Assessment of risk factors for earlier onset of sporadic Alzheimer's disease dementia
Correspondence Address: Source of Support: This work was supported by CAPES – Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, while grants from CNPq – Conselho Nacional de Desenvolvimento Científico e Tecnológico, and FAPESP – The State of São Paulo Research Foundation, were also provided for the Laboratory of
Genetics, Conflict of Interest: None DOI: 10.4103/0028-3886.149384
Source of Support: This work was supported by CAPES – Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, while grants from CNPq – Conselho Nacional de Desenvolvimento Científico e Tecnológico, and FAPESP – The State of São Paulo Research Foundation, were also provided for the Laboratory of Genetics, Conflict of Interest: None
Background: Pharmacological treatment has mild effects for patients with Alzheimer's disease dementia (AD); therefore, the search for modifiable risk factors is an important challenge. Though risk factors for AD are widely recognized, elements that influence the time of dementia onset have not been comprehensively reported. We aimed to investigate which risk factors might be related to the age of onset of AD in a sample of patients with highly variable educational levels, taking into account the Framingham risk scoring as the sole measure of vascular risk. Subjects and Methods: We included 209 consecutive late-onset AD patients to find out which factors among educational levels, coronary heart disease risk estimated by way of Framingham risk scores, history of head trauma or depression, surgical procedures under general anesthesia, family history of neurodegenerative diseases, gender, marital status and APOE haplotypes might be related to the age of dementia onset in this sample of patients with low mean schooling. Results: Mean age of AD onset was 73.38 ± 6.5 years old, unaffected by schooling or family history of neurodegenerative diseases. Patients who were APOE-ε4 carriers, married, or with history of depression, had earlier onset of AD, particularly when they were women. Coronary heart disease risk was marginally significant for later onset of AD. Conclusions: APOE haplotypes, marital status and history of depression were the most important factors to influence the age of AD onset in this sample. While midlife cerebrovascular risk factors may increase incidence of AD, they may lead to later dementia onset when present in late life.
Keywords: Alzheimer disease, cerebrovascular disorders, dementia, educational status, neurodegenerative diseases, risk factors
Treatment for patients with Alzheimer's disease dementia (AD) is still mildly effective, usually slowing cognitive decline in the earlier stages without prolonging
survival.  Thus the search for modifiable risk factors is an important challenge. Many studies have assessed risk factors for AD, particularly in highly educated populations, but elements that influence the time of dementia onset have seldom been explored. While taking into account the impact of Framingham risk scores,  we sought to investigate which of the most studied risk factors might be related to the age of onset of AD in a sample of patients with low mean schooling.
Consecutive outpatients with late-onset AD according to National Institute on Aging - Alzheimer's Association criteria  were recruited from the Behavioral Neurology Section of our university hospital during a period of 28 months. All subjects had a magnetic resonance imaging (MRI) to evaluate either medial parietal or medial, basal or lateral temporal atrophy; or, in cases of claustrophobia or use of pacemakers, a computed tomography (CT) scan to exclude vascular lesions, after which the Modified Ischemic Score  had to point to a diagnosis of AD for proper inclusion. Late-onset AD was considered when the dementia syndrome began at or after patients were 60 years old.
Participants were screened for estimated age of onset of AD, gender, marital status by the time of dementia onset, years of schooling, personal history of head trauma or depression under pharmacological treatment, amount of previous surgical interventions with general anesthesia, 10-year projected risk for coronary heart disease (CHD) according to Framingham risk scores  , use of lipid-lowering drugs, amount of family members (up to third degree) with history of AD or non-Alzheimer's neurodegenerative diseases, and APOE haplotypes. For every subject, evidence concerning age of onset of AD was determined following a review of medical records alongside an interview with the caregiver (preferably a family member), who should have frequent visits with the patient for accuracy of information regarding the time when functional impairment actually started. Risk factors were assessed in a detailed self-report or proxy report.
The Framingham risk score  is a composite score that predicts 10-year chance of CHD based on gender, age, total cholesterol, HDL cholesterol, any cigarette smoking in the previous month, and systolic blood pressure with different scoring methods according to treatment or lack of anti-hypertensive treatment. For calculation of the point total, the earliest total cholesterol and HDL-cholesterol values were accounted for by averaging at least two measurements obtained from lipoprotein analysis, preferably before patients started treatment with lipid-lowering drugs.
The Modified Ischemic Score  was chosen because it is a suitable and easy to use score for discrimination between AD and vascular dementia, composed of items from patient history, neurological examination and neuroimaging exams. Patients with lower scores have higher probability of AD, whereas higher scores point to vascular dementia. Intermediate scores suggest mixed dementia, and in such cases patients were not included in our study.
After blood samples were collected from all patients in tubes with EDTA 0.1%, genomic DNA was extracted for determination of APOE haplotypes (polymorphisms rs7412 and rs429358 assessed by way of Real-Time Polymerase Chain Reactions using TaqMan® SNP Genotyping Assays on the Applied Biosystems® 7500 Fast Real-Time PCR System). The standard protocols of the manufacturer (Applied Biosystems®, USA) were followed.
