APOE Genotype, Family History of Dementia, and Alzheimer Disease Risk: A 6-Year Follow-up Study

Background  Both family aggregation and apolipoprotein E (APOE) ε4 allele are well-known risk factors for dementia, but the relation between these two factors remains unclear.

Objective  To explore whether the risk of dementia and Alzheimer disease (AD) due to a positive family history is explained by APOE genotypes.

Design  Community-based cohort study.

Setting  The Kungsholmen district of Stockholm, Sweden.

Participants  A total of 907 nondemented people 75 years or older, followed up for 6 years to detect incident dementia and AD cases according to the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition.

Main Outcome Measures  Risk of dementia and AD by Cox proportional hazards models after controlling for several potential confounders.

Results  Subjects who had at least 2 siblings with dementia were at an increased risk of AD. Individuals with both APOE ε4 allele and at least 2 affected first-degree relatives had a higher risk of dementia and AD compared with those without these 2 factors. Similar results were obtained for history of dementia separately in parents or siblings. Among the ε4 allele carriers, subjects with 2 or more first-degree demented relatives had increased risk of dementia and AD, whereas no increased risk was detected among non–ε4 carriers.

Conclusions  Family history of dementia was associated with an increased risk of dementia and AD in this very old population, but only among APOE ε4 carriers. This suggests the existence of other genetic or environmental risk factors that may be active in the presence of the APOE ε4 allele.

The role of both family history of dementia and the apolipoprotein E (APOE) gene in the development of Alzheimer disease (AD) has been extensively investigated.¹ There is strong evidence to suggest that APOE ε4 allele carriers, as well as subjects with a family history of dementia, have an increased risk of AD.^2,3 Familial aggregation and genetic risk factors appear to be most influential in AD at relatively early ages.^4,5 However, there are reports supporting an effect of both familial aggregation and APOE ε4 even in late-onset AD,^6-8 although a lower effect in comparison with early-onset cases has been detected.⁹

It is hypothesized that APOE ε4 allele might explain the association between family history of dementia and AD. Previous studies have tried to evaluate this hypothesis, but to what extent familial aggregation is due to the association between the ε4 allele and AD remains equivocal. Some studies indicated that ε4-positive patients with AD tended to have a higher rate of family history of dementia than ε4-negative patients.^2,10,11 Conversely, patients with family history of AD are also more likely to carry the ε4 allele than patients without family history.¹² Other studies, however, showed that the APOE ε4 allele was not related to familial aggregation of AD.^13,14

Most previous analyses have been hospital-based case-control studies. Because of ascertainment bias and severe truncation of data, these studies might overestimate the effects of family history and APOE ε4 allele, especially in very old people. Only a small-scale prospective study has examined both family history of dementia and APOE ε4 allele in relation to AD risk among people 75 years or older.¹⁵

In a previous study within the Kungsholmen Project, a strong familial aggregation was detected among prevalent cases of late-onset AD, but the contribution of the APOE ε4 allele was not considered.⁷ In the present study, we examined the 6-year follow-up data from the same project to explore whether the risk of dementia and AD due to a positive family history is explained by APOE genotypes.

The study base of the Kungsholmen Project included all registered residents 75 years or older of the Kungsholmen district of Stockholm, Sweden, in October 1987. A total of 1810 subjects (74.6% of the eligible subjects) participated in the baseline survey (1987-1989).¹⁶ Of the 1473 baseline participants who were diagnosed as being nondemented by a 2-phase design, 172 subjects were excluded because of refusal or moving at first follow-up (1991-1993). Of the remaining 1301 subjects who underwent the first follow-up evaluation, 176 did not provide information on family history and 218 had missing data on APOE genotype. Therefore, the study population constituted 907 baseline nondemented individuals.

At first follow-up, 167 persons received a dementia diagnosis, and 29 individuals refused to undertake the second follow-up. Thus, 526 non-demented alive subjects were further followed for another 3 years and examined during 1994 to 1996. For subjects who had died during the first (n = 185) and second (n = 122) follow-up periods, information regarding health status was obtained from medical records and death certificates.

All phases of the project were approved by the Ethics Committee of the Karolinska Institutet, Stockholm.

A standardized protocol was used at baseline and follow-ups, which included an interview on family and personal history, a clinical examination, and psychological tests. The diagnosis of dementia was made according to the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition criteria¹⁷ following a 3-step diagnostic procedure used at the baseline survey.¹⁸ Our criteria for AD were similar to those of the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association¹⁹ for probable AD. The same diagnostic method was used for deceased subjects, based on the clinical records, discharge diagnoses, and death certificates.

