Abstract
The taxonomic approach in dementia has, to date, been categoric with a presumption that a specific underlying pathology is associated with each of the three major dementia subtype diagnoses: Alzheimer's disease (AD), vascular dementia (VaD) and Lewy body dementia. This approach is now enshrined in broadly accepted international research diagnostic criteria [1,3]. These criteria aim to separate the categories as clearly as possible so that, for example, systemic cardiovascular disease debars a diagnosis of probable AD. The validity of this approach is challenged when some patients are classified as ‘mixed’, as they show clinical features suggestive of two or more dementia subtype diagnoses.
It has been further challenged by clinico-pathological validation of these criteria using patients diagnosed in vivo, when recruited onto a community dementia register, whose brains were then made available at postmortem [4]. Until now, the criteria had only been validated in more specialist samples. From the register sample, mixed pathology was found to be much more prevalent at postmortem, particularly a combination of AD and vascular pathology, than was suggested by the in vivo diagnosis. The criteria were good at detecting ‘pure’ pathology, but then pure pathology reflected only a part of the total pathology in the sample. The result was that the sensitivity of the diagnostic criteria was considerably reduced, if the aim was to detect all of the neuropathology in the sample. The categorical distinction is further challenged by a constant, but perhaps under-emphasised, literature showing links between overt cardiovascular disease, its risk factors and AD. This literature is the subject of this review.
Atherosclerotic disease
The association between stroke and AD cannot easily be examined in clinical studies because the current concept of ‘probable’ AD does not allow the two conditions to coexist. However, it has been noted that dementia occurring after a stroke follows an AD-like clinical course in a substantial proportion of cases [5]. In assessing relationships with generalised cardiovascular disease, reliance on a history or not of exposure to vascular disease ‘outcomes’, such as myocardial infarction, is problematic because these ‘outcomes’ are not necessarily an adequate measure of underlying atherosclerotic pathology. Also, the raised mortality associated with these conditions reduces case duration and therefore prevalence in cross-sectional studies, and reduces the chances of successful follow-up in prospective work. Some research has, however, used more accurate assessments of current vascular status to provide a more reliable estimation of exposure. The Hisayama study found that electrocardiogram (ECG) abnormalities did not predict incident dementia (either VaD or AD) in a community-based population of 828 subjects over a 7-year follow-up period [6]. However, a trend towards an association among subjects lacking a family history of dementia was reported between a more narrow categorisation of ECG ischaemia and incident AD in a case-control study nested within a treatment trial of hypertension [7]. A trial fibrillation, in addition, has been reported to be associated with a broad category of AD that included cases with what was felt to be coexistent but unrelated cerebrovascular disease [8]. The Rotterdam study has been the first population-based study to attempt direct quantification of the extent of atherosclerotic disease, using ankle-brachial blood pressure ratios (reflecting peripheral vascular disease) and carotid ultrasound indices in a large community sample. They have reported increased atherosclerosis to be associated cross-sectionally with prevalent dementia and both its AD and VaD subtypes [9]. Follow-up of this cohort appears to confirm similar associations with incident AD [10].
Hypertension
Investigations of the relationship between AD and hypertension have been complicated because the direction of the association appears to change depending upon the relative timing of the assessments of blood pressure level and cognitive status.
While cross-sectional studies suggest that higher blood pressure levels in late-life are associated with better late-life cognitive function, studies with a long-term longitudinal perspective report that higher blood pressure levels in mid-life are associated with impaired cognitive function in late-life [11,12]. Cross-sectional, population-based studies examining AD as an outcome have suggested a similar pattern: AD was associated with raised blood pressure in the Kuopio study of younger (69–78-year-old) subjects [13] but with lower blood pressure in the Kungsholmen study of an older (75–101-year-old) group [14]. Of prospective studies with incident AD as an outcome, two with 5- and 7-year follow-up periods found no association with blood pressure [6,15], although in a cohort of hypertensive subjects those with the highest blood pressure levels (> 190 mmHg) and without a family history of dementia had an increased risk of AD [7]. The Göteborg Study, which had a 15-year follow up period, reported that subjects who developed AD between ages 80–85 years had had significantly higher blood pressures at age 75 but had lower blood pressures at the time of diagnosis than those who did not develop dementia [16], again consistent with the inverting of direction of association seen in cognitive impairment studies. The situation, therefore, appears to be that raised blood pressure in mid-life is a risk factor for future AD but long periods of follow-up are required to show this because the association has disappeared or reversed in age groups where dementia is most commonly studied.
Surprisingly, little well-designed research has followed on from these findings to investigate whether AD can be prevented through control of blood pressure. Recent results from the Syst-Eur systolic hypertension treatment trial have suggested a statistically significant 50% reduction in the incidence of ‘dementia’ (predominantly AD) in the treatment compared to the placebo group [17]. However, this trial was marginally powered to investigate this rare outcome, with only 11 incident cases of dementia in the treatment group and 21 in those taking placebo. More research is needed, particularly in light of negative findings from two previous large randomised controlled trials suggesting no effect of antihypertensive treatment in late-life on cognitive functioning [18,19].
Type 2 diabetes and insulin resistance
The Rotterdam study reported a cross-sectional association between type 2 diabetes and dementia as a broad diagnosis, and with the AD subtype in those treated with insulin [20]. The prospective phase of this study showed a stronger association, with risk for incident AD increased in all diabetic subjects, but again particularly strongly in those receiving insulin treatment [21]. These findings concurred with those from one other prospective study [6]. Insulin resistance has been reported to be associated with AD [13], although its role as an independent prospective risk factor remains to be evaluated. It is a precursor to type 2 diabetes and has been hypothesised to underlie some of the clustering with other vascular risk factors such as hypertension and dyslipidaemia [22].
