Prospective Study of Apolipoprotein E Genotype and Functional Outcome Following Ischemic Stroke FREE

ALLELIC VARIATION in apolipoprotein E (APOE for gene; apoE for protein) influences some forms of acute and chronic brain injury.¹ Interest in the effects of APOE polymorphism on the central nervous system arose in 1993 with the finding that the APOE ϵ4 allele is associated with late-onset familial² and sporadic Alzheimer disease.³ The ϵ4 allele seems to determine not if but when an individual will develop the disease.⁴ In addition, possession of the ϵ4 allele is clearly established as a marker of poorer outcome following acute^{5 - 7} and chronic head injury⁸ as well as intracerebral hemorrhage.^{9 - 10} There has been less consistent evidence to demonstrate that the APOE ϵ4 allele is associated with a worse outcome following ischemic stroke.

Animal studies with APOE knockout mice have shown larger infarcts in models of focal cerebral ishemia¹¹ and increased neuronal injury in models of global cerebral ischemia.^{12 - 13} Homozygous APOE ϵ3/ϵ3 and ϵ4/ϵ4 transgenic mice have also been assessed under transient focal cerebral ischemia conditions.¹⁴ Mice with the ϵ3/ϵ3 genotype had significantly smaller infarct volumes and better functional outcome than did ϵ4/ϵ4 mice. These results yield the testable hypothesis that human ϵ4 carriers may have a poorer outcome following ischemic stroke (as occurs in head injury) than non-ϵ4 carriers. However, no adverse effect on long-term survival could be demonstrated in a large retrospective study in patients with ischemic stroke¹⁰ ; instead, a modestly improved long-term survival was associated with increasing ϵ4 allele dose. This study did not, however, examine functional recovery. Survival after ischemic stroke is influenced profoundly by nongenetic factors. Because the transgenic animal data in transient focal cerebral ischemia have demonstrated such functional differences in ϵ3/ϵ3 compared with ϵ4/ϵ4 animals, we sought to determine if APOE genotype influences 1-month or 3-month functional outcome in humans following ischemic stroke.

Consecutive patients were recruited prospectively from an acute stroke unit between January 1997 and May 1998. Ischemic stroke was diagnosed when patients presented with an acute onset of a focal neurological deficit with no evidence of intracerebral hemorrhage or alternative pathologic characteristics on brain imaging. Prior to enrollment, all patients or a relative gave informed consent for the study. Clinical baseline characteristics including National Institutes of Health Stroke Scale (NIHSS) scores¹⁵ and Oxfordshire Community Stroke Project (OCSP) classifications¹⁶ were recorded at the time of admission. The study was approved by the hospital ethics committee.

Apolipoprotein E genotypes were determined from whole blood samples by physicians blind to stroke classification and outcome. Leukocyte DNA was extracted and amplified by a polymerase chain reaction.¹⁷ The product was digested using the restriction enzyme HhaI, separated on a 10% polyacrylamide gel, stained with ethidium bromide, and visualized by UV light.

Participating patients were observed prospectively. Functional outcome was measured at 1 month and 3 months using the Barthel Index (BI)¹⁸ and modified Rankin scale (mRS),^{19 - 20} blind to APOE genotype.

Differences in stroke severity using OCSP classifications and the NIHSS scores among different APOE genotype categories (ϵ3/ϵ3, ϵ2 carrying, and ϵ4 carrying patients) were assessed with a χ² test and Mann-Whitney test, respectively. Outcome categories of the BI and mRS were defined as in the National Institute of Neurological Disorders recombinant tissue plasminogen activator trial²¹ : ie, good (mRS, 0-1; BI 95-100); moderate (mRS, 2-3; BI, 55-90); poor (mRS, 4-5; BI, 0-50); and dead (mRS, 6). Outcome was compared between ϵ4 carriers and ϵ3/ϵ3 patients as well as between ϵ2 carriers and ϵ3/ϵ3 patients using the χ² test. The groups were also examined as ϵ4 and non-ϵ4 carriers with outcome dichotomized as good (mRS, 0, 1, or 2; BI>55) vs all other functional categories. Differences in the distribution of potential prognostic variables (OCSP classifications and NIHSS scores; history of hypertension, stroke or transient ischemic attack, diabetes mellitus, atrial fibrillation, ischemic heart disease, peripheral vascular disease; current cigarette smoker; current infection or hyperlipidemia; use of aspirin, calcium antagonists, angiotensin-converting enzyme inhibitors; age; and possession of the ϵ2 or ϵ4 APOE alleles) were examined in a linear regression analysis using 3-month BI and mRS scores as categorical values. Variables with P<.05 on univariate analysis were then considered in a multivariate analysis by multiple linear regression.

