Thrombolysis for Stroke, Part I: What the Drug Does and How the Window Moved

Intravenous thrombolysis is one of the few treatments in acute neurology where the pharmacology is simpler than the decision to use it. The drug is a manufactured copy of an enzyme the body already produces to dissolve a blood clot, and its molecular mechanism has been understood since the 1980s. What has changed, repeatedly, is who should receive it. The treatment window has moved three times since 1995, and the imaging that now governs its outer edge did not exist when the first and seminal trial was run. An opinion about a thrombolysis decision is therefore an opinion about a moving standard, and the first question in any such case is which version of that standard applied on the day.

Alteplase, Tenecteplase and Thrombectomy Are Three Different Things

Alteplase is recombinant tissue plasminogen activator (rt-PA), a laboratory-manufactured version of the enzyme that human endothelium (the lining of blood vessels) releases to break down clot. It works by converting plasminogen, an inactive protein circulating in blood, into plasmin, the enzyme that cuts fibrin, the protein mesh holding a clot together.

The property that makes the drug usable at all is that this conversion is inefficient in flowing blood and efficient on the surface of a clot. Fibrin is not merely the target; it is a required partner in the reaction, assembling the drug and its substrate into a working complex. The original enzyme kinetics quantify the difference. The Michaelis constant, a measure of how much raw material the reaction needs in order to run efficiently, falls from 65 micromolar in solution to 0.16 micromolar on a fibrin surface (a lower number means a more efficient reaction).¹ The practical translation is that the drug is close to inert in the bloodstream and becomes hundreds of times more effective once it lands on clot. The word doing the work in "clot-selective" is relative, and that is precisely why bleeding remains the central risk.

Tenecteplase, increasingly used in place of alteplase, is not a different class of drug. It is alteplase with three engineered amino-acid substitutions, producing fourteen-fold greater fibrin specificity, eighty-fold greater resistance to the body's natural inhibitor of the enzyme, and roughly eight-fold slower clearance from the blood.² The slower clearance is the operational point: tenecteplase is given as a single injection over seconds, while alteplase requires an initial bolus (a small dose pushed in over about a minute) followed by a sixty-minute drip. Two large randomized trials, AcT in Canada (1,600 patients) and TRACE-2 in China (1,430 patients), found tenecteplase non-inferior to alteplase on the rate of excellent functional outcome, with comparable rates of bleeding.^3,4 Non-inferiority is a narrower claim than it sounds. It means the trial ruled out the new drug being worse by more than a margin fixed in advance, five percentage points in AcT, rather than showing the two drugs to be equivalent.

The accumulated evidence has now moved the guideline, and the change has practical consequence. Across the phase 3 trials since 2019, more than 6,000 patients have been randomized between the two agents. The 2026 AHA/ASA guideline recommends either tenecteplase at 0.25 mg/kg (maximum 25 mg) or alteplase at 0.9 mg/kg (maximum 90 mg) within 4.5 hours, both at Class 1, Level of Evidence A.⁵ Neither agent is the fallback. A record showing tenecteplase is not a record showing that a substitute was used, and the higher 0.4 mg/kg tenecteplase dose is separately recommended against.

Mechanical thrombectomy is a different intervention entirely, and the distinction is worth fixing early. It is a catheter procedure that physically retrieves clot from a large artery, performed by threading a device up from an artery in the groin or wrist, and its pooled evidence shows a considerably larger effect than any thrombolytic trial has produced.⁶ Thrombectomy applies only to large-vessel occlusion (a blockage in one of the major arteries), which is a minority of ischemic strokes; thrombolysis applies broadly. A record in which both were considered is not a record of one decision made twice.

The Penumbra Is the Target

Roughly seven out of every eight strokes are ischemic, meaning an artery is blocked by a clot. The rest are hemorrhagic, meaning an artery has ruptured and is bleeding into the brain. Thrombolysis treats only the first, and giving it to the second is catastrophic. This is exactly why every patient presenting with stroke-like symptoms to any Emergency Room should undergo emergent head CT imaging: to differentiate a non-bleeding (ischemic) from a bleeding (hemorrhagic) stroke. Everything that follows in this essay concerns ischemic stroke.

