Thrombolytic Therapy for Acute Ischemic Stroke: A Very Great Honor

A 35-year-old woman is found paralyzed in a bathroom. She had been shopping with her mother for bridesmaids’ gifts. She had suffered a severe migraine during the previous 3 weeks and had been evaluated at an outside hospital. For the migraine, she had undergone non-contrast CT head scan and lumbar puncture with normal results. She had been treated with enough narcotics to induce nausea, vomiting and dehydration but eventually was discharged without headache one week before presentation. The day of admission was her first trip outside the house. After shopping a while, she complained to her mother of a strange feeling and went into the ladies’ room. When she did not return the mother investigated and found her awake but confused and paralyzed on the left side. Paramedics were summoned who surmised she could be having a stroke and contacted the closest medical center. Upon arrival at the outlying hospital, a Neurologist was summoned.

In 1990, when this case occurred, no acute treatment for stroke existed. The paramedics used generic assessment tools, because stroke-specific pre-hospital assessments were yet to be invented. There were no certified stroke centers, so she went to the closest medical center; there was no Code Stroke, no Emergency Department (ED) protocols, no tackle box with tissue-type plasminogen activator (t-PA). What did we know about acute stroke on the day this patient presented? What could we do for her?

In 1981, a key publication changed our entire approach to brain ischemia (Figure 1) by demonstrating the time course for penumbral transformation to infarction¹. Prior to this paper, clinical dogma held that stroke damage finalized within minutes of symptom onset—stroke conveyed a hopeless prognosis, immediately. In the late 1970’s, investigators documented a heterogeneous depression of cerebral blood flow (CBF) in the territory of the occluded artery. The penumbra was then defined as the brain region receiving regional cerebral blood flow (rCBF) between 2 critical values—a CBF threshold for function and a lower CBF threshold for cell death. In the penumbra, neurons were identified as “idling” to suggest that they were nonfunctional but salvageable^{1, 2}. But time was important as well; as shown in Figure 1, idling neurons did not last indefinitely. The most important aspect of Figure 1, however, was the x-axis, which was labeled in hours; idling cells in the penumbra could survive up to several hours receiving rCBF sufficient to allow survival^3–6. In other words, the penumbra involved two different parameters: blood flow and time. This influential idea suggested that over time the “core” of the infarct enlarged, eventually subsuming the penumbra, giving rise to the clinical dictum that “time is brain.” The publication of the time-course for penumbral salvage galvanized a generation of young investigators to join research labs in hopes of finding ways to salvage ischemic brain.

Figure One: Time course of penumbra evolving into infarction.

A hypothetical graph, based on data recorded in non-human primates during middle cerebral artery occlusion, which illustrates rCBF (y-axis) against time (x-axis). The horizontal line represents the rCBF threshold for function: brain areas receiving flow below this level do not function. The up-sloping line represents the rCBF threshold for infarction: ischemic brain areas receiving rCBF below this threshold, over time, will evolve into infarction. Depending on the rCBF, areas of brain could survive minutes (very low rCBF) to hours (greater rCBF). The observation that some areas of non-functioning brain could endure hours of low rCBF before infarcting implied the possibility of successful acute stroke therapy. The data underlying this revolutionary concept galvanized the research community to pursue recanalization treatments and neuroprotective therapies (reprinted from Jones et al¹ with permission. Copyright 1981 American Association of Neurosurgery.)

In 1985, Dr. Justin Zivin published a seminal paper in Science showing functional benefit of thrombolytic therapy after embolic stroke⁷. He used intravenous t-PA to restore function after embolic stroke. Around the same time, several investigators showed that thrombolysis successfully opened arteries in experimental models^8–11 as well as in small clinical case series^12–14. I joined the Zivin lab in 1985 and was very fortunate to receive support from the National Institutes of Neurological Disorders and Stroke (NINDS) under a program (the R29 grant) for encouraging clinicians to become translational scientists called First Independent Research Support or FIRST. Thanks to the NINDS and Dr. Zivin, I was able to learn basic experimental methods for studying thrombolysis, anticoagulation, embolic stroke, and hemorrhagic transformation^15–21. These papers and many others built a foundation for intravenous thrombolysis by showing efficacy and safety in animal models. This experimental foundation proved essential, because earlier human trials of intravenous thrombolysis with streptokinase caused significant harm—mainly intracerebral hemorrhages—and the wider medical community assumed intravenous thrombolysis with any agent would cause unacceptable hemorrhaging^22–27. From today’s perspective, we now know that streptokinase may be safe and effective under certain circumstances²⁸, but in the late 1980s and early 1990s we needed a robust and rigorous safety assessment of intravenous t-PA to determine whether it was safer than streptokinase²⁹.

