Browse the corpus

High blood pressure (BP) is present in 70% or more of patients with acute ischemic stroke and intracerebral hemorrhage (ICH) 1. Affected patients have a worse outcome, whether judged as early recurrence, death within a few weeks, or combined death and dependency after several months 1–4. Lowering BP might therefore reduce these events and improve functional outcome providing that cerebral perfusion is not reduced in the presence of dysfunctional cerebral autoregulation. However, recent large trials have been inconsistent and inconclusive in their results 5,6. Nitric oxide (NO) donors are candidate treatments for acute stroke: NO is a cerebral and systemic vasodilator, modulates vascular and neuronal function, and inhibits apoptosis 7. Preclinical studies of cerebral ischemia found that NO donors reduce stroke lesion size and improve regional cerebral blood flow (CBF) and functional outcome 8. Five small clinical studies of NO donors have been performed, these involving a total of 208 patients with recent stroke.

Nitric oxide (NO) donors are candidate treatments for acute stroke: NO is a cerebral and systemic vasodilator, modulates vascular and neuronal function, and inhibits apoptosis 7. Preclinical studies of cerebral ischemia found that NO donors reduce stroke lesion size and improve regional cerebral blood flow (CBF) and functional outcome 8. Five small clinical studies of NO donors have been performed, these involving a total of 208 patients with recent stroke. Intravenous sodium nitroprusside reduced BP without altering CBF and exhibited antiplatelet effects (thereby precluding its use in ICH) 9. Four pilot trials of transdermal glyceryl trinitrate (GTN) found that it lowered BP by approximately 8%; did not alter platelet function (and so could be given in ICH); did not alter middle cerebral artery blood flow velocity or regional CBF; improved aortic vascular compliance; and could be given to patients with dysphagia 10–13. No safety concerns were present in these studies, and in one small trial, ultra-acute treatment with GTN was associated with an improved functional outcome 13,14.

t alter middle cerebral artery blood flow velocity or regional CBF; improved aortic vascular compliance; and could be given to patients with dysphagia 10–13. No safety concerns were present in these studies, and in one small trial, ultra-acute treatment with GTN was associated with an improved functional outcome 13,14. On the basis of these preclinical and clinical data showing feasibility, tolerability and apparent safety of GTN, and the potential for efficacy, the large ‘Efficacy of Nitric Oxide in Stroke’ (ENOS) trial was started and is ongoing. ENOS is assessing, in a partial, factorial, prospective, randomized, single-blind, blinded-outcome design, whether to lower BP with GTN (vs. no GTN) and whether to continue (vs. stop) prestroke antihypertensive therapy. The trial commenced in 2001, and protocols for the main trial and an outline on the management of neuroimaging were published in 2006 and 2007, respectively 15,16. Several nontreatment-related and blinded analyses of the ENOS database have been published since the start of the trial 17–22. The independent Data Monitoring Committee have assessed the trial every six-months and on each occasion recommended that the trial should continue.

blished in 2006 and 2007, respectively 15,16. Several nontreatment-related and blinded analyses of the ENOS database have been published since the start of the trial 17–22. The independent Data Monitoring Committee have assessed the trial every six-months and on each occasion recommended that the trial should continue. Prior to presentation of the primary analyses in 2014, two further publications are planned, the statistical analysis plan (SAP) and a detailed listing of baseline characteristics. The accompanying Supporting Information Appendix S1 details the SAP and is presented prior to locking of the trial database (expected in late February) so that analyses are not data driven or selectively reported 23. Unusually, this SAP includes not just information on the two primary publications (GTN vs. no GTN, and continue vs. stop prestroke antihypertensive medication) but also provides detailed information on the intended baseline characteristics publication and the first set of secondary publications. The SAP also informs much of the content of the final trial report to be submitted to the Medical Research Council/Efficacy and Mechanism Evaluation Programme (EME); the final report will be submitted in the third quarter of 2014 for publication in the EME Journal, part of the National Institute for Health research collection of peer-reviewed open access journals.

e final trial report to be submitted to the Medical Research Council/Efficacy and Mechanism Evaluation Programme (EME); the final report will be submitted in the third quarter of 2014 for publication in the EME Journal, part of the National Institute for Health research collection of peer-reviewed open access journals. Importantly, the ENOS Trial Steering Committee have changed the original plan for the analysis of the primary outcome, as reported in the protocol (published in IJS) 15, from using an unadjusted binary ‘cut’ of the modified Rankin Scale (mRS 24, unadjusted comparison of mRS >2 between the treatment groups) to an adjusted ordinal analysis utilizing all seven levels of the mRS with adjustment for minimization variables. The change meant that the sample size could be reduced from 5000 patients to a minimum of 3500 patients assuming power of 90% and significance of 5%. The decision to change from dichotomous to polytomous analysis was not based on any interim analysis of the ENOS dataset; rather, it reflects the recognition that ordinal analyses are more efficient statistically (i.e. they provide improved statistical power for a given sample size) 25,26 as also shown for head injury trials 27. (The importance of this change is highlighted by recent trials that were technically neutral on their primary outcome when using a binary analysis but positive when analyzed secondarily using an ordinal analysis. 6,28) Similarly, adjusted analyses provide additional statistical power 29, are important if minimization is used during the process of randomization 30, and help address any minor imbalances present at baseline because of chance. As a result, these statistical approaches are likely to be more sensitive to any treatment effect and, as such, are recommended by the European Stroke Organization 31. The collection of all baseline data needed for covariate adjustment of the primary outcome should mean there is no need for imputation for missing data.

As a result, these statistical approaches are likely to be more sensitive to any treatment effect and, as such, are recommended by the European Stroke Organization 31. The collection of all baseline data needed for covariate adjustment of the primary outcome should mean there is no need for imputation for missing data. In the future, data from ENOS will be integrated into individual patient data meta-analyses of NO donors, and BP lowering, for acute stroke (the latter through the ‘Blood pressure in Acute Stroke Collaboration’), and made available to participating countries and the ‘Virtual International Stroke Trials Archive’ 32. Ultimately, a subset of the data will be made available over the web, as with the International Stroke Trial 33. Similarly, anonymized baseline and on-treatment neuroimaging data will be published 16. Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Appendix S1 Statistical Analysis Plan (ENOS).

Introduction This trial addresses an important focus for stroke rehabilitation: the ability of stroke survivors to recover the use of their arm and hand for everyday functional tasks such as picking up a cup, unscrewing a coffee jar lid, and doing up buttons and zippers. Limitations in performing such everyday tasks seriously affect capacity for independent living. At six-months after stroke, only 38% of people who receive rehabilitation have recovered some dexterity 1.

or everyday functional tasks such as picking up a cup, unscrewing a coffee jar lid, and doing up buttons and zippers. Limitations in performing such everyday tasks seriously affect capacity for independent living. At six-months after stroke, only 38% of people who receive rehabilitation have recovered some dexterity 1. Systematic reviews [for example 2] indicate that repetitive functional task-specific activity can enhance motor recovery. Two training modalities that have attracted considerable attention are constraint-induced movement therapy (CIMT), which consists of repetitive practice of functional tasks combined with constraint of the ipsilesional upper limb, and robot-assisted therapy (RAT), which provides repetitive practice of movements that are required for upper limb functional activity. CIMT is effective between three- and nine-months after stroke 3 but less so when given early after stroke 4. Furthermore, CIMT is suitable only where at least 10 degrees of active movement are present in the paretic thumb and two or more paretic fingers 5. This requirement for a high level of function excludes many stroke survivors 6. RAT has been used in a wider group of stroke survivors but has also been found to be no more effective than an equal dose of conventional therapy 7. Thus, the question of whether a novel treatment aimed at enhancing upper limb functional task-specific activity might be better than routine conventional physical therapy (CPT) remains unanswered.

der group of stroke survivors but has also been found to be no more effective than an equal dose of conventional therapy 7. Thus, the question of whether a novel treatment aimed at enhancing upper limb functional task-specific activity might be better than routine conventional physical therapy (CPT) remains unanswered. Functional strength training (FST) combines functional task-specific exercise and strength training, the latter of which is included because the largest impact on upper limb functional recovery after stroke may result from loss of muscle strength 8. Importantly, preliminary data indicate FST may be more effective than another therapy, movement performance therapy (MPT), at equal intensity 9. Both FST and MPT are used in routine CPT.

hich is included because the largest impact on upper limb functional recovery after stroke may result from loss of muscle strength 8. Importantly, preliminary data indicate FST may be more effective than another therapy, movement performance therapy (MPT), at equal intensity 9. Both FST and MPT are used in routine CPT. Evaluation of FST through properly designed clinical trials is crucial. We are aware, however, that a potentially serious barrier to the development of any novel treatment for poststroke motor impairment arises from poor understanding of its mechanisms of action and, in particular, whether it is likely to work in all stroke sub-groups 10. The CIMT and RAT trials recruited patients based on clinical phenotype, but in order to target therapy to those most likely to benefit, this may not be sufficient. It is increasingly recognized that large rehabilitation trials need to include more sophisticated measures of residual brain structure and function in order for researchers to understand the mechanisms of action of treatment and the characteristics of ‘responsive’ patients 10,11. For example, both the pretreatment level of brain activity in primary motor cortex during the performance of a motor task 12 and the degree of damage to descending motor white matter pathways 13 were associated with greater clinical improvement in 24 chronic stroke patients undergoing two-weeks of robot-based therapy. These studies are encouraging, but if we are to incorporate such data into models that can accurately predict therapeutic response, larger sample sizes are clearly required.

te matter pathways 13 were associated with greater clinical improvement in 24 chronic stroke patients undergoing two-weeks of robot-based therapy. These studies are encouraging, but if we are to incorporate such data into models that can accurately predict therapeutic response, larger sample sizes are clearly required. Specific objectives of this trial are the following: to determine whether CPT + FST commenced early after stroke produces greater improvements in upper limb motor recovery than CPT + MPT; to identify similarities and differences in the neural correlates of clinical improvement in response to (a) CPT + FST and (b) CPT + MPT; to determine whether any pretreatment parameters or combinations are sufficiently predictive of improvement in upper limb motor function in response to intervention to enable physical therapy to be targeted at those most likely to respond; to provide estimates of the preliminary cost-effectiveness of CPT + FST and inform the design of a subsequent pragmatic trial. Methods Design The trial will be a randomized, controlled, observer-blind trial (Fig. 1). Figure 1 Flow diagram to illustrate trial design. Study population Study criteria (combined inclusion and exclusion) are as follows: aged 18 + years; infarction in anterior cerebral circulation territory within the previous 60 days, confirmed by clinical neuroimaging; score at least 11/33 for Motricity Index pinch section 14 but unable to complete the nine-hole peg test (9HPT) 14 in 50 s or less; no obvious spatial neglect as defined by a score of 0 or 1 on Extinction and Inattention sub-scale of the NIH Stroke Scale;

infarction in anterior cerebral circulation territory within the previous 60 days, confirmed by clinical neuroimaging; score at least 11/33 for Motricity Index pinch section 14 but unable to complete the nine-hole peg test (9HPT) 14 in 50 s or less; no obvious spatial neglect as defined by a score of 0 or 1 on Extinction and Inattention sub-scale of the NIH Stroke Scale; no obvious motor dyspraxia or communication deficits as assessed by ability to imitate action with the nonparetic upper limb, as in previous pilot work 15; able, prior to the index stroke, to use the paretic upper limb to lift a cup and drink from it. Randomization The randomization sequence, generated in advance, will stratify participants by clinical center, time after stroke (up to 30 days and 31–60 days), and ability to use the paretic upper limb as assessed by the 9HPT 14 (substantial impairment = able to move one peg or less in 50 s; moderate impairment = able to move two or more pegs in 50 s). An independent telephone randomization service will maintain allocation concealment from the research team prior to randomization. An independent statistician at the Robertson Centre will generate the randomization sequence and will not be involved in developing the statistical analysis plan or programming the trial statistical analysis.

phone randomization service will maintain allocation concealment from the research team prior to randomization. An independent statistician at the Robertson Centre will generate the randomization sequence and will not be involved in developing the statistical analysis plan or programming the trial statistical analysis. Interventions The intervention phase will last for six-weeks. All participants will receive routine CPT provided by the clinical physiotherapists, as in previous pilot work 9. Training will be provided for them to record content and amount of therapy provided daily for home participants, outpatients, and inpatients. In each clinical center, the extra therapy will be provided by a research therapist responsible for either extra MPT (control) or extra FST (experimental). We will not tell clinical staff which research therapist is responsible for which treatment. This strategy is expected to minimize the potential for therapist bias, although allocation to different forms of exercise-based therapy is more difficult to conceal than, for example, an active or placebo drug, because of observable differences.

inical staff which research therapist is responsible for which treatment. This strategy is expected to minimize the potential for therapist bias, although allocation to different forms of exercise-based therapy is more difficult to conceal than, for example, an active or placebo drug, because of observable differences. Participants in both groups will take part in extra therapy, as prescribed and overseen by a research therapist through direct and nondirect contact, for up to 1·5 h a day, five-days a week for six-weeks. The research therapists will be trained to direct either FST or MPT in accordance with a standardized manual before the trial begins. Fidelity to the manuals will be assessed at the beginning and at regular points throughout the trial with little prior warning to therapists.

1·5 h a day, five-days a week for six-weeks. The research therapists will be trained to direct either FST or MPT in accordance with a standardized manual before the trial begins. Fidelity to the manuals will be assessed at the beginning and at regular points throughout the trial with little prior warning to therapists. Control intervention (CPT + MPT) MPT is the component of CPT which emphasizes movement quality rather than quantity. It is based on neurophysiological approaches for which there is evidence that they produce no effect compared with no treatment or a placebo 16. MPT involves joint and soft tissue mobilization, sensory stimulation, facilitation of muscle activity/movement, positioning, retraining normal patterns of movement, and education for the patient/carer. Emphasis is given to hands-on interventions provided by a therapist facilitating and guiding movement (therapist-dependent) to provide sensory input and to optimize postural control and joint alignment in preparation for voluntary movement. Some repetitive practice of functional tasks is included, but without systematic progression in resistance to movement. Feedback and instructions during MPT encourage an internal focus of attention on the movement performance, for example, amount and direction of elbow movement when lifting a teapot.

Control intervention (CPT + MPT) MPT is the component of CPT which emphasizes movement quality rather than quantity. It is based on neurophysiological approaches for which there is evidence that they produce no effect compared with no treatment or a placebo 16. MPT involves joint and soft tissue mobilization, sensory stimulation, facilitation of muscle activity/movement, positioning, retraining normal patterns of movement, and education for the patient/carer. Emphasis is given to hands-on interventions provided by a therapist facilitating and guiding movement (therapist-dependent) to provide sensory input and to optimize postural control and joint alignment in preparation for voluntary movement. Some repetitive practice of functional tasks is included, but without systematic progression in resistance to movement. Feedback and instructions during MPT encourage an internal focus of attention on the movement performance, for example, amount and direction of elbow movement when lifting a teapot. Experimental intervention (CPT + FST) FST involves repetitive progressive resistive exercise during goal-directed functional activity, with the therapist providing verbal prompting and feedback (therapist-independent). FST is based on the key elements of normal upper limb function (i.e., positioning the hand and then using it to manipulate objects) and is therapist-independent while maintaining participant safety. The focus is on improving the power of shoulder/elbow muscles to enable appropriate placing of the hand, improving production of appropriate force in arm and hand muscles to achieve a specific grasp, and specific resistive functional practice for wrist and finger muscles to maximize ability to manipulate objects. The initial level of load/resistance is the maximum load that still permits five repetitions of action through the available range of muscle length. Treatment is progressed systematically using repetition and increased resistance to movement by changing the limb’s relationship to gravity, changing the amount of friction to overcome, and increasing the size and weight of items (e.g., following an empty cup with a cup containing an increased volume of water). FST involves specific movements for muscle groups, upper limb gross movement patterns underlying functional activity, hand reaching/retrieval activities, hand grip activities, hand manipulation involving entire everyday movements, and using objects such as screw-top canisters, pegs, mugs, and pens. These movements are extended into more complex everyday activities such as placing different food items into a shopping bag, then lifting the bag onto a shelf; tightening/loosening nuts/bolts; opening a bottle and drinking from it; and pouring tea from a pot.

g objects such as screw-top canisters, pegs, mugs, and pens. These movements are extended into more complex everyday activities such as placing different food items into a shopping bag, then lifting the bag onto a shelf; tightening/loosening nuts/bolts; opening a bottle and drinking from it; and pouring tea from a pot. Feedback and instructions encourage an external focus of attention on the effects of movements, for example, whether the teapot has been lifted clearly off the table. Outcomes In accordance with the intention-to-treat principle, every effort will be made to include all randomized participants at outcome and follow-up whether or not they discontinue randomized treatment before the end of the intervention phase. Measurements will be made and processed by assessors blinded to treatment allocation. To assess whether blinding was achieved, we will ask assessors, at outcome and follow-up points, to guess participants’ group allocation. Agreement with actual allocation will be assessed with Cohen’s kappa. Clinical efficacy measures will be made pre-randomization (baseline), the day (± 7) after the six-week intervention phase ends (outcome), and six calendar months (± 14 days) after the index stroke (follow-up). The primary outcome measure will be the score on the Action Research Arm Test (ARAT) 17, a measure of the primary focus of both interventions – improved upper limb function.

Clinical efficacy measures will be made pre-randomization (baseline), the day (± 7) after the six-week intervention phase ends (outcome), and six calendar months (± 14 days) after the index stroke (follow-up). The primary outcome measure will be the score on the Action Research Arm Test (ARAT) 17, a measure of the primary focus of both interventions – improved upper limb function. Secondary outcome measures will be (a) Wolf Motor Function Test (WMFT) score 18, (b) hand grip force, and (c) pinch grip force. The upper limb positions for both pinch grip force and hand grip force will be standardized and the myometer set to zero after the participant’s hand/digits are positioned around the bars, ‘at rest’. The instruction given will be ‘squeeze as hard as you can’. Force values will be obtained over three trials, with the greatest value obtained used for data analysis. Explanatory measures will be made at baseline and outcome. Full training will be given to all trial center teams, with comprehensive monitoring and supervision arrangements. Data from all sites will be subject to prompt, rigorous quality control prior to statistical processing. All participants will be provided with full explanations and opportunities to ask questions, plenty of time to be made comfortable, and plenty of time to practice the tasks. Structural brain imaging will be conducted in all patients in the same location during the same session.

Explanatory measures will be made at baseline and outcome. Full training will be given to all trial center teams, with comprehensive monitoring and supervision arrangements. Data from all sites will be subject to prompt, rigorous quality control prior to statistical processing. All participants will be provided with full explanations and opportunities to ask questions, plenty of time to be made comfortable, and plenty of time to practice the tasks. Structural brain imaging will be conducted in all patients in the same location during the same session. First, a T1-weighted, 1 × 1 × 1-mm whole-brain image will be obtained using structural magnetic resonance imaging (MRI). Automated normalization, segmentation, and lesion identification will be performed as previously described 19, using spm12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12). This approach has a high sensitivity for delineating brain lesions and identifying tissue classes, thereby being useful in atrophy and white matter disease. Outputs include normalized lesion maps as well as gray and white matter maps consisting of voxel-wise values representing gray or white matter density. Second, a map of the static magnetic field will be generated to allow for offline correction of image distortions.

First, a T1-weighted, 1 × 1 × 1-mm whole-brain image will be obtained using structural magnetic resonance imaging (MRI). Automated normalization, segmentation, and lesion identification will be performed as previously described 19, using spm12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12). This approach has a high sensitivity for delineating brain lesions and identifying tissue classes, thereby being useful in atrophy and white matter disease. Outputs include normalized lesion maps as well as gray and white matter maps consisting of voxel-wise values representing gray or white matter density. Second, a map of the static magnetic field will be generated to allow for offline correction of image distortions. Third, diffusion tensor imaging (DTI) data will be acquired at 2 × 2 × 2 mm. A diffusion tensor model will be fitted to data at each voxel to allow for maps of diffusion parameters (fractional anisotropy, mean diffusivity, and eigenvalues) to be generated for each participant at each scanning session. Probabilistic tractography (http://www.fmrib.ox.ac.uk/fsl) will be used to generate pathways of interest across the group 19, in particular the corticospinal tract. Fourth, dual-echo T2-weighted and proton-density whole-brain MRI scans will be acquired at 1 × 1 × 3 mm resolution to allow for accurate delimitation of stroke volumes and for detection of other pathological changes, such as white matter hyperintensities.