A stroke inevitably leads to the loss of healthy brain tissue and nerve cells. The affected brain area can be divided into two compartments: the ischemic core, which is already infarcted (dead due to lack of blood flow) and cannot be saved. Medical interventions cannot reverse the damage in this core. Around it lies the penumbra, tissue that is underperfused and electrically silent but still structurally alive, and still salvageable if flow returns. Positron emission tomography (PET, a scan that measures blood flow and metabolism directly) places the boundaries in measurable terms: blood flow below roughly 12 mL per 100 g per minute marks tissue transitioning to necrosis, while flow between 12 and 22 mL per 100 g per minute defines the penumbra.⁷ Everything thrombolysis is attempting to do is aimed at that middle band.

The penumbra is not stable. It converts to core over hours, which is the physiological content of the phrase "time is brain." A widely cited quantitative model estimates that a typical large-vessel supratentorial stroke (one in the cerebral hemispheres, above the level of the brainstem and cerebellum) destroys 1.9 million neurons per minute, and that the average non-lacunar stroke evolves over roughly ten hours.⁸ The estimate is a model rather than a measurement, and it was built for large-vessel strokes rather than the small-vessel kind. It is fair game for cross-examination on both counts, though the direction it describes is not seriously contested.

Success is measured functionally, not radiographically. The modified Rankin Scale (mRS) grades global disability from 0 (no symptoms) through 5 (severe disability requiring constant care);⁹ modern trials add 6 for death. They generally report the proportion achieving mRS 0 to 1, meaning no significant disability, or mRS 0 to 2, meaning functional independence. Agreement between observers on the raw scale is moderate rather than excellent, which is worth knowing before a damages argument leans on a single recorded grade. A reperfused artery (one in which flow has been restored) accompanied by an mRS of 5 is not a treatment success, and charts that report the former without the latter should be read with that gap in mind.

Severity at presentation is evaluated separately using the National Institutes of Health Stroke Scale (NIHSS), a quick 15-item bedside exam performed by paramedics, emergency nurses, and neurologists. Although widely used, this scale has limitations, such as not accurately assessing the posterior circulation of the brain and sometimes producing low scores despite significant impairment. The mechanism is specific: the scale omits gait and truncal ataxia, dysphagia and cough, so a basilar-territory stroke can score low while the patient is gravely impaired.¹⁰ Therefore, in practice, the score is just one factor among others in deciding eligibility for thrombolytic treatment. Generally, higher scores indicate greater deficits: a score of 6 or below suggests a good chance of recovery, while 16 or above indicates a high risk of death or severe disability. Each additional point reduces the likelihood of an excellent outcome.¹¹ The score frequently appears in stroke assessments, with values that differ depending on who performs the evaluation, and these disagreements highlight the scale's limitations.

The Hemorrhage Rate Is a Definition, Not a Number

The principal harm is symptomatic intracranial hemorrhage: bleeding into the brain that worsens the patient's neurological condition. In the original trial, symptomatic hemorrhage within 36 hours occurred in 6.4 percent of treated patients against 0.6 percent of controls.¹²

One complication produces recurring confusion in litigation: "symptomatic hemorrhage" has no single definition. The rate depends entirely on which definition is applied. The SITS-MOST registry, which followed 6,483 patients treated in routine practice, reported a rate of 1.7 percent under its own protocol definition and 7.3 percent under the broader Cochrane definition, in the same patients.¹³ An expert quoting a hemorrhage rate without naming the definition behind it has stated a number that cannot be checked.

Net mortality is not significantly increased, though the picture is more textured than a single number. The original trial found three-month mortality of 17 percent in treated patients against 21 percent in controls (P=0.30).¹² The nine-trial pooled analysis found 90-day mortality of 17.9 percent against 16.5 percent, a hazard ratio of 1.11 (95 percent CI 0.99 to 1.25, P=0.07): a real absolute excess of early death from bleeding of roughly 2 percent, offset by three to six months by a gain in disability-free survival.¹⁴ That confidence interval does not exclude an 11 percent relative increase, and an expert who says flatly that thrombolysis does not kill anyone has overstated the evidence.

