Neocortical β‐amyloid area is associated with dementia and APOE in the oldest‐old

Apolipoprotein E (APOE) ε2 carriers may be protected from dementia because of reduced levels of cortical β‐amyloid. In the oldest‐old, however, APOE ε2 carriers have high β‐amyloid plaque scores and preserved cognition. We compared different measures of β‐amyloid pathology across APOE genotypes in the oldest‐old, and their relationship with dementia.


Introduction
Alzheimer's disease (AD) affects more than 5 million people in the United States alone, at far-reaching social and financial costs. Moreover, the incidence of dementia (mostly AD) continues to increase with age and is more than 18%/year in people age 90 years and older (the oldest-old) [1]. The relationship between clinical AD and the hallmark pathology of AD, b-amyloid plaques, is well established in the younger elderly (age, 65-90 years) [2].
However, this relationship is still equivocal in the oldestold. Several studies have suggested that b-amyloid plaques are not associated with dementia [3,4], whereas others have found that this relationship continues into the oldestold [5,6]. Based on the recent attention to b-amyloid-based biomarkers, it will become increasingly important to understand this relationship in the oldest-old, the fastest growing age group in the country.
The apolipoprotein E (APOE) gene has been identified as the major genetic risk modifier in sporadic AD. APOE ε4 carriers have increased risk of clinical and pathological AD [7,8]. Conversely, APOE ε2 carriers have consistently lower risks of dementia and clinical AD [9], but the relationship between APOE ε2 and all forms of b-amyloid neuropathology is still not clear. Some studies have found that APOE ε2 carriers have decreased levels of b-amyloid neuropathology [10,11], whereas others have no decrease [12,13]. Although the precise mechanisms by which APOE can affect the pathogenesis of AD are unknown, it has been proposed that the ApoE proteins can affect a large variety of AD processes, including tau phosphorylation, neuronal apoptosis, and b-amyloid deposition [14].
We recently found that, in the oldest-old, APOE ε2 carriers have a somewhat reduced risk of dementia but increased levels of b-amyloid pathology compared with those with APOE ε3/3 [15,16]. These studies assessed b-amyloid pathology using Consortium to Establish a Registry for Alzheimer's Disease (CERAD) plaque scores, which measured b-amyloid plaques. There is significant evidence that diffuse plaques or oligomeric b-amyloid may also be critically involved in AD and cognitive dysfunction [17].
Therefore, it is the purpose of this study to extend our initial studies of the relationship between APOE and b-amyloid pathology. Adding to our previous study [16], b-amyloid pathology will be assessed using two different measures: CERAD scoring, which is a semiquantitative measure of neuritic plaques, and a b-amyloid percent area measure, which quantifies b-amyloid via immunostaining using the NAB228 antibody and measures all assembly states of b-amyloid.

Participants
Participants were drawn from The 901 Study, a population-based longitudinal study of aging and dementia in persons age 90 years and older. These subjects are survivors from the Leisure World Cohort Study, an epidemiologic investigation of a retirement community in Orange County, CA (Leisure World, Laguna Woods) initiated during the early 1980s. The cohort is primarily white, well educated, upper middle class, and mostly female. The 1151 participants from the original cohort age 90 years and older on January 1, 2003, were invited to join The 901 Study. Although most of the recruited participants still lived in the same county (60%), many had moved to other parts of California (24%) or out of state (16%). We selected all subjects who had a completed brain autopsy and completed b-amyloid percent area analysis by October 2010, which totaled 103 subjects. However, we excluded three people because they did not have a valid Mini Mental State Examination (MMSE) score, one person for not having APOE genotyping, and, to maintain mutually exclusive groups, three participants who had the APOE genotype APOE ε2/APOE ε4. Thus, 96 people were included in the study.

