Correspondence
Stroke thrombolysis can cause potentially fatal intracerebral haemorrhage, but advocates claim the potential reduction in disability justifies this risk. Alteplase was authorised following the National Institute of Neurological Disorders and Stroke (NINDS) trial.1 A 2004 review2 raised concerns over the trial data. Outcomes across the centres differed considerably. Bias could explain the observation that the plot of outcome (modified Rankin score 0–1) against number of patients recruited does not resemble the expected symmetrical funnel (figure). The review2 highlights concerns with local randomisation resulting in imbalance of stroke severity between the two study groups. 50% of patients within the 0–90 min window were reportedly randomised between 89 and 90 min. The poor prognosis of placebo patients recruited between 91 and 133 min resulted in exceptionally high odds for a favourable outcome. When the imbalance was considered, the estimated benefits were substantially reduced.3 Unbalanced randomisation might explain why 30-day mortality was improved in NINDS but not replicated elsewhere. For blood pressure, “non-compliance with the defined protocol was substantial, and persistent, throughout the study”.2 The review2 clarified that in some instances pharmacological monitoring was done by representatives (nurses) of the sponsoring pharmaceutical firm. Incomplete blinding in alteplase trials could also produce bias.4 First, treatment very often causes visible bleeding. Second, the possibility that investigators could have distinguished either the freeze-dried or the initially frothy reconstituted alteplase from matching placebo has not been openly examined and www.thelancet.com Vol 384 August 23, 2014
reported. Third, observers monitoring outcomes potentially had knowledge of events around randomisation. Three industry sponsored trials (ECASS, ECASS II, and ECASS III), each of more than 500 participants, only improved post-stroke independence by 6%, 8%, and 5% respectively, with a fourth (ATLANTIS) reporting a 4% deficit.4 30-day mortality was similar in two studies, and 5% worse in two. The one positive trial of these four (ECASS III) had randomised 14·1% of patients with a history of stroke to placebo and 7·7% to alteplase (p=0·003).5 The authors of a Cochrane review4 conclude by suggesting that cautious physicians “may choose not to use the treatment at all”. In two pooled analyses promoting the benefit of alteplase within 4·5 h, the questionable quality of the raw data received no comment.6,7 The 2009 Cochrane review,4 echoing the 2004 NINDS review,2 suggests that these data pooling and modelling “may be incorrect”. Although a plausible spread of disability scores in observational stroke cohorts is observed (appendix figures 1 and 2), by contrast the scores from patients in the alteplase trials are unevenly distributed (appendix figures 3, 4, and 5).7–9 The trial data
within 180 min are more uneven and come predominantly from NINDS. Furthermore, observational treatment cohorts examining the 3–4·5 h window ignored data available from untreated patients—with a 3% lower mortality rate.7,8 When these studies compare less than 3 h with 3–4·5 h delays, their strikingly similar outcome profiles undermine the case for early administration (appendix figure 2).8 Consequently, results from the larger third International Stroke Trial (IST-3) were eagerly awaited.10,11 Concerns, however, exist. After initially providing drug and placebo, Boehringer Ingelheim discontinued the supply. A published explanation for this decision is awaited. The double-blind trial then became open—likely yielding positive responses from treated participants. Before the final analysis, the university trial statistician left, a new statistician was recruited externally, and he was granted access to the unmasked data before the statistical analysis plan was finalised.12 Although the plan was prepared without knowledge of the unblinded data, the trial statistician and the authors of the statistical analysis plan were contemporaneously working on the IST-3 steering and writing committees.10,12
See Editorial page 638
See Online for appendix
150
Participants per centre
Questions about authorisation of alteplase for ischaemic stroke
–40
–30
–20
100
50
–10 0 +10 Change in favourable outcome (%)
+20
+30
+40
Figure: Funnel plot of change in favourable outcome versus participants per centre in the NINDS trial Outcome was favourable if the modified Rankin score was 0–1. Data are from the t-PA Review Committee.2
Submissions should be made via our electronic submission system at http://ees.elsevier.com/ thelancet/
659
Correspondence
IST-3 showed no benefit of alteplase administered within 6 h of onset on the primary outcome of being alive and independent at 6 months.10 No trend was found relating outcome with delay to treatment. 4% excess early mortality was reflected in the Kaplan-Meier curve (see appendix in reference 10). Treated patients had a 7% higher use of high-dependency beds than did untreated patients. Eventual place of residence, unpublished to date but less prone to responder bias than modified Rankin score, indicated no benefit. Recall bias could explain improvements in the complex ordinal analysis (see table 8 in appendix in reference 11). Finally, the 6–18 month survival curves, predicted to diverge if treated survivors were less disabled, remained inseparable.9,11 These observations and concerns over balanced evaluation suggest that the evidence of benefit is precarious. Comprehensive and coordinated care is effective, but it would be prudent for regulatory bodies to carefully reconsider authorisation for the use of alteplase in ischaemic stroke. I declare no competing interests. I thank physicians in the West Midlands who have helped to stimulate this investigation, and particularly Peter Wilmshurst.