To fully qualify t-PA as safe, in light of the negativity engendered from the streptokinase trials, NINDS-funded investigators at three stroke centers examined the safety and clinical outcome following treatment with escalating doses of intravenous single-chain t-PA (alteplase) in patients with very early stroke symptoms who could be treated within 90 min of symptom onset³⁰. Ultimately 74 patients were treated with one of 7 doses, ranging from 0.35 mg/kg to 1.08 mg/kg given over 60 to 90 min (Table). Only 3 symptomatic hemorrhages occurred, all of them in patients receiving 0.95 mg/kg. In the next lower treatment group, treated with 0.85 mg/kg, 55% of patients exhibited major neurological improvement within 24 h and no hemorrhages. Thus, the investigators moved forward with a dose of 0.85mg/kg delivered as a 10% bolus followed by the remainder infused over 60 minutes. An additional dose finding study confirmed safety of the same dose when started as late as 180 minutes after stroke symptom onset³¹. Just prior to launching the pivotal NINDS trial of t-PA for acute stroke, the investigators studied 27 patients in a feasibility pilot study of 0.85mg/kg and again showed safety and feasibility³². To simplify dose calculations in the next, larger trial, the 0.85mg/kg was rounded, and as a result, the now-canonical dose of 0.9mg/kg was finalized for further study.

Table. rt-PA Dose Escalation and Patient Outcomes.

Acute ischemic stroke patients were treated with escalating doses of intravenous t-PA, also called rt-PA. In the first 3 dose tiers, t-PA was infused over 60 min with no loading bolus. The experience at each dose tier was reviewed by an independent safety and monitoring committee (John Hallenbeck, MD, PhD; Thomas Price, MD; and David Stump, MD) before proceeding to the next dose. The highest dose tier (1.08mg/kg) was discontinued because of hemorrhagic complications at lower doses occurring in a parallel study of patients treated 91–180 minutes from symptom onset^{30, 31}. For analysis of safety (Bleeds) the end points included intracerebral parenchymal hematoma (ICH, where solid clot was clearly present and displacing brain parenchyma), hemorrhagic transformation without ICH, systemic hemorrhagic complication, and death resulting from any hemorrhagic complication. Major neurological improvement (MNI) was defined retrospectively as an improvement of 4 or more points in the NIHSS. (Adapted from Brott et al³⁰, with permission. Copyright© 1992, American Heart Association).

Dose (mg/kg)	N	Bleeds (N)	MNI (N)
0.35	6	0	2
0.60	12	0	4
0.85	10	0	4
0.85*	20	0	11
0.95**	22	2	11
0.95***	3	1	1
1.08	1	0	1
Total	74	3(4%)	34 (46%)

^*10% bolus dose given

^**90 min infusion (all other dose tiers used 60 min)

^***Dropped back to lower dose tier because of bleeds.

The use of t-PA for acute stroke proved safe and effective in the NINDS Trial of t-PA for Acute Stroke (NINDS Trial), published in 1995³³. The therapy was approved by the Food and Drug Administration (FDA) in 1996, and endorsed by the American Heart Association, American Academy of Neurology, and National Stroke Association in 1997^34–36. Despite the now-ubiquitous acceptance of thrombolytic therapy for acute ischemic stroke, prior the NINDS Trial, the infrastructure to conduct such a trial did not exist—every trial site built from scratch a stroke team that could treat patients within 90 or 180 minutes of stroke symptom onset. In San Diego, we created a variety of tools that are now considered essential to managing code strokes, but prior to 1992 these did not exist³⁷. In all the other NINDS Trial sites the coordinators and investigators similarly solved a variety of cultural, logistic and medical barriers to rapid stroke treatment³⁸. Today, such tools as tackle boxes in the ED, pre-written order sets, treatment protocols, and the very term “code stroke” are so common, it may be difficult to believe they did not exist until the NINDS coordinators and investigators created them³⁹.

