Archives of Gerontology and Geriatrics 105 (2023) 104853

Contents lists available at ScienceDirect

Archives of Gerontology and Geriatrics

journal homepage: www.elsevier.com/locate/archger

Cardiometabolic diseases, polygenic risk score, APOE genotype, and risk of

incident dementia: A population-based prospective cohort study
Yanchun Chen a, 1, Yuan Zhang a, 1, Shu Li b, Lihui Zhou a, Huiping Li a, Dun Li c,
Yaogang Wang a, c, Hongxi Yang d, *
a
School of Public Health, Tianjin Medical University, Tianjin, China
b
School of Management, Tianjin University of Traditional Chinese Medicine, Tianjin, China
c
The Discipline of Integrative Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin, China
d
Department of Bioinformatics, School of Basic Medical Sciences, Tianjin Medical University, Tianjin, China

A R T I C L E I N F O A B S T R A C T

Keywords: Objective We aimed to prospective investigate the association between cardiometabolic diseases (CMDs) with
Cardiometabolic diseases dementia, and to examine whether genetic factors and CMDs jointly contribute to the incidence of dementia.
Genetic risk Methods We used data from the UK biobank of 204,646 adults aged 37-73 free of dementia at baseline.
Dementia
Genetic risk for dementia including APOE ε4 status and polygenic risk score (PRS) categorized as low, inter­
Prospective study
mediate, and high. CMDs including ischemic heart disease (IHD), stroke, and type 2 diabetes (T2D) were
confirmed by touchscreen questionnaires, medical examinations, and hospital inpatient records.
Results Over the follow-up (median: 12.5 years), 5,750 participants developed dementia. The HRs (95% CI) of
those with APOE ε4 carriers and high PRS were 3.16 (3.00-3.33) and 1.50 (1.41-1.60), respectively. The risk of
dementia was 70% higher among those with CMDs (HR: 1.70; 95% CI: 1.60-1.82). In joint effect analyses,
compared to no CMDs and APOE ε4 non-carriers, the HRs (95% CIs) of dementia were 3.53 (3.31-3.76)/2.06
(1.89-2.23) in participants with only APOE ε4 carriers and CMDs, and 5.06 (4.64-5.53) for those with APOE ε4
carriers plus CMDs. Compared to no CMDs and low PRS, the HRs (95% CIs) of dementia were 1.29 (1.19-1.40)/
1.60 (1.48-1.73) in participants with only intermediate and high PRS, and 2.00 (1.79-2.23)/2.63 (2.38-2.92) for
those with intermediate, and high PRS plus CMDs. Moreover, there were significant additive and multiplication
interactions between CMDs and APOE ε4 carriers of dementia, but only multiplication interaction was observed
for PRS.
Conclusions CMDs were associated with higher risk of dementia regardless of genetic risk for dementia.

1. Introduction underlying dementia, there is a great deal of interest in identifying

potentially modifiable risk factors, including chronic diseases, to pre­
Dementia is one of the leading causes of disability and death in later vent and manage dementia.
life. With the development of the global population aging, the preva­ Both genetic and environmental factors contribute to the develop­
lence of dementia has been increasing. It was estimated that approxi­ ment of dementia (Gatz et al., 2006). In the past decade, several
mately 47 million people suffer from dementia worldwide in 2015, and genome-wide association studies (GWASs) have identified over 30
this number will almost triple reaching 152 million in 2050 (Livingston genomic loci associated with the risk of dementia (Bis et al., 2020;
et al., 2020). Dementia, including Alzheimer’s disease (AD) and vascular Escott-Price et al., 2015; Hollingworth et al., 2011; Lambert et al., 2013;
dementia (VaD), is a complex, multisystemic disease with no curative Seshadri et al., 2010). Among them, the apolipoprotein ε4 (APOE ε4)
treatment and a heavy disease burden so far. The global societal cost of allele is the strongest genetic risk factor associated with AD (Kunkle
dementia is expected to reach as high as 2 trillion US dollars in 2030 et al., 2019). Studies had shown that APOE ε4 carriers had an estimated
(Wimo et al., 2017). Given the rapid pathophysiological progress risk of Alzheimer’s disease over 50%, compared with less than 10% for

* Corresponding author.
E-mail address: [email protected] (H. Yang).
1
These authors contributed equally to this manuscript.

https://doi.org/10.1016/j.archger.2022.104853
Received 7 September 2022; Received in revised form 27 October 2022; Accepted 29 October 2022
Available online 31 October 2022
0167-4943/© 2022 Elsevier B.V. All rights reserved.
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