Roger Shinton [email protected] See Correspondence page 659
24 Belle Vue Terrace, Hampton-in-Arden, Solihull, West Midlands B92 0AR, UK 1
2
3
4
5
660
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: 1581–87. O’Fallon M, Asplund K, Goldfrank LR, et al. Report of the t-PA Review Committee. 2004. http://stroke.nih.gov/ resources/t-pa-review-committee.htm (accessed June 10, 2014). Hoffman JR, Schriger DL. A graphic reanalysis of the NINDS trial. Ann Emerg Med 2009; 54: 329–36. Wardlaw JM, Murray V, Berge E, Del Zoppo GJ. Thrombolysis for acute ischaemic stroke. Cochrane Database Syst Rev 2009; 4: CD000213. Hacke W, Kaste M, Bluhmki E, et al. for the European Cooperative Acute Stroke Study (ECASS) investigators. Alteplase compared with placebo within 3 to 4.5 hours for acute ischemic stroke. N Engl J Med 2008; 359: 1317–29.
6
7
8
9
10
11
12
The ATLANTIS, ECASS, and NINDS rt-PA Study Group Investigators. Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials. Lancet 2004; 363: 768–74. Lees KR, Bluhmki E, von Kummer R, et al, for the ECASS, ATLANTIS, NINDS and EPITHET rt-PA Study Group Investigators. Time to treatment with intravenous alteplase and outcome in stroke: an updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET trials. Lancet 2010; 375: 1695–703. Wahlgren N, Ahmed N Davalos A, et al, Thrombolysis with alteplase 3-4·5 h after acute ischaemic stroke (SITS-ISTR): an observational study. Lancet 2008; 372: 1303–09. Slot KB, Berge E, Dorman P, Lewis S, Dennis M, Sandercock P, and the Oxfordshire Community Stroke Project, the International Stroke Trial (UK) and the Lothian Stroke Register. Impact of functional status at six months on long term survival in patients with ischaemic stroke: prospective cohort studies. BMJ 2008; 336: 376–79. The IST-3 collaborative group. The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (IST-3): a randomised controlled trial. Lancet 2012; 379: 2352–63. The IST-3 collaborative group. Effect of thrombolysis with alteplase within 6 h of acute ischaemic stroke on long-term outcomes (the third International Stroke Trial [IST-3]): 18-month follow-up of a randomised controlled trial. Lancet Neurol 2013; 12: 768–76. Sandercock P, Lindley R, Wardlaw J, Whiteley W, Murray G, and the IST-3 collaborative group. Statistical analysis plan for the third International Stroke Trial (IST-3); part of a ‘thread’ of reports of the trial. Int J Stroke 2012; 7: 186–87.
