on behalf of the TARDIS Investigators Continue or Stop: the rationale for early closedown of a clinical trial as a result of stopping rules within the Data Monitoring Committee charter Lisa J Woodhouse, Jason P Appleton, Lelia Duley, Ian Ford, Didier Leys, Stuart Pocock, Nikola Sprigg, Catherine Sudlow, Matthew Walters, and Philip M Bath; on behalf of the TARDIS Investigators

Disclosures TARDIS funded by: British Heart Foundation NIHR Health Technology Assessment programme Indirect funding from: The Stroke Association (for funding of Division of Stroke Medicine, University of Nottingham) NIHR Stroke Research Network/CRN Stroke No commercial relationships: Drugs supplied by recruiting hospitals Drugs: Individual drugs are licensed for use after stroke/TIA

Trial Triple Antiplatelet for Reducing Dependency after Ischaemic Stroke (TARDIS) International, multicentre, parallel-group, prospective, randomised, open-label, blinded-endpoint controlled trial [1] 1. TARDIS Investigators. Int J Stroke 2015;10:1159-65 www.tardistrial.org 3

Design ADC AD or C AD or C AD or C Randomised to either intensive (combination of Aspirin, Dipyridamole and Clopidogrel) or guideline (either Aspirin plus Dipyridamole or Clopidogrel alone) antiplatelet therapy Start up phase, 3.5 years April 2009 to Sept 2012 902 (aim 350+) patients from UKSRN centres Main Phase , 5 years Oct 2012 to Sept 2017 Start-up phase shows acceptable safety Expand recruitment to 4,100 patients Day 0 Day 30 Day 90 ADC AD or C R AD or C AD or C 4

Aim Assess if 30 days of intensive (triple) antiplatelet therapy is safe and effective in reducing recurrent cerebral ischaemic events after non-cardioembolic stroke or TIA in comparison with guideline antiplatelet therapy

Main outcomes of interest Primary efficacy outcome (90 days): Recurrent Stroke/TIA by severity, using mRS See poster for more information Primary safety outcome: Bleeding by severity 6

Recruitment Commenced 7 April 2009 Halted 18 March 2016 on the advice of the independent Data Monitoring Committee (DMC) 3096 participants recruited (75.5% of planned 4100)

Data Monitoring Committee (DMC) Followed a pre-defined charter Included stopping rules for safety and efficacy Did not include any stopping rules for futility

DMC: Charter stopping rules SAFETY: Primary outcome favoured the control group, P<0.01 (nominal, 2-sided) Combined outcome of fatal or non-fatal stroke or major bleeding favoured the control group, P<0.01 (nominal, 2-sided) Overall rate of symptomatic intracranial haemorrhage exceeded 2% During start-up phase, major bleeding favoured the control group, P<0.01 (nominal, 2-sided) EFFICACY: Conduct formal interim analyses, after 40% and 70% of participants completed day 90 outcome assessment Combined outcome of fatal or non-fatal stroke or major bleeding event favoured the control group, P<0.001 (2-sided)

DMC: Stopping rules Before making any decision, the DMC was asked to consider: Overall internal and external evidence Multiplicity of testing Possibility that trends in the data might be reversed with longer follow-up or increased recruitment

DMC: Meetings Reviewed unblinded data in confidence every 6 months Met on 13 occasions Recommended trial continuation for all but the last data review

Recommendation to stop was based on three observations DMC: Stop decision Recommendation to stop was based on three observations

DMC: Stop reason 1 “More than doubling of the risk of bleeding in all categories of bleeding, with a clearly statistically significant increase in the risk of the composite of major and fatal bleeds” Not present during the start-up phase (as per the pre-defined stopping rules) 38 vs 17, HR 2.16, 95% CI (1.21-3.84), p=0.009

DMC: Stop reason 2 “No compensatory evidence of benefit for the primary outcome” 100 vs 109, HR 0.90, 95% CI (0.69-1.19), p=0.47

DMC: Stop reason 3 A conditional power futility analysis suggested the trial was highly unlikely to ever demonstrate a significant difference in the primary outcome Conditional analysis performed by DMC Charter did not request or define this

DMC: Stop decision Following the DMCs recommendation: All patients on active treatment moved to control treatment Trial Steering Committee (TSC) reviewed the same data (and subsequent additional analyses) one month later Both TSC and DMC Chairs agreed that the trial should stop recruitment and complete follow-up

TSC: Additional interpretation Overall neutral findings with no difference in the net balance of death, stroke, myocardial infarction and major bleeding   Adjusted Unadjusted Outcome I G OR (95% CI) 2p Any stroke or Major/Fatal bleed 92 74 1.23 (0.89-1.69) 0.20 1.24 (0.91-1.71) 0.17 Stroke/TIA/ACS/ Death 124 139 0.87 (0.68-1.13) 0.30 0.87 (0.68-1.12) 0.29 Vascular death/NF stroke/NF MI/Major bleed 98 87 1.11 (0.82-1.50) 0.50 1.12 (0.83-1.51) 0.45 Death/Stroke/MI 90 93 0.95 (0.70-1.29) 0.75 0.96 (0.71-1.29) 0.76 Death/NF stroke/NF MI/Major bleed/Fatal bleed 111 104 1.04 (0.79-1.38) 0.77 1.06 (0.80-1.40) 0.68

Stopping: Conclusion DMC’s decision to stop was based on a combination of three observations: Significant increase in major (including fatal) bleeding No compensatory decrease in recurrent events Futility (on basis of conditional power analysis) TSC: Overall neutral findings Both DMC and TSC agreed to stop the trial with these findings > DMCs may need to recommend stopping trials in certain circumstances not defined in charter

Thank you