Temporal Alterations in Cerebrospinal Fluid Amyloid β-Protein and Apolipoprotein E After Subarachnoid Hemorrhage
Background and Purpose— The mechanism underlying the association between possession of the APOEε4 allele and less favorable outcome after subarachnoid hemorrhage (SAH) remains to be determined. After SAH the level of apolipoprotein E (apoE) in the cerebrospinal fluid (CSF) is decreased, and lower levels are associated with more severe injury and less favorable outcome. This study examined serial CSF samples to determine the time course for the decrease in CSF apoE and the relationship between CSF apoE and amyloid β-protein (Aβ), testing the hypothesis that apoE-Aβ interactions occur in vivo after SAH.
Methods— Enzyme-linked immunosorbent assay was used to assay apoE, Aβ1–40, and Aβ1–42 in serial ventricular CSF samples from 19 patients with SAH and 13 controls. CSF S100B and τ were assayed as surrogate markers of brain injury.
Results— There was a sustained decrease in CSF apoE (P<0.001) and Aβ (P<0.001) after SAH in contrast to the observed elevation in CSF S100B (P<0.001) and τ (P<0.001) concentration. There was significant correlation between CSF Aβ concentration and clinical outcome (r=0.65, P<0.01), and the decrease in CSF Aβ concentration correlated significantly with that of apoE (r=0.85, P<0.0001).
Conclusions— After SAH both apoE and Aβ levels decrease in the CSF, supporting the concept that interactions between these proteins occur in vivo. The possibility that apoE and Aβ influence outcome after SAH warrants further investigation.
Recent evidence indicating that polymorphism of the human APOE gene (ε3/3, ε3/4, ε4/4, ε2/3, ε2/4, ε2/2) influences outcome after subarachnoid hemorrhage (SAH) suggests that apolipoprotein E (apoE indicates protein; APOE, gene) may play an important role in the process of recovery after SAH.1 Several key roles have been identified for apoE in the central nervous system, including the transport of lipid and modulation of inflammatory processes and oxidative stress.2 In addition, in vivo and in vitro studies suggest that apoE promotes the formation of insoluble amyloid aggregate from amyloid β-protein (Aβ) peptides cleaved from amyloid precursor protein (APP), and this process is apoE isoform dependent, with expression of the human APOE ε4 allele being most amyloidogenic.2–4 Intriguingly, recent studies in transgenic mice expressing excess APP identified increased vulnerability to ischemic brain damage, possibly through Aβ-induced disturbance of endothelium-dependent vascular reactivity, although in these studies apoE appeared to be protective.5,6 Clearly, the relevance of these findings to human brain injury is uncertain, although the possibility that interactions between apoE and Aβ may be important after human brain injury such as SAH, in which ischemic injury adversely affects patient outcome, warrants further investigation. In the present study we report temporal alterations in the cerebrospinal fluid (CSF) concentration of apoE and Aβ after SAH, the relationship between maximal changes in CSF apoE and Aβ, and their relationship to injury severity and clinical outcome. We also report the time course for alterations in CSF S100B and τ, utilized as surrogate markers of brain injury severity.7,8
Subjects and Methods
Patients and Control Subjects
The characteristics of the SAH patients investigated have been described previously.9 The control group comprised 13 patients (mean age, 36 years; range, 16 to 61 years) with suspected shunt dysfunction or compensated chronic hydrocephalus with no history of acute brain injury or impaired level of consciousness, who required drainage/examination of ventricular CSF. Only patients with CSF found to have a normal cell count, albumin, total protein concentration, and no xanthochromia were included, so that ventricular CSF parameters were as “normal” as was feasible.
CSF was processed and analyzed for apoE, S100B, total protein, and albumin, as previously described.9 Total τ (Innogenetics), Aβ1–40, and Aβ1–42 (BioSource International) were assayed with commercial kits according to the manufacturers’ instructions.
