Abstract
Objective: To test the hypothesis that patients with Alzheimer disease (AD) who have vascular risk factors have a worse prognosis over 18 months vs those without such risk factors.
Methods: A sample of 224 people with AD and their caregivers were recruited purposively to be representative of people with dementia in terms of cognition, sex, and living situations in a longitudinal study of AD. Standardized instruments measuring cognition, functional status, and neuropsychiatric symptoms were used to collect data. Physical examination and relevant blood tests were performed.
Result: There was no difference in rate of deterioration between people with and without vascular risk factors, except in those who had a cerebrovascular accident (CVA) during the 18-month follow-up (p < 0.001). We considered possible confounders of outcome: sex, age, years of education, severity of dementia, depression, taking cholinesterase inhibitors (AChEIs), and whether those with vascular risk factors were more likely to die, but the results remained unchanged. Stopping AChEIs during the study was associated with cognitive and functional decline (p < 0.001).
Conclusions: Vascular risk factors as measured clinically and biochemically do not significantly increase deterioration at 18 months in people with Alzheimer disease who have a low burden of cerebrovascular risk factors. However, cerebrovascular events are associated with more rapid decline. Vascular risk factors may contribute to the expression of Alzheimer disease initially but are not part of the underlying etiologic process.
Most people with Alzheimer disease (AD) exhibit microvascular degeneration with up to 30% having evidence of cerebral infarction at autopsy.1,2 Epidemiologic studies have shown an association between vascular risk factors and the development of cognitive impairment.3–5 Vascular risk factors allied to the development of AD include diabetes mellitus, thrombotic episodes, atrial fibrillation, smoking, obesity, and atherosclerosis.4–10 This association may be related to shared etiology: hypoxia could influence the β amyloid cascade, vascular damage could lead to cytokine-mediated inflammation,11 or amyloid deposition in cerebral blood vessels may directly increase the risk of hemorrhagic strokes.12 Additionally, there is evidence that cerebrovascular disease and infarction in particular can increase the severity of AD.13
Elevated blood concentration of homocysteine is an independent risk factor for vascular disease14 and may be a risk factor for cognitive impairment.15 The mechanism could be through neurotoxicity,16 increasing oxidative stress, impairment of endothelial function,17 or by homocysteine acting synergistically with β amyloid to promote AD.18 Hyperhomocysteinemia is a potentially reversible condition.18
The effect on rate of progression in AD of vascular risk factors and high homocysteine could help elucidate whether vascular disease has an additive effect or is intrinsic to the pathogenesis of AD. We tested the primary hypothesis that progression of AD in terms of cognition and functioning will be significantly different over 18 months in those who have vascular risk factors (as measured by the Hachinski Ischemic score19 and the Framingham Heart Study algorithm20,21) and high levels of homocysteine vs those who do not.
Methods.
This study is part of a large longitudinal study in people with AD and their carers over an 18-month period.22,23
The relevant local research ethics committees gave ethical approval for the study. Patients with AD and their caregivers were approached through a variety of sources: their local community mental health team, dementia specialist nurses, the voluntary sector (including the Alzheimer’s Society), memory clinics, nursing and residential homes, day hospitals, day centers, and inpatient units. We then contacted caregivers by letter. Many of the patients had not yet received a diagnosis but the doctor in our research team screened and confirmed the diagnosis in all our participants. Interviews with the patients with AD and their carers were carried out at a place of their choice: usually the participants’ home.
The participants were prospectively recruited to be a representative sample of people with AD in terms of severity of cognitive impairment in the community, which was categorized as follows: mild 30%, moderate 40%, and severe 30%.24 Mild impairment was defined by Mini-Mental State Examination (MMSE) > 20, moderate as MMSE ≤ 20 and ≥ 10, and severe as MMSE < 10.25
The inclusion criteria were a standardized diagnosis of dementia26 and fulfillment of criteria for possible or probable AD,27 being aged ≥ 55 years, living in either North London or Essex, and having a caregiver (spending a minimum of 4 hours a week with the person with AD).
Exclusion criteria were having significant vascular or neurologic disease (including having a Hachinski score > 4), enduring mental illness, alcohol or drug abuse, and being unable to comply with the study assessment (due to another disease or inability to understand the national language).