Third, diffusion tensor imaging (DTI) data will be acquired at 2 × 2 × 2 mm. A diffusion tensor model will be fitted to data at each voxel to allow for maps of diffusion parameters (fractional anisotropy, mean diffusivity, and eigenvalues) to be generated for each participant at each scanning session. Probabilistic tractography (http://www.fmrib.ox.ac.uk/fsl) will be used to generate pathways of interest across the group 19, in particular the corticospinal tract. Fourth, dual-echo T2-weighted and proton-density whole-brain MRI scans will be acquired at 1 × 1 × 3 mm resolution to allow for accurate delimitation of stroke volumes and for detection of other pathological changes, such as white matter hyperintensities. A key purpose of FST is to improve the production of appropriate force in different muscles to enhance grasping and manipulation of objects by the paretic hand. We will therefore use a grip force task for functional MRI (fMRI), which is able to detect changes in brain activity corresponding to a force as low as 0.01N. As all participants will be able to produce the beginnings of prehension, they will be able to perform the fMRI grip task. Participants will be scanned while performing isometric handgrips of between 20% and 50% of paretic hand maximum grip force as measured just prior to scanning. Participants will be trained in how to perform the task prior to scanning. The experiment will be conducted according to an event-related design (auditory cues given at interstimulus interval of 7 ± 2 seconds) with actual exerted force recorded 20.

of paretic hand maximum grip force as measured just prior to scanning. Participants will be trained in how to perform the task prior to scanning. The experiment will be conducted according to an event-related design (auditory cues given at interstimulus interval of 7 ± 2 seconds) with actual exerted force recorded 20. Analysis of fMRI data will follow standard approaches using spm12. Single-participant results will include voxel-wise values for (a) magnitude of brain activity during handgrips and (b) how much brain activity is modulated by handgrip force (20%, 50%). These values are independent of one another and provide complementary information on how the brain is working to generate motor output. In addition, we will use dynamic causal modeling (DCM) 21 to measure (a) connectivity between brain regions (coupling parameters) during grip and (b) changes in connectivity between brain regions (bilinear parameters) with increasing grip force. Single pulses of transcranial magnetic stimulation (TMS), using a standard figure-of-eight coil, will be given over the hand and arm areas of the primary motor cortex of the lesioned hemisphere and, when possible, the nonlesioned hemisphere. Electromyographic recordings of motor evoked potentials (MEPs) over both contralateral and ipsilateral extensor carpi radialis and biceps brachii muscles will allow for characterization of recruitment curves, which will be constructed by measuring the amplitude of the MEP at between 100% and 130% (where possible) of active motor threshold 20,22.

s of motor evoked potentials (MEPs) over both contralateral and ipsilateral extensor carpi radialis and biceps brachii muscles will allow for characterization of recruitment curves, which will be constructed by measuring the amplitude of the MEP at between 100% and 130% (where possible) of active motor threshold 20,22. Economic evaluation Costs and effects will be monitored at baseline and outcome in order to inform the design of a future Phase III study. A preliminary estimate of cost-effectiveness will also be made. Resource items to be monitored will include input by the research therapist, length of stay in the original admission and any subsequent readmission, and other health-care and non-health-care contacts. Levels of informal care will also be monitored. The EuroQoL-5D 23 will be the main measure of effect, as it can be used to estimate the quality-adjusted life year gain. Assessment of safety Comprehensive details of the trial safety monitoring process can be found in the operational trial protocol (http://www.fastindicate.com). In addition to events that may be expected to occur in a trial of this type, pain and fatigue will be particularly monitored, as there is a possibility that either MPT or FST, carried out in addition to CPT, could be associated with overuse syndrome, expressed by experience of pain or fatigue. We will follow the Norwich Clinical Trials Unit system of documenting and reporting serious adverse events (http://www.nnuh.nhs.uk/Dept.asp?ID=681). Adverse events will be recorded from randomization to end of trial.

addition to CPT, could be associated with overuse syndrome, expressed by experience of pain or fatigue. We will follow the Norwich Clinical Trials Unit system of documenting and reporting serious adverse events (http://www.nnuh.nhs.uk/Dept.asp?ID=681). Adverse events will be recorded from randomization to end of trial. Data monitoring body Ethical approval has been provided by the National Research Ethics Service. The University of East Anglia is the trial sponsor. All necessary local R&D governance approval will be obtained for each center in advance of recruitment. This trial is registered with Current Controlled Trials (ISRCTN19090862) and the UK Clinical Research Network Study Portfolio (UKCRN ID 12967). Independent trial oversight will be provided by a trial steering committee and a data monitoring and ethics committee set up in adherence to the UK Medical Research Council Guidelines for Good Clinical Practice in Clinical Trials.

s (ISRCTN19090862) and the UK Clinical Research Network Study Portfolio (UKCRN ID 12967). Independent trial oversight will be provided by a trial steering committee and a data monitoring and ethics committee set up in adherence to the UK Medical Research Council Guidelines for Good Clinical Practice in Clinical Trials. Sample size The clustered data structure (patients grouped according to therapist within each treatment group) is accounted for in the design and analysis 24. This sample size calculation is based on actual ARAT data from our previous early-phase trial 9. Assuming an intraclass correlation coefficient of 0·01 in both treatment arms and three centers with a separate therapist for each randomized arm, a sample size of 99 participants per group would have 80% power to detect a clinically important mean difference of 6·2 in ARAT change when data were analyzed using a two-sample t-test with Satterthwaite correction, applying a 5% two-sided significance level and allowing for potentially different standard deviations in the CPT + MPT (7·9) and CPT + FST (19·3) groups. To account for clustering in the design (participants grouped according to therapist within each randomized treatment at each study site), a sample size inflation factor, 1 + (m-1)*ICC, is applied, where m is the cluster size and ICC is the intraclass correlation coefficient. Assuming that recruitment is evenly distributed across therapists, sample size is therefore inflated to 129 evaluable participants per group. The corresponding mean differences in ARAT change that would be detectable in a study of this size for ICCs of 0·02 and 0·03 would be 7·0 and 7·8 respectively, showing that the design is fairly insensitive to assumptions about the ICC. Finally, to allow for an attrition rate of 10% [7% in our previous single-center trial 9], 144 participants per group will be recruited for a total sample size of 288.

dy of this size for ICCs of 0·02 and 0·03 would be 7·0 and 7·8 respectively, showing that the design is fairly insensitive to assumptions about the ICC. Finally, to allow for an attrition rate of 10% [7% in our previous single-center trial 9], 144 participants per group will be recruited for a total sample size of 288. For the explanatory measurements, our experience indicates that of those meeting the criteria for this proposed trial, at least 70% will consent to participate, and 90% of these will complete the measurements. Thus, we anticipate 181 sets of explanatory measurements. Statistical analysis Data for all participants will be analyzed according to participants’ allocated group. A single formal analysis will take place at the end of the study. Interim data summaries will be made available to the independent data monitoring and ethics committee. With regard to missing data, we shall aim to do the following 25: (1) implement strategies to limit the amount of missing data; (2) develop an understanding of the causes for data being missing and (3), based on this, decide what the main assumption about the mechanism of missing data is; (4) conduct a statistical analysis which accounts for this assumption; and (5) perform sensitivity analyses to assess whether the conclusions depend strongly on the validity of the main assumption about the mechanism of missingness.

based on this, decide what the main assumption about the mechanism of missing data is; (4) conduct a statistical analysis which accounts for this assumption; and (5) perform sensitivity analyses to assess whether the conclusions depend strongly on the validity of the main assumption about the mechanism of missingness. Clinical efficacy Continuous outcome variables will be compared between treatment groups using a multilevel normal linear model. Change from baseline to outcomes (day 43; Fig. 1) will be modeled, adjusting for baseline value, time after stroke, and 9HPT score as patient-level covariates. Which therapist administered the treatment will be included in the model as a zero-mean random effect. We will test, by comparing the log-likelihood, whether a separate random-effect variance is required for therapists delivering each treatment arm or whether a pooled variance is sufficient. The treatment effect will be summarized using the adjusted mean difference and 95% confidence interval. Secondary analyses will include sensitivity to incomplete follow-up, descriptive analysis at each time-point, statistical modeling of follow-up measures at six-months, and a per-protocol analysis. For safety analysis, the number and percentage of participants experiencing each prespecified category of adverse event will be summarized by treatment group.

sitivity to incomplete follow-up, descriptive analysis at each time-point, statistical modeling of follow-up measures at six-months, and a per-protocol analysis. For safety analysis, the number and percentage of participants experiencing each prespecified category of adverse event will be summarized by treatment group. Neural correlates of response to CPT + FST and CPT + MPT Our main intention with these analyses is to infer differences between correlations between neural explanatory measurements and CPT + FST and those between such measurements and CPT + MPT (objective 2). Associations will be investigated between change from baseline in clinical outcomes and change from baseline in each of the explanatory measurements. We will make every possible effort to record and adjust for potential confounding baseline variables, such as baseline motor score. The relationship will be explored further via multilevel linear regression, which will include which therapist administered the treatment as a random effect., We will assess whether changes in neuroimaging and/or neurophysiological measures are strongly associated with clinical improvements in the individual participant and, in particular, whether such associations differ between the CPT + FST and CPT + MPT groups. We acknowledge the potential value of structural mean models (SMM)/causal inference (CI) and are aware of the additional complexity due to clustering in the design. Therefore, we will investigate an extension of SMM/CI as a potential exploratory analysis 26.

sociations differ between the CPT + FST and CPT + MPT groups. We acknowledge the potential value of structural mean models (SMM)/causal inference (CI) and are aware of the additional complexity due to clustering in the design. Therefore, we will investigate an extension of SMM/CI as a potential exploratory analysis 26. The relationship will be explored further via multilevel linear regression which will include therapist as a random effect. We will assess whether changes in neuroimaging and/or neurophysiological measures are strongly associated with clinical improvements in the individual participant and in particular whether such associations differ between the CPT+FST and CPT+MPT groups.

linear regression which will include therapist as a random effect. We will assess whether changes in neuroimaging and/or neurophysiological measures are strongly associated with clinical improvements in the individual participant and in particular whether such associations differ between the CPT+FST and CPT+MPT groups. Indicators of response to CPT + FST and CPT + MPT Baseline measurements considered will be (a) TMS- and DTI-based measurements related to corticospinal system integrity, (b) normalized lesion maps, (c) voxel-wise measurements of gray and white matter density, (d) voxel-wise measurements of brain activity during hand grip and its modulation by changing force, and (e) clinical variables, including motor scores. An interaction term between treatment group and each variable from (a) to (e) in turn will be added to the normal linear model for ARAT used in the clinical efficacy analysis. Continuous baseline variables will be categorized as high or low, the cutpoint being at the median of the observed data. We will adjust for the time after stroke category. Statistical significance of the interaction term will be assessed and treatment effect calculated within each of the high and low sub-groups of the interaction variable.

ariables will be categorized as high or low, the cutpoint being at the median of the observed data. We will adjust for the time after stroke category. Statistical significance of the interaction term will be assessed and treatment effect calculated within each of the high and low sub-groups of the interaction variable. Further analysis will develop a multiple regression model within each treatment group to predict change in ARAT clinical outcome using baseline measurements. This will determine the sub-set of baseline variables independently associated with treatment response and will allow for a different group of baseline predictors within each treatment group. Principal components analysis will be used to reduce dimensionality of the predictor variables while retaining a meaningful interpretation of the principal components. Trial funding FAST INdiCATE is funded by the Efficacy and Mechanism Evaluation programme (http://www.eme.ac.uk). EME reference: 10/60/30 Summary This trial aims to determine clinical efficacy of CPT + FST and CPT + MPT for enhancement of upper limb motor function early after stroke, neuro-biomechanical correlates associated with clinical improvement, and which stroke survivors may be most likely to respond to FST and which to MPT. The results are expected to inform a subsequent definitive clinical trial and also to lead to advances in knowledge of how the upper limb recovers after stroke in response to well-characterized interventions.

ed with clinical improvement, and which stroke survivors may be most likely to respond to FST and which to MPT. The results are expected to inform a subsequent definitive clinical trial and also to lead to advances in knowledge of how the upper limb recovers after stroke in response to well-characterized interventions. This report is independent research funded by the Medical Research Council (MRC) and managed by the National Institute for Health Research (NIHR) on behalf of the MRC-NIHR partnership. The views expressed in this paper are those of the authors and not necessarily those of the MRC, NHS, NIHR, or Department of Health. CJW is supported in this research by the Edinburgh MRC Hub for Trials Methodology Research and by NHS Lothian through the Health Services Research Unit Edinburgh. We are also grateful for the assistance and support provided by clinicians working in the clinical centers for this trial (Norfolk, Birmingham, and North Staffordshire), everyone working in the associated clinical brain imaging units, staff in the Clinical Trials Units in Glasgow and Norwich, and the UK Stroke Research Network.

Introduction Stroke is the second leading cause of death and a leading cause of disability worldwide (1). Cerebral infarction, due to thrombotic occlusion of a brain artery, is the most common stroke type, accounting for 65–85% of all cases. The only specific therapy of demonstrated benefit for acute ischemic stroke (AIS) is intravenous (IV) fibrinolysis with tissue plasminogen activator (tPA) up to 4·5 hours after onset. However, patients with occlusions of large, proximal, intracranial arteries are often not responsive to IV tPA, as lytic therapy achieves early reperfusion in only 13–50% of patients with occlusions in the carotid terminus and the M1 segment of the middle cerebral artery (MCA) (2–5). The Solitaire™ Flow Restoration (FR) device is a self-expanding stent retriever that restores blood flow in patients experiencing ischemic stroke because of large intracranial vessel occlusion. In multicenter registries and large clinical series, the Solitaire stent retriever has yielded high rates of reperfusion and favorable clinical outcomes (6–8). In a randomized, head-to-head device trial, compared with first-generation, coil retriever devices, use of the Solitaire™ FR was associated with superior recanalization rates, faster achievement of reperfusion, reduced intracranial haemorrhage complications, and improved final disability outcome (9).

al outcomes (6–8). In a randomized, head-to-head device trial, compared with first-generation, coil retriever devices, use of the Solitaire™ FR was associated with superior recanalization rates, faster achievement of reperfusion, reduced intracranial haemorrhage complications, and improved final disability outcome (9). The Solitaire™ with the Intention for Thrombectomy as Primary Endovascular Treatment for Acute Ischemic Stroke (SWIFT PRIME) trial is being undertaken to establish the safety and efficacy of neurothrombectomy with Solitaire in conjunction with IV tPA vs. IV tPA alone, among AIS patients treatable within six-hours of symptom onset. Methods Objective The aim of this study is to determine whether subjects experiencing an AIS due to large vessel occlusion treated with combined IV tPA and Solitaire revascularization device within six-hours of symptom onset have less stroke-related disability than those subjects treated with IV tPA alone. Design The study is a global, multicenter, two-arm, prospective, randomized, open, blinded end-point clinical trial. The study patient flow outline is shown in Fig. 1. Figure 1 Study design diagram. ASPECTS, Alberta Stroke Program Early CT score; CTA, computed tomography angiography; CTP, computed tomography perfusion imaging; DWI, diffusion-weighted imaging; FLAIR, fluid attenuated inversion recovery; FR, Flow Restoration; GRE, gradient refocused echo; IV, intravenous; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NCCT, non-contrast CT; PWI, perfusion weighted imaging; tPA, tissue plasminogen activator.

sion imaging; DWI, diffusion-weighted imaging; FLAIR, fluid attenuated inversion recovery; FR, Flow Restoration; GRE, gradient refocused echo; IV, intravenous; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NCCT, non-contrast CT; PWI, perfusion weighted imaging; tPA, tissue plasminogen activator. Patient population Entry criteria were structured to enroll patients who have AIS, moderate to severe neurologic deficits, harbor imaging-confirmed occlusions of proximal, anterior circulation arteries, do not have a large, established core infarct, have received treatment with IV tPA, and are able to undergo Solitaire neurothrombectomy within six-hours of last known well time. Full, detailed study inclusion and exclusion criteria are shown in Table 1. Table 1 Study inclusion and exclusion criteria

Patient population Entry criteria were structured to enroll patients who have AIS, moderate to severe neurologic deficits, harbor imaging-confirmed occlusions of proximal, anterior circulation arteries, do not have a large, established core infarct, have received treatment with IV tPA, and are able to undergo Solitaire neurothrombectomy within six-hours of last known well time. Full, detailed study inclusion and exclusion criteria are shown in Table 1. Table 1 Study inclusion and exclusion criteria Inclusion criteria: 1. Age 18 – 80 2. Clinical signs consistent with acute ischemic stroke 3. Prestroke Modified Rankin Score ≤ 1 4. NIHSS ≥ 8 and < 30 at the time of randomization 5. Initiation of IV tPA within 4·5 hours of onset of stroke symptoms (onset time is defined as the last time when the patient was witnessed to be at baseline), with investigator verification that the subject has received/is receiving the correct IV tPA dose for the estimated weight prior to randomization. 6. Thrombolysis in cerebral infarction (TICI) 0–1 flow in the intracranial internal carotid, M1 segment of the MCA, or carotid terminus confirmed by CT or MR angiography that is accessible to the Solitaire™ FR device. (Note: M1 segment of the MCA is defined as the arterial trunk from its origin at the ICA to the first bifurcation or trifurcation into major branches neglecting the small temporo-polar branch.) 7. Subject is able to be treated within six-hours of onset of stroke symptoms and within 1·5 hours (90 min) from CTA or MRA to groin puncture. 8. Subject is willing to conduct protocol-required follow-up visits. 9. An appropriate signed and dated informed consent form (or enrollment under exception from explicit informed consent if permitted under country regulations) 10. Subject is affiliated with a social security system (if required by individual country regulations). 11. Subject meets national regulatory criteria for clinical trial participation. Exclusion criteria: 1. Subject who is contraindicated to IV tPA as per local national guidelines. 2. Female who is pregnant or lactating or has a positive pregnancy test at time of admission. 3. As applicable by French law, subject who is a protected individual such as an incompetent adult or incarcerated person. 4. Rapid neurological improvement prior to study randomization suggesting resolution of signs/symptoms of stroke 5. Known serious sensitivity to radiographic contrast agents 6. Known sensitivity to nickel, titanium metals, or their alloys 7. Current participation in another investigation drug or device treatment study 8. Known hereditary or acquired haemorrhagic diathesis, coagulation factor deficiency.

f signs/symptoms of stroke 5. Known serious sensitivity to radiographic contrast agents 6. Known sensitivity to nickel, titanium metals, or their alloys 7. Current participation in another investigation drug or device treatment study 8. Known hereditary or acquired haemorrhagic diathesis, coagulation factor deficiency. (A subject without history or suspicion of coagulopathy does not require INR or prothrombin time lab results to be available prior to enrollment.) 9. Renal failure as defined by a serum creatinine > 2·0 mg/dl (or 176·8 μmol/l) or glomerular filtration rate (GFR) < 30. 10. Subject who requires hemodialysis or peritoneal dialysis, or who has a contraindication to an angiogram for whatever reason. 11. Life expectancy of less than 90 days 12. Clinical presentation suggests a subarachnoid haemorrhage, even if initial CT or MRI scan is normal 13. Suspicion of aortic dissection 14. Subject with a comorbid disease or condition that would confound the neurological and functional evaluations or compromise survival or ability to complete follow-up assessments. 15. Subject currently uses or has a recent history of illicit drug(s) or abuses alcohol (defined as regular or daily consumption of more than four alcoholic drinks per day). 16. Known history of arterial tortuosity, preexisting stent, and/or other arterial disease that would prevent the device from reaching the target vessel and/or preclude safe recovery of the device Imaging exclusion criteria 1. CT or MRI evidence of haemorrhage on presentation 2. CT or MRI evidence of mass effect or intra-cranial tumour (except small meningioma) 3. CT or MRI evidence of cerebral vasculitis 4. CT showing hypodensity or MRI showing hyperintensity involving greater than 1/3 of the MCA territory (or in other territories, > 100 cc of tissue) on presentation 5. *Baseline non-contrast CT or DWI MRI evidence of a moderate/large core defined as extensive early ischemic changes of Alberta Stroke Program Early CT score (ASPECTS) < 6. 6. CT or MRI evidence of a basilar artery (BA) occlusion or posterior cerebral artery (PCA) occlusion 7. CTA or MRA evidence of carotid dissection or complete cervical carotid occlusion requiring stenting at the time of the index procedure (i.e. mechanical thrombectomy) 8. Imaging evidence that suggests, in the opinion of the investigator, the subject is not appropriate for mechanical thrombectomy intervention (e.g.

