Apolipoprotein E ε4 Is Associated With the Presence and Earlier Onset of Hemorrhage in Cerebral Amyloid Angiopathy
Background and Purpose Cerebral amyloid angiopathy is an important cause of intracerebral hemorrhage in the elderly. The ε4 allele of the apolipoprotein E gene, recently established as a genetic risk for Alzheimer's disease, has also been suggested as a possible risk factor for cerebral amyloid angiopathy. We sought to determine whether this allele is specifically associated with hemorrhages related to amyloid angiopathy and whether it correlates with the age at which first amyloid angiopathy-related hemorrhage occurs.
Methods Forty-five consecutive patients presenting with lobar hemorrhage were prospectively classified according to clinical, radiological, and when available, pathological features and evaluated for apolipoprotein E genotype. They were compared with 1899 elderly patients from a population-based sample and with 18 consecutive patients with hemorrhages in deep regions typical of a hypertensive mechanism.
Results Patients with multiple hemorrhages confined to the lobar territory demonstrated a greater than twofold overrepresentation (P<.001) in frequency of the apolipoprotein E ε4 allele compared with the population-based sample. Apolipoprotein E genotypes of patients with hemorrhages in deep territories resembled the population sample. Among patients with strictly lobar hemorrhages, carriers of the ε4 allele had their first hemorrhage more than 5 years earlier than noncarriers (mean age at first hemorrhage, 73.4±8.0 versus 78.9±7.4 years; P=.033). These effects were independent of the accompanying presence of Alzheimer's disease.
Conclusions The data support a specific role for apolipoprotein E ε4 in accelerating the process that leads to amyloid angiopathy-related hemorrhage.
Cerebral amyloid angiopathy refers to deposition of congophilic material in the cerebral vasculature, particularly small and medium-sized arteries of the leptomeninges and cerebral cortex.1 2 The amyloid peptide contained in the vascular deposits of CAA is essentially identical to the Aβ found in senile plaques in AD.3 4 While vascular amyloid is often tolerated without complication, some cases of extensive CAA undergo a cascade of vascular smooth muscle cell death, fibrinoid necrosis, and microaneurysm formation that culminates in hemorrhage.5 6
CAA-related hemorrhages are characterized clinically by their occurrence in the elderly population, their predilection for cortical/corticosubcortical (lobar) locations, and their multiplicity.2 The presence of multiple lobar hemorrhages in an elderly patient without other known cause of hemorrhage has therefore been suggested as grounds for the diagnosis of “probable CAA.”7 Clinical detection of small hemorrhages in CAA can be enhanced through use of the sensitive technique of gradient-echo MRI.8
The present study investigates the role of the ε4 allele of the apoE gene in CAA-related hemorrhage. The polymorphic apoE gene exists in three major alleles, ε2, ε3, and ε4, with ε3 occurring at the highest frequency.9 The ε4 allele has emerged as an important risk factor for AD,10 11 12 13 correlating with both increased incidence as well as earlier onset of dementia.14 15 16 The ε2 allele appears conversely to associate with decreased risk for AD.17 18 19
At a pathological level, the presence of apoE ε4 correlates with increased deposition of Aβ as senile plaques in the brain parenchyma and CAA in cerebral vessels.7 11 13 20 21 In a series of 93 postmortem cases systematically graded for severity of CAA, the presence of apoE ε4 increased the odds for moderate or severe CAA by 2.9-fold relative to cases without ε4.7
Since deposition of Aβ in vessels appears to be associated with destruction of the vessel wall, we hypothesized that apoE ε4 might be a specific risk factor for CAA-related intracerebral hemorrhage.7 The ε4 allele, however, is also associated with nonamyloid vascular diseases such as atherosclerosis,22 raising the alternative possibility that ε4 might exert a more general effect on the risk of intracerebral hemorrhage.
The present study was designed to address the following questions: (1) Does apoE genotype exert a specific effect on risk of CAA-related hemorrhage or a general effect on all types of intracerebral hemorrhage? (2) Among patients with CAA-related hemorrhage, does apoE genotype affect the course of disease as measured by the age of onset?
