Erythropoietin Therapy for Subarachnoid Hemorrhage

BACKGROUND:

1. Delayed ischaemia deficits in subarachnoid haemorrhage

Seven thousand patients suffer SAH each year within the UK with young adults (<55 years) being equally affected. Cerebral vasospasm and related cerebral ischemia are the major causes of delayed morbidity and mortality in patients who survive the initial SAH episode. Previous studies have revealed that immediate low cerebral blood flow (CBF) accompanies SAH, particularly in those in coma. However, for better grade patients the low CBF is often delayed and associated with cerebral vasospasm [Knuckey 1985]. Vasospasm often precedes the onset of DID which commonly occurs in the 2nd week after SAH [Knuckey 1985]. Autoregulation is the most accomplished reflex mechanism and reflects an intrinsic ability of the cerebral vessels to dilate in an attempt to maintain constant CBF in the face of poor cerebral perfusion. Following SAH, impairment of autoregulation affects prognosis [Lam 2000]. Indeed, daily assessment of autoregulation can help to identify patients at high risk of clinical deterioration [Smielewski 1997].

Autoregulation can be conveniently measured non-invasively from the bedside using transcranial Doppler (TCD). Direct measures of middle cerebral artery flow velocities (FV) are predictive of DID; elevated mean FV (>120cms/sec) and increased ratio of mFV between MCA and extracranial internal carotid artery (ICA) (Lindegaard ratio >3) being associated with cerebral vasospasm and higher incidence of adverse events [Lindegaard 1988]. The rapid increase of FV with time is also predictive, as is the relationship between FV and spontaneous changes in blood pressure (correlation index Mx) [Czosnyka 2000].

A more sophisticated measure of abnormal cerebrovascular haemodynamics involves the transient hyperaemic response test (THRT) [Giller 1991]. This is used routinely in our department where we have demonstrated a relationship between an abnormal THRT and clinical outcome following SAH [Lam 2000]. The THRT assesses the FV response to a manual carotid compression (Figure 1) and can be repeated within a few minutes providing a reliable and readily accessible method for assessing autoregulation. When THRT is used for day-to-day assessments for the same patient, this test can provide a valuable monitor of sequential changes in autoregulation.

2. Autoregulation after SAH

Autoregulation is often impaired after SAH, even in patients presenting in good clinical grade [Voldby 1985]. An abnormal THRT response commonly develops either early during days 0 to 3 (primary), or late around days 7 to 14 (secondary) after the initial bleed. Primary abnormalities in THRT are far more common in poor grade patients, while secondary abnormalities occur more regularly in the initially good grade patients who later develop signs of DID with clinical deterioration [Ratsep 2002]. Routine and daily assessment of autoregulation with TCD helps to identify patients at higher risk for DID. This delay in the evolution of an abnormal autoregulation response in better grade patients presents an opportunity to introduce early therapeutic strategies designed to offset the potentially harmful effects. Indeed, routine use of plasma expansion is already adopted with this aim in mind [Origitano 1990].

3. Therapeutic opportunity; why recombinant human erythropoietin?

Recently attention has been focused on potential therapeutic roles for endogenous brain proteins possessing neuroprotective properties. This neuroprotective approach of using erythropoietin (EPO) is aimed at protecting potentially viable brain tissue from apoptosis [Ehrenreich 2002]. EPO, a 34-kDa hydrophobic sialoglycoprotein, is responsible for the survival, proliferation, and differentiation of committed erythroid progenitor cells. It is produced by foetal liver and adult kidney, which is accelerated during hypoxia [Lacombe 1999]. Because there is no pre-formed EPO storage, therefore, the control of EPO gene transcription involves complex interactions with hypoxia-inducible factor 1 (HIF-1), a transcription factor which is activated during hypoxia, hypoglycaemia, and seizures [Lacombe 1999].

EPO receptor (EPO-R), like other members of the haematopoietic receptor family, does not possess an endogenous tyrosine kinase activity. However, its close association with Jak2 tyrosine kinase can induce a rapid tyrosine phosphorylation on a number of proteins [Lacombe 1999]. Hypoxia can induce an increased EPO-R expression, which in turn causes increased sensitivity to EPO [Chin 2000].

