Abstract
Background: The prevalence of AD appears to vary widely in different ethnic groups. Certain risk factors for AD are well established for the general population, but there is little information regarding the relevance of these risk factors in specific ethnic groups.
Objective: The authors examined the risk of AD associated with the APOE-ε4 allele, the APOE-ε2 allele, smoking, alcohol consumption, history of hypertension, low educational level, estrogen replacement therapy, and history of head trauma with loss of consciousness among samples of white non-Hispanics (WNH) (392 AD patients, 202 normal subjects) and white Hispanics (WHIS) (188 AD patients, 84 normal controls).Design:— This was a case-control study of patients evaluated at an outpatient memory disorders clinic and control subjects recruited from a free memory screening offered to the community.
Results: Increased risk for AD was associated with the APOE-ε4 allele after controlling for age, education, and gender among WNH (OR = 3.5; 95% CI = 2.3 to 5.5) and WHIS (OR = 3.1; 95% CI = 1.7 to 5.8). No protective effect was conferred by the APOE-ε2 allele, although this relationship approached significance among WNH (p = 0.02). Low levels of education increased the risk for AD among WNH (OR = 3.1; 95% CI = 1.8 to 5.9) but not WHIS. Alcohol use and hypertension approached significance as risk factors in WNH (p < 0.05) but not WHIS. Estrogen replacement treatment approached significance as a protective factor in both ethnic groups (p < 0.05).
Conclusions: Although the APOE-ε4 allele is a risk factor for AD among WHIS and WNH, other risk factors such as low education and hypertension appear to be important only for WNH. Risk factors for AD reported or suggested previously that were not confirmed by this study include smoking and head trauma with loss of consciousness.
An elevated risk for AD associated with the apolipoprotein E epsilon 4 allele (APOE-ε4) localized on chromosome 19 is well established,1,2 as is decreased risk for AD associated with the ε2 allele of the APOE gene.1-3 Putative nongenetic risk factors for AD include age, head trauma, alcohol abuse, smoking, lack of estrogen replacement therapy (among women), hypertension, and low education. However, these results have primarily been based on white non-Hispanic (WNH) subjects. A recent study suggests that the risk for AD may be greater in Hispanics than in WNHs, with this risk attributed to factors other than the APOE locus.4 The Hispanic elderly population (age 65+) in the United States is projected to increase by 555% over the next 30 years, compared with 93% among WNH elderly.5 Thus, the investigation of risk factors for AD in distinct ethnic/cultural groups is important. However, other than the APOE-ε4 and -ε2 genotypes, AD risk factors have not been evaluated among Hispanics.2,4,6-8 Therefore, we investigated multiple risk factors for AD among samples of elderly WNHs and white Hispanics (WHIS).
Methods.
Participants.
This study included 594 WNH (392 AD patients, 202 control subjects) and 272 WHIS subjects (188 AD patients, 84 control subjects). Ethnicity was based on self-report by the patient or primary caregiver. To evaluate the risk for AD among a more homogenous Hispanic sample, patients and control subjects who were both Hispanic and black, or Hispanic and Jewish, were excluded from the study. Further, all Hispanic subjects in this study were required to speak Spanish as a primary language and to have been born in a Spanish-speaking country (75% were born in Cuba). Participants were evaluated at the Wien Center for Alzheimer’s Disease and Memory Disorders, an affiliated program of Mount Sinai Medical Center and the University of Miami School of Medicine. All patients met diagnostic criteria for possible or probable AD according to National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders criteria.9
The normal control sample included 243 subjects (85%) who were recruited at the Wien Center from a free memory screening offered to community residents. Twenty-six control subjects (9%) were patients evaluated by a neurologist (R.D.) at the Wien Center for noncognitive symptoms, who were found to be without objective cognitive deficits and whose examination did not result in a primary neurologic diagnosis. The remaining 17 control subjects (6%) were recruited as spouse controls and assessed with the free memory screening battery. All control subjects were ≥65 years old.
Instruments.
The tests used in the free memory screening evaluation included 1) the Mini-Mental State Examination (MMSE),10 with age and education adjustment,11 and 2) a four-trial recall of the three words given in the MMSE. The first trial followed the serial sevens task, which is routine in the MMSE protocol; the second trial was given at completion of the MMSE and an additional task of spelling the word world backwards; the third trial was given after the subject repeated six items from a grocery list and the administration of the Hamilton Depression Rating Scale12; and the fourth trial was given after subjects attempted to recall six items from the grocery list and the administration of a test of verbal fluency for categories. The criteria used to classify memory screening subjects as normal control subjects were 1) a score on the MMSE greater than a median score from a large population-based study, adjusted for age and education,11 and 2) a score of ≥9 (out of 12) on the four-trial recall. Recall scores of <9 have shown high sensitivity (96%) and specificity (87%) in differentiating patients with mild to moderate dementia from normal elderly control subjects (submitted manuscript).