Alteplase for ischaemic stroke—responses Roger Shinton’s Correspondence contains numerous factual inaccuracies and statements that are not consistent with the text of the cited references, and do not in our opinion merit a reconsideration of the authorisation of alteplase in acute ischaemic stroke. W M O’Fallon and colleagues’ reanalysis of the NINDS trial data did not “raise concerns”,1 but concluded, “despite an increased incidence of symptomatic intracerebral hemorrhage in t-PA treated patients and subgroup imbalances in baseline stroke severity, when the drug was administered according to the study protocol, there was a statistically significant, and clinically important, benefit of t-PA…”.1
The funnel plot of treatment effect by centre does not provide evidence of bias. The apparent outlier at the bottom left of the plot is based on 14 patients at one centre. In one treatment group, three of six patients had a favourable outcome, and in the other, one of eight had a favourable outcome. These numbers are too small for any meaningful estimate of the treatment effect. Without this outlying point, there is little suggestion of centre-to-centre variability. Importantly, the funnel plot is unadjusted, whereas O’Fallon and colleagues1 explored centre effects more formally (and appropriately) by fitting models with treatment-by-centre interactions, which showed no statistical evidence of differences in treatment effect between the centres. The nurses recording blood pressure in NINDS were not employed by the drug manufacturer (Genentech), but by the US National Institutes of Health. Genentech monitors provided Food and Drug Administration-required coding of drugs and source document verification, but were not involved in trial data collection. The initial donation of drug and placebo for IST-3 was only ever agreed for the start-up phase. Boehringer Ingelheim informed the IST-3 investigators at the end of the pilot phase that they would not provide further supplies because their priority was to invest all available resources in completion of the randomised licensing trial (ECASS III) and international treatment registry (SITS register), specifically requested by the European Medicines Agency. Blinding in all the trials other than IST-3 was not compromised by frothing of the active drug, since the placebo agent frothed in the same way (due to the arginine excipient). Moreover, the “visible bleeding” that might unblind the treating physician occurs in less than 10% of patients, and the team members treating the patient in the acute phase were not involved in the 3-month or 6-month assessments in any of the referenced trials. www.thelancet.com Vol 384 August 23, 2014
Stroke thrombolysis can cause potentially fatal intracerebral haemorrhage, but advocates claim the potential reduction in disability justifies this risk. Alteplase was authorised following the National Institute of Neurological Disorders and Stroke (NINDS) trial.1 A 2004 review2 raised concerns over the trial data. Outcomes across the centres differed considerably. Bias could explain the observation that the plot of outcome (modified Rankin score 0–1) against number of patients recruited does not resemble the expected symmetrical funnel (figure). The review2 highlights concerns with local randomisation resulting in imbalance of stroke severity between the two study groups. 50% of patients within the 0–90 min window were reportedly randomised between 89 and 90 min. The poor prognosis of placebo patients recruited between 91 and 133 min resulted in exceptionally high odds for a favourable outcome. When the imbalance was considered, the estimated benefits were substantially reduced.3 Unbalanced randomisation might explain why 30-day mortality was improved in NINDS but not replicated elsewhere. For blood pressure, “non-compliance with the defined protocol was substantial, and persistent, throughout the study”.2 The review2 clarified that in some instances pharmacological monitoring was done by representatives (nurses) of the sponsoring pharmaceutical firm. Incomplete blinding in alteplase trials could also produce bias.4 First, treatment very often causes visible bleeding. Second, the possibility that investigators could have distinguished either the freeze-dried or the initially frothy reconstituted alteplase from matching placebo has not been openly examined and www.thelancet.com Vol 384 August 23, 2014
