This was “not an overwhelming result, but very encouraging, and we knew we had only a one in twenty chance that it was a false-positive,” said Lees. “So we were very much hoping that it was a genuine result.”

“We have very limited treatment options, and we were desperately hoping that we would have something else to use to treat patients.”

Stroke is the third-leading cause of death in the UK and the USA, and is and a major cause of disability in adults. The high rate of disability resulting from stroke also leaves many patients needing expensive healthcare and daily living support for the rest of their lives.

At present, there is just one treatment that is licensed for use in ischemic stroke patients: the clot-busting drug tissue plasminogen activator (tPA). But the use of tPA is limited: the majority of patients arrive in hospital after the 3-hour treatment window for tPA, and many of those who do arrive in time have contraindications to the drug.

Many doctors also have concerns over the treatment-related risk of hemorrhagic stroke, especially in elderly patients.

As Lees noted: “We have very limited treatment options, and we were desperately hoping that we would have something else to use to treat patients.”

To be certain of having enough power to detect the size of effect seen in SAINT I in the follow-up trial, the investigators increased recruitment for SAINT II from the planned 1700 stroke patients to 3306 participants.

But after 3 months of follow-up, patients who received NXY-059 within 6 hours of stroke onset (average of 3 hours and 48 minutes) had similar functional outcomes to those who were given placebo.

The odds ratio for improved functional outcome with the study drug was 0.94, with a confidence interval of 0.83–1.06.

“It’s very disappointing when you find that it has been a false dawn,” said Lees.

“Many people will always come out of the woodwork and say, well, in retrospect, I told you so.”

Falling on the final STAIR

The results were particularly galling for the researchers because they had adopted an extremely rigorous approach to testing NXY-059. In 1999, against a background of putative neuroprotectants proving ineffective in clinical trials of stroke patients, the Stroke Treatment Academic Industry Roundtable (STAIR) had its first meeting.

STAIR produced a raft of recommendations for the preclinical testing of neuroprotectants. These included:

• Conducting randomized, blinded animal studies;

• Replicating the results of animal studies in at least two independent laboratories;

• Monitoring animals for at least a month after treatment;

• Testing promising drugs in primate models of stroke;

• Defining the most likely dose range and therapeutic window.

NXY-059 fulfilled all the STAIR criteria for preclinical testing before it moved to the clinical levels.

Lees noted that, at the time, some scientists disputed the positive effect of NXY-059 in animal stroke models. “If that’s true, then it’s worrying that they did not publish their data in advance [of clinical testing],” he said.

“On the other hand… many people will always come out of the woodwork and say, well, in retrospect, I told you so. So, it’s difficult to know what to take from that.”

He stressed that success at preclinical levels boosts, but does not guarantee, the chances of clinical success, simply because of the fundamental differences between humans and animals.

The area of stroke damage in patients is about the size of a ping-pong ball, or a tangerine, Lees explained, “so we could have been missing a benefit that was to make that smaller by, say, the size of the thickness of the skin.”

The entire brain of some rodents used in preclinical research is about the size of a grape, so a similar size of benefit would have a noticeable effect on the animals’ physical state after a stroke.

“The STAIR criteria… as they currently stand they may not be rigorous enough.”

Lees, himself a STAIR member, said that the SAINT experience “suggests that the STAIR criteria are not sufficient – as they currently stand they may not be rigorous enough.”

He spoke of a move toward conducting meta-analyses of animal study data. Ideally, this would require standardizing the methods used in the laboratory, something that may be difficult to achieve, as scientific funding currently encourages competition, rather than collaboration.

There is also room for expanding early clinical testing, doing more small studies using imaging markers as surrogates for clinical endpoints, before moving into the expensive and labor-intensive phase III trials.

The next STAIR meeting, in early 2008, will address these issues.

Statistical wrangles

The STAIR members recently gave their backing to the method of statistical analysis used by the SAINT investigators. The method involves analyzing the distribution of scores across a scale, in this case the modified Rankin Scale (mRS), to detect differences between patients given active treatment or placebo.