Upon the Neurologist’s arrival at the bedside, the patient exhibited left hemiplegia, hemineglect of the left visual space, forced right eye deviation, and left hemianesthesia. A non-contrast head CT scan showed no hemorrhage, and no signs of early ischemic damage; in retrospect, a hyperdense artery sign is seen, but at the time she presented, this finding had been described only in the imaging literature⁴⁰ and the Neurologist did not recognize its significance ⁴¹. In discussion with the patient’s mother and fiancée—an attorney—the possibility of attempting thrombolysis was offered. The family was told that although intravenous thrombolytic therapy was safe and effective for heart attack patients, and although pilot studies suggested a possible benefit in stroke patients, no data supported its safety or efficacy unequivocally in stroke patients. After an extended discussion, intravenous t-PA was administered to the patient using the protocol designed by the NINDS Investigators.

The main results of the NINDS Trial—now widely appreciated in the medical community—appeared in the New England Journal of Medicine on December 15^th, 1995³³. The day before, however, a press conference was held on the NIH campus. Media attendance exceeded anything seen previously for a stroke trial result and included medical and lay press. Key leaders of the NINDS, including John Marler and Michael Walker, introduced the study and the key speakers: Thomas Brott, James Grotta and myself. That night, a bevy of media items ran in print and on television (the public Internet and social media were uninvented). Given all the attention, we NINDS investigators thought that acute stroke treatment would change dramatically and quickly; we were very wrong. Despite approval by the FDA in 1996, intravenous thrombolytic therapy for acute ischemic stroke did not fully enter mainstream medical use for another 20 years.

In the meantime, the NINDS investigators published many papers over 10 years thoroughly describing the full study experience with the drug. In the early years after approval, skeptics argued that without precise identification of which patient subgroups respond to thrombolytic therapy, we should treat no one for fear we might treat a patient unlikely to benefit. Many commentators wanted to “target” treatment at the subgroup of patients that is most likely to respond⁴². We knew, however, that a majority of patients enjoyed some benefit from t-PA, albeit not a complete cure⁴³. We examined the original data for evidence that any particular subgroup of patients was unlikely to benefit, and therefore treatment could be withheld⁴³. A similar analysis, led by Joe Broderick, tried to identify subgroups of patients who were more likely than others to suffer hemorrhagic complications⁴⁴. The two analyses included multiple sequential statistical analyses, and various combinations of baseline data. From the analyses of baseline variables, such as age, stroke deficit, presence of diabetes, presence of prior stroke, and other important factors, no subgroup could be identified for whom thrombolytic stroke therapy could be particularly recommended or prohibited⁴³. Moreover, although symptomatic hemorrhage occurred more commonly in patients who were older or with higher NIHSS scores, there was no statistically significant interaction with treatment⁴⁴. In other words, benefit outweighed risk in all patients regardless of age or stroke severity. We showed unequivocally that clinicians should prescribe t-PA to patients who fulfill the criteria as outlined in the original protocol. Withholding therapy in any subgroup of patients—hoping to target more-likely-to-respond patients—was unwise^45–48.

The Group published additional data and analyses showing the treatment to be cost effective⁴⁹. For a theoretical cohort of 1000 t-PA-treated patients, the net cost savings in the first year were estimated to be $4 million; the sensitivity analysis from the simulations indicated a probability of 93% that the estimate was correct. Multiple studies confirmed these the cost-benefit of thrombolysis for ischemic stroke in multiple countries with very different health care delivery systems^49–52. Further studies confirmed that thrombolytic therapy saves money for the health care system overall, even after factoring in the costs of stroke teams, the evaluations of “mimic” patients, and the hemorrhages that result in some cases^{52, 53}. Thrombolytic therapy for acute ischemic stroke is one of only a few therapies that not only benefit patients directly but save the system money.

Under the lead authorship of Tom Kwiatkowksi, we confirmed a durable benefit of thrombolytic stroke therapy by following the patients in the original study for up to one year⁵⁴. The results were nearly identical to those in the original publication: the odds ratio of a near complete recovery after t-PA was 1.7 (95% CI 1.3 to 2.3), compared to placebo (p=0.0013). Looking at functional independence (Barthel Index) as the outcome, patients are 50% more likely than placebo treated patients to have minimal or no disability 1 year after thrombolytic stroke therapy. In addition to the long term functional data, the volume of infarction seen on CT scans 3 months after thrombolytic therapy was significantly reduced: median stroke volume was 25.5cm³ in placebo treated, vs. 15.5cm³ in t-PA treated, patients (p=0.039)⁵⁵.