Orolingual angioedema (swelling of the tongue and mouth, occasionally severe enough to require emergency airway management) occurred in 4.9 percent of 659 consecutively treated patients in the largest modern series. It is independently associated with ACE-inhibitor use (odds ratio 3.87, 95 percent CI 1.71 to 8.75) and with female sex (odds ratio 5.47). What that series overturned was an older claim that angioedema tracks the location of the stroke on the initial scan.¹⁵

Three Hours, Then Four and a Half

One definition governs everything that follows. The clock does not start when symptoms were noticed. It starts at the last moment the patient was observed to be at baseline, the last known well time. A patient who went to bed at eleven and woke at six with a deficit has a last known well of eleven and seven hours on the clock, whatever time the stroke actually occurred.

The treatment began with a single trial. In 1995 the NINDS rt-PA Stroke Study randomized 624 patients treated within three hours of onset and found a global odds ratio for favorable outcome of 1.7 (95 percent confidence interval 1.2 to 2.6), with treated patients at least 30 percent more likely to have minimal or no disability at three months.¹² The Food and Drug Administration approved the stroke indication in June 1996, and the three-hour window became the standard of care.

The trial has been contested since publication, and an expert who cannot discuss the criticism is not credible on the subject. The principal objection concerns an imbalance in baseline stroke severity between the treatment arms. A 2004 independent reanalysis committee examined precisely that question and found the adjusted odds ratio for favorable outcome to be 2.1 (95 percent CI 1.5 to 2.9), larger than the unadjusted figure rather than smaller.¹⁶ A 2022 formal risk-of-bias assessment nonetheless rated the randomization process at high risk of bias, noting that 11 of 16 randomization strata deviated from the expected allocation ratio.¹⁷ Both findings are part of the record.

Europe then failed twice, instructively. ECASS I tested a six-hour window at a higher dose, 1.1 mg/kg, and was negative on intention-to-treat.¹⁸ ECASS II corrected the dose to the NINDS 0.9 mg/kg, kept the six-hour window, and was negative again: 40.3 percent against 36.6 percent, P=0.277.¹⁹ With the dose corrected and the result still negative, time was left standing as the variable that mattered.

What eventually moved the window was ECASS III in 2008, which treated 821 patients between three and 4.5 hours after onset and found favorable outcome in 52.4 percent against 45.2 percent (odds ratio 1.34, 95 percent CI 1.02 to 1.76), with a number needed to treat of 14.²⁰

Two qualifications belong with that result. ECASS III entered a narrower population than routine practice inside three hours: it excluded patients over 80, those with a severity score above 25, anyone taking an oral anticoagulant, and the combination of prior stroke with diabetes. Those four are trial entry criteria rather than demonstrated contraindications, and treating them as absolute bars is a recurring analytical error. The distinction itself has since been retired. The 2026 guideline abandons the separate three-hour and three-to-4.5-hour categories that organised the 2019 edition and states a single undifferentiated window of 4.5 hours.⁵ The history still matters for any case predating that change, which is most of them.

And the label never moved with the evidence. The FDA approval of June 1996 reads within three hours, and it reads that way still. Treatment between three and 4.5 hours is guideline-driven and off-label, which is not a criticism of the practice but is a fact that shapes both the standard-of-care argument and what should have been said to the patient.

The shape of the benefit curve over time does more work in litigation than any individual trial result. A pooled analysis of individual patient data from nine randomized trials, covering 6,756 patients, resolves it directly.¹⁴ Treatment within three hours produced good outcomes in 32.9 percent against 23.1 percent of controls (odds ratio 1.75). Treatment between three and 4.5 hours produced 35.3 percent against 30.1 percent (odds ratio 1.26). Beyond 4.5 hours, the advantage was no longer statistically significant.

(An odds ratio above 1 favors treatment and below 1 favors the control group. The confidence interval is the range of values the data are consistent with, so an interval that includes 1.0 means the result cannot be separated from chance. Trials also report adjusted, unadjusted and common odds ratios, and these are not interchangeable between studies.)

