1. IST-3 collaborators meeting Professor Peter Sandercock University of Edinburgh, UK ISS conference, Capetown, October 2006

2. Outline What do we know? Analysis of existing trials of thrombolysis for stroke – Overall effects: risks and benefit – Subgroups: effect of time & age Who is treated? Current use of rt-PA in Europe & USA Do we need further randomised trials? YES Randomised trials of IV thrombolysis IST-3 protocol Progress with trial Imaging update

3. Background

4. Percent dead Diff. per Placebo Treated 1000 Streptokinase 12.0% 9.2% 28 Aspirin 11.8% 9.8% 24 P < 0.00001 for both comparisons. NOTE: The 22% reduction in the odds of death observed in ISIS-2 was identical to the estimate of benefit from a meta-analysis of all previous trials ISIS-2: 17,000 patients with acute MI

5. Rapid change in clinical practice. Thrombolysis for MI increased markedly after publication of megatrials in ‘86 & ‘87 GISSI ISIS-2 GISSI AIMS ISIS-2 ASSET Ketley and Woods Lancet 1993: 342: 891-4

6. Trials in acute stroke have been far too small “It is still not sufficiently widely appreciated just how large clinical trials need to be to detect reliably the sort of moderate, but important, differences in major outcomes that might exist (especially if effects in different subgroups are to be assessed reliably).” Collins and MacMahon Lancet 2001; 357: 373-380

7. Randomised trials of thrombolysis vs control in acute myocardial infarction Total no. patients by 1994!58,600 Randomised trials trials of thrombolysis vs control in acute ischaemic stroke Total (all agents) 5,675 rt-PA 2,700 rt-PA < 3hrs 930 rt-PA aged > 80 years 42

8. Previous trials of rt-PA

9. Review of trials of thrombolysis with rt- PA for acute ischaemic stroke Risks –Symptomatic cerebral haemorrhage –Death Benefits –Reduced ‘death or dependency’ –?reduction in massive cerebral oedema? Subgroup analyses: effect of –Time to treatment –Age –Risk factors for intracerebral haemorrhage –Appearance of CT scan

10. rt-PA < 6 hrs RISK: symptomatic intracerebral haemorrhage (SICH) More with control More with rt-PA Cochrane Database of Systematic Reviews 2004 < 3 hours 3 – 6 hours

11. rt-PA < 6hrs RISK: effects on death Non-significant 16% increase in deaths (95% CI, 6% reduction to 44% increase) rt-PA saves livesrt-PA kills

12. POSSIBLE BENEFIT of rt-PA < 6hrs ? Reduction in symptomatic cerebral oedema

13. rt-PA <6 hours: BENEFIT = reduction in ‘death or dependency’, despite risk 20% reduction with rt-PA (95% CI 7-23%) BUT the significant between- trial heterogeneity (I 2 =62%) makes result unreliable

14. Areas of uncertainty

15. Review of trials of thrombolysis with rt- PA for acute ischaemic stroke Risks –Symptomatic cerebral haemorrhage –Death Benefits –Reduced ‘death or dependency’ –?reduction in massive cerebral oedema? Subgroup analyses: effect of –Time to treatment –Age –Stroke severity –Risk factors for intracerebral haemorrhage –Appearance of CT scan

16. rt-PA trials meta-analysis. Benefit declines with increasing time to treatment, but scope for benefit up to 6h (Lancet 2004; 363: 768–74) Benefit Harm 3 hours6 hours Upper and lower 95% confidence limits Line of no effect

17. Only a small, variable proportion of patients are treated with rt-PA Countryno.no. % treated hospitals patients rt-PA (range) USA 42 1,195 4.1% (0-12%) USA 29 3,948 1.8% (0-10%) USA 137 23,058 1.6% (0-5%) Germany 104 13,440 3.0% (0-18%)

18. Effect of hospital, age and race, and presence of neurologist on likelihood of receiving thrombolysis for acute ischaemic stroke among 23,058 acute stroke patients from 137 community hospitals in USA In 35% of hospitals, no patients at all given rt-PA. Strong trends to LESS rt-PA use: –with increasing age of the patient, –if no neurologist available. Reed et al. Stroke 2001: 32; 1832-44

19. Number of older patients with acute stroke per year in UK 87,000 patients aged > 70 years 47,000 patients aged > 80 years = A big problem for acute medical services!

