aged to older adults. Alzheimers Res Ther 2023;15(1):55–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Li C, Guo J, Zhao Y, et al. Visit-to-visit HbA1c variability, dementia, and hippocampal atrophy among adults without diabetes. Exp Gerontol 2023;178:112225. [DOI] [PubMed] [Google Scholar]
- 44.Toppala S, Ekblad LL, Lotjonen J, et al. Midlife insulin resistance as a predictor for late-life cognitive function and cerebrovascular lesions. J Alzheimers Dis 2019;72(1):215–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Nam K, Kwon H, Jeong H, et al. High triglyceride-glucose index is associated with subclinical cerebral small vessel disease in a healthy population: A cross-sectional study. Cardiovasc Diabetol 2020;19(1):53–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Kellar D, Lockhart SN, Aisen P, 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(3):240–248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Walker KA, Chawla S, Nogueras-Ortiz C, et al. Neuronal insulin signaling and brain structure in nondemented older adults: The Atherosclerosis Risk in Communities Study. Neurobiol Aging 2021;97:65–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Lu R, Aziz NA, Diers K, et al. Insulin resistance accounts for metabolic syndrome-related alterations in brain structure. Hum Brain Mapp 2021;42(8):2434–2444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Roberts RO, Knopman DS, Cha RH, et al. Diabetes and elevated hemoglobin A1c levels are associated with brain hypometabolism but not amyloid accumulation. J Nucl Med 2014;55(5):759–764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Gomez G, Beason-Held LL, Bilgel M, et al. Metabolic syndrome and amyloid accumulation in the aging brain. J Alzheimers Dis 2018;65(2):629–639. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Austin BP, Nair VA, Meier TB, et al. Effects of hypoperfusion in Alzheimer’s disease. J Alzheimers Dis 2011;26 Suppl 3(Suppl 3):123–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Kapasi A, Yu L, Petyuk V, et al. Association of small vessel disease with tau pathology. Acta Neuropathol 2022;143(3):349–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Rhea EM, Leclerc M, Yassine HN, et al. State of the science on brain insulin resistance and cognitive decline due to Alzheimer’s disease. Aging Dis 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Arvanitakis Z, Wang H, Capuano AW, et al. Brain insulin signaling, Alzheimer disease pathology, and cognitive function. Ann Neurol 2020;88(3):513–525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Arvanitakis Z, Capuano AW, Wang H, et al. Brain insulin signaling and cerebrovascular disease in human postmortem brain. Acta Neuropathol Commun 2021;9(1):71–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Yarchoan M, Toledo JB, Lee EB, et al. Abnormal serine phosphorylation of insulin receptor substrate 1 is associated with tau pathology in Alzheimer’s disease and tauopathies. Acta Neuropathol 2014;128(5):679–689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Lilamand M, Bouaziz-Amar E, Dumurgier J, et al. Plasma leptin is associated with amyloid CSF biomarkers and Alzheimer’s disease diagnosis in cognitively impaired patients. J Gerontol A Biol Sci Med Sci 2023;78(4):645–652. [DOI] [PubMed] [Google Scholar]
- 58.Crane PK, Walker R, Hubbard RA, et al. Glucose levels and risk of dementia. N Engl J Med 2013;369(6):540–548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Weinstein G, Davis-Plourde KL, Conner S, 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(2):e0212293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Sood A, Capuano AW, Wilson RS, et al. Metformin, age-related cognitive decline, and brain pathology. Neurobiol Aging 2024;133:99–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Shimizu H, Shimomura Y, Hayashi R, et al. Serum leptin concentration is associated with total body fat mass, but not abdominal fat distribution. Int J Obes Relat Metab Disord 1997;21(7):536–541. [DOI] [PubMed] [Google Scholar]
- 62.Tikhonoff V, Casiglia E, Guidotti F, et al. Body fat and the cognitive pattern: A population-based study. Obesity (Silver Spring) 2015;23(7):1502–1510. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data supporting the conclusions of this article can be requested for research purposes via the Rush Alzheimer’s Disease Center Research Resource Sharing Hub at radc.rush.edu.