Optimising Clinical Trial Design for Proof of Neuroprotection in Acute Ischaemic Stroke: The SAINT Clinical Trial Programme

The development of acute ischaemic stroke therapy has proven to be a daunting task. Important advances to date have included proof of the benefits of organised care in stroke units [1, 2], recognition of the modest benefits of aspirin in acute ischaemic stroke [3,4,5], and approval of recombinant tissue plasminogen activator (rt-PA) for acute ischaemic stroke by the US Food and Drug Administration (FDA) [6]. Other agents aimed at reperfusion following acute ischaemic stroke, such as intra-arterial prourokinase [7], intravenous ancrod [8] and abciximab [9] have shown promise but have not received marketing approval. Another thrombolytic agent, desmoteplase, is undergoing further trials after a positive proof of concept study [10]. Most recently, the Merci Retriever system was approved by the FDA for clot removal from the intracranial vasculature in ischaemic stroke, although not yet proven in pivotal trials [11].

Over 50 neuroprotective agents have also been tested in randomised, controlled trials of acute ischaemic stroke [12], but none have demonstrated unequivocal efficacy. The Stroke Therapy Academic Industry Roundtable (STAIR) has developed recommendations for the preclinical and clinical evaluation of acute ischaemic stroke therapies to optimise the potential for successful translation from animal models to human stroke [13,14,15,16]. One of the most promising neuroprotective agents developed to date, NXY-059, is being evaluated in the Stroke Acute Ischaemic NXY-059 Treatment (SAINT) clinical trial programme, which was designed in accordance with the STAIR criteria.

Pharmacological therapy in acute ischaemic stroke can be divided into two broad classes: restoring cerebral blood flow (reperfusion) and therapies aimed at modifying components of the neurotoxic cascade that follows ischaemia (neuroprotection). Cerebral ischaemia triggers the pathological pathways of the ischaemic cascade that cause cell death in the ischaemic core and neuronal cell injury in the ischaemic penumbra [17]. By interfering with this ischaemic cascade, neuroprotective strategies aim to preserve viable brain cells in the ischaemic penumbra, thereby aiding recovery from stroke [17, 18]. The target for both reperfusion and neuroprotective strategies is the salvage of ischaemic penumbra, with limitation of infarct volume and hence improved outcomes.

Approaches to neuroprotection have included modulation of calcium, sodium and potassium channels, glutamate release/N-methyl-d-aspartate (NMDA) receptor antagonists, free radical scavengers, nitric oxide-related intervention, anti-inflammatory agents, growth factors, gamma-aminobutyric acid (GABA) agonists and serotonin (5-HT_1A) receptor agonists [19].

Although a broad range of neuroprotective strategies has been shown to reduce infarct size and improve functional outcome measures in animal stroke models [20], none have proven unequivocally beneficial in Phase III efficacy trials. This failed translation of neuroprotective agents is summarised in table 1[19]. Most recently, clinical development of the 5-HT_1A receptor agonist repinotan for the treatment of acute ischaemic stroke was halted following the disappointing results of a Phase IIb study [21], while a Phase II study of the astrocyte-modulating agent ONO-2506 in acute ischaemic stroke was terminated in the US following an unfavourable interim analysis [22].

The failure of neuroprotective agents in Phase III efficacy trials has been attributed to multiple factors relating to inadequate preclinical data, unfavourable compound characteristics and poor clinical trial design [23]. The animal models of stroke used to evaluate neuroprotective agents have usually involved young healthy rodents, rather than elderly animals that are more representative of typical human stroke populations that may suffer from multiple co-morbid conditions. Moreover, in contrast to the human brain, the rodent brain contains only a small amount of white matter and it is likely that the mechanisms of ischaemic damage differ between white and grey matter [24]. The heterogeneity and complexity of human stroke compared with animal infarct models is likely to be another factor [25].

