Eriksson, Ulrika K. MSc*; Bennet, Anna M. PhD*; Gatz, Margaret PhD* †; Dickman, Paul W. PhD*; Pedersen, Nancy L. PhD* †
Cardiovascular disease (CVD) and risk factors for CVD have been linked to cognitive impairment and dementia.1–5 CVD is a well-established risk factor for vascular dementia (VaD) but the association with Alzheimer disease (AD) remains unclear.
The main underlying cause of CVD is atherosclerosis.6 It has been hypothesized that atherosclerosis-induced brain hypoperfusion, oxidative stress, and/or inflammation could contribute directly to the development of the neuropathology in AD.7 However, CVD and AD are both prevalent diseases in elderly people and coexist in a large proportion of individuals with late-onset dementia. Accordingly, many elderly persons with clinical dementia have both AD-type and vascular brain lesions.8,9 An alternative explanation is thus that CVD is simply a comorbid process that increases the likelihood of a dementia diagnosis in patients with subclinical AD pathology.10
The apolipoprotein E epsilon 4 (ApoE4) allele is the most well-established genetic risk factor for sporadic AD.11 However, the role of ApoE4 in the association between CVD and AD is still unclear. ApoE4 has been linked to CVD, although this association has recently come in to question.12 Studies indicate that ApoE contributes to AD pathology through direct effects on amyloid-β processing and neurotoxicity13 and not through enhanced atherosclerosis and CVD.14,15 Some studies have also found that there is an interaction between ApoE4 and CVD on AD risk,16,17 although others have not.1,3
Familial factors are important in the development of both dementias and CVD and it has therefore also been proposed that an association between CVD and dementia could be explained by confounding from sharing of genetic or familial environmental risk factors.18 Twin pairs discordant for CVD and dementia provide an exceptional setting to control for such confounding; if the twin with dementia is more likely to have had CVD than the nondemented twin partner, this provides strong support for the explicit contribution of CVD on the development of dementia.
In this study we have analyzed a population-based sample of 2214 twins who participated in longitudinal studies of aging that included clinical dementia evaluation. First, we estimated the impact of CVD on the risk of dementia and the major subtypes of dementia diagnosis, AD, and VaD. Second, we evaluated whether the association between CVD and dementia is explained by ApoE4 genotype. Third, to evaluate whether the association between CVD and dementia can be explained by familial predisposition to disease, we performed twin pair analyses. Finally, to test whether the impact of CVD on dementia risk is detectable also at the population level, we replicated our analysis in a large register-based cohort of 18,405 individuals.
The Swedish Twin Registry (STR) is a population-based national register including over 170,000 Swedish twins born from 1886 to 2000.19 For this study we included all 2287 twins (henceforth referred to as the “clinical cohort”) with cognitive data from the 3 separate longitudinal studies of aging with clinical dementia evaluation that have been conducted within the STR: (1) the Swedish Adoption/Twin Study of Aging (SATSA),20 (2) Origins of Variance in the Old: Octogenarian Twins (OCTO-Twin),21 and (3) Aging in Women and Men (GENDER).22 All 3 studies have been described in detail elsewhere. In brief, SATSA is an ongoing study of like-sex pairs aged 50 years or more. The first face-to-face examination was conducted in 1986 to 1988 with follow-ups every 3 years (except for 1995 to 1997). There is now information on cognitive status on 1087 individuals. OCTO-Twin included 702 individuals from like-sex twin pairs where both had survived to the age of 80 years with a baseline examination in 1991 to 1994 and 4 follow-ups on a 2 year rolling schedule. GENDER consists of cognitive testing of 498 twins from unlike-sex pairs born from 1916 to 1925 with both members of the pair alive and willing to participate in 1995. Two follow-up occasions of testing took place in 1999 to 2001 and 2003 to 2004.
