Date: September 5th, 2014
Guest Skeptic: Dr. Ryan Radecki is an Assistant Professor of Emergency Medicine – The University of Texas Medical School at Houston. He has an amazing blog called Emergency Medicine Literature of Note. It is billed as musings on publications and studies relevant to EM & otherwise. You can follow Ryan on Twitter @EMLitofNote.
This is the first episode of season#3 of the Skeptics’ Guide to Emergency Medicine. The goal continues to be to cut the knowledge translation window from ten years down to one year. The SGEM does this by using social media to provide you with high quality, clinically relevant, critically appraised, evidence based information. It is all part of the Free Open Access to Meducation movement. FOAM – Medical education for anyone, anywhere, anytime.
We will continue to provide a weekly podcast critically reviewing a recent publication. The validated and reliable Best Evidence in Emergency Medicine (BEEM) appraisal tool will be used to review the manuscripts. Guest skeptics will be invited on to help and give a variety of opinions. There will be more SGEM Classics with the Swami on historic, must know emergency medicine papers. Cheesy theme music mostly from the 1980’s will be selected to help with the edumatainment factor. And of course, cool skeptical prizes will be awarded to the keenest listener who answers the trivia question at the end of each podcast.
I am also working on an exciting new segment for the SGEM tentatively called “Hot Off the Press”. Can’t provide details right now but I hope it will help cut the KT window to less than one week.
So let us start Season#3 by addressing one of the most hotly debated issues in emergency medicine, thrombolysis for acute CVA. The SGEM does not shy away from controversial issues it just follows the data.
Thrombolytic therapy for acute ischemic stroke does remains widely controversial. A “pro/con” debate in the BMJ was accompanied by a non-scientific poll with results split nearly 50/50 regarding whether the benefits of tPA were adequately proven. Many professional organizations have recommended thrombolytic therapy, with treatment under 3 hours preferred, but treatment up to 4.5 hours accepted. Other professional societies, primarily in Emergency Medicine, have been less enthusiastic about tPA.
The Australasian College of Emergency Medicine suggests additional evidence is necessary, and the Council of the American College of Emergency Physicians recently voted to revise a pro-tPA Clinical Policy statement. Alternatively, many stroke neurologists feel tPA to be beneficial in many patients currently excluded from treatment. This includes the very elderly (>80 years), mild strokes (NIHSS < 5), and time windows up to six hours from onset.
Active clinical trials are underway addressing each of these questions, as well as trials evaluating the utility of advanced imaging to select patients even further from the onset of symptoms.
Case Scenario: 78 year-old man presents to ED with right leg and arm weakness for the last four hours. He has a history of hypertension and dyslipidemia. His vitals are unremarkable except for a blood pressure of 165/95. A non-contrast CT head is performed that shows no acute intracranial pathology. Your stroke team asks you to administer thrombolysis and admit.
Question: Is thrombolysis safe and effective treatment in patients who present with signs of an ischemic stroke of less than 6 hours duration?
Background: Acute ischemic strokes represent the leading cause of disability in our society and the third most common cause of death. There have been many studies performed looking at thrombolysis for acute CVA.
We have covered the original 1995 NINDS article on Episode#70. The bottom line from that manuscript was I was skeptical that thrombolysis has benefit for acute stroke.
I presented 12 major trials for thrombolysis in CVA at the Swedish National Emergency conference early this year at SweetBEEM. To summarize there were four trials stopped due to harm or futility, six showing no benefit, and only two showing benefit. This was not enough proof for me to reject the null hypothesis.
Reference: Wardlaw JM et al. Thrombolysis for Acute Ischemic Stroke (Review). Cochrane Database Syst Rev 2014
- Population: 27 randomised trials, involving 10,187 patients, with definite ischemic stroke.
Intervention: Any thrombolytic agent – urokinase, streptokinase, rt-PA, recombinant pro-urokinase or desmoteplase.
- Comparison: Placebo.
- Outcome: Functional independence at long-term follow up, with safety outcomes of spontaneous intracerebral hemorrhage and death.
Authors’ Conclusions: Thrombolytic therapy given up to six hours after stroke reduces the proportion of dead or dependent people. Those treated within the first three hours derive substantially more benefit than with later treatment. This overall benefit was apparent despite an increase in symptomatic intracranial haemorrhage, deaths at seven to 10 days, and deaths at final follow-up (except for trials testing rt-PA, which had no effect on death at final follow-up).
Quality Checklist for Systematic Reviews:
- The clinical question is sensible and answerable. Yes
- The search for studies was detailed and exhaustive. Yes
- The primary studies were of high methodological quality. Unsure
- The assessments of studies were reproducible. No
- The outcomes were clinically relevant. Yes
- There was low statistical heterogeneity for the primary outcome. No
- The treatment effect was large enough and precise enough to be clinically significant. No
Key Results: In evaluating 27 trials, the authors perform 39 analyses, some with multiple sub-analyses. They also tend to focus primarily on those involving rt-PA, or alteplase, as the approved and guideline-recommended therapy.