We sought to determine whether—given the press of time and early, rapid treatment of patients—stroke teams inadvertently treated TIA patients. Using the NINDS thrombolysis protocol, we found that very few TIA patients were treated: in the NINDS Trial placebo group, only 2% of patients exhibited no neurologic deficit (NIHSS = 0) when examined 24 hours after stroke³³. These 2% of patients included patients with TIA and perhaps other non-stroke etiologies. All of these patients were independent in activities of daily living and none suffered hemorrhage within 3 months after their strokes⁵⁶.

Following the original NINDS Trial report, several additional publications confirmed the value of thrombolytic therapy for stroke. The first European Cooperative Acute Stroke Study (ECASS) failed to show a statistically significant benefit for t-PA on the primary outcome, but using the data methods developed for the NINDS study, a clear benefit was seen: the global odds ratio for favorable outcome was 1.5 (95% CI 1.1 to 2.0, p=0.008)⁵⁷. A confirmatory study (ECASS II)⁵⁸ also failed on its primary endpoint, the global odds ratio for a favorable outcome (score of 0 or 1) on the modified Rankin Scale, but they found significantly (p=0.024) more patients with favorable outcome (mRS 0–2) after treatment with t-PA (54.3%) than with placebo (46.0%). ECASS III closely mirrored the original NINDS Trial protocol, although patients who presented between 3 and 4.5 hours after symptom onset were enrolled. ECASS III showed that thrombolytic therapy is safe and effective for acute stroke victims treated between 3 and 4.5 hours⁵⁹. In response to ECASS III, I pointed out that the longer time window did not allow us clinicians to slow down during a Code Stroke⁶⁰ because we knew that the potential for neurologic rescue declines with every passing minute⁶¹. From the moment the patient arrives at the door every minute counts, and the only justifiable delays would be for brain imaging to exclude hemorrhage, and a few simple lab tests^62,63. The phrase ‘primum non tardare’ highlights the need to focus on time—in addition to potential risks and benefits—since lost time represented a far greater threat to the patient⁶⁴.

To assess potential benefit of rt-PA in stroke patients outside the accepted selection criteria a group of European investigators launched a trial in the late 1990s, named the third “International Stroke Trial”, or IST-3^{65, 66}. The main effect of t-PA in IST-3 was positive⁶⁷. A pre-specified meta-analysis of individual patient data from all the randomized phase 3 trials discussed above confirmed that alteplase increased the odds of a good stroke outcome irrespective of age and stroke severity, if given within 4.5 h of symptom onset⁶⁸. In the combined analysis there was a significant negative effect of time-delay on the odds of a good outcome, as summarized in Figure 2.

Figure Two. The odds ratio for good outcome following thrombolytic therapy for acute ischemic stroke declines over time.

A pooled analysis of patient-level data included all patients enrolled in 9 randomized, placebo-controlled, blinded clinical trials. The solid line indicates the point estimate of the odds ratio for success (mRS 0 or 1) while the dashed lines represent the 95% confidence intervals. The curve—widely interpreted as showing ‘time is brain’—declines over time, suggesting the odds of any single patient benefiting is greater with earlier treatment. The confidence intervals indicate that the true value of the odds ratio for benefit in patient groups treated with t-PA lies somewhere between the dashed lines, with 95% confidence. (reprinted from Emberson et al⁶⁸ with permission. Copyright© 2014 Emberson et al. Open Access article distributed under the terms of CC BY).

