APOE ɛ 4 Genotype,Amyloid,and Clinical Disease Progression in Cognitively Normal Older Adults

INTRODUCTION

In cognitively normal (CN) older adults, carriage of the apolipoprotein E (APOE) ɛ4 allele (ɛ4) increases risk for progression to dementia due to Alzheimer’s disease (AD) (hereafter abbreviated as AD-dementia) and its prodromal stage mild cognitive impairment (MCI) [1–3]. Processes by which ɛ4 increases risk for clinical disease progression are still unclear. While studies in transgenic mice report ɛ4 acts directly to disrupt amyloid-β (Aβ) clearance [4, 5], ɛ4 is also associated with disruption to central nervous system (CNS) lipid processing, increased tau hyperphosphorylation, and increased neuroinflammation [6]. Thus, ɛ4 risk for clinical disease progression could be solely through increased Aβ accumulation or through both increased Aβ accumulation and disruption of non-Aβ CNS functions.

Clinical studies support both possibilities. First, frequency of the ɛ4 allele in the general population is approximately 25% but in older AD-dementia cohorts it can reach 60% [7, 8]. Second, compared to CN ɛ4 noncarriers, CN ɛ4 carriers show higher average Aβ levels overall and a higher proportion of individuals classified with abnormally high Aβ (Aβ+) [9, 10]. Prospective neuropsychological studies of CN ɛ4 carriers show both cognitive decline and clinical disease progression occur only if adults are also Aβ+ [11–13]. Conversely, one recent postmortem study showed low numbers of Aβ plaques but high tau levels in ɛ4 carriers who died with dementia [14]. Clinical studies of older CN adults report that glucose hypometabolism is related to ɛ4 carriage but not Aβ status [15] and also observe in CN ɛ4 carriers with low levels of amyloid (Aβ-) patterns of functional connectivity equivalent to that in AD-dementia [16]. Finally, a meta-analysis of studies investigating relationships between clinical disease stage and Aβ+ showed approximately 30% of ɛ4 carriers aged above 70 years and classified clinically with mild cognitive impairment (MCI) are Aβ- [17].

Inconsistency in findings about how ɛ4 increases risk for AD-dementia may reflect differences in study methodology. For example, most epidemiological studies of early AD-dementia, by definition include individuals with comorbid diseases which themselves are associated with CNS disease independent of Aβ, or in a manner exacerbated by ɛ4 carriage, for example, cardiovascular or cerebral vascular disease [1, 18–21]. Such comorbid conditions are also often uncontrolled in postmortem studies (e.g., [14]). Furthermore, prospective neuropsychological studies of the effect of ɛ4 on clinical disease progression in Aβ- CN adults generally use small samples (i.e., n < 33) [22, 23] limiting their ability to detect subtle ɛ4 effects independent of Aβ. Conversely, clinical studies reporting ɛ4 effects independent of Aβ on neuroimaging parameters have used cross-sectional designs with small (e.g., n < 39) convenience subsamples drawn from larger studies [15, 16].

The risk of ɛ4 for clinical disease progression should therefore be studied in a CN sample, relatively free of major or uncontrolled chronic illness followed carefully over years. With the risk of ɛ4 for clinical disease progression determined, knowledge of Aβ+ levels can be incorporated into the same models to allow estimation of the extent to which any ɛ4 risk for clinical disease progression is moderated by Aβ. The aim of this study was to determine indices of risk of clinical disease progression for ɛ4 carriage in a large CN sample for whom Aβ status was unknown, but who had satisfied rigorous medical and neuropsychiatric inclusion criteria. Risk estimates were then recomputed taking into account Aβ status. The hypothesis was that in CN older adults, for whom Aβ status was unknown, ɛ4 carriage would increase the risk of clinical disease progression over 72 months. We then explored the extent to which this risk was moderated by Aβ status.

MATERIALS AND METHODS

RESULTS

During the 72-month study period, there was an average follow-up of 54 months and the total person-years at risk were 3,732. For the total sample, 43 cases (6%) of incident MCI/AD-dementia were classified by 72 months with 479 cases remaining classified as CN. There were 219 withdrawals (29%) and 24 deaths (3%). For the PET imaged sample, 28 cases (7%) of incident MCI/AD-dementia were classified by 72 months with 326 cases remaining classified as CN. There were also 61 withdrawals (14%) and 8 deaths (2%).