20. 9 am. This 85 year old man suddenly cannot speak and his right arm is weak. At 3hrs, CT confirms it is an ischaemic stroke: should he be treated? does his age affect his response to thrombolysis? ?

21. Effect of age on benefit from rt-PA Analysis of major randomised trials of rt-PA for stroke Adjustment for age did not modify the relation between benefit and time No subgroup analysis to answer the simple questions: –Does age alter the balance of risk and benefit? –Should there be an upper age limit for treatment? Lancet 2004; 363: 768–74

22. Effect of stroke severity (NIHSS) on good outcome with rt-PA 1.2 2.6 1.9 2.1 2.5 2.1 0.1 1 10 0-56-1011-1516-20> 20All Patients Baseline NIHSS Score Log (OR) good outcome Test for equal OR’s: Chi-square (4 DF) = 1.70; p = 0.79 No evidence of difference in treatment benefit of rt-PA across the five NIHSS severity groups Ingall et al Stroke. 2004;35:2418-2424.

23. Effect of age on risk? (NINDS trial): factors which predict intracerebral haemorrhage Baseline NIHSS > 20 Age > 70 years Ischaemic changes present on initial CT Glucose > 16.7 mmol/L Ingall et al Stroke. 2004;35:2418-2424.

24. 1.9 2.6 1.4 0.1 1 10 01>=2 Number of SICH Risk Factors Log (OR) Test for equal OR’s: Chi-square (2 DF) = 1.77; p = 0.41 = no evidence of a difference in risk of SICH in these three groups Outcome by no. of SICH risk factors, adjusted for co-variates

25. NINDS conclusions on SICH. Subgroup analyses suggested that some clinical characteristics were related to the occurrence of SICH However, after adjustment, the differences between subgroups were not statistically significant. Cannot predict reliably who will develop SICH Ingall et al Stroke. 2004;35:2418-2424.

26. Early ischaemia signs on CT 4 hours 24 hours Hypodensity : loss of grey/white differentiation, loss of the lentiform nucleus Swelling : effacement of sulci, compression of ventricle

27. ‘Early ischaemia’ signs on CT: effect on response to rt-PA Two randomised trials, 1,926 patients. No evidence that ‘early infarct sign’ significantly modifies effect of thrombolysis But analysis has insufficient statistical power – need larger trials Wardlaw et al, Radiology 2005: 235: 444

28. What is known: summary Limited rt-PA trial data (2,100 patients) Very limited data of effects in older people Effects on death unclear ‘Time is brain’; early treatment is best Clear ~3% risk of fatal brain haemorrhage Despite risk, potential for net benefit for selected patients up to 6hrs.

29. Current use of rt-PA in clinical practice

30. Current rt-PA approval for use in routine clinical practice Patient MUST be –able to be treated within 3 hours –aged under 80 –not have a history of prior stroke + Diabetes –not have any of the standard exclusions –NIHSS < 25 –No extensive infarction on CT There must be a discussion of risk/consent

31. Variation in use of rt-PA for acute ischaemic stroke ‘within licence’ in Europe SITS register (2003-5) March 2005

32. We need further large trials to: Determine reliably: –whether there are patients outside strict criteria of current approval who benefit < 3hrs –which type of patients benefit 3-6 hours –Balance of risk and benefit in older patients Contribute further evidence to: –persuade doubting clinicians to change practice –reduce inequalities in patient access to treatment –persuade health authorities to fund: cost of drug treatment a well-organised acute stroke service in every acute hospital

33. Current small scale randomised trials of i.v. thrombolysis

34. Third International Stroke Trial. A large randomised trial to answer the question: can a wider variety of patients be treated? Target: 6000 patients from 300 centres in 36 Countries

35. Main features of IST - 3 International, multi-centre, Prospective, Randomised, Open, Blinded Endpoints study of i.v. rt-PA vs control. Independent. Investigator-led Primary outcome: the proportion of patients alive and independent at six months (Modified Rankin 0,1 or 2) Randomisation by telephone or internet with on-line minimisation to balance key prognostic factors. Blinded central review of all scans

