The European Cooperative Acute Stroke Study III (ECASS III) looked at approximately 400 patients in an IV tPA arm who received the drug between three and four-and-a-half hours after ischemic stroke onset. (N Engl J Med 2008;359:1317.) The investigators measured clinical outcomes at 30 and 90 days using the modified Rankin Scale, and found significantly improved clinical outcomes in the IV tPA group. They acknowledged that IV tPA caused more “symptomatic intracerebral hemorrhage,” but this did not result in a difference in mortality.
If all you read is the abstract, it sounds good, even radical because it attempts to create a new standard of how to treat ischemic strokes that qualify. Unfortunately, its conclusions are problematic, if not deceptive.
The placebo and IV tPA groups are not equivalent. The most important prognostic indicator for ischemic stroke is the baseline NIH score. This was statistically worse in the placebo group. The other statistically different variable in the placebo group was “previous history of stroke.” I wonder if people with worse strokes who have more previously damaged brain might have worse outcomes at 30 and 90 days?
The symptomatic ICH is deceptively “redefined.” In the ECASS III study, the patients had relatively mild strokes that were not high risk for bleeding, and yet there was bleeding in the brain in one of four patients. The operative word in other studies and in this one is “symptomatic” intracranial hemorrhage. By the NINDS criteria for symptomatic intracranial hemorrhage, the ECASS III patients had a symptomatic head bleed rate of one in every 12 patients (7.9%). ECASS I, ECASS II, and now ECASS III are all consistent in that the IV tPA arm causes twice the amount of “symptomatic” intracranial hemorrhage as placebo in patients at lower risk for bleeding.
But listen how the ECASS III investigators spin the result: Only 2.4 percent experienced “symptomatic” intracranial hemorrhage, defined by “experts” who determined whether IV tPA was the “predominant cause of the deterioration.” And while they acknowledge that this rate is 10 times the amount found in their placebo group, they soften this difference by underlining that there is no difference in mortality, without also mentioning that the study is underpowered to show differences in mortality. If one looks at other real-world studies, the marked increases of ICH with tPA result in greater mortality. (Neurology 2006;66:1742.)
To conclude that “symptomatic” intracranial hemorrhage is only 2.4 percent “but mortality was not affected” is just a hair short of frank dishonesty.
The study also dichotomized data, which distorts conclusions. A primary measure of the outcome is the modified Rankin Scale, a 0-6 measure. For unknown reasons, the investigators turned this into a dichotomized scale with 0-1 being “favorable” and 2–6 being “unfavorable.”
A score of 2 is “slight disability,” which is lumped together with “6,” which is death. If one adjusts the scale and makes 0–2 “favorable” and 3–6 “unfavorable,” the advantage of IV tPA over placebo goes away in the intention-to-treat group.
All studies have problems that can be picked apart, but it is disappointing that a study that could be so important is plagued with such apparent errors that one wonders if they were not intentionally deceptive.
Rethinking tPA for Stroke
EMN's Special Report this month looks at the American Heart Association's decision to extend the tPA treatment window for stroke to 4.5 hours. At least one emergency medicine society is on board.
See p. 24.