Metabolic Syndrome and Cognitive Trajectories in the Framingham Offspring Study

Abstract

Metabolic syndrome (MetS) has been linked to increased risk of developing cognitive impairment and dementia including Alzheimer’s disease. It remains unclear whether and at what stage in the adult lifespan MetS and its components begin to alter the trajectory of cognitive performance. In the present study, 2,892 Framingham Offspring participants completed health assessments every four years since 1971 and underwent repeat neuropsychological testing from 1999 to 2014. We estimated the associations of levels and changes in cognitive trajectories with hazard of MetS using a joint growth/survival model. All models were adjusted for baseline age, sex, education, and smoking status. Findings showed that both mid-life and late-life MetS were associated with lower level of cognitive functioning but not cognitive trajectories. Associations were strongest among those who were nondemented and apolipoprotein (APOE) ɛ4 noncarriers. In addition, individuals with the most rapid cognitive decline were more likely to have MetS. The pattern of results showed that associations between MetS and cognition varied, depending upon whether the sample was stratified by genetic and cognitive status and whether we considered cognitive performance as a continuous variable or examined categorical groupings. Given that mid-life MetS was associated with poorer cognition at age 55, cognitive changes may occur early during the MetS process. Our findings suggest that those with MetS are at greater risk of dementia given their lower level of cognitive functioning and also suggest that MetS may be a risk factor for decline in the absence of known risk factors including the APOE ɛ4 allele.

Keywords

Aging Alzheimer’s disease apolipoprotein E cognition diabetes metabolic syndrome neuropsychology vascular risk

INTRODUCTION

Metabolic syndrome (MetS) consists of a cluster of interrelated risk factors that increases risk of cardiovascular disease and stroke, and includes abdominal obesity, dyslipidemia, hypertension, and hyperglycemia [1, 2]. Previous studies have shown that the risk of cardiovascular and cerebrovascular morbidity and mortality linked to MetS is greater than the risk associated with the individual MetS components [3 –5]. Several studies have also reported that the presence of MetS is a risk factor for developing mild cognitive impairment [6], Alzheimer’s disease (AD) [7, 8], and vascular dementia [9 –11].

Studies examining the association between MetS and cognitive performance across domains including memory and executive functioning have reported mixed results [12]. Cross-sectional studies have generally reported a link between MetS and poor cognition although some studies have shown no association [13] or shown that MetS is associated with better cognition [14, 15]. Similarly, longitudinal studies have reported both accelerated [16, 17] and decelerated cognitive decline with MetS [18]. Discrepant findings across studies may also be related to heterogeneity in the participant samples in terms of demographics including age; limited length of follow-up; use of different cognitive testing measures and reliance on cognitive screening measures [19, 20], which have been criticized for poor sensitivity [21, 22]; and study design including statistical modeling approach [12].

Given that existing data is limited and conflicting, it remains unclear whether and at what stage in the adult lifespan MetS and its components begin to alter the trajectory of normal cognitive function. A better understanding of the role of MetS in influencing this trajectory may have implications for identifying those individuals at increased risk for cognitive impairment and dementia and indicate for whom and when management of these risk factors during adulthood may be particularly helpful in delaying onset of cognitive difficulties. Using data from the Framingham Heart Study, we aimed to determine how MetS and its components affect the level and trajectory of cognitive functioning in this well-characterized cohort of community-dwelling adults. We hypothesized that MetS would relate to lower cognitive level and steeper cognitive decline. To our knowledge, no studies to date have examined whether MetS and its components are associated with within-person changes in cognitive function using as large an age range (i.e., 40–80+), as long of a follow-up period (i.e., up to 41 years for metabolic risk factors and up to 13 years with cognitive testing), and repeated comprehensive neuropsychological assessment.

MATERIALS AND METHODS

Participants

The Framingham Heart Study (FHS) is a community-based prospective cohort study began in 1948 to identify risk factors for cardiovascular disease. In the current study, participants were members of the Offspring Cohort, which includes biological children of the original FHS cohort and spouses of offspring (n = 5,124) who have undergone health examinations approximately every 4 years since 1971 [23]. The present analysis is based on the 2,892 Offspring participants who underwent neuropsychological assessments starting at Exam 7 in 1999. Follow-up data up to 2014 (Exam 9) were included in these analyses. The protocol was approved by the Institutional Review Board of Boston University Medical Center, and all participants provided written informed consent.

Neuropsychological assessment

Standardized neuropsychological tests were administered in three separate waves of testing. For this study, we constructed factor scores from the neuropsychological test battery to represent general cognitive performance, processing speed/executive function, and memory [24 –31]. In brief, factor scores for each domain were estimated from a 2-parameter logistic graded response item response theory model of the neuropsychological test battery [32], which was consistently administered over time. To construct the memory factor, we used Wechsler Memory Scale (WMS) Logical Memory immediate recall, delayed recall, and recognition; WMS Visual Reproduction immediate recall, delayed recall, and recognition; and WMS Verbal Paired Associates total immediate and delayed recall, immediate and delayed recall of easy pairs, and immediate and delayed recall of hard pairs [33]. Trail Making Tests A and B [34], WMS Digit Span Backward, Wechsler Adult Intelligence Scale (WAIS) Similarities subtest [35], Controlled Word Association Test (FAS), and Category Fluency (Animals) contributed to the processing speed/executive function factor. All the above variables together with WMS Paired Associates total learning, learning of easy pairs, learning of hard pairs, and delayed recall of hard pairs; WMS Digit Span Forward; Hooper Visual Organization Test (total correct); and 30-item Boston Naming Test [36] (number correct without cues), were used to create the composite for general cognition. We scaled each composite factor to have a mean of 50 and standard deviation (SD) of 10.

MetS classification

MetS was classified at each FHS exam using available data, according to National Cholesterol Education Program–Adult Treatment Panel III (NCEP ATP III) criteria [37] based on the presence of three or more of the following: 1) abdominal obesity (waist circumference > 102 cm in men or > 88 cm in women, or when waist circumference was not available, body mass index > 30 kg/m²); 2) elevated plasma triglycerides (≥150 mg/dl); 3) low high-density lipoprotein cholesterol (<40 mg/dL in men or<50 mg/dL in women); 4) high blood pressure (systolic/diastolic blood pressure≥130 mm Hg/≥85 mm Hg or use of antihypertensive medication); and 5) elevated blood glucose/hyperglycemia (elevated fasting plasma glucose≥100 mg/dL; hemoglobin A1c≥5.7%; or self-reported history of type 2 diabetes or diabetic treatment). Although the NCEP ATP III criteria suggested a cutpoint of≥110 mg/dL to define elevated blood glucose, we used a cutpoint of≥100 mg/dL, given that The American Diabetes Association later established≥100 mg/dL as the cutpoint above which persons have either prediabetes (impaired fasting glucose) or diabetes [38]. This cutpoint has been recommended to be applied in defining elevated glucose as a criterion for MetS [2]. MetS was set to missing for records with missing data on 3 or more of the above components. Finally, we considered a participant who ever met criteria for MetS to have MetS through the duration of their follow-up.