The interviews were conducted by trained, experienced health professionals and comprised a medical history and examination with information to allow a diagnosis of probable or possible AD to be made including a full cardiovascular and neurologic examination.27 Details of vascular risk factors were collected, including history of vascular events such as myocardial infarction (MI), cerebrovascular accidents (CVA), previous hypertension and diabetes mellitus (including details of treatment for both), past atrial fibrillation, and smoking (amount currently in grams of tobacco). Clinically measured vascular risk factors included blood pressure, which was measured after 5 minutes of sitting at both baseline and follow-up. The lowest value was used. If the reading was high, the blood pressure was repeated twice over the next 15 minutes and a mean value taken. Venous blood was collected for total cholesterol, high density lipoprotein (HDL) (using the Roche/Hitachi Modular analyser for both), and HbA1c (the Menarini HA8160 analyser with DCCT-aligned calibration level) where consent was granted. Diabetes was defined by blood glucose level of 11.1 mmol/L or over, previous diagnosis, use of hypoglycemic agent or insulin or high HbA1c level.28 Body mass index and waist/hip ratio (the waist: measured at the umbilicus, and hip: measured at the largest circumference) were measured. Plasma homocysteine specimens were stored at or below −20 within 1 hour of collection. Homocysteine concentrations were determined by pre-column derivitization followed by separation by high-performance liquid chromatography with fluorescence detection on the DS30.
Algorithms were used to generate a more global assessment of vascular disease and included the Hachinski Ischemic Score (HIS)19: a score from 0 to 7 which was used to exclude significant vascular disease for the diagnosis of AD (>4) and was also used to dichotomize patients into two vascular groups (no vascular risk = 0, vascular risk = score 1 to 4). The Framingham Heart Study21 generated a simple coronary disease predication algorithm for patients without overt vascular disease. It includes multiple variables (blood pressure, smoking, total cholesterol, HDL cholesterol, and diabetes) weighted for age and sex. It does not include exercise and obesity. It has been well validated in other populations.20 The Framingham risk score generates a percentage risk of coronary heart disease which can then be dichotomized into low and high risk groups—≤20% and >20%.
Standardized assessment interviews were carried out with caregiver and care recipient. Sociodemographic details were collected including years of education and medication treatment. Three measures of cognition were used: Mini-Mental State Examination (MMSE),25 Severe Impairment Battery (SIB),29 and the AD Assessment Scale–cognition (ADAS-cog).30,31 The Neuropsychiatric Inventory (NPI)32 scored psychological and behavioral symptoms of AD. The AD Co-operative Study Inventory–Activities of Daily Living (ADCS-ADL)33 is a caregiver-rated questionnaire to assess functional impairment, with a range of 0 to 78, where 78 implies full functioning. Depression was measured using the Cornell Scale for Depression in Dementia (CSDD).34
Power calculation.
We assumed that at 18 months, the mean change in terms of MMSE would be three points35 and mortality rates, 21%.36 We postulated that the population with no baseline vascular risk on the HIS score (0) would deteriorate one point less than the mean MMSE and the remainder (vascular risk HIS > 0) would deteriorate one point more, i.e., deteriorating two and four points. On this basis, and knowing that 50% had a HIS = 0, we calculated that we had a power of 80% at significance level of 0.05 of finding a true difference if there were 80 survivors in each group. This would require 102 in each group at baseline.
Data analysis.
SPSS 11.5 was used to enter the data and for the statistical analysis. We calculated the difference in scores from baseline to 18-month follow-up. For rate of cognitive impairment, differences in scores for the MMSE, SIB, and ADAS-cog were used. The NPI difference supplied change in level of behavioral and psychological symptoms. Difference in ADCS ADL showed change in functional abilities. We calculated the means and standard deviations of the differences in terms of neuropsychiatric symptoms, cognition, and functioning. We used t tests to examine associations in continuous data and χ2 analyses to test for significant associations in categorical data. The Mann-Whitney U test was employed to test for associations between ordinal and categorical variables if the data were not normally distributed, and t tests for independent normally distributed continuous data. We also considered associations of decline with the HIS scored continuously and using the cutpoint of 2/3. These results did not differ and we have not reported them separately. To consider independent predictors of decline, we entered the following variables into the stepwise linear regression model (for missing values in the Framingham risk score, homocysteine, BMI, and years of education, the mean was coded in): Step 1: age, years of education, and sex; Step 2: vascular measures: history of myocardial infarction, atrial fibrillation, hypertension, CVA (including over the course of the study) and diabetes mellitus, HIS (Hachinski Ischemic Score),19 smoking habit, blood pressure, Framingham risk score,20,21 and homocysteine level; Step 3: medication that improved vascular integrity included aspirin, statins, and calcium channel blockers; and Step 4: possible confounders such as regular antipsychotic, antidepressant, and cholinesterase inhibitor (AChEI) treatment at baseline and follow-up and depression as rated on the CSDD.34
Results.
