Sex differences in allostatic load profiles and incident dementia: The AGES-Reykjavik Study

Abstract

Background

Allostatic load (AL), an umbrella term for the physiological response to chronic stress, is different in women and men. AL has also been associated with all-cause dementia.

Objective

The current study investigates if AL clusters differently in men and women, and if these sex-based clusters are associated with all-cause dementia.

Methods

The study included individuals without dementia (n = 5343, 58% women, age range: 66–98 years) at baseline from the AGES-Reykjavik Study, a population-based cohort study. AL markers of cardiovascular, lipid, neuroendocrine, and inflammatory components were assessed at baseline. Clustering of AL markers was done using latent profile analysis in men and women separately to create sex-specific AL risk groups. Sex-specific Cox regressions on the sex-specific AL risk groups, adjusted for age, education, and medical and lifestyle factors, were performed to assess if the relationship between AL and all-cause, Alzheimer's, and non-Alzheimer's dementia differed per sex.

Results

All-cause dementia was diagnosed in 1099 participants during follow-up (median: 10 years). Only cardiovascular and metabolic factors differed between AL groups in men. One of the groups in women, labeled ‘Risk factors’, was associated with a lower risk of AD dementia (HR 0.75; 95% CI 0.58; 0.98) compared to the ‘Average’ group. In men, a group labeled ‘Multisystem dysregulation’, consisting of mostly individuals with diabetes, was associated with an increased risk of all-cause dementia (HR 1.75; 95% CI 1.06; 2.90).

Conclusions

AL clustered differently in men and women. Metabolic dysregulation, specifically in men, was associated with all-cause dementia.

Keywords

allostatic load Alzheimer's disease clustering dementia sex differences

Introduction

Dementia, characterized by a decline in cognitive functioning, daily functioning, and quality of life, currently impacts around 50 million people worldwide and is expected to triple in number by 2050.¹ As there currently is not a cure for dementia, focusing on modifiable risk factors can help aid in the detection of high-risk individuals as well as prevention programs. Multiple modifiable risk factors have been identified for dementia, such as cardiovascular,^2,3 metabolic,^4,5 lipid,⁶ inflammatory,^7,8 and neuroendocrine⁹ factors. These biomarkers together form the umbrella term of “allostatic load” (AL), representing wear-and-tear across multiple physiological systems after repeated stressful stimuli.¹⁰ Assessing AL rather than individual biomarkers may give a better representation of ongoing complex disease processes.¹¹

Additionally, there are significant sex differences in the epidemiology and pathophysiology of dementia.^12,13 However, the research regarding these differences is limited and rarely focuses on AL factors. AL may emerge years before symptoms of cognitive decline and may assist the identification of high-risk individuals. In our recent study, we found four distinct AL clusters through latent profile analysis using the population-based AGES-Reykjavik Study.¹⁴ Further, one specific profile was associated with incident all-cause dementia, labeled ‘Multisystem dysregulation’. However, it is not known whether AL clusters the same in both men and women, and if those sex-specific clusters are associated with incident dementia. A recent meta-analysis of 21 different cohorts assessing individual cardiovascular, metabolic, and lipid factors found no sex difference in the association of these factors with dementia.¹⁵ There is a possibility that AL factors may interact with one another differently in men and women, and therefore should be assessed via clustering techniques rather than looking at each AL marker individually.

A recent systematic review found a discrepancy in the literature regarding sex differences and AL.¹⁶ While several studies found higher AL in men compared to women, a number of studies also found no differences in AL in men compared to women. One reason for this could be due to the cut-offs that are normally used to assess AL, either individually or as a sum score. These cut-offs may be sensitive to differences in sex and may explain discrepancies between previous studies, highlighting the need to either have consistent sex-specific cut-offs or assess AL on a continuous scale. Further, by defining AL through a sum score may hide possible sex-specific interactions that may exist between AL systems. Additionally, some studies have shown that women and men may show differential biological wear-and-tear in response to chronic stress.^17,18 Differences between men and women may expand to sex differences in AL domains. Specifically, one study in the U.S. found higher immune-related AL in women and higher total, cardiovascular, and metabolic AL in men.¹⁸ Further research is needed on sex differences in AL, specifically by using a cluster-based approach, and to see if those sex-based clusters may explain sex disparities in dementia. Using a clustering approach takes into account the complex interactions that may exist between AL factors,¹⁹ and which may also differ between sexes.²⁰

This study had two aims: to investigate (1) if AL clusters differently in men and women and (2) if those AL clusters differentially associate with incident dementia.

Methods

Participants

The current study is based on data from the Age, Gene/Environment Susceptibility (AGES)-Reykjavík Study, which is a population-based cohort study of individuals 65 years or older living in the Reykjavík area in Iceland. In short, the AGES-Reykjavík Study stems from the Reykjavík Study, which began in 1967 by the Icelandic Heart Association. A total of 5764 individuals were chosen at random from the survivors of the Reykjavík Study between 2002 and 2006. Baseline cognitive and biometric assessments were collected at the Reykjavík research center. Follow-up for incident dementia diagnosis was done until 2014. More information on the AGES-Reykjavík Study can be found elsewhere.²¹

Standard protocol approvals, registrations, and patient consents

The Icelandic Data Protection Authority, the Icelandic National Bioethics Committee (VSN: 00-063), and the Institutional Review Board for the National Institute on Aging, NIH approved this study. Written informed consent was obtained from all participants.