Statistical analyses

Participant characteristics

We first characterized the sample using means and percentages and examined potential differences between participants with and without MetS using t-tests for continuous variables and chi-squared tests for categorical variables.

Associations of time to onset of MetS with cognitive trajectories

We estimated the levels and changes in cognitive trajectories with time to incident MetS using joint growth/survival models [39] adjusted by baseline age, sex, years of education, and smoking status (i.e., smoker versus nonsmoker) at the first FHS exam. These models allow us to test the effect of the survival process (i.e., time to onset of MetS) on baseline levels (at age 55) and rates of change in cognitive performance. Cognitive trajectories were estimated using linear mixed effects models including random effects for people and time. Higher scores on the cognitive factors (general cognitive performance, processing speed/executive function, memory) reflect better performance and shallower (i.e., less negative) slopes represent less decline. We used age as the time scale of interest, and included a fixed quadratic term to accommodate nonlinearity in cognitive trajectories. In addition to overall MetS, we also tested each individual component of MetS. We used maximum likelihood estimation with robust standard error estimation under the EM algorithm in Mplus software [40].

In secondary analyses, we examined the associations of time to MetS with cognitive trajectories, stratified by APOE ɛ4 status (i.e., carrier versus non-carrier) and by dementia status (i.e., ever diagnosed with all-cause dementia versus remained nondemented during the course of the study). We did not perform analyses examining specific types of dementia (e.g., AD) given that the majority (i.e., over 70%) of individuals diagnosed with dementia were diagnosed with dementia due to AD and, therefore, there were few participants with non-AD forms of dementia.

Association of mid-life and late-life binary MetS status with cognitive trajectories

To confirm results using a simpler modeling approach, we estimated the levels and changes in cognitive trajectories concurrently with binary MetS status (present versus absent) at ages 55 (mid-life) and 70 (late-life), using separate latent growth curve models. For the model in which MetS was ascertained at age 55 years, the cognitive intercept is at age 55. Similarly, for the model in which MetS was ascertained at age 70 years, the cognitive intercept is at 70 years.

Association of MetS with rapid cognitive decline

We also performed analyses to determine whether presence of MetS distinguished between two extreme groups: individuals showing rapid decline and individuals showing slow decline or stable cognition. For this analysis, we limited the dataset to participants who completed at least four study visits with cognitive testing and estimated random intercepts and slopes from a random effects model. We identified the 30% of the sample with the slowest decline and the 30% of the sample with the most rapid decline and compared these groups on demographics and incidence of MetS and its components.

Sensitivity analyses

To evaluate whether our results were driven by those individuals that were lost to follow-up versus those who remained in the study, we re-ran our main analyses using a subsample of people who had MetS data available at both ages 55 and 70 years.

RESULTS

Participant characteristics

Table 1 show participant demographics and baseline characteristics for the total sample and as a function of MetS status. The sample was, on average, 34.8 years of age at baseline MetS assessment. There were no cases of prevalent dementia at Exam 1. The analysis included 68,086.5 person-years, with mean follow-up time of 36.8 years (SD = 3.9).

Table 1

Characteristics for entire study sample and according to MetS status

	Total sample (n = 2,892)	Without MetS (n = 867)	With MetS* (n = 2025)	p
Demographic and Clinical Characteristics
Age at baseline MetS assessment, mean (SD)	34.8 (9.2)	32.2 (9.2)	35.9 (9.0)	<0.001
Age at first neuropsychological assessment, mean (SD)	62.3 (9.7)	59.6 (9.9)	63.5 (9.4)	<0.001
Education Group, n (%)				<0.001
<High school degree	110 (3.8)	17 (2.0)	93 (4.6)
High school degree	766 (26.5)	193 (22.3)	573 (28.4)
Some college	749 (25.9)	219 (25.3)	530 (26.2)
≥College degree	1260 (43.7)	436 (50.4)	824 (40.8)
Women, n (%)	1553 (53.7)	532 (61.4)	1021 (50.4)	<0.001
Baseline smoker, n (%)	1682 (60.3)	450 (55.1)	1232 (62.5)	<0.001
APOEɛ4 carrier, n (%)	638 (22.9)	172 (21.1)	466 (23.7)	0.127
Ever had (all cause) dementia, n (%)	180 (6.2)	48 (5.5)	132 (6.5)	0.316
Ever had Alzheimer’s disease, n (%)	130 (4.5)	33 (3.9)	97 (4.9)	0.242
Years in study, mean (SD)	36.8 (3.9)	36.9 (3.8)	36.8 (3.9)	0.269
Number of study visits, mean (SD)	10.1 (2.2)	9.8 (2.4)	10.3 (2.1)	<0.001
Years in study with cognition, mean (SD)	5.5 (4.1)	5.5 (4.1)	5.6 (4.2)	0.818
Number of study visits with cognition, mean (SD)	2.2 (1.2)	2.2 (1.1)	2.3 (1.2)	0.051
MetS/MetS Components, n (%)
Ever had MetS	2025 (70.0)	0 (0.0)	2025 (100.0)
Earliest age when MetS observed, mean (SD)	53.5 (13.5)		53.5 (13.5)
Ever had abdominal obesity	2182 (75.5)	399 (46.1)	1783 (88.0)	<0.001
Ever had elevated triglycerides	1596 (55.6)	151 (17.7)	1445 (71.6)	<0.001
Ever had low HDL cholesterol	2187 (75.9)	399 (46.6)	1788 (88.4)	<0.001
Ever had high blood pressure	2414 (83.5)	499 (57.6)	1915 (94.6)	<0.001
Ever had elevated fasting glucose	2608 (90.8)	643 (75.4)	1965 (97.3)	<0.001
Cognitive Function**, mean (SD)
General Cognitive Performance	54.0 (7.5)	55.5 (7.4)	53.4 (7.5)	<0.001
Memory	53.9 (8.2)	55.3 (8.2)	53.3 (8.1)	<0.001
Processing Speed/Executive Functioning	53.8 (6.8)	55.0 (6.3)	53.3 (7.0)	<0.001

*Ever met criteria for MetS during the entire study period. **Cognitive scores reflect performance at first cognitive assessment and are internally scaled within the Framingham Offspring Study. Note: There was missing data for APOE genotype (n = 111), education (n = 7), baseline smoking (n = 104), MetS status (n = 2), abdominal obesity (n = 687), elevated triglycerides (n = 1,251), low HDL cholesterol (n = 677), high blood pressure (n = 460), elevated fasting blood glucose (n = 278), age at metabolic system first observed (n = 867), memory (n = 6), and executive function (n = 8).

Seventy percent of the sample (n = 2025) developed MetS over the course of the study, and 53.5 years was the mean age when MetS was first observed. Compared to those without MetS, individuals with MetS were somewhat older at baseline (i.e., at Exam 1, which was conducted 1971–1975) and first neuropsychological assessment, more likely to be male, completed less education, have more study visits, and poorer mean cognitive performance at first neuropsychological assessment.