Information on a history of dementia among first-degree relatives (ie, parents and siblings, not children) was obtained through a structured interview. For most subjects, multiple informants were sought to supplement and verify these responses. In our study, offspring were excluded because most offspring were too young to be at risk for dementia and AD. We accepted as “demented” only the family members reported as having both memory deficits and behavioral disturbances or time or space disorientation.⁷ The family history interview collected demographics for each first-degree relative and other information on family medical history. APOE genotypes were determined by a standard procedure.

Data on demographics (age, sex, and education) and the number of siblings were collected at baseline interview. Educational levels were based on formal schooling. Information on vascular disorders at baseline, including heart disease (International Classification of Diseases, Eighth Revision, codes 410-414, 427, and 428), cerebrovascular disease (codes 430-438), and diabetes mellitus (code 250), were obtained from the computerized Stockholm Inpatient Register system that encompassed all hospitals in Stockholm since 1969.

We used 1-way analysis of variance for the comparison of continuous variables and χ² test for categorical variables. Several Cox proportional hazards models were constructed. First, we estimated the relative risk of dementia and AD associated with family history of dementia by different affected family members, controlling for age (in years), sex, education (<8 vs ≥8 years), number of siblings, vascular disease, and APOE genotypes (categorized as ε2/ε2 or ε2/ε3, ε3/ε3, and any ε4). Then, as preliminary analysis showed similar dementia risk between subjects with none and 1 affected first-degree relative or sibling, these 2 categories were combined. In all succeeding analyses, we treated family history of dementia as a dichotomous variable (≥2 vs <2 demented relatives or siblings). We examined the combined effect between different affected family members and APOE ε4. On the basis of the hypothesis that APOE ε4 allele might modify the risk effect of family history on dementia, we conducted an analysis stratifying by APOE ε4 status. In addition, sensitivity analyses for the missing APOE genotypes (n = 218) were performed by creating 2 extreme options. These analyses assumed that all subjects with missing values of APOE genotypes had either no ε4 allele (option 1) or at least 1 ε4 allele (option 2).

Among the 907 subjects, 265 developed dementia (203 with AD) during a mean follow-up of 5.1 years (SD, 2.1 years; maximum, 8.2 years). Table 1 shows baseline characteristics of the study population by number of first-degree relatives with dementia. Overall, 31.1% (n = 282) of the study population reported a family history of dementia. No obvious group differences appeared for age, sex, education, Mini-Mental State Examination score, vascular disease, and family history of Parkinson disease. Subjects with a family history of dementia, compared with those without family history, had a higher mean number of siblings (F = 10.58, P<.001) and a slightly higher proportion of ε4 allele carriers (32.6% vs 25.6%; χ² = 4.78, P = .03).

No significant difference was found in mean number of siblings, age at onset of dementia in first-degree relatives, age at death of parents, and birth order between nondemented subjects and those who became demented during the entire follow-up period.

Subjects with 2 or more demented first-degree relatives had a nonsignificantly increased risk of dementia and AD, but subjects with at least 2 affected siblings had significantly increased risk of AD, which was still marginally significant when APOE genotype was included in the model (Table 2).

The combined effect of family history of dementia and APOE ε4 allele was examined (Table 3). Subjects with both at least 2 affected first-degree relatives and APOE ε4 allele had a higher risk of dementia and AD than those in the reference group. The combined effects of either at least 1 parent or 2 siblings with dementia and APOE ε4 allele were similar to those in subjects with at least 2 first-degree relatives with dementia.

Stratum-specific analysis by APOE ε4 status showed that the risks of dementia and AD for having at least 2 first-degree relatives or siblings were elevated only in the presence of APOE ε4 allele (Table 4).

Sensitivity analyses of missing APOE genotypes led to results similar to those reported in Table 3 and Table 4 (data not shown).

In this prospective community-based study, we found that a family history of dementia continued to influence the occurrence of dementia, and AD in particular, even after 75 years of age. Our data showed a combined effect of familial aggregation of dementia and APOE ε4 allele on the risk of dementia and AD. Subjects with 2 or more demented first-degree relatives or siblings had a substantially increased risk of dementia and AD only in the presence of APOE ε4 allele.

A family history of dementia is generally considered an established risk factor for AD.^6,7 However, several prospective studies, including a pooled analysis, did not find an association between family history of dementia and the risk of AD.^20,21 By contrast, the present study showed that subjects with 2 or more demented siblings had increased risk of AD independent of several potential confounding factors. This indicates that the impact of a positive family history on dementia and AD persisted in this very old population. Our finding was in line with other earlier studies of AD,^6,7 although the risk estimation is weaker than previously reported.

