Impact of APOE4-CSF Aβ interaction on hippocampal volume loss over 1 year in MCI

Abstract

Background

The majority of studies relating amyloid pathology with brain volumes have been cross-sectional. Apolipoprotein E4 (APOE4), a genetic risk factor for Alzheimer’s disease (AD), is also associated with hippocampal volume loss. No studies have considered the effects of amyloid pathology and APOE4 together on longitudinal volume loss.

Methods

We evaluated whether an abnormal level of cerebrospinal fluid beta-amyloid (CSF Aβ) and APOE4 carrier status were independently associated with greater hippocampal volume loss over 1 year. We then assessed whether APOE4 status and CSF Aβ acted synergistically, testing the significance of an interaction term in the regression analysis. We included 297 participants: 77 cognitively normal (NC), 144 with mild cognitive impairment (MCI), and 76 with AD.

Results

An abnormal CSF Aβ level was found to be associated with greater hippocampal volume loss over 1 year in each group. APOE4 was associated with hippocampal volume loss only in the NC and MCI groups. APOE4 carriers with abnormal CSF Aβ in the MCI group acted synergistically to produce disproportionately greater volume loss than noncarriers.

Conclusion

Baseline CSF Aβ predicts progression of hippocampal volume loss. APOE4 carrier status amplifies the degree of neurodegeneration in MCI. Understanding the effect of interactions between genetic risk and amyloid pathology will be important in clinical trials and our understanding of the disease process.

1. Introduction

Fibrillar beta-amyloid (Aβ) plaques, one of the hallmarks of Alzheimer’s disease (AD), have been shown to be associated with hippocampal atrophy in multiple cross-sectional positron emission tomography (PET) studies using the amyloid ligand, Pittsburgh Compound B (PiB) [1-5]. A few studies have found similar correlations between cerebrospinal fluid (CSF) Aβ, an indirect measure of cerebral amyloid deposition [6,7], and hippocampal atrophy [8,9]. However, studies relating Aβ pathology with longitudinal volume loss have been mixed. One PiB-PET study found a strong association between brain Aβ and change in regional MRI volumes in normals, but only a trend in AD [3]. One study reported an association between CSF Aβ and the rate of hippocampal atrophy [10], although CSF p-tau was found to be a better predictor, and two other studies found no correlation between Aβ and the rate of whole brain atrophy [11,12].

The primary goal of our study was to determine whether baseline CSF Aβ level is associated with longitudinal hippocampal volume loss, incorporating data from the multicenter Alzheimer’s Disease Neuroimaging Initiative (ADNI) (www.loni.ucla.edu\ADNI). Since apolipoprotein E4 (APOE4), a well-documented genetic risk factor for developing AD [13,14], is associated with increased brain Aβ [15-18] and hippocampal atrophy [19-21], we further explored whether APOE4 modifies the relationship between abnormally low CSF Aβ and hippocampal volume loss.

2. Methods

2.1. Participants

The participants in this study were recruited through the ADNI between 2005 and 2008, a longitudinal study including 56 centers in the U.S. and Canada with the purpose of identifying biomarkers of early Alzheimer’s disease (AD) for clinical trials (www.adni-info.org). The ADNI was funded by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies, and non-profit organizations, as a 5-year public-private partnership.

2.2. APOE genotyping and clinical assessment

All participants underwent APOE genotyping at the baseline visit. Approximately 6 ml of blood were obtained from each participant in an EDTA tube, gently mixed by inversion, and shipped at ambient temperature to a single designated laboratory within 24 hours of collection for analysis.

Participants ranged in age from 55 to 90, did not have major depression or severe systemic illnesses that would interfere with participation, and did not take investigational or psychometric medications. The normal controls (NC) subjects had no memory complaint, had preserved activities of daily living, scored between 26 and 30 on a baseline Mini-Mental Status Examination (MMSE) [23], scored a 0 on the Clinical Dementia Rating scale (CDR) [24], and scored within the normal range on the Logical Memory II subscale (delayed paragraph recall) from the Wechsler Memory Scale – Revised (WMS-R Log Mem) [25]. The subjects with mild cognitive impairment (MCI) had a memory complaint that was verified by a study partner, had preserved activities of daily living, and scored between 24 and 30 on the MMSE, 0.5 on the CDR, and below the normal range on the WMS-R Log Mem. The AD subjects scored between 20 and 26 on the MMSE, between 0.5 or 1 on the CDR, and met NINCDS/ADRDA criteria for probable AD [26]. Written consent was obtained from all subjects participating in the study, and the study was approved by the institutional review board at each participating site.