The data did not have a gaussian distribution, and therefore unbalanced 1-way ANOVA was performed with the use of nonparametric tests (Kruskal-Wallis test with Dunn post hoc test) to compare the median concentration of the analyte at each time point after injury with that of the control group. From the analysis of serial CSF samples, the maximal change (abbreviated as peak for proteins that increase after SAH and trough for proteins that decrease after SAH) from the baseline value was determined in each patient and for each protein. Rank correlation and linear Spearman regression analyses were used to determine the relationship between the maximal change in CSF protein concentration and injury severity (assessed with the Glasgow Coma Scale), clinical outcome (assessed with the Glasgow Outcome Scale), and time after injury. The Mann-Whitney test was used to compare protein concentrations in CSF from patient groups dichotomized according to injury severity or clinical outcome. Probability values of <0.05 were considered significant. Statistical analysis was performed with the use of GraphPad Prism software.
CSF ApoE and Aβ After SAH
On days 3 and 4 after SAH, the concentration of apoE was significantly lower than that of the control group but thereafter increased toward control values. The decrease in CSF apoE observed in SAH CSF occurred despite the release of plasma proteins into the CSF. These data are summarized in the Table. CSF Aβ1–40 and Aβ1–42 concentrations were significantly lower in SAH CSF (Figure 1) compared with controls for all the time points investigated. In SAH CSF there was significant correlation (r=0.85, P<0.0001) of Aβ1–40trough and apoEtrough (Figure 2). The maximum decrease in Aβ1–42 concentration correlated with clinical outcome (Spearman r=−0.60, P=0.01) 3 months after SAH. Patients with unfavorable outcome after SAH had significantly lower CSF Aβ1–42 concentration (P=0.05, Mann-Whitney) than those with favorable outcome.
CSF S100B and τ After SAH
The concentration of S100B in SAH CSF was significantly elevated compared with controls for all the time points investigated (Table). Similarly, the concentration (±SD) of τ in control CSF (0.19±0.2 μg/L) was significantly lower than that in SAH CSF for all time points investigated (day 2, 3.4±4.4; day 6, 2.7±2.9; day 10, 2.9±1.9 μg/L). The peak concentration of τ in SAH CSF correlated significantly (Spearman r=−0.69, P=0.002) with Glasgow Coma Scale score. The concentration of τ in the CSF of patients in coma after SAH was significantly (P=0.01, Mann-Whitney) higher than that of patients with a Glasgow Coma Scale score >8. There was significant correlation between peak S100B (Spearman r=−0.60, P=0.02) and τ (Spearman r=−0.63, P=0.02) and Glasgow Outcome Scale score 3 months after SAH. The peak τ concentration in SAH CSF of patients with unfavorable outcome was significantly (P=0.007, Mann-Whitney) higher than that of those with favorable outcome, but this was not the case for S100B.
There was no significant difference between the concentration of the proteins assayed in brain injury CSF, injury severity, and clinical outcome according to APOE genotype. Within the small number of patients in each group, there was no significant difference in APOE genotype frequency and no correlation with injury severity or clinical outcome.
From the analysis of serial CSF samples, we found that, compared with controls, apoE, Aβ1–40, and Aβ1–42 levels are persistently lower in CSF during the acute phase after SAH. Furthermore, the decrease in CSF Aβ1–40 correlated with the decrease in apoE, and the decrease in Aβ1–42 correlated with clinical outcome. In contrast, there was a marked and persistent elevation in CSF S100B and τ levels in SAH CSF.