Assessment of Dementia
Dementia was ascertained in the clinical cohort through a 2-step procedure which entailed, first, a cognitive screening and, second, diagnostic assessment of each suspected case.23 Screening and diagnostic assessment at baseline and follow-up occasions were similar in all 3 studies and have been described in detail previously.24,25 In brief, the Mini-Mental State Examination26 was used to screen for dementia. In SATSA, all twins who screened positive for suspicion of dementia (and their twin partners) were given a complete clinical work-up including physical and neurologic evaluation, a complete history based on informant interviews, neuropsychologic assessment, laboratory tests, and neuroimaging. In OCTO-Twin and GENDER, the diagnostic assessment of dementia was based on reviews of medical records (from both inpatient and outpatient settings), the results on cognitive tests, biochemical blood values, and informant interviews. In all 3 studies, the final diagnoses of dementia were set at a multidisciplinary consensus conference. Dementia was diagnosed according to the criteria in the Diagnostic and Statistical Manual of Mental Disorders, third edition, revised27 and fourth edition28 and differentially diagnosed as AD (based on the NINCDS/ADRDA criteria),29,30 VaD (based on the NINDS-AIREN criteria),31 mixed dementia (AD+VaD), other specified dementia, or unspecified dementia.
Assessment of CVD
Information on CVD was gathered through linkage to the National Patient Register (NPR). The NPR contains data on admission dates, primary diagnosis, secondary diagnoses, and surgical procedures for all individuals admitted to hospital.32 The NPR was established in 1964 and reached complete nationwide coverage in 1987. CVD was assessed though extraction of all International Classification of Diseases (ICD) codes for angina pectoris (AP), myocardial infarction (MI), atherosclerosis, claudication, ischemic heart disease, and the surgical procedures coronary artery bypass graft and percutaneous transluminal coronary angioplasty (ICD codes 410 to 414, 420, 440, 443, 450, 453, I20 to I25, I70, I73, 3068, 3127, 3141, 3158, FNC10-60, FNC96, FND10-20, FND96, FNE00-20, FNE96, 3080, 984, FNG00, FNG02, FNG05, FNG0). Analyses used 2 categories—total CVD (including MI) and MI alone—as MI is an acute condition that requires hospitalization and has been shown to have high sensitivity and specificity in the NPR.33
Age at baseline, sex, history of smoking (ever vs. never), and level of education (less than elementary school vs. more than elementary school) were based on self-reported data. Body mass index (BMI) (25 kg/m2 or more vs. <25kg/m2) was based on clinical measurements in twins who underwent clinical dementia evaluation. Data on diabetes and stroke were derived from the NPR, ICD-codes 260, 250, E10 to E14 and 330 to 332, 334, 430, 431, 433, 434, 436, I60, I61, I63, I64, respectively. Information on ApoE genotype (at least one ApoE4 allele vs. no ApoE4 alleles) was available for 1623 twins who had donated a blood sample. In total, there were 2121 individuals with data on educational attainment, 1921 individuals with smoking data, and 1662 individuals with BMI data.
Baseline was defined as January 1, 1974 (more than half of the Swedish counties were then covered by the NPR).34 From the 2287 twins with cognitive data we excluded 12 individuals owing to death or dementia before baseline and 61 twins whose dementia onset preceded their first record of CVD in the NPR; leaving 2214 in the final study population (313 monozygous pairs, 718 dizygous pairs, 5 pairs with unknown zygosity, and 142 single twins). A part of this sample, that is, like-sexed pairs born from 1903 to 1936 (N=1053), is also included in the register-based study population (by design).
All analyses were performed using the statistical software package SAS v 9.1 and v 9.2. Relative risks were estimated as hazard ratios (HR) in a Poisson regression model (PROC GENMOD with DIST=Poisson and OFFSET=natural logarithm of time at risk). Time at risk was calculated as years from baseline to the date of the first record of dementia, last date of follow-up, other causes of dementia (eg, hydrocephalus), or end of study on December 31, 2003 [the last occasion the STR was updated against the Causes of Death Register (CDR)]. The Lexis macro35 was used to split the underlying time scale (time since entry into study). CVD was analyzed as a time-dependent variable, that is, a person negative for CVD contributes time-at-risk to the CVD negative group until the first record of CVD in the NPR, at which point that individual will start to contribute time-at-risk to the CVD positive group. Testing the proportional hazards assumption revealed that the effect of CVD on dementia risk was affected by time since CVD and a second time scale (time since CVD) was therefore introduced. HRs were thus estimated by including (time since exposure×CVD) interaction terms (split into ≤3 and >3 y since exposure) in the same model. We adjusted for correlated twin data using generalized estimating equations (REPEATED SUBJECT=identification number unique to each twin pair, TYPE=exchangeable).
