Early-Onset Alzheimer Disease in Families With Late-Onset Alzheimer Disease: A Potential Important Subtype of Familial Alzheimer Disease

Background  Genetic influences on the development of late-onset Alzheimer disease (LOAD) are heterogeneous and ill defined.

Objective  To determine the genetic risk factors for LOAD.

Design  We asked the following questions: (1) Does early-onset Alzheimer disease (EOAD) occur in families with predominantly LOAD? and (2) Does the apolipoprotein E (APOE) genotype explain the wide differences in onset age in LOAD families?

Setting  University of Washington Alzheimer Disease Research Center, Seattle.

Participants  A total of 136 kindreds and a separate group of 29 affected parent-child pairs.

Main Outcome Measures  We evaluated the kindreds with familial LOAD for the occurrence of EOAD and the affected parent-child pairs with a 20-year or more difference in the age at onset.

Results  In the 136 kindreds with LOAD, 104 had only late-onset cases (men, 36%), whereas 32 families (24%) had a combination of LOAD and EOAD cases. The 44 EOAD cases in these families accounted for 20% of cases of AD in the 32 families and 6% in all 136 families. The early-onset cases had a mean ± SD onset age of 56.1 ± 3.2 years (range, 45-59 years; men, 50%). Seven (28%) of 25 individuals with EOAD sampled did not have an APOE ε4 allele, and 2 of the earliest-onset cases were ε3/ε3. In 29 parent-child pairs with a 20-year or more difference in age at onset, 7 (35%) of the 20 children sampled did not have an APOE ε4 allele.

Conclusions  Many LOAD families (approximately 25%) have at least 1 individual with EOAD, and in these individuals, the ratio of men to women is nearly 50%, suggesting a possible subtype of familial AD. The APOE genotype plays an important role in these early-onset cases, but at least one fourth of the risk must represent the influence of other genetic and/or environmental factors. These LOAD families with early-onset cases represent an important resource for investigation of these factors.

Alzheimer disease (AD) is the most common neurodegenerative disorder that causes dementia in elderly individuals. The cause of AD is unknown in most cases. The most powerful risk factor for developing AD is age,¹ with AD affecting as many as 40% to 50% of individuals older than 85 years. Two other risk factors associated with AD are a family history of the disease and apolipoprotein E (APOE) genotype.² The frequency of the APOE ε4 allele is significantly elevated in populations with AD,³ and the presence of the APOE ε4 allele shifts the age at onset toward an earlier time point.²

Typically, AD has been divided into early-onset AD (EOAD; onset before the age of 60 years) and late-onset AD (LOAD; onset at or after the age of 60 years). Late-onset AD occurs in most patients with AD (>95%),⁴ whereas EOAD accounts for a small proportion of these patients (<5%). Demarcation between these 2 types has never been clear, and families that sometimes span these age groups remain a puzzle. The frequency of EOAD in predominantly LOAD kindreds is not known.

Specific mendelian genetic factors have been identified in patients who develop familial EOAD: namely, mutations in presenilin 1,⁵ presenilin 2,^6,7 and amyloid precursor protein.⁸ Because these specific mutations account for only a small proportion of all AD cases, clearly several additional factors have not been identified. Better characterization of this population may facilitate finding these elusive genes. We addressed this issue by reviewing 2 sets of LOAD families and looking specifically to see (1) how often EOAD occurs in predominantly LOAD families and (2) what role the APOE genotype plays in LOAD families with early-onset cases.

Kindreds were selected for this study if LOAD had been diagnosed in 2 or more living family members and, whenever possible, there was autopsy confirmation of AD in the family. Pedigrees were identified and characterized primarily by the University of Washington (Seattle) Alzheimer Disease Research Center genetics group. Kindreds were considered to have familial LOAD if the family's mean age at onset was 60 years or older. Kindreds were not included in the analysis if they contained individuals with identified mutations in amyloid precursor protein, presenilin 1, or presenilin 2 or a diagnosis of dementia other than AD. The presenilin 1 gene was sequenced in 6 individuals with the earliest onset of AD (≤55 years) from 6 different LOAD families by DNA sequencing of exons 3 to 12 in genomic DNA using standard methods with a genetic analysis system (CEQ8000; Beckman Coulter, Fullerton, Calif).⁹ Family members with non-Alzheimer neurodegenerative diagnoses, such as Parkinson disease, amyotrophic lateral sclerosis, or frontotemporal dementia, were excluded from the study.