Dementia assessment

Based on international criteria, a three-step procedure was used for dementia ascertainment. Further details can be found elsewhere.^22–24 Briefly, in step one, the total sample underwent cognitive assessment, and screen positives underwent further neuropsychological assessment. In step two, further neurologic and proxy examinations were performed on those who screened positive on neuropsychological assessments. Lastly, a multidisciplinary team of a neuropsychologist, neuroradiologist, geriatrician, and neurologist diagnosed dementia at both baseline for exclusion and at follow-up (between 2007 to 2011) based on international guidelines.²¹ Cases were also identified through medical records, nursing home records, and death certificates. For diagnoses in a nursing home, this was done for all-cause dementia and Alzheimer's disease (AD) during an intake exam and by a standardized regularly spaced exams done by all Icelandic nursing homes.²⁵ The current study assessed all-cause dementia, AD, and other dementias as outcomes.

AL measures

AL was defined based on previous research^14,26 and separated into the following categories: cardiovascular factors, lipids, metabolic factors, inflammation, and neuroendocrine factors. Systolic blood pressure and pulse pressure were used as cardiovascular factors.^27,28 High-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides were used for lipids.^26,29 Abdominal circumference and fasting glucose were used for metabolic factors.^29,30 High-sensitivity c-reactive protein (CRP) was used for the inflammatory factor.³¹ Morning and evening salivary cortisol were used for neuroendocrine factors.²⁹ More information for how the factors were assessed can be found elsewhere.¹⁴

Covariates

The covariates included age, sex, education, lifestyle, and medical variables. Age, sex, education and lifestyle variables were assessed via questionnaires at baseline. Education was divided into primary, secondary, college, or university degree. Smoking was classified as current, former, or never smoker. Alcohol use was used continuously as grams per week. Physical activity (i.e., moderate to vigorous) was categorized as never, rarely, occasionally (i.e., weekly <1 h), moderate (1–3 h per week), or high (>4 per week).³² Use of antidepressant or antihypertensive medication was classified as none or any. Stroke was defined through self-assessment or from hospital registries. Prevalence of APOE ɛ4 genotype was dichotomous and assessed via microplate array diagonal gel electrophoresis (MADGE).³³

Demographic information

Mild cognitive impairment was defined as scoring less than 1.5 standard deviations below a cut-point determined from the cohort on memory or two other cognitive domains³⁴ and diagnosed by the same multidisciplinary panel as for dementia. Diabetes mellitus was defined as a history of diabetes, use of glucose-modifying medication, or a fasting blood glucose of more than 7 mmol/L. Metabolic syndrome was based on WHO criteria.^35,36

Data analysis

We excluded those with a dementia diagnosis at baseline, leaving us with 5343 individuals for the analysis.

Missing data (max: 12%) was handled using multiple imputation (10 datasets) in Mplus (v. 6.12).^14,37 Mplus uses a Bayesian Markov chain Monte-Carlo estimation for imputation. Estimates were then pooled during the analyses.

We then assessed possible sex differences between AL factors using chi-square tests, ANOVAs, or Mann Whitney U tests. To assess interaction between profile and sex, we performed these ANOVAs on standardized AL variables to meet the assumptions of normality. We performed latent profile analysis (LPA) to determine sex-based AL profiles (i.e., stratifying by sex) using the tidyLPA (version 1.1.0)³⁸ platform in R (version 4.2.0), which runs MPlus³⁷ using the MPlusAutomation R package.³⁹ Only AL factors were used to derive the sex-based profiles. To create sex-based AL profiles, LPA uses covariance across the indicator variables (i.e., the AL variables) to find relationships among individuals.⁴⁰ We estimated 2–6 profiles to assess best model fit, based on the Akaike information criterion (AIC), the Bayesian information criterion (BIC), the sample size-adjusted BIC, and entropy. We also determined the number of profiles based on more than 1% of the sample fitting into one of the profiles. For further analyses, participants were classified based on most likely class membership.

Lastly, the sex-specific AL profiles were assessed for risk of incident all-cause dementia, AD dementia, and other dementias using Cox regressions, adjusted in a first model for age and education, and a second model additionally adjusting for lifestyle and medical variables (i.e., history of stroke, smoking, alcohol use, antidepressant and antihypertensive medication use, physical activity, and APOE ɛ4 genotype). The assumptions of Cox proportional hazards, influential observations, and nonlinearity were tested and met. We assessed competing risks (e.g., dementia-free mortality) as outcome as an additional sensitivity analysis. Lastly, we assessed as another sensitivity analysis an interaction between AL profiles and APOE ɛ4 genotype to explore if AL differs between men and women with and without an APOE ɛ4 genotype.

Results

The study sample had a mean age of 77 years (age range: 66–98 years), with 58% of them being women. During a maximum of 12 years follow-up (median: 10 years follow-up), 1099 individuals were diagnosed with all-cause dementia, with 492 cases classified as AD dementia. Baseline characteristics for the total sample, as well as stratified by sex, can be found in Table 1. All allostatic load factors differed between men and women. Systolic blood pressure, abdominal circumference, evening salivary cortisol, and glucose were significantly higher in men. Whereas pulse pressure, HDL cholesterol, LDL cholesterol, morning salivary cortisol, CRP, and triglycerides were significantly higher in women.

Table 1.

Baseline characteristics in the study sample and stratified by sex.