Co-twin Control Analysis
The co-twin control design capitalizes on pairs who are discordant for both the exposure and the outcome. In this longitudinal setting, pairs in which both twins eventually developed dementia were treated as discordant pairs; the twin with the earliest age of dementia onset was defined as the case and the partner (with a later age of onset) as the control. This is equivalent to a nested case-control study with age as the timescale and matched on twin pair. Only pairs where the partner to the demented proband was alive and nondemented at the time the proband developed dementia were included. CVD data on both twins in the pair were collected until the date the proband developed dementia. Twin pairs were analyzed using conditional logistic regression (PROC LOGISTIC with STRATA=identification number unique to each twin pair).
Register Cohort, N=18,405
We replicated our analysis in a large population-based sample of 18,405 twins (henceforth referred to as the “register cohort”) for whom dementia diagnoses were ascertained through register linkage. The register cohort included all twins born from 1903 to 1936 that responded to a questionnaire in 1963, 1967 or 1973 regarding health-related information with special attention to cardiovascular health. Cohort compilation and analyses were identical to those for the clinical cohort with the exceptions described below.
The outcome of interest was defined as hospitalization with dementia and was ascertained through linkage to the NPR and extraction of all primary and secondary diagnoses of AD (ICD codes 305, 290, 331, F00, G30), VaD (306, 293, 290E, F01), and other dementias not owing to exogenous factors or tumors. Information on dementia was complemented with data from the CDR. The CDR contains information on the date of death and underlying and contributory causes of death and the register has more than 99% overall completeness since 1961.36 A “surrogate” date of dementia hospitalization was calculated for those twins who only had a dementia diagnosis in the CDR (N=153) (thus lacking a date of dementia hospitalization) by deducting 3 years from the date of death (three years was the average time between the first hospitalization with dementia and death with dementia in those twins [N=273] who had a record of dementia in both the NPR and the CDR).
The health questionnaire data provided by the twins in 1963, 1967 or 1973 was used to (1) complement CVD data in the NPR with self-reported information on ever having had AP and/or a MI as to also include CVD occurring in younger ages and (2) extract (self-reported) covariate information similar to that in the clinical cohort.
Twins in the register cohort were followed for dementia through linkage to the NPR and CDR from the age of 60 to 65 years (depending on when the NPR became complete in the county of residence) until December 31, 2003. The upper limit of 65 years of age at baseline was imposed to minimize risk of including twins who had already developed dementia.
The study sample was compiled by including all 26,486 twins born from 1903 to 1936 who responded to questionnaires in 1967 or 1973. Individuals were then excluded owing to one or more of the following: missing CVD data in the questionnaires (N=2583), an age above 65 years at baseline (N=3797), loss to follow-up (N=151), dementia at baseline (N=64), death before baseline (N=2062), or dementia onset preceding the first record of CVD in the NPR (N=103); leaving 18,405 twins in the final study population (2767 monozygous pairs, 4839 dizygous pairs, 175 pairs with unknown zygosity, and 2843 single twins).
Of the 2275 twins initially included in the clinical cohort, 494 (21.7%) received a diagnosis of dementia, 251 at the first wave of dementia testing, and 243 at follow-up. Of the twins with a dementia diagnosis, 133 (26.9%) were at some point hospitalized with CVD, 72 before dementia onset and 61 twins after dementia onset. Of the 72 twins with CVD before dementia, 48.6% received a diagnosis of VaD and 33.3% a diagnosis of AD. Of the 61 twins with CVD after dementia, 23.0% had a diagnosis of VaD and 62.3% had a diagnosis of AD (P=0.0006). Twins diagnosed with AD were hospitalized with CVD on average 1.4 years after AD diagnosis whereas twins diagnosed with VaD were hospitalized with CVD on average 4.8 years before VaD diagnosis (P<0.0001).