those with APOE ε4 non-carriers (Genin et al., 2011). Therefore, it is Physical activity was self-reported according to the International Phys­
necessary to take separately APOE ε4 allele and polygenic risk score ical Activity Questionnaire (IPAQ) and classified as low, middle, and
(PRS) by combining other multiple risk alleles into account the risk of high. A healthy diet pattern was formulated from the Food Frequency
dementia and its subtypes, which will be predictive of incident dementia Questionnaire composed of at least 4 of the 7 elements, including fruits:
and provide quantitative measures of genetic susceptibilities. ≥3 servings/day, vegetables: ≥3 servings/day, fish: ≥2 servings/week,
Cardiometabolic diseases (CMDs) consist of a combination of disor­ processed meats: ≤1 serving/week, unprocessed red meats: ≤1.5 ser­
ders, such as obesity, hypertension, atherogenic dyslipidemia, and vings/week, whole grains: ≥3servings/day, refined grains: ≤1.5 ser­
abnormal glucose tolerance, all of which together lead to type 2 diabetes vings/day, details were provided in Appendix Table 1. Body mass index
(T2D) and cardiovascular disease (CVD) (Guo et al., 2014; Kim et al., (BMI) was obtained from the physical measurement and was calculated
2018). CMDs were defined as at least one of the three chronic diseases as weight divided by height squared and was classified into <25 kg/m2,
including stroke, ischemic heart disease (IHD), and T2D (Di Ange­ 25 to <30 kg/m2, and ≥30 kg/m2. Information on hypertension history
lantonio et al., 2015; Han et al., 2021; Qiao et al., 2021). The prevalence was defined as systolic blood pressure ≥140 mm Hg, diastolic blood
of any CMDs is about 20% and has become a comorbidity with subse­ pressure ≥90 mm Hg, or a diagnosis by doctors at baseline. Blood
quent dementia in the elderly (Keenan et al., 2016; Kivimaki et al., samples included high-density lipoprotein (HDL), low-density lipopro­
2017). Accordingly, there are emerging studies have shown that single tein (LDL), and C-reactive protein (CRP) were also collected at baseline.
and multiple CMDs are associated with dementia (Lyall et al., 2017;
Wang et al., 2020). A prospective study conducted in Swedish National 2.3. Assessment of dementia
Study on Aging and Care revealed that only one CMD was related to a
41% higher risk of dementia, whereas any two CMDs and three CMDs Incident dementia and its subtypes were ascertained using hospital
confer a 2.38-fold and 4.76-fold increased dementia risk, respectively admission records and death certificates linkage to Hospital Episode
(Wang et al., 2020). It is possible that CMDs have potential different Statistics (England), the Scottish Morbidity Record (Scotland), and the
effects in populations with different levels of genetic risk. However, to Patient Episode Database (Wales). International Classification of Dis­
our knowledge, no study has confirmed whether APOE ε4 and non-APOE eases (ICD-10) codes was used to define all-cause dementia (ICD-10
PRS interact with CMDs to modify the risk of dementia. codes: F00, F01, F02, F03, F05.1, G30, G31.1, G31.8), Alzheimer’s de­
In the present study, using a population-based longitudinal cohort of mentia (AD) (ICD-10 codes: F00, G30), and vascular dementia (VaD)
UK Biobank, we sought to (1) evaluate the associations of genetic factors (ICD-10 codes: F01).
(non-APOE polygenic risk for dementia and APOE ε4 carrier status) and
CMDs with the risk of dementia and its subtypes and (2) explore whether 2.4. Assessment of cardiometabolic diseases
and to what extent these associations are modified by genetic risk.
Cardiometabolic diseases (CMDs) were defined as stroke, ischemic
2. Methods heart disease (IHD), and type 2 diabetes (T2D). The diagnosis of stroke
and IHD was determined by self-report (Data-Field 20002, 6150), and
2.1. Study design and population medical records (ICD-10: I60-I69 for stroke, I20-I25 for IHD). T2D was
ascertained based on self-report (Data-Field 20002, code: 2443), medi­
The UK Biobank is an ongoing population-based cohort that was cal records (ICD-10: E11), glycated hemoglobin ≥6.5% (48 mmol/mol),
initiated from 2006 to 2010 in 22 assessment centers across the United and the use of anti-diabetic drugs.
Kingdom. More than 500,000 participants mostly aged 37–73 were
recruited and responded to complete touch-screen questionnaires and 2.5. Assessment of genetic risk factors
have physical measurements taken on extensive personal and de­
mographic data (Sudlow et al., 2015). Details of the study protocol have Genotyping was assayed using 2 very similar genotyping arrays, the
been published elsewhere (Collins, 2012). The present analysis was UK BiLEVE Axiom (Affymetrix) array and the UK Biobank Axiom array
limited to individuals aged ≥60 years at baseline (since most cases of (Applied Biosystems). Imputed genotypes were derived from the
dementia occur in the elderly) and individuals of white British ancestry Haplotype Reference Consortium and UK10K haplotype resource using
(genetic information available). From the 502,507 participants, we computationally efficient methods. In this study, genetic analysis was
excluded participants who were of non-white British ancestry (n = 27, conducted using version 3 published in March 2018. Genotyping quality
033), those who were younger than 60 (n = 264,394), those who had a control is centrally carried out by the UK Biobank (Bycroft et al., 2018).
diagnosis of dementia at baseline (n = 153), those who with missing data The APOE single-nucleotide polymorphisms (SNPs) were directly
on genetic risk (n = 5,906), and those who were lost to follow-up (n = genotyped by rs429358 and rs7412. We collated ε2/ε4, ε3/ε4, and ε4/ε4
375). Our final analyses included 204,646 participants (Appendix as APOE ε4 carriers and ε2/ε2, ε2/ε3, and ε3/ε3 as APOE ε4 non-
Fig. 1). carriers.
PRS was created based on GWASs to assess the cumulative genetic
2.2. Data collection risk for dementia. SNPs were selected from a previous study and the
summary-statistic data were based on individuals of European ancestry
At baseline, information on sociodemographic factors including sex, (Lourida et al., 2019). Therefore, the present study was restricted to
age, education levels, socioeconomic status, smoking status, alcohol white (British, Irish, or other white backgrounds) samples in the UK
intake frequency, physical activity, and diet pattern was collected using Biobank. The PRS was calculated across all SNPs excluding APOE ε4
a touch-screen questionnaire and interview. Education level was cate­ allele associated with dementia, for details regarding the selected SNPs
gorized as college or university degree, upper secondary, lower sec­ were provided in Appendix Table 2. To avoid inflated estimates, PRS for
ondary, vocational, and others. Socioeconomic status was derived from dementia were constructed using SNPs from external GWASs (Kunkle
the Townsend Deprivation Index (TDI) (Guillaume et al., 2016) of et al., 2019). SNPs were selected based on two criteria: the SNPs were
aggregated information on social class, employment, car availability, significantly associated with dementia risk at a genome-wide level (P
and housing and classified as low (highest quintile), middle (quintiles 2 value = 5 × 10–8); SNPs in high linkage disequilibrium (r2 > 0.2) were
to 4), and high (lowest quintile). Smoking status was defined as never, excluded. An individual-level PRS was determined as the sum of the
previous, and current smokers. Alcohol intake frequency was catego­ number of risk alleles present at each SNP weighted by the effect sizes
rized as never, special occasions only, one to three times a month, once derived from all available SNPs in the UK Biobank, and it was produced
or twice a week, three or four times a week, and daily or almost daily. by using the PLINK “–score” command. The PRS for participants were