Associations of time to MetS with cognitive trajectories across the entire sample

Table 2 shows the level and slope of cognitive trajectories for each cognitive factor (general cognitive performance, executive function, and memory), and their associations with onset of MetS from joint survival/growth models. There were no associations of onset of overall MetS and levels and change in any of the cognitive factor scores. When the individual components of MetS were examined separately, faster time to incident abdominal obesity was associated with lower levels of general cognitive performance (β= –0.671, 95% CI: –1.230, –0.112), executive function (β= –0.551, 95% CI: –1.086, –0.016), and memory (β= –0.671, 95% CI: –1.332, –0.010). None of the other individual components were associated with cognitive level. Furthermore, timing of components was not associated with cognitive trajectories.

Table 2

Joint growth/survival models for the association of time to MetS and its components with cognitive trajectories

Cognitive factor	MetS – cognitive slope	95% CI	MetS – cognitive level	95% CI	Cognitive level	95% CI	Cognitive slope	95% CI
MetS
General Cognitive Performance	–0.018	–0.063, 0.027	–0.564	–1.154, 0.026	51.402	50.212, 52.592	–0.371	–0.461, –0.281
Speed/Executive Function	–0.035	–0.082, 0.012	–0.569	–1.141, 0.003	53.245	52.073, 54.417	–0.461	–0.555, –0.367
Memory	0.000	–0.051, 0.051	–0.652	–1.344, 0.040	50.320	48.936, 51.704	–0.306	–0.406, –0.206
Abdominal Obesity
General Cognitive Performance	0.002	–0.041, 0.045	–0.671	–1.230, –0.112	51.221	50.047, 52.395	–0.374	–0.462, –0.286
Speed/Executive Function	–0.014	–0.057, 0.029	–0.551	–1.086, –0.016	53.054	51.911, 54.197	–0.468	–0.560, –0.376
Memory	0.018	–0.031, 0.067	–0.671	–1.332, –0.010	50.111	48.747, 51.475	–0.305	–0.403, –0.207
Elevated Triglycerides
General Cognitive Performance	–0.011	–0.060, 0.038	–0.423	–1.089, 0.243	51.337	50.155, 52.519	–0.373	–0.463, –0.283
Speed/Executive Function	–0.026	–0.079, 0.027	–0.508	–1.194, 0.178	53.200	52.044, 54.356	–0.464	–0.556, –0.372
Memory	0.017	–0.038, 0.072	–0.685	–1.451, 0.081	50.293	48.923, 51.663	–0.310	–0.410, –0.210
Low HDL Cholesterol
General Cognitive Performance	–0.026	–0.073, 0.021	0.018	–0.594, 0.630	51.211	50.027, 52.395	–0.369	–0.459, –0.279
Speed/Executive Function	–0.030	–0.081, 0.021	–0.233	–0.845, 0.379	53.078	51.933, 54.233	–0.461	–0.553, –0.369
Memory	–0.014	–0.065, 0.037	0.085	–0.638, 0.808	50.099	48.725, 51.473	–0.302	–0.402, –0.202
High Blood Pressure
General Cognitive Performance	–0.046	–0.101, 0.009	–0.257	–0.896, 0.382	51.359	50.146, 52.572	–0.362	–0.454, –0.270
Speed/Executive Function	–0.037	–0.094, 0.020	–0.447	–1.082, 0.188	53.257	52.040, 54.474	–0.459	–0.555, –0.363
Memory	–0.053	–0.114, 0.008	–0.195	–0.948, 0.558	50.253	48.854, 51.652	–0.293	–0.395, –0.191
Elevated Fasting Glucose
General Cognitive Performance	–0.051	–0.108, 0.006	–0.017	–0.721, 0.687	51.244	50.050, 52.438	–0.364	–0.454, –0.274
Speed/Executive Function	–0.055	–0.114, –0.004	0.112	–0.601, 0.825	53.043	51.873, 54.213	–0.456	–0.550, –0.362
Memory	–0.030	–0.093, 0.033	–0.302	–1.096, 0.492	50.210	48.824, 51.596	–0.300	–0.400, –0.200

Bolded values indicate p < 0.05. Models were adjusted for baseline age, education, sex, and smoking status (smoker versus nonsmoker) at Exam 1. The joint growth/survival models were used to evaluate time to MetS onset and its components with cognitive level and cognitive trajectories. Cognitive tests were internally scaled within the Framingham Offspring Study. MetS, metabolic syndrome; CI, confidence interval; HDL, high-density lipoprotein.

Associations of time to MetS with cognitive trajectories stratified by APOE ɛ4 status

Table 3 shows the level and slope of cognitive trajectories for each cognitive factor (general cognitive performance, executive function, and memory), and their associations with onset of MetS from joint survival/growth models stratified by APOE ɛ4 status (carrier versus noncarrier). Faster time to MetS was not associated with level or slope of the cognitive factors among APOE ɛ4 carriers. However, among APOE ɛ4 noncarriers, faster time to MetS was associated with lower levels of processing speed/executive functioning (β= –0.733, 95% CI: –1.370, –0.096) although there was no association of time to MetS and cognitive trajectories. When the individual MetS components were examined, faster time to incident abdominal obesity was associated with lower levels of general cognitive performance (β= –1.054, 95% CI: –1.695, –0.413), executive function (β= –1.039, 95% CI: –1.645, –0.433), and memory (β= –0.916, 95% CI: –1.682, –0.150). There were no other significant associations with cognition for the other components.

Table 3

Joint growth/survival models for the association of time to MetS with cognitive trajectories, stratified by APOE ɛ4 genotype and dementia status

Cognitive factor	MetS – cognitive slope	95% CI	MetS – cognitive level	95% CI	Cognitive level	95% CI	Cognitive slope	95% CI
Presence of at least one APOEɛ4 allele
General Cognitive Performance	–0.024	–0.153, 0.105	–0.752	–2.251, 0.747	52.650	49.777, 55.523	–0.477	–0.720, –0.234
Speed/Executive Function	–0.009	–0.142, 0.124	–0.605	–2.198, 0.988	54.137	51.130, 57.144	–0.557	–0.816, –0.298
Memory	–0.009	–0.146, 0.128	–1.058	–2.714, 0.598	50.344	46.996, 53.692	–0.300	–0.578, –0.022
No APOEɛ4 allele
General Cognitive Performance	–0.007	–0.056, 0.042	–0.508	–1.176, 0.160	51.162	49.843, 52.481	–0.336	–0.428, –0.244
Executive Function	–0.026	–0.077, 0.025	–0.733	–1.370, –0.096	53.339	52.049, 54.629	–0.445	–0.543, –0.347
Memory	0.003	–0.052, 0.058	–0.433	–1.231, 0.365	50.158	48.625, 51.691	–0.285	–0.389, –0.181
Developed All-cause Dementia
General Cognitive Performance	–0.140	–0.487, 0.207	5.038	–3.063, 13.139	47.239	36.394, 58.084	–1.153	–1., –0.681
Executive Function	–0.225	–0.633, 0.183	5.604	–4.604, 15.812	48.833	36.436, 61.230	–1.182	–1.725, –0.639
Memory	–0.106	–0.443, 0.231	4.291	–3.496, 12.078	45.891	34.131, 57.651	–0.915	–1.427, –0.403
Remained Nondemented
General Cognitive Performance	–0.006	–0.045, 0.033	–0.663	–1.214, –0.112	50.469	49.360, 51.578	–0.232	–0.308, –0.156
Executive Function	–0.022	–0.065, 0.021	–0.732	–1.271, –0.193	52.237	51.153, 53.321	–0.318	–0.398, –0.238
Memory	0.008	–0.037, 0.053	–0.726	–1.404, –0.048	49.608	48.261, 50.955	–0.196	–0.286, –0.106
Developed Alzheimer’s Disease
General Cognitive Performance	0.019	–0.493, 0.531	4.416	–7.663, 16.495	44.425	28.988, 59.862	–1.174	–1.840, –0.508
Executive Function	–0.017	–0.605, 0.571	3.056	–11.905, 18.017	47.771	30.980, 64.562	–1.216	–1.943, –0.489
Memory	–0.105	–0.509, 0.299	7.347	–2.518, 17.212	43.710	28.250, 59.170	–0.965	–1.594, –0.336

