Correlations between apolipoprotein E ε4 gene dose and brain-imaging measurements of regional hypometabolism

Abstract

Patients with Alzheimer's disease (AD) have abnormally low positron emission tomography (PET) measurements of the cerebral metabolic rate for glucose (CMRgl) in regions of the precuneus and the posterior cingulate, parietotemporal, and frontal cortex. Apolipoprotein E (APOE) ε4 gene dose (i.e., the number of ε4 alleles in a person's APOE genotype) is associated with a higher risk of AD and a younger age at dementia onset. We previously found that cognitively normal late-middle-aged APOE ε4 carriers have abnormally low CMRgl in the same brain regions as patients with probable Alzheimer's dementia. In a PET study of 160 cognitively normal subjects 47–68 years of age, including 36 ε4 homozygotes, 46 heterozygotes, and 78 ε4 noncarriers who were individually matched for their gender, age, and educational level, we now find that ε4 gene dose is correlated with lower CMRgl in each of these brain regions. This study raises the possibility of using PET as a quantitative presymptomatic endophenotype to help evaluate the individual and aggregate effects of putative genetic and nongenetic modifiers of AD risk.

To evaluate putative genetic and nongenetic modifiers of the risk for a psychiatric or neurological disorder, it would be helpful to identify an “endophenotype,” a measurable feature that is more closely related to disease susceptibility than the clinical syndrome itself (1). A useful endophenotype should be associated with the disorder, associated with heritable or nonheritable risk factors in clinically unaffected persons, and correlated with relative risk.

Fluorodeoxyglucose positron emission tomography (PET) measurements of the cerebral metabolic rate for glucose (CMRgl) provide a promising quantitative neuroimaging endophenotype of Alzheimer's disease (AD) risk. AD is associated with abnormally low CMRgl in the precuneus and the posterior cingulate, parietotemporal, and frontal cortex (2, 3). In patients with Alzheimer's dementia, the CMRgl reductions are progressive (2) and correlated with dementia severity (3) and predict subsequent clinical decline and the histopathological diagnosis of AD (4). In patients with mild cognitive impairment, the CMRgl reductions predict subsequent conversion to probable AD (5–7). In comparisons between nondemented carriers and noncarriers of the apolipoprotein E (APOE) ε4 allele, a susceptibility gene for late-onset AD (8, 9), which is found in almost one-fourth of the population (10), we (11–13) and others (14, 15) found that the ε4 carriers have significantly lower CMRgl in each of the same brain regions as clinically affected patients.

Extending our original findings to almost four times as many cognitively normal late-middle-aged persons with two, one, and no copies of the ε4 allele, we now demonstrate that APOE ε4 gene dose (i.e., the number of ε4 alleles in a person's APOE genotype), a heritable factor known to be associated with a progressively higher risk of AD (8, 9) and a progressively younger age at dementia onset (8), is correlated with lower CMRgl in the brain regions preferentially affected by AD. Based on these findings, we suggest how PET could be used before the possible onset of symptoms as a quantitative endophenotype to help assess the individual and aggregate effects of putative genetic and nongenetic modifiers of the risk for AD.

Methods

Subjects. Newspaper advertisements were used to recruit cognitively normal volunteers 47–68 years of age who reported a first-degree family history of probable AD, understood they would not receive any information about their APOE genotype, provided their informed consent, and were studied under guidelines approved by the human subjects committees at Banner Good Samaritan Medical Center and the Mayo Clinic. Venous blood samples were drawn and APOE genotypes characterized with analysis involving restriction fragment length polymorphisms (16).