2
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

Z-standardized and then divided into low (lowest tertile), intermediate Table 1
(tertile 2), and high (highest tertile) genetic risks. Baseline characteristics of the study population by cardiometabolic diseases
(CMDs) status.
2.6. Statistical analyses Characteristic CMDs status (n = 204646)
Total CMD-free CMDs P value
The baseline characteristics according to CMDs status were pre­ Total 204646 170531 34115
(83.3) (16.7)
sented as frequency (percentage) for categorical variables and means Age, mean (SD), year 64.13 (2.85) 64.02 (2.83) 64.72 <0.001
(standard deviations [SDs]) for continuous variables. We determined (2.86)
statistical differences between CMDs status and each characteristic using Female 107683 95485 12198 <0.001
t-tests and Chi-square tests if applicable. (52.6) (56.0) (35.8)
Education level
We first used Cox proportional hazards regression models to estimate <0.001
College/university 53315 46657 6658
the hazard ratios (HRs) and 95% confidence intervals (CIs) of genetic degree (26.1) (27.4) (19.5)
risk and CMDs with all-cause dementia, AD, and VaD. We used follow-up Upper secondary 18538 (9.1) 15833 (9.3) 2705 (7.9)
time as the timescale, which was considered at risk from the recruitment Lower secondary 47952 40724 7228
date to the first event of dementia and its subtypes, death, loss to follow- (23.4) (23.9) (21.2)
Vocational 14851 (7.3) 11847 (6.9) 3004 (8.8)
up, or September 30, 2021, whichever came first. The proportional Others 69990 55470 14520
hazards assumption was determined using the Schoenfeld residuals (34.2) (32.5) (42.6)
method and no violation was detected. Cox regression models were Socioeconomic statusa <0.001
adjusted for sex, age, education levels, socioeconomic status, BMI, 1(least deprived) 44969 38894 6075
(22.0) (22.8) (17.8)
smoking status, alcohol intake frequency, physical activity, diet pattern,
2-4 125947 105730 20217
hypertension, HDL, LDL, CRP, and the first 10 principal components of (61.5) (62.0) (59.3)
ancestry. Multiple imputations based on five replications using chained 5(most deprived) 33730 25907 7823
equations were used to handle missing covariates (Liao et al., 2014). (16.5) (15.2) (22.9)
Stratified analyses were performed to test whether the association be­ BMI (kg/m2) <0.001
59999 54573 5426
tween CMDs and the risk of dementia was modified by genetic predis­
<25
(29.3) (32.0) (15.9)
position to dementia. We also created category dummy variables to 25to<30 93088 78709 14379
quantify the magnitude of combined CMDs profiles and genetic predis­ (45.5) (46.2) (42.1)
position to dementia. Moreover, the multiplicative interactions between ≥30 51559 37249 14310
(25.2) (21.8) (41.9)
CMDs and genetic factors were examined by including an interaction
Smoking status <0.001
term reflecting both exposure and their cross-product term in the above Never 101073 87948 13125
model. Particularly, we assessed additive interactions by calculating the (49.4) (51.6) (38.5)
relative excess risk due to interaction (RERI), the attributable proportion Previous 86605 69071 17534
(AP), and the synergy index (SI). In addition, the Kaplan-Meier method (42.3) (40.5) (51.4)
Current 16968 (8.3) 13512 (7.9) 3456
was used to estimate the cumulative incidence of dementia and its (10.1)
subtypes during the 15 years of follow-up. Alcohol intake <0.001
To examine the robustness of our findings, we conducted a series of frequency
sensitivity analyses. First, to explore the influence of sex and age, we Never 48671 41778 6893
(23.8) (24.5) (20.2)
conducted stratification analyses by sex and age strata (late middle-aged
Special occasions only 45945 39400 6545
adults [60–64 years old] or young older adults [≥65 years old]). Second, (22.5) (23.1) (19.2)
to minimize the potential contribution of reverse causality to our results, One to three times a 49310 41170 8140
we repeated the main analyses by excluding participants who developed month (24.1) (24.1) (23.9)
dementia during the first 3 years of follow-up. Third, we included par­ Once or twice a week 20292 (9.9) 16753 (9.8) 3539
(10.4)
ticipants younger than 60 years at baseline in the main analyses to ac­
Three or four times a 24087 19088 4999
cess the effects of CMDs and genetic predisposition on the risk of week (11.8) (11.2) (14.7)
dementia for the whole population. Fourth, we performed analyses Daily or almost daily 16341 (8.0) 12342 (7.2) 3999
using variables without imputation for the missing data of covariates. (11.7)
Physical activity
Fifth, the broad definition of CMDs also includes obesity and hyper­ <0.001
Low 35513 27895 7618
tension (Luo et al., 2022; Mau et al., 2009). Therefore, we repeated the (17.4) (16.4) (22.3)
main analyses using the broad definition of CMDs including obesity, Moderate 85126 71166 13960
hypertension, stroke, IHD, and T2D. Finally, we used the competing risk (41.6) (41.7) (40.9)
proportional subdistribution hazards regression model proposed by Fine High 84007 71470 12537
(41.0) (41.9) (36.7)
and grey (Noordzij et al., 2013), in which death was the competing risk
Diet pattern 0.565
event. Unhealthy 110383 91933 18450
The statistical analyses were performed using STATA 15 statistical (53.9) (53.9) (54.1)
software (StataCorp) and R (version 3.6.3, R Foundation for Statistical Healthy 94263 78598 15665
(46.1) (46.1) (45.9)
Computing). Statistical significance was considered a 2-tailed P-
Hypertension <0.001
value<0.05. No 100270 93427 6843
(49.0) (54.8) (20.1)
3. Results Yes 104376 77104 27272
(51.0) (45.2) (79.9)
HDL, mean (SD), mmol/ 1.46 (0.39) 1.49 (0.39) 1.27 (0.35)
3.1. Baseline characteristics of the study population
<0.001
L
LDL, mean (SD), mmol/ 3.57 (0.91) 3.71 (0.86) 2.84 (0.83) <0.001
In total, 204,646 participants (64.1±2.9 years old, 53% female) were L
included in this analysis. Table 1 shows the baseline characteristics of CRP, mean (SD), mg/L 2.84 (4.69) 2.75 (4.56) 3.28 (5.27) <0.001
APOE ε4 status 0.005
participants according to CMDs status. Over 15 years of follow-up
(median [interquartile range] length of follow-up, 12.5 [11.6-13.2] (continued on next page)

3
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

Table 1 (continued ) participants with intermediate and high genetic risk, respectively. The
No 146892 122619 24273 HRs (95% CIs) of VaD were 1.24 (1.05-1.47) and 1.55 (1.32-1.82) in
(71.8) (71.9) (71.2) participants with intermediate and high genetic risk, respectively. The
Yes 57754 47912 9842 incidence of dementia was higher for participants with APOE ε4 carriers
(28.2) (28.1) (28.8) and high PRS (Appendix Figs. 2 and 3).
Dementia PRSb 0.272
Low 67327 56094 11233
The incidence rate and risk of dementia were also increased mono­
(32.9) (32.9) (32.9) tonically across CMDs status (Appendix Fig. 4). Participants with one or
Intermediate 68113 56652 11461 more CMDs were associated with a 1.70-fold higher risk of dementia
(33.3) (33.2) (33.6) (HR, 1.70; 95% CI: 1.60-1.82), 1.42-fold higher risk of AD (HR, 1.42;
High 69206 57785 11421
95% CI: 1.28-1.58), and 2.55-fold higher risk of VaD (HR, 2.55; 95% CI:
(33.8) (33.9) (33.5)
2.19-2.97) (Table 2). The HRs (95% CI) of dementia were 1.57 (1.47-
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided 1.68) for one, 2.53 (2.25-2.84) for two, and 4.68 (3.40-6.44) for three
by height in meters squared); HDL, high-density lipoprotein; LDL, low-density CMDs compared with participants with no CMD. The HRs (95% CI) of
lipoprotein; APOE ε4, apolipoprotein ε4 allele; PRS, polygenic risk score; CRP,
AD were 1.34 (1.20-1.50) for one, 1.88 (1.53-2.30) for two, and 3.75
C-reactive protein; SD, standard deviation.
a (2.12-6.66) for three CMDs. The HRs (95% CI) of VaD were 2.31 (1.97-
Socioeconomic status assessed with the Townsend deprivation index, which
2.71) for one, 3.88 (3.04-4.96) for two, and 9.40 (5.45-16.20) for three
combines information on social class, employment, car availability, and
housing. CMDs.
b
Dementia PRS defined according to a polygenic risk score excluding APOE ε4
as low (lowest quintile), intermediate (quintile 2 to 4), and high (highest 3.3. Joint effect of CMDs status and APOE ε4 status on dementia risk
quintile).
Fig. 1 presents the associations between dementia and the joint ex­
years; total person-years, 2,458,370), we documented 5,750 (2.8%) all- posures of CMDs status and APOE ε4 status. Compared to individuals
cause dementia, 2,432 (1.2%) AD, and 936 (0.5%) VaD. As compared with APOE ε4 non-carriers and no CMD, the HRs (95% CIs) of dementia
with participants free of CMDs, individuals with CMDs were more likely were 2.06 (1.89-2.25) in those with only CMDs, and 3.53 (3.31-3.76)
to be older, male, obese, and current smokers, more likely to have high and 5.06 (4.64-5.53) for those with CMDs plus APOE ε4 non-carriers or
frequency of alcohol intake, hypertension, high CRP, and APOE ε4 carriers, respectively. There were significant additive (RERI: 0.482, 95%
allele; less to have a university education level, high socioeconomic CI: 0.075-0.889; AP: 0.095, 95% CI: 0.020-0.169; SI: 1.134, 95% CI:
status, high physical activity, healthy diet pattern, high HDL, and high 1.022-1.258) and multiplication (P for interaction <0.001) interactions
LDL. between CMDs and APOE ε4 carriers of dementia, as well as VaD (Ap­
pendix Table 3). Additionally, we also observed gradients in the asso­
3.2. Association between dementia with genetic risk and CMDs ciations between the number of CMDs and APOE ε4 status on dementia
risk according to APOE ε4 status (Appendix Table 4). The Kaplan-Meier
Table 2 presents individual associations between dementia and ge­ survival curve showed that those with both CMDs and APOE ε4 carriers
netic risk (both APOE ε4 status and a dementia PRS). In multi-adjusted had the highest cumulative incidence of dementia (Fig. 2).
Cox regression models, participants with APOE ε4 carriers had a higher
risk for dementia (HR, 3.16; 95% CI: 3.00-3.33), AD (HR, 4.40; 95% CI: 3.4. Joint effect of CMDs status and PRS on dementia risk
4.05-4.78), and VaD (HR, 2.53; 95% CI: 2.22-2.88) compared to those
with APOE ε4 non-carriers. The risk of incident dementia was higher for Fig. 1 shows the association between dementia and the joint expo­
those with intermediate (HR, 1.19; 95% CI: 1.11-1.27) and high PRS sures of CMDs status and levels of PRS. Compared to individuals with
(HR, 1.50; 95% CI: 1.41-1.60) compared to those with a low PRS. The low PRS and no CMD, the HRs (95% CIs) of dementia were 1.29 (1.19-
HRs (95% CIs) of AD were 1.25 (1.12-1.39) and 1.75 (1.58-1.93) in 1.40) and 1.60 (1.48-1.73) in those with intermediate and high PRS plus

