Memory performance and the apolipoprotein E polymorphism in a community sample of middle-aged adults
Abstract
The apolipoprotein E genotype (APOE) is an established risk factor for Alzheimer disease, with the age-at-onset occurring earlier in individuals having at least one APOE ϵ4 allele, relative to the APOE ϵ3 or APOE ϵ2 isoforms. Moreover, nondemented older adults with the APOE ϵ4 allele also show diminished cognitive performance, particularly on tests of learning and memory, and an accelerated decline in memory performance with increasing age. The current investigation extends the study of the APOE ϵ4 allele and cognitive performance to healthy, middle-aged adults. A community sample of 220 non-Hispanic Caucasian men and women, aged 24–60 (average age = 46), were genotyped for the APOE polymorphism and completed a battery of neuropsychological tests. Multivariate analyses were conducted on measures of verbal learning and memory (e.g., learning a list of words and recalling them 30 min later), visual memory (e.g., reproducing a previously copied figure from memory), and attention span (e.g., repeating long lists of digits), after adjustments for age, and estimated IQ. Results indicated that performance on learning and memory tasks was significantly poorer in adults having any APOE ϵ4 allele, relative to adults with APOE ϵ2 and ϵ3 genotypes (P < .01). Attention span did not differ by genotype. These findings, the first in a sample of middle-aged adults, suggest that the APOE polymorphism is a marker for age-related decline in memory (detectable prior to overt, clinical manifestations of memory loss), and/or a marker for individual differences in memory ability across the life span. Am. J. Med. Genet. (Neuropsychiatr. Genet.) 96:707–711, 2000. © 2000 Wiley-Liss, Inc.
INTRODUCTION
Numerous reports have established an association of the apolipoprotein E genotype (APOE) with Alzheimer disease (AD) [for reviews see Farrer et al., 1997; Lendon et al., 1997; Roses, 1996]. The gene for apolipoprotein E is encoded on chromosome 19 and the three common isoforms, designated as ϵ2, ϵ3, and ϵ4, are distinguished by a single amino acid substitution in the receptor-binding region of apolipoprotein E [Weisgraber et al., 1982; Weisgraber et al., 1981]. Homozygosity for the APOE ϵ4 allele, which occurs in 2–3% of the Caucasian population [Mahley, 1988], is associated with a five- to eight-fold greater risk for AD, and APOE ϵ4 heterozygosity with a two- to three-fold greater risk, relative to individuals with no ϵ4 allele [Contois et al., 1996; Corder et al., 1993]. The APOE ϵ4 genotype also lowers the age at onset of AD in a dose-dependent fashion [Blacker et al., 1997; Corder et al., 1993; Meyer et al., 1998].
The APOE ϵ4 allele is also associated with lower scores on cognitive status examinations in community samples of older adults (average age ≥ 65 years) [e.g., Alstiel et al., 1997; Berr et al., 1996; Feskens et al., 1994; Henderson et al., 1995; Kuller et al., 1998; Tilvis et al., 1998], most of whom are nondemented. This is particularly evident on learning and memory tasks in community samples of elderly adults [Bondi et al., 1995; Carmelli et al., 1998; Helkala et al., 1995; Henderson et al., 1995; Reed et al., 1994; Schmidt et al., 1996; Small et al., 1999]. Further, older adults with at least one ϵ4 allele report more subjective memory complaints [Small et al., 1999] and more difficulty with activities of daily living [Albert et al., 1995] relative to adults with no ϵ4 allele. Although not always observed in cross-sectional reports [Berr et al., 1996; Caselli et al., 1999; Smith et al., 1998], the association between the APOE ϵ4 genotype and diminished performance on memory and learning tasks has been demonstrated with remarkable consistency across demented and nondemented samples ranging in age from 60 to 90 years.
In addition to these cross-sectional associations, prospective studies have shown that the presence of an APOE ϵ4 allele is associated with a decline in learning and memory performance in adults averaging over 65 years of age relative to other APOE genotypes [Bondi et al., 1995; Helkala et al., 1996; Hyman et al., 1996; Jonker et al., 1998; O'Hara et al., 1998; Small BJ et al., 1998]. Two reports did not detect an association between the APOE ϵ4 allele and a decline in memory performance [Carmelli et al., 1998; Henderson et al., 1995], but in one of these studies persons with the APOE ϵ4 allele showed worse performance on initial testing [Carmelli et al., 1998]; the other study did not report baseline data [Henderson et al., 1995]. Inconsistent findings across studies may be attributed to differences in assessment procedures, small sample sizes with inadequate statistical power, and higher mortality from all causes, including dementia, in individuals with the APOE ϵ4 allele [Tilvis et al., 1998].
Together, these results suggest that any decline in memory performance associated with the APOE ϵ4 allele may begin well before the common age of clinical impairment, or alternatively, that the APOE ϵ4 allele may be a marker of poorer memory ability across the life span. Accordingly, the present study was conducted to determine if an association between the APOE ϵ4 allele and poorer memory ability could be detected in a community sample of adults younger than 60 years of age.
