Impact of Apolipoprotein E ε4 in Alzheimer’s Disease: A Meta-Analysis of Voxel-Based Morphometry Studies

Abstract

Background and Purpose

Alzheimer’s disease (AD) is the most-prevalent form of dementia and imposes substantial burdens at the personal and societal levels. The apolipoprotein E (APOE) ε4 allele is a genetic factor known to increase AD risk and exacerbate brain atrophy and its symptoms. We aimed to provide a comprehensive review of the impacts of APOE ε4 on brain atrophy in AD as well as in mild cognitive impairment (MCI) as a transitional stage of AD.

Methods

We performed a coordinate-based meta-analysis of voxel-based morphometry studies to compare gray-matter atrophy patterns between carriers and noncarriers of APOE ε4. We obtained coordinate-based structural magnetic resonance imaging data from 1,135 individuals who met our inclusion criteria among 12 studies reported in PubMed and Google Scholar.

Results

We found that atrophy of the hippocampus and parahippocampus was significantly greater in APOE ε4 carriers than in noncarriers, especially among those with AD and MCI, while there was no significant atrophy in these regions in healthy controls who were also carriers.

Conclusions

The present meta-analysis has highlighted the significant link between the APOE ε4 allele and hippocampal atrophy in both AD and MCI, which emphasizes the critical influence of the allele on neurodegeneration, especially in the hippocampus. These findings improve the understanding of AD pathology, potentially facilitating progress in early detection, targeted interventions, and personalized care strategies for individuals at risk of AD who carry the APOE ε4 allele.

Graphical Abstract

INTRODUCTION

Alzheimer’s disease (AD) is an increasing challenge to worldwide healthcare that imposes significant burdens at the individual, family, and societal levels.¹ There are currently approximately 50 million patients with AD worldwide, with this prevalence being projected to double every 5 years and to reach 152 million by 2050.² AD is the most-prevalent form of dementia and progresses slowly, but it inevitably induces substantial neuronal loss, especially in the medial temporal lobe and hippocampus,³ which manifests as memory decline, cognitive impairment, and changes in personality and language abilities.⁴,⁵ Although the underlying causes are complex and involve genetic, environmental, and lifestyle factors,⁶ the ε4 allele of the apolipoprotein E (APOE) gene has been identified as the strongest risk factor for AD.⁷ Approximately 60% of patients with AD carry an ε4 allele.⁷,⁸ Carrying one allele increases the AD risk two- to threefold, while those carrying two alleles face a 15-fold increased AD risk.⁹ The presence of the APOE ε4 allele can also influence the age at onset, progression rate, and severity of AD.¹⁰,¹¹,¹² The current study aimed to elucidate the impacts of the APOE ε4 allele on brain structure in AD through a comprehensive review of previous neuroimaging studies.

AD is characterized by amyloid plaques¹³ and neurofibrillary tangles¹⁴ that result in synaptic and neuronal loss, and brain atrophy.³ The progressive atrophy of the hippocampus and parahippocampal gyrus particularly disrupts this memory network in AD.¹⁵ The accumulation of neurofibrillary tangles and amyloid plaques are the primary pathological features of AD that contribute to decreased neuronal activity and synaptic functionality.¹⁶,¹⁷ Individuals carrying the APOE ε4 allele notably present with increased amyloid-β42 and tau protein concentrations, which are correlated with accelerated cognitive decline.¹⁸

The APOE gene located on chromosome 19¹⁹ is crucial to APOE synthesis, which plays a vital role in the transport of cholesterol and other lipids across cells in various body tissues.²⁰ The APOE protein predominantly produced by astrocytes in the brain is critical for neuron maintenance and repair.²¹ This gene exists in three main isoforms (ε2, ε3, and ε4) that each have a unique impact on lipid metabolism and neuronal health.²² The ε4 variant is strongly associated with AD risk, playing a role in its progression by promoting amyloid β buildup and impairing the physiology of the medial temporal lobe.²³ The association between the ε4 allele and elevated brain amyloid is evident in both the early,²⁴,²⁵ and advanced stages of AD.²⁶,²⁷ This genetic variant is also associated with a reduced hippocampal volume²⁸ and more-significant memory deficits in patients with AD.²⁹