One hundred eighty nine patients were enrolled and APOE genotyped. All had follow-up assessments. There were 89 men and 100 women (mean ± SD age, 69.4 ± 11.0 years; range, 28-93 years). The baseline characteristics, including the OCSP classifications are given in Table 1. The median NIHSS score on admission was 6 (interquartile range, 3-10).

The APOE alleles were in Hardy-Weinberg equilibrium. No patient had the ϵ2/ϵ2 genotype, as indicated on the tabulation below.

Fifty-eight patients (31%) carried 1 or more ϵ4 alleles and 25 (13%) had an ϵ2 allele. There was no significant difference in the age of ϵ3/ϵ3 patients (mean age, 69 years) and ϵ4 (mean age, 68 years; P = .56) or ϵ2 carrying patients (mean age, 73 years; P = .11) at the time of stroke. The APOE allele frequencies were 0.07 for ϵ2 (n = 25 alleles), 0.77 for ϵ3 (n = 292), and 0.16 for ϵ4 (n = 61), a similar frequency distribution to that found in previously studied Scottish populations.²²

There was no statistically significant difference in baseline stroke severity among the proportion of patients in different OCSP categories for ϵ3/ϵ3 vs ϵ2 or ϵ4 carriers (Table 2). Although patients possessing an ϵ4 allele had greater median NIHSS scores on admission to the hospital than ϵ3/ϵ3 carriers, the difference was also not statistically significant (P = .10).

One-month outcome (Figure 1) was not statistically different in ϵ4 carriers and patients with the ϵ3/ϵ3 genotype for either BI (P = .64) or mRS (P = .59) scores. Similarly, there were no significant differences between ϵ2 carriers and ϵ3/ϵ3 patients (mRS, P = .19; BI, P = .87).

The results were unchanged at 3 months (Figure 2). In addition, dichotomizing around end points that signify functional independence (mRS<3 and BI>55) yielded no statistically significant differences between ϵ4 (n = 58) and non-ϵ4 carriers (n = 131) (mRS, P = .37; BI, P = .61) or for ϵ2 carriers compared with non-ϵ2 carriers (mRS, P = .19; BI, P = .32). Excluding ϵ2/ϵ4 heterozygotes (in case of divergent functional effects) from the analyses did not alter any of the results.

Univariate analyses showed that OCSP categories (P<.001), NIHSS scores (P<.001), and age (P = .002) were significantly associated with an adverse 3-month (BI) outcome. Scores for the NIHSS and age remained significant in a multivariate analysis of 3-month BI and mRS outcomes (Table 3).

Although APOE polymorphism has emerged as an important determinant of outcome following head injury^{5 - 8} and intracerebral hemorrhage,^{9 - 10} the findings of this prospective outcome study and a previous retrospective survival study in another population¹⁰ indicate that the ϵ4 allele does not exert a major adverse influence on baseline severity, functional outcome, or survival following ischemic stroke. These contrasting findings from different types of brain injury suggest that the association of the ϵ4 allele with outcome may be insult specific.

