SAN FRANCISCO—Disappointing results from the Stroke–Acute Ischemic NXY-059 Treatment (SAINT II) study may represent a turning point for neuroprotection research. Negative findings from SAINT II, which was unable to replicate the positive findings from last year’s SAINT I trial, have given researchers reason to reflect on the future course of a once-promising but as yet unfulfilled area of translational research.

“We are at an absolutely critical point in time,” said Geoffrey A. Donnan, MD, at the 2007 International Stroke Conference. “I think we are really at the crossroads of neuroprotection. It seems to me that we either have to stop altogether because of recent disappointments, or we press ahead in a very different way.” Dr. Donnan is a Professor of Neurology at the University of Melbourne and Director of the National Stroke Research Institute in Australia.

Although reasons for NXY-059’s failure in the clinical trial stage may be open to interpretation, Dr. Donnan believes that one thing is clear. “NXY-059 does not neuroprotect in humans,” he said. “I think it is as simple as that. I suspect that the positive results of SAINT I were due to the play of chance…. We have to pause and go back to the absolute basics of where we are with neuroprotection and look at how we actually select neuroprotectants for phase III clinical trials.”

FINDING THE RIGHT DRUG

The selection of neuroprotectant agents for clinical trials has been largely dependent on the efficacy of the drugs in animal model studies. In an effort to determine the quality of such studies, Dr. Donnan and colleagues compiled data from 1,026 experimental neuroprotection trials in acute stroke that were conducted between 1955 and 2003. They also devised a 10-point scoring system, based on Stroke Therapy Academic Industry Roundtable (STAIR) criteria, to quantify the quality of each study. Common problems with the animal model studies, the researchers found, included small sample sizes, the fact that they were unblinded and rarely randomized, and that frequently no statement was made about temperature control—temperature is “one of the most powerful neuroprotectants,” said Dr. Donnan. In general, his group found that “the higher the score, the more modest the outcome,” he noted. “Most studies are, what we would consider by these criteria, of fairly low quality.”

Dr. Donnan’s group also conducted a meta-analysis of the animal model studies. In one trial that used nicotinamide, for example, “the overall benefit was about a 30% [improvement in outcome],” said Dr. Donnan. “It is pretty average for neuroprotection experiments. But if you look at the difference [for patients who] have comorbidities like hypertension and diabetes, the effect is less.”

The researchers also applied their scoring system to the 10 major papers that have been published through 2003 regarding NXY-059, excluding both SAINT trials. They found that the overall mean score was 4.5 out of 10, “which is a bit disappointing,” Dr. Donnan commented. Analysis of the mean scores for each of the 10 criteria from the trials revealed that “NXY-059 was not bad for most things, but not good for random allocation to treatment or control, only 0.3 out of 1,” he noted. The scores were “not good” regarding blinded induction of ischemia (0.2). As for the appropriate animal model—aged, diabetic, or hypertensive—the mean score was 0.1, and the results were “absolutely awful” for sample size calculation (0), he said. “Most of these studies have about half a dozen animals, and the power is about 50%. No one would do a clinical study with a power of 50%. So the potential for error is enormous.”

Dr. Donnan believes that some key issues need to be resolved before future neuroprotection trials are conducted. “I think we have to stop and ask ourselves some really hard questions before we go forward,” he said. “Are we selecting drugs that really work in animal models? I want someone to reassure me that rat cells are the same as human cells in their response to cerebral ischemia. I want someone to show me that the neuroprotection you see in an animal model can also be reproduced in the same sort of model in humans. At the moment, there is no evidence whatsoever at any level that we have achieved neuroprotection in humans. Is there any evidence that neuroprotectants reach their target in humans? And of all studies, NXY-059 seems one of the most obvious examples of this, because the serum levels in both animals and humans were very similar. In fact, the serum levels were somewhat higher in humans. So, did NXY-059 actually reach its target? We do not know. No one has ever looked at this. And finally, are the human proof-of-concept models of cerebral ischemia really ideal? So, based on all these considerations, I will outline a new road map for neuroprotection research.”

MOVING FORWARD IN NEUROPROTECTION

The first step on that path, beyond choosing more efficacious drugs from preclinical studies, is devoting greater attention to neuroprotection in humans, according to Dr. Donnan. “We need to look at in vitro efficacy in human tissue at a very basic level,” he said. Oxygen deprivation and oxygen-glucose deprivation (OGD) models are one technique that could be used for demonstrating efficacy in human brain cell cultures and tissue slices. Another option, he suggested, is using fresh brain slices, which can be obtained from biopsy specimens or after patients have had surgical decompression for intracerebral hemorrhage or during temporal lobectomy. “There are significant amounts of normal human brain tissue potentially available, of course, after going through the usual ethical processes,” said Dr. Donnan. “The advantage of the OGD model is that the neurovascular bundle is more or less intact. This is important because there may be an interaction between vessels and neuronal cells or glia and neuronal cells, and these interactions might be important in the response to ischemia. These responses to ischemia, which we understand reasonably well in animal models, must be replicated in humans to be assured that we are on the right track with the neuroprotectants we select. To my knowledge, this has never been done.”

The next step, according to Dr. Donnan, is to determine if a compound is actually reaching its intended target. “It seems absurdly simple that we should be doing this, but for some unknown reason this principle has never been established in humans,” he commented. “What I’m talking about is not rocket science.” With the use of positron emission tomography (PET), he suggested, the in vivo distribution of the putative neuroprotectant needs to be examined in both healthy patients and in those with ischemic stroke. It must be clearly demonstrated, he said, that the neuroprotectant crosses the blood-brain barrier and reaches the ischemic penumbra in adequate concentrations.

Using PET to label compounds such as minocycline, for example, is “not all that difficult,” noted Dr. Donnan. “There is no reason why we shouldn’t go ahead and do this sort of thing much more often as part of a stepwise process to a logical outcome with neuroprotection.”

The next goal is achieving efficacy in novel human models of cerebral ischemia, said Dr. Donnan. He suggested preloading a neuroprotectant, either before a vascular procedure or during the high-risk phase after a transient ischemic attack or minor stroke. “We know that from the animal data, if you preload with a neuroprotectant, you get a much more powerful effect than if it is administered during later time windows. So, why not use the opportunity that we have with high-risk situations, which unfortunately exist with procedures—for example, [during] cardiac bypass surgery, carotid endarterectomy, and angioplasty? Preload them with a putative neuroprotectant, and look at outcomes in terms of a clinical end point and/or imaging. This approach, together with the use of a combination of imaging and clinical end points, will allow for smaller sample sizes.”

REACHING THE CLINICAL STAGE

Only after all of these objectives have been met should a neuroprotectant move into a standard clinical trial, advocated Dr. Donnan. “For phase II clinical trials, we should be selecting patients that have a visible target,” he said. “This target should be the ischemic penumbra, because almost all our understanding of the ischemic cascade is based on experiments involving the grey matter in the ischemic penumbral state. Hence, this is the target of our therapy, and it is only logical, therefore, that we should select patients in whom our target was present.”

While some have suggested that the quest for neuroprotection be abandoned, Dr. Donnan doesn’t agree. “I don’t think that is correct, because there are so many unanswered questions as to why we are where we are,” he said.

“I think we should press ahead, but we should do so using a different and new road map. By doing so, we will identify very early failures, or we will eventually have a neuroprotectant that actually works in humans."

NR