- Thrombolytic therapy, up to six hours, dead or dependent – 23 trials, 9,318 participants
- 2679/4891 (54.7%) allocated thrombolysis vs. 2608/4427 (58.9%) allocated control
- OR 0.85 (95% CI 0.78 to 0.93), with significant heterogeneity (I2 = 39%, P = 0.03)
- Thrombolytic therapy, up to six hours, risk of symptomatic intracranial hemorrhage – 25 trials, 10,186 participants:
- 402/5372 (7.4%) allocated thrombolysis vs. 84/4814 (1.7%) allocated control
- OR 3.75 (95% CI 3.11 to 4.51), without heterogeneity (I2 = 7%)
- Thrombolytic therapy, up to six hours, risk of death from all causes – 28 trials, 10,187 patients
- 1043/5372 (19.4%) allocated thrombolysis vs. 856/4815 (17.7%) allocated control
- OR 1.18 (95% CI 1.06 to 1.30), with significant heterogeneity (I2 = 48%, P = 0.003)
These trials included urokinase, streptokinase, and desmoteplase – and modern thrombolytic therapy is typically undertaken with alteplase, or rt-PA. How about the results for just the rt-PA trials?
- rt-PA, up to six hours, dead or dependent – 8 trials, 6729 participants
- 1830/3372 (54.2%) allocated rt-PA vs. 1947/3357 (57.9%) allocated control
- OR 0.85 (95% CI 0.77 to 0.93), with significant heterogeneity (I2 = 45%, p = 0.04)
- rt-PA, within 3 hours, dead or dependent – 6 trials, 1779 participants
- 531/896 (59.2%) allocated rt-PA vs. 603/883 (68.3%) allocated control
- OR 0.65 (95% CI 0.54 to 0.80), without heterogeneity
The authors therefore lead their published abstract conclusion with the sentence: “Thrombolytic therapy given up to six hours after stroke reduces the proportion of dead or dependent people.”
This Cochrane Review update the 2009 version through the addition of the Third International Stroke Trial (IST-3), the largest stroke trial to date. To remind listeners, IST-3 enrolled patients up to six hours, and specifically enrolled those who were not eligible for tPA under the current European license, but where the treating clinician believed tPA was promising therapy.
A minimal initial number were enrolled double-blind, but following the approval of tPA in Europe, Boehringer Ingelheim ceased supply of the study medication. This resulted in transition to an open-label design, in which both the treating clinician and the patient were aware of treatment allocation.
This introduces several forms of bias, including alteration of the type or intensity of other initial treatment based on allocation, a placebo effect after being given the “promising” therapy, and a nocebo effect if randomized to no additional treatment.
Outcome assessments at 6 months were blinded to allocation, but many occurred via telephone or postal mail, relying on un-blinded patients or family members to report functional status. This trial is subject to substantial limitations to internal validity, and its inclusion in this updated Cochrane Review – comprising over half the total number of patients treated with thrombolytic therapy – diminishes the reliability of the analysis as a result of bias in favour of thrombolytics.
There were a number of problems with IST-3:
- Pragmatic, open-label (blinding)
- Small blinded (300) favored control
- Only patients doctors thought would benefit (bias)
- Missed target by ~50% for enrollment
- After 7 years they seemed to move the goal post
- Another statistician was brought in to “persuade”
- Came up with secondary outcome with ordinal logistic regression analysis
- Primary end point was NEGATIVE
- Reported as a positive study – “despite the early hazards, thrombolysis within 6h improved functional outcome. Benefit did not seem to be diminished in elderly patients.
The SGEM did a critical review of IST-3 on Episode#29. The bottom line from that review was treatment with tPA harmed (death) 1 in 25 early, the fatal and non-fatal bleed rate when up significantly and there was no benefit seen at 6 months in the primary outcome.
The authors appropriately note many of these threats in their reporting of potential sources of bias. Only 14 of 27 included trials met criteria for high-grade concealment. Other than IST-3, most were double-blind, placebo-controlled trials – although, the authors note saline placebo is not identical in appearance to tPA, and prolonged incidental bleeding from venipuncture or gingiva in tPA patients could serve to un-blind patients or clinicians. Blinding of follow-up assessment was only explicitly declared in seven trials, which may have resulted in clinicians aware of acute phase events performing the follow-up assessment.
We mentioned earlier that several trials were stopped prematurely as a result of safety, futility, or enrolment issues. ATLANTIS A was stopped after 142 patients due to an excess of symptomatic intracranial haemorrhage. ATLANTIS B was stopped after 619 patients due to futility, and the authors note data has only been publicly presented on 547 of them.
ASK, MAST-E, MAST-I, and MELT were stopped early by their data monitoring committees. AUST was discontinued after slow recruitment. The authors do not acknowledge the distorting effect of early termination when weighting results for pooled analysis.