Previous case-control and cohort studies have examined APOE genotype as a possible genetic predisposition to ischemic stroke.^{23 - 31} Some studies have implicated the ϵ2 allele,^{24 - 25} some the ϵ4 allele,^{25 - 27} and others have reported neutral findings.^{28 - 31} A recent meta-analysis of case-control studies found a small significant overrepresentation of the ϵ4 allele in ischemic stroke patients compared with age- and sex-matched controls.³² The ϵ4 allele is also overrepresented in coronary heart diseasϵ³³ and is thought to be more atherogenic than other APOE alleles.³⁴ Individuals with the ϵ3/ϵ4 and ϵ4/ϵ4 genotypes seem to carry excess risk compared with control subjects in both coronary heart diseasϵ³³ and ischemic stroke.³² Although the results of the current study demonstrated fewer lacunar infarcts in ϵ4 carriers (29% vs 42%) and higher median NIHSS scores compared with ϵ3/ϵ3 patients (7 vs 5), neither of these was statistically significant, supporting the findings of a larger study that demonstrated no difference in APOE allele frequencies between patients with lacunar and cortical ischemic events.¹⁰

Properties of apoE that are potentially relevant to brain injury (neurotrophic,³⁵ antioxidant,³⁶ immunomodulatory,³⁷ and intracellular calcium effects³⁸ ) have been identified under in vitro conditions. Although some of these are isoform specific (neurotrophic,³⁵ antioxidant,³⁶ and intracellular calcium changes³⁸ ), the relevance in vivo of differing mechanisms has not been clearly defined. Consequently, in head injury and intracerebral hemorrhage (conditions for which possession of the ϵ4 allele seems to almost double the risk of a poor prognosis^{6 ,10} ), it is not known whether ϵ4 carriers incur more severe injury or have defective repair mechanisms compared with non-ϵ4 carriers. In the present study, baseline NIHSS scores and OCSP categories were used to assess stroke severity. Similar findings in each of the major APOE genotype groupings (ϵ3/ϵ3, ϵ4, and ϵ2 carriers) suggest that ischemic insult severity in stroke is not associated with APOE genotype.

The outcome findings of our study question how reliably one can extrapolate data from animal models of stroke to humans.³⁹ Brain apoE changes have been well described in different animal models of brain injury,^{1 ,40} including cerebral ischemia.⁴¹ The results have suggested that injury either causes neuronal expression of apoE or increases neuronal uptake of apoE.⁴¹ This has led to the description of apoE as an "injury factor."⁴² In one focal ischemic model, infarct volume and hemiparesis severity were significantly worse in transgenic ϵ4/ϵ4 mice compared with ϵ3/ϵ3 mice.¹⁴ It remains unclear if APOE genotype influences the size of human infarct volume in different types of ischemic stroke. Tomimoto et al⁴³ have demonstrated that apoE-immunoreactive axons in humans are accompanied by apoE-positive macrophages in the periphery of infarcts, although APOE genotypes were not examined in this small study. However, a recent semiquantitative assessment of neuronal damage following global ischemia in humans failed to demonstrate an APOE genotype influence.⁴⁴

The follow-up functional assessments in our study provided a means, albeit crude, of measuring repair. The neutral results suggest that the different apoE isoforms do not reflect major differences in repair mechanisms following cerebral ischemia in humans. A trend toward better functional outcome in ϵ2 carriers vs either the ϵ3/ϵ3 or ϵ4+ groups was present, but must be interpreted with caution since only 13% of the population (n = 25 patients) possessed an ϵ2 allele. Furthermore, survival and 3-month placement following ischemic stroke in a cohort of 640 patients¹⁰ were not associated with the APOE ϵ2 allele.

Although ischemic stroke is a heterogeneous condition, the number of patients in our study is much greater than in those studies of head injury^{5 - 8} or intracerebral hemorrhage^{9 - 10} (equally heterogeneous conditions) that demonstrated an influence of APOE ϵ4 on outcome. This supports the conclusion that APOE genotype does not contribute to major clinically significant differences in either injury severity or functional outcome in ischemic stroke. Our results indicate that associations between APOE genotype and outcome in humans may be insult specific.

Accepted for publication February 23, 2000.

Mark O. McCarron, MD, MA, MRCP, is supported by a Patrick Berthoud Fellowship, Kent, England.

Reprints: Keith Muir, MD, MRCP, MSc, Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow G51 4TF, Scotland (e-mail: k.muir@clinmed.gla.ac.uk).