Table 2
Cox proportional-hazards models investigating the associations between CMDs status and genetic risks with incident dementia.
Factors Total Dementia AD VaD
Cases HR (95%CI) Cases HR (95%CI) Cases HR (95%CI)
APOE ε4 status
No 146,892 2,632 1.00 (Ref.) 910 1.00 (Ref.) 475 1.00 (Ref.)
Yes 57,754 3,118 3.16 (3.00-3.33) 1,522 4.40 (4.05-4.78) 461 2.53 (2.22-2.88)
PRS
Low 67,327 1,541 1.00 (Ref.) 602 1.00 (Ref.) 244 1.00 (Ref.)
Intermediate 68,113 1,864 1.19 (1.11-1.27) 765 1.25 (1.12-1.39) 309 1.24 (1.05-1.47)
High 69,206 2,345 1.50 (1.41-1.60) 1,065 1.75 (1.58-1.93) 383 1.55 (1.32-1.82)
CMDs status
No 170,531 3,988 1.00 (Ref.) 1,804 1.00 (Ref.) 526 1.00 (Ref.)
Yes 34,115 1,762 1.70 (1.60-1.82) 628 1.42 (1.28-1.58) 410 2.55 (2.19-2.97)
Only one 29,054 1,361 1.57 (1.47-1.68) 503 1.34 (1.20-1.50) 304 2.31 (1.97-2.71)
Stroke 3,733 199 1.80 (1.55-2.08) 58 1.19 (0.91-1.54) 65 3.99 (3.06-5.20)
IHD 13,709 617 1.41 (1.29-1.55) 232 1.22 (1.05-1.41) 126 1.90 (1.54-2.34)
T2D 11,612 545 1.71 (1.55-1.88) 213 1.60 (1.37-1.86) 113 2.26 (1.81-2.81)
Any two 4,705 362 2.53 (2.25-2.84) 113 1.88 (1.53-2.30) 92 3.88 (3.04-4.96)
Stroke+IHD 1,000 80 2.43 (1.94-3.04) 17 1.21 (0.75-1.96) 23 4.29 (2.80-6.59)
Stroke+T2D 550 49 3.11 (2.34-4.14) 13 1.96 (1.13-3.40) 18 6.81 (4.20-11.04)
IHD+T2D 3,155 233 2.47 (2.15-2.84) 83 2.12 (1.68-2.68) 51 3.19 (2.34-4.34)
Three (Stroke+IHD+T2D) 356 39 4.68 (3.40-6.44) 12 3.75 (2.12-6.66) 14 9.40 (5.45-16.20)

Models were adjusted for sex, age, education levels, socioeconomic status, body mass index (BMI), smoking status, alcohol intake frequency, physical activity, diet
pattern, hypertension, high-density lipoprotein (HDL), low-density lipoprotein (LDL), and C-reactive protein (CRP), and the first 10 principal components of ancestry.
Abbreviations: HR, hazard ratio; CI, confidence interval; CMDs, cardiometabolic diseases; AD, Alzheimer’s Disease; VaD, vascular dementia; APOE ε4, apolipoprotein
ε4 allele; PRS, polygenic risk score; IHD, ischemic heart disease; T2D, type 2 diabetes.

4
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

Fig. 1. Multivariable HRs and 95% CIs of joint effect of CMDs status and genetic predisposition on dementia. Models were adjusted for sex, age, education
levels, socioeconomic status, body mass index (BMI), smoking status, alcohol intake frequency, physical activity, diet pattern, hypertension, high-density lipoprotein
(HDL), low-density lipoprotein (LDL), and C-reactive protein (CRP), and the first 10 principal components of ancestry. Abbreviations: HR, hazard ratio; CI, confidence
interval; CMDs, cardiometabolic diseases; AD, Alzheimer’s Disease; VaD, vascular dementia; APOE ε4, apolipoprotein ε4 allele; PRS, polygenic risk score.

Fig. 2. Incidence of dementia by joint effect of CMDs and APOE ε4 status. Abbreviations: HR, hazard ratio; CI, confidence interval; CMDs, cardiometabolic
diseases; AD, Alzheimer’s Disease; VaD, vascular dementia; APOE ε4, apolipoprotein ε4 allele.

no CMD, respectively, and 2.00 (1.79-2.23) and 2.63 (2.38-2.92) for developing dementia. Similar patterns can be seen between joint expo­
those with intermediate and high PRS plus CMDs, respectively. Only sures of CMDs status and PRS with AD and VaD. In addition, we also
significant multiplication interactions between CMDs and PRS for de­ found similar associations between CMDs status and dementia stratified
mentia were observed (P for interaction = 0.006) (Appendix Table 5). by dementia PRS (Appendix Table 6).
The Kaplan-Meier survival curve demonstrates clear differences in the
probability of dementia by PRS and CMDs status (Fig. 3). Participants
with both CMDs and high PRS at baseline had the highest probability of

Fig. 3. Incidence of dementia by joint effect of CMDs and PRS level. Abbreviations: HR, hazard ratio; CI, confidence interval; CMDs, cardiometabolic diseases;
AD, Alzheimer’s Disease; VaD, vascular dementia; APOE ε4, PRS, polygenic risk score.