36. Recruitment Recruitment at 28.10.06: 648 patients from 65 centres in 9 countries.

37. Recruitment by country CountryNo. centresPts.% UK22 25238% Poland412821% Norway99415% Italy125610% Belgium2458% Australia9365% Sweden9243% Austria Canada 1111 5454 1%

38. Countries in process of joining/seeking to join trial In process Czech Republic Hungary India Mexico Portugal Seeking to join Argentina Belarus Brazil Chile South Africa Switzerland Taiwan Thailand Ukraine

39. 120 patients aged over 80 192 patients aged > 80 = increased world evidence base 5 x!

40. Median = 4.1 hours

41. Report of the IST 3 Data Monitoring Committee The Data Monitoring Committee for the IST-3 trial reviewed the interim outcome data on 26 September 2006, and had no safety concerns. We would encourage all collaborators to support this important trial and the study organisers to explore all appropriate ways to achieve the required rate of recruitment as rapidly as possible. Professor Rory Collins, Chairman

42. Summary IST-3 is obtaining very important data about thrombolysis, outside the current licence It has already increased the world randomised evidence base on the effect of rt-PA in patients aged > 80, by 5 times! It is encouraging new centres to use thrombolysis in a randomised trial –Closely monitored –Adds to the randomised evidence It will provide uniquely useful data on the impact of clinical and CT findings on response to rt-PA

43. Persisting hyperdense artery sign, ASPECTS score and risk of developing mass effect: an analysis based on the first 389 patients from the IST-3 Trial. Kobayashi A, 1,2 Skowronska M 1,2, Bembenek J 3, Sandercock P 4, Kane I 4, Czlonkowska A 1,2, Lewis S 4, Wardlaw J 4, for the IST-3 Collaborative Group 1 Medical Univerity of Warsaw, Poland, 2 Institute of Psychiatry and Neurology, Warsaw, Poland, 3 Wolski Hospital, Warsaw, Poland, 4 University of Edinburgh, United Kingdom

44. Background In patients with acute ischaemic stroke, the hyperdense artery sign is a marker of fresh thrombus occluding the vessel The hyperdense artery sign –is associated with greater stroke severity –can disappear with reperfusion, –can persist if the artery remains occluded –If persistent, may predict infarct swelling No evidence that hyperdense artery on baseline CT influenced response to rt-PA in NINDS trial

45. Hyperdense artery

46. Aims of this analysis In patients randomised in IST3, to assess how often the hyperdense artery sign: –is seen on baseline CT scan –disappears by the time of second scan –is associated with ‘early ischaemic change’ and mass effect on baseline and follow-up CT Assess associations with a persistent hyperdense artery

47. IST-3 image assessment protocol Design: –randomised controlled trial of rt-PA vs control in 6000 patients with acute ischaemic stroke < 6 hrs of onset Eligibility for the trial –No clear indication for, or contraindication to, thrombolysis with iv rt-PA. CT scan –Before randomisation (MR permitted) –Repeat CT at 24-48 hrs after randomisation All scans read centrally –blinded to clinical details, treatment allocation, and later events Detailed assessment of –site, size, location of early ischaemic change & swelling –ASPECTS score –presence of of hyperdense artery –haemorrhage, non stroke lesions etc.

48. Scale to assess swelling and mass effect Wardlaw AJNR 1994

49. ASPECTS Score Each of 10 MCA territory regions scored normal/ abnormal for hypodensity, and total no. abnormal areas subtracted from 10, so most extensive visible MCA ischaemia score = 0 (worst), no visible ischaemic change = 10 (best)

50. Material and methods We included in this analysis data from all patients randomised, for whom we had on 15.12.05: –Digitised copies of baseline and follow-up CT * (we excluded patients with MR as baseline scan, or who had no follow-up CT) –The (blinded) assessment by the expert central reader of both scans entered in database Analyses presented are based on the 389 patients who met these criteria

