Putting the Patient First: Comments on Scientific Rationale for the Inclusion and Exclusion Criteria for Intravenous Alteplase in Acute Ischemic Stroke

The American Stroke Association (AHA/ASA) has released a Scientific Statement entitled “Scientific Rationale for the Inclusion and Exclusion Criteria for Intravenous Alteplase in Acute Ischemic Stroke”, which was affirmed by the American Academy of Neurology as an ‘educational tool’ and endorsed by the American Association of Neurological Surgeons and the Congress of Neurological Surgeons. This scientific statement, written by Demaerschalk et al, is published almost 20 years to the day from our publication of the National Institutes of Neurological Disorders and Stroke (NINDS) trial of recombinant tissue plasminogen activator (rt-PA) for acute ischemic stroke in the New England Journal of Medicine.¹ The day prior to the New England Journal publication, trial leadership held a press conference at the NINDS in Bethesda, MD, to announce these landmark results. My most vivid memory of that press conference is focusing really hard on not tripping over the television camera cables as we authors walked into the packed room. By prior arrangement, the elders did the talking: Tom Brott, Jim Grotta, John Marler.² During the preparation beforehand, I recall being instructed to avoid any over-statements should I be asked any questions; in the event, alas, one sports analogy did manage to find its way into some major media outlets: ‘we expected to hit a single or double and we ended up with a grand slam’. Oops.

Given that the trial showed such a phenomenal and impressive result, why wasn’t it a ‘grand slam’? Why has it taken 20 years—and to be honest the publication of successful neurothrombectomy trials—to really unleash intravenous thrombolytic therapy? The effect size (both relative and absolute) swamp anything previously seen in Neurology—uptake of this therapy should have been swift and sure.³ Did initial professional envy and a “not invented here” syndrome block wider/faster adoption? Was uptake inhibited by the self-appointed demagogues who were hawking their newsletters and bashing science they did not understand? Or did the published selection criteria pose too steep a barrier for most neurologists to embrace IV rt-PA, as suggested by Demaerschalk et al?

The scientific rationale for the inclusion and exclusion criteria discussed by the AHA/ASA writing group is based on exhaustive summarization of the literature, most of which did not exist during the writing of the original trial protocol. The inclusion and exclusion criteria used in the original trial derived from a Delphi-like consensus process, not from experimental evidence. My colleagues and I envisioned our trial to be a Phase 2 dose confirmation and safety trial only, one that would lead to a better-designed, Phase 3 pivotal confirmation of efficacy. Given the widespread fear of thrombolysis at the time (late 1980’s), we opted to err on the side of conservatism, but in no way were the original criteria derived from any realistic sense that violations would lead to bleeds or other adverse events. Rather, the thought was, let’s get safely past a small dose confirmation study and proceed to the definitive trial thereafter. Never in our wildest imagination did we foresee our conservative Phase 2 safety trial being the pivotal success that it was.

Thus, in 1996, the US Food and Drug Administration (FDA) adopted a label indication for IV rt-PA for acute ischemic stroke that incorporated our non-evidence based, overly-conservative, interim-not-intended-to-last-forever selection criteria, nearly verbatim.⁴ Some of us testified at an FDA advisory panel against using the clinical trial criteria as the indications for clinical use, for two reasons. First, as mentioned, our clinical trial criteria were based on consensus, not data, and were overly conservative. Second, we felt—as asserted by Demaerschalk et al—that the selection criteria might prove an impediment to swift adoption of the treatment.

The drug was approved, marketed, and over the intervening years the use of IV rt-PA for acute ischemic stroke has slowly but inexorably increased. Today no rational physician doubts the benefits, or that the risks are far outweighed. So is there really a need for an exhaustive review of the ‘Scientific Rationale’ for the selection criteria as published here? The writing committee felt that their work is timely and necessary and could serve to assuage the fears of practicing physicians faced with the patient in potential need of IV rt-PA. More importantly, the document explores “…some popular myths regarding treatment”, and contra-dogma myth busting is always welcome. I note in passing, however, that these myths have been busted before^{5, 6}; perhaps this time the readership is ready to reassess dogma, and accept data where before only emotion proved persuasive.

A few issues in the document require comment. First, selection criteria for a clinical trial are hard and fast; those in clinical practice are not. While the committee implies this, it is never explicitly stated. In clinical practice, a physician must use judgment in applying clinical trial results to the individual—selection criteria cannot be viewed as a ‘hard-stop’ without careful consideration of all the factors present in that patient. Whether any individual patient will benefit from the therapy is probabilistic—to the extent the patient satisfies the clinical trial entry criteria, they will have a higher likelihood of benefit. We cannot promise any single patient they will benefit, but we can predict they are more or less likely to benefit. Thus, trial entry criteria such as age limits or severity scores serve only to point the clinician toward patients more likely to benefit; such limits should not be used to exclude individual patients from therapy. For example, consider age: as the writing committee suggests, “Age is one of the most important factors influencing the risk of stroke…” On the other hand, we know that age has little to do with the potential for benefit from IV rt-PA for acute ischemic stroke. As we showed in 1997, although the absolute benefit of rt-PA declines with increasing age, the relative benefit of rt-PA compared to placebo is independent of age.⁷ The IST-3 investigators confirmed this.⁸ In other words, even though the absolute response rate declines with age, potential for treatment benefit does not. Therefore, despite prevailing dogma that older patients do not benefit from IV rt-PA, or are more likely to bleed, the writing committee correctly recommends that clinicians avoid using blindly any upper age limit to exclude patients.