on 7. CTA or MRA evidence of carotid dissection or complete cervical carotid occlusion requiring stenting at the time of the index procedure (i.e. mechanical thrombectomy) 8. Imaging evidence that suggests, in the opinion of the investigator, the subject is not appropriate for mechanical thrombectomy intervention (e.g. inability to navigate to target lesion, moderate/large infarct with poor collateral circulation, etc.). * Before Revision F, this criterion stated: ‘Core Infarct and hypoperfusion: a) MRI- or CT-assessed core infarct lesion greater than 50 cc; b) Severe hypoperfusion lesion (10 sec or more Tmax lesion larger than 100 cc); c) Ischemic penumbra < 15 cc and mismatch ratio ≤1·8’. CT, computed tomography; CTA, computed tomography angiography; DWI, diffusion-weighted imaging; ICA, internal carotid artery; INR, international normalized ratio; IV, intravenous; MCA, middle cerebral artery; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator.

raphy angiography; DWI, diffusion-weighted imaging; ICA, internal carotid artery; INR, international normalized ratio; IV, intravenous; MCA, middle cerebral artery; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator. During the course of the trial, a revision in imaging entry criteria was made. At launch, the study employed a ‘target mismatch’ strategy, using multimodal computed tomography (CT) or magnetic resonance imaging (MRI), including perfusion sequences, to identify patients with salvageable tissue (10). Subsequently, to accommodate sites with limited perfusion imaging capability and ensure accelerated treatment delivery (11), Revision F of the study changed imaging entry criteria to employ a ‘small to moderate core’ strategy, using Alberta Stroke Program Early CT score (ASPECTS) ratings of CT or magnetic resonance (MR) images (12).

to accommodate sites with limited perfusion imaging capability and ensure accelerated treatment delivery (11), Revision F of the study changed imaging entry criteria to employ a ‘small to moderate core’ strategy, using Alberta Stroke Program Early CT score (ASPECTS) ratings of CT or magnetic resonance (MR) images (12). At sites in which standard imaging practice includes CT or MR angiography (MRA) (and, under the initial study entry criteria, perfusion imaging), patients may directly be enrolled in the randomized phase of the trial. At sites where this imaging is not standard practice, subjects are first enrolled in a screening phase and the imaging studies are obtained. A subject is considered enrolled into the screening phase of the study after the informed consent form has been signed or country-specific requirements have been met for enrollment without explicit informed consent in emergency circumstances. A subject is considered enrolled in the randomized phase of the study after a treatment allocation is assigned from the randomization system.

study after the informed consent form has been signed or country-specific requirements have been met for enrollment without explicit informed consent in emergency circumstances. A subject is considered enrolled in the randomized phase of the study after a treatment allocation is assigned from the randomization system. Randomization Subjects will be randomly assigned in a 1 : 1 fashion to one of two treatment arms: (1) IV tPA and Solitaire device or (2) IV tPA alone. The number of treatments and controls will be balanced within investigational sites and by baseline National Institutes of Health Stroke Scale (NIHSS) severity (≤17 vs. >17), age (<70 years vs. ≥70 years at the time of randomization), and occlusion location (M1 vs. all other). Subject allocation to treatment will be accomplished by using an interactive web response or interactive voice response system. Intervention Endovascular treatment The Solitaire device revascularization procedure is summarized in Fig. 2. Among endovascular arm patients, to facilitate speed of intervention, two target procedural time targets are employed. Time from acquisition of the final study-qualifying image (perfusion CT or MR sequence prior to Revision F; CT or MRA sequence under Revision F) to groin puncture is optimally targeted to be less than 70 min and should be no greater than 90 min. Time from groin puncture to first deployment of the Solitaire device is targeted to be less than 20 min.

inal study-qualifying image (perfusion CT or MR sequence prior to Revision F; CT or MRA sequence under Revision F) to groin puncture is optimally targeted to be less than 70 min and should be no greater than 90 min. Time from groin puncture to first deployment of the Solitaire device is targeted to be less than 20 min. Figure 2 Solitaire revascularization device procedure. FR, Flow Restoration; FU, follow-up; IFU, instructions for use; TICI, thrombolysis in cerebral infarction. For the intervention, the proceduralist may use the Solitaire™ FR device or the Solitaire™ 2 revascularization device. The proper study device size is selected per device-specific instructions for use. If deemed appropriate by the neurointerventionalist, IV sedation or general anesthesia may be administered to assure subject comfort and safety. Several retrieval passes with Solitaire revascularization devices, if needed, may be performed, up to: (1) a maximum of three retrievals in the same vessel and (2) a maximum of two retrievals per device. Procedural angiography Among subjects randomized to IV tPA + Solitaire device group, angiography images obtained during the procedure will consist of the following: (1) Baseline angiogram: obtained prior to device deployment while assessing the clot location; (2) Post-device use angiogram: obtained immediately after each pass of study device use; and (3) Final post-procedural angiogram: obtained after all treatments have been completed.

the procedure will consist of the following: (1) Baseline angiogram: obtained prior to device deployment while assessing the clot location; (2) Post-device use angiogram: obtained immediately after each pass of study device use; and (3) Final post-procedural angiogram: obtained after all treatments have been completed. CT or MRI At screening, non-contrast CT (NCCT) or gradient refocused echo MRI are used to exclude haemorrhage, NCCT or diffusion-weighted imaging (DWI) MRI to identify early ischemic changes using the ASPECTS scale, and CT angiography (CTA) or MRA to assess for target vascular occlusion. In patients transferred from another facility, updated screening imaging at the study hospital must be obtained to qualify the patient for the trial. At all sites prior to Revision F and with Revision F at sites where perfusion CT or MR imaging is local standard of care, perfusion images will also be obtained at screening and will be processed locally using RApid processing of PerfusIon and Diffusion (RAPID), an operator-independent system for processing of perfusion weighted imaging (PWI) and DWI images (13). RAPID will generate PWI and DWI maps, segment the PWI and DWI lesions, and calculate lesion volumes within 10 min of scan completion.

ill be processed locally using RApid processing of PerfusIon and Diffusion (RAPID), an operator-independent system for processing of perfusion weighted imaging (PWI) and DWI images (13). RAPID will generate PWI and DWI maps, segment the PWI and DWI lesions, and calculate lesion volumes within 10 min of scan completion. Follow-up multimodal CT or MRI will be obtained among all study patients at 27 h (±6 h) from time of randomization to assess any presence of haemorrhage, recanalization of the occluded artery, reperfusion of the ischemic region, and infarct growth. At all sites prior to Revision F and with Revision F at sites where perfusion CT or MR imaging is local standard of care, perfusion images will also be acquired at 27 h. Follow-up and assessment of clinical outcomes Table 2 shows the schedule of study visits. A 27-h post-randomization visit includes an NIHSS examination and CT or MR imaging. Subsequent follow-up visits occur at 7–10 days (or discharge if earlier), 30 days, and 90 days, and include the modified Rankin Scale (mRS) assessing global disability, the Barthel Index assessing instrumental activities of daily living, the NIHSS exam assessing neurologic deficit, and the European Quality of Life Scale (EuroQoL) assessing health-related quality of life. In addition, to assess health economics outcomes, a Resource Utilization Questionnaire will collect information on use of healthcare resources through 90 days, including the index hospitalization, subsequent inpatient and outpatient care, and readmissions. Table 2 Study visits

Follow-up and assessment of clinical outcomes Table 2 shows the schedule of study visits. A 27-h post-randomization visit includes an NIHSS examination and CT or MR imaging. Subsequent follow-up visits occur at 7–10 days (or discharge if earlier), 30 days, and 90 days, and include the modified Rankin Scale (mRS) assessing global disability, the Barthel Index assessing instrumental activities of daily living, the NIHSS exam assessing neurologic deficit, and the European Quality of Life Scale (EuroQoL) assessing health-related quality of life. In addition, to assess health economics outcomes, a Resource Utilization Questionnaire will collect information on use of healthcare resources through 90 days, including the index hospitalization, subsequent inpatient and outpatient care, and readmissions. Table 2 Study visits Assessment method Screening Procedure 27 h post-randomization 7–10 days or discharge 30-day follow-up 90-day follow-up Unscheduled follow-up Range 0 0 ± 6 h 0 ± 7 days ± 15 days N/A Pregnancy test w – – – – – – Blood labs w – – – – – – NIH Stroke Scale w – w w w w w Prestroke modified Rankin Scale w – – – – – – Modified Rankin Scale – – – w w w w Barthel Index – – – w – w w MRI or NCCT w – w – – – – ASPECTS w – – – – – – MRA or CTA w – w – – – – Catheter angiography – w – – – – – Resource utilization – – – w w w w EuroQol 5D-5L – – – – w w – Concomitant medications w w w w w w w Adverse events w ww w w w w w Optional: Applicable only to sites with DWI/PWI or CTP imaging as local standard of care at initial evaluation: PWI MR or CT perfusion w – w – – – – RAPID imaging processing w – w – – – – ASPECTS, Alberta Stroke Program Early CT score; CT, computed tomography; CTA, computed tomography angiography; EuroQoL, European Quality of Life Scale; MR, magnetic resonance; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NCCT, non-contrast computed tomography; NIH, National Institutes of Health; RAPID, RApid processing of PerfusIon and Diffusion.

omputed tomography; CTA, computed tomography angiography; EuroQoL, European Quality of Life Scale; MR, magnetic resonance; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; NCCT, non-contrast computed tomography; NIH, National Institutes of Health; RAPID, RApid processing of PerfusIon and Diffusion. Adverse event categorization All adverse events will be validated and categorized for severity and relatedness by a clinical events committee, comprised of three expert physicians independent of the investigational sites. Relatedness categories will include: (1) study disease-related: event clearly attributable to underlying disease state with no temporal relationship to the device, treatment, or medication; (2) concomitant disease-related: event attributable to disease other than the study disease with no temporal relationship to the device, treatment, or medication; (3) IV tPA-related: event clearly attributable to IV tPA medication with no temporal relationship to the device or treatment; (4) procedure-related: event has strong temporal relationship to the procedure or treatment of the device implantation or any user handling; (5) primary study device-related: event has a strong temporal relationship to the Solitaire device and alternative etiology is less likely; (6) ancillary device-related: any device other than Solitaire™ FR, such as microcatheter or guidewire; (7) device unknown: device related but unable to attribute a specific device; (8) other; and (9) unknown.

ated: event has a strong temporal relationship to the Solitaire device and alternative etiology is less likely; (6) ancillary device-related: any device other than Solitaire™ FR, such as microcatheter or guidewire; (7) device unknown: device related but unable to attribute a specific device; (8) other; and (9) unknown. Imaging core laboratory The Core Imaging Laboratory will provide an independent and unbiased assessment of imaging-related entry criteria and end-points. Entry criteria assessed by the Core Lab on initial CT or MRI will include presence of small core using ASPECTS score, presence of target penumbral pattern, and presence of proximal vessel occlusion in the internal carotid artery (ICA) or M1 MCA. Among endovascular arm patients, the Core Lab will assess the catheter angiograms for vascular reperfusion following device use, measured by the thrombolysis in cerebral infarction (TICI) scale. On 27-h CT or MRI scans, the Core Lab will assess outcome ASPECTS score, infarct volume, recanalization defined by CTA or MRA TICI grades, and presence of haemorrhagic transformation. In patients undergoing perfusion imaging at 27 h, the Core Lab will assess infarct volume and reperfusion ratio.

nfarction (TICI) scale. On 27-h CT or MRI scans, the Core Lab will assess outcome ASPECTS score, infarct volume, recanalization defined by CTA or MRA TICI grades, and presence of haemorrhagic transformation. In patients undergoing perfusion imaging at 27 h, the Core Lab will assess infarct volume and reperfusion ratio. Haemorrhages on CT/MRI will be radiologically classified according to the following categories: HT 1 – small petechiae within ischemic field without mass effect; HT 2 – confluent petechiae within ischemic field without mass effect; PH1 – hematoma within ischemic field with some mild space-occupying effect but involving ≤ 30% of the infarcted area; PH2 – hematoma within ischemic field with space-occupying effect involving > 30% of the infarcted area; RIH – any intraparenchymal haemorrhage remote from the ischemic field; IVH – intraventricular haemorrhage; and SAH – subarachnoid haemorrhage. All Core Lab readings will be performed independently by two experienced readers. For key measures, discrepancies will be resolved by a third, independent reader. The Core Lab will assess all CT and MR imaging blinded to treatment assignment. Assessment of catheter angiographic images for revascularization will not be blinded, as this evaluation will only be done for subjects in the IV tPA plus Solitaire FR treatment arm.

epancies will be resolved by a third, independent reader. The Core Lab will assess all CT and MR imaging blinded to treatment assignment. Assessment of catheter angiographic images for revascularization will not be blinded, as this evaluation will only be done for subjects in the IV tPA plus Solitaire FR treatment arm. Primary outcome The primary study end-point is the degree of disability or dependence at 90 days as assessed by the mRS. A global measure of disability, the mRS comprises of seven grades ranging from 0 (no symptoms) to 6 (death). The mRS will be assessed in a formally operationalized manner by use of the Rankin Focused Assessment – Ambulation (RFA-A). The 90-day mRS will be assessed by study personnel certified in the scoring of the mRS using the RFA-A and will be blinded to treatment assignment. Secondary outcomes The study has three secondary clinical efficacy end-points: (1) death due to any cause at 90 days; (2) functional independence as defined by mRS score ≤ 2 at 90 days; and (3) change in NIHSS score at 27 ± 6 h post-randomization. The study has four technical efficacy end-points: (1) volume of cerebral infarction as measured by a CT or MRI scan at 27 ± 6 h post-randomization; (2) reperfusion measured by reperfusion ratio on CT or MRI scan 27 ± 6 h post-randomization; (3) arterial revascularization measured by TICI 2b or 3 following device use; and (4) correlation of RAPID-assessed core infarct volume with 27 ± 6 h post-randomization stroke infarction in subjects who achieved TICI 2b–3 reperfusion without intracranial haemorrhage.

ratio on CT or MRI scan 27 ± 6 h post-randomization; (3) arterial revascularization measured by TICI 2b or 3 following device use; and (4) correlation of RAPID-assessed core infarct volume with 27 ± 6 h post-randomization stroke infarction in subjects who achieved TICI 2b–3 reperfusion without intracranial haemorrhage. Study safety end-points are: (1) all serious adverse events and (2) symptomatic intracranial haemorrhage (SICH) at 27 ± 6 h post-randomization. SICH is defined as any PH1, PH2, RIH, SAH, or IVH associated with a four-point or more worsening on the NIHSS within 24 h. Health economic evaluation end-points will include features of care for the index stroke (length of stay, discharge to home or any other type of facility and cost of that care, rehabilitation services, home health services), readmissions due to subsequent stroke, and calculation of quality-adjusted life year assessment, using utilities derived from the EuroQoL 5D-5L assessment. In conjunction with external data on long-term stroke outcomes, these data will be used to estimate the incremental cost-effectiveness of Solitaire treatment for the target population. Sample size The primary effectiveness end-point in this study is 90-day global disability assessed via the blinded evaluation of mRS, analyzed using simultaneous success criteria on: (1) the overall distribution of mRS (Rankin shift) and (2) the proportion of subjects achieving functional independence, defined as mRS of 0 to 2.

ze The primary effectiveness end-point in this study is 90-day global disability assessed via the blinded evaluation of mRS, analyzed using simultaneous success criteria on: (1) the overall distribution of mRS (Rankin shift) and (2) the proportion of subjects achieving functional independence, defined as mRS of 0 to 2. The statistical hypothesis on Rankin shift is that the distribution of mRS in subjects randomized to the IV tPA plus Solitaire will be more favorable than the distribution in the IV tPA only group. For this purpose, the entire distribution 0 to 6 of mRS values will be considered except that categories 5 and 6 are collapsed into a single group. Additionally, to measure benefit in terms of functional independence, a simultaneous requirement for success is that the difference in the proportion of patients with mRS 0–2 outcomes nominally meets a prespecified minimum dependent on the evaluable sample size at the time of the assessment.

lapsed into a single group. Additionally, to measure benefit in terms of functional independence, a simultaneous requirement for success is that the difference in the proportion of patients with mRS 0–2 outcomes nominally meets a prespecified minimum dependent on the evaluable sample size at the time of the assessment. Power and sample size are determined by the mRS shift analysis and are computed by assuming that the true proportions of subjects with various mRS outcomes at the 90-day follow-up visit are as presented in Table 3. The tPA-only outcome distributions are those observed with IV tPA use in the two NINDS (National Institute of Neurologic Disorders and Stroke) tPA trials, restricted to those subjects with baseline NIHSS of at least 8 but less than 30 to correspond with the current study's inclusion criteria. The Solitaire outcome distributions are based on data collected in heterogeneous settings including the use and non-use of IV tPA. Included in these rates is adjustment for an expected proportion of 10% of subjects randomized to the Solitaire group who will be unable for anatomical reasons to be treated with the randomized device; this cohort of subjects will remain in the Solitaire group per intent to treat but are assumed to have outcomes similar to those randomized to tPA only. Table 3 Hypothesized true outcomes for sample size calculations

Power and sample size are determined by the mRS shift analysis and are computed by assuming that the true proportions of subjects with various mRS outcomes at the 90-day follow-up visit are as presented in Table 3. The tPA-only outcome distributions are those observed with IV tPA use in the two NINDS (National Institute of Neurologic Disorders and Stroke) tPA trials, restricted to those subjects with baseline NIHSS of at least 8 but less than 30 to correspond with the current study's inclusion criteria. The Solitaire outcome distributions are based on data collected in heterogeneous settings including the use and non-use of IV tPA. Included in these rates is adjustment for an expected proportion of 10% of subjects randomized to the Solitaire group who will be unable for anatomical reasons to be treated with the randomized device; this cohort of subjects will remain in the Solitaire group per intent to treat but are assumed to have outcomes similar to those randomized to tPA only. Table 3 Hypothesized true outcomes for sample size calculations Randomized group mRS 0 mRS 1 mRS 2 mRS 3 mRS 4 mRS 5–6 Solitaire plus IV tPA 19·7% 18·2% 18·3% 9·4% 12·5% 21·9% IV tPA alone 11·0% 21·6% 8·1% 15·7% 14·4% 29·2% IV, intravenous; mRS, modified Rankin Scale; tPA, tissue plasminogen activator.

