Age and Apolipoprotein E*4 Allele Effects on Cerebrospinal Fluid β-Amyloid 42 in Adults With Normal Cognition FREE

Lower concentration of a long form of the β-amyloid protein ending at amino acid 42 (Aβ₄₂) in cerebrospinal fluid (CSF) is a consistent biomarker of Alzheimer disease (AD).^{1 - 2} In contrast, CSF Aβ₄₀ concentration is unaffected by AD. Aging and presence of theapolipoprotein E*4 (APOE*4) allele are the 2 strongest risk factors for AD. The presence of the APOE*4 allele interacts with aging to lower by 10 to 15 years the age of clinical dementia onset.³ Neuropathological studies have demonstrated that years before the onset of clinical dementia, plaques of insoluble aggregated Aβ protein, the neuropathological hallmark of AD, begin to accumulate in brain tissue.⁴ The Aβ peptide is generated by proteolytic processing of the amyloid precursor protein into a series of fragments 38 to 42 amino acids long; Aβ₄₂ represents about 10% of synthesized Aβ but is by far the earliest and predominant Aβ species deposited in plaques in AD.⁵ Decreased concentration of Aβ₄₂ in CSF is a consistent biomarker of AD^{1 - 2} and likely reflects deposition of Aβ₄₂ in plaques.⁶ This Aβ deposition is considered central to the pathogenesis of AD. In transgenic mice that overexpress mutant amyloid precursor protein, Aβ₄₂ deposition in the brain increases with age and parallels a decrease in CSF Aβ₄₂ concentration.⁷ Animal and human neuropathology studies suggest that the presence of the APOE*4 allele hastens AD onset by modulating the deposition and clearance of Aβ to favor the formation of plaques.⁸

Herein, we estimated in adults with normal cognition the effects of age and APOE genotype on biomarkers related to the deposition in brain of Aβ. Specifically, we determined CSF Aβ₄₂ concentration and APOE genotype in 184 healthy adults without dementia across a broad age range. We also measured CSF Aβ₄₀ concentration to determine whether effects on CSF Aβ₄₂ levels are selective. We hypothesized that CSF Aβ₄₂ concentration would begin to decline years before the age at which clinical AD commonly presents; that this decline would be substantially accentuated by the presence of the APOE*4 allele; and that neither age nor presence of the APOE*4 allele would affect CSF Aβ₄₀ concentrations.

All procedures were approved by the institutional review boards of the participating institutions; all subjects provided written informed consent. Ninety-four men and 90 women (age range, 21-88 years [mean ± SD age, 50 ± 20 years]) underwent detailed clinical and laboratory evaluation and had no clinically significant abnormalities. Subjects had Mini-Mental State Examination⁹ scores between 26 and 30 (mean ± SD, 28.8 ± 1.5), Clinical Dementia Rating Scale¹⁰ scores of 0, and no evidence or history of cognitive or functional decline. For subjects older than 50 years, scores on delayed recall were higher than a cutoff of 1.5 SD lower than age-adjusted means (for both the Logical Memory¹¹ and New York University paragraph tests¹² ). We collected CSF in the morning after an overnight fast using Sprotte 24-g atraumatic spinal needles. Samples with more than 500 red blood cells per milliliter were excluded. Samples were frozen immediately on dry ice and stored at −80°C until assay. The Aβ₄₂ and Aβ₄₀ concentrations were measured in the 10th milliliter of CSF collected using sensitive, well-validated sandwich enzyme-linked immunosorbent assays.¹³ The APOE genotypes were performed by a restriction digest method.^{14 - 15}

We examined the relationship between Aβ₄₂ (or Aβ₄₀) concentration and age and APOE genotype using a linear regression model of Aβ₄₂ (or Aβ₄₀) concentration on age and APOE*4 status. Age was modeled as an orthogonal quadratic polynomial to separate out linear from quadratic trends in the Aβ peptide–age relationship. Interaction with APOE*4 status was also modeled to determine if trends differed by presence or absence of the APOE*4 allele. To assess selectivity of changes in Aβ₄₂ concentration, we also examined if the ratio of Aβ₄₂-Aβ₄₀ concentration was influenced by age and the APOE*4 allele. The ratio was modeled directly as a dependent variable on age and APOE*4 status and, indirectly, using a regression of Aβ₄₂ concentration on age and APOE*4 status with Aβ₄₀ concentration as an additional covariate. Because Aβ₄₀ and Aβ₄₂ are both produced by cleavage from amyloid precursor protein by β and γ secretase, there should be a fixed ratio of secretion of these forms of Aβ into CSF. We hypothesized that different disposition of these molecules would be demonstrated as changes in the ratio of Aβ₄₂-Aβ₄₀ concentration.

Lower levels of Aβ₄₂ were associated with older age and with presence of the APOE *4 allele. There was a significant difference in the relationship between age and Aβ₄₂ concentration by APOE*4 allele status (interaction between age and APOE*4 status, P = .01). The relationship for APOE*4-positive subjects was strongly linear with Aβ₄₂ concentration decreasing as age increased. This linear relationship was absent for APOE*4-negative subjects. The quadratic component of the age term indicated that there was a change in the trend in the relationship between age and Aβ₄₂ concentration for both groups but the shape of the curves differed markedly (Figure, A). For APOE*4-negative subjects, mean Aβ₄₂ concentration rose slightly until approximately age 50 years, then fell slightly with increasing age. In contrast, for APOE*4-positive subjects, Aβ₄₂ concentration declined slightly in younger subjects, then declined rapidly beginning between ages 50 and 60 years. Graphical assessment confirms that changes in trend occurred between ages 50 to 60 years for both APOE*4-positive and APOE*4-negative subjects, but the slope of subsequent CSF Aβ₄₂ concentration decline was markedly greater in the APOE*4-positive subjects.