51. Frequency of hyperdense artery on baseline and follow-up CT Present on baseline scan 152 (39%) Present on follow-up scan102 (26%) Persisted (seen on 1 st & 2 nd scan) 88 (23%) Present on baseline, disappeared by 2 nd scan 64 (16%)

52. Baseline CT: hyperdense artery and early ischaemic changes & swelling Hyperdense artery present n (%) Hyperdense artery absent n (%) p Loss of grey/white matter definition128 (84%)101 (43%) <0.001 Loss of basal ganglia outline98 (64%)56 (24%) <0.001 Hypodensity27 (18%)19 (8%) 0.004 Any mass effect96 (63%)58 (24%) <0.001 ASPECTS score, mean5.06.3 <0.001

53. Follow-up CT: ischaemic change and swelling in patients with and without persisting hyperdense artery Hyperdense artery persisted n (%) Hyperdense artery disappeared n (%) p Loss of grey/white definition81 (92%)50 (82%)0.06 Loss of basal ganglia outline71 (81%)42 (69%)0.2 Hypodensity81 (92%)53 (87%)0.3 Any sign of mass effect80 (91%)51 (84%)0.2 ASPECTS score (mean)3.44.90.002

54. Summary In IST-3, the hyperdense artery was seen on baseline CT in 152/389 patients (39%) In 64/152 (42%) the hyperdense artery was no longer visible on the follow-up CT Compared with patients without hyperdense artery sign (HAS), patients with persisting HAS showed more extensive ischaemic change on both the baseline and follow-up CT IST-3 will therefore have substantial data to assess the influence of the hyperdense artery sign and early ischemic change on the response to i.v. rt-PA

55. Lots of new features on the website! www.ist3.com

56. Study protocol

57. Hypothesis #1 Intravenous thrombolysis with rt-PA administered to a wide range of patients with acute ischaemic stroke within six hours of symptom onset will increase the proportion of patients alive and independent at six months.

58. Hypothesis #2 The balance of risks and benefits of thrombolysis may be modified by clinically important variables, including: –age –delay in treatment, –stroke severity, –initial brain scan appearances.

59. Main features of the IST - 3 International, multi-centre, prospective, randomised open treatment blinded end-point design (PROBE) study. Large and pragmatic - 6000 patients. Eligibility criteria simple Blinded assessment of outcome at 6 months. Intention to treat analysis. Pharmaceutical industry not involved in trial development, operation or data management.

60. Exclusion Criteria (1) Contraindications to thrombolysis: Major surgery, trauma or gastrointestinal or urinary tract haemorrhage within the previous 21 days. Arterial puncture at a non-compressible site within the previous 7 days. Currently on oral anticoagulant or intravenous heparin therapy with abnormal coagulation (APPT and/or INR). Known defect of clotting or platelet function. Women of childbearing potential (unless pregnancy impossible) or known to be breastfeeding.

61. Exclusion Criteria (2) Not specified but might include: Prognosis very poor regardless of therapy; likely to be dead within months. Unlikely to be available for follow-up (eg, no fixed home address, visitor from overseas).

62. Eligibility Criteria (1) General Indications: Patients older than 16 years of age. Symptoms and signs of clinically definite acute stroke (mild, moderate or severe). A clearly established time of stroke onset that will allow thrombolysis to be given within six hours. CT or MR brain scanning has excluded intracranial haemorrhage.

63. Eligibility Criteria (2) If your patient : Has a clear indication for thrombolysis: Treat with rt-PA. Has a clear contraindication to thrombolysis: DO NOT treat with rt-PA. Thrombolysis is ‘promising but unproven,’ consider RANDOMISING the patient in IST- 3.

64. Trial procedure: Brain imaging Mandatory initial scan (CT or MRI) before randomisation to rule out haemorrhage and stroke mimics. Patients eligible even if early ischaemic changes visible. Repeat CT at 24-48 hrs after randomisation.

65. Trial procedure: Ethics and Consent Each participating centre must obtain approval from the appropriate Ethics Committee / IRB Before randomisation, consent must be obtained from the patient or their relatives (or legal representative).

66. Randomisation (1) 24 hour computerised randomisation service: telephone or via ist3 website (www.ist3.com) Enter key demographic items and baseline data Computer checks data for consistency. Treatment allocated immediately.