Smoking
Early studies examining associations between AD and a smoking history were chiefly of case control design using prevalent cases. They were likely to be subject to both selection bias and survival bias through raised case mortality among smokers [23,24]. Population-based research has not supported earlier reports of a protective effect on AD and, if anything, has suggested the opposite. Smoking was found to be associated with an increased risk of AD in those hypertensive subjects who lacked a family history of dementia [7]. The Canadian Study of Health and Ageing reported that heavy smokers had an increased risk of AD and an earlier age of onset [25], and the Rotterdam study reported that current smokers had a greater than twofold risk of incident AD [26].
Cholesterol levels
For cholesterol as with blood pressure the direction of the association with AD may change depending on the timing of the assessment of lipid levels. Lipid levels have generally been reported to be lower in AD than in non-demented subjects of similar ages [9,13] and the Hisayama study found no association with incident dementia over 7 years [6]. However, a recent historical cohort study reported that subjects who went on to develop AD had higher cholesterol levels (after adjusting for apolipoprotein E genotype) recorded 20–30 years earlier, than did those who had not developed the disease. In men, these raised cholesterol levels had decreased prior to the clinical onset of dementia [27]. The Rotterdam study has also reported a decreased risk of AD associated with fish consumption [28]. A prospective association with raised cholesterol may thus be masked by metabolic changes and hypocholesterolaemia early in the course of dementia.
Apolipoprotein E interactions
Possession of the apolipoprotein E ∊4 allele is widely recognised as a risk factor for AD. It has also been reported to be a risk factor for cardiovascular disease, although this association may not be apparent in subjects surviving into old age [10,29]. The Rotterdam study reported that there appeared to be an interaction between possession of this allele and severity of atherosclerosis in determining the risk of AD [9], in that the subjects most at risk were those with vascular disease who also possessed the ∊4 allele. In contrast, a study of the survivors among an elderly hypertensive cohort found no such interaction. Controlling for vascular risk factors marginally increased the strength of the association between possession of the ∊4 allele and the onset of dementia and AD [30].
Reports of interactions between the APOE genotype and environmental risk factors for AD such as head injury [31] and Herpes Simplex virus exposure [32] suggest that the effect of this allele on risk for AD may be to modify vulnerability to environmental insults, possibly through effects on neuronal repair mechanisms. Smoking appears to be principally a risk factor for AD in those who do not possess the APOE ∊4 allele [26], perhaps because there are both beneficial (neurotransmitter) and harmful (vascular) effects of smoking which balance each other out in ∊4 positive subjects. Insulin resistance similarly appeared in one study to be principally associated with AD in ∊4 non-carriers [13], which may be due to raised mortality associated with a combination of glucose intolerance and possession of this allele [33].
Possible mechanisms of association
A raised risk of AD has been reported in association with various vascular risk factors. The recency of many of these findings may reflect a change in approach to dementia subtype classification by research groups, in that such risk factors are no longer seen as necessarily exclusion factors for the diagnosis of AD. Diagnosis of AD was clinically determined in most of these studies with little use either of imaging techniques to quantify subclinical cerebrovascular disease, or postmortem follow-up to confirm the diagnosis pathologically. If a true link between vascular disease and AD exists then various possibilities exist for mechanisms of association [34]. One of these is that vascular disease plays a direct role in inducing Alzheimer pathology. Some support exists for this in that postmortem findings from non-demented subjects show increased Alzheimer changes (principally amyloid deposition) in those with cardiovascular disease or hypertension [35]. Amyloid deposition has been demonstrated secondary to ischaemia in animal models [36], abnormalities in the vasculature have been reported in AD [37] and plausible but hypothetical links exist between cerebrovascular disease and AD through inflammatory mechanisms, blood-brain barrier disturbances and vascular amyloid deposition. Protein glycation secondary to glucose intolerance may be an additional mechanism for the increased risk of AD in type 2 diabetes [38].
An alternative or additional mechanism of association may be that cerebrovascular disease acts in parallel to accelerate the clinical onset of AD, precipitating the manifestation of dementia at relatively mild stages of AD pathology. This interaction is suggested by postmortem findings from the Nun study where subjects with Alzheimer pathology were more likely to have had clinical dementia 2–4 years earlier in life if ischaemic changes were also present [39]. White matter abnormalities may play a role in this respect with cerebrovascular disease affecting fronto-subcortical cognitive function, increasing the likelihood of memory disturbance secondary to hippocampal damage in AD becoming clinically manifest.
A third possibility is that common factors may underlie both AD and vascular disease. Oxidative damage, for instance, has been hypothesised as mediating the initiation and progression of both atherosclerotic and Alzheimer pathology [40,41]. Alternatively, insulin resistance has been found to be associated with cardiovascular risk factors and AD, as mentioned above. Recent evidence has suggested potentially important similarities between the intracellular pathways of insulin metabolism and those involved in the phosphorylation of tau protein (underlying neurofibrillary tangle formation in AD), raising the possibility of common, possibly genetically determined, abnormalities underlying both AD and ‘Syndrome X’ disorders [38].
Conclusions
There is increasing evidence in support of the hypothesis that cardiovascular disease and associated risk factors play a role in the pathogenesis of AD as well as other types of dementia linked by definition to vascular disease processes. At this stage, the mechanisms by which vascular disease and AD are linked remain uncertain. However, it is an area of intense, international research interest, so that a better understanding of their association should emerge in due course. What is paramount is that the conceptual framework concerning dementia, AD and vascular disease should become much more flexible, so that new, perhaps multifactorial aetiological models can be proposed and tested.