	Total population (n = 5343)	Men (n = 2246, 42%)	Women (n = 3097, 58%)	p
Age (y)	77 (6)	77 (6)	77 (6)	0.46
Education, college & university	1450 (27%)	726 (32%)	724 (23%)	<0.001
Current smoker	652 (12%)	260 (12%)	392 (13%)	<0.001
Alcohol use (g/week)*	3 (16)	6 (26)	2 (8)	<0.001
Physical activity, moderate/high	1694 (32%)	791 (35%)	903 (29%)	<0.001
History of stroke	348 (7%)	155 (7%)	193 (6%)	0.33
MCI at baseline	550 (10%)	275 (12%)	275 (9%)	<0.001
Metabolic syndrome	1687 (32%)	664 (30%)	1023 (33%)	0.01
Diabetes	671 (13%)	363 (16%)	308 (10%)	<0.001
Antihypertensive medication	2539 (48%)	968 (43%)	1571 (51%)	<0.001
Antidepressant medication	766 (14%)	236 (11%)	530 (17%)	<0.001
APOE ɛ4 genotype	1471 (28%)	647 (29%)	824 (27%)	0.07
Systolic blood pressure (mmHg)*	140 (27)	141 (26)	140 (27)	0.03
Pulse pressure (mmHg)*	67 (23)	65 (22)	68 (25)	<0.001
Abdominal circumference (cm)*	101 (15)	102 (13)	99 (17)	<0.001
HDL (mmol/L)*	1.5 (0.6)	1.4 (0.5)	1.7 (0.6)	<0.001
LDL (mmol/L)*	3.5 (1.5)	3.2 (1.4)	3.6 (1.4)	<0.001
Morning cortisol (nmol/L)*	17 (16)	16 (14)	18 (17)	0.003
Evening cortisol (nmol/L)*	2 (3)	3 (3)	2 (3)	<0.001
C-reactive protein (mg/L)*	1.9 (2.9)	1.8 (2.6)	2.0 (3.1)	0.001
Triglycerides (mmol/L)*	1.1 (0.7)	1.0 (0.7)	1.1 (0.7)	<0.001
Fasting glucose (mg/dL)*	5.5 (0.8)	5.7 (0.9)	5.5 (0.8)	<0.001

Means and standard deviations or numbers and percentages are shown, else for * indicating median and interquartile range. MCI: mild cognitive impairment; APOE: apolipoprotein E; HDL: high-density lipoprotein; LDL: low-density lipoprotein. Imputed values are shown for: 6% missing data on education, 1% on metabolic syndrome, 1% on systolic and diastolic blood pressure, 1% on abdominal circumference, < 1% on HDL and LDL cholesterol, 9% on morning and evening cortisol, < 1% on c-reactive protein, < 1% on triglycerides, < 1% on glucose, < 1% on APOE ɛ4 genotype, 3% on smoking status, 4% on alcohol use, 12% on physical activity, 2% on stroke history, and 3% on antihypertensive use.

Based on model fit, the four-class model for both men and women was chosen (see Supplemental Table 1). Entropy increased in the four-class model for both men and women, and then decreased; therefore, we chose for the smallest amount of profiles with the best fit statistics to prevent overfitting. This also allowed for direct comparison between AL profiles across sexes. Fit statistics seemed slightly better in men compared to women (Supplemental Table 1). ANOVAs revealed a significant interaction between sex and profile membership on standardized systolic blood pressure (F = 27.22, p < 0.001), pulse pressure (F = 33.64, p < 0.001), abdominal circumference (F = 37.19, p < 0.001), HDL (F = 78.32, p < 0.001), LDL (F = 5.81, p = 0.003), CRP (F = 9.22, p < 0.001), triglycerides (F = 33.18, p < 0.001), and glucose (F = 37.25, p < 0.001). However, there was no significant interaction between sex and profile membership on morning cortisol (F = 2.51, p = 0.08) or evening cortisol (F = 3.53, p = 0.01). Illustration of the averaged z-scores per AL factor, profile, and sex can be found in Figure 1. In comparison to the profiles found in the total population, larger discrepancies between them were found with the profiles found in women compared to those found in men. Standardized means and confidence intervals per AL factor and profile in the total population and in women and men separately are found in Supplemental Figure 1. Proportions of individuals above clinical cut-offs per AL factor and profile are shown in Supplemental Table 2.

Figure 1.

Averaged AL factor z-score and 95% confidence intervals per sex-stratified profile. SBP: systolic blood pressure; DBP: diastolic blood pressure; PP: pulse pressure; HDL: high density lipoprotein; LDL: low-density lipoprotein; TG: triglycerides; CRP: c-reactive protein. Panels are on different scales for visualization purposes.

AL differences between profiles in women

In women, the profile with the highest proportion we labeled ‘Risk factors’ (46%), as this group had increased levels of risk factors (i.e., having metabolic syndrome, prevalent type 2 diabetes, taking anti-hypertensives, taking anti-depressants) compared to the ‘Average AL’ profile that had the second highest proportion (32%). The ‘Average AL’ profile was labeled as such as it had the least proportion of individuals with AL factors above their clinical cut-offs (Supplemental Table 2). The third profile we labeled ‘High cardiovascular dysregulation’ (19%), as it was characterized by high systolic blood pressure and pulse pressure (Table 2). The fourth profile we labeled ‘Multisystem dysregulation’ (3%), as it was characterized by higher levels of abdominal circumference, evening cortisol, CRP, triglycerides, and glucose. Further, lower morning cortisol was also seen in this profile (Table 3). All allostatic load factors differed between all four profiles in women.

Table 2.

Baseline characteristics in the study sample stratified by AL profile, in women (n = 3097).