67. Patients allocated thrombolysis Admit patient to acute stroke unit Drug: recombinant tissue-type plasminogen activator (rt-PA, Alteplase - Boehringer Ingelheim). Dose: total dose of 0.9mg per kg of body weight up to a maximum of 90mg. Regimen: 10% of the total dose given as an intravenous bolus delivered over one minute, with the remainder infused over the next 60 minutes. Avoid aspirin and heparin for 24 hours All other standard treatments as usual Clinical monitoring according to IST-3 protocol

68. Patients allocated control Admit patient to acute stroke unit Patient must NOT be given rt-PA. Aspirin can be started immediately All other standard treatments as usual Clinical monitoring according to IST-3 protocol

69. Monitoring vital signs and neurological status after randomisation Every 15 minutes for 2 hours (use manual not automated method for BP) Every 30 minutes for the next 6 hours Hourly for a further 6 hours 4 hourly for the next 36 hours If there is any cause for concern, review, report, document and increase observation frequency accordingly

70. Trial procedure: Haemorrhages during thrombolysis Stop infusion immediately if significant bleeding (intra- or extracranial) occurs. Monitor blood pressure, maintain circulating blood volume, transfuse blood as appropriate for patients with major extracranial bleeding. +/- neurosurgical opinion for intracranial haemorrhages.

71. Data collection at 7 days Demographic data. Treatments given in hospital Major clinical events within 7 days Contact details of patient to permit follow-up at 6 months (family doctor, address of relatives etc.). Send pre-randomisation and follow-up CT scans to Edinburgh (you can send via email, web, or post)

72. Data collection at six months Is the patient alive and independent? Modified Rankin, quality of life Assessed blind with a validated postal or telephone questionnaire.

73. Events within 7 days Deaths from any cause. Symptomatic intracranial haemorrhage (fatal and non-fatal). Asymptomatic intracranial haemorrhage (repeat brain imaging data). Major extracranial haemorrhage (fatal, or requiring transfusion or operation). Recurrent ischaemic stroke.

74. Data analyses: ‘intention to treat’ Subsidiary analyses, subdivided by : Clinical stroke syndrome (OCSP) Presence or absence of atrial fibrillation Pre-randomisation scan appearances Pre-randomisation antiplatelet use Baseline risk Blood pressure at randomisation Antihypertensive use following randomisation Age Sex

75. Sample size: 6,000 patients Assuming a power of 80%, an alpha level of 5%, with 6000 patients, mostly treated between 3 & 6 hours of onset, the trial could detect a 3% absolute difference in the primary outcome (the proportion of patients dead or dependent at 6 months). This absolute difference is clinically worthwhile, is consistent with the effect size observed among patients randomised between 3 & 6 hours of stroke onset in the Cochrane review of the rt-PA trials. It is also comparable with the absolute benefit seen with thrombolytic therapy for acute MI. If 3500 patients were recruited, the trial could detect a 4% absolute difference in the primary outcome. A sample size of 1000 patients could detect a 7% absolute difference in the primary outcome, which is consistent with the effect size among patients randomised within 3 hours of stroke in the Cochrane review.

76. Which imaging method to select patients for thrombolysis?

77. Do we need to see ‘mismatch’ with CT Perfusion/MR-DWI/PWI? –NO! 1.No reliable evidence that patients with mismatch benefit from thrombolysis – DIAS trial + 2.50% of patients without mismatch get lesion growth, so could still benefit from treatment 3.Extra imaging increases delays to start of treatment 4.No consensus on definition of mismatch or how to assess perfusion

78. Do we need to see a blocked artery with CTA, MRA, IAA? – NO! Dogma: “only treat a blocked artery…” Block may be below resolution of imaging (No obvious block on CTA, MRA, IAA does not exclude a blocked vessel)? Lacunar stroke - Lacunar infarcts in NINDS benefited as much as other stroke subgroups Its not just the big vessels – what about the microcirculation – it needs to be unblocked too.