Power and sample size are determined by the mRS shift analysis and are computed by assuming that the true proportions of subjects with various mRS outcomes at the 90-day follow-up visit are as presented in Table 3. The tPA-only outcome distributions are those observed with IV tPA use in the two NINDS (National Institute of Neurologic Disorders and Stroke) tPA trials, restricted to those subjects with baseline NIHSS of at least 8 but less than 30 to correspond with the current study's inclusion criteria. The Solitaire outcome distributions are based on data collected in heterogeneous settings including the use and non-use of IV tPA. Included in these rates is adjustment for an expected proportion of 10% of subjects randomized to the Solitaire group who will be unable for anatomical reasons to be treated with the randomized device; this cohort of subjects will remain in the Solitaire group per intent to treat but are assumed to have outcomes similar to those randomized to tPA only. Table 3 Hypothesized true outcomes for sample size calculations Randomized group mRS 0 mRS 1 mRS 2 mRS 3 mRS 4 mRS 5–6 Solitaire plus IV tPA 19·7% 18·2% 18·3% 9·4% 12·5% 21·9% IV tPA alone 11·0% 21·6% 8·1% 15·7% 14·4% 29·2% IV, intravenous; mRS, modified Rankin Scale; tPA, tissue plasminogen activator. Sample size and power are computed incorporating a group sequential analysis plan with five interim analyses for efficacy, futility, and safety. Under this group sequential analysis plan, with a one-sided alpha level set at 0·025, 750 subjects with an evaluable primary endpoint provide 90% power for testing the study's primary effectiveness hypothesis; assuming attrition of 10% for the primary end-point, the total randomized sample size is up to 833 while the expected randomized sample size under the alternative hypothesis is approximately 477 (Table 4).

cts with an evaluable primary endpoint provide 90% power for testing the study's primary effectiveness hypothesis; assuming attrition of 10% for the primary end-point, the total randomized sample size is up to 833 while the expected randomized sample size under the alternative hypothesis is approximately 477 (Table 4). Table 4 Group sequential analysis Evaluable sample size Stopping for safety Stopping for efficacy Stopping for futility Two-sided alpha for mortality Two-sided alpha for Rankin shift Effect size Δ for mRS 0–2 Effect size φ for mRS mean value Effect size Δ for mRS 0–2 200 0·0036 0·0200 12·0% 0·00 0·0% 300 0·0058 0·0125 10·0% 0·00 0·0% 400 0·0094 0·0150 9·0% 0·10 n/a 500 0·0147 0·0150 8·0% 0·14 n/a 600 0·0203 0·0150 6·0% 0·14 n/a 750 (final) 0·0340 0·0350 5·0% n/a n/a mRS, modified Rankin Scale. Statistical analyses Statistical testing of the primary end-point will be conducted using the Cochran–Mantel–Haenszel test for the shift in Rankin scores, augmented by the requirement of functional independence cited above. Type I and Type II error will be computed via simulation and overall alpha will be controlled at a one-sided level of 0·025. The primary end-point analysis and the testing strategy based on the group sequential design will be conducted using the intention to treat (ITT) population. If the results are favorable, a second analysis of the primary end-point will be conducted using the Food and Drug Administration (FDA) evaluable cohort (those having received IV tPA within three-hours of stroke symptom onset) following a step-down approach.

l be conducted using the intention to treat (ITT) population. If the results are favorable, a second analysis of the primary end-point will be conducted using the Food and Drug Administration (FDA) evaluable cohort (those having received IV tPA within three-hours of stroke symptom onset) following a step-down approach. Group Sequential Analysis: Interim analyses will be performed when 200 subjects from the ITT population have provided evaluable primary effectiveness data and then after each subsequent 100 subjects, to a maximum of 750 subjects with evaluable data (i.e. 200, 300, 400, 500, 600, and 750). Table 4 provides the group sequential boundaries including minimum acceptable statistical criteria at each look (including the final analysis). Additionally, a safety stopping rule is defined under which the trial is halted for a substantial mortality difference in either possible direction at the various interim looks. This rule does not impact effectiveness findings.

cluding minimum acceptable statistical criteria at each look (including the final analysis). Additionally, a safety stopping rule is defined under which the trial is halted for a substantial mortality difference in either possible direction at the various interim looks. This rule does not impact effectiveness findings. Study organization and funding Study conduct is overseen by the executive and steering committees and the sponsor. The executive committee will be led by the global principal investigator (PI) and will include a global imaging and workflow PI, an EU (European Union) national PI, US and EU interventional PIs, and an interventional advisor. The steering committee will be comprised of recognized experts in the treatment of stroke and representatives from leading enrolling sites. The executive committee, assisted by the steering committee, will oversee all clinical trial activities including protocol development and protocol amendment during study conduct. Covidien is the study sponsor and source of funding, and will perform monitoring at each site to ensure protection of the rights of subjects, the safety of subjects, and the quality and integrity of the data collected and submitted according to FDA and EMA (European Medicines Agency) regulation.

t during study conduct. Covidien is the study sponsor and source of funding, and will perform monitoring at each site to ensure protection of the rights of subjects, the safety of subjects, and the quality and integrity of the data collected and submitted according to FDA and EMA (European Medicines Agency) regulation. Data Safety Monitoring Board The Data Safety Monitoring Board will be comprised of individuals who are independent of the investigational sites and who have expertise in multiple disciplines, including neurology, neurosurgery, interventional neuroradiology, and biostatistics/epidemiology. The independence of the members will be maintained and bias minimized by blinding, to the extent possible, all members to individual subject and treating center identity when reviewing study data. This board shall provide recommendations to the sponsor regarding stopping/continuing enrollment in the study, including carrying out planned formal interim analyses.

ntained and bias minimized by blinding, to the extent possible, all members to individual subject and treating center identity when reviewing study data. This board shall provide recommendations to the sponsor regarding stopping/continuing enrollment in the study, including carrying out planned formal interim analyses. Discussion The SWIFT PRIME trial will provide definitive information on the efficacy and safety of the Solitaire revascularization devices when added to IV tPA, in comparison with therapy with IV tPA alone. The study was designed to incorporate lessons from trials of earlier generation endovascular interventions that failed to demonstrate treatment benefit (14–17). Unlike those prior studies, SWIFT PRIME tests a highly effective recanalization device, the Solitaire stent retriever, which achieves recanalization much more frequently and rapidly than earlier generation therapies (9). SWIFT PRIME ensures that all enrolled patients have appropriate target occlusions by mandating CTA or MRA imaging at entry. SWIFT PRIME is enrolling only patients with occlusion locations that are distinctively responsive to endovascular therapy, by including ICA and M1 MCA lesions that respond poorly to IV tPA and excluding M2 and more distal MCA lesions that frequently benefit from IV tPA alone. SWIFT PRIME is identifying patients who have salvageable brain left to save at the time of enrollment by requiring small core size on ASPECTS or RAPID imaging. SWIFT PRIME is ensuring that all patients who clinicians expect may respond to therapy will be enrolled by requiring that sites commit to enroll all eligible study patients. SWIFT PRIME is minimizing progression of infarct that occurs during delays between the time of study enrollment and endovascular reperfusion through rigorous focus on management efficiency and intensive quality improvement, training, and feedback about interventional workflow during the trial. Lastly, SWIFT PRIME is providing equivalent concomitant therapy in the two treatment arms, using full-dose tPA in both study groups, rather than a reduced dose in the neuroendovascular arm.

on management efficiency and intensive quality improvement, training, and feedback about interventional workflow during the trial. Lastly, SWIFT PRIME is providing equivalent concomitant therapy in the two treatment arms, using full-dose tPA in both study groups, rather than a reduced dose in the neuroendovascular arm. SWIFT PRIME enrolled its first patient on December 30, 2012. When completed, SWIFT PRIME will provide pivotal data allowing assessment of the efficacy and safety of the Solitaire revascularization device for reperfusion of AIS caused by large intracranial vessel occlusion. Study personnel Global PI Jeffrey L. Saver, MD Executive committee Hans-Christoph Diener, MD (EU PI), Mayank Goyal, MD (global imaging and workflow PI), Elad Levy, MD (US interventional PI), Alain Bonafé, MD (EU interventional PI), Vitor Mendes Pereira, MD (EU interventional PI), and Reza Jahan, MD (interventional advisor) Steering committee Gregory W. Albers, MD (global penumbral imaging investigator), Christophe Cognard, MD, David Cohen, MD, Werner Hacke, MD, Olav Jansen, MD, Tudor G. Jovin, MD, Heinrich P. Mattle, MD, Raul G. Nogueira, MD, Adnan H. Siddiqui, MD, and DiLeep R. Yavagal, MD Data safety and monitoring committee Rüdiger von Kummer, MD (chair), Wade Smith, MD, Francis Turjman, MD, and Scott Hamilton, PhD Clinical events committee Arun Amar, MD (chair), Nerses Sanossian, MD, and Yince Loh, MD

Steering committee Gregory W. Albers, MD (global penumbral imaging investigator), Christophe Cognard, MD, David Cohen, MD, Werner Hacke, MD, Olav Jansen, MD, Tudor G. Jovin, MD, Heinrich P. Mattle, MD, Raul G. Nogueira, MD, Adnan H. Siddiqui, MD, and DiLeep R. Yavagal, MD Data safety and monitoring committee Rüdiger von Kummer, MD (chair), Wade Smith, MD, Francis Turjman, MD, and Scott Hamilton, PhD Clinical events committee Arun Amar, MD (chair), Nerses Sanossian, MD, and Yince Loh, MD Site investigators and coordinators S. Starkman, MD (PI), J. Guzy (UCLA/Ronald Reagan UCLA Medical Center); W. Clark, MD (PI), S. Jamieson (Oregon Health and Science University); V. Reddy, MD (PI), L. Baxendell (University of Pittsburgh Medical Center); A. Siddiqui, MD (PI), S. Hall (University of Buffalo Neurosurgery/Buffalo General Hospital); D. Yavagal, MD (PI), K. Ramdas (University of Miami/Jackson Memorial Hospital); T. Devlin, MD (PI), K. Barton (Chattanooga Center for Neurologic Research/Erlanger Hospital); B. Jagadeesan, MD (PI), D. Hildebrandt (Hennepin Country Medical Center); B-F. Fitzsimmons, MD (PI), T. Larson (Medical College of Wisconsin/Froedtert Memorial Lutheran Hospital); R. Ecker, MD (PI), L. Connolly (Maine Medical Center); R. Budzik, MD (PI), M. Taylor (Ohio Health Research Institute/Riverside Methodist Hospital); I. Acosta, MD (PI), E. Bonwit (Florida Hospital); E. Deshaies, MD (PI), M. Villwock (State University of New York Upstate); M. Jumaa, MD (PI), T. Hendrickson (Promedica Toledo Hospital); C. Ramsey, MD (PI), S. Renfrow (Central Baptist Hospital); M.S. Hussain, MD (PI), A. Richmond (Cleveland Clinic Cerebrovascular Center, NI); J. Carpenter, MD (PI), J. Domico (West Virginia University Hospital); V. Deshmukh, MD (PI), M. Rodriguez (Providence Brain and Spine Institute); A. Puri, MD (PI), M. Howk (University of Massachusetts Medical Center); R. Nogueira, MD (PI), S. Doppelheuer (Emory University/Grady Medical Center); D. Lopes, MD (PI), C. Anton (Rush University Medical Center); C. Martin, MD (PI), B. Brion (Saint Luke's Hospital); H. Farid, MD (PI), L. Cross (St. Jude Medical Center); A. Hassan, DO (PI), L. Jones-Fullingim (Valley Baptist Medical Center-Harlingen); A. Malek, MD (PI), M. Smith, P. Beck (Tenet Health Systems); A. Bonafé, MD (PI), M. Moynier (CHU de Montpellier – Hôpital Gui de Chauliac); O. Jansen, MD (PI), S. Krieter (Universitätsklinikum Kiel- Abteilung für Neuroradiologie); JF. Arenillas, MD (PI), FJ Reyes Muñoz (Hospital Clinico Universitario de Valladolid); R. du Mesnil de Rochemont, MD (PI), H. Braun (Klinikum der Johann Wolfgang Goethe-Universität); L. Remonda, MD (PI), M.

ital Gui de Chauliac); O. Jansen, MD (PI), S. Krieter (Universitätsklinikum Kiel- Abteilung für Neuroradiologie); JF. Arenillas, MD (PI), FJ Reyes Muñoz (Hospital Clinico Universitario de Valladolid); R. du Mesnil de Rochemont, MD (PI), H. Braun (Klinikum der Johann Wolfgang Goethe-Universität); L. Remonda, MD (PI), M. Guglielmetti (Kantonsspital Aarau); C. Weimar, MD (PI), M. Dietzold (Universitätsklinikum Essen); K. Hansen, MD (PI), T. Hyldal (Rigshospitalet, Copenhagen University Hospital); P. Papanagiotou, MD (PI), H. Merdivan (Klinikum Bremen-Mitte); M. Killer-Oberpfalzer, MD (PI), A. Jedlitschka (Universitätsklinikum Christian Doppler Klinik Salzburg); P. Ringleb, MD (PI), I. Ludwig (Universitätsklinikum Heidelberg/ Neurologische Klinik und Poliklinik); G. Reimann, MD (PI), K. Burg (Klinikum Dortmund); C. Brekenfeld, MD (PI), G. Wortmann (Universitätsklinikum Hamburg-Eppendorf); S. Prothmann, MD (PI), B. Schwaiger (Klinikum rechts der Isar – Technische Universität München); H-P Haring, MD (PI), K. SØrensen (Landes-Nervenklinik Wagner-Jauregg); and G. Andersen, MD (PI), E. Bach (Aarhus University Hospital)

Stroke and transient ischemic attack (TIA) are complicated by a high risk of recurrence during the first few hours and days after the index event; a risk that can be mitigated with early and multimodal prophylaxis (1,2). Two mega-trials demonstrated that early antiplatelet therapy with aspirin is effective at reducing recurrence after ischemic stroke (3,4). A meta-analysis of small trials suggested that dual antiplatelet therapy given within 72 h of onset was superior to monotherapy in reducing the early risk of recurrence (5); this systematic review suggested that the composition of antiplatelets (aspirin, clopidogrel, dipyridamole) was less important than using two rather than one agent. Subsequently, the large CHANCE trial found that combined aspirin and clopidogrel was superior to aspirin alone in preventing stroke recurrence by 90 days in Chinese patients with minor ischemic stroke or TIA when randomized within 24 h of onset (6,7).

pyridamole) was less important than using two rather than one agent. Subsequently, the large CHANCE trial found that combined aspirin and clopidogrel was superior to aspirin alone in preventing stroke recurrence by 90 days in Chinese patients with minor ischemic stroke or TIA when randomized within 24 h of onset (6,7). If dual therapy is superior to monotherapy for acute treatment/prophylaxis, then intensive/triple antiplatelet therapy (aspirin + clopidogrel + dipyridamole) might be better still providing the risk of recurrence is high and bleeding does not become excessive. A series of ‘proof-of-mechanism’ and ‘proof-of-concept’ studies have investigated this approach (8–12). In vitro studies found that triple therapy was most effective in inhibiting platelet aggregation, platelet–leucocyte conjugation, and leucocyte activation (8–10). In multiway crossover phase I and II trials comparing short-term administration of mono, dual and triple antiplatelet therapy, the combination of aspirin + clopidogrel, with or without dipyridamole, was most potent in inhibiting platelet function ex vivo in both normal volunteers and participants with previous stroke/TIA (11,12). A small parallel group trial of intensive therapy in participants with stroke reported that triple therapy (vs. aspirin alone) was feasible to administer for up to 24 months (13) although there was a nonsignificant trend to increased bleeding with intensive treatment. Chronic triple treatment may be useful in clinical practice in participants at very high risk of recurrence, defined as recurrence on dual antiplatelet therapy (14).

pirin alone) was feasible to administer for up to 24 months (13) although there was a nonsignificant trend to increased bleeding with intensive treatment. Chronic triple treatment may be useful in clinical practice in participants at very high risk of recurrence, defined as recurrence on dual antiplatelet therapy (14). On the basis of these preclinical and clinical data showing feasibility, tolerability and apparent safety of intensive/triple antiplatelets, and the potential for efficacy, the large ‘Triple Antiplatelets for Reducing Dependency after Ischaemic Stroke’ (TARDIS) was started and is ongoing. TARDIS is assessing, in a prospective, randomized, open-label, blinded-outcome design, the safety and efficacy of intensive vs. guideline antiplatelet therapy. The trial commenced in 2009 and will reach 50% of its planned recruitment of 4,100 patients during 2014. The independent Data Monitoring Committee has assessed unblinded data from the trial on eight occasions to date and, on each occasion, recommended that TARDIS should continue. The accompanying Supporting Information Appendix S1 details the statistical analysis plan (SAP) and is published during recruitment and well before final data cleaning and locking of the trial database so that analyses are not data driven or selectively reported (15). As for the ENOS trial (16), this SAP includes not just information on the planned primary publications but also provides detailed information on the intended baseline characteristics publication.

ore final data cleaning and locking of the trial database so that analyses are not data driven or selectively reported (15). As for the ENOS trial (16), this SAP includes not just information on the planned primary publications but also provides detailed information on the intended baseline characteristics publication. TARDIS will be reported as both a prevention trial, i.e. efficacy of intensive antiplatelet agents for reducing the frequency and severity of recurrent stroke and TIA (primary aim), and an acute intervention trial, i.e. efficacy in shifting functional outcome. TARDIS is using a novel primary outcome based on both the frequency and severity of recurrent strokes. Conventionally, vascular prevention trials just count recurrent events. However recurrent events may be mild, severe or fatal, and this information can allow ordered categorical outcomes to be defined: fatal event/severe event/moderate event/mild event/no event. Analysis of such polytomous outcomes is more efficient statistically, i.e. they provide improved statistical power for a given sample size, or allow a trial to be smaller for a given power, as shown in an empirical re-analysis of published vascular prevention trials (17,18). This approach follows that used for the design and analysis of trials in acute stroke (19,20). Similarly, adjusted analyses provide additional statistical power (21), are important if minimization is used during the process of randomization (22), and help address any minor imbalances present at baseline due to chance. As a result, these statistical approaches are likely to be more sensitive to any treatment effect and, as such, are recommended by the European Stroke Organisation (23). The use of these approaches, and inclusion of TIA as part of the spectrum of outcomes, allows the size of TARDIS to be almost halved from ∼8000 patients. The collection of all baseline data needed for covariate adjustment of the primary outcome should mean there is no need for imputation for missing data.

isation (23). The use of these approaches, and inclusion of TIA as part of the spectrum of outcomes, allows the size of TARDIS to be almost halved from ∼8000 patients. The collection of all baseline data needed for covariate adjustment of the primary outcome should mean there is no need for imputation for missing data. Guidelines for use of antiplatelet agents after stroke have changed during the conduct of the trial. At the start of the trial, the UK National Institute for Health and Clinical Excellence (NICE) had recommended the use of aspirin and dipyridamole for secondary prevention (24) and the initial trial protocol defined this as the guideline comparator. However, the widespread availability of inexpensive generic clopidogrel, and significant randomized evidence supporting the use of clopidogrel after stroke, led NICE to update their earlier guidance in 2010 with a recommendation that clopidogrel should be used for secondary prophylaxis. As a result, the protocol was updated and allowed the use of either combined aspirin and dipyridamole, or clopidogrel alone, as the comparator. Investigators are allowed to choose whether randomization includes one or both comparators and this choice can be made separately for stroke and TIA. Choices can be changed but not within 48 h so they cannot influence randomization for a particular patient.

rin and dipyridamole, or clopidogrel alone, as the comparator. Investigators are allowed to choose whether randomization includes one or both comparators and this choice can be made separately for stroke and TIA. Choices can be changed but not within 48 h so they cannot influence randomization for a particular patient. This SAP also informs some of the content of the final trial report to be submitted to the NIHR Health Technology Appraisal (HTA) program; the final report will be submitted for publication in the HTA Journal, part of the NIHR collection of peer-reviewed open access journals. In the future, data from TARDIS will be integrated into individual patient data meta-analyses of antiplatelet agents in acute stroke, and made available to participating countries, and the ‘Virtual International Stroke Trials Archive’ (VISTA) (25). Ultimately, a subset of the data will be made available over the web, as with the International Stroke Trial (26). Similarly, anonymized baseline and on-treatment neuroimaging data will be published (27). Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher's website. Appendix S1  Statistical analysis plan (TARDIS).

Introduction There is evidence that the presence of cardiovascular disease pathology can increase the future risk of Alzheimer's disease (AD) and cognitive decline 1. White matter hyperintensities (WMH) and cerebral microbleeds (CMB) are generally considered to reflect cerebrovascular burden in ageing. They are manifestations and markers of cerebral small‐vessel disease and often co‐occur 2. The mechanisms underlying significant association between cardiovascular and neurodegenerative pathology are unclear; however, there are three main hypotheses 1. Firstly, it is possible that cardiovascular diseases and AD/cognitive decline share common risk factors and are not mechanistically related 1. Secondly, it is possible that cardiovascular burden may expedite progression of AD/cognitive decline through promoting atherosclerosis and accumulations of amyloid‐beta plaques, and/or (thirdly) by increasing vulnerability to such pathology and lowering the threshold at which cognitive decline becomes apparent behaviorally, even in the absence of a mechanistic link 1, 3.

ay expedite progression of AD/cognitive decline through promoting atherosclerosis and accumulations of amyloid‐beta plaques, and/or (thirdly) by increasing vulnerability to such pathology and lowering the threshold at which cognitive decline becomes apparent behaviorally, even in the absence of a mechanistic link 1, 3. Two genetic risk factors for AD and age‐related cognitive decline are in the APOE and TOMM40 gene loci 4. Two recent meta‐analyses reported no overall significant association between APOE ε4 and WMH in older adults (P > 0·05). Paternoster et al. 5 did not stratify analyses by whether participants were generally healthy or not (total n = 8546), whereas Schilling et al. 6 did (‘healthy’ n = 8405). However, these reports were not independent because some common data were included in both. In contrast, two meta‐analyses reported a significant association between APOE ε4 and CMB: Schilling et al. 6 (‘healthy’ n = 5387; P = 0·002) and Maxwell et al. 7 (total n = 7351; P < 0·01). These two reports also had a degree of overlapping datasets. A more recent individual study 8 reported similar results (n = 1965, P < 0·05).