DISCUSSION

The results of this study support the hypothesis that in CN older adults, for whom Aβ status is unknown, ɛ4 carriage increases the risk of clinical disease progression over 72 months, even with the known risks for AD from demographic and clinical characteristics taken into account. First with Aβ status unknown, CN older adults without significant or uncontrolled cardiovascular, endocrine, or inflammatory illness who were ɛ4 carriers and noncarriers were equivalent in their proportions of females, estimated premorbid IQ, level of education, and level of depressive symptoms. While ɛ4 carriers were one year younger than noncarriers, this difference was very small in terms of disease progression models (e.g., see [40]). In this total sample, age, sex, premorbid IQ, level of education, and level of depressive symptoms were each associated with clinical disease progression by 72 months.

Prospective analysis showed that in the total sample, the risk of clinical disease progression increased by 1.05 times for every additional year of age, 1.62 times for males, 1.19 times for every increase in level of depressive symptoms, and decreased by 0.76 times for each increase in education level. These demographic and clinical risk factors for disease progression are consistent with those identified in other large epidemiological studies of risk for AD-dementia [41–46]. More important for the current study, however, was the observation that even with these clinical and demographic risk factors taken into account, CN ɛ4 carriers were 2.66 times more likely than noncarriers to progress to a clinical classification of MCI/AD-dementia within the 72-month study period. While ɛ4 carriage has been found previously to increase risk for classification of MCI or AD-dementia in epidemiological studies, the finding that it also predicts disease progression in the healthy AIBL sample highlights the importance of APOE ɛ4 as a risk factor for AD even in the absence of cardiovascular disease or uncontrolled systemic illness.

To evaluate the extent to which risk for progression to MCI/AD-dementia conferred by ɛ4 was due to the risk of ɛ4 for high Aβ levels, analyses were repeated in a large CN group who had also undergone Aβ PET neuroimaging. First, estimates of group demographic and clinical characteristics in the PET imaged sample were equivalent to that in the total sample when considered according to APOE status. Furthermore, proportions of ɛ4 carriers were equivalent in the total sample and the PET imaged sample. This observation is consistent with findings from previous studies and is proposed to reflect the direct relationship between ɛ4 and amyloid accumulation [10, 47–49]. As was identified for the total sample, demographic and clinical characteristics including sex, level of education, and level of depressive symptoms also conferred risk for progression to MCI/AD-dementia in the PET imaged sample and these risks remained when ɛ4 carriage was added to the predictive model. Unlike for the total sample, with Aβ status known, age did not increase risk of clinical disease progression (i.e., in the PET imaged sample). This finding is expected given the well-established and strong association between age and Aβ accumulation in CN older adults [10, 51].

The addition of Aβ status showed that high Aβ levels increased risk for clinical disease progression such that Aβ+ individuals were 2.11 times more likely than Aβ- individuals to progress to MCI/AD-dementia within 72 months. Furthermore, with Aβ status included in predictive models, ɛ4 carriage no longer predicted risk of clinical disease progression. Thus, Aβ assumed the risk for clinical disease progression that had been conferred initially by ɛ4 carriage. These data suggest strongly that ɛ4 risk for clinical disease progression is indirect, in that ɛ4 carriage increases risk for high Aβ, which in turn increases risk for clinical disease progression [6].

The finding in the total sample that ɛ4 increased risk for clinical disease progression to MCI/AD-dementia in the context of unknown Aβ status is consistent with knowledge of ɛ4 risk for dementia from epidemiological samples [1, 44], although the mechanisms by which ɛ4 acts to increase risk of disease progression are still not understood. Recent reviews of the biology of APOE in the context of AD suggest two possible mechanisms for this process. The first mechanism proposes that ɛ4 increases risk of AD through reducing Aβ clearance. The second mechanism proposes that in addition to reducing Aβ clearance, ɛ4 also affects cerebrovascular integrity and increases inflammatory processes that themselves give rise to neurodegeneration, independent of the effects of Aβ [6]. The current results support the first mechanism that risk of ɛ4 carriage for AD is mediated through the effects of ɛ4 on Aβ accumulation. The current finding that in CN adults ɛ4 does not add further predictive value for clinical disease progression once Aβ status is taken into account is consistent with findings from previous prospective experimental biomarker studies showing that in the absence of Aβ+, ɛ4 carriage is not associated with cognitive decline [22, 23]. The current results are not consistent with data from neuroimaging studies concluding that ɛ4 influences AD processes independent of Aβ [15, 16]. It is possible, however, that findings of neuronal abnormalities related to ɛ4 but independent of Aβ reflect that ɛ4 carriage doesdisrupt brain structure and function but that these effects are too subtle to manifest clinically. This hypothesis will require prospective neuroimaging and clinical studies of a relatively rare group of CN older people.