5
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

3.5. Additional analyses individuals with APOE ε4 non-carriers and APOE ε4 carriers, compared
to those with APOE ε4 non-carriers and no stroke (Llewellyn et al.,
The results were not substantially changed when the analysis was 2010). However, little evidence of associations between any CMDs, PRS,
stratified by sex and age strata (Appendix Tables 7 and 8), when de­ and dementia development was observed. A prior study conducted by
mentia occurred within the first 3 years after recruitment were excluded Karlsson et al. including 13,231 Swedish twins found that there was no
(Appendix Table 9), when we included those younger than 60 years at genetic overlap between CVD and AD, while CVD had a stable higher
baseline (Appendix Table 10), when we repeated the main analyses risk of dementia across quartiles of PRS for coronary artery disease
using variables without imputation (Appendix Table 11), when the main (Karlsson et al., 2017). It could be conjectured that the link between
analyses were conducted using the broad definition of CMDs (Appendix CVD and dementia was not due to genetic overlap, but rather the result
Table 12), and when the competing risk of death was considered (Ap­ of shared exposure risk factors. We found that individuals with CMDs
pendix Table 13). were at greater risk of incident dementia regardless of their genetic risk.
In other words, CMDs prevention is critical across all genetic risk groups.
4. Discussion To the best of our knowledge, our study is the first to demonstrate the
joint effect of CMDs and genetic factors on dementia risk.
In this large-scale prospective study, we found that CMDs were The precise mechanisms underlying the interaction between CMDs
associated with a 1.7-fold higher risk of dementia. Both APOE ε4 and and genetic factors on dementia remain unclear. Dementia is the result
PRS were independently associated with dementia risk. CMDs were of a complex interplay between genetic predisposition and exposure to
consistently associated with an increased risk of dementia regardless of environmental factors, like any other phenotype. Common genetic
genetic risk (APOE ε4 status or PRS). The cumulative incidence of de­ variants including APOE ε4 related to dementia regulate lipid homeo­
mentia differed substantially across the joint effect of CMDs and genetic stasis through mediating lipid transport, which might have an effect on
risk. Participants with APOE ε4 carriers or high PRS plus CMDs had the the regulation of lipid, cholesterol, and insulin (Liu et al., 2013; Mahley
highest probability of developing dementia. In addition, there were and Rall, 2000; Matsuzaki et al., 2010). Existing evidence suggests that
significant additive and multiplication interactions between CMDs and long-term cardiac vascular disease and diabetes, in addition to hyper­
APOE ε4 carriers with dementia, but only multiplication interaction was lipidemia and hypercholesterolemia, may lead to systemic atheroscle­
observed for PRS. These findings provide evidence that dementia risk rosis (Qiu and Fratiglioni, 2015). Dysfunction of lipid metabolism is the
can be reduced by the prevention of CMDs across all categories of ge­ pathological basis of atherosclerosis, which will affect the occurrence of
netic predisposition. dementia (Ou et al., 2020). Another explanation is that APOE ε4 is also a
APOE genotype had a large effect on late-onset dementia, especially risk factor for CMDs, suggesting that this allele and CMDs might have
for AD (Deming et al., 2017). Recently, PRS combining multiple genetic compounding effects on dementia by vascular and neuronal damage
risk alleles with small individual effects has been also predictive of (Araki et al., 2003; Kalmijn et al., 1996).
incident dementia (Adams et al., 2015; Marden et al., 2014). Findings
from a prospective cohort of 1,211 participants conducted by the Fra­ 4.1. Strengths and limitations
mingham Heart Study (FHS) revealed that APOE ε4 allele and high PRS
had a 2.3-fold and 2.5-fold increase in the risk of dementia (Peloso et al., The clear strength of this study was using a relatively representative
2020). In the current study, we noted positive associations between UK general population with a long follow-up period. Additionally, major
APOE ε4 allele and PRS with dementia, and with APOE ε4 carriers confounding factors that may obscure the true associations were
having a higher risk of AD than PRS reaching more than four times. adjusted in the analyses. More importantly, we examined the in­
Numerous epidemiologic studies have suggested the associations teractions and joint effects between CMDs and APOE ε4 and non-APOE
between individual CMDs, such as stroke, IHD, and T2D, and increased PRS on dementia respectively. Despite these strengths, several limita­
dementia risk (Barbiellini Amidei et al., 2021; Grande et al., 2020; tions must be acknowledged. First, CMDs were assessed at baseline,
Koton et al., 2022). Previous studies in the UK Biobank also have found which may not reflect potential changes in the trajectory of the disease
that single CMDs phenotypes (including T2D, stroke, and IHD) were during the follow-up period. Second, this was an observational study,
associated with higher risks of dementia (Dong et al., 2022; van Gennip and causality was not allowed to be demonstrated. Third, the possibility
et al., 2021). However, evidence regarding associations between com­ of reverse causality could not be excluded, although our results were
bined CMDs with dementia and its subtypes is scarce. To our knowledge, similar after excluding the first 3 years of follow-up. Fourth, participants
only one cross-sectional study conducted in UK Biobank including 474, who volunteered in the UK Biobank cohort were more health-conscious
129 participants has found single and multiple CMDs (including hy­ than the general population, which may have underestimated the inci­
pertension, coronary artery disease, and diabetes) were associated with dence of dementia (Fry et al., 2017). Finally, our findings are only from
a higher risk of cognitive impairment (Lyall et al., 2017). Although this individuals of European ancestry, which might affect the generalization
analysis considered the effect of combined CMDs, some limitations of our findings to different nationalities and ethnicities.
should be mentioned. First, this study was largely limited by the
cross-sectional design, whereby causal inferences cannot be made. 4.2. Conclusion and implications
Second, this study analyzed the effect of CMDs only for cognitive
impairment, but not for dementia and its subtypes. Additionally, it had In this large-scale prospective cohort study, our findings indicated
not explored whether the association between CMDs and dementia was that CMDs were associated with an increased risk of dementia regardless
modified by genetic risk of dementia. In the current study, among CMDs of genetic risk including APOE ε4 and non-APOE polygenic risk. From a
status, CMDs for three had a significantly higher risk of dementia, which public health perspective, our results highlight the importance of the
further supplemented the current recommendation to pay more atten­ prevention of CMDs to lower the risk of dementia in people with genetic
tion to the prevention of morbidities, therefore maintaining a lower risk predisposition. Future studies are warranted to investigate the potential
of developing dementia. mechanism of dementia in relation to CMDs and genetic risk.
Several studies have suggested associations between individual
CMDs, genetic factors, and dementia. A cross-sectional study of 1000 Funding
participants aged 55 years or older from China reported that there was a
multiplicative interaction between T2D and APOE ε4 on dementia H.Y. received a grant from the National Natural Science Foundation
development (Zhen et al., 2018). Llewellyn et al. suggested that stroke of China (No.72104179). Y.W. received grants from the National Nat­
was associated with about a 2.2-fold and 15-fold increased risk for ural Science Foundation of China (No. 91746205), and Major Science