Subjects and Methods
Evaluation of Patients
Forty-five consecutive patients (22 men, 23 women) with lobar hemorrhage presenting to one of the participating institutions (Massachusetts General Hospital, Brigham and Women's Hospital, New England Medical Center, or Boston University Medical Center) were enrolled in this study. Subjects were excluded if they were younger than 50 years or if they had another known cause of hemorrhage (excessive warfarin [international normalized ratio >3.0], antecedent head trauma or ischemic stroke, intracerebral tumor, vascular malformation, vasculitis, blood dyscrasia, or coagulopathy). The 27 patients with definite/probable CAA-related hemorrhage described in the present study include 11 patients reported previously.7 The lobar hemorrhage patients were compared for clinical features and apoE genotype with 18 consecutive individuals (10 men, 8 women) older than 50 years presenting with nonlobar (basal ganglia, thalamus, pons, or cerebellum) hemorrhages (“deep hemorrhage”). One of the 45 lobar hemorrhage patients and one of the 18 deep hemorrhage patients were black; all other patients were white.
The control population consisted of a population-based sample of 1899 individuals older than 65 years living in two Iowa counties.23 This population resembled the patients with hemorrhage in racial background (white) and mean age (79.08 years; see ages of hemorrhage patients in Table 1⇓). Although women predominated in the Iowa control population (669 men, 1229 women), apoE genotype did not vary with sex.
Patients with lobar hemorrhage underwent full clinical examination including MRI scan (see below) and were prospectively classified according to number and location of hemorrhages and, when available, pathological data. Those patients demonstrated to have either (1) lobar hemorrhage with pathological evidence of CAA6 or (2) multiple hemorrhages restricted to the cortical/corticosubcortical regions (Fig 1⇓) were classified as “definite/probable CAA-related hemorrhage” according to previously defined criteria.7 24 Patients with only a single lobar hemorrhage were designated as “possible CAA-related hemorrhage.” Our analysis revealed a third group of patients with hemorrhages in both lobar and deep locations, defined as “mixed hemorrhages” (Fig 2⇓).
The routine radiological evaluation of patients with lobar hemorrhage included gradient-echo MRI sequences. This technique enhances the magnetic susceptibility (and resultant signal dropout) produced by chronic blood products, thereby heightening sensitivity for hemorrhage.25 Multiplanar gradient-echo images (repetition time, 749 to 750 milliseconds; echo time, 50 milliseconds; number of excitations, 2; flip angle, 10°; time, 3:41) were obtained in the axial plane, as described previously.8
Clinical features (history of hypertension or dementia and age at first hemorrhage) were recorded before determination of genotype. Age at first hemorrhage was calculated according to initial clinical presentation with hemorrhage. The presence of hypertension was defined by previous use of antihypertensive medication or a requirement for antihypertensive therapy during hospitalization persisting at least 1 to 2 weeks beyond the date of hemorrhage. History of dementia, defined as progressive decline in memory, language, or other cognitive functions (abstract thinking, praxis, executive function) before cerebral hemorrhage, was determined through interview by the treating neurologist with family members.
Determination of apoE genotype was performed by restriction enzyme digestion of an apoE polymerase chain reaction product derived from blood samples, as described previously.7 Genotype analysis for all samples (case and control subjects) was performed and recorded by a single investigator who had no knowledge of the patient's clinical features.
This study was performed with approval and in accord with the guidelines of institutional review boards and with informed consent of subjects or family members.
Determination of odds ratios and comparison of apoE allele frequency (proportion of chromosomes in which an allele is present) between patients with different types of hemorrhage or between hemorrhage patients and the control population were performed with 2×2 tables with the use of Fisher's exact test for significance. Similar methods were used for comparison of the incidence of hypertension or dementia. Odds ratios are presented with 95% CIs as determined by Cornfield estimate. Ages at first hemorrhage were normally distributed (Shapiro-Francia test for normality) and compared by Student's t test. All analyses were performed with Stata software (Stata Corp), and all significance tests were two-tailed.
We evaluated 45 consecutive patients older than 50 years presenting with lobar hemorrhage and compared them with 18 consecutive patients older than 50 years with hemorrhage in nonlobar locations (deep hemorrhage) and with a population-based elderly control population. Twenty-seven of the 45 patients with lobar hemorrhage met criteria for definite/probable CAA-related hemorrhage (see “Subjects and Methods”), 8 with pathological evidence (biopsy or postmortem) for CAA, the remainder with radiological documentation of multiple lobar hemorrhages (Fig 1⇑). Another 12 patients evaluated radiologically and with only a single lobar hemorrhage were classified as possible CAA. The other 6 patients with an unexpected pattern of both lobar and deep hemorrhages (Fig 2⇑) were designated as having mixed hemorrhages.