Advances in neuroimaging, and particularly in magnetic resonance imaging, have been indispensable in identifying and quantifying brain atrophy in AD.³⁰ Voxel-based morphometry (VBM) is a neuroimaging analysis technique that can identify focal differences in brain volume using statistical voxel-wise comparisons.³¹ Activation likelihood estimation (ALE) is a widely used algorithm for coordinate-based meta-analyses in neuroimaging.³² This involves assessing the convergence of activation points reported across different investigations by modeling these points as probability distributions. ALE treats reported activation foci as spatial probability distributions centered at specific coordinates. ALE maps are created by computing the union of activation probabilities for each voxel. The algorithm applies a permutation procedure to distinguish true convergence of foci from random clustering. Previous ALE meta-analyses of AD using VBM have focused on comparing AD patients with healthy controls (HCs), and have revealed atrophy of the hippocampus and parahippocampal gyrus.³³,³⁴,³⁵ However, despite numerous previous studies on AD, there is a notable gap in research comparing brain atrophy between APOE ε4 carriers and noncarriers.

The present study aimed to fill this gap using ALE meta-analysis to synthesize data across studies, to provide a comprehensive understanding of the impact of the APOE ε4 allele on brain structure during various stages of cognitive decline. We compared brain atrophy between APOE ε4 carriers and noncarriers across AD, mild cognitive impairment (MCI), and HC groups. We applied ALE to data obtained in multiple studies to identify significant patterns of brain atrophy and thereby provide new insights into the relationship between the APOE ε4 allele and AD pathology.

METHODS

Literature search, selection criteria, and quality appraisal

We conducted a comprehensive literature search of PubMed and Google Scholar from October 2023 to February 2024 that aimed to find structural neuroimaging studies reported on in English and published at any time that compared APOE ε4 carriers with noncarriers. Our search strategy involved the following combinations of keywords: 1) (Alzheimer) AND (voxel-based morphometry or VBM) AND (apolipoprotein or APOE), and 2) (mild cognitive impairment or MCI) AND (voxel-based morphometry or VBM) AND (apolipoprotein or APOE). This search strategy yielded 83 studies, of which 12 were deemed to be relevant after applying the following inclusion criteria: 1) using VBM as the imaging modality; 2) using brain coordinates in a standardized stochastic space, either the Montreal Neurological Institute (MNI) or Talairach template; 3) comparing groups between APOE ε4 carriers and noncarriers; 4) investigating atrophy of gray matter; and 5) performing a whole-brain analysis. The initial screening process yielded 33 studies that were eligible for assessment, of which 21 were excluded for the following reasons: 1) comparative studies (n=6), 2) reviews (n=1), 3) case reports (n=1), 4) no genetic information (n=4), 5) no contrast information (n=2), 6) white-matter investigations (n=2), 7) no whole-brain analysis (n=2), or 8) nonhumans (n=3).

Fig. 1 illustrates the inclusion and exclusion criteria applied during the study selection process using a flow diagram based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 for new systematic reviews.³⁶ M.B., Z.I., and J.J. conducted all of the selection steps. The studies that met the criteria after the screening process were included in the meta-analysis.

Fig. 1. PRISMA flow diagram of the study selection process. AD, Alzheimer’s disease; HC, healthy control; MCI, mild cognitive impairment; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

ALE analysis

The impact of APOE ε4 status on brain atrophy was investigated comprehensively using a meta-analysis with the ALE algorithm in BrainMap GingerALE (version 3.0.2; https://brainmap.org).³² The experiments were categorized into AD, MCI, and HC groups (Table 1). Separate ALE meta-analyses were performed on four subexperiments: 1) with all patient experiments combined (AD+MCI), and 2) in each of the AD, MCI, and HC groups. Some studies included multiple experiments, and so the total number of experiments (n=25) exceeded the total number of studies (n=12). All Talairach coordinates were transformed into MNI space using the transform called icbm2tal developed by Laird et al.³⁷ The probability criterion was set as an uncorrected p value of <0.001 with 1,000 permutations. We reported significant clusters with a corrected p value of <0.05 for the false discovery rate (FDR) with the minimum cluster volume set to 200 mm³.