6
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

and Technology Project of Public Health in Tianjin (No. Deming, Y., Li, Z., Kapoor, M., Harari, O., Del-Aguila, J. L., Black, K., Carrell, D., Cai, Y.,
Fernandez, M. V., Budde, J., Ma, S., Saef, B., Howells, B., Huang, K. L., Bertelsen, S.,
21ZXGWSY00090).
Fagan, A. M., Holtzman, D. M., Morris, J. C., Kim, S., , … Cruchaga, C., &
Alzheimer’s Disease Neuroimaging, Initiative., Alzheimer Disease Genetic
Ethics approval Consortium (ADGC). (2017). Genome-wide association study identifies four novel
loci associated with Alzheimer’s endophenotypes and disease modifiers. Acta
Neuropathology, 133, 839–856.
The UK Biobank has full ethical approval from the North West Multi- Dong, C., Zhou, C., Fu, C., Hao, W., Ozaki, A., Shrestha, N., Virani, S. S., Mishra, S. R., &
Center Research Ethics Committee, the National Information Gover­ Zhu, D. (2022). Sex differences in the association between cardiovascular diseases
and dementia subtypes: a prospective analysis of 464,616 UK Biobank participants.
nance Board for Health and Social Care in England and Wales, and the
Biology of Sex Differences, 13, 21.
Community Health Index Advisory Group in Scotland. Di Angelantonio, E., Kaptoge, S., Wormser, D., Willeit, P., Butterworth, A. S., Bansal, N.,
O’Keeffe, L. M., Gao, P., Wood, A. M., Burgess, S., Freitag, D. F., Pennells, L.,
Peters, S. A., Hart, C. L., Haheim, L. L., Gillum, R. F., Nordestgaard, B. G.,
Availability of data and materials Psaty, B. M., Yeap, B. B., , … Danesh, J., & Emerging Risk Factors Collaboration.
(2015). Association of cardiometabolic multimorbidity with mortality. JAMA, 314,
The datasets used and/or analyzed during the current study are 52–60.
Escott-Price, V., Sims, R., Bannister, C., Harold, D., Vronskaya, M., Majounie, E.,
available from the corresponding author on reasonable request. Badarinarayan, N., Morgan, K., Passmore, P., Holmes, C., Powell, J., Brayne, C.,
Gill, M., Mead, S., Goate, A., Cruchaga, C., Lambert, J. C., van Duijn, C., Maier, W.,
CRediT authorship contribution statement Ramirez, A., Holmans, P., Jones, L., Hardy, J., Seshadri, S., Schellenberg, G. D.,
Amouyel, P., Williams, J., & GERAD/PERADES, IGPA Consortia. (2015). Common
polygenic variation enhances risk prediction for Alzheimer’s disease. Brain, 138,
Yanchun Chen: Formal analysis, Methodology, Writing – original 3673–3684.
draft, Writing – review & editing. Yuan Zhang: Formal analysis, Writing Fry, A., Littlejohns, T. J., Sudlow, C., Doherty, N., Adamska, L., Sprosen, T., Collins, R., &
Allen, N. E. (2017). Comparison of sociodemographic and health-related
– review & editing. Shu Li: Formal analysis, Writing – review & editing. characteristics of UK Biobank participants with those of the general population.
Lihui Zhou: Methodology, Writing – review & editing. Huiping Li: American Journal of Epidemiology, 186, 1026–1034.
Methodology, Writing – review & editing. Dun Li: Methodology, Gatz, M., Reynolds, C. A., Fratiglioni, L., Johansson, B., Mortimer, J. A., Berg, S.,
Fiske, A., & Pedersen, N. L. (2006). Role of genes and environments for explaining
Writing – review & editing. Yaogang Wang: Methodology, Funding
Alzheimer disease. Archives of General Psychiatry, 63, 168–174.
acquisition, Writing – review & editing. Hongxi Yang: Conceptualiza­ Genin, E., Hannequin, D., Wallon, D., Sleegers, K., Hiltunen, M., Combarros, O.,
tion, Investigation, Formal analysis, Methodology, Supervision, Writing Bullido, M. J., Engelborghs, S., De Deyn, P., Berr, C., Pasquier, F., Dubois, B.,
Tognoni, G., Fievet, N., Brouwers, N., Bettens, K., Arosio, B., Coto, E., Del Zompo, M.,
– review & editing.
Mateo, I., Epelbaum, J., Frank-Garcia, A., Helisalmi, S., Porcellini, E., Pilotto, A.,
Forti, P., Ferri, R., Scarpini, E., Siciliano, G., Solfrizzi, V., Sorbi, S., Spalletta, G.,
Valdivieso, F., Vepsalainen, S., Alvarez, V., Bosco, P., Mancuso, M., Panza, F.,
Declaration of Competing Interest Nacmias, B., Bossu, P., Hanon, O., Piccardi, P., Annoni, G., Seripa, D., Galimberti, D.,
Licastro, F., Soininen, H., Dartigues, J. F., Kamboh, M. I., Van Broeckhoven, C.,
None. Lambert, J. C., Amouyel, P., & Campion, D. (2011). APOE and Alzheimer disease: a
major gene with semi-dominant inheritance. Molecular Psychiatry, 16, 903–907.
Grande, G., Ljungman, P. L. S., Eneroth, K., Bellander, T., & Rizzuto, D. (2020).
Acknowledgements Association between cardiovascular disease and long-term exposure to air pollution
with the risk of dementia. JAMA Neurolology, 77, 801–809.
Guillaume, E., Pornet, C., Dejardin, O., Launay, L., Lillini, R., Vercelli, M., Mari-
The authors would like to express their gratitude to the participants Dell’Olmo, M., Fernandez Fontelo, A., Borrell, C., Ribeiro, A. I., Pina, M. F.,
and staff involved in data collection and management in the UK Bio­ Mayer, A., Delpierre, C., Rachet, B., & Launoy, G. (2016). Development of a cross-