The diagnosis of AD was based on clinical examinations of available affected individuals, as well as medical records and family history. Autopsy results were obtained in 24% of affected persons, all of which confirmed the diagnosis of AD. Clinical criteria for AD were those suggested by McKhann et al.¹⁰ Age at onset was determined to be that age at which family members and records agreed that the individual first began showing signs of memory loss or behavioral changes.¹¹ Because age-at-onset estimations are always somewhat arbitrary, age at death was also analyzed as a definitive end point.

We assessed 2 sets of pedigrees. Group 1 contained 136 kindreds with LOAD identified as part of a larger genetic linkage study. Group 2 contained 34 families selected prospectively from those being evaluated at the University of Washington Alzheimer Disease Research Center for whom an approximate 20-year or more difference in age at onset between the affected parent and affected child was noted. Five of these families were excluded because they had already been identified in group 1, leaving 29 families for analysis in group 2. Differences between means were analyzed by a 2-tailed t test. Significance was set at P<.05, and results are presented as mean ± SD.

We identified 136 families that consisted of 751 individuals with AD (men, 38%) (Table 1). These individuals were divided into kindreds with only LOAD (104 families) (group 1A) or LOAD families with EOAD cases (32 families) (group 1B). Autopsies were obtained in 177 individuals (24%), and all confirmed the diagnosis of AD.

In 104 kindreds of the 136 pedigrees, all affected members developed AD at or after the age of 60 years. These families consisted of 533 individuals with AD, and affected women significantly outnumbered affected men (men, 36%; P <.001)). Two hundred seventy-eight patients were available for a blood sample, and 110 (21%) underwent autopsy, confirming the diagnosis of AD.

The mean age at AD onset could be estimated in 385 individuals and was 73.1 ± 5.3 years (range, 60-96 years), the mean age at death was 82.2 ± 4.8 years (range, 65-104 years; n = 325), and the mean duration of disease was 9.6 ± 4.7 years (range, 1-30 years; n = 229).

Thirty-two (24%) of the 136 kindreds had at least 1 individual with onset of AD before the age of 60 years (Table 1 and Figure). These families consisted of 218 individuals with AD, with no significant difference between numbers of affected women and men (men, 44%; P=.10). One hundred twenty-three individuals were available for blood sampling. Sixty-seven individuals (31%) had autopsy results that confirmed the diagnosis of AD. The presenilin 1 gene was sequenced in 6 individuals from 6 different families with the earliest-onset cases (onset, 45-55 years). No mutations were found in the presenilin 1 gene.

The mean age at onset of AD could be estimated for 187 individuals and was 67.0 ± 2.8 years (range, 45-86 years), the mean age at death was 76.6 ± 3.4 years (range, 55-97 years; n = 162), and the mean duration of disease was 10.2 ± 2.2 years (range, 3-28 years; n = 141).

Forty-four (20%) of the 218 individuals (men, 50%) in these 32 families developed AD before the age of 60 years, giving an overall frequency of EOAD of 6% in the larger group of 136 kindreds (44/751). Twenty-five (57%) of the 44 individuals with EOAD were available for a blood sample, and 17 (39%) had autopsy confirmation of AD.

A total of 25 individuals with EOAD from 20 of the 32 group 1B LOAD and EOAD families were sampled for the APOE allele (Table 2). Eighteen individuals (72%) had 1 or more ε4 alleles, and 7 (28%) had no ε4 allele. Two of the earliest 3 cases of AD (onset ages, 45 and 48 years) were homozygous for APOE ε3 (Figure 1).

The second set of 29 pedigrees was selected because of a wide range in AD onset between the affected parent and affected child. The difference was 20 years or more in 27 of the affected parent-child pairs and 17 to 18 years in the other 2 (Table 3).

Thirty-seven parents had a diagnosis of AD. The mean age at onset of AD in these parents was 78.7 ± 5.7 years (range, 66-90 years; men, 32%), and the mean age at death was 85.1 ± 6.1 years (range, 76-100 years; n = 30). Nine parents had APOE genotyping, and 8 had confirmation of AD at autopsy. Surprisingly, in 8 families (28%) both parents carried the diagnosis of AD.

The total number of affected children and their unaffected siblings in these 29 kindreds was 103. Of these siblings, 44 (43%) were affected by AD. In addition to the 29 index cases, 15 other siblings had AD and 2 of these also had an approximately 20-year or more difference in disease onset when compared with the affected parent. Thus, 31 children from the 29 pedigrees had an approximately 20-year earlier onset than the affected parent.