Determination of clinical diagnosis
As members of The 901 Study, all participants received a neurological examination and neuropsychological testing every 6 months, including the MMSE and other tests described previously [18]. Medical history information was obtained and included comorbidities such as depression, stroke, congestive heart failure, atrial fibrillation, and Parkinson's disease. In addition, most participants had available medical records and neuroimaging (computed tomography/magnetic resonance imaging) that were used for the clinical diagnosis. Clinical diagnoses were determined by a consensus diagnostic conference, using all available information and blinded to APOE genotype. Dementia diagnosis was established using Diagnostic and Statistical Manual of Mental Disorders, 4th edition, criteria. Clinical AD diagnoses were established using National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association criteria [19] for possible or probable AD. The institutional review board of the University of California, Irvine, approved all procedures.

Determination of pathological diagnosis
Before dissection, the whole brain was weighed. One hemisphere of each individual was selected for use in the final neuropathology diagnosis based on the neurological examiner's impression of any asymmetry in clinical neurological features. The hemisphere corresponding to any such clinical features was selected for examination. Thioflavin S staining was performed on midfrontal and superior temporal cortex sections, and CERAD criteria were used to determine neuritic plaque scores [20]. For analysis, semiquantitative plaque scores were given numeric values, so that none 5 0, sparse 5 1, moderate 5 2, and frequent 5 3.
For the b-amyloid percent area measure, midfrontal and superior temporal cortex sections were stained using antib-amyloid (NAB228, an antibody binding to Ab 1-11 ). A representative section of NAB228-stained tissue was selected (holes in the tissue were not included in stained areas), the multithresholder maximum entropy algorithm was applied, and b-amyloid percent area was determined. Detailed methods have been described previously [21,22]. In all analyses, the mean value of the two cortical areas was used.

Statistical analysis
The levels of neuropathology between demented and nondemented subjects within each APOE genotype were analyzed using Student t tests. Because of the small number of homozygous individuals for APOE ε2 and APOE ε4, these individuals were grouped with the heterozygous participants for analysis. We then examined the relationship between age at death, post mortem interval, gender, and education to APOE genotype to determine whether these variables should be included as covariates in subsequent analyses. Although none of the variables were associated with the outcome variables, age at death and gender were retained in the model as covariates to maintain statistical continuity with other APOE studies. Logistic regression analyses were performed to examine the relationship of APOE genotype to several clinical and neuropathological outcomes. The clinical outcomes analyzed were presence or absence of dementia, presence or absence of possible or probable AD, and MMSE score less than 24 (vs 24). The neuropathological outcomes analyzed were CERAD plaque score (frequent vs none/sparse/moderate) and b-amyloid percent area (above the median vs below the median). Because they are continuous variables, we also used linear regression to examine the effects of APOE on MMSE scores and b-amyloid percent area. We used the data from the visit closest to the time of death for analyses of MMSE score. All statistical analyses were performed using SAS (version 9.2, SAS Institutes Inc., Cary, NC) and SPSS (version 19, IBM, Armonk, NY).