APOE genotype analysis.
APOE genotypes were determined at the Center for Research in Neurodegenerative Diseases, University of Toronto, Canada, and the Roskamp Laboratory, Department of Psychiatry, University of South Florida, Tampa, using methods previously described.13,14 Risk was assessed for the APOE-ε4 allele by combining the ε3/ε4 and ε4/ε4 genotypes. The protective role of the APOE-ε2 allele was assessed with the ε2/ε3 genotype. There were no ε2/ε2 subjects among our samples of AD patients and control subjects. For the genotypic risk, the ε3/ε3 genotype was used as a reference. Subjects with the ε2/ε4 genotype (n = 19; 2%) were excluded from the analyses of risk associated with the APOE-ε4 and APOE-ε2 alleles to create more homogenous risk factors. They were, however, included in other analyses, i.e., the calculation of allele frequencies.
Measures.
Information on risk factors was self-reported in control subjects and assessed in patients via knowledgeable family members. History of smoking was assessed as 1) never smoked, 2) <1 pack/week, 3) <1 pack/day, 4) ∼1 pack/day, 5) >1 pack/day. The data were recoded as ever (smoked) versus never (smoked), consistent with methods previously used.15 History of alcohol consumption was assessed as 1) never consumed, 2) <1 drink/month, 3) <1 drink/week, 4) <1 drink/day, 5) 1 to 2 drinks/day, 6) >2 drinks/day or heavy drinker. Subjects who reported drinking >2 drinks per day were classified as having a history of significant alcohol use. History of head trauma resulting in loss of consciousness was determined by a semistructured interview that included questions about the number of episodes of head injury (e.g., motor vehicle accident, fall from a height), the age at time of head injury, history of any loss of consciousness and its duration, and external injuries or sequelae from each head injury. Any head injury after the onset of AD was excluded. History of estrogen replacement therapy was assessed with a semistructured interview that addressed specifically whether estrogen or other female hormones had been used, age when they were first used, and duration of use (continuous or discontinuous). A history of oophorectomy and the age at menopause were also obtained. Hypertension was defined as history of elevated blood pressure, and past and current use of any medication for treatment of hypertension (not necessarily before the onset of AD). Education was defined as years of formal schooling, with low education categorized as ≤10 years.
Statistical methods.
Allele frequencies for the APOE-ε2, -ε3, and -ε4 alleles were formulated by counting alleles and calculating sample proportions. The risk for AD was assessed using logistic regression analyses that yielded age-, education-, and gender-adjusted odds ratios for AD in the two ethnic groups. Putative risk factors included APOE-ε4 genotype, alcohol use, head trauma with loss of consciousness, hypertension, and low education. Putative protective factors included APOE-ε2 genotype, smoking, and estrogen replacement therapy. The alpha level was set at 0.01 for the logistic regression analyses, to control for potentially inflated family-wise Type I error rates. Fisher’s exact tests were used for all cross-tabulations. Total mean scores for clinical and demographic variables between the ethnic groups were compared by independent sample Student’s t-tests. The alpha level was set at 0.05 for these analyses.
Results.
Table 1 shows a summary of demographic data across ethnic and cognitive groups. A series of Student’s t-tests indicated significant ethnic differences with respect to age, education, and MMSE scores. Compared to WNH, WHIS were 1) younger, 2) less educated, and 3) more cognitively impaired, according to their MMSE scores. Table 2 shows the prevalence rates of the explored risk factors for AD by ethnicity among AD patients and control subjects. Among control subjects, a history of smoking was more common for WNH than WHIS (Fisher’s exact test, p = 0.0004), whereas low education was more prevalent among WHIS than WNH (Fisher’s exact test, p < 0.0001).
Demographic and cognitive variables by ethnic and diagnostic group*
Prevalence of nongenetic risk factors for AD by ethnicity among AD patients and controls
APOE allele frequencies for the two ethnic groups are given in tables 3 and 4⇓. Table 3 shows that the frequency of the ε4 allele was greater in AD patients than in control subjects in both ethnic groups. No difference in ε4 frequency was observed between ethnic groups among AD patients or control subjects. The APOE-ε2 allele frequency between the ethnic groups was not different in patients or control subjects (see table 4). The ε2 allele frequency was higher among control subjects than in AD patients for WNH, but not for WHIS.