Both EPO and EPO-R are ubiquitously distributed in neural tissues [Digicaylioglu 1995]. Except larger vessels, EPO-R is abundantly distributed in astrocytes, neurons, and cerebral capillary endothelia [Marti 1996, Yamaji 1996, Bernaudin 1999, Brines 2000]. Immunohistochemistry shows that these distributions of EPO-R are consistent with the anatomical location of blood-brain barrier (BBB) [Brines 2000]. Such a condensed distribution of EPO-R within and surrounding the cerebral capillaries implies that a small amount of EPO is sufficient to mediate physiological function in the central nervous system [Chin 2000]. The increased production of EPO in response to hypoxia and anemia by astrocytes and neurons corresponds to that which occurs in kidney and liver [Masuda 1994]. Thus, during hypoxia, EPO acts directly on cerebral capillary endothelial cells as an angiogenic factor through a paracrine mechanism from the astrocytes and an endocrine mechanism from the kidney [Digicaylioglu 1995].

EPO protects nervous tissue under any conditions characterised by a relative deficiency of ATP [Brine 2000]. This neuroprotective effect is achieved by several mechanisms.

1. EPO can induce expression of several glycolytic enzymes, which re-directs energy metabolism toward favouring survival during oxygen deprivation [Digicaylioglu 1995].

2. During cerebral ischemia, the N-methyl-D-aspartate (NMDA) receptor-mediated glutamate toxicity is the major cause of neuron death [Morishita 1997]. Glutamate toxicity is in part mediated by nitric oxide (NO). EPO up-regulates the expression of antioxidant enzymes on a transcriptional level in vascular smooth muscle cells [Kusano 1999], and can potentially protect neurons from NO-generating agents [Sakanaka 1998]. Furthermore, EPO can suppress the increase in Ca2+ concentration induced by glutamate and the nitric oxide (NO)-induced apoptosis [Morishita 1997].

EPO has been shown to provide neurotrophic [Konishi 1993] and angiogenic effects [Yamaji 1996], and it has the potential for neural regeneration [Siren 2001]. These effects are dose-dependent and unrelated to the nerve growth factor (NGF) [Konoshi Y 1993]. The distributions of EPO and EPO-R correspond to the sequence of histopathological changes of cerebral ischemia/infarction. In acute hypoxia (cardiac arrest), strong immunoreactivity of EPO can be found in vascular endothelium, while the EPOR is expressed strongly in neurons [Siren 2001]. After cerebral ischemia, EPO plays an important role in angiogenesis and glial reaction. EPO immunoreactivity appears in the endothelium, intravascular inflammatory cells and neurons of the penumbra. Also EPO-R is expressed in neurons, astrocytes and endothelial cells [Bernaudin 1999, Siren 2001]. Because the expression of EPO-R precedes the up-regulation of EPO synthesis [Bernaudin 1999], the sensitivity to EPO increases [Chin 2000]. In contrast, these changes are not detectable in non-ischemic cortex [Bernaudin 1999], indicating an increased potential for EPO to act where neurons are at risk [Siren 2001]. By protecting endothelial cells from apoptosis, EPO can increase cerebral blood flow, improve microcirculation of the penumbra, and improve tissue oxygenation by angiogenesis, thus significantly reduce the infarct volume [Bernaudin 1999, Siren 2001]. This angiogenesis corresponds to the finding that under hypoxic conditions such as exposure to high altitudes, the brain increases the mean microvascular density [Lamanna JC 1992]. In old infarcts, EPO and EPO-R are strongly expressed in reactive astrocytes. Thus the EPO system may participate in the repair process [Siren 2001]. The persistent up-regulation of EPO production by astrocytes after stroke also provides rapidly available sources of EPO to make neurons more tolerate ischemia [Siren 2001]. These endogenous neuroprotective mechanisms are insufficient to cope with acute injuries, as severe ischemic insult can exhaust the EPO/EPO-R neuroprotective capacity of the brain or the latency of de novo synthesis is too long. Clinical application of EPO should be administered rapidly to provide additional neuroprotection [Brines 2000].

Epoetin beta is a 165-aa glycoprotein manufactured by using recombinant DNA technology, which contains the identical amino acid sequence of isolated human EPO and possesses the same biological activity. It is similar to the endogenous EPO except for minor differences in the pattern of 4 carbohydrate chains. [Brine 2000] The recombinant human EPO has been widely used for treating anaemia associated with chronic renal insufficiency, HIV infection, cancer, and surgery, and has an excellent safety record in wide clinical practice.