Table 1. The 12 studies included in the meta-analysis.

Study	Condition	Participants
Chang et al.,58 2019	AD	APOE ε4 carriers=38, noncarriers=59
Forno et al.,59 2023	AD	APOE ε4 carriers=51, noncarriers=36
Brys et al.,57 2009	MCI	APOE ε4 carriers=13, noncarriers=11
Novellino et al.,66 2019	MCI	APOE ε4 carriers=40, noncarriers=55
You et al.,67 2021	MCI	APOE ε4 carriers=22, noncarriers=38
Alexander et al.,56 2012	HC	APOE ε4 carriers=14, noncarriers=10
Goltermann et al.,60 2019	HC	APOE ε4 carriers=22, noncarriers=38
Honea et al.,62 2009	HC	APOE ε4 carriers=14, noncarriers=39
Honea et al.,63 2009	HC	APOE ε4 carriers=34, noncarriers=29
Honea et al.,61 2010	HC	APOE ε4 carriers=14, noncarriers=39
Nao et al.,65 2017	HC	APOE ε4 carriers=22, noncarriers=38
Mosconi et al.,64 2014	HC	APOE ε4 carriers=19, noncarriers=33

AD, Alzheimer’s disease; APOE, apolipoprotein E; HC, healthy control; MCI, mild cognitive impairment.

NeuroSynth

NeuroSynth³⁸ (http://neurosynth.org) was employed to connect focal brain atrophy identified in the VBM meta-analysis with associated functional systems and related cognitive functions. This approach treats the brain as an interconnected network in which structural damage in one region can affect related functional systems.³⁹ By combining these analyses we aimed to obtain a more-comprehensive understanding of the interplay between brain structure and function in patients with AD carrying APOE ε4. We selected the peak coordinates from each cluster identified in our meta-analysis as a seed region to create meta-analytic coactivation maps and to identify keywords related to these maps. NeuroSynth generates a z-score map that indicates the probability that a specific term was identified in a research article in association with an activation site. We applied a voxelwise threshold to the NeuroSynth outcomes at a significance level of 0.01 while adjusting for the FDR. Up to five keywords with z-scores >0 were selected.

RESULTS

ALE meta-analysis

Twelve studies involving 25 experiments with 1,135 participants and yielding 164 foci met the relevant inclusion criteria and were included in the analysis. The results are summarized in Fig. 2 and Table 2. The primary analysis with all patients (AD+MCI; 10 experiments, 51 foci, and 608 patients) revealed that patients carrying the APOE ε4 allele exhibited significant atrophy of the bilateral hippocampus and parahippocampal regions relative to noncarriers (Fig. 2A). Further detailed analyses by patient category revealed that in patients with AD (6 experiments, 35 foci, and 405 ADs), APOE ε4 carriers exhibited notable atrophy of the bilateral hippocampus, the right parahippocampal gyrus, and the posterior cingulate cortex (PCC) (Fig. 2B). In patients with MCI (4 experiments, 16 foci, and 203 MCIs), carriers of the APOE ε4 allele exhibited atrophy of the bilateral parahippocampal gyrus and the globus pallidus (Fig. 2C). Among HCs (15 experiments, 113 foci, and 527 HCs), those carrying the APOE ε4 allele exhibited atrophy of the right superior temporal gyrus (Fig. 2D).

Fig. 2. Results of the ALE meta-analysis: all patients (AD+MCI) carriers<noncarriers (A), AD carriers<noncarriers (B), MCI carriers<noncarriers (C), and HC carriers<noncarriers (D). Color bar indicates z-scores. AD, Alzheimer’s disease; ALE, activation likelihood estimation; HC, healthy control; MCI, mild cognitive impairment.

Table 2. Results of the ALE meta-analysis of APOE carriers<noncarriers.