The 61 twins with a dementia onset preceding hospitalization with CVD were excluded, leaving 433 twins with dementia in the analysis. Dementia subtype was available for all twins and the diagnosis were as follows: 242 possible or probable AD, 113 possible or probable VaD, 11 mixed (AD+VaD), and 67 not otherwise specified dementia (Table 1).
During the follow-up period until December 31, 2003, 409 (18.5%) twins in the clinical cohort were admitted at least once to hospital with a primary or secondary diagnosis of CVD. AP and MI made up the largest proportion of CVD diagnoses, 115 (28.1%) had a diagnosis of both AP and MI, 134 (32.8%) had AP but not MI, 107 (26.2%) had an MI but not AP, and 53 (13.0%) had only a diagnosis of atherosclerosis, claudication, ischemic heart disease, coronary artery bypass graft, or percutaneous transluminal coronary angioplasty. Stroke was not included in the CVD variable, but 105 (25.7%) individuals of those positive for CVD (N=409) were also hospitalized with stroke, compared with 207 (11.5%) of the non-CVD (N=1805) part of the sample (P<0.0001).
To evaluate if our findings were related to the effect of ApoE4 on the timing or severity of CVD, we looked at age at first hospitalization with CVD and the number of hospitalizations with CVD. We found no significant differences in timing of CVD between carriers and noncarriers of ApoE4: ApoE4 carriers were on average 74.5 years at first hospitalization with CVD and noncarriers 74.8 years (P=0.78). Nor did we find a significant difference in the number of hospitalizations (in those ever hospitalized): ApoE4 carriers were hospitalized with CVD or stroke on average 4.5 times and noncarriers 4.4 times (P=0.89).
Risk of Dementia, AD and VaD
CVD was associated with an almost doubled risk of dementia [HR 1.83, 95% confidence interval (CI) 1.23-2.72] during the first 3 years since the first hospitalization with CVD and a more than 3-fold increased risk of a diagnosis of VaD (HR 3.64, 95% CI 2 .01-6.57) whereas there was no significant increased risk of a diagnosis of AD (HR 1.48, 95% CI 0.83-2.64) (Table 2). Relative risk estimates decreased with longer follow-up time since hospitalization with CVD (interaction term HR 0.97 per year, P=0.22).
Including ApoE4 genotype as a covariate in the model had no effect on the relative risk of CVD on dementia/AD. However, there was a significant interaction between ApoE4 and CVD on the risk of AD (interaction term HR 2.98, P=0.05) but not on all dementias (1.31, P=0.40) or VaD (0.76, P=0.61). Stratifying on ApoE4 genotype revealed that the association between CVD and dementia/AD during the first 3 years since hospitalization with CVD was present only among carriers of at least one ApoE4 allele, HR 2.37 (1.34-4.22) for all dementias and 2.39 (1.15-4.96) for AD [in ApoE4 noncarriers: 1.45 (0.76-2.75) and 0.76 (0.24-2.42) for all dementias and AD, respectively]. Risk estimates for VaD were similar regardless of ApoE4 status, 3.36 (1.15-9.79) and 3.90 (1.73-8.82) in the ApoE4 positive and negative groups, respectively.
Twins under the age of 75 years at the first hospitalization with CVD had a higher risk of any dementia (2.61, 1.25-5.44) and VaD (7.68, 3.26-18.08) compared with twins 75 years of age and older, 1.57 (0.99-2.51) and 2.08 (0.96-4.54), respectively. For AD, the trend was reversed HR 0.54 (0.08-3.90) before the age of 75 years and 1.66 (0.93-2.99) thereafter.
We also performed additional analyses that had no effect on the relative risk estimates: (1) to evaluate an effect of CVD on dementia risk that is not mediated by stroke, we reran our analysis excluding prevalent stroke cases and censoring incident stroke; (2) we excluded fatal MIs (defined as death within a month of hospitalization with MI); and (3) we adjusted for sex, educational level, BMI, smoking, and diabetes in those with complete covariate information (N=1215).
Risk of Dementia, AD, and VaD in Co-twin Control Analyses
CVD was more likely to have been present in the twin who developed dementia than in the twin partner who survived to the same age (Tables 3, 4). The results were very similar to those found in the survival analysis of the full clinical cohort; the most elevated odds ratios were observed for VaD and there were no significant findings for AD. Among identical twin pairs, no twin without CVD developed VaD. This result is consistent with the full twin sample, but makes it impossible to estimate an odds ratio.