79. Further technique-specific issues CT Perfusion/CTA Contrast agents delay clot lysis times by >50% (Morcos Eur Radiol 2005;15:1463, Pislaru J Am Coll Cardiol 1998;32:1102, Dehmer J Am Coll Cardiol 1995;25:1069) DWI/PWI 20-30% of acute stroke patients cannot be scanned for medical reasons or MR contraindications (Hand et al JNNP 2005, Barber et al JNNP 2005, Singer Neurology 2004;62:1848) IST imaging strategy Plain CT is the primary method, but MRI permitted. Need to gain maximal information from CT Imaging sub-studies planned

80. CT scanning

81. web-based CT reading and feedback system: Log on to www.neuroimage.co.ukwww.neuroimage.co.uk Register Do first 20 scans (2 batches) - get 1 CPD credit -get feedback Do all six batches - get 5 CPD credits IST-3: Training to read CT scans what you, the reference standard, five experts and all other specialties said about that scan, and a follow-up scan to see where the infarct appeared

82. IST-3 Imaging: Training materials to read scans

83. Early infarct signs on CT 4 hours24 hours Hypodensity : loss of grey/white differentiation, loss of the lentiform nucleus Swelling : effacement of sulci, squashing the ventricle

84. IST-3. CT scan at baseline: Joanna Wardlaw’s opinion Acute ischaemic change72% Hyperdense artery44% Periventricular lucencies40% Old vascular lesion39% Normal4% Non-stroke lesion*1% Haemorrhage0% categories are not mutually exclusive: a patient may have more than one feature. *both lesions were small incidental meningiomas

85. IST3 plans for CT analysis and international CT reading panel and its methods

86. Maximising information from CT. Need to clarify relationships between: Treatment risk (haemorrhage) and: presence of ‘early infarct’ signs white matter lesions old infarcts/haemorrhages Treatment benefit and ‘early infarct’ signs how much density change = irreversible infarction? is swelling without hypodensity important? how important is dense artery sign?

87. CT ‘at 2 hrs’ after onset. On closer questioning, true onset was at least 8 hrs before CT. 48 hrs How hypodense?

88. This is ‘marked hypodensity’: Lessons from this case 1. Infarct has very clearly demarcated edges, and marked hypodensity = ?irreversible infarction 2. Patients with right hemisphere symptoms may not notice when their stroke starts due to “neglect” (anosognosia). 3. Always review the patient’s history if CT suggests that the infarct is older than the stated time of onset would suggest.

89. Baseline < 3 hours Early left basal ganglia hypodensity Follow-up Haemorrhagic transformation Is this too much hypodensity?

90. Swelling without density change? (CT at 2 hrs) 5 hrs: swelling + density change

91. 4 days – infarct in density change and swelling areas IST3 00188 5 hrs - then swelling + density change

92. Dense artery sign: what does it predict?

93. IST3 00338 24 hours later – persistent dense MCA (thrombus) Dense MCA & haemorrhagic transformation 4.5 hours after acute left hemiparesis

94. IST3 00343 48 hours later – persistent MCA occlusion Persistent dense MCA & massive swelling Acute right hemiparesis 4 hours ago

95. Dense artery sign: what does it predict? A persistently dense MCA at 24-48 hrs may predict adverse events (bleeding or massive swelling) IST-3 will establish whether rt-PA causes disappearance of dense MCA sign and reduces these adverse outcomes

96. What about microhaemorrhages, which can be seen on CT (sometimes)? CT MR GRE

97. Old haemorrhage on CT, and rt-PA White linear density; slit-like hole haemosiderin

98. Purpose of central CT reading. Clarify relationships between: Early infarct signs and haemorrhage risk Early infarct signs and treatment benefit - how much density change? - how much swelling? White matter lesions & haemorrhage risk Old infarcts/haemorrhages and risk Appearance/disappearance of dense artery sign and response to treatment

99. IST3 – plans for CT reading Central collection = potentially 12, 000 CT’s! Electronic storage Reading by panel of 20 experts Web-based CT reading system developed from ACCESS study All scans read once Subset read by all readers CT reading advisory group met 6 th October, panel of 20 experts being assembled Preliminary analyses on first 500 patients being submitted for Lancet

100. Edinburgh