Elevated Plasma β-Amyloid Peptide Aβ42 Levels, Incident Dementia, and Mortality in Down Syndrome FREE

Alzheimer disease (AD) is associated with deposition of extracellular β amyloid in neuritic plaques and vessel walls, as well as intracellular accumulation of neurofibrillary tangles. β-Amyloid peptides Aβ₄₀ and Aβ₄₂, the 2 major species of β amyloid, are generated from the amyloid precursor protein by sequential proteolytic cleavage by β and δ secretases.¹ Previous studies have demonstrated that brain levels of Aβ₄₂ increase early in the development of dementia and decrease with cognitive decline.² In cerebrospinal fluid, Aβ₄₂ levels are reduced in patients with AD. In the elderly with mild cognitive impairment, low cerebrospinal fluid Aβ₄₂ and high tau concentrations predict incipient AD.³ However, the relation of plasma concentration of Aβ₄₂ and Aβ₄₀ is less consistent. Although elevated plasma Aβ₄₂ levels in nondemented participants have been associated with increased risk for AD, mild cognitive impairment, and cognitive decline,^{2 ,4 - 7} a recent large prospective study found that high levels of Aβ₄₀ in conjunction with low levels of Aβ₄₂ predicted the development of dementia.⁸

Individuals with Down syndrome (DS) have increased levels of Aβ₄₀ and Aβ₄₂ peptides in plasma and an increased risk for AD.^{9 - 17} The increased levels of β-amyloid peptides have been attributed to triplication and overexpression of the gene for β-amyloid precursor protein, located on chromosome 21.¹⁸ Previously, we reported that plasma Aβ₄₂ levels in demented adults with DS were selectively increased compared with levels in nondemented adults with DS. We also found that Aβ₄₂ levels were elevated in demented and nondemented adults with the apolipoprotein E (APOE) ε4 allele.¹² However, all previous studies in adults with DS have used cross-sectional analyses to examine the relation of plasma β-amyloid peptides to prevalent dementia. In the present study, we expanded the sample used in our previous cross-sectional analysis and conducted a longitudinal study to examine the relation of plasma levels of Aβ₄₂ and Aβ₄₀ with the incidence of dementia and mortality risk during 5.5 years of follow-up.

A community-based sample of 207 adults with cytogenetically confirmed DS was studied. All individuals were 45 years or older, resided in New York State, and were participating in a larger longitudinal study of aging in adults with mental retardation. Participants were recruited with the help of state and voluntary service provider agencies. Subjects were eligible to participate in the study if a family member or a correspondent provided informed consent, and participants also signed a form acknowledging their willingness to participate. The participation rate was 74.6%. Recruitment, informed consent, and study procedures were approved by the institutional review boards of the New York State Institute for Basic Research in Developmental Disabilities and of Columbia University Medical Center.

Assessments included evaluations of cognition, functional and vocational abilities; behavioral/psychiatric conditions; and health status. Assessments were repeated at 14 to 18 months for 4 cycles of data collection. Cognitive function was evaluated with a test battery designed for use with individuals varying widely in their levels of intellectual functioning, as described previously.¹⁹ Participants showing declines in cognition or in adaptive behavior were evaluated by the study neurologist to confirm the presence of dementia and to determine the presence or absence of medical or psychiatric conditions other than AD that might result in or mimic dementia. Structured interviews were conducted with caregivers to collect information on changes in cognitive function, adaptive behavior, and medical history. Past and current medical records were reviewed for all participants. Vital status was obtained from follow-up interviews and from the National Death Index and was collected through October 2005.