Results
The demographic information of the participants is summarized in Table 1. The participants were predominantly female and were highly educated. Participants' median age at death was 96.8 years. APOE ε2 was present in 13.5% of participants (one person was APOE ε2 homozygous) and APOE ε4 in 18.8% (one person was APOE ε4 homozygous). Dementia was present in 59.4% of the participants, whereas clinical AD was present in 44.8%. The median percent amyloid area was 3.2% in APOE ε2 carriers and 10.6% in APOE ε4 carriers.
The relationship between APOE, dementia status, and b-amyloid neuropathology is summarized in Fig. 1. CERAD plaque scores did not differentiate demented vs nonde-mented participants with APOE ε2 and APOE ε3/3, but b-amyloid percent area did differentiate demented vs nondemented participants in all three APOE genotypes.
The dissociations between CERAD plaque score and b-amyloid percent area for participants with an APOE ε2 gene are shown in Table 2. All but one of the participants with a b-amyloid percent area value below the median had moderate or frequent CERAD plaque scores. One participant had a low CERAD plaque score, but a b-amyloid percent area score slightly above the median (6.6%).
The associations between APOE genotype and clinical and neuropathological outcomes are summarized in Table 3. Neither APOE ε2 nor APOE ε4 was associated significantly with any clinical outcomes (dementia, clinical AD, or MMSE score , 24). However, all the odds ratios (ORs) for APOE ε2 were less than one, suggesting lower odds for all the outcomes; all the ORs for APOE ε4 were more than one, suggesting higher odds for all clinical outcomes. For analyses of neuropathological outcomes, APOE ε4 carriers were significantly more likely to have frequent b-amyloid plaques (OR, 5.11) and high b-amyloid percent area (OR, 11.50) than APOE ε3/3 carriers. In contrast, APOE ε2 carriers were significantly more likely to have frequent b-amyloid plaques (OR, 4.51), and were less likely, although not significantly, to have high b-amyloid percent area (OR, 0.42) than APOE ε3/3 carriers. When MMSE score was analyzed as a continuous variable, APOE ε2 carriers did slightly better on the MMSE (B 5 2.20, P 5 .48), whereas APOE ε4 carriers did slightly worse (B 5 21.72, P 5 .52) than APOE ε3/3 carriers, but the results were not significant. When amyloid area was analyzed as a continuous variable, APOE ε2 carriers had slightly (but not significantly) lower b-amyloid percent areas (B 5 20.48, P 5.79), whereas APOE ε4 carriers had significantly higher b-amyloid percent

Discussion
In this study of the oldest-old, we found that b-amyloid percent area but not CERAD plaque score was associated with dementia in all three APOE genotypes. These findings may help explain the surprising dissociation between AD neuropathology and dementia that we previously found in APOE ε2 carriers in this cohort [16]. In that study, APOE ε2 carriers demonstrated significantly higher CERAD plaque scores than APOE ε3/3 carriers, despite having decreased odds of dementia.
In the current study, both APOE ε2 and APOE ε4 carriers had very high CERAD plaque scores. Previous studies have confirmed that APOE ε4 is associated with increased AD neuropathology, especially b-amyloid plaques [7]. The mechanism behind this increase in b-amyloid pathology is still being determined [14]. In contrast, there is little evidence in the literature that APOE ε2 is associated with increased b-amyloid plaque scores. Previously, in this oldest-old cohort, we described that APOE ε2 carriers have increased plaque scores, as measured by CERAD [16]. However, other researchers examining younger cohorts have found APOE ε2 to be associated with decreased b-amyloid pathologies, including b-amyloid plaque density, neuropil thread formation, and amyloid angiopathy [11]. It is possible, therefore, that the extreme age of the subjects in this cohort may be involved in the different neuropathological levels in APOE ε2 carriers.
In contrast to the CERAD plaque measure, the b-amyloid percent area measure was strongly associated with dementia and was not elevated in APOE ε2 carriers compared with APOE ε3/3 carriers. The APOE ε2 carriers, despite having high CERAD plaque scores, had relatively low levels of amyloid as measured by b-amyloid percent area. These findings  suggest that CERAD plaque scores may not be the best way to measure the relevant b-amyloid pathology in oldest-old APOE ε2 carriers. An APOE ε2-specific mechanism may exist that increases the numbers of CERAD plaques, while actually reducing b-amyloid in other forms and therefore overall b-amyloid levels. In addition, APOE ε2 carriers may have an accelerated formation of b-amyloid into plaques to protect against the more destructive oligomeric form of b-amyloid. The b-amyloid percent area measures many different assembly states of b-amyloid, including diffuse plaques and b-amyloid oligomers. It may be a better marker for overall b-amyloid pathology, or may be measuring other, more relevant forms of b-amyloid, especially in the oldest-old. It is likely that a measure of all types of b-amyloid (such as the NAB228 percent area) relates better than staging to in vivo amyloid measures, because positron emission tomographic scanning may visualize only the more fibrillar b-amyloid deposits, but further research is needed to elucidate this relationship.
There are several limitations to this study that warrant discussion. This analysis focused on the relationship between APOE genotype, dementia, and hallmark AD neuropathology, but did not address additional pathologies such as vascular lesions, hippocampal sclerosis, or Lewy bodies. These other pathologies are perhaps involved in the association between APOE ε2 genotype and dementia, but this analysis focused specifically on the methodology of measuring b-amyloid neuropathology. Because AD neuropathology alone becomes increasingly rare with advanced age [23], subjects with mixed pathology may be more relevant to clinical situations. Nevertheless, future analyses should examine to what extent these other pathologies are related to APOE and dementia. It is also possible that APOE ε2 may offer protection against dementia because of a mechanism that is completely independent of neuropathology, such as affecting brain reserve or neuroplasticity, which the current findings do not address. Another limitation to the study was the inclusion of non-AD cases in the analysis. In the oldest-old, many of the non-AD cases have high levels of AD pathology, and we felt that their inclusion was warranted because their clinical diagnosis may have been incorrect or they could have had multiple dementia etiologies including AD. In addition, the current cohort comprises a predominantly female, white population of high educational level, characteristics that could potentially limit the generalizability of our results. However, according to the U.S. Census, most people age 90 years and older are women, white, and at least high school graduates [24]. Thus, despite the lack of representation of minority subjects and having mostly women in the cohort, the composition of the current cohort does reflect that of people age 90 years and older in the United States.
The current findings suggest that the relationship among APOE, AD neuropathology, and cognition in the oldest-old is a complex one. Although APOE ε4 carriers are more likely to have dementia and b-amyloid neuropathology, APOE ε2 carriers show a dissociation between cognition and b-amyloid neuropathology when measured using CERAD plaque scores. However, if b-amyloid pathology is measured using b-amyloid percent area, the relationship between b-amyloid and cognition is restored, and APOE ε2 carriers have better cognition and less b-amyloid percent area. Thus, it appears that b-amyloid continues to be important for cognition after age 90; however, CERAD plaque scores may no longer be the best measure of this neuropathology.