There was also a significant interaction between age and APOE*4 in the regression model of Aβ₄₀ concentration (P = .02). In contrast to Aβ₄₂ concentration, Aβ₄₀ concentration did not change with age in APOE*4-positive subjects (P = .68) and increased linearly with age in APOE*4-negative subjects (P<.001), with neither relationship having a significant quadratic component (Figure, B). Regression of the ratio of Aβ₄₂-Aβ₄₀ concentration showed a significant trend with age (P<.001); the ratio values were stable until about age 50 years, then the ratio decreased sharply as age increased (data not shown). This decrease was stronger for APOE*4-positive subjects, but this difference was not quite statistically significant (interaction between age and APOE*4 status, P = .13). These results were confirmed by a regression of Aβ₄₂ concentration on age and APOE*4 status, adjusting for Aβ₄₀ concentration.

We also examined whether CSF Aβ₄₂ levels were related to scores on delayed recall tests and on the Mini-Mental State Examination, using partial correlation controlling for age. None of these test scores showed a significant association at P<.05.

A previous study that measured CSF Aβ₄₂ concentration found a nonlinear relationship with age,¹⁶ while another showed no relationship with age.¹⁷ However, in these previous studies, APOE genotypes were not determined and characterization to assure that subjects were cognitively normal was minimal. Studies in which APOE genotyping was performed have demonstrated decreased CSF Aβ₄₂ concentration in APOE*4-positive vs APOE*4-negative subjects. No age × APOE*4 interactions were seen, but the age ranges of subjects in these studies were limited (mean ± SD age, 59 ± 8 years¹⁸ and 58 ± 7 years¹⁹ ) and young subjects were not included. To our knowledge, the present study is the only study that examined effects of the APOE*4 allele across a broad age range. It is also the only study, to our knowledge, that measured both Aβ₄₀ as well as Aβ₄₂ concentrations to assess the specificity of these effects on the more pathogenic 42 amino acid–length Aβ species. Our results suggest that the APOE*4 allele is associated with selective alteration of the fate of Aβ₄₂, but not Aβ₄₀, that results in decreasing concentration of Aβ₄₂ in CSF during the normal adult life span. In persons with the APOE*4 allele, decline in CSF Aβ₄₂ concentration possibly begins in young adulthood, followed by marked acceleration of this decline beginning in midlife—decades before clinical manifestations of AD. β-amyloid 40, although also found in amyloid plaques, did not show the same association of decreased CSF concentration with age and presence of the APOE*4 allele. Because Aβ₄₂ and Aβ₄₀ are produced by the same secretases, the concentrations of these forms of Aβ are initially in equilibrium. Our results are therefore consistent with differential aggregation and deposition in the brain, clearance, or binding to carrier molecules that selectively affects Aβ₄₂. These results do suggest that sequestration of Aβ₄₂ in the brain occurs early in APOE*4 allele carriers, bolstering evidence for this as a key initiating factor in AD pathogenesis. Measuring Aβ concentration in CSF provides an indirect estimation of the net effect of production, clearance, aggregation, and deposition; therefore, we cannot determine which of these factors is most important. Our results are consistent with the findings of a recent study of asymptomatic adult carriers of presenilin mutations who had low levels of CSF Aβ₄₂, lending further support to a decrease in CSF Aβ₄₂ concentration as a preclinical biomarker for AD.²⁰

These findings have implications for the preclinical diagnosis of AD, as well as for treatment. Follow-up of cohorts such as ours will be important to affirm our cross-sectional findings and to assess whether subjects who fall in the lowest part of the age- and APOE-adjusted range for CSF Aβ₄₂ concentration have the highest risk of developing AD. It will also be important to correlate CSF Aβ₄₂ concentration with methods of assessing brain amyloid deposition in vivo, such as positron emission tomographic scanning with the Pittsburgh Compound-B.²¹ Therapeutic strategies aimed at prevention of AD may need to be applied in early midlife or even younger ages to have maximal effect on amyloid deposition. Primary prevention trials for AD targeting elderly persons may be too late to affect the early stages of disease pathology.

Correspondence: Elaine R. Peskind, MD, VA Puget Sound Health Care System, S-116MIRECC, 1660 S Columbian Way, Seattle, WA 98108 (peskind@u.washington.edu).

Accepted for Publication: December 23, 2005.

Author Contributions:Study concept and design: Peskind, Farlow, Raskind, and Galasko. Acquisition of data: Peskind, Quinn, Kaye, Clark, Farlow, DeCarli, Raskind, Schellenberg, Lee, and Galasko. Analysis and interpretation of data: Peskind, Li, Shofer, Kaye, Farlow, Raskind, Lee, and Galasko. Drafting of the manuscript: Peskind, Li, Shofer, Farlow, Raskind, Lee, and Galasko. Critical revision of the manuscript for important intellectual content: Peskind, Li, Shofer, Quinn, Kaye, Clark, Farlow, DeCarli, Raskind, Schellenberg, and Galasko. Statistical analysis: Li and Shofer. Obtained funding: Peskind, Raskind, and Galasko. Administrative, technical, and material support: Peskind, Kaye, Farlow, Raskind, Schellenberg, Lee, and Galasko. Study supervision: Peskind, Kaye, Farlow, and Lee.

Funding/Support: This study was supported by grants AG05136, AG08419, AG08017, AG10124, AG10133, AG23185, and M01 RR00034 from the US National Institute on Aging; the National Alzheimer's Coordinating Center; Friends of Alzheimer's Research; Alzheimer's Association of Western and Central Washington; and the Department of Veterans Affairs.

Role of the Sponsor: None of the funding sources had a role in study design; collection, analysis, and interpretation of data; writing of the report; or the decision to submit the paper for publication.