Reanalysis of alteplase for stroke stirs controversy
http://www.100md.com
《英国医生杂志》
An independent panel, appointed by a leading US institute to reanalyse the data regarding stroke treatment, has published its findings which they say support the use of thrombolytic agent alteplase (tPA, also known as rt-PA) "to treat patients with acute ischaemic stroke within three hours of onset under the NINDS tPA trial protocol."
Data was reanalysed after an article in the BMJ (2002;324:723-9) sparked off controversy. The original study data reviewed by the panel was published in the New England Journal of Medicine more than nine years ago (1995;333:1581-7).
Critics take issue, however, with the printed conclusions of the analysis published in this month抯 issue of Stroke (2004;35:2418-24).
Jerome Hoffman, professor of medicine and emergency medicine at the University of California in Los Angeles and a leading critic of calls for widespread use of thrombolysis for stroke, said, "I抦 not sure whether the authors of this paper mean to give this therapy a thumbs up, or are taking a more conservative stance, and simply reporting that their reanalysis doesn抰 find evidence to refute the conclusions of the original study.
"While they themselves acknowledge several important limitations, they are looking at this one study only, and not considering the multiple other negative trials; they are not addressing the difference between efficacy in a very sophisticated trial and effectiveness in community practice; and their analysis couldn抰 have been expected to find statistical differences in critical baseline differences, because of woefully inadequate statistical power. But most readers, unfortunately, will surely look at the paper抯 abstract and conclude, simply, that it has validated the findings of NINDS."
"On the other hand," Professor Hoffman added, "no one looking at the data itself could fail to notice the obvious and major imbalances, which had to influence the results. Among patients treated between 91-180 minutes—the only clinically relevant group—more than 80% of patients with the mildest category of stroke at baseline, who had a wonderful prognosis regardless of treatment, happened to be in the tPA group, while almost two thirds of those almost certainly doomed to a terrible outcome, because they had the worst baseline stroke, happened to be in the placebo group."
W Michael O扚allon, professor of biostatistics and chairman of the independent committee, told the BMJ that the panel was not charged with evaluating other studies, saying, "We addressed this particular study . We reviewed other studies but we didn抰 do a detailed comparison of how other studies compared to the NINDS trial."
Professor O扚allon also said that the panellists reanalysed the data in response to "multiple concerns" raised by critics "including the question of baseline imbalance." The panel, he said, found that the "imbalance did not invalidate the results of the trial."
David Schriger, professor of emergency medicine at the University of California in Los Angeles, who along with Professor Hoffman is currently analysing the raw data from NINDS, told the BMJ, "Imagine you are underneath two airplanes that cross directly above you at different altitudes. You might think they collided. So might someone standing a few feet away from you. Someone in a third plane at the same altitude as one of the others would be quite certain that no collision occurred. The original NINDS analysis and the reanalysis are like the two people who are standing near each other. They confirm that from that perspective it appears that the drug may be efficacious, but not necessarily effective. This is only one perspective however, and viewed from other perspectives the drug抯 efficacy is not at all clear. Said another way, the reanalysis checked the math of the original analysis, but not the method."
But Professor O扚allon defends the panel抯 recommendation that health professionals "work collaboratively to develop guidelines to ensure appropriate use of alteplase in acute ischemic stroke patients." He said, "The FDA approved tPA. We are not the ones who approved it. What we examined, and what we found, is that there is no evidence that tPA, if used according to the NINDS protocol, should not be used."
Professor Hoffman responded, "I believe this is a classic error as it confuses failure to show an effect and the claim that no effect exists. Perhaps they haven抰 proved that the NINDS protocol should not be used, but that抯 hardly the same as claiming that it should."(New York Jeanne Lenzer)
Data was reanalysed after an article in the BMJ (2002;324:723-9) sparked off controversy. The original study data reviewed by the panel was published in the New England Journal of Medicine more than nine years ago (1995;333:1581-7).
Critics take issue, however, with the printed conclusions of the analysis published in this month抯 issue of Stroke (2004;35:2418-24).
Jerome Hoffman, professor of medicine and emergency medicine at the University of California in Los Angeles and a leading critic of calls for widespread use of thrombolysis for stroke, said, "I抦 not sure whether the authors of this paper mean to give this therapy a thumbs up, or are taking a more conservative stance, and simply reporting that their reanalysis doesn抰 find evidence to refute the conclusions of the original study.
"While they themselves acknowledge several important limitations, they are looking at this one study only, and not considering the multiple other negative trials; they are not addressing the difference between efficacy in a very sophisticated trial and effectiveness in community practice; and their analysis couldn抰 have been expected to find statistical differences in critical baseline differences, because of woefully inadequate statistical power. But most readers, unfortunately, will surely look at the paper抯 abstract and conclude, simply, that it has validated the findings of NINDS."
"On the other hand," Professor Hoffman added, "no one looking at the data itself could fail to notice the obvious and major imbalances, which had to influence the results. Among patients treated between 91-180 minutes—the only clinically relevant group—more than 80% of patients with the mildest category of stroke at baseline, who had a wonderful prognosis regardless of treatment, happened to be in the tPA group, while almost two thirds of those almost certainly doomed to a terrible outcome, because they had the worst baseline stroke, happened to be in the placebo group."
W Michael O扚allon, professor of biostatistics and chairman of the independent committee, told the BMJ that the panel was not charged with evaluating other studies, saying, "We addressed this particular study . We reviewed other studies but we didn抰 do a detailed comparison of how other studies compared to the NINDS trial."
Professor O扚allon also said that the panellists reanalysed the data in response to "multiple concerns" raised by critics "including the question of baseline imbalance." The panel, he said, found that the "imbalance did not invalidate the results of the trial."
David Schriger, professor of emergency medicine at the University of California in Los Angeles, who along with Professor Hoffman is currently analysing the raw data from NINDS, told the BMJ, "Imagine you are underneath two airplanes that cross directly above you at different altitudes. You might think they collided. So might someone standing a few feet away from you. Someone in a third plane at the same altitude as one of the others would be quite certain that no collision occurred. The original NINDS analysis and the reanalysis are like the two people who are standing near each other. They confirm that from that perspective it appears that the drug may be efficacious, but not necessarily effective. This is only one perspective however, and viewed from other perspectives the drug抯 efficacy is not at all clear. Said another way, the reanalysis checked the math of the original analysis, but not the method."
But Professor O扚allon defends the panel抯 recommendation that health professionals "work collaboratively to develop guidelines to ensure appropriate use of alteplase in acute ischemic stroke patients." He said, "The FDA approved tPA. We are not the ones who approved it. What we examined, and what we found, is that there is no evidence that tPA, if used according to the NINDS protocol, should not be used."
Professor Hoffman responded, "I believe this is a classic error as it confuses failure to show an effect and the claim that no effect exists. Perhaps they haven抰 proved that the NINDS protocol should not be used, but that抯 hardly the same as claiming that it should."(New York Jeanne Lenzer)