	Average AL (n = 980, 32%)	Risk factors (n = 1406, 45%)	High cardiovascular dysregulation (n = 614, 20%)	Multisystem (n = 97, 3%)	p
Age (y)	77 (6)	76 (6)	79 (6)	76 (5)	<0.001
Education, college & university	284 (29%)	299 (21%)	122 (20%)	19 (20%)	<0.001
Current smoker	132 (13%)	194 (14%)	54 (9%)	12 (12%)	0.06
Alcohol use (g/week)*	2 (10)	2 (8)	2 (8)	0 (3)	0.01
Physical activity, moderate/high	333 (34%)	384 (27%)	161 (26%)	25 (26%)	<0.001
History of stroke	43 (4%)	99 (7%)	45 (7%)	6 (6%)	0.12
MCI at baseline	86 (9%)	127 (9%)	51 (8%)	11 (11%)	0.83
Metabolic syndrome	38 (4%)	671 (48%)	226 (37%)	88 (91%)	<0.001
Diabetes	24 (2%)	130 (9%)	59 (10%)	95 (98%)	<0.001
Antihypertensive medication	384 (39%)	728 (52%)	394 (64%)	65 (67%)	<0.001
Antidepressant medication	147 (15%)	280 (20%)	82 (13%)	21 (22%)	0.001
APOE ɛ4 genotype	267 (27%)	359 (26%)	177 (29%)	20 (21%)	0.15
Systolic blood pressure (mmHg)*	139 (20)	131 (20)	169 (19)	144 (25)	<0.001
Pulse pressure (mmHg)*	67 (18)	60 (19)	95 (16)	73 (23)	<0.001
Abdominal circumference (cm)*	91 (14)	104 (15)	99 (15)	106 (17)	<0.001
HDL (mmol/L)*	2.1 (0.5)	1.5 (0.4)	1.6 (0.5)	1.4 (0.5)	<0.001
LDL (mmol/L)*	3.5 (1.3)	3.8 (1.4)	3.7 (1.5)	2.9 (1.5)	<0.001
Morning cortisol (nmol/L)*	19 (17)	17 (16)	16 (17)	16 (20)	<0.001
Evening cortisol (nmol/L)*	2 (2)	2 (3)	3 (3)	3 (4)	<0.001
C-reactive protein (mg/L)*	1.2 (1.6)	2.6 (3.8)	2.2 (3.1)	3.3 (4.0)	<0.001
Triglycerides (mmol/L)*	0.8 (0.3)	1.4 (0.7)	1.1 (0.6)	1.8 (1.1)	<0.001
Fasting glucose (mg/dL)*	5.2 (0.5)	5.6 (0.8)	5.5 (0.7)	9.3 (2.3)	<0.001

Table 3.

Associations between AL profiles in women and dementia.

	No. of cases	All-cause dementia	No. of cases	AD dementia	No. of cases	Other dementias
Model 1
Average AL	229	Ref.	112	Ref.	117	Ref.
Risk factors	275	0.87 (0.73; 1.03)	120	0.75 (0.58; 0.98)	155	0.97 (0.76; 1.23)
High cardiovascular	154	0.99 (0.81; 1.22)	67	0.92 (0.67; 1.25)	87	1.10 (0.83; 1.46)
Multisystem	23	1.33 (0.87; 2.04)	12	1.32 (0.73; 2.41)	11	1.35 (0.73; 2.51)
Model 2
Average AL	229	Ref.	112	Ref.	117	Ref.
Risk factors	275	0.83 (0.69; 0.99)	120	0.75 (0.58; 0.98)	155	0.91 (0.71; 1.16)
High cardiovascular	154	0.98 (0.79; 1.21)	67	0.90 (0.66; 1.23)	87	1.07 (0.80; 1.43)
Multisystem	23	1.44 (0.94; 2.23)	12	1.48 (0.81; 2.71)	11	1.41 (0.75; 2.63)

AL: allostatic load; AD: Alzheimer's disease, model 1 is adjusted for age and education, model 2 is adjusted for smoking, alcohol use, physical activity, history of a stroke, hypertension, antidepressant use, and APOE ɛ4 genotype.

For demographic factors, highest age was seen in the ‘High cardiovascular dysregulation’ profile, highest education in the ‘Average AL’ group, lowest alcohol use in the ‘Multisystem dysregulation’ profile, highest physical activity in the ‘Average AL’ profile, highest prevalence of metabolic syndrome in the ‘Multisystem dysregulation’ profile, highest prevalence of type 2 diabetes in the ‘Multisystem dysregulation’ profile, and highest use of antidepressant medication in the ‘Multisystem dysregulation’ profile. Smoking, stroke prevalence, mild cognitive impairment at baseline, and APOE ɛ4 genotype did not differ between the four profiles (Table 2).

AL differences between profiles in men

The profile in men with the highest prevalence (44%) we labeled ‘Low cardiovascular dysregulation’, as it was characterized by low systolic blood pressure and low pulse pressure compared to the other profiles (Table 4). The profile with the second highest prevalence (41%) we labeled ‘Average AL’, as it was characterized by average values across all AL variables (Supplemental Figure 2, Table 3). The third profile we labeled ‘High cardiovascular dysregulation’ (11%), as it was characterized by high systolic blood pressure and pulse pressure. Lastly, there was a profile we labeled ‘Multisystem dysregulation’ (4%) that was characterized by higher abdominal circumference, high triglycerides, and high glucose. Morning cortisol, evening cortisol, and CRP did not differ between the four profiles in men (Table 4).

Table 4.

Baseline characteristics in the study sample stratified by AL profile, in men (n = 2246).