, two meta‐analyses reported a significant association between APOE ε4 and CMB: Schilling et al. 6 (‘healthy’ n = 5387; P = 0·002) and Maxwell et al. 7 (total n = 7351; P < 0·01). These two reports also had a degree of overlapping datasets. A more recent individual study 8 reported similar results (n = 1965, P < 0·05). In the above reports, the positive association between APOE ε4 and CMB burden was not completely consistent; other genetic predictor variables may exert independent effects. The TOMM40 rs10524523 (‘523’) variable length poly‐T repeat has been significantly associated with brain phenotypes such as cognitive decline, independent of APOE genotype 9. The TOMM40 523 locus is characterized by a variable number of T residues (poly‐T repeats) that can be grouped into ‘short’ (<20; ‘S’), ‘long’ (20–29; ‘L’), and ‘very long’ (≥30; ‘VL’) 10. Roses 11 plotted histograms showing the distributions of poly‐T repeat lengths in different APOE genotypes: ε3/ε3, ε3/ε4, and ε4/ε4. The poly‐T repeat was strongly linked with the APOE ε haplotype; ε4 is linked to L, with ε3 linked to either S or VL alleles 4; investigating the effects of TOMM40 523 genotype on brain‐related phenotypes may explain some of the heterogeneity in the possible APOE ε4 and WMH/CMB association. This study therefore aims to contribute a large amount of relevant genetic APOE/TOMM40 and WMH/CMB brain imaging data, from a sample of community‐dwelling older adults.

g the effects of TOMM40 523 genotype on brain‐related phenotypes may explain some of the heterogeneity in the possible APOE ε4 and WMH/CMB association. This study therefore aims to contribute a large amount of relevant genetic APOE/TOMM40 and WMH/CMB brain imaging data, from a sample of community‐dwelling older adults. Methods Sample and genotyping The Lothian Birth Cohort 1936 (LBC1936) is a longitudinal sample of generally healthy community‐dwelling older adults 12. All participants were born in 1936, and most resided in the Edinburgh area of Scotland in older age. The sample received detailed cognitive, medical, and demographic assessments at the Wellcome Trust Clinical Research Facility (Edinburgh; http://www.wtcrf.ed.ac.uk) at age ∼73 years. Participants underwent detailed brain MRI around the same time 13 (mean interval = 65·0 days, SD = 39·5). Of the 866 LBC1936 participants that attended clinic assessment, 700 completed neuroimaging (mean age = 72·70, SD = 0·7). Details of LBC1936 recruitment and assessment, including aspects of possible selection bias and attrition, can be found in two cohort protocol papers by Deary et al. 12, 14. Participants were genotyped by TaqMan assay for APOE ε (Applied Biosystems, Carlsbad, CA, USA) using DNA isolated from whole blood 12. TOMM40 523 was genotyped by the laboratory of Dr. Ornit Chiba‐Falek (Duke University) as described previously 15. Brain MRI

Methods Sample and genotyping The Lothian Birth Cohort 1936 (LBC1936) is a longitudinal sample of generally healthy community‐dwelling older adults 12. All participants were born in 1936, and most resided in the Edinburgh area of Scotland in older age. The sample received detailed cognitive, medical, and demographic assessments at the Wellcome Trust Clinical Research Facility (Edinburgh; http://www.wtcrf.ed.ac.uk) at age ∼73 years. Participants underwent detailed brain MRI around the same time 13 (mean interval = 65·0 days, SD = 39·5). Of the 866 LBC1936 participants that attended clinic assessment, 700 completed neuroimaging (mean age = 72·70, SD = 0·7). Details of LBC1936 recruitment and assessment, including aspects of possible selection bias and attrition, can be found in two cohort protocol papers by Deary et al. 12, 14. Participants were genotyped by TaqMan assay for APOE ε (Applied Biosystems, Carlsbad, CA, USA) using DNA isolated from whole blood 12. TOMM40 523 was genotyped by the laboratory of Dr. Ornit Chiba‐Falek (Duke University) as described previously 15. Brain MRI Participants underwent whole brain structural MRI, acquired using a GE Signa Horizon 1·5 T HDxt clinical scanner (General Electric, Milwaukee, WI, USA) equipped with a self‐shielding gradient set (33 mT/m maximum gradient strength) and manufacturer‐supplied eight‐channel phased‐array head coil, lasting around 70 min. In addition to standard structural T2‐, T2*‐, and FLAIR‐weighted MRI, the imaging protocol included a high‐resolution T1‐weighted volume sequence acquired in the coronal plane with field‐of‐view of 256 × 256 mm, imaging matrix 192 × 192 (zero‐filled to 256 × 256), 160 1·3‐mm thick slices giving 1 × 1 × 1·3‐mm voxel dimensions 13. The repetition, echo, and inversion times were 10, 4, and 500 ms respectively. The detailed protocol for WMH/CMB image processing, and intracranial/total brain volume measurement, is published by Wardlaw et al. 13. WMH volumes were calculated from binary masks generated by an in‐house‐developed and validated software tool written in MATLAB that applies a technique named Multispectral Colouring Modulation and Variance Identification: 1936 [MCMxxxVI 16 ]. Visual scoring of WMH was also performed using the Fazekas scale by experienced neuroradiologists.

ulated from binary masks generated by an in‐house‐developed and validated software tool written in MATLAB that applies a technique named Multispectral Colouring Modulation and Variance Identification: 1936 [MCMxxxVI 16 ]. Visual scoring of WMH was also performed using the Fazekas scale by experienced neuroradiologists. Microhemorrhages (i.e. CMBs) were coded for number and distribution using a simplified version of the Brain Observer MicroBleed Scale [BOMBS 17 ], which considers microbleeds as small homogenous round foci of low signal intensity on T2*‐weighted images, of less than 10 mm in diameter. This rating scale is used to record the number of observed definite or possible microbleeds in the right/left hemispheres, delineated into bleeds <5 mm and 5–10 mm. Because of the relatively low frequency of CMB's in the LBC1936 sample, we examined the presence of ≥1 definite/possible microbleeds, strictly lobar microbleeds, strictly deep or infratentorial microbleeds. Any significant findings were reanalyzed as definite microbleeds only. Inter‐ and intra‐rater reliability standards have been reported in previous work 13. Genotyping was performed blind to imaging (and vice versa). Imaging lesions were defined according to STRIVE recommendations 2. Of the 700 participants that completed brain MRI, 25 had one or more lacunar infarcts, and given this low frequency, we did not consider this variable further.

en reported in previous work 13. Genotyping was performed blind to imaging (and vice versa). Imaging lesions were defined according to STRIVE recommendations 2. Of the 700 participants that completed brain MRI, 25 had one or more lacunar infarcts, and given this low frequency, we did not consider this variable further. Statistical analysis Age in days and gender were included as covariates. An online calculator was used to perform tests of Hardy–Weinberg equilibrium and determine minor allele frequencies (http://www.had2know.com/academics/hardy‐weinberg‐equilibrium‐calculator‐3‐alleles.html). Volumetric WMH data were transformed with a natural logarithmic function to provide a more normal distribution. Data were analyzed with the IBM SPSS statistics program (version 17; IBM, Armonk, NY, USA). Univariate general linear models tested the effects of separate APOE and TOMM40 genotypes upon imaging variables. Specifically, the effects of APOE ε4 ‘risk’ allele presence vs. absence (i.e. ε2/ε4; ε3/ε4; ε4/ε4 vs. ε2/ε2; ε2/ε3; ε3/ε3) were tested. To assess the independent effects of TOMM40 523 genotype, we tested for effects of S allele frequency vs. pooled L/VL alleles (simply L*; S/S vs. S/L* vs. L*/L*) in participants with the ‘neutral’ APOE ε3/ε3 genotype 9 and then in the ε3/ε4 genotype subgroup, because TOMM40 523 genotype may interact with the ε4 allele 18.

assess the independent effects of TOMM40 523 genotype, we tested for effects of S allele frequency vs. pooled L/VL alleles (simply L*; S/S vs. S/L* vs. L*/L*) in participants with the ‘neutral’ APOE ε3/ε3 genotype 9 and then in the ε3/ε4 genotype subgroup, because TOMM40 523 genotype may interact with the ε4 allele 18. Results Of the 866 LBC1936 participants that attended clinic assessment, 700 completed neuroimaging. Participants were excluded from analysis if they had Mini‐Mental State Exam scores below 24 or had not completed the test 19. [A cutoff of 24 was used because this is considered the lowest possible score within the range of ‘no cognitive impairment’, and is a common approach 20.]. This left 694 participants, of whom 624 and 636 had successful genotyping for APOE ε and TOMM40 523, respectively. Allele frequencies were in Hardy–Weinberg equilibrium for APOE (ε2 = 7·4%, ε3 = 77·0%; ε4 = 15·6%) and TOMM40 (S = 40·9%, L = 15·3%, and VL = 43·9%; both P > 0·05). Allele frequencies are shown in Table S1. There were no significant associations between APOE ε4 presence (vs. absence), or TOMM40 523 genotype (in any analyses), and WMH/CMB variables (all P > 0·05; Tables S2–S3).

Results Of the 866 LBC1936 participants that attended clinic assessment, 700 completed neuroimaging. Participants were excluded from analysis if they had Mini‐Mental State Exam scores below 24 or had not completed the test 19. [A cutoff of 24 was used because this is considered the lowest possible score within the range of ‘no cognitive impairment’, and is a common approach 20.]. This left 694 participants, of whom 624 and 636 had successful genotyping for APOE ε and TOMM40 523, respectively. Allele frequencies were in Hardy–Weinberg equilibrium for APOE (ε2 = 7·4%, ε3 = 77·0%; ε4 = 15·6%) and TOMM40 (S = 40·9%, L = 15·3%, and VL = 43·9%; both P > 0·05). Allele frequencies are shown in Table S1. There were no significant associations between APOE ε4 presence (vs. absence), or TOMM40 523 genotype (in any analyses), and WMH/CMB variables (all P > 0·05; Tables S2–S3). Discussion Our findings align with previous meta‐analyses in observing no significant APOE/WMH association 5, 6. In terms of CMB, this report does not align with recent meta‐analyses that concluded significant deleterious effects of APOE ε4 6, 7. Of those meta‐analyses, Schilling et al. 6 observed a significant effect of APOE ε4 in general populations only – and not in pooled samples with neurological or vascular disorders; the lack of association in LBC1936 may be because of the relatively good health of the sample. Our findings do align with two individual studies of generally healthy older adults (included in the meta‐analyses) which reported no significant APOE ε4 present (vs. absent) effects [by Sveinbjornsdottir et al. 21 (n = 1725) and Jeerakathil et al. 22 (n = 368)]. There was no evidence of an independent effect of TOMM40 523 genotype here.

gn with two individual studies of generally healthy older adults (included in the meta‐analyses) which reported no significant APOE ε4 present (vs. absent) effects [by Sveinbjornsdottir et al. 21 (n = 1725) and Jeerakathil et al. 22 (n = 368)]. There was no evidence of an independent effect of TOMM40 523 genotype here. Previous significant associations in individual CMB reports may perhaps reflect a degree of type 1 error, particularly in smaller samples. Several studies report broader age ranges than examined here (71–74 years; SD = 0·7) 7. Any effect of age on CMB may be via processes associated with age; controlling for age statistically is unlikely to completely eradicate these effects 23, so wide age ranges could possibly contribute to spurious genetic associations. The BOMBS instrument allows raters to note CMBs as either definite or possible [a cautious category to avoid misclassifications of mimics 17 ]. It would be interesting to examine if previously reported significant APOE‐CMB associations are affected when analyzed to incorporate possible microbleeds/mimics.

Previous significant associations in individual CMB reports may perhaps reflect a degree of type 1 error, particularly in smaller samples. Several studies report broader age ranges than examined here (71–74 years; SD = 0·7) 7. Any effect of age on CMB may be via processes associated with age; controlling for age statistically is unlikely to completely eradicate these effects 23, so wide age ranges could possibly contribute to spurious genetic associations. The BOMBS instrument allows raters to note CMBs as either definite or possible [a cautious category to avoid misclassifications of mimics 17 ]. It would be interesting to examine if previously reported significant APOE‐CMB associations are affected when analyzed to incorporate possible microbleeds/mimics. It is possible that the sample size examined here is not sufficiently powered to detect any possibly modest effects of APOE or TOMM40 genotypes on WMH/CMB 5. It is possible that the LBC1936 sample is generally healthier when compared with other samples, exacerbated by a selection bias where healthier participants were more likely to attend brain MRI assessment 24. Generally, the LBC1936 sample is slightly restricted in range towards the upper end of general mental ability and socioeconomic status 14. In addition, APOE ε4 genotype has previously been associated with earlier mortality and cardiovascular disease 25: it is possible that a selection bias exists whereby healthier participants are more likely to attend cognitive or brain imaging assessment, and this may contribute to our finding no effect of APOE/TOMM40 genotypes on WMH/CMB phenotypes with MRI.

usly been associated with earlier mortality and cardiovascular disease 25: it is possible that a selection bias exists whereby healthier participants are more likely to attend cognitive or brain imaging assessment, and this may contribute to our finding no effect of APOE/TOMM40 genotypes on WMH/CMB phenotypes with MRI. Maxwell et al. 7 estimated with a ‘fail‐safe N calculation’ that null studies including at least 7700 participants would be required to attenuate their meta‐analysis APOE ε4/CMB association (reported P = 0·01) to non‐significance (i.e. P > 0·05). Further independent studies will help to define the more exact nature and strength of any APOE/CMB association in generally healthy populations. Authors' contributions Designed the experiments: D. M. L., J. M. W., D. J. P., I. J. D. Analyzed the data/wrote the manuscript: D. M. L. Interpreted the results, critically revised content, and approved the final version: All authors. Supporting information Table S1. Frequency statistics for APOE/TOMM40 poly‐T repeat genotypes. Click here for additional data file. Table S2. APOE/TOMM40 genotypes and white matter hyperintensities/cerebral microbleeds: association statistics. Click here for additional data file. Table S3. TOMM40 genotypes and white matter hyperintensities/cerebral microbleeds in APOE subgroups: association statistics. Click here for additional data file.

Click here for additional data file. Table S2. APOE/TOMM40 genotypes and white matter hyperintensities/cerebral microbleeds: association statistics. Click here for additional data file. Table S3. TOMM40 genotypes and white matter hyperintensities/cerebral microbleeds in APOE subgroups: association statistics. Click here for additional data file. Acknowledgements The participation of LBC1936 members is gratefully acknowledged. The work was undertaken within the University of Edinburgh Centre for Cognitive Ageing and Cognitive Epidemiology (http://www.ccace.ed.ac.uk), part of the cross‐council Lifelong Health and Wellbeing Initiative (G0700704/84698).

Introduction and rationale Antiplatelet therapy is a standard strategy for prevention of recurrent noncardioembolic stroke, in addition to lifestyle modifications and risk factor management. A previous meta-analysis found that conventional antiplatelet agents reduced the relative risk for stroke by about 20% in patients with a history of ischemic stroke (IS) or transient ischemic attack (TIA) 1. However, more effective antiplatelet therapy is needed in patients with intracranial arterial stenosis, carotid stenosis, or multiple vascular risk factors, because these conditions are associated with a higher risk of recurrence 2–6. Since antiplatelet agents act through a range of mechanisms, the combination of dual agents can produce more effective stroke prevention. For example, the combination of aspirin with extended-release dipyridamole showed a superior effect on the prevention of recurrent IS, compared with aspirin alone 7,8. However, the efficacy of this combination could not be verified in a double-blind trial in Japan 9, and dipyridamole remains unapproved in Japan for prophylaxis of IS recurrence. Although relatively short-term combinations of aspirin and clopidogrel seem to prevent IS recurrence in patients with IS/TIA more effectively than aspirin alone 10–13, long-term usage is not generally recommended because it increased the incidence of serious hemorrhage and produced little reduction in the incidence of vascular events 14–16.

ort-term combinations of aspirin and clopidogrel seem to prevent IS recurrence in patients with IS/TIA more effectively than aspirin alone 10–13, long-term usage is not generally recommended because it increased the incidence of serious hemorrhage and produced little reduction in the incidence of vascular events 14–16. In the initial Cilostazol Stroke Prevention Study (CSPS) 17, this phosphodiesterase 3 inhibitor was shown to decrease IS recurrence without increasing serious bleeding, as compared with placebo. Cilostazol also decreased stroke [IS, intracerebral hemorrhage (ICH), or subarachnoid hemorrhage (SAH)] and halved serious bleeding as compared with aspirin in the second CSPS (CSPS 2) 18. In addition, a combination of cilostazol with aspirin for IS patients with intracranial arterial stenosis was investigated in three small studies 19–21; the Trial Of Cilostazol in Symptomatic intracranial arterial Stenosis (TOSS) 19 showed reduced stenosis progression, and all three studies found relatively few bleeding events, as compared with aspirin alone or aspirin plus clopidogrel.

with intracranial arterial stenosis was investigated in three small studies 19–21; the Trial Of Cilostazol in Symptomatic intracranial arterial Stenosis (TOSS) 19 showed reduced stenosis progression, and all three studies found relatively few bleeding events, as compared with aspirin alone or aspirin plus clopidogrel. In patients with peripheral arterial disease, the addition of cilostazol to a therapy with aspirin and/or clopidogrel did not increase bleeding times, compared with either agent alone 22. In another study of patients with peripheral arterial disease, the incidence of hemorrhagic events was comparable in subjects receiving cilostazol and in those receiving placebo 23. In a meta-analysis of randomized controlled trials in patients with a drug-eluting stent, the incidence of hemorrhagic events in the group receiving triple antiplatelet therapy (cilostazol plus aspirin and clopidogrel) was similar to that in the group receiving dual antiplatelet therapy (DAPT) with aspirin and clopidogrel 24. Thus, the addition of cilostazol may enhance the preventive effects of classical antiplatelet agents (aspirin, clopidogrel) in high-risk IS patients, without increasing hemorrhagic risk.

aspirin and clopidogrel) was similar to that in the group receiving dual antiplatelet therapy (DAPT) with aspirin and clopidogrel 24. Thus, the addition of cilostazol may enhance the preventive effects of classical antiplatelet agents (aspirin, clopidogrel) in high-risk IS patients, without increasing hemorrhagic risk. Animal studies have suggested that cilostazol protected mice from cerebral bleeds by reducing matrix metalloproteinase (MMP)-9 activity, and by preventing blood-brain barrier opening by inhibiting loss of claudin-5 expression 25–27. Cilostazol also increased intracellular cAMP concentration, thereby promoting the barrier function of tight junctions in brain capillary endothelial cells. Since the cAMP-dependent induction of claudin-5 expression is implicated in the promotion of endothelial cell tight junctions, the protection of endothelial cells may be explained by the phosphodiesterase-III inhibitory activity of cilostazol. The Cilostazol Stroke Prevention Study for Antiplatelet Combination (CSPS.com; (ClinicalTrials.gov identifier: NCT01995370) will investigate the preventive effect of DAPT including cilostazol on IS recurrence in high-risk patients with noncardioembolic IS. Patients receiving cilostazol together with aspirin or clopidogrel will be compared with those treated with aspirin or clopidogrel monotherapy.

