In longitudinal analyses of 1,138 subjects (6,292 person-years of follow-up) we found that the risk of AD increased with the number of vascular risk factors, diabetes, hypertension, heart disease, and current smoking. We also found that different combinations of risk factors were associated with a high risk of AD. Diabetes and smoking were the strongest risk factors.
The role of vascular risk factors in vascular dementia seems clear. Vascular dementia is related to stroke (46
) and may be caused by small and large vessel disease (48
) associated with diabetes, heart disease and hypertension (50
). The role of vascular risk factors in AD is controversial (2
). The main putative mechanism in the pathogenesis of AD is the deposition of amyloid beta (Aβ) in the brain (55
), and it is thought that putative risk factors for AD act directly through this pathway (56
). The association between vascular risk factors to AD may not be causal, and could be explained by the incidental coexistence of common disorders in the elderly, or by misclassification of cases of vascular or mixed dementia as AD(16
). Vascular risk factors are known to be related to stroke, and stroke has been shown to be associated with AD (26
), but the mechanisms relating cerebrovascular disease to AD remain to be elucidated. Our results show strong associations between the number of vascular risk factors and AD and support an important role for vascular disease in its pathogenesis.
Hypertension may cause AD through cerebrovascular disease. Hypertension is a risk factor for subcortical white matter lesions (WMLs) found commonly in AD (59
). Hypertension is related to increased vascular permeability with protein extravasation (62
), a common finding in brain parenchyma in AD (63
). Blood pressure was increased 10–15 years before the onset of both AD and vascular dementia in one study (65
), but it was found to be lower in old individuals with dementia (66
). Others have found no association between hypertension and cognitive impairment (67
), and there is conflicting data on the effect of antihypertensive treatment on cognition (69
). Isolated hypertension did not have a strong association with AD in our data, but hypertension clustered with diabetes, heart disease, or smoking did show a higher risk of AD. It is possible that hypertension increases the risk of AD in the presence of other risk factors.
Heart disease can lead to cognitive impairment through cerebral hypoperfusion or embolism (71
) and is also known to be linked with the APOE-
4 allele, a known risk factor for AD (72
). The Rotterdam study (74
) observed an 1.8-fold increased risk of AD in patients with atrial fibrillation. There is a higher frequency of cerebral beta-amyloid-containing senile plaques among individuals with coronary artery disease compared to age-matched controls without heart disease (56
). Heart disease alone did not significantly increase the risk of AD, but subjects with heart disease clustered with diabetes or hypertension had a higher risk of AD.
Diabetes may affect cognition and increase the risk of dementia via oxidative stress, protein glycosilation, and ischemia (75
). Type 2 diabetes is associated with hyperinsulinemia (17
), and peripheral insulin is transported to the CNS across the blood brain barrier (76
). Insulin receptors have been found in the hippocampus (80
), the part of the brain first affected by AD (81
), indicating the potential for peripheral insulin to cause direct injury in AD. Insulin degrading enzyme in the brain is a regulator of extracellular amyloid beta levels (82
) inhibited by insulin (83
). Insulin also has a role in the regulation of phosphorylation of Tau protein, the main component of neurofibrillary tangles (80
). Peripheral insulin infusion in humans increases the levels of amyloid beta in CSF (85
), further suggesting an important role of hyperinsulinemia in AD pathogenesis. These observations are supported by several epidemiologic studies linking hyperinsulinemia (19
), and diabetes (6
) to an increased risk of AD. Diabetes was strongly related to a higher risk of AD in isolation, or when clustered with other vascular risk factors. The wealth of epidemiologic and mechanistic data relating diabetes and AD make it a strong putative risk factor for AD and our results strongly support this notion. Smoking is an important cardiovascular and cerebrovascular risk factor (87
) and could increase the risk of AD through cerebrovascular disease. There have been conflicting data about the association between smoking and AD (88
), but prospective studies have found an increased risk of AD in smokers. A study from the Netherlands found an association between smoking and a higher risk of AD (89
) among persons without the APOE-
4 allele. A study from Northern Manhattan found a higher risk of AD among current smokers with the APOE-
4 allele, and no increased risk among smokers who had quit (21
). Current smoking in the absence of other risk factors was strongly related to a higher risk of AD in our data. Furthermore, the presence of current smoking in clusters with other risk factors appreciably increased the risk of AD as compared with smoking alone or other risk factors in isolation, providing compelling support for a role of smoking in increasing the risk of AD.
Vascular risk factors are seldom found in isolation and often coexist (90
). The usual statistical approach of examining each risk factor individually while adjusting for others may result in the elimination of any real association because of wrong assumptions of confounding (18
). The fact that different risk factors potentially affect the AD process through different direct and indirect pathways as described above raises the possibility that these risk factors act in additive or synergistic manners. We cannot directly address this with our data, but current knowledge on mechanisms related to AD suggests that the associations between different clusters of risk factors and AD that we demonstrate may be explained by the combination of different mechanistic pathways.
There are several potential explanations for our findings. One is that vascular risk factors are associated with a higher risk of vascular dementia and not AD, and that our results are explained by misclassification (16
). The definition of AD has a sensitivity of over 90 percent but a specificity of approximately 50 percent using pathological diagnosis as the gold standard (91
), which can result in misclassification of other types of dementia, including vascular dementia, as AD. We addressed this issue by examining only probable
AD as the outcome, and the association with clusters of risk factors remained strong, suggesting that misclassification is an unlikely explanation for our findings. Another potential explanation is confounding. Lower educational attainment and minority status (African American or Hispanic) are related to a higher risk of dementia (39
) and to a higher prevalence of diabetes, hypertension, and heart disease (20
). We addressed this by stratifying our analyses by ethnic group and education and it did not change our results appreciably. It is also possible that subjects who developed AD were more likely to acquire vascular risk factors due to age or to processes related to preclinical AD. We addressed this possibility by doing analyses excluding subjects with shorter follow-ups, and the results were essentially unchanged. Another potential explanation is chance. This seems unlikely given the strength of our findings and the high level of significance. One important consideration in our data is that the vascular risk factors in our final analyses were ascertained by self-report and we lacked sub-clinical measures of disease, such as echocardiography data. This is likely to underestimate the real prevalence of disease (38
), as is the case with diabetes. The prevalence of self reported diabetes in our sample is comparable to previous reports for the same age groups and ethnic composition (92
). However, the prevalence of diabetes in the general population is higher than what is diagnosed (92
), and self reported diabetes underestimates the true prevalence. We lacked data on the precise duration and severity of the vascular risk factors and there may be considerable measurement error in their estimation; thus, our results may be biased towards the null, and assuming no confounding, other sources of bias, or chance findings, the strong results of our study are likely an underestimation of the true associations between vascular risk factors and AD. Another important consideration is that the cohort in this study is comprised of subjects 65 years and older, with a high prevalence of vascular risk factors, and the results should be interpreted in this context; in fact, persons in the final sample seemed to have a worse vascular risk factor profile than those who were lost to follow-up, with the exception of current smoking. The relationship between vascular risk factors in middle age and AD in later life is likely to be different than what we report due to biases related to survival, and to changes in the measurement of risk factors with aging(2
). Our findings support an important role of modifiable vascular risk factors in the development of AD in the elderly.