bank. This research has been conducted using the UK Biobank Resource cultural deprivation index in five European countries. Journal of Epidemiology and
Community Health, 70, 493–499.
under Application Number 83974. Guo, F., Moellering, D. R., & Garvey, W. T. (2014). The progression of cardiometabolic
disease: Validation of a new cardiometabolic disease staging system applicable to
obesity. Obesity (Silver Spring), 22, 110–118.
Supplementary materials Han, Y., Hu, Y., Yu, C., Guo, Y., Pei, P., Yang, L., Chen, Y., Du, H., Sun, D., Pang, Y.,
Chen, N., Clarke, R., Chen, J., Chen, Z., Li, L., Lv, J., & China Kadoorie Biobank
Supplementary material associated with this article can be found, in Collaborative, G. (2021). Lifestyle, cardiometabolic disease, and multimorbidity in a
prospective Chinese study. European Heart Journal, 42, 3374–3384.
the online version, at doi:10.1016/j.archger.2022.104853.
Hollingworth, P., Harold, D., Sims, R., Gerrish, A., Lambert, J. C., Carrasquillo, M. M.,
Abraham, R., Hamshere, M. L., Pahwa, J. S., Moskvina, V., Dowzell, K., Jones, N.,
References Stretton, A., Thomas, C., Richards, A., Ivanov, D., Widdowson, C., Chapman, J.,
Lovestone, S., , … Williams, J., & Alzheimer’s Disease Neuroimaging, Initiative.
(2011). Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP
Adams, H. H., de Bruijn, R. F., Hofman, A., Uitterlinden, A. G., van Duijn, C. M.,
are associated with Alzheimer’s disease. Nature Genetics, 43, 429–435.
Vernooij, M. W., Koudstaal, P. J., & Ikram, M. A. (2015). Genetic risk of
Kalmijn, S., Feskens, E. J., Launer, L. J., & Kromhout, D. (1996). Cerebrovascular disease,
neurodegenerative diseases is associated with mild cognitive impairment and
the apolipoprotein e4 allele, and cognitive decline in a community-based study of
conversion to dementia. Alzheimer’s & Dementia : The Journal of The Alzheimer’s
elderly men. Stroke, 27, 2230–2235.
Association, 11, 1277–1285.
Karlsson, I. K., Ploner, A., Song, C., Gatz, M., Pedersen, N. L., & Hagg, S. (2017). Genetic
Araki, S., Koya, D., Makiishi, T., Sugimoto, T., Isono, M., Kikkawa, R., Kashiwagi, A., &
susceptibility to cardiovascular disease and risk of dementia. Translational Psychiatry,
Haneda, M. (2003). APOE polymorphism and the progression of diabetic
7, e1142.
nephropathy in Japanese subjects with type 2 diabetes: results of a prospective
Keenan, T., Zhao, W., Rasheed, A., Ho, W. K., Malik, R., Felix, J. F., Young, R., Shah, N.,
observational follow-up study. Diabetes Care, 26, 2416–2420.
Samuel, M., Sheikh, N., Mucksavage, M. L., Shah, O., Li, J., Morley, M., Laser, A.,
Barbiellini Amidei, C., Fayosse, A., Dumurgier, J., Machado-Fragua, M. D., Tabak, A. G.,
Mallick, N. H., Zaman, K. S., Ishaq, M., Rasheed, S. Z., Memon, F. U., Ahmed, F.,
van Sloten, T., Kivimaki, M., Dugravot, A., Sabia, S., & Singh-Manoux, A. (2021).
Hanif, B., Lakhani, M. S., Fahim, M., Ishaq, M., Shardha, N. K., Ahmed, N.,
Association between age at diabetes onset and subsequent risk of dementia. JAMA,
Mahmood, K., Iqbal, W., Akhtar, S., Raheel, R., O’Donnell, C. J., Hengstenberg, C.,
325, 1640–1649.
Marz, W., Kathiresan, S., Samani, N., Goel, A., Hopewell, J. C., Chambers, J.,
Bis, J. C., Jian, X., Kunkle, B. W., Chen, Y., Hamilton-Nelson, K. L., Bush, W. S.,
Cheng, Y. C., Sharma, P., Yang, Q., Rosand, J., Boncoraglio, G. B., Kazmi, S. U.,
Salerno, W. J., Lancour, D., Ma, Y., Renton, A. E., Marcora, E., Farrell, J. J., Zhao, Y.,
Hakonarson, H., Kottgen, A., Kalogeropoulos, A., Frossard, P., Kamal, A.,
Qu, L., Ahmad, S., Amin, N., Amouyel, P., Beecham, G. W., Below, J. E., , …
Dichgans, M., Cappola, T., Reilly, M. P., Danesh, J., Rader, D. J., Voight, B. F., &
Farrer, L. A., & Alzheimer’s Disease Sequencing, Project. (2020). Whole exome
Saleheen, D. (2016). Causal assessment of serum urate levels in cardiometabolic
sequencing study identifies novel rare and common Alzheimer’s-Associated variants
diseases through a Mendelian Randomization Study. Journal of the American College
involved in immune response and transcriptional regulation. Mol Psychiatry, 25,
of Cardiology, 67, 407–416.
1859–1875.
Kim, Y., Keogh, J., & Clifton, P. M. (2018). Nuts and cardio-metabolic disease: a review of
Bycroft, C., Freeman, C., Petkova, D., Band, G., Elliott, L. T., Sharp, K., Motyer, A.,
meta-analyses. Nutrients, 10.
Vukcevic, D., Delaneau, O., O’Connell, J., Cortes, A., Welsh, S., Young, A.,
Kivimaki, M., Kuosma, E., Ferrie, J. E., Luukkonen, R., Nyberg, S. T., Alfredsson, L., …
Effingham, M., McVean, G., Leslie, S., Allen, N., Donnelly, P., & Marchini, J. (2018).
Jokela, M. (2017). Overweight, obesity, and risk of cardiometabolic multimorbidity:
The UK Biobank resource with deep phenotyping and genomic data. Nature, 562,
Pooled analysis of individual-level data for 120 813 adults from 16 cohort studies
203–209.
from the USA and Europe. Lancet Public Health, 2, e277–e285.
Collins, R. (2012). What makes UK Biobank special? Lancet, 379, 1173–1174.