4. Why 90,000 IU Epoetin beta?

The safety and efficacy of epoetin beta for treating acute stroke (within 8 hours) in humans have been confirmed in a double blind, randomized, placebo-controlled clinical trial, which uses high dose intravenous epoetin beta (3,3000 IU/50 ml/30min/day, three days, with a cumulative dose 100,000IU) [Ehrenreich 2002]. No associated changes in blood pressure, haematocrit, haemoglobin, and red blood cell count are found in this trial [Ehrenreich 2002]. Only the reticulocyte count increases 29.2 ±10.0% on the 4th day after the last dose of EPO [Ehrenreich 2002]. This systemic administration of EPO also has led to a 60- to 100-fold increase in CSF levels and does not require a breakdown of the blood-brain barrier (BBB) [Ehrenreich 2002]. Patients who are treated with EPO have a lower and earlier surge of serum S100 (secreted by reactive astrocytes and functional disturbances of membrane integrity of brain cell membrane) and thereafter return to within the normal limits sooner than the placebo [Ehrenreich 2002]. This earlier return of serum S100 towards the normal level implies that the inflammatory processes and/or BBB integrity can be improved more rapidly by systemic EPO therapy [Ehrenreich 2002]. The beneficial effects on stroke patients continue throughout 30 days, particularly for those with moderate to severe strokes [Ehrenreich 2002].

By activation of endothelial EPO receptors, the circulating EPO may appose inflammatory pathways in cerebral arteries and attenuate vasospasm induced by SAH. In vivo studies have found that systemic administration of EPO (i.p. 1000units/kg) immediately after SAH can improve survival and motor functions [Buemi 2000]. The systemic delivery of EPO has the advantage that it is universally available to the capillary endothelium in the whole brain, in contrast to intraventricular injection, which is highly localized and not practical in clinical settings [Brines 2000]. Because the concentration of EPO in CSF is independent of the serum level and the extent of BBB disruption in patients suffering from aneurysmal SAH, the EPO in CSF is predominantly produced by the brain in order to match its need [Springborg 2003]. Therefore, an early activation of endothelial EPO-R may be a potential therapeutic strategy, which in fact has been proved by a reduction of S100 protein in CSF (a marker of brain damage) and restoration of cerebral autoregulation (for at least 48 hours) after systemic administration of recombinant human EPO [Springborg 2002]. Therefore, 90,000 Iuof epoetin beta will be use to achieve the maximal effects on CBF and autoregulation in these SAH patients to be studied.

QUESTIONS TO BE ANSWERED:

We wish to address the following hypotheses for SAH patients:

1. Systemic treatment with epoetin beta (90,000IU) is safe following SAH

2. Systemic treatment with epoetin beta reduces the incidence of abnormal FV and cerebral autoregulation following SAH.

3. Systemic treatment with epoetin beta reduces the incidence of acute neurological deterioration following SAH If primary endpoints are improved with epoetin beta therapy, the data will be used to formulate the power calculation for design of a Phase III clinical outcome study. However, we will assess clinical outcome at the end of the trial and at discharge to detect any indication of adverse effect after EPO therapy.

Randomisation procedure:

Following informed consent, patients with angiography-positive aneurismal SAH will be randomised to receive either intravenous epoetin beta 30,000IU or placebo (0.9% saline) 50ml/30min, three times in the first week after SAH (total dose 90,000IU). The number in each group will be 40. For blinding, the Pharmacy Manufacture Unit (PMU) will prepare and number identical vials containing either saline (0.9% NaCl) or epoetin beta reconstituted in saline. The vials will be randomly assigned to patients upon enrollment with the contents of each vial known only by the PMU. Trial medication will be started as soon as possible within 72 hours of the ictus. As approximately 70% of aneurysms will be treated with open clipping, and the remainder with endovascular coils, we do not consider the method of treatment to represent a contaminating factor, but it will be included in the final analysis. Location, size, and morphology of the culprit aneurysm are not believed to affect outcome in our institution.

Following randomisation and start of trial therapy the clinical management of each patient will be as routine. Arterial blood pressure will be continuously monitored (Finapress, or via radial arterial line).

Safety:

The full blood cell count, reticulocyte count, blood viscosity, coagulation profile, serum biochemistry, serum iron levels, and C-reactive protein (CRP) at the time of admission will be checked as clinical routine. Although EPO has effects of erythropoiesis and thrombopoiesis, associated deterioration or adverse events have not been observed in short-term treatment [Ehrenreich 2002]. However, in the face of any abnormalities the trial drug will be stopped and the safety committee informed. A safety committee (chaired by Dr Ken Smith, Consultant nephrologist) will review the safety data at monthly intervals or, if concerns arise, on a patient-by-patient basis.