	Average AL (n = 898, 40%)	Low cardiovascular (n = 1010, 45%)	High cardiovascular dysregulation (n = 246, 11%)	Multisystem (n = 92, 4%)	p
Age (years)	77 (5)	76 (6)	79 (6)	76 (6)	<0.001
Education, college & university	295 (33%)	326 (32%)	75 (30%)	30 (33%)	0.03
Current smoker	94 (10%)	133 (13%)	21 (9%)	12 (13%)	0.01
Alcohol use (g/week)*	6 (26)	6 (26)	10 (27)	6 (25)	0.05
Physical activity, moderate/high	317 (35%)	360 (36%)	82 (33%)	32 (35%)	0.26
History of stroke	67 (7%)	66 (7%)	14 (6%)	8 (9%)	0.81
MCI at baseline	116 (13%)	114 (11%)	33 (13%)	12 (13%)	0.78
Metabolic syndrome	259 (29%)	247 (24%)	81 (33%)	77 (84%)	<0.001
Diabetes	122 (14%)	98 (10%)	51 (21%)	92 (100%)	<0.001
Antihypertensive medication	419 (47%)	325 (32%)	171 (70%)	53 (58%)	<0.001
Antidepressant medication	100 (11%)	105 (10%)	19 (8%)	12 (13%)	0.49
APOE ɛ4 genotype	280 (31%)	287 (28%)	54 (22%)	27 (29%)	0.05
Systolic blood pressure (mmHg)*	150 (13)	128 (14)	178 (17)	143 (22)	<0.001
Pulse pressure (mmHg)*	73 (12)	54 (11)	98 (12)	66 (19)	<0.001
Abdominal circumference (cm)*	102 (13)	101 (13)	101 (14)	109 (16)	<0.001
HDL (mmol/L)*	1.4 (0.5)	1.4 (0.5)	1.4 (0.5)	1.1 (0.4)	<0.001
LDL (mmol/L)*	3.3 (1.4)	3.3 (1.3)	3.1 (1.4)	3.0 (1.4)	0.01
Morning cortisol (nmol/L)*	16 (13)	16 (15)	17 (15)	18 (17)	0.77
Evening cortisol (nmol/L)*	3 (3)	3 (3)	3 (3)	3 (3)	0.25
C-reactive protein (mg/L)*	1.8 (2.6)	1.8 (2.6)	1.9 (2.5)	1.8 (2.5)	0.92
Triglycerides (mmol/L)*	1.0 (0.6)	1.0 (0.6)	1.1 (0.7)	1.7 (1.2)	<0.001
Fasting glucose (mg/dL)*	5.7 (0.8)	5.6 (0.7)	5.8 (0.9)	9.6 (2.6)	<0.001

With regards to demographics, age was highest in the ‘High cardiovascular dysregulation’ profile, education levels were highest in the ‘Multisystem dysregulation’ profile, smoking levels were highest in the ‘Multisystem dysregulation’ profile, metabolic syndrome and diabetes mellitus type 2 prevalence were highest in the ‘Multisystem dysregulation’ profile, and anti-hypertensive use was highest in the ‘High cardiovascular dysregulation’ profile. Alcohol use, physical activity, stroke prevalence, mild cognitive impairment at baseline, antidepressant medication use, and APOE ɛ4 genotype did not differ between the four groups (Table 4).

Sex-specific AL profiles on dementia risk

For the AL profiles created among women, only the ‘Risk Factors’ group was associated with a lower risk of dementia, only in the fully-adjusted model (HR 0.83; 95% CI 0.69; 0.99, p = 0.04). For AD dementias, the ‘Risk Factors’ profile was associated with a decreased risk of AD dementias (HR 0.75; 95% CI 0.58; 0.98, p = 0.03) and remained after further adjustment for covariates (HR 0.75; 95% CI 0.58; 0.98, p = 0.03). For other dementias, no profiles were associated with increased or decreased risk, even after further adjustment for covariates (Table 3).

For the AL profiles in men, the ‘Multisystem dysregulation’ profile was associated with an increased risk of all-cause dementia (HR 1.75; 95% CI 1.06; 2.90, p = 0.03), compared to the ‘Average AL’ profile. After the adjustment of multiple confounders, the association remained (HR 1.75; 95% CI 1.06; 2.90, p = 0.03). No other profiles were associated with all-cause dementia. No profiles were associated with AD dementia, also after further adjustment for covariates. For other dementias, the ‘Multisystem dysregulation’ profile was associated with an increased risk of other dementias (HR 2.33; 95% CI 1.24; 4.38, p = 0.01), which remained after further adjustment of covariates (HR 2.23; 95% CI 1.18; 4.21, p = 0.01). No other profiles were associated with other dementias (see Table 5).

Table 5.

Associations between AL profiles in men and dementia.

	No. of cases	All-cause dementia	No. of cases	AD dementia	No. of cases	Other dementias
Model 1
Average AL	167	Ref.	77	Ref.	90	Ref.
Low cardiovascular	186	1.17 (0.95; 1.44)	84	1.17 (0.86; 1.60)	102	1.15 (0.87; 1.53)
High cardiovascular	48	0.93 (0.67; 1.28)	14	0.64 (0.36; 1.13)	34	1.18 (0.79; 1.75)
Multisystem	17	1.73 (1.05; 2.86)	6	1.34 (0.58; 3.08)	11	2.33 (1.24; 4.38)
Model 2
Average AL	167	Ref.	77	Ref.	90	Ref.
Low cardiovascular	186	1.16 (0.94; 1.43)	84	1.17 (0.85; 1.60)	102	1.18 (0.88; 1.57)
High cardiovascular	48	1.01 (0.73; 1.40)	14	0.67 (0.37; 1.20)	34	1.24 (0.83; 1.85)
Multisystem	17	1.75 (1.06; 2.90)	6	1.44 (0.62; 3.34)	11	2.23 (1.18; 4.21)

When assessing competing risks, we found an increased risk for the ‘Multisystem dysregulation’ group in both women and men for dementia-free mortality (Supplemental Table 3 and Supplemental Table 4). When looking at the interaction between AL profiles and APOE ɛ4 genotype, there was no evidence for interaction in women or men on AL profile and APOE ɛ4 genotype (Supplemental Table 5).

Discussion

The current study aimed to explore sex differences in clustering AL as well as its relationship to incident dementia. All AL variables discriminated the clusters in women, whereas in men, only cardiovascular, lipid, and metabolic factors discriminated between the clusters. Further, in men the ‘Multisystem dysregulation’ group was associated with almost a two-times higher risk of incident all-cause dementia. When clustering AL in both men and women together, this profile was also associated with increased risk of incident dementia.¹⁴ However, when assessing sex-specific AL clusters, this profile was only associated with an increased risk in men, highlighting the importance of assessing sex differences in dementia. Lastly, in women, a decreased risk was seen in the ‘Risk factors’ profile, specifically for AD dementia.