Demography.
We interviewed 224 people with AD at baseline. At 18 months, 167 (74.6%) people completed follow-up. A total of 48 (21.4%) had died. Eight (3.6%) refused to take part. One (0.4%) had moved too far away to be interviewed. Participants who died were older (mean: 84.4 vs 80.0, p < 0.001), more cognitively impaired on MMSE (mean: 10.8 vs 15.8, p < 0.001), and more likely to be living in 24-hour care accommodation (24 [50.0%] vs 48 [28.7%], p < 0.01), but those who refused were no different from the rest of the population in these terms. Furthermore, those who died were no more likely to have a history of diabetes, hypertension, or treatment with antihypertensives, previous cerebrovascular accident (CVA), or difference in HIS scores from the rest of the sample. Mean years of education in the sample at baseline were 9.41 with a range of 1 to 16. Table 1 shows the demographic information for the two groups, baseline and 18-month follow-up. Table 2 shows the range and severity of the sample based on the MMSE.
Table 1 Demographic data at baseline (B/L) and follow-up of patients with Alzheimer disease
Table 2 Severity of impairment on the Mini-Mental State Examination (MMSE) at baseline (B/L) and follow-up
Vascular demographics.
At baseline, 11 (4.9%) had a history of myocardial infarction (MI); no one had an MI in the following 18 months. At baseline, 15 (6.7%) had a history of atrial fibrillation, and 27 (12.1%) had diabetes. Sixity-one (27.2%) were taking antihypertensives (β blockers, ACE inhibitors) and 25 (11.2%) were taking a calcium channel blocker. A total of 103 (46%) had never smoked, 81 (36.2%) were former smokers, 31 (13.9%) were currently classed as smokers. Nineteen (8.4%) had a history of CVA at baseline with 7 (3.1%) reporting a CVA during the 18 months follow-up. Fifty-three (31.6%) were overweight according to BMI with 28 (16.1%) being obese. Waist/hip ratio mean was 0.9 with a range of 0.8 to 1.0 for men and 0.84 with range of 0.7 to 1.0 for women.
Vascular factors and deterioration.
The mean changes in cognition are shown in table 3. Changes in score for behavioral and psychological symptoms and functional ability are shown in table 4. In the MMSE, the mean difference was three points for the cohort as had been predicted.35 HIS was coded as a dichotomous variable. At follow-up, 70 (41.9%) had a score of 0 with no significant vascular history. A total of 97 (58.1%) scored 1 or over (maximum of 4). There was no association with rate of deterioration in terms of cognition, functioning, or neuropsychiatric symptoms. Total risk for Framingham prediction was only available in 130 of our sample at 18 months as some participants did not consent to having their blood taken. In our sample, 32 (24.6%) were classed as high risk with 98 (75.4%) as low risk. There were no significant associations with deterioration. The only cardiovascular factor that was associated with increased cognitive deterioration was having a CVA during the 18-month study period (p < 0.001).
Table 3 Deterioration in measures of cognition for vascular algorithms and other known risk factors in Alzheimer disease
Table 4 Deterioration in functional ability and neuropsychiatric symptoms for vascular algorithms and other known risk factors in Alzheimer disease
For the other risk factors considered—hypertension (systolic or diastolic) at baseline and follow-up or history of hypertension at baseline, smoking, hypercholesterolemia and hyperhomocysteinemia, history of myocardial infarction and atrial fibrillation, body mass index and waist/hip ratio, and diabetes (NIDDM, IDDM)—there was no association with neuropsychiatric symptoms or cognitive or functional decline over any of the measures. Treatment with statin medication, aspirin, or calcium channel blockers was not associated with deterioration.
Mediators and confounders.
