Tissue plasminogen activator for acute ischemic stroke

Authors: NINDS r-tPA stroke study group.
Source: New England Journal of Medicine 333:1581-7. December 14, 1995.
Institutions: National Institute of Neurological Disorders and Stroke; multiple, mainly academic medical centers in the US.
Financial support: NINDS.



Overall study design: Study conducted in two identical parts, results then combined. Part I: patients randomized to tPA vs. placebo, assessed for effect of tPA 24 hours after stroke (and then 3 months later). Assuming no major deleterious effect of tPA in Part I (at 24 hours), more patients were then to be enrolled in part II: same randomization, also assessed at 24 hours and 3 months. Both parts were stratified into two equal groups by time from onset of symptoms to treatment: 0-90 mins and 90-180 mins. Although both parts were conducted identically, the primary outcome in part I was the percentage of patients with significant improvement at 24 hours, in part II it was the percentage of patients without significant disability at 3 months.

Eligibility: main criteria: ischemic stroke with clearly defined time of onset, CT scan without hemorrhage. Main exclusions related to risk of hemorrhage (recent surgery, abnormal coags) and BP > 185/110.

Treatment: Placebo or recombinant tPA (0.9 mg/Kg, 10% bolus, remainder over 60 mins).

Measurements: At baseline and 24 hours: NIH stroke scale (NIHSS), a measurement of neurologic deficit on a 0-42 point scale. At three months: three measurements of function: Barthel Index (ability to perform ADL's), modified Rankin scale (overall function) and Glasgow outcome scale (outcome, including death); the NIHSS; and a global test statistic derived from these four measurements.

Repeat CT scans performed at 24 hours and 7-10 days, as well as whenever any symptoms suggested hemorrhage. Major systemic bleeding and deaths recorded.


In part I, 291 patients randomized, equally divided into tPA vs placebo as well as 0-90 mins vs 91-180 mins. In part II, 333 patients randomized, also equally distributed.

Authors' Discussion

tPA yielded significant benefits at 3 months compared with placebo. Other recent, large trials of thrombolytic therapy in stroke yielded no benefit; according to the authors this is because these trials treated many patients more than 3 hours after symptom onset, and/or because of higher doses of thrombolytics and poorer BP control.


G. del Zoppo from the Scripps Institute points out, in an editorial, that although the European Cooperative Acute Stroke Study, another randomized tPA study, found no benefit for tPA, when the results were re-analyzed excluding 109 patients with protocol violations, a similar improvement in the Rankin score was obtained. Dr. del Zoppo also points out that very early intervention may decrease the incidence of treatment-related hemorrhage. He also notes that studies looking only at short-term improvement or mortality may miss a later improvement in functional results, as found in this study.


The two most problematic issues that this study raises, for me, are the necessity to treat within 3 hours of symptom onset including a CT scan and the fact that patients with worse deficits at baseline had a higher tendency to hemorrhage.

Obtaining a quality CT scan, read by a qualified radiologist, so soon after the onset of symptoms is very difficult. It might require triaging probable stroke patients towards institutions that are set up to handle this sort of protocol. A significant number of patients don't receive thrombolytic therapy for acute MI because of delays in diagnosis; MI patients have a window of opportunity felt to be around 6 hours, as a rule, and don't usually require a procedure prior to treatment, making their treatment much easier than that of CVA patients who, apparently, need to be caught within 3 hours and clearly need a scan before therapy.

In this study, patients who had treatment-related intracerebral bleeding had worse baseline neurological deficits. If it turns out that this sub-group of patients does worse with thrombolysis, then this therapy might have to be withheld from the group with the worst outcomes. This would not negate the benefit of thrombolysis for those with lesser deficits, but might mean that even very early intervention would not greatly impact on the disastrous strokes that are so feared.

See also article from Lancet on streptokinase in ischemic stroke.


Reader comments

June 15, 1996

Letters to the Editor about this article from the NEJM

Subject: Rx of CVA with TPA
Date: Fri, 13 Dec 1996
From: arkj@paknet1.pk

TPA is not available in Pakistan and would be too expensive in any case. Can streptokinase or any other thrombolytic be used?

Date: Sun, 02 Mar
From: Mario Cicone <mciccone@nt.net>

In the same issue of NEJM, (Dec 14, 1995, pp 1588-1593), a study by Richard Kay et. al. (abstract) looked at the use of Low Molecular Weight Heparin (LMWH) for the treatment of ischemic stroke. It seems to me that this study was favourable towards its use in the first 48 hours post symptoms. There has been very little criticism in subsequent letters to the Editor regarding this study.

I don't get it. Why has all the focus been on the NINDS t-PA study and not on this one? Should we not be looking to use LMWH in acute ischemic stroke??


Date: June 2, 1999
From: The Bear [ysehgal@cancom.net]

Whenever I see a "drug-company" trial of an expensive drug, I get suspicious.  Doesn't it bother anyone that the outcome measure was changed at three months (they completely made-up a "global score" because the actual outcome measures wouldn't have been positive).  This is totally a matter of data manipulation to prove a predetermined goal and has nothing to do with science.  The lack of effect has been confirmed by several studies into TPA and Streptokinase, and yet people still jump on the tPA bandwagon, and the danger of intracerebral hemorrhage has been confirmed again and again.  This is another reason for journals to have a requirement to disclose drug-company funding.

Yogi Sehgal,
Terrace Bay ON, Canada
reply to: yogman1@hotmail.com

In the article, it is not specified whether the "global statistic" was developed before data analysis or not. However, most of the individual measurements which were components of the global statistic also reached statistical significance.  Thus, it is not likely that the outcome measure was changed at three months, since the actual component outcome measures were, in fact, positive.  --mj

June 30, 2000
From: Charles A. Pilcher, MD

As best as I can calculate it, the NNT for tPA in stroke is about 25% as great as the NNH in the NINDS study. In other words, we'll make 1 patient worse for every 4 we make marginally better. And "worse" means death for half of those made "worse."

The June 21, 2000, issue of JAMA editorializes on stroke centers, where the NNT is estimated to be about 18. I suspect the NNT for Coronary Care Units is a bit better than 18, like maybe 1?

I remain open but skeptical. We're undoubtedly on the right track, and this is only the first battle in the war. To read from most authors, one would think the war was already won.

Charles A. Pilcher, MD, FACEP 
Kirkland, WA


I sympathize with your feelings, although the analogy with CCU's doesn't quite hold up.  If the NNT for CCU's were truly 1, this would imply that every single patient admitted to a CCU would have done worse if not admitted to such a unit, clearly not the case.  Of course, you are right that the "war" is nowhere near won with IV thrombolytics as treatment for stroke.   --mj

Journal Club on the Web home page

Submit a comment about this article

Site Meter