To determine the occurrence of dementia and dementia subtypes, information from all available sources was reviewed. Diagnosis was made in a consensus conference regarding the presence or absence of dementia and its cause. Following recommendations of the AAMR-IASSID (American Association on Mental Retardation and the International Association for the Scientific Study of Intellectual Disabilities) Working Group for the Establishment of Criteria for the Diagnosis of Dementia in Individuals With Developmental Disability,²⁰ we classified participants (1) as demented if they had a history of progressive memory loss and functional decline during a period of at least 1 year, if no other medical or psychiatric conditions that might mimic dementia were present (eg, untreated hypothyroidism or stroke) and if a clinical diagnosis of AD had been made by a neurologist or a psychiatrist familiar with this population (n = 77), and (2) as nondemented if they were without cognitive or functional decline (n = 114) or if they showed some cognitive and/or functional decline but not of sufficient magnitude to meet criteria for dementia (n = 16). Demented participants were characterized by a decline of at least 20% on a memory test, evidence of decline in 1 other area of cognition, and a decline of at least 15% on an adaptive behavior scale. Participants classified as demented showed substantial and consistent decline during the course of follow-up. In a few cases, there was caregiver concern about decline in function but only small declines in cognitive function or adaptive abilities, and these participants were also classified as nondemented (n = 3). Alzheimer disease was the predominant form of dementia, accounting for 96.1% of the cases. Participants with evidence of vascular or other forms of dementia, detected during the neurological evaluations or from clinical histories, were excluded from the analysis (n = 3), leaving 204 participants for analysis. Age at meeting criteria for dementia was used to estimate age at onset of AD, recognizing that it is difficult to document the onset of initial symptoms in this population with precision.

A 10-mL venous blood sample (menadione-EDTA lavender-top tubes) was collected at baseline. Plasma levels of Aβ₄₂ and Aβ₄₀ were measured blind to dementia status using a combination of monoclonal antibody 6E10 (specific to an epitope present on 1-16 amino acid residues of Aβ) and rabbit antisera R165 (vs Aβ₄₂) and R162 (vs Aβ₄₀) in a double-antibody sandwich enzyme-linked immunosorbent assay as described previously.^{6 ,11 - 12} The detection limit for these assays was 5 pg/mL. Levels of Aβ_1-42 and Aβ_1-40 from each sample were measured twice using separate aliquots. The correlation between the first and second Aβ₄₂ and Aβ₄₀ measurements was substantial for both peptides (r = 0.86 and r = 0.96 for Aβ₄₂ and Aβ₄₀, respectively [P<.001]), and we used the mean of both measurements in the statistical analyses.

APOE genotyping was performed as described in a previous study²¹ by means of standard polymerase chain reaction–restriction fragment length polymorphism methods using HhaI (CfoI) digestion of an APOE genomic polymerase chain reaction product spanning the polymorphic (cys/arg) sites at codons 112 and 158. Acrylamide gel electrophoresis was used to assess and document the restriction fragment sizes.²² Participants were classified according to the presence or absence of an APOE ε4 allele.

In preliminary analyses, we used the χ² test to analyze categorical variables and the t test for unpaired samples, 2-sided, and analysis of variance to compare characteristics of participants by tertiles of Aβ₄₂ and Aβ₄₀ levels. We then used the t test and analysis of variance to compare levels of Aβ₄₂ and Aβ₄₀ by dementia status, vital status, and other demographic characteristics. Among those who were not demented at baseline, we used Kaplan-Meier life-table methods and Cox proportional hazards models to estimate cumulative incidence and the hazard ratio (HR) (with 95% confidence intervals [CIs]) of dementia by tertile of β-amyloid peptide levels, first in univariate models (model A) and then in models that adjusted for age, sex, level of mental retardation, and the presence of the APOE ε4 allele (model B). Among all participants, we also used Kaplan-Meier life-table methods and Cox proportional hazards models to estimate cumulative survival and the HR of death associated with tertiles of plasma Aβ₄₂ and Aβ₄₀ levels. The time-to-event variable was the time since baseline. Level of mental retardation was classified into the following 2 groups: mild/moderate (IQ, 35-70) and severe/profound (IQ, ≤34), based on IQ scores obtained before the onset of dementia.²³ Participants were also classified according to the presence or absence of an APOE ε4 allele.