For statistical analysis, we employed three multiple regression models: in model 1, independent variables were schooling, gender (male or female), marital status (married versus non-married), history of head trauma, history of depression under pharmacological treatment, amount of previous surgical interventions under general anesthesia, amount of family members (up to third degree) either with history of AD or non-Alzheimer's neurodegenerative diseases, number of copies of APOE-ε4 (0, 1 or 2), and 10-year CHD risk according to Framingham risk scores  ; in models 2 (males) and 3 (females), independent variables were the same as in model 1, except for the inclusion of raw Framingham risk scores  in replacement of 10-year CHD risk and gender. The dependent variable was always the age of onset of AD. Regarding marital status, we considered as non-married all patients who were single, divorced or widowers by the time of dementia onset. The Mann-Whitney test was employed to assess differences between genders concerning age of onset of the dementia syndrome. The threshold of significance was set at P < 0.05.
This study is part of the research project 1067/10 (CAAE 0540.0.174.000-10) approved by the Ethics Committee of our university hospital on August 2010. All invited patients and their legal representatives agreed to participate in the research and signed the Informed Consent Form before the evaluation, with no exceptions.
A total of 217 patients were recruited; five patients (2.3%) had to be excluded due to incomplete data, and another three (1.4%) had to be excluded due to onset of dementia earlier than 60 years of age; thus, the final sample was 209 patients. Mean age of onset of dementia was 73.38 ± 6.5 years (range 60-88 years-old): 73.64 ± 6.6 years for females and 72.85 ± 6.3 years for males (P = 0.3466). Mean schooling was 4.2 ± 3.7 years (range 0-15 years): 3.7 ± 3.3 years for females and 5.3 ± 4.2 years for males. Some patients had undergone up to seven surgical procedures under general anesthesia, while family members with AD ranged from none to 14 (mean 0.5 ± 1.2) and family members with non-Alzheimer's neurodegenerative diseases ranged from none to two (mean 0.1 ± 0.4). Mean Framingham risk scores  were 19.7 ± 2.5 (range 13-25) for females and 14.9 ± 1.2 (range 11-17) for males. Overall mean 10-year risk for CHD was 14.27% ± 7.3% (range 2%-30%): 11.48% ± 6.5% (range 2%−30%) for females and 20.07% ±5.1% (range 8%−30%) for males. Complete demographic profiles and genetic results for APOE haplotypes are given in [Table 1].
[Table 2] shows results from multiple regression models. Disparities in age of AD onset were more significant for females, in whom independent variables explained 15.2% of variations. Patients who were married, with history of depression, and APOE-ε4 carriers had earlier onset of AD, particularly when they were women. Each copy of APOE-ε4 led to an overall onset of dementia almost 2 years earlier, reaching almost 4.5 years earlier for women with both copies of APOE-ε4. The 10-year projected risk for CHD was marginally significant for age of AD onset, whereas higher raw Framingham risk scores  had significance for later onset of AD for females. For most patients, we were able to estimate the value of Framingham risk scores  before the start of the dementia syndrome. Nevertheless, nearly 40% of our patients were using lipid-lowering drugs when such scores were assessed, thus adding a possible bias for our results.
More than a third of worldwide cases of AD are attributable to cerebrovascular risk factors, whereas almost one-fifth are attributable to low education.  While APOE-ε4 and depression have been known to affect the age of onset of AD, , lifetime cerebrovascular risk factors seem to have a mild combined effect for earlier onset of AD,  thus justifying the search for suitable scales to measure their impact. Associations between risk factors such as midlife hypertension and dementia have been established in several studies, while it has been suggested that anti-hypertensive therapy may reduce incidence of AD.  APOE-ε4 carriers tend to have more AD neuropathologic lesions and coronary artery atherosclerosis than non-carriers, suggesting an interrelation between AD and CHD.  Nevertheless, education, occupation and lifetime leisure activities increase cognitive reserve regardless of APOE haplotypes, but do not affect AD pathology. 
In this study, APOE-ε4 carriers expectedly had earlier onset of AD, confirming the role of genetically mediated neurodegeneration in the evolution of the dementia syndrome. Though routine APOE testing is not recommended, genetic counseling is suggested for all patients with sporadic AD who have strong family history of neurodegenerative diseases.  Accumulation of amyloid-β in the arterial walls may contribute to oxidative stress and chronic hypoperfusion, mediating the relationship between cerebrovascular risk and AD. 
History of depression on pharmacological treatment has resulted in onset of AD more than 2.5 years earlier, reaching almost 3.2 years earlier for women. It has been shown that patients with bipolar disorder have smaller hippocampi and tend to experience more cognitive decline, possibly due to higher cortisol levels, whereas electroconvulsive therapy and use of antidepressants tend to increase hippocampal volume.  Variants of CR1, the complement receptor 1 gene, have been associated with both sporadic AD and major recurrent depression, suggesting a link between these neuropsychiatric syndromes.  Nevertheless, it is still unknown whether depression is a risk factor for dementia or a prodromal symptom, while its association with AD does not seem to be explained by concomitant vascular risk factors. 