A recent systematic review found that men may have higher total AL than women,¹⁶ but it was unclear which specific aspects of AL may drive this sex difference. The current study found dominant discrimination of endocrine and immune factors when clustering in women, whereas cardiovascular and metabolic factors were the most important factors in men during clustering. These findings are in line with a recent narrative review by Longpré-Poirier, Dougoud,⁴¹ that suggested these sex differences in AL as well. This highlights the need for classification of AL through clustering rather than using sum scores or predefined cut-offs when assessing sex differences.

The current study found a sex-specific effect solely of the ‘Multisystem dysregulation’ group on all-cause dementia in men only. While a previous study assessed sex differences in individual risk factors for dementia,¹⁵ the AL factors assessed in that study (i.e., cholesterol and abdominal circumference), that were increased in the ‘Multisystem dysregulation’ group, showed no differences in sex on incident dementia. This discrepancy could be explained by the clustering of risk factors in the current study. The clustering of AL factors into risk profiles could more accurately reflect multimorbidity and dementia risk.⁴² It is important to note that almost everyone in this group had type 2 diabetes, and the increased risk on incident dementia could also be explained by the presence of type 2 diabetes. When adding type 2 diabetes as an additional confounder to the fully-adjusted model, this indeed drove the relationship between the ‘Multisystem dysregulation’ group and incident dementia. The previous studies on sex differences regarding diabetes and dementia have thus far been mixed.^15,43,44 A previous study⁴³ found a difference between early- and late-onset of diabetes, with increased risk seen during midlife onset compared to later-life onset of type 2 diabetes. Lastly, our competing risk analysis found an increased risk of dementia-free mortality in the ‘Multisystem dysregulation’ group in women. This is in line with a previous meta-analysis showing an increased relative risk of mortality in women with type 2 diabetes compared to men.⁴⁵ There is a possibility that women in this risk profile had a lower life expectancy, and therefore died before receiving a dementia diagnosis.

Lastly, the current study found a decreased risk of AD dementia in the ‘Risk factors’ profile in women compared to the ‘Average AL’ profile. While many women in this profile showed multiple risk factors, such as being on anti-hypertensive medication, their blood pressure was comparable to the ‘Average AL’ group. Previous literature has shown that antihypertensive medication use may be protective for dementia.⁴⁶ There is a possibility that the anti-hypertensive medication is acting as a protective factor for incident dementia in this group. Previous literature has also found that women show a protection against the negative effects of hypertension and metabolic dysregulation, which may be explained by sex differences in the renin-angiotensin system.⁴⁷ Further, a large majority of the group also was physically active, which could also be preventing dementia.⁴⁸ Lastly, we speculate that this group had other cognitive resilience factors, such as certain social factors⁴⁹ or lifelong wealth,⁵⁰ that could have compensated for their biological risk factors and therefore reduced their overall risk. It would be of interest for future studies to assess the role of cognitive reserve when assessing sex differences of allostatic load on dementia. Previous studies have suggested that cognitive reserve may protect against allostatic load⁵¹ and may be further implicated by sex differences.⁵²

Strengths of the study include the LPA methodology to assess AL on a continuous scale, a large, community-based cohort of approximately equal proportion of men and women to assess sex differences, as well as a long follow-up time for incident dementia assessment. Further, we used imputation to handle missing data and include multiple confounders. However, the study also came with some limitations. The subtyping of other dementias was not done in all participants, particularly those diagnosed in nursing homes; therefore, our results on AD and other dementias cannot be inferred with categorical certainty. Further, the ‘Multisystem dysregulation’ profile consisted of a small proportion of the study population, and therefore may have been underpowered. Replication studies are therefore warranted. We also did not have information on diet, which could have been driving the null association and lowered triglycerides observed⁵³ in the ‘High cardiovascular’ group. Also, it is important to note that the AGES-Reykjavik Study is ethnically homogeneous. As psychological stress, AL, sex, and ethnicity have been shown to intersect, it is crucial to assess sex differences in AL in marginally underrepresented populations as well, who may be disproportionately affected by stressors.^54,55

The current study revealed distinct patterns in AL between women and men. Future studies assessing AL should prioritize including interactions with sex. Metabolic dysregulation, specifically in men, was associated with a higher risk of all-cause dementia. These findings advance our understanding of the biological underpinnings of sex differences in dementia and highlight metabolic disruption in men as a potential target for early intervention and prevention strategies.

Supplemental Material

sj-docx-1-alz-10.1177_13872877251375944 - Supplemental material for Sex differences in allostatic load profiles and incident dementia: The AGES-Reykjavik Study

Supplemental material, sj-docx-1-alz-10.1177_13872877251375944 for Sex differences in allostatic load profiles and incident dementia: The AGES-Reykjavik Study by Emma L Twait, Lotte Gerritsen, Vilmundur Gudnason, Lenore J Launer and Mirjam I Geerlings in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

We acknowledge the employees of the Icelandic Heart Preventive Clinic for their contribution to data collection as well as the participants of the AGES-Reykjavik Study.

ORCID iD

Mirjam I Geerlings

Ethical considerations

The Icelandic Data Protection Authority, the Icelandic National Bioethics Committee (VSN: 00-063), and the Institutional Review Board for the National Institute on Aging, NIH approved this study.

Consent to participate

Written informed consent was obtained from all participants.

Consent for publication

Not applicable

Author contributions

Emma L Twait: Conceptualization; Formal analysis; Visualization; Writing – original draft.

Lotte Gerritsen: Methodology; Writing – review & editing.