Associations of time to MetS with cognitive trajectories stratified by dementia status

Table 3 shows the level and slope of cognitive trajectories for each cognitive factor (general cognitive performance, executive function, and memory), and their associations with onset of MetS from joint survival/growth models stratified by dementia status (ever developed all-cause dementia versus remained nondemented throughout the course of the study). Faster time to MetS was not associated with level or slope of the cognitive factors among those individuals who developed dementia (all-cause dementia or AD). However, among those who remained nondemented, those individuals with faster time to MetS had lower levels on general cognitive performance (β= –0.664, 95% CI: –1.214, –0.112), processing speed/executive functioning (β= –0.732, 95% CI: –1.271, –0.193), and memory (β= –0.726, 95% CI: –1.404, –0.048) although there was no association of time to MetS and cognitive trajectories.

When the individual MetS components were examined, among those who remained nondemented, faster time to incident abdominal obesity was associated with lower levels of general cognitive performance (β= –0.681, 95% CI: –1.204, –0.158), executive function (β= –0.608, 95% CI: –1.112, –0.104), and memory (β= –0.688, 95% CI: –1.329, –0.047). In addition, faster time to elevated triglycerides was associated with lower levels of general cognitive performance (β= –0.755, 95% CI: –1.370, –0.140), executive function (β= –0.773, 95% CI: –1.402, –0.144), and memory (β= –1.032, 95% CI: –1.779, –0.285). There were no other significant associations with cognition for the other components.

Associations of presence of MetS in mid-life and late-life with cognitive level and trajectories

Table 4 shows the association of time-invariant components of MetS and cognition in mid-life (i.e., at age 55) and in late life (i.e., at age 70). Those with MetS in mid-life showed lower levels of general cognitive performance, processing speed/executive function, and memory at age 55 relative to those without mid-life MetS. Those with MetS in late-life showed lower levels of general cognitive performance and processing speed/executive function at age 70 compared to those without late-life MetS. Mid-life abdominal obesity was associated with lower levels of all cognitive factors although late-life abdominal obesity was not associated with cognitive level. High mid-life triglycerides were associated with lower levels of all cognitive factors, whereas high late-life triglycerides were associated with lower levels of general cognitive performance and processing speed/executive function. Low mid-life HDL cholesterol was associated with lower levels of all cognitive factors although low late-life HDL cholesterol was associated with lower level of processing speed/executive function only. Both high mid-life and late-life blood pressure was associated with lower levels of all cognitive factors compared to those without elevated blood pressure. Neither high mid-life nor late-life fasting blood glucose was associated with cognitive levels. Furthermore, neither mid-life MetS nor late-life MetS overall was associated with cognitive trajectories although mid-life and late-life abdominal obesity, mid-life elevated triglycerides, and mid-life low HDL cholesterol were each associated with slower rate of cognitive decline.

Table 4

Latent growth curve models for binary predictor of MetS at ages 55 and 70 and cognitive trajectories

Cognitive factor	MetS – cognitive slope	95% CI	MetS – cognitive level	95% CI	Cognitive level	95% CI	Cognitive slope	95% CI
MetS at 55
General Cognitive Performance	0.045	–0.018, 0.108	–1.376	–2.133, –0.619	58.411	57.562, 59.260	–0.407	–0.468, –0.346
Speed/Executive Function	0.021	–0.044, 0.086	–1.319	–2.085, –0.553	59.336	58.485, 60.187	–0.411	–0.474, –0.348
Memory	0.056	–0.013, 0.125	–1.590	–2.472, –0.708	58.774	57.806, 59.742	–0.443	–0.510, –0.376
MetS at 70
General Cognitive Performance	0.026	–0.033, 0.085	–0.792	–1.415, –0.169	52.503	51.776, 53.230	–0.403	–0.468, –0.338
Speed/Executive Function	0.008	–0.053, 0.069	–1.248	–1.863, –0.633	53.553	52.814, 54.292	–0.402	–0.471, –0.333
Memory	0.047	–0.018, 0.112	–0.609	–1.248, 0.030	52.182	51.441, 52.923	–0.447	–0.520, –0.374
Abdominal Obesity at 55
General Cognitive Performance	0.126	0.065, 0.187	–1.533	–2.260, –0.806	58.320	57.509, 59.131	–0.426	–0.483, –0.369
Speed/Executive Function	0.103	0.038, 0.168	–1.650	–2.385, –0.915	59.340	58.529, 60.151	–0.433	–0.492, –0.374
Memory	0.125	0.058, 0.192	–1.444	–2.293, –0.595	58.540	57.607, 59.473	–0.455	–0.518, –0.392
Abdominal Obesity at 70
General Cognitive Performance	0.094	0.033, 0.155	0.077	–0.572, 0.726	51.981	51.273, 52.689	–0.438	–0.503, –0.373
Speed/Executive Function	0.068	0.003, 0.133	–0.113	–0.766, 0.540	52.844	52.123, 53.565	–0.436	–0.503, –0.369
Memory	0.090	0.023, 0.157	0.045	–0.612, 0.702	51.800	51.085, 52.515	–0.466	–0.537, –0.395
Elevated Triglycerides at 55
General Cognitive Performance	0.067	0.006, 0.128	–1.464	–2.219, –0.709	58.461	57.630, 59.292	–0.416	–0.477, –0.355
Speed/Executive Function	0.044	–0.021, 0.109	–1.454	–2.209, –0.699	59.409	58.576, 60.242	–0.420	–0.483, –0.357
Memory	0.061	–0.008, 0.130	–1.448	–2.332, –0.564	58.719	57.766, 59.672	–0.445	–0.512, –0.378
Elevated Triglycerides at 70
General Cognitive Performance	0.020	–0.037, 0.077	–0.735	–1.333, –0.137	52.404	51.714, 53.094	–0.396	–0.461, –0.331
Speed/Executive Function	0.004	–0.055, 0.063	–1.118	–1.714, –0.522	53.379	52.671, 54.087	–0.398	–0.465, –0.331
Memory	0.053	–0.010, 0.116	–0.586	–1.203, 0.031	52.113	51.403, 52.823	–0.446	–0.517, –0.375
Low HDL Cholesterol at 55
General Cognitive Performance	0.064	0.007, 0.121	–1.139	–1.884, –0.394	58.452	57.576, 59.328	–0.423	–0.486, –0.360
Speed/Executive Function	0.071	0.010, 0.132	–1.658	–2.407, –0.909	59.715	58.841, 60.589	–0.441	–0.506, –0.376
Memory	0.064	0.001, 0.127	–1.038	–1.900, –0.176	58.647	57.640, 59.654	–0.452	–0.521, –0.383
Low HDL Cholesterol at 70
General Cognitive Performance	–0.012	–0.073, 0.049	–0.512	–1.135, 0.111	52.346	51.611, 53.081	–0.379	–0.450, –0.308
Speed/Executive Function	–0.016	–0.079, 0.047	–1.100	–1.717, –0.483	53.518	52.779, 54.257	–0.390	–0.464, –0.316
Memory	–0.015	–0.082, 0.052	–0.400	–1.043, 0.243	52.062	51.300, 52.824	–0.408	–0.486, –0.330
High Blood Pressure at 55
General Cognitive Performance	0.016	–0.043, 0.075	–1.167	–1.937, –0.397	58.618	57.656, 59.580	–0.398	–0.467, –0.329
Speed/Executive Function	0.025	–0.036, 0.086	–1.338	–2.110, –0.566	59.663	58.701, 60.625	–0.416	–0.487, –0.345
Memory	0.013	–0.050, 0.076	–1.278	–2.160, –0.396	58.966	57.874, 60.058	–0.428	–0.502, –0.354
High Blood Pressure at 70
General Cognitive Performance	–0.046	–0.117, 0.025	–1.462	–2.270, –0.654	53.274	52.333, 54.215	–0.345	–0.425, –0.265
Speed/Executive Function	–0.041	–0.114, 0.032	–1.614	–2.394, –0.834	54.161	53.236, 55.086	–0.361	–0.445, –0.277
Memory	–0.062	–0.138, 0.014	–1.652	–2.463, –0.841	53.236	52.301, 54.171	–0.363	–0.451, –0.275
Elevated Fasting Glucose at 55
General Cognitive Performance	–0.005	–0.066, 0.056	–0.193	–0.999, 0.613	57.942	56.978, 58.906	–0.383	–0.452, –0.314
Speed/Executive Function	–0.023	–0.088, 0.042	0.041	–0.790, 0.872	58.679	57.705, 59.653	–0.381	–0.455, –0.307
Memory	0.001	–0.064, 0.066	–0.478	–1.411, 0.455	58.440	57.321, 59.559	–0.420	–0.496, –0.344
Elevated Fasting Glucose at 70
General Cognitive Performance	–0.008	–0.086, 0.070	–0.810	–1.672, 0.052	52.720	51.750, 53.690	–0.380	–0.466, –0.294
Speed/Executive Function	–0.021	–0.103, 0.061	–0.816	–1.690, 0.058	53.476	52.486, 54.466	–0.379	–0.471, –0.287
Memory	0.034	–0.054, 0.122	–0.697	–1.606, 0.212	52.417	51.400, 53.434	–0.448	–0.546, –0.350