At the baseline assessment, 30 of the 204 participants had prevalent AD. During the course of follow-up (mean ± SD duration, 3.9 ± 1.1 years), 44 participants (21.6%) developed AD. The mean ± SD time from baseline to onset of dementia was 2.1 ± 1.2 years. Baseline levels of plasma Aβ₄₂ and Aβ₄₀ were correlated with each other (r = 0.55 [P = .001]). In cross-sectional analyses, levels of Aβ₄₂, but not Aβ₄₀, were modestly related to age at baseline (r = 0.22 [P = .002] and r = 0.10 [P = .17], respectively). The relation of age and Aβ₄₂ level was similar among prevalent cases (r = 0.24 [P = .20]) and those who remained nondemented throughout the follow-up period (r = 0.16 [P = .07]), but was lower among incident cases (r = 0.08 [P = .59]). There was no substantial relation between age and Aβ₄₀ level in any group. Table 1 and Table 2 present demographic characteristics by tertiles of Aβ₄₂ and Aβ₄₀ peptide levels, respectively, at baseline. Participants in the highest tertile of Aβ₄₂ peptide level were significantly older than those in the 2 lower tertiles, more likely to have prevalent dementia, and more likely to have subsequently died (Table 1). Participants in the highest 2 tertiles of Aβ₄₂ peptide level were more likely to have developed AD than those in the lowest tertile (Table 1) (P < .05). Participants in the highest tertile of Aβ₄₀ peptide level were older than those in the lowest and middle tertiles, but the difference did not reach statistical significance (P = .18) (Table 2). Tertile of Aβ₄₀ peptide level was not related to dementia status or risk of death (Table 2). Levels of Aβ₄₂ and Aβ₄₀ did not differ by sex, level of mental retardation, or presence of the APOE ε4 allele (Tables 1 and 2).

Plasma Aβ₄₂ levels were highest among those with prevalent dementia at baseline, followed by those who developed AD during the course of the study, and lowest among those who remained dementia free throughout the follow-up period (25.82 ± 7.5 vs 24.10 ± 5.9 vs 22.76 ± 6.1 pg/mL, respectively [P = .04]). In contrast, plasma Aβ₄₀ levels did not differ by dementia status (153.12 ± 50.7 vs 162.50 ± 55.1 vs 171.46 ± 54.3 pg/mL, respectively [P = .18]). Plasma Aβ₄₂ levels were also higher at baseline in those who died during the course of follow-up than in those who survived (25.8 ± 7.6 vs 22.8 ± 5.7 pg/mL [P = .01]), but plasma Aβ₄₀ levels did not differ between those who died and those who survived (164.70 ± 56.6 vs 155.93 ± 51.0 pg/mL, respectively [P = .69]).

Among participants who were nondemented at baseline, those in the middle and highest tertiles of plasma Aβ₄₂ levels were more than 2.0 times as likely to develop AD as those in the lowest tertile, after adjustment for age at baseline, sex, level of mental retardation, and presence of the APOE ε4 allele (HR, 2.6 [95% CI, 1.2-5.9; P = .02] for those in the middle tertile of Aβ₄₂ level; HR, 2.0 [95% CI, 0.9-4.6; P = .10] for those in the highest tertile of Aβ₄₂ level) (Table 3 and Figure 1). Although participants in the middle tertile of Aβ₄₂ level had the highest risk of developing AD (HR, 2.6), the effect of the middle and highest tertiles was similar. Therefore we repeated the analysis combining the middle and highest tertiles. The HR for AD in the combined middle and highest tertiles compared with the lowest tertile was 2.3 (95% CI, 1.1-4.8; P = .02). In contrast, there was no relation between plasma Aβ₄₀ level and the risk of incident AD (Table 3).

Among all participants, those in the highest tertile of plasma Aβ₄₂ level were more likely to die during follow-up than those in the lowest tertile (HR, 2.4 for the highest vs the lowest tertile of Aβ₄₂ level [95% CI, 1.1-5.1; P = .02]), whereas the risk of death did not differ for those in the middle tertile of Aβ₄₂ level and those in the lowest tertile (HR, 1.2 [95% CI, 0.5-2.8]) (Table 4 and Figure 2). Because dementia is associated with an increased risk of death, we repeated this analysis including only individuals who never developed dementia and found a comparable result, although the result was not statistically significant owing to a loss of power (HR for mortality in those with the highest tertile of Aβ₄₂ level, 2.7 [95% CI, 0.5-15.6]) (data not shown).