Compared with the clinical cohort, the register cohort was 5.1 years older at baseline (P<0.0001), 3.8 years younger at first CVD (P<0.0001), 1.7 years younger at dementia onset (P<0.0001), 7.8 years younger at death (P<0.0001), had 8.1 years shorter follow-up time (P<0.0001), a higher burden of CVD (P<0.0001), a larger proportion of men (P<0.0001), and ever smokers (P<0.0001) (Table 1).
Age-adjusted HRs are shown in Table 2. CVD was associated with an almost doubled risk of dementia during the first 3 years since CVD and a 43% increased risk during the following 10 years. HRs decreased with longer follow-up time (interaction term HR 0.97 per year, P<0.0001) and the age-adjusted relative risk of dementia attributable to CVD after more than 13 years declined to 1.25 (1.04-1.50). Excluding individuals with a history of stroke at baseline and censoring follow-up time after an incident stroke had no effect on the associations between CVD and dementia. Including stroke as a covariate in survival analysis had a marginal effect on the estimates, HR of CVD on dementia decreased from 1.48 (1.14-1.94) to 1.42 (1.08-1.85) for the entire follow-up period.
Additional adjustments for sex, educational level, smoking, BMI, and diabetes did not abolish the association. The multi-adjusted relative risk (HR with 95% CI) of dementia during the first 3 years since CVD was 1.65 (1.27-2.13) and during the following 10 years 1.42 (1.17-1.74).
In 2 large population-based samples with long-term follow-up we were able to show that nonstroke CVD can double the risk of late-life dementia. However, nonstroke CVD was only a significant risk factor for a diagnosis of AD in individuals carrying the ApoE4 allele. Also, by using a co-twin control design, we could show that a twin with dementia is more likely to have had CVD than is his or her nondemented identical twin partner, suggestion that the association between CVD and dementia is not confounded by genetic and early life environmental factors. We were also able to show, consistent with findings from the Rotterdam study,3 that the relative risk of dementia owing to CVD is attenuated over time, perhaps as a consequence of the impact of CVD on mortality risk.
Recent data show that late-life dementia often represents a mix of AD and vascular pathology.37,38 Our results raise a question whether CVD might skew differential diagnosis of dementia toward VaD. In our sample, one-fourth of twins diagnosed with AD were at some point in life hospitalized with CVD, but <40% of them had been hospitalized before their AD diagnosis compared with VaD, where more than 70% were hospitalized before their VaD diagnosis (data not shown). It is thus possible that the impact of CVD on the clinical manifestation of AD is underestimated owing to diagnostic bias.
In this study we were interested in estimating an effect of atherosclerotic disease on the risk of dementia/AD that is not mediated by the acute trauma to the brain that is stroke.39 Thus, we did not include stroke in our definition of CVD. However, given that coronary heart disease and peripheral vascular disease are associated with stroke incidence,40 we also performed additional analyses to adjust for the possible indirect effect of stroke on our estimates. We excluded prevalent stroke cases, censored incident stroke cases, and also included stroke as a covariate in the model. None of these adjustments had a substantial effect on our risk estimates. Our data thus suggest that CVD is not only a risk factor for dementia/AD in that it is associated with stroke but that, even in the absence of stroke, atherosclerotic disease seems to be a risk factor for dementia/AD.
Our study further suggests that CVD increases the risk of dementia through the CVD per se. Similar results in the cohort analyses and the within twin pair analyses suggest that the association between CVD and dementia is not confounded by genetic or early life environmental factors. If confounding was present and the association was attributable to genes or early life exposures in common to both diseases, any (false) positive associations found in the cohort analyses would not be seen in the co-twin control analyses.
