A walk down memory lane reminds us that we have yet to see a stroke study of sufficient size to end the tPA debate. And even when we had such data, it was largely ignored. Let’s study our history to avoid repeating mistakes.
A walk down memory lane reminds us that we have yet to see a stroke study of sufficient size to end the tPA debate. And even when we had such data, it was largely ignored. Let’s study our history to avoid repeating mistakes.
I don’t believe that anyone knowingly repeats history, particularly bad history. We either don’t know about the past or don’t recognize that we are traveling a path that others have already forged. Despite the fact that the trails and foot prints left behind by tPA for the treatment of myocardial infarction are worn, if you look closely enough, there is clear evidence that prior researchers, clinicians and pharmaceutical demigods have come before us. Although I won’t belabor the specific point, it is worth stating that I am still unsettled by the notion of tPA for ischemic stroke. That indigestion led me to revisit the history of this drug, particularly with respect to its first application, acute myocardial infarction. I think we might start to see a pattern here – a trail to avoid. See what you think.
First, you might be thinking that this is a futile line of thought. Isn’t tPA in AMI accepted medical care? Yes, but even if we acknowledge that tPA for AMI is the current standard of care for reperfusion when PCI is not available, there are far too many examples of doctors becoming convinced a treatment was the right thing to do, only to later conclude the opposite. My list includes: MAST trousers, high dose steroids for spinal cord injury, sodium bicarbonate for metabolic acidosis, calcium in cardiac arrest, COX-2 inhibitors, Xygris for sepsis and Swan-Ganz catheters to name just a few. So, it behooves us to always go back and ask whether our conclusions about the best practices were, in fact, based on the best evidence. Sometimes the most compelling evidence becomes available long after decisions have been made, and such is never fully considered.
Just because the fight was abandoned in the 1990s doesn’t mean that tPA was really better than its predecessor streptokinase. Let’s go back in time and take a second look at the streptokinase/tPA debate. Most of the studies comparing tPA to streptokinase were equivalency trials, attempting to prove that tPA was as effective, but not superior to, streptokinase. Despite nearly 10 times the cost and an increased risk of ICH in those older than 70 years old, tPA was ultimately determined to be as effective. Streptokinase was ultimately abandoned as a thrombolytic for AMI. But, was that decision prompted by the evidence, or clever marketing?
Looking at a sample of the literature from the late 1980s and early 1990s, several authors felt that the case for tPA for AMI had not been made. Before there were any concerns about the skyrocketing cost of medicine, these authors were making the case for a cost/benefit analysis that is very necessary today.
“Assuming equal safety and efficacy, streptokinase may remain the drug of choice for most patients because of its lower cost, ” wrote M. W. Rich in the American Heart Journal in 1987.1
“Based on analysis of the published megatrials, SK is a more cost-effective thrombolytic agent for patients with acute MI than tPA.”2
GUSTO-I, historically, could be considered the turning point, where tPA won the battle over SK. However, GISSI-2 and ISSIS-3 showed no benefit from tPA with comparison to streptokinase. GISSI-2 included 12,490 patients, tPA versus streptokinase, with and without heparin. There was no difference in the combined end point of death plus severe left ventricular damage3. ISSIS-3 randomized 41,299 patients to receive either tPA or streptokinase with or without heparin. All patients received aspirin as well. Again, no differences were noted in 35 day and 6 month mortality rates4.
Regarding stroke, the NINDS trial, in 1995, was the first reported “positive trial.” The later tPA v. streptokinase trials were designed to prove superiority. GUSTO-I, historically, could be considered a corollary to NINDS. But let’s take a closer look at GUSTO-I. This study looked at 41,021 patients. The primary difference in this study was using the accelerated delivery process (e.g. given over 90 minutes with two-thirds in the first 30 minutes). The primary end point was 30-day mortality, which was 6.3% and 7.4% for accelerated tPA with intravenous heparin and streptokinase with intravenous tPA, respectively. They reported this as a 14% reduction in mortality. However, caution needs to be exercised reporting a percent of a percentage. Really, the difference was 1.1%. A one-year follow up study was performed finding the same difference in mortality. The 1-year mortality rate was 9.1% v. 10.1% for tPA v. streptokinase, respectively5.