Clinical ratings, neuropsychological tests, volumetric MRI, and fluorodeoxyglucose PET were performed as described (7) in APOE ε4 homozygotes, heterozygotes, and noncarriers who denied an impairment in memory or other cognitive skills, had scores of at least 28 on the Folstein MiniMental State Examination and <10 on the Hamilton Depression Rating Scale, did not satisfy criteria for a current psychiatric disorder using a structured psychiatric interview, did not use centrally acting medications for at least 2 weeks before their PET session, and had a normal neurological examination. The APOE ε4 heterozygotes (all with the ε3/ε4 genotype) and ε4 noncarriers were individually matched to each ε4 homozygote for their gender, age (within 3 years), and educational level (within 2 years). The 160 subjects included 36 APOE ε4 homozygotes, 46 ε4 heterozygotes, and 78 ε4 noncarriers (60 with the ε3/ε3 genotype and 18 with the ε2/ε3 genotype). [Eleven 11ε4 homozygotes, 11 ε4 heterozygotes, and 22 ε4 noncarriers were included in our previous reports (11, 12).]

Brain Imaging. Volumetric T₁-weighted MRI and PET were performed as described (11, 13, 17). PET was performed with the 951/31 ECAT scanner (Siemens, Knoxville, TN), a transmission scan, the i.v. injection of 10 mCi (1 Ci = 37 GBq) of ¹⁸F-fluorodeoxyglucose, and a 60-min dynamic sequence of emission scans as the subjects, who had fasted for at least 4 h lay quietly in a darkened room with their eyes closed and directed forward. For whole brain measurements, CMRgl (mg/min per 100 g) was calculated by using the PET images, an image-derived radiotracer input function, plasma glucose levels, and a graphical method (18). Regional analyses were performed by using the PET images (in counts) acquired during the last 30 min.

An automated algorithm [SPM99, Wellcome Department of Cognitive Neurology, London (19)] was used to linearly and nonlinearly deform each person's PET image into the coordinates of a standard brain atlas, normalize data for the variation in whole brain measurements by using proportionate scaling, and generate statistical parametric maps of the CMRgl reductions in the APOE ε4 carriers and the correlations between APOE ε4 gene dose and lower CMRgl (P < 0.005, uncorrected for multiple comparisons). The statistical maps were superimposed onto “AD search regions,” which were used to investigate the predicted effects of the ε4 allele on CMRgl in brain regions preferentially affected by AD, and a spatially standardized volume-rendered MRI. The AD search regions were defined as AD-related reductions in regional-to-whole brain CMRgl; were determined by the comparison among previously reported patients with Alzheimer's dementia and normal controls (2); and were located in the precuneus, posterior cingulate cortex, and parietotemporal cortex (P < 0.005, uncorrected for multiple comparisons), and the frontal cortex (P < 0.05, uncorrected for multiple comparisons). To reduce Type I errors, the maximal correlation between APOE ε4 gene dose and lower CMRgl in each AD search region was corrected for the number of resolution elements in that region using the small-volume correction procedure in SPM99. In post hoc analyses, the subjects' coregistered MRI and PET images, SPM99, and a partial-volume correction algorithm (20) were used to generate and compare statistical parametric maps with and without correction for individual differences in brain tissue volume (e.g., the combined effects of brain atrophy and image blurring).

Results

There were minimal differences in the three subject groups' demographic features, clinical ratings, and neuropsychological scores (Table 1). There were no significant differences in their measurements of whole brain CMRgl (mean ± SD, 5.60 ± 0.96 mg/min per 100 g in the ε4 homozygotes, 5.65 ± 0.82 mg/min per 100 g in the ε4 heterozygotes, and 5.50 ± 1.09 mg/min per 100 g in the ε4 noncarriers, ANOVA, P = 0.79).

Table 1. Subject characteristics, clinical ratings, and neuropsychological scores.