Bolded values indicate p < 0.05. Models were adjusted for baseline age, education, sex, and smoking status (smoker versus nonsmoker) at Exam 1. Cognitive tests were internally scaled within the Framingham Offspring Study. MetS, metabolic syndrome; CI, confidence interval; HDL, high-density lipoprotein.

Association of presence of MetS with rapid versus slow cognitive decline

Table 5 shows demographic characteristics including prevalence of MetS as a function of “slow decline” (i.e., 30% of the sample with the slowest decline on the general cognitive performance factor) and “rapid decline” (i.e., the 30% of the sample with the most rapid decline on the general cognitive performance factor). Those showing rapid decline were significantly more likely to have MetS. In addition, compared to slow decliners, individuals showing rapid decline were somewhat older at baseline MetS assessment and first neuropsychological assessment, more likely to be male, completed less education, have fewer years in study and fewer study visits, and poorer mean cognitive performance at first neuropsychological assessment. There was no difference in mean age at which MetS was first observed for slow versus rapid decliners. Figure 1 shows cognitive trajectories for those participants characterized as “slow” versus “rapid” decliners.

Table 5

Characteristics for sample according to slow versus rapid cognitive decline

	Slow decliners (n = 857)	Rapid decliners (n = 865)	p
Demographic Characteristics
Age at baseline, mean (SD)	33.0 (8.3)	36.9 (9.9)	<0.001
Age at first neuropsychological assessment, mean (SD)	60.5 (8.6)	64.5 (10.7)	<0.001
Education Group, n (%)			<0.001
<High school degree	4 (0.5)	83 (9.7)
High school degree	110 (12.8)	341 (39.8)
Some college	160 (18.6)	235 (27.4)
≥College degree	585 (68.1)	198 (23.1)
Women, n (%)	509 (59.3)	406 (47.0)	<0.001
APOEɛ4 carrier, n (%)	183 (22.3)	200 (24.1)	0.392
Ever had dementia, n (%)	4 (0.5)	156 (18.0)	<0.001
Years in study, mean (SD)	37.5 (3.3)	35.9 (4.4)	<0.001
Number of study visits, mean (SD)	2.5 (1.2)	2.2 (1.3)	<0.001
Years in study with cognition, mean (SD)	6.4 (3.7)	4.9 (4.3)	<0.001
Number of study visits with cognition, mean (SD)	2.4 (1.2)	2.2 (1.3)	<0.001
MetS/MetS* Components, n (%)
Ever had MetS	556 (64.9)	644 (74.5)	<0.001
Earliest age when MetS observed, mean (SD)	53.7 (13.0)	53.1 (13.5)	0.449
Cognitive Function**, mean (SD)
General Cognitive Performance	60.9 (3.4)	46.3 (7.3)	<0.001
Memory	61.2 (4.3)	45.8 (7.4)	<0.001
Executive Functioning	57.9 (3.5)	48.4 (8.5)	<0.001

MetS, metabolic syndrome; APOE, apolipoprotein E; SD, standard deviation. *Ever met criteria for MetS during the entire study period. **Cognitive scores reflect performance at first cognitive assessment and are internally scaled within the Framingham Offspring Study.

Fig.1

Graph of model-estimated slopes depicting trajectories of change in the general cognitive performance factor by age for individual participants showing slow decline (i.e., 30% of sample showing the slowest decline/most stable performance) versus rapid decline (i.e., 30% of sample showing the most rapid decline). Individuals with MetS were significantly more likely to show rapid decline relative to their counterparts without MetS (p < 0.001).

Sensitivity analyses

We estimated joint growth/survival models (Supplementary Tables 1 and 2) and latent growth curve models (Supplementary Table 3) among the 1,601 individuals with MetS data at both ages 55 and 70 years. Results from these supplementary tables correspond to Tables 2, 3, and 4, respectively. The majority of findings remained similar. Specifically, the lack of findings between overall MetS and cognitive level or slope from joint growth/survival models were maintained (Supplementary Table 1). The associations between abdominal obesity and cognition in these models were attenuated. Although the associations of abdominal obesity with general cognitive performance (β= –0.592, 95% CI: –1.421, 0.237) and memory did not reach statistical significance (β= –0.662, 95% CI: –1.656, 0.332), the point estimates were close to the corresponding estimates in Table 2 (general cognitive performance: β= –0.671, 95% CI: –1.230, –0.112; memory: β= –0.671, 95% CI: –1.332, –0.010). In this subsample, there was a significant association between earlier onset of elevated fasting blood glucose and steeper decline in speed/executive function (β= –0.074, 95% CI: –0.147, –0.001). We re-ran this analysis excluding those individuals with a diagnosis of diabetes, and the association remained significant (β= –0.076, 95% CI: –0.149, –0.003).