Cluster		Coordinates (x, y, z)	ALE score	z	p
All (AD+MCI) carriers<noncarriers
	Parahippocampal gyrus	28, -34, -2	0.034	7.362	<0.000001
	Hippocampus	-26, -14, -18	0.015	4.448	0.000004
	Parahippocampal gyrus	26, -14, -20	0.017	4.776	0.000001
	Hippocampus	32, -22, -10	0.018	5.176	<0.000001
AD carriers<noncarriers
	Parahippocampal gyrus	26, -34, -2	0.027	6.667	<0.000001
	Hippocampus	32, -22, -10	0.018	5.309	<0.000001
	Posterior cingulate cortex	-8, -52, 26	0.018	5.279	<0.000001
	Hippocampus	-26, -39, -2	0.018	5.296	<0.000001
MCI carriers<noncarriers
	Parahippocampal gyrus	-24, -14, -18	0.014	4.855	0.000001
	Globus pallidus	-18, -8, -12	0.011	4.463	0.000004
	Parahippocampal gyrus	26, -14, -18	0.020	6.185	<0.000001
	Parahippocampal gyrus	26, -34, -2	0.015	4.966	<0.000001
HC carriers<noncarriers
	Superior temporal gyrus	54, -30, 8	0.025	5.655	<0.000001
	Middle temporal gyrus	56, -34, 2	0.019	4.840	0.000001

AD, Alzheimer’s disease; ALE, activation likelihood estimation; APOE, apolipoprotein E; HC, healthy control; MCI, mild cognitive impairment.

NeuroSynth

To identify the functional networks underlying the patterns observed in our meta-analysis, specifically comparing AD carriers and noncarriers, we utilized NeuroSynth to perform a meta-analytic coactivation analysis. The results are summarized in Fig. 3. Analysis of the coactivation map for the parahippocampal gyrus cluster (MNI coordinates: 26, −34, −2) revealed connectivity with the bilateral hippocampus, parahippocampal gyrus, and PCC (Fig. 3A). The hippocampal clusters (MNI coordinates: 32, −22, −10 and −26, −39, −2) were primarily linked with the bilateral hippocampus (Fig. 3B and C). The coactivation analysis showed that the PCC cluster (MNI coordinate: −8, −52, 26) was associated with the medial prefrontal cortex, precuneus, PCC, bilateral angular gyrus, middle temporal gyrus, and hippocampus (Fig. 3D). Additionally, word clouds generated by NeuroSynth meta-analysis decoding revealed that the most frequently associated terms with the brain activity patterns in these coactivation maps included “episodic,” “memory,” “encoding,” and “retrieval” (Fig. 3E).

Fig. 3. NeuroSynth results. Meta-analytic coactivation maps of the parahippocampal gyrus cluster (A), right hippocampus cluster (B), left hippocampus cluster (C), and PCC cluster (D). Word cloud illustrating the findings of a NeuroSynth meta-analysis revealing the terms most commonly associated with the brain activity patterns of seed regions identified when comparing AD carriers and noncarriers (E). Color bar indicates z-scores. AD, Alzheimer’s disease; MNI, Montreal Neurological Institute; PCC, posterior cingulate cortex.

DISCUSSION

The present meta-analysis aimed to determine how the APOE ε4 allele influences brain atrophy in AD, in order to obtain insights into the mechanisms underpinning brain atrophy in this condition. Our examinations of the effects of the APOE ε4 allele in individuals at various stages of clinical progression of AD and MCI have revealed how this allele impacts the hippocampus and parahippocampus, which are areas vulnerable to neurodegenerative changes. We found significantly more atrophy of the hippocampus and parahippocampus in APOE ε4 carriers than in noncarriers, especially within the AD and MCI groups, while HC carriers showed no significant atrophy in these regions. These findings are consistent with previous research highlighting the susceptibility of the hippocampus to early degenerative changes in AD, in which it plays a critical role in memory formation, storage, and retrieval.³,³³,³⁴,³⁵ The observed atrophy in these critical memory regions among APOE ε4 carriers suggests that the allele contributes to the exacerbation of memory loss in AD, supporting the notion that this genetic variant is a significant risk factor for accelerated neurodegeneration.⁷,⁸,⁴⁰