Variable	ε4 homozygotes n = 36	ε4 heterozygotes n = 46	ε4 noncarriers n = 78	P value
Age	55.6 ± 5.2	56.1 ± 4.2	56.5 ± 4.7	0.52
Gender (F/M)	23/13	30/16	49/29	0.96
Years of education	15.8 ± 2.5	16.0 ± 2.6	15.8 ± 2.3	0.99
MMSE	29.7 ± 0.6	29.9 ± 0.4	29.6 ± 0.7	0.28
HAM-D	2.5 ± 3.3	1.7 ± 1.9	3.9 ± 4.7	0.08
AVLT
Total learning	47.6 ± 8.0	48.5 ± 9.4	50.2 ± 10.2	0.30
Short-term memory	9.4 ± 2.5	10.4 ± 2.5	10.3 ± 3.4	0.07
Long-term memory	8.7 ± 2.9	10.0 ± 2.9	9.8 ± 3.4	0.05
Complex figure test
Copy	34.9 ± 1.5	33.9 ± 2.9	34.3 ± 1.9	0.10
Recall	18.4 ± 6.1	19.0 ± 6.6	17.6 ± 6.1	0.60
Boston naming test	57.1 ± 3.1	56.8 ± 2.9	56.7 ± 3.4	0.67
WAIS-R
Information	11.9 ± 2.2	12.3 ± 2.1	11.2 ± 2.1	0.06
Digit span	11.3 ± 2.2	11.6 ± 2.9	11.3 ± 2.7	0.81
Block design	11.9 ± 2.7	12.2 ± 2.7	11.9 ± 2.5	0.83
Arithmetic	11.9 ± 2.4	12.3 ± 2.5	11.4 ± 2.7	0.47
Similarities	12.4 ± 2.2	12.5 ± 2.0	11.8 ± 1.9	0.28
COWAT	44.4 ± 11.0	43.8 ± 11.5	47.7 ± 10.3	0.23
WMS-R orientation	13.8 ± 0.4	13.9 ± 0.4	13.7 ± 0.4	0.51

MMSE, Mini Mental State Exam; HDRS, Hamilton Depression Rating Scale; AVLT, Auditory verbal learning Test; WAIS-R, Wechsler Adult Intelligence Scale-Revised; COWAT, Controlled Oral Word Association Test; WMS-R, Wechsler Memory Scale-Revised.

The ε4 carriers had abnormally low CMRgl in regions of the posterior cingulate, precuneus, parietotemporal, and frontal regions previously implicated in patients with Alzheimer's dementia (P < 0.05 after correction for multiple comparisons). Moreover, APOE ε4 gene dose was significantly correlated with lower CMRgl in each of these brain regions (Fig. 1; Table 2). Further supporting the relevance of these findings to the pathophysiology of AD, significant correlations between ε4 gene dose and lower CMRgl were limited to the vicinity of the AD search regions, and significant correlations between ε4 gene dose and higher CMRgl were not observed anywhere in the brain. Because the CMRgl reductions in the ε4 carriers and their correlations with ε4 gene dose remained significant after correction for partial-volume averaging, and because the resulting CMRgl reductions and correlations were not significantly different from those found without this correction, the PET findings do not appear to be solely attributable group differences in brain tissue volume (e.g., the combined effects of brain atrophy and image blurring).

Fig. 1.

Correlations between APOE ε4 gene dose and lower CMRgl (shown in blue, P < 0.005, uncorrected for multiple comparisons) in cognitively normal late-middle-aged persons are projected onto the lateral and medial surfaces of the left and right cerebral hemispheres and shown in relationship to CMRgl reductions (purple and dark blue) found in previously described patients with Alzheimer's dementia (2) (purple and dark blue). Significant correlations were observed bilaterally in the precuneus (PCu), posterior cingulate (PC), left and right parietotemporal (PT), and left frontal (F) regions preferentially affected by AD (dark blue), and in adjacent PC, PCu, PT, and F regions (light blue).

Table 2. Location and magnitude of most significant correlations between APOE ε4 gene dose and regional hypometabolism.

	Atlas coordinates, mm*
Brain region	x	y	z	Correlation coefficient	P value†
Posterior cingulate	-1	-54	23	0.27	2.8 × 10-4
Precuneus	13	-80	48	0.29	1.3 × 10-4
Left parietotemporal	-43	-69	42	0.34	6.9 × 10-6
Right parietotemporal	39	-71	33	0.26	3.2 × 10-3
Left frontal	-45	1	37	0.23	1.5 × 10-3

The data were extracted from voxels associated with the most significant correlations in each predetermined AD search region.