When analyses were stratified by dementia status (ever versus never developing dementia during the study) and APOE ɛ4 status (carrier versus noncarriers) significant associations between MetS and general cognitive performance and memory among those who remained nondemented were maintained. The association between MetS and executive function for those without the APOE ɛ4 allele was attenuated and no longer reached statistical significance (Supplementary Table 2).

Finally, significant associations of the binary predictor of MetS (present versus absent) at age 55 with levels of general cognitive performance, speed/executive function, and memory were maintained, as were associations of MetS at age 70 with levels of general cognitive performance and speed/executive function. Most of the significant associations of individual MetS components at age 55 and at age 70 with cognitive level were maintained, although some findings between individual components and cognitive slope were attenuated (Supplementary Table 3).

DISCUSSION

In this large prospective longitudinal study of community-dwelling individuals spanning middle to old age, onset of MetS was associated with lower level of general cognitive performance, processing speed/executive function, and memory abilities. These associations were strongest among those who were nondemented and who were at a relatively lower risk of developing dementia by virtue of not possessing a copy of the APOE ɛ4 allele. MetS was not associated with cognitive trajectories in the overall sample although when restricting the sample to those individuals showing the most rapid decline and those showing the slowest decline/most stable cognition, individuals with the most rapid decline were significantly more likely to have MetS. When we re-ran our main analyses using a subsample of people who had MetS data available at both ages 55 and 70 years, most findings remained similar, although some associations were attenuated. Overall, our results did not appear to be driven by differences between those individuals that were lost to follow-up versus those who remained in the study.

Our findings suggest that the influence of MetS on cognitive performance and cognitive decline may be evident in subgroups of study populations in which there is no association between these variables in the overall sample. Although MetS was not associated with cognitive trajectory, it was associated with lower cognitive level among nondemented participants and those without an APOE ɛ4 allele. Our findings suggest that those with MetS are at greater risk of cognitive impairment and dementia given their lower level of cognitive functioning. That is, individuals with MetS are closer to meeting thresholds for cognitive impairment.

Our analyses examining associations of mid-life MetS and cognition (at age 55) as well as associations of late-life MetS and cognition (at age 70) showed that MetS and several of its components relate to poorer cognitive performance compared to those without MetS during both mid-life and late-life. There were a greater number of significant associations between presence of MetS and poorer cognition during mid-life compared to late-life. Although several studies have shown that mid-life vascular risk factors relate to poorer cognition later in life [41, 42], very few studies have had data to examine the associations of both mid-life and late-life vascular risk factors with comprehensive neuropsychological data spanning multiple domains collected at multiple time points. The present findings help elucidate the association of vascular risk factors and cognitive functioning across the adult lifespan.

Our current results corroborate findings from another previous study in which we observed that, in a sample of 1,493 older adults, diabetes was associated with poorer baseline in memory, language, processing speed/executive functioning, and visuospatial abilities but no difference in rate of cognitive decline [43]. Findings of poorer cognitive performance but equal rates of change suggest that separation of slopes occurred earlier in life. Of note, in latent growth curve models, presence of abdominal obesity, elevated triglycerides, and low HDL cholesterol were associated with lower level of cognitive performance but also slower rate of decline. It is possible that the presence of vascular risk factors causes an insult that leads to lower cognitive performance in the absence of accelerated decline or that an additional factor related to MetS and related conditions (e.g., lower cognitive reserve) may explain the pattern of findings [43].

Our results are consistent with the findings of several previously published reports showing that MetS may relate to poorer cognition particularly in certain subgroups (e.g., cognitive risk groups, individuals with persistent MetS) [12 , 44]. Notably, the association between MetS and poorer cognitive performance was strongest among those participants who may be considered to be at relatively low risk of cognitive dysfunction, given that they were without an APOE ɛ4 allele and were nondemented. This is in line with previous work from the Framingham Heart Study that found that the diabetes-related risk of developing AD was higher among those without major AD risk factors, including absence of an APOE ɛ4 allele [45]. A plausible biological explanation for these findings comes from evidence showing that the brains of individuals without an APOE ɛ4 allele may be more vulnerable to the negative effects of hyperinsulinemia [46]. This pattern of findings may also suggest that genetic and other risk factors may be the primary drivers for those with the APOE ɛ4 allele and those with dementia.

Our findings differed, depending on whether cognition is treated as a continuous variable or a categorical variable. When we considered only those with the most extreme cognitive change, we observed a significant association between MetS and cognitive decline. Notably, those who were faster decliners had fewer follow up visits, compared to slower decliners, which may have introduced a differential bias. Those individuals with longer follow up may be representative of healthy older adults who are less vulnerable to the detrimental effects of MetS and its components. In addition, when compared to slow decliners, individuals who declined fast, completed fewer years of education and showed poorer mean cognitive performance at their first neuropsychological assessment, which are risk factors for future cognitive decline. As such, although MetS was associated with cognitive decline, we cannot be certain that this relationship is causative. Also, given that we found associations between midlife MetS and lower cognitive level but not rate of cognitive change, confounding by early-life factors that influence both MetS status and cognitive performance is possible. Although we adjusted for educational attainment in our primary analyses, we did not have a good measure of early-life conditions (e.g., socioeconomic status [SES]) available for this study. Future research should examine the influence of early-life SES on MetS-cognition associations.

The main goal of our manuscript was to examine associations between MetS and cognition, given in part that MetS is common and its components tend to cluster. Although we were not primarily interested in specific components of MetS, in secondary analyses, we evaluated the associations of MetS components with cognitive performance to evaluate whether one or a few of the components might be driving associations with overall MetS. The associations of time to MetS with cognitive level (presented in Table 2) appear to be primarily driven by obesity. The associations of MetS (present versus absent) at ages 55 and 70 years with cognitive levels (presented in Table 4) were significant for overall MetS as well as for the components of obesity, elevated triglycerides, low HDL cholesterol, and high blood pressure. In terms of these latter analyses, multiple components agreed with the overall MetS findings, although there were no instances of complete overlap where both overall MetS and all individual components showed significant associations with cognitive level. It is possible that the combination of factors that comprise MetS is not uniquely relevant to cognitive outcomes and may instead reflect the utility of examining several health factors. Future research should compare MetS to other aggregate risk scores and also examine MetS characteristics, i.e., duration and complications in relation to cognitive outcomes. Nonetheless, given that each of its components is potentially modifiable, MetS may be useful in identifying those at higher risk for cognitive impairment and motivating individuals toward behavioral change in an effort to reduce risk.