Vilmundur Gudnason: Project administration; Writing – review & editing.

Lenore J Launer: Project administration; Writing – review & editing.

Mirjam I Geerlings: Methodology; Supervision; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The AGES-Reykjavik study was funded by the Icelandic Heart Association, National Institute of Aging contracts (N01-AG-12100 and HHSN271201200022C), the Intramural Program at National Institute of Aging, and Althingi (the Icelandic Parliament). This study was supported by grants from Alzheimer Nederland (WE.03-2017-06, PI Geerlings) and (WE.03.2021-09, PI Geerlings). This work is also part of the BIRD-NL consortium funded by the Dutch Medical Research Council (ZonMw) as part of the National Dementia Strategy 2021–2030 by the Dutch Ministry of Health, Welfare and Sport (grant number: 1051003210005).

Declaration of conflicting interests

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

Data availability statement

Data from the AGES-Reykjavik study are available through collaboration (AGES_data_request@hjarta.is) under a data usage agreement with the IHA.

Supplemental material

Supplemental material for this article is available online.

References

Livingston

Huntley

Sommerlad

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

McGrath

Beiser

DeCarli

, et al. Blood pressure from mid- to late life and risk of incident dementia. Neurology 2017; 89: 2447–2454.

Walker

Sharrett

, et al. Association of midlife to late-life blood pressure patterns with incident dementia. JAMA 2019; 322: 535–545.

Qureshi

Collister

Allen

, et al. Association between metabolic syndrome and risk of incident dementia in UK biobank. Alzheimers Dement 2024; 20: 447–458.

Xue

, 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.

Gong

Harris

Peters

SAE

, et al. Serum lipid traits and the risk of dementia: a cohort study of 254,575 women and 214,891 men in the UK biobank. EClinicalMedicine 2022; 54: 101695.

Fernandes

Tábuas-Pereira

Duro

, et al. C-reactive protein as a predictor of mild cognitive impairment conversion into Alzheimer's disease dementia. Exp Gerontol 2020; 138: 111004.

Feng

Wang

Zeng

, et al. Predictors of cognitive decline in older individuals without dementia: an updated meta-analysis. Ann Clin Transl Neurol 2023; 10: 497–506.

Ouanes

Popp

. High cortisol and the risk of dementia and Alzheimer's disease: a review of the literature. Front Aging Neurosci 2019; 11: 43.

10.

McEwen

. Physiology and neurobiology of stress and adaptation: central role of the brain. Physiol Rev 2007; 87: 873–904.

11.

Adedeji

Holleman

Juster

R-P

, et al. Longitudinal study of Alzheimer's disease biomarkers, allostatic load, and cognition among memory clinic patients. Brain Behav Immun Health 2023; 28: 100592.

12.

Ferretti

Martinkova

Biskup

, et al. Sex and gender differences in Alzheimer's disease: current challenges and implications for clinical practice: position paper of the dementia and cognitive disorders panel of the European academy of neurology. Eur J Neurol 2020; 27: 928–943.

13.

Aggarwal

Mielke

. Sex differences in Alzheimer's disease. Neurol Clin 2023; 41: 343–358.

14.

Twait

Basten

Gerritsen

, et al. Late-life depression, allostatic load, and risk of dementia: the AGES-Reykjavik study. Psychoneuroendocrinology 2023; 148: 105975.

15.

Gong

Harris

Lipnicki

, et al. Sex differences in dementia risk and risk factors: individual-participant data analysis using 21 cohorts across six continents from the COSMIC consortium. Alzheimers Dement 2023; 19: 3365–3378.

16.

Kerr

Kheloui

Rossi

, et al. Allostatic load and women's brain health: a systematic review. Front Neuroendocrinol 2020; 59: 100858.

17.

Ali

Nitschke

Cooperman

, et al. Systematic manipulations of the biological stress systems result in sex-specific compensatory stress responses and negative mood outcomes. Neuropsychopharmacology 2020; 45: 1672–1680.

18.

Mair

Cutchin

Kristen Peek

. Allostatic load in an environmental riskscape: the role of stressors and gender. Health Place 2011; 17: 978–987.

19.

D'Amico

Amestoy

Fiocco

. The association between allostatic load and cognitive function: a systematic and meta-analytic review. Psychoneuroendocrinology 2020; 121: 104849.

20.

Juster

McEwen

Lupien

. Allostatic load biomarkers of chronic stress and impact on health and cognition. Neurosci Biobehav Rev 2010; 35: 2–16.

21.

Harris

Launer

Eiriksdottir

, et al. Age, gene/environment susceptibility–Reykjavik Study: multidisciplinary applied phenomics. Am J Epidemiol 2007; 165: 1076–1087.

22.

Sigurdsson

Aspelund

Kjartansson

, et al. Incidence of brain infarcts, cognitive change, and risk of dementia in the general population: the AGES-Reykjavik Study (Age Gene/Environment Susceptibility-Reykjavik Study). Stroke 2017; 48: 2353–2360.

23.

Saczynski

Sigurdsson

Jonsdottir

, et al. Cerebral infarcts and cognitive performance: importance of location and number of infarcts. Stroke 2009; 40: 677–682.

24.

Qiu

Cotch

Sigurdsson

, et al. Cerebral microbleeds, retinopathy, and dementia: the AGES-Reykjavik Study. Neurology 2010; 75: 2221–2228.

25.

Jørgensen

el Kholy

Damkjaer

, et al. “RAI"–an international system for assessment of nursing home residents. Ugeskr Laeger 1997; 159: 6371–6376.

26.

Seeman

Epel

Gruenewald

, et al. Socio-economic differentials in peripheral biology: cumulative allostatic load. Ann N Y Acad Sci 2010; 1186: 223–239.

27.