Author disclosures
Dr Lee has received funding for travel and honoraria from Takeda Pharmaceutical Company Ltd.; has received speaker honoraria from Pfizer, Inc, BMS, and Merck; may accrue revenue on patents re: Modified avidin-biotin technique; Method of stabilizing microtubules to treat Alzheimer's disease; Method of detecting abnormally phosphorylated tau; Method of screening for Alzheimer's disease or disease associated with the accumulation of paired helical filaments; Compositions and methods for producing and using

Acknowledgments
This work was funded by the National Institute on Aging grants R01AG21055, P50AG016573, and R01AG10124. We thank the University of California at Irvine tissue repository and, in particular, Ron Kim for the brain autopsies, as well as The 901 Study participants and families for their participation in the study.

RESEARCH IN CONTEXT
1. Systematic review: We searched PubMed for reports with the search items APOE Alzheimer's disease, APOE Alzheimer neuropathology, and APOE amyloid. We targeted studies that examined the relationships among apolipoprotein E (APOE), dementia, and Alzheimer's disease (AD) neuropathology.
2. Interpretation: These findings help explain the dissociation between AD neuropathology and dementia previously found in APOE ε2 carriers. If b-amyloid pathology is measured using b-amyloid percent area, the relationship between b-amyloid and cognition is restored. Thus, it appears that b-amyloid continues to be important for cognition after age 90; however, CERAD plaque scores may no longer be the best neuropathological measure.
3. Future directions: The current study demonstrated that b-amyloid percent area correlated with cognition better than CERAD plaques. Because b-amyloid percent area measures all assembly states of b-amyloid, it is unknown which states cause cognitive deficits seen in AD. Future research should use more specific measures of neuropathology to help identity these specific assembly states and to determine how they affect cognition.