Several studies suggested that susceptibility to AD was determined by interaction of APOE genotypes with other genetic and environmental factors.^4,22,23 One study found that individuals with APOE ε4 allele or a family history of dementia had a 9-fold increased age-specific risk of dementia than did those with neither of them.¹⁵ Another study demonstrated that family clustering of AD was due largely to factors other than APOE genotype.²⁴ In agreement with previous reports,^2,10 our study showed that having both family history of dementia and APOE ε4 allele caused an increased risk of dementia and AD, particularly for subjects with either at least 1 parent or 2 first-degree relatives or siblings with dementia.

In our study, positive family history of dementia, in the absence of APOE ε4 allele, was not associated with dementia and AD risk. On the contrary, other studies suggested that family history was a risk factor for dementia, independent of APOE ε4 allele.^22,25 The effects of a family history on the development of dementia and AD among subjects with and without APOE ε4 allele may be age dependent. Our sample consisted of subjects with a minimum age of 75 years at entry. In addition, there were few subjects with 2 or more affected first-degree relatives or siblings and too few with dementia among the ε4-negative group. Thus, the statistical power may not be enough to detect a relevant association. An obvious familial aggregation observed only among the APOE ε4 carriers may suggest that there are some additional family-linked factors that may be active in the presence of APOE ε4 allele. This is consistent with the notion that the APOE ε4 allele is responsible for only a part of the familial risk. Therefore, other genetic or shared environmental factors need to be explored.

There are several caveats to the interpretation of our results. First, misclassification in the assessment of dementia history in first-degree relatives might have occurred because of inaccurate reporting, especially concerning those of older relatives. However, information on family history was collected through multiple informants, which was shown to be reliable in other studies.²² Second, there may be a cohort effect for different risk estimates, in that parents would be less likely to be reported as having dementia than siblings because of lack of awareness of the disease when the parents were alive.²⁶ To test this hypothesis, we analyzed the data separately for subjects with a history of dementia in different first-degree family members (either parent or siblings). The results showed that the relative risks were higher for affected siblings vs parents in our study. Furthermore, our results were similar to those from other studies.⁶ Third, it is reported that the impact of familial or genetic factors on AD and dementia diminishes with increasing age.²⁷ Because of the old age of our study sample, the estimated relative risks in our study were lower than those of previous studies.^3,6,15 Thus, our findings may not be generalizable to the younger-old population. Finally, we were concerned with the 17% of eligible subjects who did not undergo genotyping for APOE. To assess possible bias in the results due to the relatively high dropout rate of subjects with APOE genotypes, 2 extreme situations were simulated in sensitivity analyses. The results did not show substantial difference from the original findings.

In conclusion, our results suggested that family history of dementia and APOE genotypes might still be involved in the development of dementia and AD among the very old. The effect of family history on dementia and AD was detectable only among subjects with the APOE ε4 allele. This implies that other familial (genetic or environmental) risk factors for dementia and AD might be active among APOE ε4 carriers.

Correspondence: Chengxuan Qiu, MD, PhD, Stockholm Gerontology Research Center, Olivecronas väg 4, Box 6401, S-113 82 Stockholm, Sweden (chengxuan.qiu@neurotec.ki.se).

Accepted for Publication: April 14, 2004.

Author Contributions:Study concept and design: Huang, Qiu, von Strauss, Winblad, and Fratiglioni. Acquisition of data: von Strauss, Winblad, and Fratiglioni. Analysis and interpretation of data: Huang, Qiu, and Fratiglioni. Drafting of the manuscript: Huang, Qiu, and Fratiglioni. Critical revision of the manuscript for important intellectual content: Qiu, von Strauss, Winblad, and Fratiglioni. Statistical analysis: Huang, Qiu, and Fratiglioni. Obtained funding: Huang, Qiu, Winblad, and Fratiglioni. Administrative, technical, and material support: von Strauss, Winblad, and Fratiglioni. Study supervision: Qiu and Fratiglioni.

Funding/Support: This study was supported by research grants from the Swedish Research Council, the Gun and Bertil Stohne Foundation, and the Gamla Tjänarinnor Foundation, Stockholm, and the Swedish Alzheimer Foundation, Lund.

Acknowledgment: We thank our colleagues in the Kungsholmen Project study group for their cooperation in data collection and management.