ClinicalTrials.gov identifier: NCT01995370) will investigate the preventive effect of DAPT including cilostazol on IS recurrence in high-risk patients with noncardioembolic IS. Patients receiving cilostazol together with aspirin or clopidogrel will be compared with those treated with aspirin or clopidogrel monotherapy. Methods Design The CSPS.com is a multicenter, randomized, open-label parallel-group trial. All trial centers are required to receive approval from the relevant ethics committee before initiation of the trial. Written informed consent must be obtained from each patient before enrollment. This trial is registered with ClinicalTrials.gov (NCT01995370) and the UMIN Clinical Trials Registry (000012180). Figure 1 shows a flowchart of the trial design. Table 1 shows the schedule of observations, examinations, and assessments. Figure 1 Trial design flowchart. Change in the dose of aspirin or clopidogrel will not be permitted after informed consent is obtained. Cilostazol treatment can be started with 100 mg/day, provided the dose is increased to 200 mg/day within 15 days. Table 1 Trial schedule

Methods Design The CSPS.com is a multicenter, randomized, open-label parallel-group trial. All trial centers are required to receive approval from the relevant ethics committee before initiation of the trial. Written informed consent must be obtained from each patient before enrollment. This trial is registered with ClinicalTrials.gov (NCT01995370) and the UMIN Clinical Trials Registry (000012180). Figure 1 shows a flowchart of the trial design. Table 1 shows the schedule of observations, examinations, and assessments. Figure 1 Trial design flowchart. Change in the dose of aspirin or clopidogrel will not be permitted after informed consent is obtained. Cilostazol treatment can be started with 100 mg/day, provided the dose is increased to 200 mg/day within 15 days. Table 1 Trial schedule Assessment Period (Date of onset) Registration Start date of observation 1 M 3 M 6 M 12 M Thereafter, every 6 months Completion of observation (*) Informed consent ○ Demographics ○ Modified Rankin Scale (mRS) ○ ○ ○ ○ ○ ○ ○ Compliance status of trial drugs ○ ○ ○ ○ ○ ○ ○ Concomitant medication ○ ○ ○ ○ ○ ○ ○ Blood pressure ○ ○ ○ ○ ○ ○ ○ Head MRI ○ Head MRA, T2*WI △ Carotid artery imaging (US, CTA, MRA) △ Laboratory test (blood) ○ Laboratory test (urine) ○ Chest X-ray ○ ECG ○ Adverse event * When patients are withdrawn from the trial, or one-year after the start of the protocol treatment in the last patient. ○, required; △, optional.

Assessment Period (Date of onset) Registration Start date of observation 1 M 3 M 6 M 12 M Thereafter, every 6 months Completion of observation (*) Informed consent ○ Demographics ○ Modified Rankin Scale (mRS) ○ ○ ○ ○ ○ ○ ○ Compliance status of trial drugs ○ ○ ○ ○ ○ ○ ○ Concomitant medication ○ ○ ○ ○ ○ ○ ○ Blood pressure ○ ○ ○ ○ ○ ○ ○ Head MRI ○ Head MRA, T2*WI △ Carotid artery imaging (US, CTA, MRA) △ Laboratory test (blood) ○ Laboratory test (urine) ○ Chest X-ray ○ ECG ○ Adverse event * When patients are withdrawn from the trial, or one-year after the start of the protocol treatment in the last patient. ○, required; △, optional. Patient population Patients who have developed noncardioembolic IS between 8 and 180 days before the start of the protocol treatment are the target population of the trial. Inclusion and exclusion criteria are listed in Table 2. Patients are asked to continue taking clopidogrel or aspirin alone as an antiplatelet and could not change the agent once providing informed consent. Stenosis (≥50%) of a major intracranial or extracranial artery, or two or more of the vascular risk factors listed in Table 2, is essential as an indicator of a high risk of IS recurrence. Table 2 Inclusion and exclusion criteria Inclusion criteria: 1) Clinical diagnosis of noncardioembolic IS that developed between 8 and 180 days before the start of the protocol treatment 2) A responsible lesion identified on MRI 3) Age 20–85 years 4) Taking clopidogrel or aspirin alone as antiplatelet therapy when providing informed consent 5) At least one of the following (a–c):

vents, adverse drug reactions, and severe or life-threatening hemorrhage as defined in the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) classification, which includes ICH and hemorrhage resulting in substantial hemodynamic compromise requiring treatment 28. Sample size An annual IS recurrence rate of 4% is assumed in the monotherapy group, based on data from patients receiving aspirin monotherapy in the Japanese Aggrenox Stroke prevention versus Aspirin Programme (JASAP) study and Effective Vascular Event REduction after STroke (EVEREST) 9,29. According to Lakatos and Lan's method 30, at least 1688 patients per group would be required for the primary analysis using the log-rank test to detect a 30% relative risk reduction in the DAPT group with 80% power, a two-sided type I error of 0·05, a recruitment period of 2·5 years, and a maximum follow-up period of 3·5 years. Assuming an annual discontinuation rate of 5%, the target number of randomized patients is therefore 2000 per group, or 4000 in total. The 4000 patients will be enrolled in Japan between October 2013 and March 2016.

Inclusion criteria: 1) Clinical diagnosis of noncardioembolic IS that developed between 8 and 180 days before the start of the protocol treatment 2) A responsible lesion identified on MRI 3) Age 20–85 years 4) Taking clopidogrel or aspirin alone as antiplatelet therapy when providing informed consent 5) At least one of the following (a–c): a. ≥50% stenosis of a major intracranial artery (to the level of A2, M2, or P2) b. ≥50% stenosis of an extracranial artery (the common carotid artery, internal carotid artery, vertebral artery, brachiocephalic artery, or subclavian artery) c. Two or more of the following risk factors – Age ≥ 65 years – Diabetes mellitus – Hypertension – Peripheral arterial disease – Chronic kidney disease – History of IS (excluding the index IS for this trial) – History of ischemic heart disease – Smoking (only current smokers) 6) Considered to be able to visit the trial site for ambulatory care throughout the observation period 7) Written informed consent by the patient Exclusion criteria:1) High-risk sources of cardioembolism, according to the TOAST classification 2) Using any anticoagulants 3) Contraindication to MRI examination, such as claustrophobia or implanted pacemaker 4) Scheduled to undergo any surgery, such as percutaneous angioplasty, stent placement, and bypass grafting, during the trial period 5) Drug-eluting coronary stent implanted within one-year 6) History of symptomatic nontraumatic intracranial hemorrhage, any other hemorrhagic disease, bleeding predisposition, or blood clotting disorders 7) History of hypersensitivity to cilostazol 8) Congestive heart failure or uncontrolled angina pectoris

4) Scheduled to undergo any surgery, such as percutaneous angioplasty, stent placement, and bypass grafting, during the trial period 5) Drug-eluting coronary stent implanted within one-year 6) History of symptomatic nontraumatic intracranial hemorrhage, any other hemorrhagic disease, bleeding predisposition, or blood clotting disorders 7) History of hypersensitivity to cilostazol 8) Congestive heart failure or uncontrolled angina pectoris 9) Thrombocytopenia (platelet count, ≤ 100,000/mm3) 10) Severe liver or renal dysfunction 11) Pregnant, breast-feeding, or of child-bearing potential 12) Malignant tumor requiring treatment 13) Aspirin user meeting any of the following criteria: History of hypersensitivity to aspirin or salicylic acid analogues Current peptic ulcer Aspirin-induced asthma or its history 14) Clopidogrel user with a history of hypersensitivity to clopidogrel 15) Participating in any other clinical studies 16) Unsuitable for trial enrollment, as judged by the investigator TOAST, the Trial of Org 10172 in Acute Stroke Treatment. Randomization The data center will randomize the patients to either monotherapy group or DAPT groups using a block-randomization scheme, with the research site as a randomization unit. After providing informed consent, each patient will be identified by a linkable patient identification code, and registered via a web-based registration system, along with information relevant to his or her eligibility. If the patient is considered eligible, the allocated therapy for the patient will be notified through this system. Patients and investigators will be aware of the treatment allocation.

ent identification code, and registered via a web-based registration system, along with information relevant to his or her eligibility. If the patient is considered eligible, the allocated therapy for the patient will be notified through this system. Patients and investigators will be aware of the treatment allocation. Treatments Patients in the monotherapy group will receive oral aspirin (81 or 100 mg) or clopidogrel (50 or 75 mg), once daily. Clopidogrel at 50 mg is approved for older (for example ≥75 years old) or low-weight patients (≤50 kg body weight) in Japan. Patients in the DAPT group will receive oral cilostazol (100 mg, twice daily; the recommended dose in Japan) and either aspirin (81 or 100 mg) or clopidogrel (50 or 75 mg), once daily. To prevent adverse drug reactions such as headache and tachycardia, cilostazol treatment can be started at 100 mg/day, and increased to 200 mg/day within 15 days. If the patient experiences minor cilostazol-related adverse reactions including headache, palpitation, nausea, and vertigo, the trial can be continued using the reduced cilostazol dose of 100 mg/day. Change in the choice of these three antiplatelet medications will not be permitted after informed consent is obtained. Temporary suspension of the trial treatment is permitted during invasive treatment such as surgery, but cannot exceed four-weeks. Use of any concomitant treatment with any antiplatelet or anticoagulant agent (other than trial drugs) will be prohibited.

dications will not be permitted after informed consent is obtained. Temporary suspension of the trial treatment is permitted during invasive treatment such as surgery, but cannot exceed four-weeks. Use of any concomitant treatment with any antiplatelet or anticoagulant agent (other than trial drugs) will be prohibited. Patients will be withdrawn from the trial when they experience any cardiovascular events corresponding to the primary and secondary outcomes, or when the trial treatment has been suspended for more than four consecutive weeks. Primary outcome The primary outcome is a recurrence of IS, with the symptoms lasting for at least 24 h. Secondary outcomes Any stroke (IS, ICH, and SAH) SAH or ICH IS or TIA Death from any cause Stroke, myocardial infarction (MI), or vascular death All vascular events, including stroke, MI, and other vascular events (e.g., aortic dissection; aortic rupture; pulmonary embolism; heart failure, angina pectoris or peripheral artery disease requiring hospitalization; revascularization of coronary artery, aorta, cephalocervical artery, and peripheral arteries). Safety outcomes are adverse events, adverse drug reactions, and severe or life-threatening hemorrhage as defined in the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) classification, which includes ICH and hemorrhage resulting in substantial hemodynamic compromise requiring treatment 28.

ed type I error of 0·05, a recruitment period of 2·5 years, and a maximum follow-up period of 3·5 years. Assuming an annual discontinuation rate of 5%, the target number of randomized patients is therefore 2000 per group, or 4000 in total. The 4000 patients will be enrolled in Japan between October 2013 and March 2016. Statistical analysis Efficacy and safety analyses will be performed on the basis of the intention-to-treat (ITT) population. The treatment groups will be compared using the log-rank test. Cox proportional hazard models will be used to calculate the hazard ratio and its 95% confidence intervals for the DAPT group, compared with the monotherapy group. The hazard ratio will be adjusted for age, sex, type of IS (atherothrombotic or lacunar), and modified Rankin Scale (mRS). Annual recurrence rates will be estimated using the person-year method and the Kaplan–Meier method, and their 95% confidence intervals will be calculated using the approximate Poisson and Greenwood methods. Subgroup analyses will be performed following stratification by age, sex, antiplatelet agents (aspirin, clopidogrel), type of IS (atherothrombotic or lacunar stroke), stenosis of extracranial arteries, mRS, medical history and complications (hypertension, diabetes mellitus, dyslipidemia, coronary heart disease, peripheral arterial disease, chronic kidney disease, IS), smoking status, obesity, and microbleeds. Tests for interaction between the treatment arm and subgroup will be performed using the Cox proportional hazards model. Interim assessment of the safety and efficacy end-points will be conducted when the cumulative person-year exposure has reached 50% of the target. Early termination of the trial will be decided primarily on the basis of the Haybittle-Peto ad hoc method 31,32. Although no termination criteria are specified with regard to the safety end-points, the independent data monitoring committee (IDMC) may recommend early termination based on the incidence of adverse events within the DAPT group.

he trial will be decided primarily on the basis of the Haybittle-Peto ad hoc method 31,32. Although no termination criteria are specified with regard to the safety end-points, the independent data monitoring committee (IDMC) may recommend early termination based on the incidence of adverse events within the DAPT group. Study organization and funding Otsuka Pharmaceutical Co., Ltd. sponsored the trial implementation in the Japan Cardiovascular Research Foundation under an agreement between the two parties. Subject registration, data management, and statistical analysis are delegated to a contract research organization (EPS Corporation). This trial will be conducted at approximately 400 sites under the guidance of the steering committee. The enrolled patients have to pay partly their own costs for drugs prescribed in this study depending on the type of their own health insurance. The principal investigator is Takenori Yamaguchi, who is attached to the National Cerebral and Cardiovascular Center. The final trial protocol was prepared by the protocol committee, and it will be amended, if required. Any event related to the primary and secondary end-points will be reviewed by the event review committee blindly to antiplatelet medications. The statistical analysis committee will be responsible for proposing the randomization scheme and providing expert opinions on the sample size and the analysis plan. It will be also responsible for generating the interim and final analysis plans.

viewed by the event review committee blindly to antiplatelet medications. The statistical analysis committee will be responsible for proposing the randomization scheme and providing expert opinions on the sample size and the analysis plan. It will be also responsible for generating the interim and final analysis plans. The IDMC will recommend trial discontinuation/continuation or protocol amendment based on annual reviews of patient accrual, serious adverse events, incidence of adverse events, and on the interim efficacy analysis comparing the two treatment arms. The individuals who played a critical role in the planning, supervision, and conduct of this trial are listed in the end of this article.

ocol amendment based on annual reviews of patient accrual, serious adverse events, incidence of adverse events, and on the interim efficacy analysis comparing the two treatment arms. The individuals who played a critical role in the planning, supervision, and conduct of this trial are listed in the end of this article. Discussion Prevention of IS recurrence in high-risk patients remains an important clinical problem, since monotherapy with conventional antiplatelet agents is only modestly effective. The CSPS.com will evaluate whether DAPT including cilostazol is similarly safe and more effective in preventing IS recurrence and other vascular outcomes than aspirin or clopidogrel monotherapy in high-risk patients. This trial will resolve several clinical questions regarding preventive antiplatelet therapy for recurrent IS. First, this will be the first trial to explore the efficacy and safety of the combined use of a thienopyridine (clopidogrel) and a phosphodiesterase inhibitor (cilostazol). Second, DAPT including cilostazol may be particularly effective for Asian patients, since they frequently have intracranial atherosclerosis 33 and the TOSS showed the potential of aspirin plus cilostazol for regression of intracranial arterial stenosis 19. Since patients with stenosis (≥50%) of a major intracranial or extracranial artery can be included in the study regardless of existence or number of vascular risk factors, we expect that around half or more of the included patients will have atherothrombotic stroke. Third, DAPT including cilostazol may be safe and fit for long-term use, based on the results of previous clinical and hemostatic studies 20–25. Since Asian ethnicity is a possible risk factor for ICH 34,35, avoidance of ICH during antithrombotic therapy is essential for Asian stroke patients.

Table 1 Baseline characteristics of the patients with stroke and (for comparison) community‐dwelling older subjects without stroke Original study Stroke Study 1 Stroke Study 2 Stroke Study 3 Normal older subjectsc Total sample n = 298a n = 97a n = 264a n = 517 Patients with recent small subcortical infarct on DWI 87b 32 75 n/a Age years (mean ± SD; range) 69 ± 12 64·5 ± 12·1; 36–91 65·9 ± 11·6; 39–96 72·7 ± 0·7; 71–74·2 Male (%) 44 (60) 21 (66) 48 (64) 273 (53) Diabetes (%) 8 (9) 4 (14) 9 (12) 52 (10) Hypertension (%) 41 (47) 20 (61) 60 (80) 244 (47) History of cardiovascular disease (%) 17 (20) 3 (10) 12 (16) 139 (27) Median WMH volume ml (IQR) 15·3 (25·1) 16·2 (28·1) 19·7 (34·0) 6·9 (11·5) Median recent small subcortical infarct volume ml (IQR) 0·9 (1·1) 0·7 (0·7) 0·65 (0·8) n/a Number of lacunes 49 10 28 n/a Median lacune volume ml (IQR) in those with lacunes 0·17 (0·27) 0·22 (0·16) 0·13 (0·22) n/a a Indicates the total number of patients in the original study including patients with nonlacunar stroke. b Six of 87 patients lacked FLAIR images for WMH mapping and registration failed in two patients. c These control subjects had never had a stroke. n/a, not applicable.

Introduction One quarter of patients presenting with ischaemic stroke have ‘lacunar’ or ‘small vessel’ stroke 1. The responsible lesion is an acute lacunar (or recent small subcortical) infarct, which appears hyperintense on diffusion‐weighted magnetic resonance imaging (DWI‐MRI) 2. These recent small subcortical infarcts evolve over a few weeks to months to leave either a small cavity (lacune) or a small noncavitated hyperintense lesion on FLAIR or T2 that resembles a white matter hyperintensity (WMH) 3, 4, 5. Thus, long term, the appearance can be identical to the WMH and lacunes 6 that are commonly seen on MRI in older people scanned for other reasons but who have never experienced any stroke symptoms. New lacunes and WMH may also appear on follow‐up imaging in patients who have not noticed any new symptoms 3. Recent small subcortical ischaemic stroke, WMH, and lacunes are part of the spectrum of small vessel disease (SVD) 7, 8. It is unclear why some SVD lesions cause the patient to present with stroke and so may lead to earlier diagnosis of SVD, whereas others, with very similar imaging appearances, accumulate covertly. We tested the hypothesis that the location of SVD lesions in the brain determines, at least in part, why some cause acute lacunar stroke neurological symptoms, whereas others can develop in the brain covertly, even becoming quite numerous and leading to cognitive impairment, but without apparently causing discrete stroke symptoms that might lead the patient to seek medical attention earlier. Methods Subjects

We tested the hypothesis that the location of SVD lesions in the brain determines, at least in part, why some cause acute lacunar stroke neurological symptoms, whereas others can develop in the brain covertly, even becoming quite numerous and leading to cognitive impairment, but without apparently causing discrete stroke symptoms that might lead the patient to seek medical attention earlier. Methods Subjects Stroke patients: We used data from three prospective consecutive nonoverlapping studies of patients presenting with acute stroke to one regional hospital, who underwent MRI: Study 1 from 2002 to 2005 9, Study 2 from 2005 and 2007 10, and Study 3 from 2010 to 2013. From these studies, we selected all patients who had presented with an acute lacunar stroke clinical syndrome according to the Oxfordshire Community Stroke Project Classification which does not use risk factors to diagnose the stroke syndrome 11 (i.e. pure motor hemiparesis, pure sensory syndrome, sensorimotor syndrome, dysarthria‐clumsy hand syndrome, or ataxic hemiparesis), with a DWI‐MRI‐positive recent small subcortical infarct consistent with this index stroke less than 2‐cm maximum axial diameter and no past history of stroke: 87 patients from Study 1, 32 from Study 2, and 75 from Study 3 (total: 194). Patients with more than one DWI‐MRI positive lesion were excluded. All patients were assessed by experienced stroke physicians, and had baseline demographics, full medical history, examination, and investigations for stroke 9, 10. Written informed consent was obtained from all patients; all studies had Research Ethics Committee approval (Study 1, LREC2001/4/46; Study 2, LREC 2002/8/64; Study 3, LREC 09/S1101/54).

experienced stroke physicians, and had baseline demographics, full medical history, examination, and investigations for stroke 9, 10. Written informed consent was obtained from all patients; all studies had Research Ethics Committee approval (Study 1, LREC2001/4/46; Study 2, LREC 2002/8/64; Study 3, LREC 09/S1101/54). In order to map the corticospinal tracts from diffusion tensor imaging (DTI) and for age‐ and population‐relevant image registration purposes, we used data from 517 community‐dwelling older subjects aged 71–74 years without history or imaging evidence of stroke or incidental finding, the Lothian Birth Cohort 1936 (LBC1936) (www.lothianbirthcohort.ed.ac.uk/) 12. The presence of WMH alone was not excluding. Written informed consent was obtained from all participants. The study was approved by the Lothian (REC 07/MRE00/58) and Scottish Multicentre (MREC/01/0/56) Research Ethics Committees. MR brain image acquisition All MRI data were acquired on the same 1.5T GE Signa Horizon HDxt MRI scanner (General Electric, Milwaukee, WI, USA) using self‐shielding gradients (maximum gradient 33 mT/m) and an eight‐channel phased‐array head coil. The scanner underwent daily quality assurance tests to maintain operational standards. The stroke studies had T1‐weighted (T1‐W) sagittal, T2‐W, Fluid Attenuated Inversion Recovery (FLAIR), gradient echo (GRE), and DWI axial imaging with virtually identical sequences 9, 10; the LBC1936 had T2‐W, fluid attenuation inversion recovery (FLAIR), T2*‐W axial, and T1 volume coronal MR images and DTI 12 (Table S1).

stroke studies had T1‐weighted (T1‐W) sagittal, T2‐W, Fluid Attenuated Inversion Recovery (FLAIR), gradient echo (GRE), and DWI axial imaging with virtually identical sequences 9, 10; the LBC1936 had T2‐W, fluid attenuation inversion recovery (FLAIR), T2*‐W axial, and T1 volume coronal MR images and DTI 12 (Table S1). Image processing All image processing was performed blind to clinical details. We masked the recent small subcortical infarct seen on DWI on the FLAIR image using Analyze 11.0TM (http://www.analyzedirect.com/Analyze/). We used the FLAIR image for consistency of registration with the WMH maps and to avoid acute stroke lesion distortion from registration and blooming artifacts. One trained observer segmented the WMH using a validated semiautomated technique, MCMxxxVI 13 (http://sourceforge.net/projects/bric1936/), a multispectral method that fuses two MRI sequences into the red/green/blue color space and performs minimum variance quantization to separate tissues and lesions. Any segmentation errors were corrected manually. The intraclass correlation coefficient was 0·964, P < 0·01, two‐tailed indicating excellent observer reliability 13. We use the term ‘WMH’ for convenience to include FLAIR hyperintensities in cerebral hemispheric white matter, deep grey matter, and brain stem 2. We identified lacunes (round or ovoid cavities of CSF signal on FLAIR, T1‐W and T2‐W images with diameters ≥3 mm2) semi‐automatically by thresholding the FLAIR images using Analyze 11.0TM.

