|
HAVE
WE PROVEN THAT STROKE THROMBOLYSIS WORKS? ENTHUSIASTS AND SKEPTICS DUKE IT OUT
PHOENIX—Systemic tissue plasminogen activator (t-PA) is the only treatment approved for acute ischemic stroke in the United States. In the United Kingdom, t-PA is even less widely used than in North America, where the time-to-treatment window of three hours after symptom onset, as well as the availability of specialists to administer the treatment, has limited its general adoption outside of expert centers. Some researchers there feel that the results seen in the National Institutes of Neurological Disorders and Stroke (NINDS) trial that established the efficacy of t-PA need further replication, and that more randomized evidence is needed.
Charles P. Warlow, MD, Professor of Medical Neurology at the University of Edinburgh, is one of them. In a recent editorial in BMJ, he and colleague Joanna Wardlaw, MD, Professor of Neuroradiology at the University of Edinburgh, asserted that “what is good enough for heart attacks is still not good enough for brain attacks.”
The balance of risk and benefit of any treatment is “best determined by randomized trials—facts, not opinions and theories,” they said. But while randomized evidence on the use of t-PA in heart attacks is available on some 60,000 patients with myocardial infarctions, that number for stroke is only 5,675. While Dr. Warlow argued that the NINDS trial enrolled too few patients and perhaps had some randomization inequities that allowed a disproportionate number of patients with favorable prognosis into the treatment arm, that trial nevertheless demonstrated a benefit of treatment—a roughly 36% relative reduction of the risk of death or dependency—but also a fourfold increase in the risk of intracerebral hemorrhage in patients treated with t-PA.
When asked why t-PA should not be given “at least to those patients who are prepared to trade an immediate risk for long-term benefit,” the researchers replied that there are at least two good reasons for caution: first, they noted, the evidence for the three-hour time window is based on only 957 patients, of whom two-thirds were enrolled the NINDS study. “Experience tells us that relying on one positive trial to alter treatment policy is unwise, and that meta-analysis of a small number of trials is an uncertain art. To change the treatment of stroke radically around the world requires the reassurance of consistent results in several trials with much larger sample sizes, as happened for heart attack.”
Second, they argued that more information is needed on a wider range of patients: only 42 patients in the randomized trials of t-PA, for example, were older than 80, and only a few had lacunar strokes. “To know what sort of patients are most likely to benefit, or be harmed, will require reasonably sophisticated modeling of baseline factors, only one of which we know about so far—more than or less than three hours from onset of symptoms,” they said. “This will require several thousand more randomized patients.”
A CALL FOR MORE DATA
Although “enthusiasts and skeptics” have debated the need for more data, t-PA has failed to have the effect on outcomes anticipated by the NINDS results, they pointed out. “Even among enthusiasts such as ourselves, it has proved very difficult to implement thrombolysis, requiring rapid triage in emergency departments, five consultant neurologists and stroke physicians on an on-call rota, very rapid access to CT, and multidisciplinary after-care. Others find it equally difficult; some find it impossible in the United Kingdom and elsewhere.”
This disagreement between enthusiasts and skeptics has “played into the hand” of conservative health service managers, who “rightly” want more evidence before funding hyperacute stroke services, Drs. Warlow and Wardlaw wrote. “Advocating more trials does not mean that existing results cannot be believed—far from it,” they added. “It simply means we need much more detail.”
More data will be available in the years ahead from such trials as the third International Stroke Trial (IST-3), of which both authors are on the management committee, as well as the ECASS-3 (European Cooperative Acute Stroke Study), EPITHET (Echoplanar Imaging Thrombolysis Evaluation Trial), and DIAS (Duplex Sonography in Acute Stroke) trials. Currently, t-PA is licensed in Europe, and while patients treated within the license must be recorded, “registering use of t-PA will not add to the randomized evidence.
“Sadly for patients with stroke, the heart attack doctors are not just in the lead, they are out of sight [in] evaluating and now establishing intra-arterial chemical and physical approaches to unblocking the pipes [quickly],” Drs. Warlow and Wardlaw concluded. “Maybe brain doctors have been a bit too clever, deflected into seeking the holy grail of neuroprotection—expensive and, so far, unsuccessful. Basic plumbing should come first.”
GREEN LIGHT/RED LIGHT
More recently, at the American Stroke Association’s 28th International Stroke Conference, Dr. Warlow touched on some of these points again. In reviewing the current state of investigation with a number of stroke treatments, he used a “green light/red light” analogy, with red denoting situations in which treatment was suggested and being used in a haphazard way, amber for when randomized controlled trials were being done, and green for when the treatment was generally accepted. Using this approach, Dr. Warlow gave t-PA only an amber light. “I don’t think the evidence is yet convincing enough to change routine practice worldwide,” he said. “We are still at amber, although some of you have jumped the light.”
More and bigger trials are under way with t-PA but are delayed, he feels, by competition for patients from the “far better-funded” industry neuroprotection trials, and “hostility from grant application referees who believe in thrombolysis and cannot tolerate the genuine uncertainty of others. It is, in fact, a general problem in refereeing large multicenter grant applications, because there will be few referees independent of the study, and those who are may be hostile.”