The authors also downplay the potential bias resulted to trial funding and author affiliation. They note “8/27 trials were run by companies that make the clot-dissolving drugs”, but go on to state most participants come from trials funded by Government or charity sources. However, this statement is only true based on the inclusion of IST-3. The majority of trials comprising the remaining participants, including most trials evaluating tPA, were funded by Genentech and Boehringer Ingelheim. All tPA trial reports were co-authored by individuals declaring conflict-of-interest with one of the two manufacturers.
All trials were performed in specialized stroke centers, and patients were assessed by expert neurologists prior to enrolment and treatment. The generalizability of any of these findings to many practice settings is limited, and observational studies show mixed results regarding safety and rate of treatment of stroke mimics.
The simple summary of this review boils down to the reliability of the evidence, rather than the analyses of the authors. However, after nearly two decades of tPA, most clinicians’ opinions are fully crystalized. Regardless of the conclusions or analyses in this article, few will change practice.
Comment on author’s conclusion compared to our conclusion: The authors write:“Thrombolytic therapy given up to six hours after stroke reduces the proportion of dead or dependent people. Those treated within the first three hours derive substantially more benefit than with later treatment.”
In fact, those treated with tPA after three hours derived NO benefit, according to their analyses. Analysis 1.19 reports death or dependency at follow-up with an OR of 0.93 (95% CI 0.83 to 1.04) versus control. Therefore, even accepting the data at face value, the authors present a misleading conclusion through inappropriate partitioning of data.
The remaining author conclusions, regarding a definite increase in symptomatic intracranial hemorrhage, deaths at 7 to 10 days, and deaths at final follow-up, accurately reflect consistent harms present across all trials.
The authors also call for additional trials aimed at identifying the latest time window, treatment population, reduction in harms from intracranial hemorrhage, and the appropriate environment for practice. However, it may be more appropriate to perform additional studies to narrow the treatment population to maximize benefit, rather than expand the treatment population to the very limits of risk/benefit balance.
The Bottom Line: No true believers – on either side of the issue – are harmed by this publication. Thrombolysis for CVA…I remain skeptical
Case Resolution: You reassess the patient and he seems to be improving clinically. You discuss the potential benefits and risk of thrombolysis. By now it is now over 4.5hrs and the treatment window has closed. The patient is not given tPA and admitted for further care.
This is a challenging case and we are clearly in a realm of uncertainty. There are few risk stratification and prognostic tools out there to predict outcome with tPA. But this patient is far from that idea patient population. His age, hypertension and dyslipidemia are clinical features likely increasing his chance of intracranial hemorrhage.
In these trials in a typically younger population the symptomatic intracranial hemorrhage rate was 7.4%. Just roughly estimating his chance of ICH after tPA is probably 10-12% or even higher with a corresponding increase in mortality.
His unilateral weakness is improving. Is this a transient event? It is difficulty to say so close to the onset of symptoms. Stroke mimics and TIAs also do great with tPA but they also do great without treatment. The level of disability is an important consideration. A patient with some weakness might be reasonable to expect to regain some function with rehabilitation, whereas profound hemiparesis is almost certain to be completely disabled for the remained of his life.
Many individuals will actually make the decision to take that substantial risk of bleeding in order to gain an even smaller chance of independent living. Assessment of pre-stroke functional status is also an important consideration. Then as you say this patient’s presents near the end of this proposed therapeutic window. In general, and in this specific case, I would say the risks even exceed the theoretical benefits.
Clinical Application: I will continue to offer tPA, as recommended by the American Heart Association guidelines, and tacitly mandated by medicolegal consequences if not offered. The manner in which patients are provided informed consent for this therapy dramatically influences their likelihood of electing to receive treatment.
What do I tell my patient? Thrombolytic therapy for acute ischemic stroke is controversial. The risks – severe bleeding and with neurologic worsening – are consistent and known, occurring in about 1 in 12 patients. The benefits are uncertain, and most studies suggestive of benefit were sponsored by drug companies or otherwise biased. Other trials showing harm or no benefit were stopped early when hoped-for results did not materialize. However, a number of patients in all trials using tPA – about 1 in 10 – received noticeable functional benefit from this therapy when given quickly after a stroke. Doubts persist about the reliability of this data, and, unfortunately, it is unclear which patients have the greatest chances of benefit or harm from this therapy. The choice to receive this therapy remains an individual patient decision.
Keener Kontest: Who did the first study for thrombolysis is stroke and when was it published?
If you know the answer than send me an email to TheSGEM@gmail.com with “keener” in the subject line. The first person to correctly answer the question will receive a cool skeptical prize.
- Upcoming conferences:
Thank you to Dr. Ryan Radecki for being the first guest skeptic on Season3 of the SGEM. I hope this was a good way to launch Season#3. You can look forward to a weekly skeptical review of the EM literature. As mentioned, Swami will be back with some Classic papers, expect more great skeptical guests, I will be recording SGEM Journal Club at Queen’s University next month, there will be the occasional 10,000 foot view of an EM/MedEd/EBM issue and a brand new SGEM segment coming soon.
Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.