Poor design of the clinical Phase III trials of neuroprotective agents may also have contributed to the lack of evidence for the efficacy of many novel compounds that have been investigated [23]. The only therapy proven to be effective in acute ischaemic stroke, rt-PA, has a treatment window of just 3 h [6]. In contrast, most previous trials of neuroprotective agents have allowed enrolment 6 to 8 h or more following stroke onset, despite preclinical evidence of much shorter time windows [15]. Many trials have enrolled patients with stroke severities that render them uninformative with regard to treatment effect. For example, patients with mild stroke are likely to do well, while patients with severe stroke are likely to do poorly, irrespective of intervention. Additionally, most studies of neuroprotective agents have been underpowered and so were unable to detect any modest, yet clinically significant benefits of the study agent [26]. This is particularly important given that the benefits of neuroprotection are likely to be considerably smaller than those observed in trials of reperfusion [27], so that sample size is an important issue.

In some trials, study agents have been tested at doses below those shown to be neuroprotective in animal models because of dose-limiting side effects, such as psychoactive effects, QT prolongation or liver function test abnormalities [28]. Two large Phase III trials of the competitive NMDA antagonist selfotel were terminated prematurely because of an apparent lack of clinical efficacy [29]. In these trials, selfotel was administered at lower doses than those required to achieve neuroprotective plasma concentrations in animal stroke models, because of adverse effects. Finally, the use of outcome measures not sufficiently sensitive to assess modest but clinically relevant differences in functional outcome may also have contributed to the large number of trial failures [30].

The preclinical STAIR recommendations for the evaluation of neuroprotective agents (table 2) [14] emphasise the importance of rigorously testing these agents in preclinical stroke models designed to investigate dose ranges, the time window of treatment effects, histological and functional outcomes, sustained treatment effects and safety, before progressing to Phase III efficacy trials. It is also recommended that first-in-class agents be investigated in larger species/primate models, after demonstration of a biological action in rodents. These recommendations were extended in 2003 to include the extent of histological protection, subcortical protection, white matter protection, the length of treatment and efficacy as monotherapy [31].

The clinical guidelines developed by the STAIR advocate that neuroprotective agents be administered to patients using analogous conditions shown to produce significant histological and functional protection in animals [15]. Hence, patients should be enrolled as soon as possible after stroke onset to match the time window proven in animal models. The doses used in humans should produce plasma levels shown to be neuroprotective in animal models. Limiting enrolment to patients with moderate baseline deficits (e.g. National Institutes of Health Stroke Scale [NIHSS] scores of 7–22) would be expected to enhance the probability of detecting a clinical benefit, given the high placebo response in patients with mild stroke and the greater difficulty in attaining improvement in those severely affected. Additionally, a meaningful and reliable outcome measure should be selected. A global statistical approach is recommended, whereby the treatment effect is measured using multiple pre-defined outcome measures. As discussed previously, the trial should be powered with an adequate sample size to detect any small, yet clinically significant, benefits of the study agent [32].

NXY-059 is a novel free radical-trapping neuroprotectant that acts at several points in the ischaemic cascade. It has been shown to be neuroprotective in rat models of transient and permanent focal cerebral ischaemia [33,34,35] and to reduce infarct volume and improve functional recovery in a primate model of permanent focal ischaemia [36]. NXY-059 is the first compound to have progressed to Phase III clinical efficacy trials on the basis of having met all the benchmark STAIR preclinical criteria (table 2).

An initial clinical study demonstrated that NXY-059 was well tolerated by stroke patients at plasma concentrations that provided neuroprotection in a rat model of transient focal ischaemia (8 to 40 µmol/l) [37]. A further Phase IIa study investigated the tolerability of NXY-059 at the higher plasma concentrations (50 to 150 µmol/l) needed to achieve neuroprotection in the rat model of permanent focal ischaemia [38]. In this double-blind, dose-escalation study, 135 patients with acute stroke were randomised to NXY-059 at a loading dose of 915 mg over 1 h followed by 420 mg/h for 71 h, NXY-059 at 1,820 mg for 1 h followed by 844 mg/h for 71 h, or placebo. NXY-059 was well tolerated at, and above, concentrations shown to be neuroprotective in the rat model. The unbound NXY-059 plasma concentration at steady state in the high-dose group (260 µmol/l) exceeded the concentration needed to achieve neuroprotection in the rat model (fig. 1) and in a subsequent primate model of permanent focal ischaemia [36].

