KISSIMMEE, FLA—NXY-059, a free-radical–trapping agent, significantly reduced disability in patients who had had an acute ischemic stroke, according to results from the Stroke–Acute Ischemic NXY Treatment (SAINT I) study. Patients who received the drug within six hours of stroke onset were 20% more likely to improve at 90 days than were those who took a placebo. Improvement was measured by score on the modified Rankin scale, the study’s primary outcome measure. Further analysis revealed that a higher percentage of patients who used the drug became symptom-free and were able to walk without assistance after a stroke.

SAINT I becomes the first positive, properly powered phase III trial of a neuroprotectant drug, with SAINT II currently under way in the hope of confirming those benefits. The findings also mark the long-awaited link between animal science and human studies in neuroprotection and stroke, ending a pattern in which numerous agents have reduced infarct size in various animal models but produced disappointing findings in subsequent clinical trials.

"It is absolutely clear that we have a statistically significant, clinically important effect at 90 days, which was even more [robust] when we looked earlier on," said principal investigator Kennedy R. Lees, MD, in a symposium held at the 2006 International Stroke Conference, a week after the breakthrough results were published in the February 9 New England Journal of Medicine. "NXY-059 is a drug that is well tolerated and safe. Its side effect seems to be one that we find highly desirable: to reduce hemorrhagic transformation in the presence of tissue plasminogen activator [t-PA]…. I believe that we do have a clear possibility of achieving neuroprotection, something that would be useful to many, many patients, rather than just a few." Dr. Lees is a Professor of Cerebrovascular Medicine at the University of Glasgow in the United Kingdom.

A new approach to stroke treatment, which in the case of NXY-059 involves disruption of the ischemic cascade through neuroprotection, could have a profound effect on the number of patients who could benefit, Dr. Lees asserted. Currently, the only FDA-approved therapy for treating patients with acute ischemic stroke is IV t-PA. Although thrombolysis is effective, only about 5% of stroke patients actually receive t-PA, primarily due to its three-hour treatment window and possible risk of intracerebral hemorrhage accompanying its use. "What we need is something that may have a slightly more modest effect but that can be given to a large number of people, because that will have greater impact in public health terms," said Dr. Lees.

NOVEL DRUG, NOVEL METHODOLOGY

SAINT I was a randomized placebo-controlled trial involving 1,722 patients with acute ischemic stroke, conducted at 158 hospitals in 24 countries. Participants were assigned to receive a 72-hour infusion of placebo or IV NXY-059 within six hours of stroke onset. The mean time from stroke onset to treatment was three hours and 46 minutes, as each center’s investigators were required to keep the average time to treatment within four hours. In animal models, "the drug was clearly effective at four hours and a bit less clearly effective at five, and so on," reasoned Dr. Lees. The initial infusion rate was 2,270 mg per hour, reduced after one hour to 480 to 960 mg (32 to 64 mL) per hour for an additional 71 hours.

The NXY-059 and placebo groups were roughly equal regarding baseline prognostic variables. Approximately 29.4% of the placebo group versus 28% of the NXY-059 group received t-PA. In both groups mean age was about 68, and 55% of participants were men. NIH Stroke Scale (NIHSS) scores at study entry were comparable, and there were no significant differences in medical history between groups. "We made sure that the groups were well balanced in terms of their age and stroke severity, NIHSS score, use of t-PA, etc," said Dr. Lees.

The study’s primary end point was the score on the modified Rankin scale at 90 days or the last rating. Dr. Lees explained the reasoning for choosing the Rankin scale and the statistical analysis used in computing the data, theorizing that analyzing the entire distribution of scores on the Rankin scale is more powerful and more relevant to a neuroprotectant than is the use of dichotomization, a method used in previous stroke trials.

"We chose what we thought was the most robust of the outcome measures that were available—the modified Rankin scale," said Dr. Lees. "With the Rankin scale, because it covers a wide range of disability and because patients who come in with a severe stroke are very unlikely to make it back to full recovery, particularly with a neuroprotectant drug, we decided we would analyze this by looking at the overall shift in the scale…. We used the Cochran-Mantel-Haenszel test to see if we could show a shift in that distribution, rather than picking one single hurdle and expecting patients to jump over that. That was our single primary outcome measure, to be taken at 90 days. In addition, though, if that was positive, we could go on and look at the Rankin cut in different ways, cut at different dichotomies, and we could look at it at earlier time points, at seven and 30 days.

"Assuming the Rankin was positive, we could go on to our co-primary variable, which was the second in our hierarchy of tests, the NIHSS. We were going to analyze this by change from baseline score. We had NIHSS as measured at seven and 90 days. And if that was positive, we could go on and look at the Barthel Index, mainly at 90 days, but also at seven and 30 days, and on down through some of the more qualitative measures."