One trained observer segmented the WMH using a validated semiautomated technique, MCMxxxVI 13 (http://sourceforge.net/projects/bric1936/), a multispectral method that fuses two MRI sequences into the red/green/blue color space and performs minimum variance quantization to separate tissues and lesions. Any segmentation errors were corrected manually. The intraclass correlation coefficient was 0·964, P < 0·01, two‐tailed indicating excellent observer reliability 13. We use the term ‘WMH’ for convenience to include FLAIR hyperintensities in cerebral hemispheric white matter, deep grey matter, and brain stem 2. We identified lacunes (round or ovoid cavities of CSF signal on FLAIR, T1‐W and T2‐W images with diameters ≥3 mm2) semi‐automatically by thresholding the FLAIR images using Analyze 11.0TM. To compare SVD lesion locations, we aligned all brain images in standard space. We used the Mahalanobis distance 14 to derive the population‐representative ‘average’ brain from the LBC1936 cohort, that is, the subject whose intracranial, brain, ventricular, and WMH volumes were all closest to the cohort median (Method S1) 12, 15. We registered the representative LBC1936 subject's T1‐W volume and FLAIR images to standard space using the MNI‐152 isotropic T1‐W 1‐mm brain template and linear rigid body registration (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FLIRT). We registered the individual stroke patients' or LBC1936 subjects' images and lesion masks to the representative brain in standard space using affine linear transformation. We mapped the stroke patients' recent small subcortical infarct, WMH, and lacune masks to the standard space brain. We summed the individual masks to generate group spatial probability density (PD) maps, where the PD in each voxel represents the proportion of the population with a recent small subcortical infarct, WMH, or lacune involving that voxel. Thus, if all patients had a WMH in the same voxel, the PD for WMH in that voxel would be 1; if half the population had a WMH in that voxel, the PD would be 0·5. Note that all patients contributed to the PD of recent small subcortical infarcts and WMH, but only 87 patients contributed to the PD of lacunes; thus, the PD of lacunes is calculated considering only the 87 patients who had them.

at voxel would be 1; if half the population had a WMH in that voxel, the PD would be 0·5. Note that all patients contributed to the PD of recent small subcortical infarcts and WMH, but only 87 patients contributed to the PD of lacunes; thus, the PD of lacunes is calculated considering only the 87 patients who had them. To quantify the anatomical location of the lesion probabilities, we generated a computational template of standard subcortical structures (Fig. S1) using FSL FIRST (http://www.fmrib.ox.ac.uk) for subcortical grey matter and a DTI white matter atlas 16 for standard major white matter tracts. We placed 2‐mm3 regions‐of‐interest on the standard structures and measured the PD in each. We checked the computationally determined PDs using manual region‐of‐interest placement (Supplementary Methods). We also extracted the corticospinal tracts from the LBC1936 Cohort using DTI data processed with probabilistic neighborhood tractography 17 (BedpostX/ProbTrackX algorithm) and TractoR software (http://www.tractor‐mri.org.uk/.17) excluding tracts with failed processing: the corticospinal tracts were reliably segmented in 507/517 (98%) subjects. Finally, to check for recent small subcortical infarct–WMH overlap, we subtracted the former from the latter masks.

BedpostX/ProbTrackX algorithm) and TractoR software (http://www.tractor‐mri.org.uk/.17) excluding tracts with failed processing: the corticospinal tracts were reliably segmented in 507/517 (98%) subjects. Finally, to check for recent small subcortical infarct–WMH overlap, we subtracted the former from the latter masks. Statistical analysis All parameters assessed were nonparametric; therefore, descriptive data are reported as medians and interquartile ranges (IQRs). We compared the lesion distributions using the Friedman's two‐way anova by ranks in IBM SPSS Statistics version 21 (IBM Corporation, Armonk, NY, USA; http://www‐01.ibm.com/software/analytics/spss/products/statistics/). Results Image registration performed well for 192/194 stroke patients (99%), and FLAIR data were incomplete/absent for six patients and corrupted with severe artifacts on one; so, the WMH and recent small subcortical infarct maps use data from 188/194 patients and for lacunes from 187/194 patients. The characteristics of the stroke patients were similar in the three contributing stroke studies (Table 1). The median recent small subcortical infarct volume was 0·79 ml (IQR = 0·9). The median WMH volume was 17·4 ml (IQR = 29·6) in the stroke patients and 6·9 ml (IQR = 11·5) in the LBC1936 subjects (P < 0·001). Table 1 Baseline characteristics of the patients with stroke and (for comparison) community‐dwelling older subjects without stroke Original study Stroke Study 1 Stroke Study 2 Stroke Study 3 Normal older subjectsc Total sample n = 298a n = 97a n = 264a

Results Image registration performed well for 192/194 stroke patients (99%), and FLAIR data were incomplete/absent for six patients and corrupted with severe artifacts on one; so, the WMH and recent small subcortical infarct maps use data from 188/194 patients and for lacunes from 187/194 patients. The characteristics of the stroke patients were similar in the three contributing stroke studies (Table 1). The median recent small subcortical infarct volume was 0·79 ml (IQR = 0·9). The median WMH volume was 17·4 ml (IQR = 29·6) in the stroke patients and 6·9 ml (IQR = 11·5) in the LBC1936 subjects (P < 0·001). Table 1 Baseline characteristics of the patients with stroke and (for comparison) community‐dwelling older subjects without stroke Original study Stroke Study 1 Stroke Study 2 Stroke Study 3 Normal older subjectsc Total sample n = 298a n = 97a n = 264a n = 517 Patients with recent small subcortical infarct on DWI 87b 32 75 n/a Age years (mean ± SD; range) 69 ± 12 64·5 ± 12·1; 36–91 65·9 ± 11·6; 39–96 72·7 ± 0·7; 71–74·2

Male (%) 44 (60) 21 (66) 48 (64) 273 (53) Diabetes (%) 8 (9) 4 (14) 9 (12) 52 (10) Hypertension (%) 41 (47) 20 (61) 60 (80) 244 (47) History of cardiovascular disease (%) 17 (20) 3 (10) 12 (16) 139 (27) Median WMH volume ml (IQR) 15·3 (25·1) 16·2 (28·1) 19·7 (34·0) 6·9 (11·5) Median recent small subcortical infarct volume ml (IQR) 0·9 (1·1) 0·7 (0·7) 0·65 (0·8) n/a Number of lacunes 49 10 28 n/a Median lacune volume ml (IQR) in those with lacunes 0·17 (0·27) 0·22 (0·16) 0·13 (0·22) n/a a Indicates the total number of patients in the original study including patients with nonlacunar stroke. b Six of 87 patients lacked FLAIR images for WMH mapping and registration failed in two patients. c These control subjects had never had a stroke. n/a, not applicable. 2015 World Stroke OrganizationThe commonest location of the recent small subcortical infarcts was in the superior aspect of the mid‐posterior limb of the internal capsule (PD 0·2/mm3) (Fig. 1, Table 2). The probability progressively decreased from there medially to the lateral thalamus (PD 0·17/mm3), superiorly to the corona radiata adjacent to the body of the lateral ventricle (PD 0·15/mm3), laterally to the medial–posterior aspect of lentiform nucleus (PD 0·08/mm3), and inferiorly to the midbrain/superior brainstem (PD 0·05–0·07/mm3). Comparison with the DTI‐derived corticospinal tract map (Fig. 1) showed that the recent small subcortical infarcts occurred almost exclusively in the motor or sensory corticospinal tracts.

e medial–posterior aspect of lentiform nucleus (PD 0·08/mm3), and inferiorly to the midbrain/superior brainstem (PD 0·05–0·07/mm3). Comparison with the DTI‐derived corticospinal tract map (Fig. 1) showed that the recent small subcortical infarcts occurred almost exclusively in the motor or sensory corticospinal tracts. Figure 1 View in the axial, coronal, and sagittal planes to compare the distribution of recent small subcortical infarcts (left) and lacunes (middle) in stroke patients. The location of the corticospinal tracts as derived from tractography (right) is shown for comparison. Also, see Fig. S2. figure2015 World Stroke OrganizationTable 2 Probability of finding recent small subcortical infarcts, WMH, or lacunes in each brain region in patients with stroke, and of finding WMH in the community‐dwelling older subjects without stroke (for comparison with the WMH in stroke patients) Brain region † Probability density of voxels in each region being affected by the SVD lesions Patients with stroke Normal older subjects Recent small subcortical infarct ‡

figure2015 World Stroke OrganizationTable 2 Probability of finding recent small subcortical infarcts, WMH, or lacunes in each brain region in patients with stroke, and of finding WMH in the community‐dwelling older subjects without stroke (for comparison with the WMH in stroke patients) Brain region † Probability density of voxels in each region being affected by the SVD lesions Patients with stroke Normal older subjects Recent small subcortical infarct ‡ Lacunes WMH WMH Right Left Right Left Right Left Right Left Caudate 0·05 0·12 0·01 0·01 0·56 0·54 0·49 0·49 Putamen 0·16 0·15 0·03 0·05 0·18 0·22 0·09 0·11 Globus pallidus 0·09 0·11 0·02 0·04 0·16 0·12 0·10 0·08 Thalamus 0·14 0·15 0·005 0·005 0·20 0·26 0·18 0·26 Hippocampus 0·005 0·005 0·01 0·01 0·09 0·14 0·14 0·17 Superior aspect of posterior limb of internal capsule 1 0·2 0·2 0·05 0·05 0·1 0·15 0·01 0·01 Mid posterior limb internal capsule and adjacent lateral thalamus and medial lentiform nucleus 2 0·17 0·17 0·04 0·04 0·14 0·14 0·07 0·07 Corona radiata lateral to body of lateral ventricle 3 0·15 0·15 0·005 0·005 0·39 0·43 0·11 0·11 Corona radiata superior to lateral ventricle 4 0·14 0·14 0·005 0·005 0·57 0·53 0·57 0·57 Medial lentiform nucleus 5 0·06 0·08 0·005 0·005 0·02 0·02 0·01 0·01 Centrum semiovale 0·02 0·02 0·005 0·005 0·62 0·62 0·6 0·6 Retro‐lentiform part of the internal capsules 0·02 0·02 0·04 0·04 0·26 0·26 0·12 0·12 Anterior limb internal capsule 6 0 0·05 0·005 0 0·05 0·05 0·02 0·07 Adjacent to anterior horns of the lateral ventricles 0 0 0·005 0·005 0·77 0·77 0·70 0·70 Optic radiations 0 0 0 0·005 0·55 0·49 0·47 0·38 External capsules 0·1 0·1 0·05 0·05 0·25 0·25 0·12 0·12 Midbrain/superior brainstem* 7 0·07 0·005 0·02 0·01 Brainstem* 0·02 0·005 0·14 0·03 The probability density refers to the proportion of voxels in the particular region‐of‐interest that had a lesion in the stroke patients or (for WMH) in the community‐dwelling older subjects without stroke. For example, a probability of 0·005 (or 1/188) indicates that only one subject out of the 188 stroke patients had a voxel affected by a lesion in this region. These are also adjusted by the number of individuals, within the population, that contributed data to the probability map. For example, whereas the whole stroke population had recent small subcortical infarct lesion(s), only 87 had lacunes. The probability density of lacunes is calculated considering only the 87 patients who had them. *Structure not separated into right and left due to small size.

t contributed data to the probability map. For example, whereas the whole stroke population had recent small subcortical infarct lesion(s), only 87 had lacunes. The probability density of lacunes is calculated considering only the 87 patients who had them. *Structure not separated into right and left due to small size. †Numbers in italics refer to region numbers in Fig. S1. ‡As recent small subcortical infarcts were largely confined to the primary motor/sensory pathways, their median probability value across all subcortical regions was 0, and their mean probability value and standard deviation were of the orders of 10−6 and 10−4, respectively. 2015 World Stroke OrganizationIn patients with stroke, the WMH were distributed symmetrically between the hemispheres in white matter, in external capsules, basal ganglia, and brainstem (Fig. 2) but were infrequent where the recent small subcortical infarcts were most frequent (anova, P < 0·01; Table 2). Subtraction of the recent small subcortical infarcts from the WMH map confirmed that there was little overlap in location (Fig. 3). The WMH anatomical distribution was similar in the stroke patients and the LBC1936 subjects (Fig. 2, Table 2), although WMHs were more extensive in the stroke patients (ANOVA, P < 0·008; also Fig. S3). Figure 2 WMH distribution of 188 stroke patients (left) and (for comparison) in 517 similarly aged subjects without stroke (right) in axial, coronal, and sagittal planes. Burden of WMH is less in stroke‐free subjects although the distribution is similar. See also Fig. S3.

2015 World Stroke OrganizationIn patients with stroke, the WMH were distributed symmetrically between the hemispheres in white matter, in external capsules, basal ganglia, and brainstem (Fig. 2) but were infrequent where the recent small subcortical infarcts were most frequent (anova, P < 0·01; Table 2). Subtraction of the recent small subcortical infarcts from the WMH map confirmed that there was little overlap in location (Fig. 3). The WMH anatomical distribution was similar in the stroke patients and the LBC1936 subjects (Fig. 2, Table 2), although WMHs were more extensive in the stroke patients (ANOVA, P < 0·008; also Fig. S3). Figure 2 WMH distribution of 188 stroke patients (left) and (for comparison) in 517 similarly aged subjects without stroke (right) in axial, coronal, and sagittal planes. Burden of WMH is less in stroke‐free subjects although the distribution is similar. See also Fig. S3. figure2015 World Stroke OrganizationFigure 3 Subtraction image to determine the difference in distribution of the recent small subcortical infarcts with respect to WMH (left) and lacunes (right) in stroke patients. Red–yellow indicates WMH (left hand image) or lacunes (right hand image); green–blue regions indicate the recent small subcortical infarcts that remain after subtraction from WMH and lacune maps (both sides). Images show a little overlap in location between the clinically symptomatic vs. covert lesions.

troke patients. Red–yellow indicates WMH (left hand image) or lacunes (right hand image); green–blue regions indicate the recent small subcortical infarcts that remain after subtraction from WMH and lacune maps (both sides). Images show a little overlap in location between the clinically symptomatic vs. covert lesions. figure2015 World Stroke OrganizationLacunes were identified in 87/187 stroke patients (Figs 1 and S2), median volume 0·14 ml (IQR 0·19 ml, range 0·03–0·8 ml; Table 1). Lacunes were most frequent in the posterior external capsules (PD 0·05/mm3) and lateral parts of posterior lentiform nuclei (PD 0·04/mm3), that is, lateral to the most frequent location of the recent small subcortical infarcts. Lacunes were also present in the caudate (PD 0·01/mm3), hippocampus (PD 0·01/mm3), and in parts of thalamus (PD 0·005/mm3) where no acute symptomatic lesions occurred (difference in distribution of acute SSI and lacunes ANOVA, P < 0·03; Fig. 3).

al to the most frequent location of the recent small subcortical infarcts. Lacunes were also present in the caudate (PD 0·01/mm3), hippocampus (PD 0·01/mm3), and in parts of thalamus (PD 0·005/mm3) where no acute symptomatic lesions occurred (difference in distribution of acute SSI and lacunes ANOVA, P < 0·03; Fig. 3). Discussion The distribution of the three SVD lesion types indicates that the lesions that present with acute stroke symptoms are located mostly in the corticospinal tracts, particularly in the posterior limb of the internal capsule. Here, perhaps even a small lesion can cause symptoms because interruption of primary motor or sensory function where the tracts are tightly packed would readily be noticed by the patient. In contrast, the WMH and lacunes, which had never caused discrete symptoms, were more common in other subcortical regions where interruption of some fibers might be less likely to be noticed by the patient. This could explain why recent small subcortical infarcts come rapidly to medical attention through causing a ‘stroke’, whereas many WMH and lacunes can accumulate over time without individually causing enough symptoms for the patient to seek medical attention until sufficient brain damage has occurred to trigger presentation with cognitive or mobility problems. Alternatively, if seeking medical attention, their symptoms may not be recognized as of importance 18, 19.

n accumulate over time without individually causing enough symptoms for the patient to seek medical attention until sufficient brain damage has occurred to trigger presentation with cognitive or mobility problems. Alternatively, if seeking medical attention, their symptoms may not be recognized as of importance 18, 19. We included almost 200 patients with acute lacunar stroke symptoms and a DWI‐visible acute lesion to ensure correct identification of the index stroke, all recruited prospectively. We used voxel‐based region‐of‐interest sampling on a purpose‐designed standard anatomical template that represents subcortical white and grey matter. We used a study‐specific population‐relevant brain template from subjects scanned on the same scanner to optimize image registration and minimize lesion distortion. However, we did not examine lesion distribution of individual lacunar stroke syndromes as our main purpose was to compare SVD lesions that presented with stroke to those that did not – such an analysis of subtypes of lacunar syndrome would require a larger sample. We did not find any DWI‐positive lesions that were asymptomatic as have been seen in some patients with recent small subcortical infarcts 20 or with recent cerebral hemorrhage 21. However, in our experience and the literature, these asymptomatic DWI positive small subcortical infarcts are infrequent, perhaps present in less than 5% of patients, and therefore, it will take some time to accumulate enough subjects, even in multicentre studies, to be able to plot their distribution reliably. Serial longitudinal studies with detailed symptom ascertainment and repeat scanning will be required to examine symptomatology as new lesions accumulate and in individual lacunar subsyndromes. Nor did we have long‐term follow‐up images on our patients to see if any lacunes that resulted from late‐stage cavitation of any of the recent small subcortical infarcts were of different size to the asymptomatic lacunes, as another possible explanation for a difference in symptoms. However, C Miller Fisher and others did not find that size of lacunes measured at post mortem was related to symptoms 22, 23. Further long‐term follow‐up is required to address this question.

tical infarcts were of different size to the asymptomatic lacunes, as another possible explanation for a difference in symptoms. However, C Miller Fisher and others did not find that size of lacunes measured at post mortem was related to symptoms 22, 23. Further long‐term follow‐up is required to address this question. Our findings support: the findings of C Miller Fisher's meticulous dissections suggesting that lesions in the internal capsule were more likely to have caused symptoms in life than those in other brain regions 22, and by other limited post mortem data 23; patients with CADASIL where asymptomatic lacunes affected the thalamus and lentiform nucleus 24; and a previous study using T2‐weighted MR showing that lesions associated with stroke symptoms were more often in the internal capsule posterior limb and adjacent basal ganglia than were similar‐appearing lesions in patients without stroke 25. The different distribution of the lacunes and WMH in these sporadic SVD patients is also not dissimilar to differences in lacunes and WMH distributions noted previously in patients with CADASIL 26. WMH volume of the stroke patients was two to three times that of the stroke‐free LBC1936 subjects, despite the similar median age, perhaps reflecting the higher prevalence of hypertension in stroke patients. A higher WMH burden in stroke patients than in subjects without stroke has also been noted previously 27, 28.