Strengths of the current study include a large, well-characterized, community-based sample; prospective study design; ongoing follow-up for over four decades; comprehensive neuropsychological assessment at multiple time points; and novel psychometric and modeling approach. This study also has limitations including that our sample was predominantly white, generally medically healthy, and relatively well-educated, which may affect generalizability.

Despite these limitations, the present findings have important implications. We found that associations between MetS and cognition varied, depending upon whether the sample was stratified by genetic and cognitive status, whether we considered cognitive performance as a continuous variable or examined categorical groupings (rapid versus slow decline), and when individual MetS components were considered. The present findings found a subgroup (nondemented older adults and those without the APOE ɛ4 allele) for whom the associations between MetS and cognition are particularly prominent. Importantly, previous studies suggest that MetS may be reversible with changes in behavior, such as diet [44]. Moreover, pilot clinical trials of treatment with intranasal insulin have reported promising results for improvement in cognition, preserved brain volume, and reduction in the tau-P181/Aβ₄₂ ratio among older adults with amnestic mild cognitive impairment and AD, which suggests that insulin may modify AD-related pathophysiologic processes [47, 48], thereby reducing risk of late-life cognitive impairment. Besides consideration for certain subgroups having more vulnerability to MetS-association cognitive dysfunction, interventions and clinical trials should evaluate age and specific components of MetS. This will be increasingly important with the aging of the population and increasing rates of obesity, diabetes, and other metabolic conditions. In sum, our findings emphasize the need to consider MetS as a potential risk factor for cognitive decline, highlight that age of onset may modify the association between MetS and cognition, and suggest that MetS may be a risk factor for decline in the absence of known risk factors including the APOE ɛ4 allele.

Footnotes

ACKNOWLEDGMENTS

Funding for this study comes from the Alzheimer’s Association (VMF-14-318524 to RA, AARG-18-566254 to KJB); VA Clinical Science Research & Development (Career Development Award-2 1IK2CX000938 to KJB); National Institute on Aging Intramural Research Program (NMA); Framingham Heart Study’s National Heart, Lung, and Blood Institute contract (N01-HC-25195; HHSN268201500001I); grants from the National Institute on Aging (R01-AG016495, R01-AG008122, R01-AG033040) and from the National Institute of Neurological Disorders and Stroke (R01-NS017950); and the Dana Foundation. The authors thank the staff, volunteers, and participants of the Framingham Heart Study for their important contributions.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Grundy

, Cleeman

, Daniels

, Donato

, Eckel

, Franklin

, Gordon

, Krauss

, Savage

, Smith

Jr. , Spertus

, Costa

(2005) Diagnosis and management of the metabolic syndrome: An American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation 112, 2735–2752.

Grundy

, Brewer

Jr , Cleeman

, Smith

Jr , Lenfant

(2004) Definition of metabolic syndrome: Report of the National Heart, Lung, and Blood Institute/American Heart Association conference on scientific issues related to definition. Circulation 109, 433–438.

Malik

, Wong

, Franklin

, Kamath

, L’Italien

, Pio

, Williams

(2004) Impact of the metabolic syndrome on mortality from coronary heart disease, cardiovascular disease, and all causes in United States adults. Circulation 110, 1245–1250.

Ford

(2005) Risks for all-cause mortality, cardiovascular disease, and diabetes associated with the metabolic syndrome: A summary of the evidence. Diabetes Care 28, 1769–1778.

Chen

, Bai

, Yeh

, Chiu

, Pan

(2006) Influence of metabolic syndrome and general obesity on the risk of ischemic stroke. Stroke 37, 1060–1064.

, Feng

, Nyunt

, Feng

, Gao

, Lim

, Collinson

, Chong

, Lim

, Lee

, Yap

(2016) Metabolic syndrome and the risk of mild cognitive impairment and progression to dementia: Follow-up of the Singapore Longitudinal Ageing Study Cohort. JAMA Neurol 73, 456–463.

Vanhanen

, Koivisto

, Moilanen

, Helkala

, Hanninen

, Soininen

, Kervinen

, Kesaniemi

, Laakso

, Kuusisto

(2006) Association of metabolic syndrome with Alzheimer disease: A population-based study. Neurology 67, 843–847.

Razay

, Vreugdenhil

, Wilcock

(2007) The metabolic syndrome and Alzheimer disease. Arch Neurol 64, 93–96.

Solfrizzi

, Scafato

, Capurso

, D’Introno

, Colacicco

, Frisardi

, Vendemiale

, Baldereschi

, Crepaldi

, Di Carlo

, Galluzzo

, Gandin

, Inzitari

, Maggi

, Capurso

, Panza

(2010) Metabolic syndrome and the risk of vascular dementia: The Italian Longitudinal Study on Ageing. J Neurol Neurosurg Psychiatry 81, 433–440.

10.

Kalmijn

, Foley

, White

, Burchfiel

, Curb

, Petrovitch

, Ross

, Havlik

, Launer

(2000) Metabolic cardiovascular syndrome and risk of dementia in Japanese-American elderly men. The Honolulu-Asia aging study. Arterioscler Thromb Vasc Biol 20, 2255–2260.

11.

Raffaitin

, Gin

, Empana

, Helmer

, Berr

, Tzourio

, Portet

, Dartigues

, Alperovitch

, Barberger-Gateau

(2009) Metabolic syndrome and risk for incident Alzheimer’s disease or vascular dementia: The Three-City Study. Diabetes Care 32, 169–174.

12.

Siervo

, Harrison

, Jagger

, Robinson

, Stephan

(2014) Metabolic syndrome and longitudinal changes in cognitive function: A systematic review and meta-analysis. J Alzheimers Dis 41, 151–161.

13.

Gatto

, Henderson

, St John

, McCleary

, Hodis

, Mack

(2008) Metabolic syndrome and cognitive function in healthy middle-aged and older adults without diabetes. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 15, 627–641.

14.

Laudisio

, Marzetti

, Pagano

, Cocchi

, Franceschi

, Bernabei

, Zuccala

(2008) Association of metabolic syndrome with cognitive function: The role of sex and age. Clin Nutr 27, 747–754.

15.

Luo

, Yang

, Hao

, Yue

, Dong

(2013) Cross-sectional study examining the association between metabolic syndrome and cognitive function among the oldest old. J Am Med Dir Assoc 14, 105–108.

16.

McEvoy

, Laughlin

, Barrett-Connor

, Bergstrom

, Kritz-Silverstein

, Der-Martirosian

, von Muhlen

(2012) Metabolic syndrome and 16-year cognitive decline in community-dwelling older adults. Ann Epidemiol 22, 310–317.

17.

Raffaitin

, Feart

, Le Goff

, Amieva

, Helmer

, Akbaraly

, Tzourio

, Gin

, Barberger-Gateau

(2011) Metabolic syndrome and cognitive decline in French elders: The Three-City Study. Neurology 76, 518–525.

18.

van den Berg

, Biessels

, de Craen

, Gussekloo

, Westendorp

(2007) The metabolic syndrome is associated with decelerated cognitive decline in the oldest old. Neurology 69, 979–985.

19.

, Niti

, Yap

, Kua

, Ng

(2008) Metabolic syndrome and cognitive decline in chinese older adults: Results from the Singapore Longitudinal Ageing Studies. Am J Geriatr Psychiatry 16, 519–522.