The role of ApoE4 in the association between CVD and AD has long been subject to debate. In line with previous findings,14,15 this study supports the hypothesis that ApoE4 is a risk factor for AD through other pathways than CVD as the 3-fold elevated risk of AD attributable to ApoE4 remained also when adjusting for CVD (data not shown). However, we did find evidence that ApoE genotype modifies the effect of CVD on AD risk (CVD was only a risk factor for AD in ApoE4 carriers). This effect modification was not attributable to indicators of CVD severity, such as an earlier onset of CVD or more frequent hospitalizations for CVD; there were no significant differences in age at hospitalization with CVD or the number of hospitalizations with CVD between ApoE4 carriers and noncarriers. Our interpretation of these findings is that carriers of the ApoE4 allele are more vulnerable to the burden of CVD and less resilient to withstand the impact of atherosclerotic disease on cognition.
There are some limitations to this study that deserve mentioning. Given the relatively high age of the study population at the beginning of follow up, it is likely that a proportion of those classified as negative for CVD at baseline are, in fact, CVD positive. If such misclassification is present, the true relative risk of dementia owing to CVD would be higher than what we have presented here. Furthermore, twins had to survive to a certain age to be eligible for participation. As it is likely that individuals affected by both CVD and dementia will be removed from the source population at a faster rate than individuals with only one (or none) of the diseases, it is possible that we have underestimated the proportion of individuals with both CVD and dementia.
Epidemiologic research using national disease registers such as the NPR and CDR offers the advantage of cost-effective large population-based studies over a long period. For acute conditions, like MI, the register data has both high sensitivity and specificity33 but for nonacute conditions, like dementia, the sensitivity is usually lower. However, validation studies suggest that the registers are suitable for studies of dementia but that presence of false negative cases (low sensitivity) might dilute the risk estimates toward the null.41
In conclusion, this study shows that nonstroke CVD is a risk factor for AD in genetically vulnerable individuals carrying the ApoE4 allele. With twin analyses, we have also shown that the association between CVD and dementia is not mediated by genetic or early life environmental factors in common to both disorders. However, we think it is important to separate the etiologic from the clinical implications of this study. Given that (1) dementia often represents a mix of AD and vascular pathology and (2) patient ApoE genotype is not usually known, CVD prevention should be of interest in all individuals as a possible means of reducing risk for dementia, irrespective of dementia subtype.
1. Newman AB, Fitzpatrick AL, Lopez O, et al. Dementia and Alzheimer's disease incidence in relationship to cardiovascular disease in the Cardiovascular Health Study cohort. J Am Geriatr Soc. 2005;53:1101–1107.
2. Qiu C, Winblad B, Marengoni A, et al. Heart failure and risk of dementia and Alzheimer disease: a population-based cohort study. Arch Intern Med. 2006;166:1003–1008.
3. van Oijen M, de Jong FJ, Witteman JC, et al. Atherosclerosis and risk for dementia. Ann Neurol. 2007;61:403–410.
4. Breteler MM, Claus JJ, Grobbee DE, et al. Cardiovascular disease and distribution of cognitive function in elderly people: the Rotterdam Study. BMJ. 1994;308:1604–1608.
5. Rosengren A, Skoog I, Gustafson D, et al. Body mass index, other cardiovascular risk factors, and hospitalization for dementia. Arch Intern Med. 2005;165:321–326.
6. Lusis AJ. Atherosclerosis. Nature. 2000;407:233–241.
7. Casserly I, Topol E. Convergence of atherosclerosis and Alzheimer's disease: inflammation, cholesterol, and misfolded proteins. Lancet. 2004;363:1139–1146.
8. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Pathological correlates of late-onset dementia in a multicentre, community-based population in England and Wales. Lancet. 2001;357:169–175.
9. Chui HC, Zarow C, Mack WJ, et al. Cognitive impact of subcortical vascular and Alzheimer's disease pathology. Ann Neurol. 2006;60:677–687.
10. Schneider JA, Wilson RS, Bienias JL, et al. Cerebral infarctions and the likelihood of dementia from Alzheimer disease pathology. Neurology. 2004;62:1148–1155.
11. Farrer LA, Cupples LA, Haines JL, et al. Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. ApoE and Alzheimer Disease Meta Analysis Consortium. JAMA. 1997;278:1349–1356.
12. Bennet AM, Di Angelantonio E, Ye Z, et al. Association of apolipoprotein E genotypes with lipid levels and coronary risk. JAMA. 2007;298:1300–1311.