2.3. CSF analysis

As described in the ADNI protocol (www.adni-info.org), all 56 participating centers were asked to perform lumbar punctures on at least 20% of their participants. Approximately half of the participants recruited at each center underwent lumbar puncture for cerebrospinal fluid analysis. CSF samples were banked and batch-processed at a single laboratory, as described previously [27]. Briefly, lumbar puncture was performed with a 20- or 24-gauge spinal needle at the baseline visit after an overnight fast. The CSF samples were then transferred into polypropylene transfer tubes, frozen on dry ice within an hour after collection, and shipped on dry ice overnight to a single designated laboratory. After thawing for 1 hour at room temperature and gentle mixing, 0.5 ml aliquots were prepared from these samples. The aliquots were then stored in bar code-labeled polypropylene vials at −80°C and measured using the xMAP Luminex platform (Luminex Corp, Austin, TX) with Innogenetics (INNOBIA AlzBio3, Ghent, Belgium) immunoassay kit-based reagents, which included the monoclonal antibody specific for Aβ_1-42 (4D7A3).

In our analysis, the baseline CSF Aβ level was dichotomized as either abnormal, i.e. reflective of underlying Alzheimer’s pathology, or normal (Figure 1). It was previously published that using a threshold CSF Aβ value of 192 pg/ml yielded a sensitivity of 96% for detecting Alzheimer’s disease, based on a sample of non-ADNI normal controls and subjects with Alzheimer’s disease using the same CSF assay [28]. Furthermore, this cutoff value showed 91% agreement with evidence of brain amyloid using Pittsburgh Compound B in positron emission tomography imaging [29].

Figure 1. Association between baseline CSF Aβ level and 1-year change in hippocampal volumes.

The CSF Aβ level of less than 192 pg/ml (as delineated by the solid line) is considered abnormal in this study, i.e. reflective of underlying Alzheimer’s pathology. Greater than 192 pg/ml is considered normal.

2.4. MRI acquisition

Participants underwent the following standardized 1.5 T MRI protocol (http://www.loni.ucla.edu/ADNI/Research/Cores/index.shtml): two T₁-weighted MRI scans, using a sagittal volumetric magnetization prepared rapid gradient echo (MP-RAGE) sequence, with an echo time (TE) of 4 ms, repetition time (TR) of 9 ms, flip angle of 8°, and acquisition matrix size of 256 × 256 × 166 in the x-, y- and z-dimensions with a nominal voxel size of 0.94 × 0.94 × 1.2 mm [30].

2.5. MRI post-processing

The raw Digital Imaging and Communications in Medicine MRI data were downloaded from the Laboratory of Neuro Imaging (LONI) Image Database Archive (http://www.loni.ucla.edu/ADNI/Data/index.shtml). The images were aligned, skull-stripped, and segmented using FreeSurfer software, version 4.3 (http://surfer.nmr.mgh.harvard.edu/) [31]. Bilateral hippocampal volumes, obtained from this segmentation, were averaged in the analyses. The change in hippocampal volumes over 1 year was calculated by subtracting the baseline hippocampal volume from the volume at follow-up and normalized by the time difference.

2.6. Statistical analyses

We excluded 33 subjects who carried at least one APOE2 allele to avoid confounding the analysis, since APOE2 is believed to be protective against development of AD and slow rates of hippocampal atrophy [32,33]. Our final cohort thus includes 297 subjects who had a lumbar puncture and at least 2 MRI scans, spaced 1 year apart – 77 NC, 144 MCI, and 76 with probable AD (Table 1).

Table 1.

Group characteristics

	NC		p- value	MCI		p- value	AD		p- value
	APOE 3/3	APOE 3/4 or 4/4**		APOE 3/3	APOE 3/4 or 4/4**		APOE 3/3	APOE 3/4 or 4/4**
N	55	22		65	79		24	52
Age (years)	76 (5.0)	76 (6.3)	0.37	76 (8.5)	73 (6.5)	0.03*	77 (9.1)	74 (7.1)	0.19
Female (%)	55	36	0.21	26	44	0.04*	42	42	0.91
Education (years)	16 (2.4)	16 (3.4)	0.84	16 (2.9)	16 (2.9)	0.82	15 (5.3)	14 (4.0)	0.14
CSF Aβ (pg/ml)	209 (48.8)	147 (43.5)	<0.001*	188 (59.8)	141 (38.8)	<0.001*	169 (53.6)	130 (29.3)	0.001*
Baseline hippocampal volume (mm3)	6752 (731.5)	6691 (724.2)	0.70	5904 (1072.9)	5599 (923.8)	0.06	5433 (1466.4)	5073 (792.0)	0.50
MMSE	29 (1.0)	29 (0.9)	0.45	27 (1.8)	27 (1.8)	0.49	23 (2.0)	24 (1.8)	0.68
Unadjusted 1-year change in hippocampal volume (mm3)	−57.8 (179.2)	−160.5 (150.9)	0.01*	−103.5 (176.0)	−199.9 (166.2)	0.003**	−231.0 (227.3)	−252.1 (174.1)	0.80

Data shown are means (SD).