APOE ε4 allele frequencies by ethnicity and diagnosis
APOE ε2 allele frequencies by ethnicity and diagnosis
Table 5 gives the results of the logistic regression analyses, which showed several significant risks for AD after adjusting for age, education, and gender. Risk for AD was associated with the APOE-ε4 allele among WNH (OR = 3.5; 95% CI = 2.3 to 5.5; p < 0.0001) and WHIS (OR = 3.1; 95% CI = 1.7 to 5.8; p = 0.0004). The decreased risk for AD among carriers of the APOE-ε2 allele in the WNH sample approached significance (OR = 0.4; 95% CI = 0.2 to 0.9; p = 0.02) but was nonsignificant in the WHIS sample. Logistic regression analyses did not show an increased risk for AD associated with smoking or head trauma with loss of consciousness. Estrogen replacement therapy approached significance as a protective factor in both WNH (OR = 0.6; 95% CI = 0.1 to 1.0; p = 0.04) and WHIS (OR = 0.4; 95% CI = 0.2 to 1.0; p = 0.04). Hypertension approached significance among WNH (OR = 1.5; 95% CI = 1.0 to 2.3; p = 0.03) but was nonsignificant among WHIS. In addition, a history of significant alcohol consumption approached significance among WNH (OR = 2.5; 95% CI = 1.1 to 6.7; p = 0.04) but not WHIS. There was an elevated risk for AD associated with low education among WNH (OR = 3.1; 95% CI = 1.8 to 5.9; p = 0.0002), but not WHIS.
Odds ratios for AD by risk factors and ethnicity
Discussion.
This study identified several risk factors for AD among samples of WNH and WHIS. These factors were similar across groups, despite several demographic differences between the two ethnic samples. However, differences in risk factors between the ethnic groups were also apparent.
The results show that the increased risk for AD associated with the APOE-ε4 allele was consistent across both ethnic groups. The elevated risk associated with the ε4 allele among Hispanics and WNHs is in agreement with considerable published research.1,2,4 Among WNH, the observed ε4 allele frequencies of 0.28 in AD patients and 0.14 in normal control subjects are consistent with the frequencies of 0.30 in AD patients and 0.11 in normal control subjects reported in the large population-based Framingham study composed primarily of WNH subjects.1
In the WHIS sample, the ε4 allele frequencies of 0.26 for AD patients and 0.12 for control subjects compare favorably with those reported in two studies of AD from Spain. Adroer et al.6 reported an ε4 allele frequency of 0.29 among patients and 0.06 among control subjects. Similarly, Ibarreta et al.7 observed an ε4 allele frequency of 0.34 in AD patients and 0.04 in control subjects. These findings are in accord with a recent meta-analytic study indicating that the ε4 allele may increase the risk for AD across distinct ethnic groups.2 However, our results contrast with a recent study reporting a lack of association between the ε4 allele and AD in Hispanics residing in New York City.4
One explanation for this difference may be that whereas the New York City study4 included mainly Hispanics of Dominican and Puerto Rican origin, our Hispanic sample was primarily Cuban. Variability in the prevalence of the ε4 allele has been noted within ethnic subgroups, especially African populations,16 and it is reasonable to expect differences between Hispanic subgroups. For this reason, we attempted to create a more homogenous group by excluding Hispanics of African or Jewish ancestry. Another explanation for differences between the current investigation and the New York City study4 may be related to the sampling procedures. For example, clinic-based investigations among Caucasians have a found a stronger APOE-ε4 association with AD than have population-based studies.2 Clinic samples may overestimate the association of APOE-ε4 with AD because of subject selection bias.17
The ε2 allele frequency was nearly twice as high in control subjects than in AD patients in both ethnic groups; however, only the differences among non-Hispanics approached significance. Previous reports among non-Hispanics have shown a protective effect of the ε2 allele against the risk of AD.2-4 However, the results among Hispanics have been inconsistent.2,4,6,7,18 These inconsistencies may be due to insufficient power or possibly to ethnic diversity among Hispanic subgroups. Future research incorporating more Hispanics should help clarify this issue.
Our data indicate that low education represents a significant risk factor for AD in WNH subjects. This is consistent with most reports, including population-based studies of Caucasians in several countries.19,20 The mechanisms by which low education is associated with AD are of considerable interest.21 For instance, higher education may have a direct protective effect by increasing cognitive and synaptic reserves. Education may also be a surrogate for other risk factors. In our study, low education was not associated with AD among WHIS subjects. Post-hoc analysis showed that WHIS control subjects had 10.0 ± 4.9 years of education, compared with 10.3 ± 5.4 years of education among WHIS AD patients (Student’s t-test = 0.4, p = 0.66). The circumstances associated with Cuban emigration to the United States may have affected this finding. For example, many elderly Cuban American immigrants have had to work in occupations that were below their level of education. If low education serves as a surrogate for other risks for AD, such as occupational exposure, then the protective effect associated with high education would be obscured in this sample. Because this study is the first, to our knowledge, to investigate the relation of AD and education among Hispanics, continued research in this area is important.