Primary outcome measures: Vasospasm and abnormalities in AR:

Trial patients will be examined daily with TCD (DWL, Germany) using a 2-MHz probe mounted on a purposed head frame for two weeks since SAH ictus. The systolic, diastolic, and mean FV will be recorded (trans-temporal) by a single user (MT). Vasospasm will be defined as mean FV > 120 cm/sec and Lindegaard ratio >3 [Lindegaard 1988]. The regression index (Mx) between mean FV and spontaneous changes in ABP will be calculated. Two carotid compressions lasting 5 seconds will be performed. The criteria for an acceptable THRT includes a sudden decrease in middle cerebral artery FV at the onset of compression, a stable TCD signal during compression, and a minimum of 30% decrease in FV with no blood pressure instability [Smielewski 1997]. The THRT ratio (THRR) is calculated using the formula: THRR = FVs (hyperaemia) / FVs(baseline), where FVs denotes systolic FV (Figure 1). THRR is classified as normal (≥l.10) or impaired (<1.10), and will be repeated 2 minutes later. The average value of the two tests will be recorded. Quality issues concerning the THRT response have been extensively evaluated in this laboratory [Smielewski 1997].

Secondary endpoint measures: Development of DID:

The clinical progress of each patient will be monitored daily. The development of a focal neurological deficit and/or a drop in the Glasgow Coma Scale by 2 points or more following vasospasm will be the criteria adopted to define an episode of DID [Pickard 1989]. Other factors which can affect consciousness, including systemic infection and infection of central nervous system, will be recorded for final analysis. Clinical outcomes will be assessed at the end of the trial and at the time of discharge. Durations of hospitalisation and NCCU stay will be observed.

Statistical Analysis:

The analysis will be performed by using statistical software, STATA (Texas, USA). Data will be expressed as mean±standard deviation. The multivariate analysis will be performed on the timings of vasospasm and abnormal AR, which will be computed as the number of days between the date of diagnosis of SAH and the date of the first detection of vasospasm and abnormal autoregulation in respect of variables likely to influence their occurrence: age, WFNS grade, Fisher's grade on CT scans, and treatment procedures, intraventricular haemorrhage and/or hydrocephalus by using the Cox Proportional Hazards regression. The measure of outcome will be the hazards ratio (HR), which expresses the probability of vasospasm or abnormal autoregulation for a specific category relative to the placebo group. Variables associated with EPO treatment (haematocrits, erythrocyte counts, reticulocyte counts, thrombocyte counts, serum iron levels, and viscosity) will be collected. The improvement in fit due to each variable will be tested for statistical significance at the 5% level with the likelihood ratio test. The test for departure from a linear trend (one degree of freedom) was used to assess the potential linear trend in certain categorical variables. Daily results from each patient will also be averaged to produce a patient-oriented database to satisfy the independence assumption for linear regression. The day-to-day variations in parameters will be calculated as the standard deviation divided by the average value from all examinations in each patient.

Power analysis:

Our previous observations indicate that vasospasm and abnormal THRT occur in 64% and 90% SAH of patients respectively. To demonstrate a 50% reduction in duration of vasospasm or abnormal autoregulation with a power of 80% and a significance level of 5% we will require examinations from 80 patients. This number of patients also has an 83% chance of demonstrating a 50% reduction in the incidence of clinical DID in the first 2 weeks following the ictus. The study is not powered to show a reduction in clinical endpoints although trends will be sought to identify any potential adverse effects.

DETAILS OF ANY DIFFICULTIES THAT CAN BE FORESEEN

1. Patient recruitment:

The Neurosurgical Unit at Addenbrookes treats 80 aneurysmal SAH patients per annum. Previous studies in our unit have shown high ascertainment, and we were the largest contributor to the International Hypothermia for Aneurysm Surgery Trial (IHAST). We do not foresee difficulties with recruitment.

2. Safety issues:

The safety and efficacy of using high dose intravenous epoetin beta (100,000IU) for treating human acute stroke (within 8 hours) have been confirmed [Ehrenreich 2002]. The therapeutic margin of EPO is very wide. Even at very high serum levels no adverse effects have been observed [NeoRecormon information sheet, Roche].

3. Assessment of AR using TCD:

Quality control for THRT assessment has been vigorously addressed, failure to complete the test is rare [Ratsep 2002]. The incidence of significant carotid atherosclerotic disease precluding safe carotid compression is less than 1% [Lam 2000]. The test is well tolerated and we know of no patient who requested to withdraw from the investigation in previous studies due to discomfort etc.

FUTURE RESEARCH

Results from this phase II study will be used to design a phase III clinical outcome trial using the Spontaneous Intracranial Haemorrhage Group mechanism. Any potential benefits may extrapolate to other cerebrovascular conditions where low CBF states and loss of autoregulation occur (e.g. acute cerebral vasculitis).