^*Significant at the α = 0.05 level.

^**Number of (APOE3/4, APOE4/4) carriers in each group: NC (20, 2), MCI (63, 16), AD (34, 18).

All statistical analyses were programmed in R, version 2.9.2 (www.r-project.org). Model assumptions were assessed with plots of residuals. APOE genotype was dichotomized into APOE4 carriers (E3/4 or E4/4) and noncarriers (APOE3/3). Age, baseline hippocampal volume, gender, and years of education were included as covariates in every model.

We first determined whether an abnormal baseline CSF Aβ level and APOE4 carrier status were independently associated with 1-year change in hippocampal volumes in all stages, after adjusting for covariates, using ordinary least squares regression. If both risk factors were significantly associated with volume loss, we then tested for interaction between APOE4 and CSF Aβ. To do so, we centered CSF Aβ on its mean to reduce collinearity and included an interaction term between APOE4 and CSF Aβ, which was considered significant at the α = 0.05 level.

3. Results

3.1. Group characteristics

The group characteristics are summarized in Table 1. Mean CSF Aβ was significantly lower in APOE4 carriers compared to noncarriers at each clinical stage, consistent with previous literature [15-18]. The APOE4 MCI group was slightly younger and included more women. No significant differences in MMSE were seen between carriers and noncarriers. Without adjusting for covariates, the change in raw hippocampal volumes over 1 year was significantly different by APOE4 status in the NC and MCI groups, but not in AD.

3.2. Association between CSF Aβ and 1-year change in hippocampal volumes

Participants with an abnormally low CSF Aβ level had greater volume loss in all groups. In the NC group, participants with a low CSF Aβ level had a 138 mm³ greater 1-year volume loss than those with a normal CSF Aβ level (p < 0.001). In the MCI group, participants with a low CSF Aβ level had a 71 mm³ greater volume loss than those with a normal CSF Aβ level (p = 0.03). In the AD group, participants with a low CSF Aβ level had a 300 mm³ greater 1-year volume loss than those with a normal CSF Aβ level (p < 0.001).

3.3. Association between APOE4 and 1-year change in hippocampal volumes

Participants who carried at least one APOE4 allele had greater volume loss in the NC and MCI groups. In the NC group, APOE4 participants had a 121 mm³ greater 1-year volume loss than those without an APOE4 allele (p < 0.007). In the MCI group, APOE4 participants had a 76 mm³ greater volume loss than those without an APOE4 allele (p = 0.01). In the AD group, APOE4-positive and APOE4-negative participants demonstrated no difference in 1-year volume loss (p = 0.66).

3.4. Testing for APOE4-CSF Aβ interaction in NC and MCI

Since both APOE4 and a low CSF Aβ level were associated with greater volume loss in the NC and MCI groups, we then tested for an APOE4-CSF Aβ interaction in these groups. No significant APOE4-CSF Aβ interaction was seen in the NC group (β = 138, p = 0.19). There was however a significant interaction between APOE4 and CSF Aβ in the MCI group (β = −181, p = 0.02). Compared to APOE4-noncarriers with normal CSF Aβ, APOE4-carriers with abnormal CSF Aβ had 88 mm³ greater volume loss over 1 year (p = 0.02). Compared to APOE4-noncarriers with abnormal CSF Aβ, APOE4-carriers with abnormal CSF Aβ had 97 mm³ greater volume loss over 1 year (p = 0.004).

4. Discussion

The major findings of this study are: 1) an abnormally low baseline CSF Aβ level, suggestive of underlying Alzheimer’s pathology, predicted 1-year change in hippocampal volumes in all groups; 2) APOE4 carriers demonstrate greater hippocampal volume loss only in the NC and MCI groups; 3) APOE4 and low CSF Aβ are synergistic risk factors, such that APOE4 carrier status amplifies the predicted 1-year volume loss beyond that predicted by a low CSF Aβ alone.