MATERIALS AND METHODS
Participants
Three hundred forty-three adults between the ages of 24 and 60 were enrolled in a study of cardiovascular disease risk factors and cholesterol-lowering. Community volunteers were recruited from southwestern Pennsylvania and approximately half of the sample had LDL-cholesterol levels ≥160 mg/dl. Individuals with a history of stroke, significant head trauma, or other neurologic disorders were not enrolled. Other exclusion criteria included a diagnosis of a psychotic disorder, use of psychotropic, glucocorticoid, or hypolipidemic medication, and in women, pregnancy or lactation. All participants completed a battery of neuropsychological tasks, including measures of memory, attention, psychomotor speed, and executive function. The study protocol was approved by the University of Pittsburgh Biomedical IRB and informed consent was obtained. Near the completion of the project, participants were contacted to provide tissue for DNA extraction. DNA was obtained on 81% of the sample; the remainder could not be located or declined to provide a sample. APOE genotype, demographic, and neuropsychological data were complete for 220 Caucasian individuals. Analyses were restricted to Caucasians due to known differences in APOE allele frequencies across ethnic groups [Maestre et al., 1995; Tang et al., 1998] and small numbers of African Americans, Asians, and Hispanics in the current sample. Forty-nine percent of the sample was female (n = 107).
Genotyping
Genomic DNA was extracted from whole blood by the method of Miller et al. [1988] or from epithelial cells from buccal mucosa using the Puregene kit (Gentra Systems, Minneapolis, MN). APOE genotyping was carried out using the polymerase chain reaction based method described by Hixson and Vernier [1990]. Genotypes were assigned by direct comparison to samples of known genotype analyzed in parallel with test samples.
Neuropsychological Testing
A variety of neuropsychological tests was administered to assess memory, attention, psychomotor performance, and executive function. To test the association of the APOE polymorphism with cognitive performance, we selected for analysis seven measures reflecting learning and/or memory retrieval processes. Measures were also chosen based upon their similarity to measures used in previously published studies of elderly, nondemented individuals to document the association between the APOE polymorphism and memory performance. These measures included a verbal associative learning test (Verbal Learning Total), in which subjects study 12 unrelated word pairs across four trials (range = 0–48), and recall them 30 min later (Verbal Learning Delayed Recall) (range = 0–12); a Figure Delayed Recall test, which required subjects to reproduce from memory a complex figure (Rey or Taylor) that they had copied 30 min earlier (range = 0–36). The fourth measure was the percentage of words identified correctly as “new” or “recurring” in a continuous recognition test format (Recurring Words − %correct).
Digit Symbol Recall was derived from the Wechsler Adult Intelligence Scale-Revised (WAIS-R) [Wechsler, 1981]. The Digit Symbol Substitution Subtest requires subjects to recode numbers into symbols using a coding key that pairs numbers to shapes. Digit symbol recall refers to the number of digit symbol pairs (range = 1–9) subjects could recall immediately after completing the test. Two additional measures were derived from the WAIS-R and included digit span–longest forward span, and digit Span–longest backward span. During the digit span subtest, subjects are required to listen to a string of numbers and repeat them. Digit span–longest forward span refers to the maximum number of digits (range = 2–8) the subject was able to repeat accurately in the order presented and digit span–longest backward span refers to the maximum number repeated backwards (range = 2–8).
Data Reduction
A principal components analysis with varimax rotation on the seven memory and attention measures yielded two components, accounting for 61% of the variance. The first component included verbal learning, verbal recall, figure recall, and digit symbol recall (Learning/Memory). The second component included digit span–longest forward span, digit span–longest backward span and percentage correct on the recurring words task (Attention Span). All dependent measures were examined for normality and four measures were transformed prior to statistical analysis, including verbal learning total, verbal learning delayed recall, digit symbol recall, and recurring words − %correct.
Statistical Analysis
Demographic characteristics of subjects having any APOE ϵ4 allele were compared by t-test to subjects of all other genotypes containing no APOE ϵ4 allele. To test the hypothesis that the APOE ϵ4 genotype would be associated with poorer memory performance in middle-aged adults, performance scores on the two task clusters were subjected to two 2 × 2 (APOE ϵ4 Allele: Any vs. None × Sex: Women, Men) multivariate analyses of variance (MANOVA). Age, education, income, and estimated WAIS-R IQ (based on the sum of age-scaled Vocabulary, Information, and Block Design subtest scores) [Brooker and Cyr, 1986] were included as covariates in initial analyses. Income level and years of education were not significantly associated with the dependent measures in either model and were dropped from the final analyses.
