Neuroprotection: Lessons from a Spectrum of Neurological Disorders

Neuroprotection, or modifying disease progression by slowing or preventing neuronal loss, remains a challenge for both the chronic cell death of the neurodegenerative diseases and also the acute cellular insults associated with ischemic and hemorrhagic stroke.

In the field of stroke, neuroprotection can be direct, using pharmaceutical agents or physiological strategies that inhibit the biochemical, metabolic and cellular consequences of ischemic injury, or indirect, by restoring blood flow and oxygen supply to ischemic but noninfarcted (salvageable) brain tissues (1). A broader definition of stroke neuroprotection includes the use of drugs that decrease the incidence of new or recurrent strokes, or which mitigate brain damage should stroke occur (i.e., prophylactic neuroprotection). In patients with brain hemorrhage, neuroprotection could be achieved by limiting hematoma growth.

Given the huge public health burden of stroke, the National Institute of Neurological Disorders and Stroke (NINDS) has spent over $200 million on over 20 stroke clinical trials since 1977 (2). A comprehensive review (3) found that 178 acute stroke trials were conducted in the 20th century, of which nearly 70% were fully or partly sponsored by industry at a cost of billions of dollars. The failure of over 50 neuroprotective drugs in stroke clinical trials has led to skepticism about the very feasibility of ‘direct’ neuroprotection; an important exception is the use of nimodipine in subarachnoid hemorrhage (4). Yet, there have been notable successes with ‘indirect’ reperfusion strategies, for example the NINDS r-tissue plasminogen activator (r-tPA) (5) and PROACT (6) trials. The results of desmoteplase in acute ishemic stroke-1 (DIAS-1) (the phase II trial of desmoteplase, a new thrombolytic agent) (7) and the trial of recombinant activated Factor VII in acute intracerebral hemorrhage (8) appear promising. In addition, we have witnessed remarkable successes with ‘prophylactic’ neuroprotection using antiplatelet agents, warfarin, statins, and angiotensin converting enzyme (ACE) inhibitors, and interventions such as carotid endarterectomy. Other stroke trials such as the EC-IC bypass, WARSS, and VISP trials have also advanced the field by demonstrating that certain interventions may not be as widely effective as previously believed. What, then, has limited our success with ‘direct’ neuroprotectants in stroke, and what lessons can be learned from neuroprotection in other neurological diseases?

The challenges of developing effective neuroprotective agents are underscored by several unanswered questions across the general field of neuroprotection. What is the cause of cell death? Several pathogenic mechanisms underlie neuronal degeneration including mitochondrial dysfunction, oxidative stress, excitotoxicity, inflammation, and possibly other mechanisms (9, 10). Thus, a single agent directed at only one pathogenic process may not be sufficient, and combination therapies may yield additive or perhaps even synergistic benefit. For the majority of the neurodegenerative diseases, combination therapies targeting different mechanisms have not yet been pursued, and preclinical data are limited. An additive neuroprotective effect has been observed in a rodent model of amyotrophic lateral sclerosis (ALS) with the combination of minocycline and creatine (11) as well as with the combination of a cyclooxygenase-2 (COX-2) inhibitor and creatine (12); a clinical trial evaluating combination therapies in ALS will soon be underway (P. Gordon, personal communication). Several agents found to be neuroprotective in animal models of various neurodegenerative diseases have failed to show efficacy in subsequent clinical trials. In such negative trials, questions regarding optimal dosing and delivery to the central nervous system (CNS) as well as the stage of the disease at which the neuroprotective agent is administered must be considered. In addition, the predictive value of the animal model must be critically assessed. Is the endpoint selected for preclinical studies relevant to the clinical trial endpoint? For example, does prolongation of the rodent lifespan translate to improved functional capacity for the patient? Moreover, what is the best clinical trial endpoint? There are few accepted biomarkers or surrogate endpoints available for use in clinical trials, and clinical outcome measures often include time to the occurrence of a clinical event, or a change in a clinical scale (13). However, such measures may lend themselves to confounds, such as a symptomatic effect of the agent under study as was seen in the DATATOP study in Parkinson's disease (14). The difficulties of using clinical endpoints are increasingly recognized in trials for multiple sclerosis, where magnetic resonance imaging (MRI) is often being evaluated as a surrogate outcome measure (15). Despite these challenges, a measurable clinical improvement remains the ultimate test of success for neuroprotective agents and is essential for obtaining Food and Drug Administration (FDA) approval. However, pivotal efficacy trials require large sample sizes, which have high cost and limited feasibility. How can agents be evaluated efficiently in clinical trials? To improve chances of success in phase III trials, phase II futility designs are being applied to neuroprotective trials in Parkinson's disease, allowing the use of small sample sizes and short-duration studies to distinguish agents worthy of further testing in Phase III efficacy trials from those which are unlikely to yield benefit (16).

How are the lessons learned from previous stroke and nonstroke neuroprotection trials being applied to current acute ischemic stroke trials? Data from stroke thrombolysis trials have emphasized the concept of narrow therapeutic time windows, which was overlooked in several neuroprotective trials. Previous inadequacies in preclinical testing have been addressed, and recommendations for future drug development and clinical trials have been proposed (17). The importance of optimizing patient selection as per the specific characteristics of the drug (e.g. effects on white versus gray matter, time window for efficacy) is now widely recognized. The promising results of the DIAS-1 trial (7), and the burgeoning of trials such as DEFUSE, EPITHET, and ROSIE which utilize baseline MRI perfusion-diffusion ‘mismatch’ as an inclusion criterion, show the great interest in being able to select patients with ‘salvageable’ tissue (the target tissue for neuroprotection), and the potential of using neuroimaging tools to rapidly identify such tissue. The planned 3200 patient target enrollment for the SAINT-2 trial of NXY-059 (as compared to the average of 186 patients in previous neuroprotective trials (3)) shows that ongoing trials are being designed with larger sample sizes to ensure the trial is powered to detect what may be a small benefit. To address the challenges of patient recruitment and the high costs of drug development, the current focus is on conducting multi-center trials, sharing data, and developing stroke trials consortia, for example the recent NINDS-sponsored ‘Specialized Program of Translational Research in Acute Stroke’ (SPOTRIAS) initiative. In addition, the utility of surrogate outcome measures, such as stroke lesion volume, in reducing sample size requirements is being evaluated. Recent trials are investigating combination therapy with drugs having actions on different cell death pathways (18). Furthermore, the concept of the ‘neurovascular unit’, which emphasizes the need to salvage not only neurons but also glia, cellular processes, basement membranes, endothelium, and other structures, is now better recognized (19). Drugs that decrease reperfusion injury, reduce postischemic hemorrhage, and inhibit downstream targets in cell death cascades are being developed to increase the safety and efficacy of proven therapies such as tPA. In addition to developing new drugs, efforts are underway to expand the stroke thrombolysis time window using simple, widely available methods such as normobaric hyperoxia therapy (20) and to increase the efficacy of thrombolysis using transcranial Doppler ultrasound (21). The influence of physiological parameters such as glucose, blood pressure and temperature on stroke outcome is now better understood. Induced hypothermia has already shown efficacy in global cerebral ischemia (22) and is being tested in ischemic stroke. Finally, previous trials have provided important insights into the issues of study design and selection of outcome measures that are being applied in ongoing trials (23).

Clearly, there are common themes across neuroprotection trials. Lessons learned from non-stroke neuroprotection trials, and the early efforts in acute stroke neuroprotection, have yielded significant advances that will have application not only in future acute stroke trials, but also in the field of stroke recovery and across a spectrum of nonischemic neurological conditions. The future is bright!