Second, the writing committee exhaustively summarizes ALL available data, discussing together data from clinical trials as well as observational data collected in registries and case series. When it comes to evaluating a putative therapy, data from uncontrolled case series and registries is close to useless. Large registries help us answer questions such as “what happens to older patients after stroke”. Without randomized controls, such registries contribute nothing to answering questions such as “do older patients respond to rt-PA as younger patients do”. That is, if you compare rt-PA treatment response in old vs. young patients without controls, you will find a lower proportion of responders among the aged group. The uncritical observer might therefore conclude rt-PA is less effective in the aged. As Demaerschalk et al eloquently illustrate, despite a lower overall absolute benefit the effect of rt-PA treatment compared against placebo is at least as strong as among younger patients,⁹ Clinicians are inundated with review articles, subgroup analyses, and pseudo meta-analyses that get this point dead wrong, so it is understandable that confusion arises at the bedside. The data presented by Demaerschalk et al clearly debunk the dogma here. As another example, consider stroke severity, as scored with the National Institutes of Health Stroke Scale (NIHSS). Popular dogma says that more severe patients (greater NIHSS score at presentation) respond less well to rt-PA, compared to less severe patients (lower NIHSS). Again, while the overall rate of good outcomes is much lower in severe patients, the response rate to rt-PA is the same or perhaps even better.^{7, 9} The writing committee clearly and rigorously summarizes the relevant data and confirms there should be no severity exclusion for selecting patients for IV rt-PA.

Third, the risk of intracerebral hemorrhage after IV rt-PA causes significant angst at the bedside. When we published the NINDS rt-PA for Acute Stroke Trial, the vast majority of neurologists were, in a word, “old school”. We were highly skilled in the arts of history taking, physical exam, lesion localization and diagnosis. We were not, as a group, prepared or trained for the demands of rapid-fire assessment and time-compressed decision making using only a very limited data set. One could argue that the decision to administer rt-PA intravenously is one of the most challenging decisions in medicine: there is very little time to assemble the data (history, physical, lab, imaging); the drug has some risk; and once it’s administered it cannot be retrieved. Most other interventions allow the physician some wiggle room once the intervention begins. Not so with an intravenous medication that has a serum half-life of a few minutes—once we make the decision to infuse, there is no going back. One of the few things more frightening to most physicians than the brain is blood, so when it comes to blood IN the brain, rational physicians become more emotional. Neurologists can bring clarity, and data, to the bedside. Despite popular dogma and fears, the majority of symptomatic intracerebral hemorrhages (sICH) occur in the patients with moderate to severe lesions already.¹⁰ It is critically important to consider the risk of sICH, but it is also critical to calmly and dispassionately consider risks vs. benefits. As fully described in this AHA/ASA statement, the risks are far outweighed by benefits in the majority of patients.

Fourth, this statement carefully and rigorously puts to rest any doubts about treating “mild” patients. Far too often, suitable candidates are deprived of a chance at recovery due to the mistaken assessment of a ‘mild’ deficit. In landmark papers, investigators showed that of patients excluded from IV rt-PA treatment on the grounds of “too mild” symptoms, over 1/3 were dead or disabled 3 months later.¹¹ There is no lower boundary on the NIHSS stroke scale, below which one should not treat.^{12, 13} A significant contribution to this problem is that the NIHSS fails to detect cognitive dysfunction. Also, deficits need to be put in the context of the patient’s occupation and avocation. A visual field cut in a truck driver could be career-ending, yet score only 1 or 2 points on the NIHSS. The statement recommends treatment for any deficit considered disabling. Similarly, excluding patients who are ‘rapidly improving’ skirts disaster. The statement provides a superb summary of this complex issue and if one were to read only one section, this would be it.

And so on. From pregnancy to platelets, this statement provides a thorough, one-stop shopping opportunity to read the data comprehensively and comprehend the safety of IV rt-PA in the presence of one or more “exclusion” criteria. If I could sum up the wealth of these data in one phrase, it would be that this statement puts patient benefit first, ahead of dogma or mythology. It has taken 20 years, but facts known from the very beginning have finally overcome uncritical opposition, fear, and conflicts of interest. Additionally, we as a specialty have progressed from diagnosis to intervention: more patients are being treated; more neurologists graduate from training with competence in thrombolysis; more medical students are interested in joining our ranks. This statement makes clear a lesson known from the earliest days of thrombolytic therapy: the risk of withholding treatment from eligible patients far outweighs the risk of giving the drug.¹⁴ We seek to avoid doing harm, and must be mindful that on occasion, doing nothing is more harmful than attempting a treatment with a favorable risk/benefit profile.

One final thought: the debate over IV rt-PA illustrates to me a larger question of the role of science in society. Recently, a young Emergency Physician told me he did not “believe” in thrombolysis, and quoted some of the popular myths and dogma discussed here and elsewhere^{5, 6}. I thought his choice of the word “believe” to be quite telling, as if one could chose to believe in the data one liked best. Can we choose to believe in gravity? I suppose so, but at some risk. In popular culture, we hear about choosing whether to “believe” in such science as climate change or childhood vaccination, so maybe it is unsurprising that younger physicians might adopt a similar posture towards data. It seems to me that little else threatens us more than the pernicious idea that individuals (and legislatures) have the freedom to choose whether they will ‘believe’ in some scientific findings and not others. Demagoguery and thoughtless opposition to IV rt-PA cost thousands of patients a chance at living disability free, and cost the US hundreds of thousands of dollars.¹⁵ What will be cost to all of us if unscientific ‘belief’ continues to dominate political and social discourse?