In the past decade, vascular disease and vascular risk factors, in particular diabetes, hypertension, smoking, heart disease, dyslipidemia, and obesity, have been implicated in the risk of LOAD.13
While the specific mechanisms through which they affect the risk of LOAD are largely unclear, vascular risk factors may directly affect the deposition of amyloid-βprotein—the main putative culprit—in the brain32,33
or they could be related through cerebrovascular disease. However, there is also evidence that they may act through other mechanisms. Diabetes may increase the risk of dementia via oxidative stress or protein glycosylation.34
Type 2 diabetes is associated with hyperinsulinemia, and peripheral insulin is transported to the central nervous system across the blood-brain barrier.35,36
Insulin receptors have been found in the hippocampus,37
the part of the brain first affected by AD,38
indicating the potential for peripheral insulin to cause direct injury in AD. Insulin-degrading enzyme in the brain is a regulator of extracellular amyloid-βlevels39
inhibited by insulin.40
Insulin also has a role in the regulation of phosphorylation of tau protein, the main component of neurofibrillary tangles.37
Heart disease can lead to cognitive impairment through cerebral hypoperfusion or embolism41
and is also known to be linked with the APOE
ε4 allele, a known risk factor for AD.42,43
Smoking may augment cholinergic metabolism by upregulation of cholinergic nicotinic receptors in the brain.44
Cholinergic deficits, characterized by reduced levels of acetylcholine and nicotinic receptors, are found in AD.45
Hypertension may also contribute to a blood-brain barrier dysfunction, which has been suggested to be involved in the etiology of AD46
or through the formation of free oxygen radicals.46
In addition, vascular risk factors increase the risk of cerebrovascular disease, and cerebrovascular disease seems to lower the threshold of amyloid pathology necessary to manifest dementia.47
We aimed to develop a risk score that allows clinicians to determine the risk of developing dementia in elderly populations and that can be used in genetic research to adjust for a compound variable of nongenetic risk factors. The dementia risk score we developed in the present study was based on the regression coefficients of age, sex, education, ethnicity, APOE ε4 genotype, and several common vascular risk factors that were individually associated with LOAD risk in our study (ie, age, sex, education, ethnicity group, APOE ε4 genotype, diabetes, hypertension, HDL-C levels, and WHR). The resulting risk score predicted dementia in this elderly population well: the probability of LOAD increased with a higher vascular risk score and a greater number of risk factors.
It is important to note that inclusion of additional factors on which we did not have information such as plasma amyloid-β measurements, cerebrospinal fluid biomarkers, neuroimaging markers, inflammation markers, or homocysteine levels may have further improved the predicitivity of the dementia risk score. There is evidence that changes in plasma amyloid β,48
regional cerebral blood flow,49
cerebral blood volume,50–52
hippocampal volumes, posterior cingulate gyrus 1H protein magnetic resonance spectroscopy metabolites, white matter hyperintensity load,53
presence of cortical and subcortical infarctions,54
and inflammation are associated with an increased risk of dementia,61–68
and inclusion of some or all of these factors in future risk scores may improve predicitivity of the score. However, our score uses variables that are readily available in most studies and easy to calculate.
The score values were derived from βcoefficients of the logistic regression model. A more accurate reflection of risk would be the sum of the original coefficients. However, this calculation would be less practical for clinical use. In our study the area under the curve was comparable with that of the model with original βcoefficients, suggesting that our simple scoring method did not result in an important loss of information.
Another important consideration in the use of our score is that the study sample was derived from participants in an urban multiethnic elderly community with a high prevalence of risk factors for mortality and dementia. They were aged 65 years or older at baseline and were followed up on average for 4.2 years. Thus, the score is applicable to the prediction of dementia risk in approximately 4 years only among elderly people who survive for about the next 4 years. It is possible that survival bias has affected our results, as it is reasonable to postulate that those who died before inclusion in the study had worse risk factor profiles than persons surviving up to inclusion in the study. If dementia were more prevalent during life in persons who died before enrollment in the study than in persons who were included in the study, our score would underestimate the effects of the individual risk factors. If those who died before inclusion in the study had worse risk factor profiles while dementia was less prevalent, the specificity of our risk score could be reduced. It is also important to note that there is a potential of misclassification of other types of dementia as LOAD. The diagnosis of LOAD was a consensus diagnosis based on clinical information that includes strokes and other features of clinical presentation such as time of dementia in relation to stroke. While imaging was used to confirm the clinical diagnosis of stroke in most cases, we did not have information on subclinical infarcts or white matter hyperintensities. However, we addressed the issue of misclassification of vascular dementia as LOAD by conducting sensitivity analysis using probable LOAD only. Compared with analyses in which we used probable and possible LOAD, the results did not change, indicating that misclassification of vascular dementia as LOAD is unlikely to explain our findings. The main reason we considered LOAD and not vascular dementia or all dementia as an outcome is that the purpose of this score is to be used as a simple summary score for adjusting for known risk factors in epidemiologic studies of LOAD, including LOAD genetics.
This study has important strengths. It is a prospective cohort study designed for the diagnosis of cognitive decline that has complete clinical and neuropsychological evaluation at each interval. Our study has sensitive measures of cognitive change in several specific domains including memory. In addition, we had the ability to diagnose dementia and cognitive impairment without dementia at baseline, thus allowing us to observe an unbiased sample.
Today there is no curative treatment for LOAD, which emphasizes the importance of primary prevention. In particular, vascular risk factors have been implicated in the risk of LOAD, suggesting that, at least partly, the same measures of primary prevention are implicated for the prevention of both LOAD and cardiovascular events. Our dementia risk score, which is based on the vascular risk profile in late life, predicts dementia well in our data set. For all vascular risk factors included in the score (diabetes, hypertension, dyslipidemia, and obesity), therapeutic interventions are widely available. While additional studies in other populations are needed to validate and further develop the score, our study suggests that this score could be a valuable tool to identify individuals in late life who might be at risk and benefit from lifestyle and therapeutic interventions. In addition, it could be used in genetic research of dementia to adjust for nongenetic confounding.