20.

Yaffe

, Kanaya

, Lindquist

, Simonsick

, Harris

, Shorr

, Tylavsky

, Newman

(2004) The metabolic syndrome, inflammation, and risk of cognitive decline. JAMA 292, 2237–2242.

21.

de Jager

, Schrijnemaekers

, Honey

, Budge

(2009) Detection of MCI in the clinic: Evaluation of the sensitivity and specificity of a computerised test battery, the Hopkins Verbal Learning Test and the MMSE. Age Ageing 38, 455–460.

22.

Ostrosky-Solis

, Lopez-Arango

, Ardila

(2000) Sensitivity and specificity of the Mini-Mental State Examination in a Spanish-speaking population. Appl Neuropsychol 7, 25–31.

23.

Feinleib

, Kannel

, Garrison

, McNamara

, Castelli

(1975) The Framingham Offspring Study. Design and preliminary data. Prev Med 4, 518–525.

24.

Gross

, Sherva

, Mukherjee

, Newhouse

, Kauwe

, Munsie

, Waterston

, Bennett

, Jones

, Green

, Crane

(2014) Calibrating longitudinal cognition in Alzheimer’s disease across diverse test batteries and datasets. Neuroepidemiology 43, 194–205.

25.

Gross

, Power

, Albert

, Deal

, Gottesman

, Griswold

, Wruck

, Mosley

Jr , Coresh

, Sharrett

, Bandeen-Roche

(2015) Application of latent variable methods to the study of cognitive decline when tests change over time. Epidemiology 26, 878–887.

26.

Gross

, Mungas

, Crane

, Gibbons

, MacKay-Brandt

, Manly

, Mukherjee

, Romero

, Sachs

, Thomas

, Potter

, Jones

(2015) Effects of education and race on cognitive decline: An integrative study of generalizability versus study-specific results. Psychol Aging 30, 863–880.

27.

Gross

, Jones

, Fong

, Tommet

, Inouye

(2014) Calibration and validation of an innovative approach for estimating general cognitive performance. Neuroepidemiology 42, 144–153.

28.

Fieo

, Mukherjee

, Dmitrieva

, Fyffe

, Gross

, Sanders

, Romero

, Potter

, Manly

, Mungas

, Gibbons

(2015) Differential item functioning due to cognitive status does not impact depressive symptom measures in four heterogeneous samples of older adults. Int J Geriatr Psychiatry 30, 911–918.

29.

Curran

, McGinley

, Bauer

, Hussong

, Burns

, Chassin

, Sher

, Zucker

(2014) A moderated nonlinear factor model for the development of commensurate measures in integrative data analysis. Multivariate Behav Res 49, 214–231.

30.

Sisco

, Gross

, Shih

, Sachs

, Glymour

, Bangen

, Benitez

, Skinner

, Schneider

, Manly

(2015) The role of early-life educational quality and literacy in explaining racial disparities in cognition in late life. J Gerontol B Psychol Sci Soc Sci 70, 557–567.

31.

Schneider

, Gross

, Bangen

, Skinner

, Benitez

, Glymour

, Sachs

, Shih

, Sisco

, Manly

, Luchsinger

(2015) Association of vascular risk factors with cognition in a multiethnic sample. J Gerontol B Psychol Sci Soc Sci 70, 532–544.

32.

Samejima

(1969) Estimation of latent ability using a response pattern of graded scores. Psychometrika Monograph: No. 17.

33.

Wechsler

(1945) A standardized memory scale for clinical use. J Psychol 19, 87–95.

34.

Reitan

, Wolfson

(1985) The Halstead-Reitan Neuropsychological Test Battery. Neuropsychology Press, Tucson, AZ.

35.

Wechsler

(1939) Wechsler-Belllevue Intelligence Scale, Psychological Corporation.

36.

Kaplan

, Goodglass

, Weintraub

(1983) The Boston Naming Test. Lea & Febiger, Philadelphia.

37.

(2001) Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 285, 2486–2497.

38.

Genuth

, Alberti

, Bennett

, Buse

, Defronzo

, Kahn

, Kitzmiller

, Knowler

, Lebovitz

, Lernmark

, Nathan

, Palmer

, Rizza

, Saudek

, Shaw

, Steffes

, Stern

, Tuomilehto

, Zimmet

(2003) Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 26, 3160–3167.

39.

McArdle

, Small

, Backman

, Fratiglioni

(2005) Longitudinal models of growth and survival applied to the early detection of Alzheimer’s disease. J Geriatr Psychiatry Neurol 18, 234–241.

40.

Muthen

, Muthen

(1998-2012) Mplus User’s Guide. Muthen & Muthen, Los Angeles, CA.

41.

Whitmer

, Sidney

, Selby

, Johnston

, Yaffe

(2005) Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 64, 277–281.

42.

Bangen

, Beiser

, Delano-Wood

, Nation

, Lamar

, Libon

, Bondi

, Seshadri

, Wolf

, Au

(2013) APOE genotype modifies the relationship between midlife vascular risk factors and later cognitive decline. J Stroke Cerebrovasc Dis 22, 1361–1369.

43.

Bangen

, Gu

, Gross

, Schneider

, Skinner

, Benitez

, Sachs

, Shih

, Sisco

, Schupf

, Mayeux

, Manly

, Luchsinger

(2015) Relationship between type 2 diabetes mellitus and cognitive change in a multiethnic elderly cohort. J Am Geriatr Soc 63, 1075–1083.

44.

Akbaraly

, Singh-Manoux

, Tabak

, Jokela

, Virtanen

, Ferrie

, Marmot

, Shipley

, Kivimaki

(2010) Overall diet history and reversibility of the metabolic syndrome over 5 years: The Whitehall II prospective cohort study. Diabetes Care 33, 2339–2341.

45.

Akomolafe

, Beiser

, Meigs

, Au

, Green

, Farrer

, Wolf

, Seshadri

(2006) Diabetes mellitus and risk of developing Alzheimer disease: Results from the Framingham Study. Arch Neurol 63, 1551–1555.

46.

Craft

, Asthana

, Schellenberg

, Cherrier

, Baker

, Newcomer

, Plymate

, Latendresse

, Petrova

, Raskind

, Peskind

, Lofgreen

, Grimwood

(1999) Insulin metabolism in Alzheimer’s disease differs according to apolipoprotein E genotype and gender. Neuroendocrinology 70, 146–152.

47.

Craft

, Claxton

, Baker

, Hanson

, Cholerton

, Trittschuh

, Dahl

, Caulder

, Neth

, Montine

, Jung

, Maldjian

, Whitlow

, Friedman

(2017) Effects of regular and long-acting insulin on cognition and Alzheimer’s disease biomarkers: A pilot clinical trial. J Alzheimers Dis 57, 1325–1334.

48.

Claxton

, Baker

, Hanson

, Trittschuh

, Cholerton

, Morgan

, Callaghan

, Arbuckle

, Behl

, Craft

(2015) Long-acting intranasal insulin detemir improves cognition for adults with mild cognitive impairment or early-stage Alzheimer’s disease dementia. J Alzheimers Dis 44, 897–906.