In absolute terms, which is the form that matters when the question is what a particular patient lost, treating within three hours leaves roughly 10 additional patients per hundred alive and free of disability, and treating between three and 4.5 hours roughly 5 per hundred. The benefit does not switch off at a threshold; it declines continuously and disappears into the noise somewhere after four and a half hours. Registry data covering 58,353 patients puts the same relationship in operational units: every 15 minutes of faster treatment was associated with lower in-hospital mortality, less symptomatic hemorrhage, and greater likelihood of walking independently at discharge.²¹

Imaging Replaced the Clock, in Selected Patients

The 2018 and 2019 expansions did not extend the clock for everyone. It replaced the clock, in selected patients, with a picture of the penumbra.

Roughly 14 to 27 percent of strokes have no known onset time, most commonly because the patient woke with symptoms, and those patients were categorically excluded for two decades. WAKE-UP tested a magnetic resonance imaging signature (a lesion visible on one sequence but not yet on another, a pattern that dates the stroke to less than about 4.5 hours) as a substitute for a known onset time, and found favorable outcome in 53.3 percent against 41.8 percent (adjusted odds ratio 1.61).²²

EXTEND then tested the clock itself. Patients treated between 4.5 and nine hours after onset, or on waking, were selected by perfusion imaging showing salvageable tissue, and achieved favorable outcome in 35.4 percent against 29.5 percent (adjusted risk ratio 1.44), with symptomatic hemorrhage in 6.2 percent against 0.9 percent.²³ A pooled analysis of the three perfusion-selected trials, 414 patients in total, found an adjusted odds ratio of 1.86 for excellent outcome, alongside an adjusted odds ratio of 9.7 for symptomatic hemorrhage, ten events against one, on a confidence interval running from 1.23 to 76.55.²⁴ The direction is not in doubt; the magnitude is barely estimated at all.

The 2026 guideline carries this forward, and fences it more tightly than the trials did. Perfusion-selected thrombolysis between 4.5 and nine hours, or on waking within nine hours of the midpoint of sleep, is now recommended at Class 2a specifically for patients who are not eligible for endovascular thrombectomy.⁵ That restriction was not in the guideline as first published; it was added by the correction issued afterwards.²⁵ A separate Class 2b recommendation reaches from 4.5 to 24 hours for large-vessel occlusion with salvageable penumbra, again only where thrombectomy cannot be delivered.⁵

Those criteria are arithmetic, and counsel should insist on the actual numbers. Automated software estimates an irreversibly injured core from cerebral blood flow below 30 percent of normal brain, and a critically underperfused volume from a contrast-arrival delay (time-to-maximum) exceeding six seconds. In EXTEND, a patient qualified only if the ratio of underperfused tissue to core exceeded 1.2, the absolute difference exceeded 10 mL, and the core was under 70 mL.²³ Those thresholds are the eligibility standard, and they are computed rather than judged. The relevant question is therefore not whether a clinician believed there was salvageable tissue, but whether perfusion imaging was obtained at all, and what it showed.

What the Evidence Will and Will Not Carry

Three calibrations matter.

The extended window rests on a small evidence base with a real bleeding cost. The pooled perfusion-selected population is 414 patients, against 6,756 in the pooled conventional-window analysis, and the same pooling that produced the favorable odds ratio produced a roughly ten-fold increase in symptomatic hemorrhage, estimated with almost no precision (95 percent CI 1.23 to 76.55).²⁴ This is the clearest illustration in the field of why a point estimate quoted without its interval is not a finding. Of the three constituent trials, one was positive and two, ECASS4-EXTEND and EPITHET, did not reach significance on their primary endpoints.²⁴

Extending the window further is contested rather than settled, and the reason matters. TIMELESS tested tenecteplase between 4.5 and 24 hours in 458 perfusion-selected patients with large-vessel occlusion and found no benefit, with a common odds ratio of 1.13 (95 percent CI 0.82 to 1.57).²⁶ The guideline reads that null result against thrombectomy access: TIMELESS patients did receive rapid thrombectomy, whereas the positive trial in the same window enrolled patients who were not offered it in time.⁵ The honest statement is therefore narrower than either side usually makes it. Late thrombolysis has not been shown to add anything for a patient who is getting a timely thrombectomy, and it retains a role for the patient who cannot.