We considered a number of confounders, namely age, sex, psychotropic medication (AChEI, antipsychotic), and depression as measured on the CSDD. Neither years of education, age, nor sex was associated with rate of decline. There was no difference in the rate of deterioration between the groups prescribed an AChEI throughout (82; 57.4%) and those who were not. However, when those who were prescribed AChEI from the start and those who had never been prescribed AChEI were compared to participants who had their AChEI stopped during the 18 months (16; 9.6%), there was a difference between the groups in rate of deterioration in cognition and functioning (MMSE: 7.4 vs 2.7; p < 0.01, SIB: 32.1 vs 8.0; p < 0.05, ADAS-cog: 17.4 vs 4.8; p < 0.01, ADCS-ADL: 24.9 vs 11.0; p < 0.001) but not in reduction of neuropsychiatric symptoms. Depression at baseline was associated with death during the follow-up period (15; 31.1% vs 33; 19.0%, p < 0.05) but there was no association with rate of deterioration. There was no difference in rates of death between the vascular groups (HIS > 1:28 [23.0%] vs HIS < 1:20 [22.5%]).
Regression analysis.
The first step in stepwise linear regression analysis included demographic factors (age, sex, and years of education). None of these data significantly predicted deterioration. Of the vascular risk factors for Step 2 (HIS score, diabetes, CVAs, history of MI and atrial fibrillation, body mass index, history of hypertension, blood pressure, Framingham risk score, cholesterol level, and smoking) only having a CVA during the study period was an independent predictor of deterioration. Step 3, adding medication prescribed at baseline that improved vascular integrity (aspirin, statins, and calcium channel blockers), made no significant change to the model. Step 4 included the possible confounders such as being prescribed an antipsychotic, treatment with an AChEI, or stopping the AChEI during the study period. For the complete model, having a vascular event (CVA) during the 18-month follow-up period contributed considerably (p < 0.001) to the rate of deterioration in cognition and functioning along with ceasing the AChEI (p < 0.01). For neuropsychiatric symptoms, there was less of an association with CVA (p < 0.01) and stopping the AChEI had no effect on the variance.
Discussion.
Clinical and biochemical indicators of vascular disease were not associated with rate of decline in cognition, neuropsychiatric symptoms, or functioning in AD over 18 months. The strength of our study is that we used several well-validated clinical and biochemical measures to rate vascular risk and clinical outcome. In addition, although people with high levels of cholesterol, blood pressure, and diabetes die earlier from cardiovascular events, which could have removed them from our analysis during the follow-up period, introducing a survival bias,11,37,38 this is unlikely as the baseline data for those who died did not show increase in any vascular markers.
Cerebral infarction during the follow-up period was associated with greater cognitive and neuropsychiatric decline. This is consistent with other studies that have shown CVAs to contribute significantly to the clinical severity of cognitive impairment in AD.13 Stopping AChEIs in our study was significantly associated with poorer outcome in terms of worsened cognitive and functional performance. However, it is difficult to draw conclusions about direction of causality as we did not collect data about the reasons why the AChEI had been stopped—these might have included lack of efficacy but this result may indicate a potential withdrawal effect.
There are a number of limitations to our study. The accuracy of clinical measures as a reflection of cerebrovascular disease is questionable.39 We have however measured them in a number of ways, including biochemically, and the lack of association remains robust. Patients who develop AD show decreases in plasma cholesterol levels preceding development of cognitive symptoms which could obscure the past effects of higher cholesterol levels in earlier life.40 Our study, based on clinical criteria and exclusion, does not have neuropathologic confirmation of diagnosis and may have included some cases of non-AD dementia. To avoid these, we used rigorous criteria so that the participants’ clinical features conformed to the best validated AD criteria3 and that they had scores of below 5 on the HIS. Thus we may have missed some people who have AD with higher levels of vascular disease and it is possible that in those relatively rare cases, vascular risk affects rate of progression. We have, however, used the generally accepted research criteria for AD and our sample will be similar to others. Our numbers in the subgroup analyses using Framingham criteria and homocysteine are small but the results are consistent with our well powered primary analysis. Dietary insufficiency is a recognized sequela of AD and may result in higher than expected levels of homocysteine once the disease is established, thereby negating our study’s ability to demonstrate differences within the sample.
Acknowledgment
The authors thank the patients with AD and their families, friends, and other carers who participated in the study. They also thank the consultant psychiatrists and community mental health teams along with the staff of the nursing and residential homes for support and help.
Footnotes
-
Editorial, see page 1326
See also page 1363
Disclosure: Lundbeck SA funded the data collection but did not participate in preparation, review, or approval of the manuscript. C.K. and G.L. have received grants from Lundbeck SA in excess of $10,000.
Received February 24, 2006. Accepted in final form July 5, 2006.
References
Disputes & Debates: Rapid online correspondence
- Relationship of vascular risk to the progression of Alzheimer disease
- Reply from the authors
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.