Randomized trials are the benchmark and should be started early in the development of new treatments, “before the intervention becomes epidemic. Indeed, it would help if new treatments were made available only in randomized trials, before too many patients were damaged by too many interventions that simply don’t work,” he said.
“As Professor David Sackett remarked, surgical trials with controls tend to lack enthusiasm, while surgical trials with enthusiasm tend to lack controls,” Dr. Warlow noted. “Enthusiasts must realize that they will never get their enthusiasms into practice unless the skeptics are allowed to test them in trials.”
THE ENTHUSIASTS STRIKE BACK
While none would assail Dr. Warlow’s accomplishments and expert opinion in many areas, his conclusion about t-PA was met with some consternation on this side of the Atlantic. At the same meeting, Joseph P. Broderick, MD, Chair of the Department of Neurology at the University of Cincinnati Medical Center, took the opportunity to disagree pointedly with Dr. Warlow’s assessment, although with good humor. He, too, looked back at some of the changes in the burden of stroke that have occurred since the “dawn of CT” scanning, now some 25 years ago, and his sentiments were broadly similar to Dr. Warlow’s in that he asserted the need for stroke research to “get moving.” He even used the same red light/green light analogy used by Dr. Warlow to describe the state of certainty surrounding various current and burgeoning stroke treatments.
However, Dr. Broderick, who was a co–principal investigator on the landmark NINDS trial, confessed to perhaps being a little disadvantaged when it comes to assessing the readiness of t-PA treatment to be widely adopted. “I realize now that I’m color-blind,” he said, “because some of the things that [Dr. Warlow] considers amber, I consider green.”
SAFETY FIRST
Alastair Buchan, MD, Professor in the Department of Clinical Neurosciences at the University of Calgary—and by his own assessment an enthusiast about the use of t-PA—pointed out that there is no question from a physiologic point of view that as long as the brain is not irreparably damaged, restoring blood flow “is the right thing to do.” He argued that in the UK, however, stroke patients still cannot be evaluated and scanned in a timely way, and not having participated in either the NINDS trial or in the ECASS studies of t-PA, outside of a few expert centers, most community hospitals still do not have a sophisticated approach to stroke treatment.
“The carotid endarterectomy studies which are very cohesive, very coherent, and very persuasive, were based on 20 or 30 years’ experience in both doing angiography and measuring carotid disease and doing surgery, and people were very good at picking patients, measuring the disease, performing the procedure, before they ever got to a randomized trial,” Dr. Buchan said. The University of Calgary, he pointed out in an interview with Neurology Reviews, has spent the past seven years training those in practice to recognize strokes, to measure the extent of injury and choose the most appropriate patients for treatment in a safe and competent way. “If you say, ‘the data are unconvincing, we must have larger trials,’ I would say ‘that’s fine, but the trial should be focused on improving our understanding of the disease and its response to treatment, rather than testing the abilities of the provision of health care.’ ”
EXPERIENCE AND SKILL
Safe and effective treatments assessed by clinical trials eventually become available in general clinical practice, but “I’d say it’s like surgery, and you don’t want people doing surgery unless they know how to do it,” Dr. Buchan added.
“I don’t think that a really sensible trial should be done by people who are not experienced with the treatment under investigation. I worry intensely when those who are enthusiastic about trials use people who are not enthusiastic about the treatment to get the answer to the question, because the treatment itself is not being assessed—rather people’s ability to provide the intervention.” Before centers are allowed to randomize patients in thrombolytic trials, they should have to demonstrate that their center has good outcomes within standard safety margins, he proposed. Otherwise, “they will be treating people out to six or even eight hours, whose scans they’re not able to interpret,” and end up with high hemorrhage rates. “The only effective treatment for acute ischemic stroke would then be called into question as being dangerous and ineffective, when in fact it doesn’t invalidate the concept that reperfusing brains of people with a good-looking scan and vascular occlusion is efficacious,” he added.
In their own case, University of Calgary researchers, who have had excellent results using a new method of reading CT scans called ASPECTS in combination with hypothermia and thrombolysis, are now reluctant to randomize all stroke patients, Dr. Buchan said. Instead, he proposed to Dr. Warlow and other IST-3 collaborators in Canada that they would only randomize those patients who they were genuinely uncertain of whether or not to treat. Those who they felt should be treated would go into a prospective registry to ensure the results in this group would be available for comparison with the randomized groups, a sort of nonrandomized control group.
The central argument is not whether something works, but how one demonstrates that something works, Dr. Buchan concluded. He also referred to Dr. Sackett’s quote, “An enthusiastic surgical trial lacks controls; a controlled surgical trial lacks enthusiasm.” In the case of t-PA, he says, “those who want to control and randomize have got to be more enthusiastic.”
NR
—Susan Jeffrey
Suggested Reading
Warlow C, Wardlaw J. Therapeutic thrombolysis for acute ischaemic stroke: What is good for heart attacks is still not good enough for brain attacks. BMJ. 2003; 326:233-234.
Return to table of contents
|
|