In adults with DS, plasma levels of Aβ₄₂, but not Aβ₄₀, were related to risks of both AD and death. However, we also found elevated plasma Aβ₄₂ levels in participants with prevalent AD; whereas, in the population without DS, plasma levels of Aβ₄₂ have been found to be highest in the preclinical state, declining shortly after onset of dementia.^{4 - 5 ,24 - 26} The association between elevated β amyloid levels and prevalent dementia may be mediated, at least in part, by increasing age. Levels of Aβ₄₂ increased with age in adults with DS, and this relationship was strongest for those with prevalent dementia, who were also older, on average, than their nondemented peers. Increasing β amyloid species in plasma with age may be a peripheral reflection of the balance between β amyloid production and clearance that in the brain contributes to age-related β amyloid deposition and AD risk. However, there was no relation between age and Aβ₄₂ levels in nondemented participants who developed dementia during the follow-up period, suggesting an independent effect of elevated Aβ₄₂ levels on the risk of dementia. Among our prevalent cases, 74.8% had reported onset of dementia within 3 years of the baseline assessment. This suggests the possibility that high preclinical levels of Aβ₄₂ peptide might decline more slowly among affected adults with DS. It will require longitudinal measurement of β-amyloid peptide levels over more extended periods of follow-up to determine the pattern of change with disease progression in this population.

Among participants without dementia at baseline, those in the highest 2 tertiles of plasma Aβ₄₂ levels were more than twice as likely to develop AD during the study period than those in the lowest tertile of plasma Aβ₄₂ level. Figure 1 shows that increased incidence of AD was apparent after approximately 1.5 years of follow-up, but did not show further increases after 4.5 years of follow-up. Earlier age at onset and increased risk of AD was similar for participants in the 2 highest Aβ₄₂ tertile levels, suggesting a threshold effect. For adults with DS, these findings, together with the striking lack of effects for Aβ₄₀, support the hypothesis that individual differences in β-amyloid processing or deposition, distinct from overexpression of amyloid precursor protein, may influence the pathogenesis of AD.

Our findings are similar to those showing plasma levels of Aβ₄₂ that are higher in nondemented elderly individuals without DS who subsequently developed late-onset sporadic AD than in those who remained free of dementia.^{5 - 6} Among nondemented elderly individuals, plasma levels of Aβ₄₂ were also increased in women with mild cognitive impairment who are at high risk of progression to AD,⁴ and high baseline levels and greater reductions in plasma levels of Aβ₄₂ during follow-up have been associated with greater cognitive decline.⁷ In contrast, a recent prospective study⁸ with 10 years of follow-up found that high levels of Aβ₄₀ in conjunction with low levels of Aβ₄₂ predicted the development of dementia. Differences between study results may be related to sampling, assay methods, or timing of the sample collection in relation to the preclinical period or to the stage of disease progression, and further work with longitudinal cohorts is needed to resolve this issue.