RESULTS
Allele frequencies for the APOE ϵ2, ϵ3, and ϵ4 alleles were 0.07, 0.78, and 0.15, respectively, which are consistent with the frequencies in other North American Caucasian samples [Davignon et al., 1988]. Twenty-eight percent of the sample (n = 61) had at least one APOE ϵ4 allele, and 4 individuals were APOE ϵ4 homozygotes. As can be seen in Table I, the two groups identified for analysis were quite similar and did not differ by sex, education, smoking status, consumption of alcohol, depressive symptoms, or estimated IQ (Ps > .50). Individuals with one or more copies of the ϵ4 allele were slightly older than individuals with no ϵ4 allele (P = .02). The MANOVA for the cluster of Learning/Memory tasks revealed that performance differed significantly by genotype, F(4, 211) = 3.7, P = .006, and sex, F(4, 211) = 2.6, P = .04. Univariate analyses indicated that individuals with one or more APOE ϵ4 alleles were less able to reproduce a complex figure from memory (P = .02) and learned fewer words on the associative learning task (P = .04), relative to individuals with no ϵ4 allele. Women learned and recalled more words than men on the verbal learning task (Ps < .01). The interaction effect for Genotype × Sex was nonsignificant (P = .45). Multivariate analyses on attention span cluster did not reveal a significant effect of genotype or sex (Ps > .26). Results were identical when the analyses were repeated using test scores that had been adjusted for age, education, income, and estimated WAIS-R IQ. Nontransformed scores on all seven measures are presented in Table II.
| No ϵ4 | ϵ4 | |
|---|---|---|
| Age | 44.7 (0.69) | 47.7* (1.06) |
| Female (%) | 48 | 49 |
| Education (years) | 15.6 (.23) | 15.3 (.41) |
| Employed (%) | 79 | 82 |
| Incomea | 4.9 (.14) | 4.7 (.19) |
| Alcoholic drinks/week | 3.7 (.47) | 3.4 (.82) |
| Current smokers (%) | 14 | 20 |
| Hamilton Depression Rating Scale | 2.6 (.30) | 2.5 (.44) |
| WAIS-R Estimated IQb | 108.2 (.85) | 108.3 (1.8) |
| n | 159 | 61 |
- * P = .02.
- a Income coded on a 1–7 scale.
- b WAIS-R IQ estimated from vocabulary, information, and block design age-scaled scores.
| No ϵ4 | ϵ4 | |
|---|---|---|
| Learning/Memory (P = .02) | ||
| Verbal learning total | 27.5 (1.0) | 23.0 (1.7) |
| Verbal learning delayed recall | 8.4 (0.3) | 7.7 (0.5) |
| Digit symbol recall | 7.2 (0.2) | 7.1 (0.3) |
| Figure delayed recall | 18.9 (0.6) | 15.8 (1.0) |
| Attention span (P = .27) | ||
| Digit span-forward long span | 7.1 (0.1) | 6.9 (0.2) |
| Digit span-backward long span | 5.4 (0.1) | 5.1 (0.2) |
| Recurring words—%correct | 82.3 (1.0) | 83.7 (1.9) |
DISCUSSION
In the current study, we observed an association between the APOE ϵ4 allele and poorer performance on a cluster of learning and memory tests in a sample of Caucasian adults, 24 to 60 years of age. These data are consistent with cross-sectional data showing that learning and/or memory abilities are worse among elderly individuals with an APOE ϵ4 allele [Bondi et al., 1995; Carmelli et al., 1998; Helkala et al., 1995; Henderson et al., 1995; Reed et al., 1994; Schmidt et al., 1996; Small et al., 1999] relative to other APOE genotypes. To our knowledge, these findings are the first to document diminished memory performance in a group of individuals younger than 60 years. It is notable that the association between the APOE polymorphism and cognitive performance measures was specific to learning and memory abilities. That is, performance on measures of attention did not differ by genotype. Further, the association between diminished learning and memory performance and the APOE ϵ4 allele was detected in the absence of group differences in estimated IQ.
The presence of an APOE ϵ4 allele among nondemented elderly adults who show performance on memory tests that is outside the normal range is predictive of the progression to clinically significant disease [Petersen et al., 1995]. Further, the neuropathological changes associated with AD (i.e., development of neurofibrillary tangles and amyloid plaques) may occur earlier and/or develop more rapidly among individuals with the APOE ϵ4 genotype [Ohm et al., 1995]. Consistent with these observations, several reports have shown lesser hippocampal volume [Plassman et al., 1997; Soininen et al., 1995] and decreased blood flow in the temporal and parietal lobes of the brain [Reiman et al., 1996; Small et al., 1995] in cognitively normal middle-aged adults who have at least one APOE ϵ4 allele. Although a link between these neuropathological changes and memory decline has not been demonstrated conclusively, it is possible that memory impairment associated with AD also begins to appear well before such deficits attain clinical significance.
The results from the current study are cross-sectional, however, and we are unable to evaluate the hypothesis that the APOE ϵ4 allele is associated with a decline in cognitive performance. An alternative explanation for these results is that the APOE polymorphism is a marker of individual differences in memory ability across the life span. Quantitative genetic studies show that there is considerable heritability in memory performance (e.g., 50%) among older adults [Finkel and McGue, 1993, 1998; McClearn et al., 1997]. Further, longitudinal studies suggest that genetic effects make a larger contribution to continuity, rather than change, in individual differences in cognitive performance across the adult life span [Pederson, 1996; Plomin, 1986]. These hypotheses await further testing in larger, age-stratified normative samples as well as longitudinal investigations beginning at least in early adulthood.