The finding that an abnormally low CSF Aβ predicted 1-year hippocampal volume loss is consistent with the predominantly cross-sectional literature, which describes an association between amyloid pathology and hippocampal atrophy [1-5, 8-10]. Some have postulated that the large extracellular amyloid plaques disrupt cortico-hippocampal pathways, leading to neurodegeneration [2]. Another hypothesis is that insoluble plaques detected in cerebrospinal fluid are an indirect marker of soluble Aβ oligomers, which may be the inciting agent in Alzheimer’s disease, by disrupting hippocampal synapses and promoting volume loss [34, 35].

The second finding that APOE4 is associated with greater longitudinal hippocampal volume loss in the NC and MCI groups is also compatible with prior literature. Numerous studies suggest that APOE4 carriers demonstrate increased vulnerability to developing AD, which is manifested through neurodegeneration [36-39]. A reason for the lack of increase volume loss among APOE4 carriers in the AD group may be that, although APOE4 carriers develop AD at an earlier age [13], once the disease is clinically apparent in an individual, APOE4 no longer alters the course of the disease. The lack of a significant difference in hippocampal volumes among APOE4 carriers and noncarriers with AD has also been reported in prior studies [40,41].

Finally, the finding that the presence of a genetic risk factor, APOE4, amplifies the association between CSF Aβ and progressive hippocampal volume loss in MCI is novel. One possible explanation for this is that the APOE4 carriers with low CSF Aβ are more likely to have Alzheimer’s pathology. Although an abnormally low CSF Aβ is highly sensitive for detecting brain amyloid associated with AD, it is not an entirely specific for AD [28]. Some of the participants with low CSF Aβ may have frontotemporal dementia and would not demonstrate the same degree of hippocampus-specific volume loss as prodromal Alzheimer’s patients [42]. However, this argument would also be true among NC subjects, in which no interaction was demonstrated.

A second explanation for the APOE4-CSF Aβ interaction in MCI could be explained by a temporal progression of pathological mechanisms resulting from the APOE4 genotype. Early on when subjects demonstrate normal cognition, the predominant effect of APOE4 appears to be to increase brain amyloid deposition, as reported by numerous prior studies [15-18]. Since, both APOE4 carrier status and a low CSF Aβ, as defined by our cutoff value, reflect greater brain amyloid, no interaction is seen in our NC group. However, once cognitive impairment is evident clinically, as in the MCI group, the effects of Aβ and APOE4 on pathogenesis of Alzheimer’s disease may diverge, thus resulting in disproportionately greater volume loss in those with both risk factors. Indeed, APOE4 has been found to be associated with an inability to repair synaptic damage, more rapid promotion of other neurotoxic species, such as tau, susceptibility to oxidative stress, and promotion of inflammatory cascades [17], beyond simply increasing levels of brain amyloid. Further work examining this interaction is warranted.

A third possible explanation is that both APOE4 and a low CSF Aβ are markers of disease progression. According to the literature, only 10-15% of individuals with MCI will convert to AD each year [43]. The other 85-95% of stable MCI individuals may be more likely have higher levels of CSF Aβ and be APOE4-negative, thus resulting in slower hippocampal volume loss.

Several study limitations deserve mention. First, the ADNI was designed to mimic a trial population, so participants were more educated, more Caucasian, and had fewer comorbidities than a community-based cohort [22]. The generalizability of our conclusions is thus controversial, and the length of follow-up was short. Second, this was a secondary analysis of the cohort, so there were different proportions of APOE4 carriers individuals at each clinical stage. Overall, the NC and AD groups had about half the number of participants as the MCI group, resulting in reduced power to detect differences. Rather than take a sample with balanced proportions, we wanted to include all available data. Furthermore, an allelic dose-dependent effect of APOE4 could not be explored, since only two NC were homozygous for APOE4, and the MCI and AD had imbalanced proportions of heterozygotes and homozygotes. Third, we only included hippocampal volumes as a marker of structural change to limit the number of comparisons. Inclusion of other limbic or whole brain markers would potentially detect more APOE4 effects not described in our analysis. Further prospective studies are needed to validate our findings.

In summary, we demonstrated that baseline CSF levels of Aβ are predictive of near-term hippocampal volume loss. The strengths of this study include the recruitment of participant from multiple centers, longitudinal follow-up, and consideration of all 3 clinical stages. We further raised the possibility of an APOE4-CSF Aβ interaction on longitudinal hippocampal atrophy among MCI participants. As interest grows in using hippocampal atrophy as an outcome in clinical trials, it will be important to consider how varying risk factors and biomarkers interact and influence the progression of neurodegeneration.