^*The coordinates were obtained from Talairach and Tournoux (24). x is the distance to the right (+) or left (-) of the midline, y is the distance anterior (+) or posterior (-) to the anterior commissure, and z is the distance superior (+) or inferior (-) to a horizontal plane through the anterior and posterior commissures

^†The reported significance levels were computed before statistical correction for multiple comparisons. Correlations in the posterior cingulate, precuneus, left parietotemporal, and left frontal regions remained significant (P < 0.05) after correcting for multiple comparisons in the respective AD search region. Correlations in the posterior cingulate, precuneus, and left parietotemporal region remained significant (P < 0.05) after further correcting for the number of search regions

Discussion

This study examined inverse correlations between three levels of genetic risk for AD (i.e., zero, one, or two copies of the APOE ε4 allele) and regional CMRgl in cognitively normal late-middle-aged persons. As predicted, progressively increased risk for AD was associated with progressively decreased CMRgl in and only in the posterior cingulate, precuneus, parietotemporal, and frontal regions preferentially affected in patients with Alzheimer's dementia. Studies in patients with mild cognitive impairment (5–7) and the younger asymptomatic relatives of patients with autosomal dominant early-onset forms of AD (21) suggest that the relationship between AD risk and the observed reductions in regional CMRgl is not limited to the APOE ε4 allele. Based on these findings, we suggest that PET could provide a quantitative presymptomatic neuroimaging endophenotype to help evaluate the individual and aggregate effects of putative genetic and nongenetic modifiers of the risk for AD.

As a complement to observational studies of older Alzheimer's dementia cases and controls, which typically require retrospective assessment of antecedent risk modifiers, our proposed endophenotype could permit the concurrent and prospective evaluation of these putative risk modifiers before the onset of symptoms. It could help address the potentially confounding effects of differential survival related to the putative risk modifier (a problem raised, for instance, in observational studies of the APOE genotype, gender, educational level, and tobacco use) and biased or inaccurate recall of the putative risk modifier (for instance, recalling one's presymptomatic use, timing, dose, or duration of hormone replacement therapy or dietary supplements or one's presymptomatic level of aerobic exercise); and it could provide easier access to accurate presymptomatic measurements of the putative risk modifier (e.g., serum cholesterol, lipid, or homocysteine levels). Furthermore, it could provide information about both the individual and net effects of suggested risk modifiers on a quantitative indicator of disease susceptibility. As a complement to prospective longitudinal cohort studies, this endophenotype could provide information about the putative risk factor without having to study many more subjects or wait many years to determine whether or when they develop symptoms. Indeed, it could be used prospectively to assess the individual or aggregate effects of the putative risk factors on subsequent CMRgl decline. [Supporting the last possibility, cognitively normal ε4 carriers have significantly greater 2-year declines in regional CMRgl than ε4 noncarriers (15, 22), and ε4 gene dose is significantly correlated with 2-year CMRgl declines in these locations.o] Finally, this endophenotype could help overcome the confounding effects of individual differences in cognitive reserve capacity, which may compensate for AD neuropathology and mask its clinical expression (23).

Although APOE ε4 gene dose was significantly correlated with hypometabolism in postulated brain regions, there was considerable overlap among the three genetic groups in their individual measurements. As we have noted (11), neither genetic testing for APOE genotype nor PET is clinically indicated to predict a cognitively normal person's risk of AD. This information does not yet determine with sufficient accuracy whether or when a person might develop AD and may be associated with psychological and social risks, and established prevention therapies are not yet available. Our ongoing longitudinal study promises to determine the extent to which APOE ε4 gene dose, brain imaging measurements, and other relevant data predict subsequent rates of cognitive decline and conversion to mild cognitive impairment and AD.

Conclusion

This article describes a relationship between three levels of genetic risk for AD and CMRgl reductions in brain regions that are preferentially affected by this disorder. It raises the possibility of using PET as a quantitative presymptomatic endophenotype for the cost-effective evaluation of the individual and aggregate effects of putative genetic and nongenetic modifiers of AD risk. If suitable measurements can be developed, it may be possible to extend this paradigm to other brain-imaging and nonimaging endophenotypes of AD risk, to the study of putative modifiers of the risk for other neurological and psychiatric disorders, and to the study of putative modifiers of the rate of brain aging, itself a risk factor for many age-related disorders.