Despite good success with intravenous thrombolysis, since not all patients benefited, additional treatments such as endovascular therapy emerged as well. Endovascular therapy for acute ischemic stroke began with intra-arterial thrombolysis⁶⁹ but catheter based thrombectomy attracted further development. After a false start with early generation thrombectomy devices^70–72, in 2014 endovascular thrombectomy was shown to be safe and effective for treating acute ischemic stroke⁷³. In 2015 at the AHA/ASA sponsored International Stroke Conference in Nashville, endovascular thrombectomy became standard-of-care for patients with an appropriate large vessel occlusion presenting within 6 or 8 hours of symptom onset^74–76. It has been said—and only half in jest—that while it took 20 years for intravenous thrombolysis to become standard-of-care, it took only 20 minutes for thrombectomy! Whether or not thrombectomy candidate patients should be treated also with intravenous thrombolysis remains an open question. And while t-PA is safe and highly effective for acute ischemic stroke, it is far from perfect. New-generation thrombolytics have been tested, including microplasmin, desmoteplase, reteplase, and tenecteplase. Tenecteplase, also called TNK-type t-PA, is as effective in vivo as alteplase^{21, 77}. Due a longer plasma half-life the drug can be administered as a bolus, avoiding a prolonged infusion⁷⁸. Pilot studies⁷⁹ and larger follow up studies tested TNK against standard dose rt-PA^{80, 81}. Hopefully efforts to replace standard rt-PA with TNK-tPA will be successful.

If one examines the confidence intervals in Figure 2, one notes the lower 95% confidence limit crosses unity (no benefit) around 5 hours, and the point estimate hits unity around 6.5 hours. Many interpret this data to mean that t-PA should be offered only to patients treatable under 5 or 6 hours from symptom onset. If one looks at the upper 95% confidence limit, however, one notes the possibility that the odds ratio for beneficial effect could be as high as 1.2 at 6 hours. In other words, the data suggest that many patients with salvageable brain tissue would be deprived of useful treatment if physicians strictly adhered to the 4.5-hour time limit. Unfortunately, the data also suggest that many patients might not have salvageable brain tissue even earlier than 4.5 hours after stroke onset. Thus, tenacious investigators such as Greg Albers pioneered the use of imaging-guided patient selection for thrombectomy to identify salvageable patients based on remaining tissue perfusion rather than time (Figure 3)^{82, 83}. Others have shown that MRI scanning can also be used to select patients for intravenous t-PA⁸⁴.

Figure Three. Perfusion Imaging with automated software.

Selected cases from the author’s institution evaluated with commercially available perfusion analysis software designed to identify areas of mismatch (RAPID CTP™, iSchemaView, Palo Alto, CA). Given that time serves imperfectly to select patients for recanalization therapy, automated analysis of brain perfusion maps effectively identifies patient likely to benefit from recanalization therapy beyond standard time windows. The color-coded mapping of brain regions with low flow (green color; defined as perfusion time-to-peak greater than 6 seconds) and with probable infarction (pink color; defined as cerebral blood flow < 30% the contralateral value).

About 30 minutes after the t-PA bolus, while the remainder of the dose was infusing, the patient suddenly grabbed her head and complained of severe headache. Fearing an intracerebral hemorrhage, a non-contrast head CT was done immediately; no hemorrhage was found. At that point, the Neurologist realized the patient had grabbed her head with the previously plegic left arm. The sudden headache indicated successful reperfusion. She went on to make a nearly complete recovery. Her wedding was held as scheduled, and she went on to return to her job and have children.

The illustrative case presented here is real and emphasizes the reason we do acute stroke therapy. Stroke practitioners know what it feels like to drive home late at night after saving a patient with t-PA or thrombectomy. We know that feeling is truly amazing and unlike much of anything else we do in Neurology. Deciding to give thrombolysis is hard, maybe among the most difficult treatment decisions in medicine, given the risks involved and the compressed time frame available. It takes a team, and it has been my very great honor to work with several teams, both clinical and research. But for some of us, our work involved more than giving t-PA, and certainly my colleagues who performed the NINDS Trial, also worked in a manner that could be characterized as pioneering. In the 1990’s, when there was no treatment for stroke and thrombolysis was considered dangerous, giving t-PA was akin to being a test pilot: the assignment was to take a brand-new machine up into the air for the first time to find out if it would crash and burn. We NINDS Investigators were honored to be allowed to study a drug that many thought too dangerous for safe human use. We invented new systems of care and created totally new concepts, like Code Stroke. I hope many more students, residents and Fellows stay interested in that pioneering part of Vascular Neurology, the test pilot part. In medical research, it is our honor to be allowed to ignore the established dogma, so that we can create something new that did not exist before. We are honored by funding agencies who support our work. We are honored by our patients and their families, who allow us to enroll them into our trials. Throughout, it has been my profound honor to work on this most significant medical advance called thrombolysis, to work with amazing clinical and research teams, and it has been my profound honor to receive the Feinberg award.