The mean age at onset in the 31 affected children was 52.9 ± 5.7 years (range, 42-67 years; men, 45%), the mean age at death was 63.6 ± 10.3 years (range, 49-87 years; n = 18), and the mean difference in disease onset between the children and parents was 26.9 ± 6.1 years (range, 17-41 years). Twelve children had an autopsy result that confirmed AD.

Twenty (65%) of the 31 children and 9 (24%) of the parents were available for APOE sampling (Table 4). The APOE ε4 allele was present in 67% of the parents and 65% of the children. Seven (35%) of the sampled affected children did not have an APOE ε4 allele. Remarkably, in 8 families both parents were affected with AD. In 7 of these kindreds, the APOE status of the children was known, and 4 affected children in 3 families did not have an APOE ε4 allele.

The important findings of this study are that (1) approximately 25% of the LOAD kindreds have family members with EOAD; (2) in the LOAD families with EOAD cases, the male-female ratio is 1:1.3, whereas in the LOAD-only families it is 1:1.8 (P<.001); and (3) approximately one fourth of the EOAD cases did not have an APOE ε4 allele, including some of the earliest-onset cases. These findings were confirmed in the group 2 families with a large difference in onset age between affected parent and child.

This study identifies differences between families with only LOAD and those with a combination of LOAD and EOAD. Our findings suggest that genetic factors that contribute to the development of AD are concentrated in the latter families, as supported by other groups.^12,13 Silverman et al¹² found that the role of genetic risk factors is greatest with an earlier age at onset in the proband, whereas Rademakers et al¹³ found a novel locus for AD linked to chromosome 7q36 in a LOAD family with members who had EOAD.

An estimated 3 to 5 additional unidentified genes contribute to the development of LOAD.¹⁴ Although APOE ε4 plays an important role in lowering the age at onset in LOAD, clearly other genetic influences besides APOE contribute to the development of the disease. Chromosomal regions identified as containing candidate genes include regions on chromosomes 2,¹⁵ 9,¹⁶ 10,¹⁷ 12,¹⁸ 15,¹⁵ and 19.¹⁹ Progress in finding these elusive genes has been disappointing.

It is striking that EOAD occurs more frequently than expected in predominantly LOAD pedigrees given that EOAD makes up only a small proportion of all AD cases in the general population. Also, among the group 1B (LOAD/EOAD) pedigrees, the expected sex difference seen in LOAD was not present. The cause for this lack of sex separation is not clear but may indicate a stronger influence of genetic factors equally affecting both sexes.

We suggest that LOAD families with EOAD cases may represent a discrete subgroup in which genetic risk factors for AD have been concentrated. These families should be considered an important resource in identifying the elusive LOAD genes.

Correspondence: Thomas D. Bird, MD, 182-GRECC, VA Puget Sound Health Care System, 1660 S Columbian Way, Seattle, WA 98108 (tomnroz@u.washington.edu).

Accepted for Publication: March 29, 2006.

Author Contributions:Study concept and design: Brickell, Steinbart, and Bird. Acquisition of data: Steinbart, Rumbaugh, Payami, Schellenberg, Van Deerlin, Yuan, and Bird. Analysis and interpretation of data: Brickell, Steinbart, and Bird. Drafting of the manuscript: Brickell, Steinbart, Rumbaugh, and Bird. Critical revision of the manuscript for important intellectual content: Brickell, Rumbaugh, Payami, Schellenberg, Van Deerlin, Yuan, and Bird. Statistical analysis: Brickell and Payami. Obtained funding: Brickell, Payami, and Bird. Administrative, technical, and material support: Steinbart, Schellenberg, Van Deerlin, Yuan, and Bird. Study supervision: Bird.

Funding/Support: This study was supported by National Institute on Aging/National Institutes of Health grants P50 AG 005136-22 and P30 AG 008017, Department of Veterans Affairs research funds, the Alzheimer's Association grant IRG 96-042, and the Neurological Foundation of New Zealand.

Additional Information: Families included in our study were referred from Oregon Health Sciences University, Portland; University of British Columbia, Vancouver; University of Minnesota, Minneapolis; University of Southern California, Los Angeles; University of California–Davis; Colorado Neurological Institutes, Denver; Marshfield Clinic, Marshfield, Wis; Indiana University, Indianapolis; and Washington University, St Louis, Mo.