The December 9, 1995 Lancet contains an article reporting on a randomized clinical trial of streptokinase, aspirin, both or neither in the treatment of acute ischemic stroke. The trial was halted prematurely because of excess early mortality and cerebral hemorrhage in the streptokinase groups. In a separate one-page article, two of the seven-member steering committee present a dissenting interpretation of the trial's results. Finally, in the same issue is an editorial adding background to the controversy.
Eligibility: Admission to hospital within 6 hours of onset of sudden development of focal neurological deficit; no clear contraindication to aspirin or streptokinase; CT scan prior to randomization.
Treatment groups: 2x2 randomization, stratified by center, to treatment with aspirin vs placebo and streptokinase vs placebo. Aspirin: 300 mg daily, through day 10. Streptokinase: 1.5 MU over 1 hour intravenously.
Follow-up: Observation, clinical f/u during 10 days hospitalization. Second CT scan, preferably at day 5. Follow up at 6 months by telephone, disability assessed using a modified Rankin scale; levels 1 and 2 (insignificant or slight disability) deemed favourable outcome; levels 3-5 deemed unfavourable outcome.
Endpoints: Primary endpoints: Effects of ASA vs no ASA and SK vs no SK on combined death and disability at six months. Three secondary analyses: 1) the three treatment groups vs placebo; 2) in-hospital mortality and intracerebral hemorrhage; 3) subgroup analyses. Plan was to randomize 1500 patients; interim analyses to be performed with interruption of trial if any therapy was favored by over 3 standard deviations.
622 patients were randomized into 4 equal groups. The trial was halted after interim analysis revealed a significantly increased incidence of in-hospital morbidity/mortality associated with streptokinase. Among 313 patients allocated to streptokinase (alone or with ASA), there were 83 deaths and 25 symptomatic cerebral hemorrhages. Among 309 patients who did not receive streptokinase, there were 36 deaths and 4 cerebral hemorrhages. The OR for 10-day case fatality, streptokinase vs. no streptokinase was 2.7 (95% CI 1.7-4.3).
At six months, there was no statistically significant effect of either SK or ASA on combined death and disability (63% for SK, 65% for no SK).
Looking at each treatment group vs neither drug, there were significant differences for: 10-day case fatality (SK+ASA higher); cerebral hemorrhage (SK alone and with ASA higher); 6 month disability (SK+ASA lower); 6 month case fatality (SK+ASA higher).
The authors describe a "decrease" in 6-month death and disability in the SK group as a 9% reduction in the odds of death and disability, and as a 24% reduction in the odds for SK alone vs placebo alone, although none of these comparisons reached statistical significance. They describe the increase in 10-day case fatality attributable to SK as an absolute increase of 15%.
The authors also state that there is a higher excess 10-day case fatality increase in the SK plus ASA group than in the SK group alone, based on the individual treatment vs. placebo analysis.
In a dissenting opinion from two members of the trial's steering committee, Drs Tognoni and Roncaglioni emphasize that the trial stopped short at 40% of its original planned enrollment, and thus that data analysis should be limited to the larger groups (SK vs no SK; ASA vs no ASA) and not extended to the smaller individual treatment groups, with insufficient numbers. They recommend a large well-planned trial, with clear provision of risk information to participants and with sponsorship by national health/medical research bodies.
The dissenting comments presented in the same issue relate mainly to how the authors of the main paper analyze subgroups that were initially intended to be much larger.
In my opinion, in addition to these statistical problems the authors of the report put a very positive "spin" on results that are not significant, for instance by referring to "decreases" when describing differences that do not reach statistical significance and by using odds-ratios vs. absolute percent changes to inflate/deflate their numbers.
Nevertheless, these results are not inconsistent with the NINDS study recently published (and reviewed here). That study found no statistically significant difference in mortality initially, but a significant benefit in disability at 3 months. The main difference between the two studies is the much shorter delay from symptom onset to treatment in the NINDS study. Thus, both studies show a "better" result several months out, compared with the early results. Very early treatment seems to be key to obtaining benefit from thrombolytic therapy in acute ischemic stroke.
Finally, an editorial in the same issue of the Lancet gives a fascinating behind-the-scenes look at how the conflict that arose between members of the study's steering committee came out into the open.
This is another example that cerebral infarctions "differ" from MI. It has been documented repeatedly that delay in treatment, or large area of infarction (especially when we see hypodense lesion on CT scan) do expose the patients to a high risk of cerebral hemorrhage. This in my mind puts some black clouds on the treatment because it will restrict it to the patients who will present within 3 hours and probably less than that with a relatively small deficit to start with, with the presence of an "expert neuroradiologist" to be absolutely sure that the CT is normal. Bratina et al. (Stroke 1995;26:409-414) reported current emergency department management of strokes in Houston:
- average time to arrival to ED is 115 m
- from arrival to exam by physician : 28 m
- from arrival to CT scan: 100 m
from above numbers, a major overhaul of emergency response system should be made before treatment will be of any success. Considering that we needed around 50.000 patients in MI thrombolytic trials to prove there is a benefit, I think we have a long way to go before proving that t-PA or streptokinase will have any role in the management of acute ischemic stroke.
July 18, 1996
The results of a similarly designed trial, MAST-E, were just published in the NEJM (July 18, 1996). They confirm the results of this study. MAST-E, from France and the UK, was a double-blind trial of 1.5 M IU of streptokinase vs. placebo (no aspirin arms) and was also halted prematurely, after randomization of 310 patients, because of excess 10-day mortality (34% vs. 18.2 %) and intra-cerebral bleeding in the treatment group. The combined outcome of death and disability at 6 months was similar in the two groups in MAST-E, also. Average time to treatment was 4.5 hours.
Journal Club on the Web home page
Submit a comment about this article