The Third International Stroke Trial: IST-3

The Cochrane systematic review of the 11 complete trials of rt-PA for thrombolysis encompasses 3,977 total patients.  IST-3 enrolled 3,035, nearly doubling our cohort of randomized data.  Unfortunately, this influx of new data does very little to resolve any of the outstanding issues regarding stroke care.

Before even looking at the results, it’s particularly important to wade through the dense study design and methods – and realize this is a non-blinded study in which patients were enrolled if the treating clinician was “uncertain of the benefits or harms of TPA”.  Considering this study began back in 2003, prior to ECASS III, a large chunk of their enrolled patients fell into the 3-4.5 hour time frame, with the remaining majority falling into the up to six hour limit.  The other major area of interest this study was intended to evaluate was the efficacy and safety in patients aged >80 years of age, of which they enrolled 1,616.  And, in a shocking twist, this study actually manages to enroll TPA and control cohorts with nearly identical baseline variables.

IST-3 is negative for the primary endpoint, which is the proportion of patients functionally independent at six months (Oxford Handicap Score 0-2, a scoring system similar to the Modified Rankin Score), with a 95% CI of 0.95 to 1.35.  On ordinal secondary analysis, there are non-significant trends towards improvements in OHS favoring rt-PA, which is probably what you’ll hear when people refer to IST-3 as “positive.”

Then, regarding the patients aged >80, there is a trend towards benefit with TPA, CI 0.97-1.88.  Unfortunately, in a neutral study, that means there is actually a trend towards harm in ages <80, CI 0.67-1.26.  Likewise, between 4.5-6 hours, there is a trend towards benefit with TPA, CI 0.89-1.93.  Therefore, between 3 and 4.5 hours, there is a trend towards harm with TPA, CI 0.50-1.07.  TPA is also essentially neutral or trends towards harm up until NIHSS 14, with more pronounced benefit shown in severe strokes.

Interestingly enough, the “blinded” phase of the study trended towards favoring control, CI 0.42-1.98, while the open phase favored TPA, CI 0.89-1.45.

So, what does this all mean?  It means, there’s still plenty of shades of grey open for interpretation and discussion.  Indeed, when added into the systematic review, IST-3 brings several of the previously significant benefits back into the nonsignificant range.  To me, this reinforces what I’ve been arguing for awhile – that the focus shouldn’t be on massive expansion of TPA eligibility, but specifically targeting those who have the best benefit/harm profile.

As with any major stroke trial, many of the investigators have financial associations with Boehringer Ingelheim.

“The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (the third international stroke trial [IST-3]): a randomised controlled trial”

5 thoughts on “The Third International Stroke Trial: IST-3”

  1. We know that all stroke patients will die, just as all people reading this comment will die. If an intervention affects mortality, eventually that effect will wane as patients age because the outcome of death is inevitable. Generally, the further we travel in time from the randomization point, the less effect we can expect an intervention to exert on an inevitable outcome like mortality. And correspondingly, the closer in time to the randomization point, the more responsible we would expect the intervention to be for such an outcome. Additionally, the older the studied population the sooner we would expect mortality curves to approximate. In IST-3, the study population was old, with 57% above the age of 80. Percent dead mortality measures were performed at 7 days and 6 months. Six months after randomization, mortality rates were equivalent in the intervention and control arms (about 27%), but at 1 week nearly 60% more patients had died with tpa than with placebo. This amounted to an absolute risk of death of 4%, or a number needed to harm of 1 in 25. The temporality argument warns us about a high mortality rate early despite approximation of mortality curves late. The eye-test does too: 87 year-old patients dying 5.5 months after receiving a drug doesn’t seem to equate with a dramatic mortality difference seen in the first week. As death is inevitable, mortality measures are in actuality only a gauge of the delay of death. A more telling statistic than percent dead at a point in time would be days of survival during a given time period. We do not have this data from IST-3, but if we extrapolate 7-day data, we can imagine the placebo group to have survived a significantly greater number of days than the tpa group during the study period. If we use only the data from IST-3, it seems we would be obligated to inform our patients in the face of their emergent tpa decision, that stroke is bad, that 1 in 4 patients can expect to be dead in 6 months, that if we do nothing your chance of death within a week is 1 in 14, and that if we administer tpa it is 1 in 9.

    To designate the intervention as harmful requires acceptance of an assumption, that death is undesirable. This is generally assumed, but it could be persuasively argued that severe neurologic disability is worse than death. Unfortunately, we do not have the data from IST-3 to determine if tpa reduced “undesirable survival” in favor of death or desirable survival in a quantifiable manner such as “days of desirable survival.” But in anticipation of such a claim, I would ask are we really at the point where tpa advocates would be willing to take this leap? To promote their miracle drug as one that kills patients to leave them less disabled? There are cheaper ways.

  2. Love your comment Greg Press! You are alluding to the affect referred to as "harvesting" or "mortality displacement". Essentially, TPA killed off the people who would have died anyways, to reach similar mortality rates at 6 months.

    Also – why were "days in the acute care setting of the hospital" (which costs a lot of money) not compared between groups? My suspicion is that they were increased. People with hemorrhages don't go to rehab facilities very quickly. And there were 5.8 more per 100 patients!

Comments are closed.