The standard of care is dated. A patient presenting in 2006 was outside the window at 3.5 hours; the same patient in 2012 was inside it; the same patient waking with symptoms in 2015 was ineligible, and in 2020, with the right imaging, may not have been. Any opinion about what should have happened must be anchored to what was established practice on the date of the event, and to what that particular hospital could actually obtain at that hour. In the United States the operative document is the AHA/ASA guideline for the early management of acute ischemic stroke, and there are distinct editions, each superseding the last. The 2026 edition states in terms that it replaces both the 2018 guideline and the 2019 update.^27,5 The right question in discovery is which edition was in force on the date of care, not which one is current now. Guidelines also carry published corrections, and the 2026 edition already has one.²⁵

Which means the useful question is rarely whether thrombolysis works. It is what the evidence supported on the date in question, and what that hospital could actually obtain at that hour.

The pharmacology has been settled since 1995. What counts as the right decision has not stopped moving since.

Part II takes up where this leaves off: the ways the decision itself fails. Never treated despite qualifying, and treated despite an exclusion, run in opposite directions and are litigated at least as often as delay. Underneath both sits the stroke that was never recognized, and delay turns out not to be a category of its own.

References

Footnotes

1. Hoylaerts M, Rijken DC, Lijnen HR, Collen D. Kinetics of the activation of plasminogen by human tissue plasminogen activator. Role of fibrin. J Biol Chem. 1982;257(6):2912-2919. doi:10.1016/S0021-9258(19)81051-7 ↩

2. Keyt BA, Paoni NF, Refino CJ, et al. A faster-acting and more potent form of tissue plasminogen activator. Proc Natl Acad Sci U S A. 1994;91(9):3670-3674. doi:10.1073/pnas.91.9.3670 ↩

3. Menon BK, Buck BH, Singh N, et al. Intravenous tenecteplase compared with alteplase for acute ischaemic stroke in Canada (AcT): a pragmatic, multicentre, open-label, registry-linked, randomised, controlled, non-inferiority trial. Lancet. 2022;400(10347):161-169. doi:10.1016/S0140-6736(22)01054-6 ↩

4. Wang Y, Li S, Pan Y, et al. Tenecteplase versus alteplase in acute ischaemic cerebrovascular events (TRACE-2): a phase 3, multicentre, open-label, randomised controlled, non-inferiority trial. Lancet. 2023;401(10377):645-654. doi:10.1016/S0140-6736(22)02600-9 ↩

5. Prabhakaran S, Gonzalez NR, Zachrison KS, et al. 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke. 2026;57:e316-e436. doi:10.1161/STR.0000000000000513 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

6. Goyal M, Menon BK, van Zwam WH, et al. Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials. Lancet. 2016;387(10029):1723-1731. doi:10.1016/S0140-6736(16)00163-X ↩

7. Heiss WD. The ischemic penumbra: how does tissue injury evolve? Ann N Y Acad Sci. 2012;1268:26-34. doi:10.1111/j.1749-6632.2012.06668.x ↩

8. Saver JL. Time is brain—quantified. Stroke. 2006;37(1):263-266. doi:10.1161/01.STR.0000196957.55928.ab ↩

9. van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19(5):604-607. doi:10.1161/01.STR.19.5.604 ↩

10. Alemseged F, Rocco A, Arba F, et al. Posterior National Institutes of Health Stroke Scale improves prognostic accuracy in posterior circulation stroke. Stroke. 2022;53(4):1247-1255. doi:10.1161/STROKEAHA.120.034019 ↩

11. Adams HP Jr, Davis PH, Leira EC, et al. Baseline NIH Stroke Scale score strongly predicts outcome after stroke: a report of the Trial of Org 10172 in Acute Stroke Treatment (TOAST). Neurology. 1999;53(1):126-131. doi:10.1212/WNL.53.1.126 ↩

12. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;333(24):1581-1588. doi:10.1056/NEJM199512143332401 ↩ ↩² ↩³

13. Wahlgren N, Ahmed N, Dávalos A, et al. Thrombolysis with alteplase for acute ischaemic stroke in the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST): an observational study. Lancet. 2007;369(9558):275-282. doi:10.1016/S0140-6736(07)60149-4 ↩

14. Emberson J, Lees KR, Lyden P, et al. Effect of treatment delay, age, and stroke severity on the effects of intravenous thrombolysis with alteplase for acute ischaemic stroke: a meta-analysis of individual patient data from randomised trials. Lancet. 2014;384(9958):1929-1935. doi:10.1016/S0140-6736(14)60584-5 ↩ ↩²

15. Pinho J, Alves JM, Oliveira L, et al. Orolingual angioedema after thrombolysis is not associated with insular cortex ischemia on pre-thrombolysis CT. J Neurol Sci. 2016;369:48-50. doi:10.1016/j.jns.2016.07.043 ↩

16. Ingall TJ, O'Fallon WM, Asplund K, et al. Findings from the reanalysis of the NINDS tissue plasminogen activator for acute ischemic stroke treatment trial. Stroke. 2004;35(10):2418-2424. doi:10.1161/01.STR.0000140891.70547.56 ↩

17. Garg R, Mickenautsch S. Risk of selection bias assessment in the NINDS rt-PA stroke study. BMC Med Res Methodol. 2022;22(1):172. doi:10.1186/s12874-022-01651-4 ↩

18. Hacke W, Kaste M, Fieschi C, et al. Intravenous thrombolysis with recombinant tissue plasminogen activator for acute hemispheric stroke: the European Cooperative Acute Stroke Study (ECASS). JAMA. 1995;274(13):1017-1025. doi:10.1001/jama.274.13.1017 ↩

19. Hacke W, Kaste M, Fieschi C, et al. Randomised double-blind placebo-controlled trial of thrombolytic therapy with intravenous alteplase in acute ischaemic stroke (ECASS II). Lancet. 1998;352(9136):1245-1251. doi:10.1016/S0140-6736(98)08020-9 ↩

20. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med. 2008;359(13):1317-1329. doi:10.1056/NEJMoa0804656 ↩

21. Saver JL, Fonarow GC, Smith EE, et al. Time to treatment with intravenous tissue plasminogen activator and outcome from acute ischemic stroke. JAMA. 2013;309(23):2480-2488. doi:10.1001/jama.2013.6959 ↩

22. Thomalla G, Simonsen CZ, Boutitie F, et al. MRI-guided thrombolysis for stroke with unknown time of onset. N Engl J Med. 2018;379(7):611-622. doi:10.1056/NEJMoa1804355 ↩

23. Ma H, Campbell BCV, Parsons MW, et al. Thrombolysis guided by perfusion imaging up to 9 hours after onset of stroke. N Engl J Med. 2019;380(19):1795-1803. doi:10.1056/NEJMoa1813046 ↩ ↩²

24. Campbell BCV, Ma H, Ringleb PA, et al. Extending thrombolysis to 4·5-9 h and wake-up stroke using perfusion imaging: a systematic review and meta-analysis of individual patient data. Lancet. 2019;394(10193):139-147. doi:10.1016/S0140-6736(19)31053-0 ↩ ↩² ↩³

25. Correction to: 2026 guideline for the early management of patients with acute ischemic stroke: a guideline from the American Heart Association/American Stroke Association. Stroke. 2026;57:e461-e467. doi:10.1161/STR.0000000000000530 ↩ ↩²

26. Albers GW, Jumaa M, Purdon B, et al. Tenecteplase for stroke at 4.5 to 24 hours with perfusion-imaging selection. N Engl J Med. 2024;390(8):701-711. doi:10.1056/NEJMoa2310392 ↩

27. Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke: 2019 update to the 2018 guidelines for the early management of acute ischemic stroke. Stroke. 2019;50(12):e344-e418. doi:10.1161/STR.0000000000000211 ↩