The efficacy and safety of NXY-059 in patients with acute ischaemic stroke is being evaluated in the SAINT clinical trial programme. The trial programme comprises two large double-blind, randomised, placebo-controlled, multi-centre Phase III studies, SAINT I and SAINT II. SAINT I was conducted at more than 150 centres and enrolled approximately 1,700 patients in Europe, Australia, South Africa and Asia [39]. SAINT II is also an international trial, with an analogous protocol and has increased the sample size to 3,200 patients.

The two trials were designed in accordance with the STAIR criteria (table 3) to assess the efficacy of intravenous NXY-059 on disability and neurological recovery and to evaluate safety when administered within 6 hours of stroke onset [40]. Acute ischaemic stroke patients with a NIHSS score of ≥6 and limb weakness are eligible for study inclusion. Given that animal model data have shown that NXY-059 provides significant neuroprotective effects at 4 to 5 h, with trends at 6 h [33, 34, 36], the SAINT clinical trials were designed to support regulatory approval for a 6-h treatment window, thereby providing treatment for a broad patient group. To enhance the probability of success for the trial programme, a novel approach is used to obtain a mean treatment window <4 h (as used in rat and primate models of permanent focal ischaemia [34, 36]) while ensuring a distribution of patients up to 6 h. Forced stratification is used to ensure that the average time until the start of study drug is <4 h at each centre. Hence, at each site, the first randomised patient is required to have a time to start of study drug of <4 h; the target time for each subsequent patient is supplied by an interactive voice response system to maintain an average time of <4 h.

The dosing regimen (2,270 mg for 1 h followed by 960 mg/h for 71 h) was selected on the basis of the results of the Phase IIb study discussed earlier [38]. The primary outcome measure is the modified Rankin scale (mRS) [41] at 90 days or last rating, analysed across the whole distribution of scores, adjusted for stratification variables. It was prespecified that the analysis of the efficacy outcomes was ordered hierarchically and formal statistical testing was performed only if the preceding endpoint was significant. The highest ranked endpoint in the hierarchy after the mRS was for the SAINT I trial the change from baseline in the NIHSS score [42], termed co-primary in the protocol. Further secondary outcomes include functional outcome assessed with the Barthel Index [43], impact on daily activities (using the patient-rated Stroke Impact Scale) [44], and quality of life (using the EQ-5D patient-rated instrument) [45]. Training and certification in the mRS and NIHSS has been conducted to ensure that the scales are used effectively and consistently within the trials. The sample size calculation was based on the assumption of a clinically relevant treatment effect equivalent to a common odds ratio of 1.3 (90% power, 5% two-sided significance). NXY-059 significantly improved the overall distribution of the mRS disability score compared to placebo (p = 0.038) in the SAINT I trial [39]. The incidence and profile of adverse events for NXY-059 was comparable to placebo.

The many previous failed trials of neuroprotective strategies have provided valuable lessons for the design of future studies in acute ischaemic stroke. All the key criteria embodied in the STAIR recommendations for the preclinical and clinical development of acute stroke therapies have been adopted for the evaluation of NXY-059. These include the proof of a potent neuroprotective action in rodent transient and permanent stroke models; efficacy in a primate model in both histological and functional outcomes; safety with successful achievement of neuroprotective plasma concentrations in human stroke matching these preclinical models; and design of two pivotal clinical trials with appropriate sample sizes and optimal trial design to test the safety and efficacy of NXY-059.

Results from SAINT I demonstrate that NXY-059 significantly improved the overall distribution of the mRS scores compared to placebo, with reduction in disability at 90 days after treatment within 6 h of stroke onset.

I thank Dr Louise Picken, from Complete Medical Communications, who provided medical writing support on behalf of AstraZeneca.