The presence of the APOE ε4 allele has been associated with a reduced hippocampal volume, which is particularly evident in older individuals with AD and MCI, with an average volume decrease of about 4% in the hippocampus per APOE ε4 allele.⁴¹,⁴²,⁴³ Based on findings from large-scale genome-wide association study meta-analyses, Lupton et al.⁴⁴ further confirmed the link between the APOE ε4 allele and hippocampal volume reduction in patients with AD and MCI, indicating that the hippocampus is smaller in carriers of the ε4 allele than in noncarriers. Moreover, Kerchner et al.⁴⁵ reported that patients with MCI and AD who were also APOE ε4 carriers had thinner hippocampal subregions and showed worse episodic memory abilities relative to noncarriers. Moreover, our analyses using NeuroSynth revealed the hippocampal network that is associated with memory function. Located in the medial temporal lobe, the hippocampus is among the first regions to be damaged in AD.³ This early vulnerability is attributable to the critical roles that this brain structure plays in encoding new memories and navigating spatial environments.⁴⁶ The parahippocampal gyrus, which surrounds the hippocampus, supports memory encoding and retrieval by processing environmental contextual information.⁴⁷ Together these structures are crucial components of the brain’s memory network that facilitate the transition of short-term memories to long-term storage.⁴⁸ Our meta-analysis corroborated these genomic and neuroimaging findings, demonstrating the significant influence of the APOE ε4 allele on hippocampal atrophy in AD and the related memory impairments.

Furthermore, our analysis revealed atrophy of the PCC among patients with AD that was greater in those carrying the APOE ε4 allele than in noncarriers. The PCC is an integral part of the Papez circuit, which also includes the hippocampus, mammillary bodies, thalamus, and parahippocampal gyrus.⁴⁹ This circuit plays pivotal roles in memory formation and emotional regulation, and is implicated in the deterioration of episodic memory and spatial navigation ability seen in AD.⁵⁰ The PCC is also a central element of the default-mode network, which is known to be affected by AD pathology.⁵¹,⁵² Studies have repeatedly found that the volume of the PCC⁵² and its functional connectivity⁵¹ are disrupted in AD. Importantly, both AD and MCI patients who carry the APOE ε4 allele have been observed to exhibit loss of gray matter⁵³ and impaired functional connectivity⁵⁴ in the PCC. Consistent with previously obtained evidence, our findings suggest the influence of the APOE ε4 allele on PCC abnormalities, supporting the notion of PCC atrophy as a distinctive cortical feature of AD.⁵⁵ Overall our results highlight the contribution of the APOE ε4 allele to the genetic risk of brain atrophy and the cognitive deficits associated with AD.

Despite the robust associations observed, caution is required when interpreting the results of this study due to its limitations, which include a relatively small sample due to the scarcity of studies directly comparing APOE ε4 carriers with noncarriers. This study did not have a mechanistic design, and so further studies are needed to better understand whether the APOE ε4 allele has a direct or indirect impact on AD pathology, focal brain atrophy/neurodegeneration, or conversion of MCI to AD, or on disease progression. However, our analyses provide a foundational basis for future research in this field, suggesting the importance of genetic screening in tailoring potential prevention and treatment strategies for AD. APOE genetic testing is important for monitoring amyloid therapies such as those using donanemab and lecanemab, and it could also be useful for investigating whether amyloid therapy can slow down the hippocampal atrophy in carriers vs. noncarriers. Recognizing the multifactorial nature of AD, further studies are also essential to unravel the complex interplay of genetic and neuronal factors underlying the disease, in order to facilitate the development of more-effective interventions and a deeper understanding of its pathophysiology.

By applying a meta-analysis to structural neuroimaging data, we have demonstrated how this genetic variant contributes to the understanding of disease and its impact on brain structure. By identifying specific brain atrophy patterns associated with the APOE ε4 allele, our meta-analysis revealed the significant link between genetic factors and neurodegeneration, and hence potential pathways for targeted interventions and treatments. The recognition that APOE ε4 carriers in AD exhibit distinct patterns of brain atrophy, especially in the hippocampus and parahippocampus, will enable more-accurate risk assessments and individualized approaches to mitigating neurodegenerative progression.

Availability of Data and Material

All data generated or analyzed during the study are included in this published article.