SIGNS OF NEUROPROTECTION

The researchers found that NXY-059 significantly improved the overall distribution of scores on the modified Rankin scale, compared with placebo (odds ratio, 1.2). Significant differences were also observed when the scores were categorized as 0 (no residual symptoms) to 1, versus 2 to 3, versus 4 to 5 (bed-bound). Results were even more significant at seven days (odds ratio, 1.31) and at 30 days (odds ratio, 1.27). In addition, among patients who received NXY-059, 4.4% more became asymptomatic (Rankin score, 0), and 3.7% more were able to walk without help (Rankin score, 0 to 3), compared with those who took a placebo.

The more robust results at earlier time points were not unexpected, noted Dr. Lees. "Let’s face it—If you take a group of 70-year-old people, even without stroke, and follow them up for a long period of time, some of them are going to have a stroke, some are going to have a heart attack, cancer, [etc]," he said. "The same will happen even over a three-month follow-up period for this population, and that’s just what we saw: at seven days, a more profound effect than we saw at 30 days, which was more profound than the 90-day outcome."

SAFETY ISSUES

"There were no problems with safety," said Dr. Lees. "Mortality was identical in the two groups. Safety in terms of adverse events and serious adverse events was very reassuring, with a slightly lower proportion of these events in the NXY treatment group compared with placebo, but no statistically significant differences." Among the adverse events, pyrexia was the most common, occurring in 19% of both groups. The lone marginal difference was that slightly more patients who received NXY-059 (6.4%) had hypokalemia during the infusion (versus 4.4% for the placebo group).

"The serious adverse event that we might have worried about was in the t-PA–treated population—were we causing any differences in hemorrhagic transformation?" noted Dr. Lees. Instead, the researchers found that when given along with t-PA, NXY-059 reduced the risk of ischemic-to-hemorrhagic transformation from 27.3% to 15.4% and lowered the risk of symptomatic intracranial hemorrhage from 6.4% to 2.5%: "a very nice effect there, a hint of biological activity even on the safety measures," commented Dr. Lees.

OTHER OUTCOME MEASURES

NXY-059 did not significantly improve other outcome measures that were evaluated. For example, the drug had no effect regarding the difference between the two treatment groups in the average change from baseline score on the NIHSS at 90 days. A post hoc analysis showed that although complete neurologic recovery (NIHSS score, 0) was more common in the NXY-059 group (22% versus 17% for placebo), the difference between the two groups in frequency of excellent neurologic outcomes (NIHSS score, 0 or 1) was not statistically significant (33% for NXY-059 and about 31% for placebo), for an odds ratio of 1.13.

"But when we looked at seven days, [we found] an encouraging result with statistical significance and a much higher odds ratio, much more like what we saw with the Rankin scale at that earlier time point and, in fact, consistent with what you would expect," said Dr. Lees. "The NIHSS was designed to measure severity, not outcome."

The Barthel Index also was not significant at 90 days but was strongly positive at seven and 30 days, added Dr. Lees. "So all these scales are very similar," he said. "They are tracking together." Regarding other secondary end points such as scores on the Stroke Impact Scale and the European Quality of Life–5 Dimensions, "the overall message here is that they were not strongly encouraging … but again, a trend in the right direction," he said.

A TRUE EFFECT

"It looks to me as if there is a true effect here that is present exactly when we would expect it to be, and it is being diluted by all the other things that happen after a stroke as time goes on," commented Dr. Lees. "But there is no inconsistency in my mind between the post hoc analysis of the NIHSS and the primary end point. We met the primary end point in this trial on the Rankin…. We did not find any interactions across the time window or the use of t-PA, etc, that would be driving this. We failed on the NIHSS as [measured by] change from baseline, but in every other way that we look at the other outcome measures, they are all moving in the same direction as the Rankin.

"It’s the first neuroprotective trial to be positive in its prestated primary end point, the modified Rankin scale, which is the one that is widely recognized to be the best," he continued. "We are reducing disability."

PERSPECTIVE ON STATISTICAL ANALYSIS

The use of the Rankin scale shift analysis in the SAINT I trial is likely to change the way that future stroke trials are conducted, but it may also raise some concerns. "It is very interesting to have this new method," said the symposium’s co-moderator, Julien Bogousslavsky, MD, of the Universitaire Vaudois in Lausanne, Switzerland. "But there is also a danger, probably, of overemphasizing this Rankin scale shift, because perhaps some of us may think that the main result of the trial is to have found the efficacy of this method, rather than the efficacy of the drug as compared to previous neuroprotective agents that have been tested."

Jeffrey Saver, MD, Professor of Neurology at the University of California, Los Angeles, School of Medicine, sought to put the SAINT I trial outcome and statistical methods into perspective. Dr. Saver, an investigator in the SAINT II trial though he was not involved in the SAINT I study, noted that the record of failure in past neuroprotectant trials led to a paradigm shift in the field at the turn of the century. The SAINT I trial, he said, could be considered the "first postmodern neuroprotective trial," in which treatment was started early, the trial was large, the right patients were enrolled, and a newer statistical analytic technique was used.

