The arguments I raised are actually those I most commonly find on EM blogs and anti-tPA EM docs on reddit, etc. I actually read a couple of SGEM articles before writing my original post to refresh myself on the form those arguments take these days. If you saw a point that was a true straw-man misrepresentation, please point it out. But declaring "straw-man!" and then not specifying where isn't particularly helpful. I do admit that the arguments I summarized were done in a glib fashion which may have reflected not only my opinion of the quality of the arguments, but also the quality of the people typically making those arguments.
By all means, if there's a particularly compelling argument I've left out that you'd like to discuss, make it or link us to it. I don't pretend to have exhaustively responded to every anti-tPA argument ever made, as my experience of doing so on various threads full of EM people has often resembled attempting to fact-check a Trump debate or fighting the Lernaean Hydra.
This was a very reasonable concern in the early days of tPA - in particular a concern for NINDS (I don't recall similar issues for ECASS-3). This prompted
reanalysis with covariation for baseline stroke severity among other relevant factors (finding adjusted OR of good outcome similar to unadjusted OR), as well as pooled analyses once other trial data became available. The fact that the NINDS results appeared to hold up when applicable subgroups from ECASS, ATLANTIS, and IST-3 were added in further argues that this is not a spurious effect from unbalanced groups.
That isn't clear at all. IST-3 showed increased short-term mortality, but remember that the point of that study was to use far more generous inclusion criteria (particularly temporally). NINDS did not show worsened mortality (actually had a trend toward improved mortality at all time points), and
pooled analyses have not shown worsened mortality at meaningful timepoints. Acute mortality is too noisy to extrapolate in stroke just as acute NIHSS changes are - the natural history of the disease involves fluctuations, and onset-to-treatment times are unsalvageably confounded by the influence of stroke severity on how quickly strokes are recognized and brought in.
NINDS was plenty definitive to alter assessments of clinical equipoise and futility analyses for other concurrent studies. That's supposition on my part, though.
The pre-NINDS studies that were stopped early for harm are generally not those included in pooled analyses. Post-NINDS studies like ATLANTIS and IST-3 that were negative had data collection set up with the intention of being compatible with NINDS and ECASS for later analysis. Skew toward positivity in those studies would only be a reasonable assertion if randomization had produced very different group sizes or characteristics, which nobody has argued to my knowledge.
Yes, which is why generating compatible data from other trials for purposes of pooled analyses was so important, and why
the AHA leans on those pooled and meta-analyses rather than just on NINDS-2.
Those two things aren't really mutually exclusive. The history of acute stroke trials is full of studies that overlapped in time frame but each generation built off the others, from the early days of heparinizing strokes and using therapeutic windows derived from ACS literature, to gradually refining until we reached NIHSS. Had NIHSS been a negative study based on the part 1 data alone, it just would have been the next trial that would have looked at the part 2 subgroup outcomes and designed their inclusion criteria and outcomes accordingly, and we would have been giving tPA since 1998 instead of 1995. This is an altogether different problem than the implied multiple comparisons argument raised by EM blogs, and that argument is further dismantled by the extensive metaanalysis literature by Cochrane and others.
I don't know where you practice, but where I did residency and still practice, the residents have the original NIHSS criteria on their stroke cards for the 0-3 hour window, and only minor modifications to the ECASS criteria for the 3-4.5 hour window. Stroke protocols can vary by institution, but when I was a chief on the stroke service I took calls from a very wide geographic area and never remember anyone trying to push tPA in a cowboyish fashion while ignoring evidence-based criteria. Neurologists and cowboys tend to be a non-overlapping Venn diagram, in case you haven't met many of us.
Furthermore, I don't know where you're getting your exclusion criteria - posterior fossa strokes were not excluded from NINDS, nor were dissections or prior strokes (outside of 3 months).
Also, just as a stylistic aside - referring to stroke as "CVA" is about as correct as referring to diabetes as "the sugar" and leaves a similar impression with most adequately trained neurologists you'll meet. In what other organ system is it acceptable to refer to an ischemic event and a hemorrhagic event with the same lumped abbreviation? Can you imagine signing a patient out to your colleague and saying "well it's either mesenteric ischemia or a huge lower GI bleed, I can't be arsed to think about the difference so let's just call it a GIVA"?
My understanding of the SK data is that the bleed rate was so much higher than anything seen in any alteplase study that it drowned out any improvement signal and had to be stopped very quickly. And yes - the pathophys in stroke is very, very different than ACS. We've learned that the hard way in stroke, as every time we've assumed heart = brain we've been burned, going back to the bad old days of heparinizing acute strokes. Brain tends to bleed into itself, heart doesn't. Lose 5% of myocardial tissue and unless you're really unlucky from a conduction standpoint, you're looking at some mild apical hypokinesis or similar, while losing 5% of brain tissue is quite likely to leave you disabled.
ATLANTIS and ECASS-III differed by more than that 1/2 hour - ATLANTIS had a different endpoint (NIHSS <1 - kind of dumb in retrospect but this study lasted forever and went through revisions based on NINDS outcomes), was originally designed for 6 hour time window before revisions, and included people ECASS didn't - the biggest being that NIHSS > 25 were in ATLANTIS but not ECASS-III. It's not as simple as that 4.5-5 hour window being bad (although that's about where pooled analyses show OR reaching 1), and it takes some pretty high-level statistics and data from multiple studies to suss that out. Plus, if we were actually cherry-picking, wouldn't we figure out ways to throw away the negative studies instead of including them in our pooled data?
If you want to argue against the 3-4.5 hour window, however, I will admit that the data there is on much weaker footing and it's even more important to be wise in those situations.
NINDS actually did show short term benefits - mean NIHSS of 14 in both groups went to 12 in the placebo group and 8 in the tPA group at 24 hours out. Wasn't statistically significant because that window is incredibly noisy in both directions, hence why we rely on the 90 day data. However every subsequent analysis I've seen has found trends in that direction.
FYI, thrombectomy is also noisy. For every amazing recovery I've seen (and I've seen some amazing ones, believe me) there are multiple others that don't do much or worsen.
Part of the reason is that the data is simply stronger than most in your specialty appear willing to admit these days. Part of the reason is that all of the diagnostic testing that might better risk-stratify people takes time, and that's a problem. There actually has been a fair amount of work on looking for risk factors for poor outcome post-tPA such as microbleeds on SWI, but the evidence that this helps more people than it hurts based on the amount of time involved isn't there.
I think a lot of us are asking the same thing. The current AHA guidelines have a weak recommendation for choosing it over alteplase, particularly in mild stroke and those going for thrombectomy. The data I've seen looks pretty decent, though I haven't done anything approaching a full review. However it's not part of our stroke protocol, and I've never seen someone come here after getting it elsewhere. My guess is that this is a matter of inertia more than anything, but I would have to check with my friends that do stroke full-time. I'm just a lowly movement disorders guy.