Previously, we reported preliminary findings associating low CSF apoE concentration with more severe injury and less favorable outcome after SAH.9 We speculated that possible mechanisms for the depletion of apoE from SAH CSF include increased receptor-mediated uptake of apoE containing lipoproteins and/or interactions between apoE and other hydrophobic molecules such as Aβ. The findings in the present study provide indirect evidence for interactions between Aβ and apoE in vivo after SAH. While the finding that CSF Aβ decreases after SAH requires further evaluation, there are some parallels with findings in the CSF of patients with Alzheimer disease, in which reduced concentrations of Aβ in the CSF are attributed to increased deposition within the brain parenchyma as amyloid plaque.10 Intriguingly, the concentration of Aβ in the CSF of patients with ischemic stroke, outcome from which is not influenced by APOE, is not significantly different from Aβ levels in control CSF.11
The observation that outcome after SAH was less favorable in patients with low CSF Aβ is of interest because the possibility that Aβ peptides participate in processes triggered by SAH has not been extensively investigated. Importantly, apoE isoform–dependent properties have been identified for Aβ peptides, which may be relevant after SAH. First, vasoactive properties have been identified for Aβ peptides, and promotion of endothelin-1–mediated vasospasm by Aβ appears to be apoE isoform dependent.12 Second, ischemic susceptibility of transgenic APP mice appears to be dependent on apoE isoform, possibly by modulation of the inflammatory response induced by middle cerebral artery occlusion.6 Furthermore, Aβ appears to diffuse freely within neural tissue, raising the possibility that Aβ cleared via the cerebral vasculature may become deposited within smooth muscle, as in cerebral amyloid angiopathy, resulting in intimal narrowing and alterations in cerebral blood flow.13,14 Finally, Aβ peptide induces free radical–mediated oxidative stress, and neuronal vulnerability appears to be APOE dependent.15
The decrease in apoE and Aβ concentrations in the CSF after acute brain injury is in contrast to increased CSF concentrations of S100B and τ. S100B is a nonspecific protein marker, the concentration of which has been correlated with injury severity and outcome after several types of acute brain injury.7 The high τ concentration in CSF from SAH patients is consistent with τ being a nonspecific marker of axonal injury.8 Although the maximum CSF τ concentration after SAH occurred later than that of S100B, there was correlation with both injury severity and clinical outcome, suggesting that τ may have a role as a biochemical marker for stratification of injury severity and prediction of outcome.
Clearly, this preliminary study is limited by difficulties in obtaining CSF from substantial numbers of SAH and control subjects, which would be necessary to detect effects attributable to APOE genotype and marker protein concentration. Furthermore, intraventricular drainage of CSF tends to be performed in comatose SAH patients who have less favorable outcomes, limiting the power to detect the relationship between CSF marker protein, injury severity, and clinical outcome. Assaying the concentration of a substance in the ventricular CSF of healthy normal controls is not ethical, limiting access to CSF available from patients investigated for possible shunt dysfunction or hydrocephalus. These patients had no clinical or CSF evidence to suggest acute brain injury, and no evidence for shunt dysfunction or raised intracranial pressure was identified.
In summary, we report substantial decreases in the CSF concentration of apoE and Aβ after SAH, supporting the concept that apoE-Aβ interactions occur in vivo after SAH. These data merit further investigation to determine whether apoE-Aβ aggregates are deposited acutely in the neuropil after SAH or become trapped in perivascular channels or blood vessel walls, resulting in local injury. Increased understanding of the roles played by apoE and Aβ after SAH may extend the potential clinical application of treatment modalities, targeting these proteins for the prevention and treatment of chronic neurodegenerative disorders such as Alzheimer disease.
This study was supported by a Wellcome Trust clinical research training fellowship (A.K.) and by the Medical Research Council (J.N.). We thank the patients and their relatives, the medical, nursing, laboratory, and research colleagues, and the hospital ethics committee for their support with this study.
- Received May 2, 2003.
- Revision received June 19, 2003.
- Accepted July 30, 2003.
Niskakangas T, Ohman J, Niemela M, Ilveskoski E, Kunnas TA, Karhunen PJ. Association of apolipoprotein e polymorphism with outcome after aneurysmal subarachnoid hemorrhage: a preliminary study. Stroke. 2001; 32: 1181–1184.
Holtzman DM, Bales KR, Tenkova T, Fagan AM, Parsadanian M, Sartorius LJ, Mackey B, Olney J, McKeel D, Wozniak D, Paul SM. Apolipoprotein E isoform-dependent amyloid deposition and neuritic degeneration in a mouse model of Alzheimer’s disease. Proc Natl Acad Sci U. 2000; 97: 2892–2897.
Zhang F, Eckman C, Younkin S, Hsiao KK, Iadecola C. Increased susceptibility to ischemic brain damage in transgenic mice overexpressing the amyloid precursor protein. J Neurosci. 1997; 17: 7655–7661.
Koistinaho M, Kettunen MI, Holtzman DM, Kauppinen RA, Higgins LS, Koistinaho J. Expression of human apolipoprotein E downregulates amyloid precursor protein-induced ischemic susceptibility. Stroke. 2002; 33: 1905–1910.
Kay A, Petzold A, Kerr M, Keir G, Thompson E, Nicoll J. Decreased cerebrospinal fluid apolipoprotein E after subarachnoid hemorrhage: correlation with injury severity and clinical outcome. Stroke. 2003; 34: 637–642.