13. Mahley RW, Weisgraber KH, Huang Y. Apolipoprotein E4: a causative factor and therapeutic target in neuropathology, including Alzheimer's disease. Proc Natl Acad Sci U S A. 2006;103:5644–5651.
14. Prince M, Lovestone S, Cervilla J, et al. The association between ApoE and dementia does not seem to be mediated by vascular factors. Neurology. 2000;54:397–402.
15. Slooter AJ, Cruts M, Ott A, et al. The effect of ApoE on dementia is not through atherosclerosis: the Rotterdam Study. Neurology. 1999;53:1593–1595.
16. Beeri MS, Rapp M, Silverman JM, et al. Coronary artery disease is associated with Alzheimer disease neuropathology in ApoE4 carriers. Neurology. 2006;66:1399–1404.
17. Hofman A, Ott A, Breteler MM, et al. Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimer's disease in the Rotterdam Study. Lancet. 1997;349:151–154.
18. Stampfer MJ. Cardiovascular disease and Alzheimer's disease: common links. J Intern Med. 2006;260:211–223.
19. Lichtenstein P, De Faire U, Floderus B, et al. The Swedish Twin Registry: a unique resource for clinical, epidemiological and genetic studies. J Intern Med. 2002;252:184–205.
20. Finkel D, Pedersen NL. Processing speed and longitudinal trajectories of change for cognitive abilities: The Swedish Adoption/Twin Study of Aging. Aging Neuropsychol Cogn. 2004;11:325–345.
21. McClearn GE, Johansson B, Berg S, et al. Substantial genetic influence on cognitive abilities in twins 80 or more years old. Science. 1997;276:1560–1563.
22. Gold CH, Malmberg B, McClearn GE, et al. Gender and health: a study of older unlike-sex twins. J Gerontol B Psychol Sci Soc Sci. 2002;57:S168–S176.
23. Gatz M, Fiske A, Reynolds CA, et al. Sex differences in genetic risk for dementia. Behav Genet. 2003;33:95–105.
24. Dahl A, Berg S, Nilsson SE. Identification of dementia in epidemiological research: a study on the usefulness of various data sources. Aging Clin Exp Res. 2007;19:381–389.
25. Gatz M, Pedersen NL, Berg S, et al. Heritability for Alzheimer's disease: the study of dementia in Swedish twins. J Gerontol A Biol Sci Med Sci. 1997;52:M117–M125.
26. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189–198.
27. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. DSM-III-R, 1987; American Psychiatric Association.
28. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. DSM-IV, 1994; American Psychiatric Association.
29. McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology. 1984;34:939–944.
30. Dubois B, Feldman HH, Jacova C, et al. Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol. 2007;6:734–746.
31. Roman GC, Tatemichi TK, Erkinjuntti T, et al. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology. 1993;43:250–260.
34. Eriksson UK, Gatz M, Dickman PW, et al. Asthma, eczema, rhinitis and the risk for dementia. Dement Geriatr Cogn Disord. 2008;25:148–156.
37. Aguero-Torres H, Kivipelto M, von Strauss E. Rethinking the dementia diagnoses in a population-based study: what is Alzheimer's disease and what is vascular dementia? A study from the kungsholmen project. Dement Geriatr Cogn Disord. 2006;22:244–249.
38. Barker WW, Luis CA, Kashuba A, et al. Relative frequencies of Alzheimer disease, Lewy body, vascular and frontotemporal dementia, and hippocampal sclerosis in the State of Florida Brain Bank. Molecular pathways to neurodegeneration. Alzheimer Dis Assoc Disord. 2002;16:203–212.
39. Leys D, Henon H, Mackowiak-Cordoliani MA, et al. Poststroke dementia. Lancet Neurol. 2005;4:752–759.
40. Lloyd-Jones D, Adams R, Carnethon M, et al. Heart disease and stroke statistics—2009 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 2009;119:e21–e181.
41. Jin YP, Gatz M, Johansson B, et al. Sensitivity and specificity of dementia coding in two Swedish disease registries. Neurology. 2004;63:739–741.
© 2010 Lippincott Williams & Wilkins, Inc.