The role of alcohol abuse as a risk factor for AD was not supported in the current study, although this relation approached significance among WNH. The prevalence of this potential risk factor was low among control subjects, being 4% among WNH and 1% among WHIS. Hence, statistical power may have been compromised, especially among WHIS. A recent study suggests that alcohol use is associated with an earlier age at onset in AD patients with a history of heavy smoking and a low genetic susceptibility for AD.22 However, other studies have not indicated a relationship between alcohol intake and AD.23-24
We did not observe an increased risk of AD associated with a history of head trauma with loss of consciousness in either ethnic group. The relationship remained nonsignificant after the criteria were expanded to include head injury without loss of consciousness. History of head trauma has been associated with AD in several case-control studies, with the risk of head injury ranging from 1.525 to 5.5.26 However, other studies have shown no relationship between these variables,27,28 consistent with our results. The lack of association in the current study may be due to the low prevalence of head trauma with loss of consciousness, with prevalence rates of 5% among WNH and 8% among WHIS.
Studies have also addressed the relationship of smoking to AD, with equivocal findings. A meta-analytic study based on 11 case-control studies of AD showed an inverse association between risk for AD and smoking; decreased risk for AD was linked with increased smoking.24 However, a recent population-based incidence study from Europe revealed that smokers had a relative risk of 2.3 for AD, compared with those who had never smoked.29 Our findings are in accord with other investigations that have not implicated smoking as a protective or risk factor for AD.24,30
There is increasing evidence that estrogen replacement therapy may reduce the risk for AD in women.31 In the current study, there was a trend for estrogen replacement therapy to be related to reduced odds for AD in both ethnic groups. Tang et al.32 showed a linear relationship between estrogen use and protection against AD, with increased protection associated with greater use. In the current study, the median duration of use was only 2 years, comparable with a national study that showed a median length of use of 3 years.33 The true benefit of estrogen replacement therapy may thus be greater than our results indicate.
In the current study, hypertension was not observed to be a risk factor for AD, although this relationship approached significance among WNH subjects. Prior research has shown it to be a risk factor for the onset of vascular dementia,34,35 although until recently hypertension has not been linked with AD. Studies by Skoog et al.34 in whites and Launer et al.36 in Japanese Americans show that a history of hypertension may increase the risk for AD. The findings of the current study are, however, in accord with results reported by Rockwood et al.37 that do not show a relationship between hypertension and AD.
There are several limitations to the current study. Because the results are based on a clinic sample, the findings may not be representative of AD patients in the community. However, the mechanisms by which these variables confer risk for dementia are unlikely to differ between clinic- and population-based studies. The findings in WHIS subjects may not be indicative of Hispanics in general, as our sample was primarily Cuban American. This group has been shown to differ from other Hispanic groups on factors such as education,38 smoking,39 and health status.40 The current investigation did not include family history, a variable that has consistently been shown to be a risk factor for AD.41 However, the APOE-ε4 allele, which may be responsible for much of the risk for AD associated with a positive family history of dementia, was included in the current study.
Additionally, the method of selecting control subjects primarily from a memory screening program may have resulted in the inclusion of some subjects in the incipient stages of dementia. However, memory complaints are very common in community-dwelling elderly subjects,42 and research has shown that such complaints in persons with normal cognition are not related to subsequent dementia.43 More compelling evidence that the control subjects in the current study are appropriate is shown by their ε4 allele frequency. We observed an ε4 allele frequency of 0.14 among both samples of elderly control subjects, which compares favorably with the frequencies of 0.14 and 0.11 reported for Caucasian and Hispanic controls, respectively, in the recent APOE meta-analysis.2
This study shows that risk factors as well as protective factors for AD are largely consistent across WHIS and WNH. However, the expression of the effects of these factors may change according to ethnic variations in the prevalence of genetic and environmental variables. For example, estrogen use, alcohol consumption, and level of education have been shown to be different among distinct ethnic groups. These factors may alter the prevalence of AD in the ethnic population and thus should be explored independently in future studies of AD. Since the prevalence of AD may be greater in elderly ethnic minorities than in Caucasians,4 continued research is warranted to identify such factors.
- Received April 20, 1998.
- Accepted October 17, 1998.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