Carbone

. Neighborhood perceptions and allostatic load: evidence from Midlife in the United States study. Health Place 2020; 61: 102263.

28.

Rodriquez

Kim

Sumner

, et al. Allostatic load: importance, markers, and score determination in minority and disparity populations. J Urban Health 2019; 96: 3–11.

29.

Forrester

Leoutsakos

Gallo

, et al. Association between allostatic load and health behaviours: a latent class approach. J Epidemiol Community Health 2019; 73: 340–345.

30.

Mauss

Jarczok

Fischer

. The streamlined Allostatic Load Index: a replication of study results. Stress 2016; 19: 553–558.

31.

Smith

Maloney

Falkenberg

, et al. An angiotensin-1 converting enzyme polymorphism is associated with allostatic load mediated by C-reactive protein, interleukin-6 and cortisol. Psychoneuroendocrinology 2009; 34: 597–606.

32.

Mijnarends

Koster

Schols

, et al. Physical activity and incidence of sarcopenia: the population-based AGES-Reykjavik study. Age Ageing 2016; 45: 614–620.

33.

Gudnason

Sigurdsson

Steingrimsdottir

, et al. Association of apolipoprotein E polymorphism with plasma levels of high density lipoprotein and lipoprotein (a), and effect of diet in healthy men and women. Nutr Metab Cardiovasc Dis 1993; 3: 136–136.

34.

Lopez

Becker

Jagust

, et al. Neuropsychological characteristics of mild cognitive impairment subgroups. J Neurol Neurosurg Psychiatry 2006; 77: 159–165.

35.

Zimmet

. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 1998; 15: 539–553.

36.

Auðunsson

Elíasson

Steingrímsson

, et al. Diffuse idiopathic skeletal hyperostosis in elderly Icelanders and its association with the metabolic syndrome: the AGES-Reykjavik study. Scand J Rheumatol 2021; 50: 314–318.

37.

Muthén

. Mplus user’s guide: Statistical analysis with latent variables, user’s guide. Los Angeles, LA: Muthén & Muthén, 2017.

38.

Rosenberg

Beymer

Anderson

, et al. tidyLPA: an R package to easily carry out Latent Profile Analysis (LPA) using open-source or commercial software. J Open Source Softw 2018; 3: 978.

39.

Hallquist

Wiley

. Mplusautomation: an R package for facilitating large-scale latent variable analyses in M plus. Struct Equ Modeling 2018; 25: 621–638.

40.

Ferguson

Moore

EWG

Hull

. Finding latent groups in observed data: a primer on latent profile analysis in Mplus for applied researchers. Int J Behav Dev 2020; 44: 458–468.

41.

Longpré-Poirier

Dougoud

Jacmin-Park

, et al. Sex and gender and allostatic mechanisms of cardiovascular risk and disease. Can J Cardiol 2022; 38: 1812–1827.

42.

Chen

Zhou

Huang

, et al. Multimorbidity burden and developmental trajectory in relation to later-life dementia: a prospective study. Alzheimers Dement 2023; 19: 2024–2033.

43.

Zhou

Dong

Xie

, et al. Sex-specific associations between diabetes and dementia: the role of age at onset of disease, insulin use and complications. Biol Sex Differ 2023; 14: 9.

44.

Moran

Gilsanz

Beeri

, et al. Sex, diabetes status and cognition: findings from the study of longevity in diabetes. BMJ Open Diabetes Res Care 2021; 9: e001646.

45.

Wang

O’Neil

Jiao

, et al. Sex differences in the association between diabetes and risk of cardiovascular disease, cancer, and all-cause and cause-specific mortality: a systematic review and meta-analysis of 5,162,654 participants. BMC Med 2019; 17: 136.

46.

Ding

Davis-Plourde

Sedaghat

, et al. Antihypertensive medications and risk for incident dementia and Alzheimer's disease: a meta-analysis of individual participant data from prospective cohort studies. Lancet Neurol 2020; 19: 61–70.

47.

White

Fleeman

Arnold

. Sex differences in the metabolic effects of the renin-angiotensin system. Biol Sex Differ 2019; 10: 31.

48.

Zhang

Cong

, et al. Effect of physical activity on risk of Alzheimer's disease: a systematic review and meta-analysis of twenty-nine prospective cohort studies. Ageing Res Rev 2023; 92: 102127.

49.

Duffner

Deckers

Cadar

, et al. Social relationship factors, depressive symptoms, and incident dementia: a prospective cohort study into their interrelatedness. Psychol Med 2024; 54: 3115–3125.

50.

Ellwood

Quinn

Mountain

. Psychological and social factors associated with coexisting frailty and cognitive impairment: a systematic review. Res Aging 2022; 44: 448–464.

51.

García-Moreno

Cañadas-Pérez

García-García

, et al. Cognitive reserve and anxiety interactions play a fundamental role in the response to the stress. Front Psychol 2021; 12: 673596.

52.

Arenaza-Urquijo

Boyle

Casaletto

, et al. Sex and gender differences in cognitive resilience to aging and Alzheimer's disease. Alzheimers Dement 2024; 20: 5695–5719.

53.

Gjuladin-Hellon

Davies

Penson

, et al. Effects of carbohydrate-restricted diets on low-density lipoprotein cholesterol levels in overweight and obese adults: a systematic review and meta-analysis. Nutr Rev 2019; 77: 161–180.

54.

Moore

Bevel

Aslibekyan

, et al. Temporal changes in allostatic load patterns by age, race/ethnicity, and gender among the US adult population; 1988-2018. Prev Med 2021; 147: 106483.

55.

Beckie

. A systematic review of allostatic load, health, and health disparities. Biol Res Nurs 2012; 14: 311–346.

Supplementary Material

Please find the following supplemental material available below.

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

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

0.25 MB

0.00 MB