This was declared a victory, and for most, the debate ended. However, some questioned the validity of the interpretation of the GUSTO-I results. In short, when you have such a large sample, is a 1% difference in mortality just as likely to be due to chance? It took a huge number of patients to show a very marginal difference. In addition, with similar numbers in ISSIS-3, no such difference was observed. Perhaps, this difference was related to accelerating the regimen. Nonetheless, a clear benefit from tPA v. streptokinase was not proved.
Skeptics continued to question the data, but most finally either gave up or retired. So, despite all of our assumptions to the contrary, a meta-analysis of 14 trials, including GUSTO-I, GISSI-2 and ISSIS-3 evaluating over 142,000 patients showed no mortality difference between tPA and streptokinase6.
Similar to Gusto-I for AMI, The NINDS trial in 1995 was the break out study for the use of tPA in stroke. But the trials using tPA in ischemic stroke were not equivalency trials, such as the ones that compared tPA to SK in AMI. There was nothing to compare to this new therapy. Similar to the first studies of tPA in AMI, tPA in stroke had its questionable outcomes (i.e. ECASS 1, JAMA. 1995 Oct 4;274(13):10 and ECASS 2, Lancet 1998; 352: 1245-1251.). As a result of these studies, more than one author raised doubts of tPA’s efficacy in stroke7. Others have reevaluated NINDS and even ECASS-III, the two reportedly positive studies for tPA and stroke, and have found differing results or opinions8,9.
A favorably reported meta-analysis was published by Lansberg in 2009. However, this analysis was limited to the efficacy of tPA in the 3-4.5 hour window and only included the ECASS I, II, III and ATLANTIS trials. The total number of patients was only 1,622 patients. The authors concluded that the less favorable ATLANTIS study results were due to chance. “The lack of efficacy of tPA therapy in the 3 to 4½ hr cohort of the ATLANTIS study is likely due to chance as there are no obvious differences in study design between ATLANTIS and the ECASS studies to account for this lack of efficacy.”10 Isn’t it just as likely that the benefit seen in ECASS-III was due to chance? They didn’t seem to consider this, biasing their evaluation toward efficacy.
We have yet to see a stroke study of sufficient size to end this debate. What is a worrisome piece of history is that even when we had such data with GUSTO-I, showing no real difference, this data was largely ignored and tPA became the new standard for thrombolysis in AMI. The unintended consequence of that move was that it opened the door for future indications for this drug. Had streptokinase won the battle, would NINDS have ever occurred? We’ll never know.
Although there are clearly limitations to the corollary I’ve d
escribed (this is a selective review of the literature), I submit there are similarities which we may be ignoring. The subject matter is similar, and the research history is similar. Are we traveling down the same road that Genentech paved for us in the 1990s?
References
1. Rich MW. TPA: is it worth the price? Am Heart J. 1987 Nov;114(5):1259-61.
2. Intravenous thrombolytic therapy in myocardial infarction: an analytical review Clin Cardiol. 1993 Apr;16(4):283-92.
3. Lancet. 1990 Jul 14;336(8707):65-71.
4. Lancet. 1992 Mar 28;339(8796):753-70.
5. GUSTO-I Circulation. 1996 Sep 15;94(6)
6. QJM: An International Journal of Medicine Volume 96, Issue 2 Pp. 103-113.
7. Bath P. Alteplase not yet proven for acute ischaemic stroke. Lancet 1998; 352: 1238-1239.
8. Hoffman JR, et al. A graphic reanalysis of the NINDS Trial. Ann Emerg Med. 2009 Sep;54(3):329-36, 336.e1-35. Epub 2009 May 23.
9. Lyden P. Thrombolytic therapy for acute stroke–not a moment to lose. N Engl J Med. 2008 Sep 25;359(13):1393-5.
10. Stroke. 2009 July; 40(7): 2438–2441.
Kevin Klauer, DO, EJD Editor-in-chief of Emergency Physicians Monthly, CMO of Emergency Medicine Physicians, Vice Speaker of the ACEP Council.
1 Comment
Kevin,
I agree. This always struck me as nothing more than great marketing, perhaps even disingenuous (to be kind) great marketing.
FYI, I think there might be a mis-print above regarding GUSTO-1. It says, “The primary end point was 30-day mortality, which was 6.3% and 7.4% for accelerated tPA with intravenous heparin and streptokinase with intravenous tPA, respectively. “
I believe the sentence should end with “streptokinase with intravenous HEPARIN, respectively.”