"But what are we to make of this primary outcome measure of the shift in the Rankin scale?" asked Dr. Saver. "What does it mean to us as clinicians? For those of us who went through all the prior negative neuroprotective trials, healthy skepticism makes us think, Is there some statistical trickery taking place here, rather than a true biologic effect?… Why is the shift analysis behaving more sensitively, and is it a number we can make sense of?"

Essentially, the shift analysis allows for the detection of improvement anywhere along the continuum of outcomes. In other words, "Does the treatment make the patient better, to some degree?" commented Dr. Saver.

Not all transitions on the Rankin scale are worthwhile achievements, Dr. Saver pointed out. "We know … that of healthy, elderly Americans, 47% of them consider a severe stroke outcome as worse than death. So for these individuals, going from a Rankin 6 to a Rankin 5 is not a better outcome."

It is also unlikely that a drug is going to make every patient better by exactly the same small amount or that it will make one patient better by a large amount, offered Dr. Saver. "Instead, the most plausible scenario is going to be a combination of jumps of modest, small, and incremental degree that reflect what we know about the nature of the disease and the nature of drug action."

According to Dr. Saver, a group of experts examined the SAINT I data and estimated that the biologically most plausible number of patients needed to treat for one patient to show improvement across the entire Rankin scale with NXY-059 treatment was 9.8. "The lowest possible number needed to treat with the SAINT I distributions is about 8," he said. "The maximal possible needed to treat is 17." By comparison, Dr. Lees’ group reported that about eight patients would need to be treated for one patient’s score to improve by 1 point.

"When you consider that there is no definite evidence of harm with this drug and one out of 10 is improving, I suspect that most physicians and patients would judge this therapy as worthwhile," commented Dr. Saver. "Of course, this has to be confirmed by a second trial."

FUTURE IMPLICATIONS

The SAINT I trial will also have an impact on future research and has a number of important implications. Eng Lo, PhD, Head of the Neuroprotection Research Laboratory at the Massachusetts General Hospital in Charlestown, said that the SAINT I trial’s success may lead to new questions and perspective on the validity of experimental models, the feasibility of combination stroke therapy, and whether reperfusion is necessary, as well as new pathways and targets to investigate. "Perhaps it is very rational to think of targeting oxidative stress as a way of therapeutically intervening," he said.

According to Marc Fisher, MD, a Professor of Neurology at the University of Massachusetts Medical School, Worcester, a Rankin scale shift approach is likely to lead to the identification of more subtle treatment effects in future trials. "The number needed to treat will be much smaller than with the dichotomized outcome," he said. "The shift of 1 point, for example, is clinically meaningful in my perspective, because going from 4 to 3 means you can walk versus not walk."

Another key issue, according to Dr. Fisher, is whether this shift analysis, "if it were retrospectively evaluated in other neuroprotective trials, could have identified neuroprotective agents that we’ve discarded [because they had] a dichotomous outcome [but] are likely to show benefits to some modest extent. Can a shift approach be applied to other stroke outcome scales?... Can a modified Rankin shift approach be used in a global outcome measure with either dichotomized end points or other shift end points of other scales?"

Dr. Fisher believes that the Rankin scale shift approach will likely supplant dichotomized outcome measures in most acute stroke trials. "It will allow us to identify more subtle but clinically meaningful treatment effects in future trials," he said. Ultimately, he added, if the findings are replicated and NXY-059 is approved by the FDA, the drug could be used routinely with IV t-PA, especially in patients who are at high risk for intracerebral hemorrhage.

Future acute stroke trials will also have to address the problem in conducting placebo-controlled trials with a potentially approved neuroprotectant administered between zero and six hours, Dr. Fisher pointed out. Additional issues, he offered, include: "Before reperfusion, does NXY-059 potentially expand the window for either thrombolytic or mechanical reperfusion? With penumbral imaging, will NXY-059 have enhanced benefits at later time points beyond the six-hour potential approval that we may see in a couple of years?"

Gregory J. del Zoppo, MD, of the Scripps Research Institute in La Jolla, California, noted in the Perspective section published along with the SAINT article that effective therapies for ischemic stroke may need to provide protection to the entire neurovascular unit. "Discovering how such ‘neurovascular protection’ can salvage neurons may well lead us to the next frontier in the treatment of stroke," Dr. del Zoppo commented.

NR

—Colby Stong

Suggested Reading
1. Del Zoppo GJ. Stroke and neurovascular protection. N Engl J Med. 2006;354:553-555.
2. Lees KR, Zivin JA, Ashwood T, et al. NXY-059 for acute ischemic stroke. N Engl J Med. 2006;354:588-600.

Return to table of contents