reported. Third, observers monitoring outcomes potentially had knowledge of events around randomisation. Three industry sponsored trials (ECASS, ECASS II, and ECASS III), each of more than 500 participants, only improved post-stroke independence by 6%, 8%, and 5% respectively, with a fourth (ATLANTIS) reporting a 4% deficit.4 30-day mortality was similar in two studies, and 5% worse in two. The one positive trial of these four (ECASS III) had randomised 14·1% of patients with a history of stroke to placebo and 7·7% to alteplase (p=0·003).5 The authors of a Cochrane review4 conclude by suggesting that cautious physicians “may choose not to use the treatment at all”. In two pooled analyses promoting the benefit of alteplase within 4·5 h, the questionable quality of the raw data received no comment.6,7 The 2009 Cochrane review,4 echoing the 2004 NINDS review,2 suggests that these data pooling and modelling “may be incorrect”. Although a plausible spread of disability scores in observational stroke cohorts is observed (appendix figures 1 and 2), by contrast the scores from patients in the alteplase trials are unevenly distributed (appendix figures 3, 4, and 5).7–9 The trial data
within 180 min are more uneven and come predominantly from NINDS. Furthermore, observational treatment cohorts examining the 3–4·5 h window ignored data available from untreated patients—with a 3% lower mortality rate.7,8 When these studies compare less than 3 h with 3–4·5 h delays, their strikingly similar outcome profiles undermine the case for early administration (appendix figure 2).8 Consequently, results from the larger third International Stroke Trial (IST-3) were eagerly awaited.10,11 Concerns, however, exist. After initially providing drug and placebo, Boehringer Ingelheim discontinued the supply. A published explanation for this decision is awaited. The double-blind trial then became open—likely yielding positive responses from treated participants. Before the final analysis, the university trial statistician left, a new statistician was recruited externally, and he was granted access to the unmasked data before the statistical analysis plan was finalised.12 Although the plan was prepared without knowledge of the unblinded data, the trial statistician and the authors of the statistical analysis plan were contemporaneously working on the IST-3 steering and writing committees.10,12
See Editorial page 638
See Online for appendix
150
Participants per centre
Questions about authorisation of alteplase for ischaemic stroke
–40
–30
–20
100
50
–10 0 +10 Change in favourable outcome (%)
+20
+30
+40
Figure: Funnel plot of change in favourable outcome versus participants per centre in the NINDS trial Outcome was favourable if the modified Rankin score was 0–1. Data are from the t-PA Review Committee.2
Submissions should be made via our electronic submission system at http://ees.elsevier.com/ thelancet/
659
Correspondence
IST-3 showed no benefit of alteplase administered within 6 h of onset on the primary outcome of being alive and independent at 6 months.10 No trend was found relating outcome with delay to treatment. 4% excess early mortality was reflected in the Kaplan-Meier curve (see appendix in reference 10). Treated patients had a 7% higher use of high-dependency beds than did untreated patients. Eventual place of residence, unpublished to date but less prone to responder bias than modified Rankin score, indicated no benefit. Recall bias could explain improvements in the complex ordinal analysis (see table 8 in appendix in reference 11). Finally, the 6–18 month survival curves, predicted to diverge if treated survivors were less disabled, remained inseparable.9,11 These observations and concerns over balanced evaluation suggest that the evidence of benefit is precarious. Comprehensive and coordinated care is effective, but it would be prudent for regulatory bodies to carefully reconsider authorisation for the use of alteplase in ischaemic stroke. I declare no competing interests. I thank physicians in the West Midlands who have helped to stimulate this investigation, and particularly Peter Wilmshurst.
Roger Shinton [email protected] See Correspondence page 659
24 Belle Vue Terrace, Hampton-in-Arden, Solihull, West Midlands B92 0AR, UK 1
2
3
4
5
660
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: 1581–87. O’Fallon M, Asplund K, Goldfrank LR, et al. Report of the t-PA Review Committee. 2004. http://stroke.nih.gov/ resources/t-pa-review-committee.htm (accessed June 10, 2014). Hoffman JR, Schriger DL. A graphic reanalysis of the NINDS trial. Ann Emerg Med 2009; 54: 329–36. Wardlaw JM, Murray V, Berge E, Del Zoppo GJ. Thrombolysis for acute ischaemic stroke. Cochrane Database Syst Rev 2009; 4: CD000213. Hacke W, Kaste M, Bluhmki E, et al. for the European Cooperative Acute Stroke Study (ECASS) investigators. Alteplase compared with placebo within 3 to 4.5 hours for acute ischemic stroke. N Engl J Med 2008; 359: 1317–29.