7
Y. Chen et al. Archives of Gerontology and Geriatrics 105 (2023) 104853

Koton, S., Pike, J. R., Johansen, M., Knopman, D. S., Lakshminarayan, K., Mosley, T., Matsuzaki, T., Sasaki, K., Tanizaki, Y., Hata, J., Fujimi, K., Matsui, Y., … Iwaki, T. (2010).
Patole, S., Rosamond, W. D., Schneider, A. L. C., Sharrett, A. R., Wruck, L., Insulin resistance is associated with the pathology of Alzheimer disease: The
Coresh, J., & Gottesman, R. F. (2022). Association of ischemic stroke incidence, Hisayama study. Neurology, 75, 764–770.
severity, and recurrence with dementia in the atherosclerosis risk in Communities Mau, M. K., Sinclair, K., Saito, E. P., Baumhofer, K. N., & Kaholokula, J. K. (2009).
Cohort Study. JAMA Neurolology, 79, 271–280. Cardiometabolic health disparities in native Hawaiians and other Pacific Islanders.
Kunkle, B. W., Grenier-Boley, B., Sims, R., Bis, J. C., Damotte, V., Naj, A. C., Boland, A., Epidemiologic Reviews, 31, 113–129.
Vronskaya, M., van der Lee, S. J., Amlie-Wolf, A., Bellenguez, C., Frizatti, A., Noordzij, M., Leffondre, K., van Stralen, K. J., Zoccali, C., Dekker, F. W., & Jager, K. J.
Chouraki, V., Martin, E. R., Sleegers, K., Badarinarayan, N., Jakobsdottir, J., (2013). When do we need competing risks methods for survival analysis in
Hamilton-Nelson, K. L, Moreno-Grau, S., … Alzheimer Disease Genetics Consortium nephrology? Nephrology Dialysis Transplantation, 28, 2670–2677.
(ADGC), The European Alzheimer’s Disease Initiative (EADI), Cohorts for Heart and Ou, M., Li, X., Zhao, S., Cui, S., & Tu, J. (2020). Long non-coding RNA CDKN2B-AS1
Aging Research in Genomic Epidemiology Consortium (CHARGE). (2019). Genetic contributes to atherosclerotic plaque formation by forming RNA-DNA triplex in the
meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and CDKN2B promoter. EBioMedicine, 55, Article 102694.
implicates Abeta, tau, immunity and lipid processing. Nature Genetics, 51, 414–430. Peloso, G. M., Beiser, A. S., Satizabal, C. L., Xanthakis, V., Vasan, R. S., Pase, M. P.,
Lambert, J. C., Ibrahim-Verbaas, C. A., Harold, D., Naj, A. C., Sims, R., Bellenguez, C., Destefano, A. L., & Seshadri, S. (2020). Cardiovascular health, genetic risk, and risk
DeStafano, A. L., Bis, J. C., Beecham, G. W., Grenier-Boley, B., Russo, G., Thorton- of dementia in the Framingham Heart Study. Neurology, 95, e1341–e1350.
Wells, T. A., Jones, N., Smith, A. V., Chouraki, V., Thomas, C., Ikram, M. A., Qiao, Y., Liu, S., Li, G., Lu, Y., Wu, Y., Ding, Y., & Ke, C. (2021). Role of depressive
Zelenika, D., Vardarajan, B. N., , … Amouyel, P., & European Alzheimer’s Disease symptoms in cardiometabolic diseases and subsequent transitions to all-cause
Initiative (EADI); Genetic and Environmental Risk in Alzheimer’s Disease; mortality: An application of multistate models in a prospective cohort study. Stroke
Alzheimer’s Disease Genetic Consortium; Cohorts for Heart and Aging Research in and Vascular Neurology, 6, 511–518.
Genomic Epidemiology. (2013). Meta-analysis of 74,046 individuals identifies 11 Qiu, C., & Fratiglioni, L. (2015). A major role for cardiovascular burden in age-related
new susceptibility loci for Alzheimer’s disease. Nature Genetics, 45, 1452–1458. cognitive decline. Nature Reviews Cardiology, 12, 267–277.
Liao, S. G., Lin, Y., Kang, D. D., Chandra, D., Bon, J., Kaminski, N., Sciurba, F. C., & Seshadri, S., Fitzpatrick, A. L., Ikram, M. A., DeStefano, A. L., Gudnason, V., Boada, M.,
Tseng, G. C. (2014). Missing value imputation in high-dimensional phenomic data: Bis, J. C., Smith, A. V., Carassquillo, M. M., Lambert, J. C., Harold, D.,
imputable or not, and how? BMC Bioinformatics, 15, 346. Schrijvers, E. M., Ramirez-Lorca, R., Debette, S., Longstreth, W. T., Jr.,
Liu, C. C., Liu, C. C., Kanekiyo, T., Xu, H., & Bu, G. (2013). Apolipoprotein E and Janssens, A. C., Pankratz, V. S., Dartigues, J. F., Hollingworth, P., … CHARGE
Alzheimer disease: Risk, mechanisms and therapy. Nature Reviews. Neurology, 9, Consortium, GERAD1 Consortium, EADI1 Consortium. (2010). Genome-wide
106–118. analysis of genetic loci associated with Alzheimer disease. JAMA, 303, 1832–1840.
Livingston, G., Huntley, J., Sommerlad, A., Ames, D., Ballard, C., Banerjee, S., Brayne, C., Sudlow, C., Gallacher, J., Allen, N., Beral, V., Burton, P., Danesh, J., Downey, P.,
Burns, A., Cohen-Mansfield, J., Cooper, C., Costafreda, S. G., Dias, A., Fox, N., Elliott, P., Green, J., Landray, M., Liu, B., Matthews, P., Ong, G., Pell, J., Silman, A.,
Gitlin, L. N., Howard, R., Kales, H. C., Kivimaki, M., Larson, E. B., Ogunniyi, A., Young, A., Sprosen, T., Peakman, T., & Collins, R. (2015). UK biobank: an open
Orgeta, V., Ritchie, K., Rockwood, K., Sampson, E. L., Samus, Q., Schneider, L. S., access resource for identifying the causes of a wide range of complex diseases of
Selbaek, G., Teri, L., & Mukadam, N. (2020). Dementia prevention, intervention, and middle and old age. PLoS Medicine, 12, Article e1001779.
care: 2020 report of the Lancet Commission. Lancet, 396, 413–446. van Gennip, A. C. E., Stehouwer, C. D. A., van Boxtel, M. P. J., Verhey, F. R. J., Koster, A.,
Llewellyn, D. J., Lang, I. A., Matthews, F. E., Plassman, B. L., Rogers, M. A., Kroon, A. A., Köhler, S., van Greevenbroek, M. M. J., Wesselius, A., Eussen, S.,
Morgenstern, L. B., Fisher, G. G., Kabeto, M. U., & Langa, K. M. (2010). Vascular Backes, W. H., Jansen, J. F., Schram, M. T., Henry, R. M. A., Singh-Manoux, A., & van
health, diabetes, APOE and dementia: The aging, demographics, and memory study. Sloten, T. T. (2021). Association of Type 2 Diabetes, according to the number of risk
Alzheimer’s Research & Therapy, 2, 19. factors within target range, with structural brain abnormalities, cognitive
Lourida, I., Hannon, E., Littlejohns, T. J., Langa, K. M., Hypponen, E., Kuzma, E., & performance, and risk of dementia. Diabetes Care, 44, 2493–2502.
Llewellyn, D. J. (2019). Association of lifestyle and genetic risk with incidence of Wang, Z., Marseglia, A., Shang, Y., Dintica, C., Patrone, C., & Xu, W. (2020). Leisure
dementia. JAMA, 322, 430–437. activity and social integration mitigate the risk of dementia related to
Luo, Y., He, L., Ma, T., Li, J., Bai, Y., Cheng, X., & Zhang, G. (2022). Associations between cardiometabolic diseases: A population-based longitudinal study. Alzheimer’s &
consumption of three types of beverages and risk of cardiometabolic multimorbidity Dementia : The Journal of the Alzheimer’s Association, 16, 316–325.
in UK Biobank participants: a prospective cohort study. BMC Medicine, 20, 273. Wimo, A., Guerchet, M., Ali, G. C., Wu, Y. T., Prina, A. M., Winblad, B., Jonsson, L.,
Lyall, D. M., Celis-Morales, C. A., Anderson, J., Gill, J. M., Mackay, D. F., Liu, Z., & Prince, M. (2017). The worldwide costs of dementia 2015 and comparisons
McIntosh, A. M., Smith, D. J., Deary, I. J., Sattar, N., & Pell, J. P. (2017). Associations with 2010. Alzheimer’s & Dementia : The Journal of the Alzheimer’s Association, 13,
between single and multiple cardiometabolic diseases and cognitive abilities in 474 1–7.
129 UK Biobank participants. European Heart Journal, 38, 577–583. Zhen, J., Lin, T., Huang, X., Zhang, H., Dong, S., Wu, Y., Song, L., Xiao, R., & Yuan, L.
Mahley, R. W., & Rall, S. C., Jr. (2000). Apolipoprotein E: far more than a lipid transport (2018). Association of ApoE genetic polymorphism and Type 2 diabetes with
protein. Annual Review of Genomics And Human Genetics, 1, 507–537. cognition in non-demented aging chinese adults: A Community Based Cross-
Marden, J. R., Walter, S., Tchetgen Tchetgen, E. J., Kawachi, I., & Glymour, M. M. Sectional Study. Aging and Disease, 9, 346–357.
(2014). Validation of a polygenic risk score for dementia in black and white
individuals. Brain and Behavior, 4, 687–697.