Among all participants, those in the highest tertile of plasma Aβ₄₂ level at baseline were at more than twice the risk of dying during the study period compared with those in the lowest tertile, whereas those in the middle tertile of Aβ₄₂ level did not have increased risk of death. The increased risk of death for those in the highest tertile may be related, at least in part, to increased mortality in demented participants. Cognitive impairment has been shown to be strongly associated with mortality in nondemented and demented elderly individuals,^{27 - 30} and dementia is also associated with increased risk of death among adults with DS.^{31 - 33} Among elderly individuals without DS, baseline plasma Aβ₄₂ levels were higher in participants with recent onset of AD who died compared with those who survived or those without AD.⁵ The investigators suggested that higher levels of Aβ₄₂ may reflect a more advanced disease or a more aggressive form of AD.⁵ However, in this study, the increased risk of death appeared to be unrelated to the presence of frank dementia. An alternative hypothesis is that high Aβ₄₂ levels are related to poorer overall health status. Therefore, we examined the frequency of seizures, stroke, cancer, hypothyroidism, diabetes, hypertension, osteoporosis, heart disease (all heart disease and, separately, myocardial infarction, coronary artery disease, atrial fibrillation, and congestive heart failure [CHF]) by tertile of Aβ₄₂ level among all participants. Only the frequency of CHF was significantly higher among those in the highest tertile of Aβ₄₂ level (lowest tertile, 1.5%; middle tertile, 2.9%; and highest tertile, 10.3% [P = .04]). Also, the highest frequency of CHF was found in those with prevalent dementia (nondemented, 3.1%; incident dementia, 2.3%; and prevalent dementia, 16.7%). However, when we repeated the Cox models for mortality, including the presence or absence of CHF as a covariate, we found that this had no substantive effect on results (HR without CHF, 2.4 [95% CI, 1.1-5.1]; HR with CHF as a covariate, 2.2 [95% CI, 1,02-4.7]) (data not shown). Thus, the relation of Aβ₄₂ levels to the declines associated with dementia and death is unlikely to be due to differences among our groups in health status.

The relation of the APOE ε4 allele to plasma β-amyloid peptide levels in individuals with DS seems less clear. In a previous study,¹² we found that mean plasma levels of Aβ₄₂ were higher in adults with DS with an APOE ε4 allele than in those without and were highest when dementia and an ε4 allele were both present. Another study⁹ found that plasma levels of Aβ₄₂ and Aβ₄₀ peptides in adults with DS were independent of the presence of an APOE ε4 allele, and this was the case in the present study. Both previous studies used cross-sectional samples with younger participants. With continued follow-up, the relation of APOE to dementia and to plasma β-amyloid peptides may become attenuated, as older individuals without the APOE ε4 allele and with lower levels of plasma β-amyloid peptides develop dementia.

Because adults with DS overexpress amyloid precursor protein and have onset of dementia 10 to 20 years before elderly adults without DS, the argument can be made that our findings may have limited generalizability to the general population. However, studies of DS and AD have suggested that underlying biological mechanisms of interest are consistent in individuals with DS and other elderly populations.³⁴

It is worthwhile to consider whether an elevated plasma Aβ₄₂ peptide level may serve in clinical settings as a biological marker sensitive to the development and progression of AD. However, the substantial overlap in the distribution of Aβ₄₂ levels between demented and nondemented groups, together with changes in levels that may be associated with disease progression, would make interpretation of individual results difficult.⁵ The origin of β-amyloid in plasma is unknown, and deposition of Aβ₄₂ in brain tissue is unlikely to result directly from increased plasma levels. Although diagnostic applications are not evident, it seems clear that the brain and plasma levels of Aβ₄₂ are sensitive to alterations in β amyloid processing that contribute to AD pathogenesis,⁵ and it will be important to gain a thorough understanding of the molecular mechanisms involved.

Correspondence: Nicole Schupf, PhD, Taub Institute for Research on Alzheimer's Disease and the Aging Brain, PO Box 16, 630 W 168th St, New York, NY 10032 (ns24@columbia.edu).

Accepted for Publication: October 25, 2006.

Author Contributions: Dr Schupf had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Schupf, Zigman, and Mayeux. Acquisition of data: Schupf, Pang, Zigman, Silverman, and Mehta. Analysis and interpretation of data: Schupf, Patel, Zigman, Silverman, and Mayeux. Drafting of the manuscript: Schupf and Patel. Critical revision of the manuscript for important intellectual content: Pang, Zigman, Silverman, Mehta, and Mayeux. Statistical analysis: Schupf, Patel, Zigman, and Silverman. Obtained funding: Schupf, Zigman, and Silverman. Administrative, technical, and material support: Pang, Zigman, Silverman, Mehta, and Mayeux. Study supervision: Zigman.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grants AGO14673, HD35897, HD37425, and AGOO7232 from the National Institutes of Health, by the National Down Syndrome Society, and by funds provided by New York State through its Office of Mental Retardation and Developmental Disabilities.