It has been stated that marriage is linked to better physical health, well-being and economic benefit, contributing a protective factor against risk of dementia.  In our sample, married patients showed a trend for earlier onset of dementia, a finding that may be due to the fact that the presence of a partner leads to earlier detection of cognitive and functional impairment, possibly hinting at an earlier diagnosis. It is difficult to predict whether marriage actually has an impact on the risk of AD, but it might also be possible that factors related to lower risk do not necessarily translate into a trend for later onset.
Schooling had no influence over the age of AD onset. An earlier study with highly educated individuals also showed no relation between years in full time education and AD onset.  On the other hand, cognitive activity and vascular health may lessen the risk of AD even in APOE-ε4 carriers, explaining to some extent why not all APOE-ε4 carriers develop AD.  It is possible that factors that confer a greater cognitive reserve may be important for rates of cognitive decline,  but not for the onset of the dementia syndrome. Nonetheless, the association between schooling and risk of dementia that other studies have found might reflect confounding by socioeconomic factors.
Male patients had higher 10-year risk for CHD than female patients, but there was no significant difference in age of AD onset between genders. Though patterns of cognitive decline and development of AD do not seem to depend on gender,  elderly women have higher rates of systemic hypertension, diabetes mellitus and hyperlipidemia than similarly aged men.  The Cache County Study  also showed that vascular factors differentially increase risks for AD and vascular dementia according to gender. Cerebrovascular risk seems to have a complex relationship with AD that might not be the same in all populations.
There was a trend for higher Framingham risk scores  to increase the age of onset of AD only in women, while the 10-year projected risk for CHD was marginally significant for later onset of AD in all patients. Though midlife cardiovascular risk might be more important for AD pathogenesis,  it seems feasible that cerebrovascular risk factors are protective against AD in late life, possibly by enhancing cerebral perfusion. Regarding hypertension, it has been shown that midlife anti-hypertensive treatment may reduce incidence of AD by prevention of blood pressure elevation, but might also reduce cerebral damage attributed to hypotension later in life. 
It has been demonstrated that higher systolic and diastolic pressures are associated with increased risk of dementia in elderly Nigerian patients, while anti-hypertensive treatment may lower such risk.  On the other hand, it has been reported that patients with AD usually present with systolic blood pressure readings in the normal range in the years prior to diagnosis, while higher systolic blood pressure in late life reduces the risk of incident AD.  Since Framingham risk scores  take into account only systolic blood pressure in the point total, a deeper evaluation of the relationship between arterial hypertension and AD onset is also required in our population. To the best of our knowledge, ours is the first study to assess the impact of Framingham risk scores  over age of AD onset.
An earlier study  showed that vascular risk factors measured in midlife and late life are associated with an increased risk of vascular dementia in a dose-dependent fashion, but not with AD. However, this effect was much smaller than the effect of copies of APOE-ε4 in that study, confirming that genetic features supersede acquired cerebrovascular risk factors in the development of mechanisms of neurodegeneration. Cerebrovascular risk factors may be assessed either in isolation or in combination, with different results according to the approach , ; hence, a specific and sensitive scale that correlates cerebrovascular risk with AD is required.
Limitations of this study include the fact that Framingham risk scores  were employed as the only measure of vascular risk, and could not be estimated in midlife or by the time of onset of the dementia syndrome (several patients had such scores calculated after the onset of AD). Many patients were already using lipid-lowering therapy when the CHD risk was calculated, possibly resulting in an underestimation of such risk. Lipid-lowering drugs may lower total cholesterol levels while increasing HDL cholesterol,  thus modifying Framingham risk scores.  Even though anti-hypertensive drugs may also lower blood pressure and modify the risk of CHD, this effect is compensated when accounting for Framingham risk scores.  Moreover, we did not assess the impact of dietary patterns or physical activity over age of dementia onset and cardiovascular risk,  another bias if we take into account the fact that aerobic exercise may enlarge hippocampal size  and also increase HDL cholesterol levels,  thus modifying Framingham risk scores.  Finally, in view of the fact that this was not a community-based study, results may not be generalizable to the entire population of our city.
We conclude that APOE haplotypes were the most important factors to influence the age of onset of AD in this sample, translating into a major impact of genetically mediated mechanisms of neurodegeneration in comparison with other risk factors, whereas married status and history of depression were also strongly linked with earlier onset of this dementia syndrome. Cerebrovascular risk factors may increase the risk of AD, but seem to confer protection leading to later dementia onset when present in late life. Effective public health policies should consider these results to help forestall the inception of neurodegenerative mechanisms in the brain.
We acknowledge the financial support by CAPES - Coordenaηγo de Aperfeiηoamento de Pessoal de Nνvel Superior, as well as the help of members from the multidisciplinary team at the Behavioral Neurology Section of Hospital Sγo Paulo, and the staff at the Laboratory of Genetics of the Federal University of Sγo Paulo (UNIFESP). Grants from CNPq - Conselho Nacional de Desenvolvimento Cientνfico e Tecnolσgico, and FAPESP - The State of Sγo Paulo Research Foundation, were also provided for the Laboratory of Genetics. We have no conflicts of interest related to this study.
[Table 1], [Table 2]