SAINT I represented the first time this statistical method had been used in a phase III stroke trial, and it met with some criticism.

“That surprised us enormously, that something that had been in the statistical literature and that entire statistical programs had been developed to deliver, that it should then be criticized for use [in the SAINT trials],” said Lees.

Stroke trials have traditionally dichotomized outcome scales, looking, for example, at the proportion of patients who achieve a mRS score of 0–3, indicating functional independence, rather than 4–6, which indicates functional dependence or death.

“That surprised us enormously, that something that had been in the statistical literature and that entire statistical programs had been developed to deliver, that it should then be criticized for use [in the SAINT trials],”

Lees stressed that this type of analysis remains the most appropriate when testing a treatment that is designed to save lives. But when assessing an intervention intended to improve the neurologic outcomes of all patients treated, assessing the distribution of scores is far more appropriate.

For example, dichotomizing the mRS between 3 and 4 points would not account for patients who remained disabled but survived (mRS score of 4–5, rather than 6), or those who moved from a mRS score of 1 to 0. Recent research has highlighted the importance of the latter scenario, showing that patients who achieve a mRS of 0 are significantly less costly to care for than those with a score of 1.

“So there’s many reasons for thinking that the individual grades in the Rankin are meaningfully different between each other, in terms of both cost, and things that matter to the patient,” said Lees.

He noted that, aside from the STAIR members, US Food and Drug Administration representatives at the meeting were also enthusiastic about the analytic method used in the SAINT trials.

“So it had backing, really, from the groups that matter,” he said. “The regulators and the combined academics and pharmaceutical industry people.”

The way forward

With the neuroprotection field suffering in the wake of SAINT II, Lees sees three directions in which acute stroke research can continue in the near term.

Firstly, there is the hope that brain imaging will allow physicians to select stroke patients likely to benefit from acute treatments. This may help select patients for clinical trials, and could also expand the treatment window for tPA therapy.

Secondly, there are other thrombolytic drugs in the final stages of clinical testing, such as desmoteplase, which could target blood clots without the risk for systemic side effects, and the snake venom ancrod.

Thirdly, Lees said that efforts continue to improve consistency within clinical trials, to address the fact that two patients who are of similar age and have similar stroke severity when admitted to hospital can have very different outcomes 3 months later.

“Some of that is because they do genuinely differ in how they respond over the 3 months before we measure them,” he said.

“Some of it is also a difference in how the clinician assesses the response at the end of 3 months.”

This was partly addressed by the SAINT I investigators, who developed a DVD-based program to train all participating physicians in use of the mRS.

“The field is not completely dead but it’s taken a bit of a hit.”

There are still a few treatments that may prove beneficial in stroke patients. Activated protein C has shown neuroprotective capabilities in animal stroke models, and may also reduce the risk for thrombolysis-related bleeding. The drug, which is already approved to treat sepsis, is entering the first stages of clinical testing.

And other treatment approaches are being tested, such as using microbubbles and ultrasound to break up clots, or infrared lasers to promote neuron survival.

“The field is not completely dead,” said Lees, “but it’s taken a bit of a hit.”

He said that, while the negative SAINT II results were disheartening, this is to be expected, on occasion, as the design of clinical trials allows for a false-positive result one in 20 times, and a false-negative finding one in every five or 10 times.

So he emphasized that the SAINT experience vindicates the insistence of the regulatory authorities on the need for two phase III trials or, alternatively, one very large one with an extremely stringent statistical analysis.

“So I think, really the lesson is that we’ve done things correctly, we’ve been disappointed in this case, but we still have a condition which is devastating, needs treatment, and, you know, I think we carry on with that.”

What are the next steps for stroke research after the failed SAINT II trial? Visit MedWire Open to read and add comments on this interview.

Link to MedWire News story:
SAINT II publication confirms 'false dawn' for neuroprotection
Link to NEJM publication:
NXY-059 for the Treatment of Acute Ischemic Stroke