One possible threat to the conclusion that ɛ4 carriage increases risk of AD-dementia by increasing Aβ levels is that most studies examining this issue utilize convenience samples, selected specifically to exclude individuals with substantial and intercurrent cardiovascular, endocrine, and inflammatory disease. Given that ɛ4 is also a risk factor for cardiovascular disease [6], the exclusion of individuals with cardiovascular disease or cardiovascular risk factors may result in the exclusion of individuals for whom ɛ4 carriage confers some risk for MCI/AD-dementia independent of Aβ. One way to test the contribution of non-Aβ processes would be to examine the risk for MCI/AD-dementia imposed by ɛ4 and Aβ in an epidemiological sample where cardiovascular disease or risk factors are not excluded. The Mayo Clinic Study of Aging (MCSA) [21] is currently the largest epidemiological study of risk for MCI/AD-dementia in CN adults whose Aβ and ɛ4 carrier status is known and where cardiovascular disease or risk factors are not excluded. For example, of the subjects providing baseline data for the MCSA, 79.9% had hypertension, 33.5% coronary heart disease, and 20% had a history of stroke, atrial fibrillation, or diabetes [21]. Despite this substantial cerebrovascular risk, studies of the MSCA observe that when Aβ levels are unknown, ɛ4 carriage is highly predictive of disease progression over an average of two years [12]. However, with both ɛ4 and Aβ added to predictive models, disease progression reflects Aβ status but not ɛ4 carriage. Thus like in the current study, MCSA investigators also concluded that the effect of ɛ4 on cognitive decline is mediated by the effect of Aβ. Further research in samples with varying degrees of cardiovascular or even cerebrovascular disease is now necessary to determine reliably how vascular risk factors moderate relationships between ɛ4, Aβ, and disease progression.

Some methodological limitations need to be considered when interpreting the current findings. The AIBL sample is not community or population based with recruitment requiring satisfaction of strict inclusion/exclusion criteria (i.e., reflected in the demographic and clinical characteristics in Table 1). This specificity was required to allow determination of early biological changes that occur in AD. Another limitation, most likely related to the sample, was the relative low rate of incident MCI/AD-dementia cases that occurred during the 72-month study period. Therefore, the hypothesis that with Aβ status known, APOE ɛ4 does not contribute risk for disease progression in CN older adults should be challenged in samples with higher comorbid cardiovascular risk factors and disease, or even frank cerebrovascular disease. Second, the Cox proportional hazards model used here assumed that censored cases, including withdrawal from study and death, were unrelated to clinical disease progression [34]. However, it is possible that individuals with AD symptoms, even mild ones, may have been biased toward dropping out of the study or even of dying. This highlights further the complex relationship between predisposing factors and the clinical outcome of AD. Additionally, while the increased risk for progression to MCI/AD-dementia associated with APOE ɛ4 after adjustment for Aβ was not statistically significant, the effect was increased and warrants further challenge in subsequent studies in other cohorts, or in the same cohort followed for a longer period.

In summary, the current study demonstrated that in a sample relatively free of cardiovascular disease and cardiovascular risk factors, ɛ4 carriage imposes risk for AD in CN older adults through its effect on increasing Aβ levels. That is, ɛ4 is a risk factor for Aβ accumulation, which in turn increases risk for clinical disease progression. Hence, APOE genotyping can help increase accurate identification of at-risk individuals and select those who are prime candidates for AD prevention trials (e.g., the TOMMORROW Study) [52].