6
7
8
9
10
11
12
The ATLANTIS, ECASS, and NINDS rt-PA Study Group Investigators. Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials. Lancet 2004; 363: 768–74. Lees KR, Bluhmki E, von Kummer R, et al, for the ECASS, ATLANTIS, NINDS and EPITHET rt-PA Study Group Investigators. Time to treatment with intravenous alteplase and outcome in stroke: an updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET trials. Lancet 2010; 375: 1695–703. Wahlgren N, Ahmed N Davalos A, et al, Thrombolysis with alteplase 3-4·5 h after acute ischaemic stroke (SITS-ISTR): an observational study. Lancet 2008; 372: 1303–09. Slot KB, Berge E, Dorman P, Lewis S, Dennis M, Sandercock P, and the Oxfordshire Community Stroke Project, the International Stroke Trial (UK) and the Lothian Stroke Register. Impact of functional status at six months on long term survival in patients with ischaemic stroke: prospective cohort studies. BMJ 2008; 336: 376–79. The IST-3 collaborative group. The benefits and harms of intravenous thrombolysis with recombinant tissue plasminogen activator within 6 h of acute ischaemic stroke (IST-3): a randomised controlled trial. Lancet 2012; 379: 2352–63. The IST-3 collaborative group. Effect of thrombolysis with alteplase within 6 h of acute ischaemic stroke on long-term outcomes (the third International Stroke Trial [IST-3]): 18-month follow-up of a randomised controlled trial. Lancet Neurol 2013; 12: 768–76. Sandercock P, Lindley R, Wardlaw J, Whiteley W, Murray G, and the IST-3 collaborative group. Statistical analysis plan for the third International Stroke Trial (IST-3); part of a ‘thread’ of reports of the trial. Int J Stroke 2012; 7: 186–87.
Alteplase for ischaemic stroke—responses Roger Shinton’s Correspondence contains numerous factual inaccuracies and statements that are not consistent with the text of the cited references, and do not in our opinion merit a reconsideration of the authorisation of alteplase in acute ischaemic stroke. W M O’Fallon and colleagues’ reanalysis of the NINDS trial data did not “raise concerns”,1 but concluded, “despite an increased incidence of symptomatic intracerebral hemorrhage in t-PA treated patients and subgroup imbalances in baseline stroke severity, when the drug was administered according to the study protocol, there was a statistically significant, and clinically important, benefit of t-PA…”.1
The funnel plot of treatment effect by centre does not provide evidence of bias. The apparent outlier at the bottom left of the plot is based on 14 patients at one centre. In one treatment group, three of six patients had a favourable outcome, and in the other, one of eight had a favourable outcome. These numbers are too small for any meaningful estimate of the treatment effect. Without this outlying point, there is little suggestion of centre-to-centre variability. Importantly, the funnel plot is unadjusted, whereas O’Fallon and colleagues1 explored centre effects more formally (and appropriately) by fitting models with treatment-by-centre interactions, which showed no statistical evidence of differences in treatment effect between the centres. The nurses recording blood pressure in NINDS were not employed by the drug manufacturer (Genentech), but by the US National Institutes of Health. Genentech monitors provided Food and Drug Administration-required coding of drugs and source document verification, but were not involved in trial data collection. The initial donation of drug and placebo for IST-3 was only ever agreed for the start-up phase. Boehringer Ingelheim informed the IST-3 investigators at the end of the pilot phase that they would not provide further supplies because their priority was to invest all available resources in completion of the randomised licensing trial (ECASS III) and international treatment registry (SITS register), specifically requested by the European Medicines Agency. Blinding in all the trials other than IST-3 was not compromised by frothing of the active drug, since the placebo agent frothed in the same way (due to the arginine excipient). Moreover, the “visible bleeding” that might unblind the treating physician occurs in less than 10% of patients, and the team members treating the patient in the acute phase were not involved in the 3-month or 6-month assessments in any of the referenced trials. www.thelancet.com Vol 384 August 23, 2014