Patients with definite/probable CAA-related hemorrhages were generally older at first hemorrhage, less likely to be hypertensive, and more likely to have preexisting dementia than the patients with deep hemorrhage (Table 1⇑). The differences reached statistical significance only for age at first hemorrhage. The group with single lobar hemorrhages (ie, possible CAA) resembled the probable CAA group in age of onset and proportion with hypertension, while the patients with mixed hemorrhages resembled the deep hemorrhage group.
ApoE genotype was determined for each patient by polymerase chain reaction and restriction enzyme digestion7 without knowledge of clinical phenotype (Table 2⇓). Genotypes from the hemorrhage patients were compared with those from an elderly population-based sample analyzed in the same laboratory.23 As observed previously,7 frequency of the apoE ε4 allele was significantly increased in patients with definite/probable CAA-related hemorrhage, occurring at greater than twice the frequency as the control population. Presence of an ε4 allele was associated with a threefold (95% CI, 1.4 to 6.4; P<.01) excess odds of definite/probable CAA relative to its absence; odds ratio for ε4 homozygotes was 6.1 (95% CI, 1.9 to 20.0; P<.02). The effect of the ε4 allele on CAA was independent of its role in AD, since apoE ε4 was elevated to the same degree (frequency=0.34; P<.002) in the subset of definite/probable CAA patients without a history of dementia.
An increase in ε4 allele frequency of similar magnitude (and borderline statistical significance) was also noted in the group diagnosed with possible CAA. The deep and mixed hemorrhage groups resembled the control population in apoE ε4 allele frequency (odds ratio for carriers of ε4=0.98 for deep hemorrhage relative to the control population). Overall, frequency of apoE ε4 was significantly greater in those patients with hemorrhages restricted to the lobar regions (0.31) than in those with hemorrhage present in deep territories (0.13; P<.05).
Patients with definite/probable CAA-related hemorrhage also demonstrated significantly increased frequency of apoE ε2 compared with the control population (Table 2⇑). Carriers of the ε2 allele demonstrated an odds ratio of 2.3 (95% CI, 1.1 to 5.2; P<.05) for definite/probable CAA. ApoE ε2 (and ε3) frequencies in the possible CAA patients resembled the definite/probable CAA group, while the deep and mixed hemorrhage groups again followed the frequencies in the control population.
Among the 39 patients with definite/probable or possible CAA-related hemorrhage, those who carried the ε4 allele had significantly earlier hemorrhages than those without (Table 3⇓). First hemorrhage occurred at a mean age of 73.4 years in carriers of the ε4 allele compared with 78.9 years in noncarriers (P=.033). Among 11 patients with first lobar hemorrhage at age 70 years or older, 9 were carriers of ε4. The effect of the ε4 allele on age at first hemorrhage was again independent of its effect on AD since it remained significant (P<.01) in the subset of patients without history of dementia (data not shown). Only five ε4 homozygotes were present, so that no meaningful comparison with ε4 heterozygotes was possible. No relationship was evident between the presence of apoE ε2 and age of first hemorrhage (Table 3⇓; P=.84).
Our data support a specific role for apoE ε4 in accelerating the process that leads from deposition of amyloid β-peptide to vascular damage and intracerebral hemorrhage. In particular, the overrepresentation of this allele is seen to apply to CAA-related hemorrhage but not to nonlobar hemorrhages. In addition, we find that carriers of apoE ε4 present with CAA-related hemorrhage at an earlier age than those without this allele. These effects appear to be independent of the role of apoE ε4 in AD.
The ε2 allele, reported to protect from AD,17 18 19 appears instead to be modestly overrepresented in patients with CAA-related hemorrhage. This observation was first reported by Nicoll and colleagues,26 who noted an increased frequency of ε2 (although not ε4) in cases of CAA. An association between ε2 and CAA-related hemorrhage is unexpected, particularly in the absence of significant correlation between ε2 and extent of vascular amyloid deposition.7 This finding does not appear to be a result of incorrect diagnosis of CAA since allele frequencies in the subset of 8 patients with pathological evidence for CAA were similar to the total set of definite/probable cases (0.38 for ε4 and 0.19 for ε2). It will be important to look for replication of this finding in other data sets and to assess whether the presence of ε2 (or the absence of ε3) might predispose to CAA-related hemorrhage.
