You may have seen these two trials. The most important thing to be said about these is that they are not trials.

They are reports of voluntary registry data (caveat – the european license dictated the setting up of these registries but that doesn’t mean it’s enforced) of patients who recieved thrombolysis. This is purely observational data of patients who someone thought should get tPA.

Despite this the authors do multiple comparisons with historical controls – chosen favorably from amongst the prior trials.

This is comparing apples and oranges. I hope it’s clear that the patients getting tPA in the SITS papers are different patients than those in the other papers. Therefore it is not fair to claim comparable efficacy.

This is a major objection to the thrombolytic literature in general, and in particular it’s combination in meta-analyses. The NNT summary covers this idea well.

The ATLANTIS trials

The ATLANTIS trial was presented in two parts. Initially it was intended to be one trial, looking at tPA form 0-6 hours.

Two things happened:

  1.  Once the NINDS study was published, the investigators scapped the idea of giving placebo to the o-3 hr patients as this was considered unethical.
  2. the safety monitoring committee found concerning results in the 5-6 hr patients given tPA.

Thus Atlantis A was a trial of 0-6 hrs that was stopped and Atlantis B was a re-launch taking patients at 3-5hours.



had similar inclusion/exclusion criteria to NINDS, but allowed for the investigators to exclude patients with “Any other condition that the investigator feels would pose a significant hazard to the patient if rtPA therapy were initiated”


They had two primary outcomes (again this is a little bit naughty)

  • a reduction in NIHSS at 24hrs and 30 days
  • a reduction in infarct size

I would contend that both of these outcomes scream SURROGATE MARKER.


most were thrombolysed after 4 hrs. This is understandable – it was the early 90s after all; there weren’t many stroke teams about.

The strokes were less severe (a trend in every trial since NINDS) with a median NIHSS of 10 or 11 (it was 14 in NINDS)

Of the patients with an NIHSS>20 who got tPA every one of them (16 of them) was dead by 90 days.

I hope you’re convinced that this was a thoroughly negative trial by now



They took patients from 3-5 hours and changed the inclusion criteria subtly. Patients with a CT showing greater than 1/3 of the MCA territory involved were excluded. The basic effect is to exclude the bigger strokes. This is now ubiquitous in the newer trials.


they had the ambitious primary outcome of an NIHSS of 0-1 (on a 42 point scale!) at 90 days. In the secondary outcomes they had the modified rankin scale that we’re probably more interested in.


they recruited 600 patients, most of whom who got lysis at about 4 hrs or so.

Again they were less severe strokes – NIHSS of 10 in this trial.

Their primary outcome makes fascinatiing reading

  • 32% of placebo patients and 34% of tPA patients had an NIHSS of 1 or less at 90 days.

That deserves comment, even though it’s a recurrent finding in all the trials. People with strokes sometimes do great no matter what we do. Or perhaps more accurately. People with strokes will often (if you count a 1/3 of the time as often) do really well with good nursing care and good stroke management. Let me quote the cochrane review of stroke units (comprising, dedicated nursing, medical, OT and physio and feedings care.)

Stroke patients who receive organised inpatient care in a stroke unit are more likely to be alive, independent, and living at home one year after the stroke

Over all in the studies the placebo group often does well (MRS <2 in 35-45% of placebo patients). We all know people with severe strokes who are “prisoners in their own bodies” but this is probably not as high a percentage as you thought it was. They are also not likely to be eligible for tPA anyhow (NIHSS greater than 25 or a nasty CT)

As for secondary outcomes, like that pesky one – death, placebo won with mortality of 7% for placebo and 11% for tPA

These trials are important when it comes to reading ECASS III as it claimes postive effects on patients treated beyond the 3 hr mark, whereas these trials showed that people did really poorly beyond the 3 hr mark.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.