Clinical and radiological evaluation, including routine use of gradient-echo MRI, was found to divide patients into groups with specific patterns of apoE genotype. Those patients with hemorrhages restricted to the cortical/corticosubcortical regions, whether multiple or single, demonstrated an overrepresentation of the ε4 allele not present in the mixed or deep hemorrhage groups. While it is speculative to infer pathophysiology without more pathological data, the similarity in genotype between the patients with possible CAA and definite/probable CAA suggests that a major proportion of isolated lobar hemorrhages in the elderly may indeed be due to CAA. Conversely, the mixed hemorrhages may not be due to CAA since they followed the genotype pattern of the general population. The pathogenesis of this unexpected (but apparently not rare) radiographic picture remains to be clarified by pathological data.
The ability of gradient-echo MRI to distinguish between genotypically distinct groups of patients supports its use in the diagnostic evaluation of lobar hemorrhage. This sensitive technique has previously been shown to identify accompanying petechial hemorrhages in a majority of patients who present with lobar hemorrhages.8
Is there clinical utility for determination of apoE genotype in any individual patient with cerebral hemorrhage? As in AD,10 apoE ε4 is neither necessary nor sufficient for the occurrence of CAA-related hemorrhage; indeed, the great majority of individuals in the general population with ε4 do not clinically manifest CAA. It is possible that determination of genotype might be helpful in evaluating those relatively young (ie, aged <70 years) patients with suspected CAA, since the absence of ε4 in these patients may prove sufficiently unusual to place the diagnosis of CAA in doubt. Another question to be addressed by future research is whether apolipoprotein E genotype might bear on the decision to initiate anticoagulation in the elderly.
Selected Abbreviations and Acronyms
|CAA||=||cerebral amyloid angiopathy|
This study was supported by an American Academy of Neurology research fellowship (Dr Greenberg) and National Institutes of Health grants AG05134 and AG12406 (Dr Hyman). We are grateful to Pamela W. Schaefer, MD, for assistance in interpretation of gradient-echo MRI scans; Joseph Locasio, PhD, for assistance with statistical methods; Jill Kaplan, MD, and Karen Furie, MD, for help in identifying patients with hemorrhage; and G. William Rebeck, PhD, for helpful discussions.
Reprint requests to Steven M. Greenberg, MD, PhD, Vincent-Burnham 811, Massachusetts General Hospital, Boston, MA 02114. E-mail firstname.lastname@example.org.
- Received January 18, 1996.
- Revision received April 23, 1996.
- Accepted May 13, 1996.
- Copyright © 1996 by American Heart Association
Vinters HV. Cerebral amyloid angiopathy. a critical review. Stroke. 1987;18:311-324.
Kase CS. Cerebral amyloid angiopathy. In: Kase CS, Caplan LR, eds. Intracerebral Hemorrhage. Boston, Mass: Butterworth-Heinemann; 1994:179-200.
Greenberg SM, Finklestein SP, Schaefer PW. Petechial hemorrhages accompanying lobar hemorrhages: detection by gradient-echo MRI. Neurology. 1996;46:1751-1754.
Mahley RW. Apolipoprotein E: cholesterol transport protein with expanding role in cell biology. Science. 1988;240:622-630.
Saunders AM, Strittmatter WJ, Schmechel D, St George-Hyslop PH, Pericak-Vance MA, Joo SH, Rosi BL, Gusella JF, Crapper-MacLachlan DR, Alberts MJ, Hulette C, Crain B, Goldgaber D, Roses AD. Association of apolipoprotein E allele epsilon 4 with late-onset familial and sporadic Alzheimer's disease. Neurology. 1993;43:1467-1472.
Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, Pericak-Vance M. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science. 1993;261:921-923.
Corder EH, Saunders AM, Risch NJ, Strittmatter WJ, Schmechel DE, Gaskell PCJ, Rimmler JB, Locke PA, Conneally PM, Schmader KE, Small GW, Roses AD, Haines JL, Pericak-Vance MA. Protective effect of apolipoprotein E type 2 allele for late onset Alzheimer disease. Nat Genet. 1994;7:180-184.
Talbot C, Lendon C, Craddock N, Shears S, Morris JC, Goate A. Protection against Alzheimer's disease with apoE ε2. Lancet. 1994;343:1432-1433.
Schmechel DE, Saunders AM, Strittmatter WJ, Crain BJ, Hulette CM, Joo SH, Pericak VM, Goldgaber D, Roses AD. Increased amyloid beta-peptide deposition in cerebral cortex as a consequence of apolipoprotein E genotype in late-onset Alzheimer disease. Proc Natl Acad Sci U S A.. 1993;90:9649-9653.
Hyman BT, Gomez-Isla T, Briggs M, Chung H, Nichols S, Kohout F, Wallace R. Apolipoprotein E and cognitive change in an elderly population. Ann Neurol. In press.