6. WMH volume of the stroke patients was two to three times that of the stroke‐free LBC1936 subjects, despite the similar median age, perhaps reflecting the higher prevalence of hypertension in stroke patients. A higher WMH burden in stroke patients than in subjects without stroke has also been noted previously 27, 28. Other factors may explain why some SVD lesions cause stroke symptoms while others develop covertly. Speed of onset (e.g. WMH) may develop more gradually than recent small subcortical infarcts. Severity of tissue damage (e.g. recent small subcortical infarcts) may be more destructive to the tissue than many WMH, as suggested by the former's higher rate of cavity formation 3. Multidirectionality of fibres (e.g. centrum semiovale white matter fibers) is more multidirectional than in the internal capsule, possibly ‘diluting’ the impact of a small lesion across several functions. Size (e.g. WMH) may be smaller than recent small subcortical infarcts, although the range of recent small subcortical infarcts includes lesions of 5 mm or less 29. Recent reports indicate that WMHs are associated with awareness of cognitive decline 19 and subtle physical symptoms 18, 30, suggesting that they are not as ‘silent’ as previously thought. Differences in size of lacune and symptoms require evaluation in further studies, as just discussed. Meanwhile, our anatomical analysis provides further evidence that presentation with an acute lacunar stroke heralds a diffuse brain disease 7, 8 and provides methods to map subcortical structures for future detailed longitudinal studies.

e of lacune and symptoms require evaluation in further studies, as just discussed. Meanwhile, our anatomical analysis provides further evidence that presentation with an acute lacunar stroke heralds a diffuse brain disease 7, 8 and provides methods to map subcortical structures for future detailed longitudinal studies. Supporting information Table S1. MR imaging sequence details for the cohorts analysed in this work. Fig. S1. Representative brain showing the subcortical grey matter structures, indicated in orange and yellow in the left and right hemispheres respectively, and the standard sampling points (numbered red dots) used to measure the PD of recent small subcortical infarcts. Fig. S2. Slice‐by‐slice distribution of recent small subcortical infarcts (left) and lacunes (right) in patients with stroke. Fig. S3. WMH in patients with stroke (left) and in 517 community‐dwelling older subjects without stroke (right). Method S1. A comparison of brain location of acute symptomatic infarcts vs. ‘silent’ small vessel lesions. Click here for additional data file. Acknowledgements We thank participants in all studies that contributed data to this analysis, Caroline Jackson for coordination of the ESS, Catherine Murray for recruitment of the participants in the LBC1936 Study, and the radiographers and staff at the Brain Research Imaging Centre.

Introduction Aspirin has been shown to be effective for secondary prevention post myocardial infarction and stroke,1 but there is uncertainty about its role in primary prevention populations, including those with cardiovascular risk factors (e.g. diabetes mellitus).2 While a large number of the general population elect to take a daily aspirin for primary prevention of stroke,3 there is disagreement in current guidelines about the use of aspirin for primary prevention.4 The guidelines are based on interpretation of previous meta-analytic findings, which report a modest benefit for selected high risk patients, mostly related to a small absolute reduction in non-fatal myocardial infarctions in elderly patients, which is offset by an increased risk of major gastrointestinal and hemorrhagic stroke.5 The American Heart Association and American Stroke Association's guidelines on primary prevention of stroke give a moderate IIa recommendation for aspirin's use in primary prevention of stroke in high risk groups.6 A survey of U.S. adults between 45 and 75 years showed that, of the people taking aspirin, 81% were taking aspirin for primary prevention of cardiovascular disease. Two-thirds of these aspirin users reported stroke prevention as the primary indication.3

spirin's use in primary prevention of stroke in high risk groups.6 A survey of U.S. adults between 45 and 75 years showed that, of the people taking aspirin, 81% were taking aspirin for primary prevention of cardiovascular disease. Two-thirds of these aspirin users reported stroke prevention as the primary indication.3 Since the publication of the United States Preventive Services Task Force meta-analysis and recommendations for aspirin in primary prevention,4 additional large randomized control trials focusing on older adults,7,8 diabetes,9 and moderate cardiovascular risk10 have reported their results. In this meta-analysis of randomized controlled trials evaluating aspirin for primary prevention of cardiovascular disease, we sought to determine the summary effect of aspirin on primary prevention of stroke and other cardiovascular outcomes. Methods Cumulative meta-analysis To reduce research waste,11 we (CJ and SR) extracted data from two previous meta-analyses: one of randomized controlled trials of aspirin in primary prevention of cardiovascular disease12 and the other of bleeding risks with aspirin for primary prevention of cardiovascular disease.5 We considered these meta-analyses of sufficiently high quality to avoid the need to repeat them. We limited our search to dates not included in these reviews (2015–2018). We (CJ and RM) repeated primary data extraction independently for all papers to confirm accuracy and resolved any inconsistencies by consensus (CJ, RM and SR).

red these meta-analyses of sufficiently high quality to avoid the need to repeat them. We limited our search to dates not included in these reviews (2015–2018). We (CJ and RM) repeated primary data extraction independently for all papers to confirm accuracy and resolved any inconsistencies by consensus (CJ, RM and SR). Selection criteria We performed a systematic review according to published guidelines from the Cochrane Collaboration13 and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).14 We selected randomized controlled trials of aspirin for primary prevention of cardiovascular disease. We included all trials with: participants older than eighteen years, evaluated aspirin therapy versus placebo, randomized controlled trials, blinded outcome assessment, no history of cardiovascular disease, greater than one-year follow-up and published as full reports. We did not exclude trials based on neuroimaging requirements for outcome assessment (stroke). We limited our search to published, peer-reviewed studies in English. The search was not limited to a patient group or aspirin dose.

ory of cardiovascular disease, greater than one-year follow-up and published as full reports. We did not exclude trials based on neuroimaging requirements for outcome assessment (stroke). We limited our search to published, peer-reviewed studies in English. The search was not limited to a patient group or aspirin dose. Search strategy We developed a search strategy for the PUBMED and EMBASE databases (Supplementary Figures I and II). The databases were searched from January 2015 to November 2018. Two reviewers (CJ and RM) independently screened titles and abstracts using the Rayann web application.15 Full texts were sourced for relevant articles. Inclusion criteria were assessed independently, and the final list was agreed by consensus. We also screened the reference list of similar review articles and earlier published meta-analyses obtained in our search. The protocol for the systematic review was registered on PROSPERO, the international prospective register of systematic reviews.

assessed independently, and the final list was agreed by consensus. We also screened the reference list of similar review articles and earlier published meta-analyses obtained in our search. The protocol for the systematic review was registered on PROSPERO, the international prospective register of systematic reviews. Data extraction We used a standardized data collection form (available on request). For each study, we extracted the title, year of publication, aspirin dose, active and control numbers, non-fatal stroke and hemorrhagic stroke. We did not pre-specify a definition for stroke. Instead, we used the definition reported by each individual paper. We included both ischemic and hemorrhagic stroke in our definition of non-fatal stroke. We also extracted non-fatal myocardial infarction, all-cause mortality, cardiovascular mortality and major gastrointestinal bleeding. Reviewers (CJ, RM and SR) independently extracted data, compared for inconsistencies, and merged into a final data set.

emic and hemorrhagic stroke in our definition of non-fatal stroke. We also extracted non-fatal myocardial infarction, all-cause mortality, cardiovascular mortality and major gastrointestinal bleeding. Reviewers (CJ, RM and SR) independently extracted data, compared for inconsistencies, and merged into a final data set. Data synthesis and analysis We present a descriptive analysis of each individual trial (Table 1). We calculated odds ratio (OR) and 95% confidence intervals from individual studies. Weighted pooled treatment effects were calculated using a random effects model. The variability across studies due to heterogeneity was estimated with the I2 statistic. We calculated the incident density rate for each study outcome by dividing the event totals by the person years of follow-up. We meta-analyzed the incident density rates to obtain pooled estimates and their 95% confidence intervals. Net-benefit was calculated as the risk difference between the benefits of all-cause mortality and non-fatal events (myocardial infarction and stroke), minus the harm of increasing major gastrointestinal bleeding and hemorrhagic stroke. Statistical analysis was performed using the Metafor package16 on R Statistical Software (V3.4.3). Table 1. Study description, stroke and other cardiovascular events.

e mortality and non-fatal events (myocardial infarction and stroke), minus the harm of increasing major gastrointestinal bleeding and hemorrhagic stroke. Statistical analysis was performed using the Metafor package16 on R Statistical Software (V3.4.3). Table 1. Study description, stroke and other cardiovascular events. Study Name Dose (mg/day) Followup (years) Total Subjects Mean Age (years) Female (%) DM (%) Non-fatal Stroke (Aspirin) Non-fatal Stroke (Control) Haem. Stroke (Aspirin) Haem. Stroke (Control) All-cause Mortality (Aspirin) All-cause Mortality (Control) CV Mortality (Aspirin) CV Mortality (Control) Non-fatal Myocardial Infarction (Aspirin) Non-fatal Myocardial Infarction (Control) Major GI Bleeding (Aspirin) Major GI Bleeding (Control) Definition of GI Bleed BMD, 27 500 6 5139 NR 0 2 61 27 13 6 270 151 119 59 80 41 3 3 Fatal GI bleeds PHS I, 26 162.5 5 22071 NR 0 – 110 92 23 12 217 227 81 83 129 213 49 28 Required major transfusion, death HOT, 21 75 3.8 18790 61.5 47 8 NR NR 14 15 284 305 133 140 68 113 77 37 Major and Fatal TPT, 22 75 6.8 2540 57.5 0 NR 16 25 2 0 113 110 49 49 47 73 6 2 Required transfusion and/or surgery PPP, 23 100 3.6 4495 64.4 57.46 16.5 15 18 2 3 62 78 17 31 15 22 17 5 GI bleeding WHS, 20 50 10.1 39876 54.6 100 2.6 198 244 51 41 609 642 120 126 184 181 129 94 Required transfusion or caused death JPAD, 24 100 4.37 2539 64.5 45.37 100 27 27 5 3 34 38 1 10 12 9 4 0 Required a transfusion JPPP, 25 100 5 14464 70.5 57.7 34 – – 28 15 297 303 58 57 20 38 103 31 Serious bleeds requiring transfusion or hospitalisaiton ARRIVE, 10 100 5 12546 63.9 29.55 0 – – 8 11 160 161 38 39 88 98 4 2 ASCEND, 9 100 7.4 15480 63.2 37.4 100 202 229 55 45 748 792 197 217 191 195 137 101 Transfusion or hospitalisation to control bleeding ASPREE, 7 100 4.7 19114 74 56 10.8 195 203 42 32 558 494 78 81 156 168 162 102 Bleeding that led to transfusion, hospitalization, surgery, or death

55 0 – – 8 11 160 161 38 39 88 98 4 2 ASCEND, 9 100 7.4 15480 63.2 37.4 100 202 229 55 45 748 792 197 217 191 195 137 101 Transfusion or hospitalisation to control bleeding ASPREE, 7 100 4.7 19114 74 56 10.8 195 203 42 32 558 494 78 81 156 168 162 102 Bleeding that led to transfusion, hospitalization, surgery, or death Results In total, 11 randomized controlled trials were eligible that recruited 157,054 participants and reported 1920 non-fatal strokes and 426 hemorrhagic strokes. Additionally, there were 2141 non-fatal myocardial infarctions, 6653 deaths, 1783 cardiovascular deaths and 1096 major gastrointestinal bleeds. Our updated search results found 1841 studies from PUBMED and 1807 studies from EMBASE, 613 duplicate studies were removed, 3032 studies were excluded after title and abstract screening, leaving 3 studies for inclusion (Supplementary Figure III). Three of the trials included in previous meta-analysis were excluded due to prior cardiovascular disease, two trials with participants having peripheral vascular disease17,18 and one trial with nearly half having previous cardiovascular disease.19 Of the included trials, nine were trials of aspirin at a dose of 100 mg or less.7,9,10,20–25 The mean follow-up across all studies was 5.58 years. The mean age was 63.79 years. 39.13% of the participants were female.

ripheral vascular disease17,18 and one trial with nearly half having previous cardiovascular disease.19 Of the included trials, nine were trials of aspirin at a dose of 100 mg or less.7,9,10,20–25 The mean follow-up across all studies was 5.58 years. The mean age was 63.79 years. 39.13% of the participants were female. Non-fatal stroke Nine trials reported non-fatal stroke.7,9,19,20,22–27 Non-fatal stroke occurred in 941 (1.19%) patients in the aspirin group and 979 (1.26%) patients in the control group. Aspirin use for primary cardiovascular prevention was not associated with a significant decrease in non-fatal stroke (odds ratio, 0.94; 95% CI, 0.85 to 1.04) (Figure 1). The P value for heterogeneity was 0.38, I2 = 16.2%, Q = 8.53, and degrees of freedom = 8. A sensitivity analysis including only studies with imaging requirement for diagnosis of stroke7,20,22,23,25 was also non-significant for aspirin benefit on non-fatal stroke in primary prevention (odds ratio, 0.90; 95% CI, 0.79 to 1.02). Figure 1. Aspirin for primary cardiovascular prevention and benefit for non-fatal stroke. Forest plot for non-fatal stroke. Forest plot showing the effect of aspirin therapy on non-fatal stroke. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect.

ing the effect of aspirin therapy on non-fatal stroke. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect. Hemorrhagic stroke Eleven trials reported hemorrhagic stroke.7,9,10,20–27 Hemorrhagic stroke occurred in 243 (0.30%) patients in the aspirin group and 183 (0.24%) patients in the control group. Aspirin use for primary cardiovascular prevention was associated with a significant increase in hemorrhagic stroke (odds ratio, 1.29; 95% CI, 1.06 to 1.56) (Figure 2). The P value for heterogeneity was 0.77, I2 = 0.0%, Q = 6.52, and degrees of freedom = 10. Figure 2. Aspirin for primary cardiovascular prevention and benefit for hemorrhagic stroke. Forest plot for hemorrhagic stroke. Forest plot showing the effect of aspirin therapy on hemorrhagic stroke. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect.

g the effect of aspirin therapy on hemorrhagic stroke. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect. All stroke Eleven trials reported all stroke.7,9,10,20–27 All stroke occurred in 1277 (1.61%) patients in the aspirin group and 1297 (1.67%) patients in the control group. Aspirin use for primary cardiovascular prevention was not associated with a significant decrease in all stroke (odds ratio, 0.95; 95% CI, 0.88 to 1.03). The P value for heterogeneity was 0.47, I2 = 2.4%, Q = 9.67, and degrees of freedom = 10.

patients in the aspirin group and 1297 (1.67%) patients in the control group. Aspirin use for primary cardiovascular prevention was not associated with a significant decrease in all stroke (odds ratio, 0.95; 95% CI, 0.88 to 1.03). The P value for heterogeneity was 0.47, I2 = 2.4%, Q = 9.67, and degrees of freedom = 10. Non-fatal myocardial infarction Eleven trials reported non-fatal myocardial infarction.7,9,10,19–27 Non-fatal myocardial infarction occurred in 990 (1.25%) patients in the aspirin group and 1151 (1.48%) patients in the control group. Aspirin use for primary cardiovascular prevention was associated with a significant decrease in non-fatal myocardial infarction (odds ratio, 0.80; 95% CI, 0.69 to 0.94) (Figure 3). The P value for heterogeneity was 0.00, I2 = 62.9%, Q = 27.34, and degrees of freedom = 10. Figure 3. Aspirin for primary cardiovascular prevention and benefit for non-fatal myocardial infarction. Forest plot for non-fatal myocardial infarction. Forest plot showing the effect of aspirin therapy on non-fatal myocardial infarction. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect.

of aspirin therapy on non-fatal myocardial infarction. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect. All-cause mortality Eleven trials reported all-cause mortality.7–10,19–27 All-cause mortality occurred in 3352 (4.23%) patients in the aspirin group and 3301 (4.25%) patients in the control group. Aspirin use for primary cardiovascular prevention was not associated with a significant decrease in all-cause mortality (odds ratio, 0.97; 95% CI, 0.92 to 1.03) (Figure 4). The P value for heterogeneity was 0.44, I2 = 20.0%, Q = 10.02, and degrees of freedom = 10. Figure 4. Aspirin for primary cardiovascular prevention and benefit for all-cause mortality. Forest plot for all-cause mortality. Forest plot showing the effect of aspirin therapy on all-cause mortality. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect.

the effect of aspirin therapy on all-cause mortality. The squares and bars represent the mean values and 95% confidence intervals of the effect sizes, while the size of the squares reflects the weight of the studies. The combined effects appear as diamonds and the vertical dashed line represents the line of no effect. Cardiovascular mortality Eleven trials reported cardiovascular mortality.7–10,17–27 Cardiovascular mortality occurred in 891 (1.12%) patients in the aspirin group and 892 (1.15%) patients in the control group. Aspirin use for primary cardiovascular prevention was not associated with a significant decrease in cardiovascular mortality (odds ratio, 0.94; 95% CI, 0.85 to 1.03) (Supplementary Figure IV). The P value for heterogeneity was 0.60, I2 = 0.0%, Q = 8.29, and degrees of freedom = 10. Major gastrointestinal bleeding Eleven trials reported major gastrointestinal bleeding.7,9,10,20–27 Major gastrointestinal bleeding occurred in 691 (0.87%) patients in the aspirin group and 405 (0.52%) patients in the control group. Aspirin use for primary cardiovascular prevention was associated with a significant increase in major gastrointestinal bleeding (odds ratio, 1.83; 95% CI, 1.43 to 2.35) (Supplementary Figure V). The P value for heterogeneity was 0.01, I2 = 61.5%, Q = 23.55, and degrees of freedom = 10.

%) patients in the control group. Aspirin use for primary cardiovascular prevention was associated with a significant increase in major gastrointestinal bleeding (odds ratio, 1.83; 95% CI, 1.43 to 2.35) (Supplementary Figure V). The P value for heterogeneity was 0.01, I2 = 61.5%, Q = 23.55, and degrees of freedom = 10. Net clinical effect Table 2 reports the pooled estimates with confidence interval for population density incidence rates for non-fatal stroke and hemorrhagic stroke in the aspirin group, control group and the difference in incidence rates between the two groups. Non-fatal myocardial infarction, non-fatal stroke and major gastrointestinal bleeding are also reported. To determine a net clinical effect, we calculated the risk difference between benefit and harm, benefit from reduction in non-fatal stroke (0.16 per 1000 person years; 95% CI, −1.07 to 1.39) and non-fatal myocardial infarction (0.54 per 1000 person years; 95% CI, −0.83 to 1.91) and harm from increase in major gastrointestinal bleeding (−0.49 per 1000 person years; 95% CI, −1.23 to 0.25) and increase in hemorrhagic stroke (−0.12 per 1000 person years; 95% CI, −0.50 to 0.26). The overall net-benefit was non-significant at 0.09 per 1000 person years (95% CI, −1.93 to 2.11). Table 2. Incidence rates per 1000 person years

major gastrointestinal bleeding (−0.49 per 1000 person years; 95% CI, −1.23 to 0.25) and increase in hemorrhagic stroke (−0.12 per 1000 person years; 95% CI, −0.50 to 0.26). The overall net-benefit was non-significant at 0.09 per 1000 person years (95% CI, −1.93 to 2.11). Table 2. Incidence rates per 1000 person years Aspirin Control Difference Estimate 95% CI Estimate 95% CI Estimate 95% CI NNT/NNH Stroke Non-fatal stroke 2.59 1.84 to 3.63 2.75 2.01 to 3.77 0.16 −1.07 to 1.39 6250.00 Hemorrhagic stroke 0.51 0.36 to 0.72 0.39 0.28 to 0.54 −0.12 −0.50 to 0.26 −8333.33 Other outcomes Non-fatal myocardial  infarction 2.24 1.53 to 3.29 2.78 1.89 to 4.08 0.54 −0.83 to 1.91 1851.85 Cardiovascular mortality 2.01 1.25 to 3.23 2.33 1.53 to 3.56 0.32 −1.05 to 1.69 3125.00 All-cause mortality 7.67 5.64 to 10.42 8.02 5.94 to 10.84 0.35 −3.02 to 3.72 2857.14 Major gastrointestinal  bleeding 1.04 0.55 to 1.96 0.55 0.29 to 1.02 −0.49 −1.23 to 0.25 −2040.82 Note: The incidence rates for 1000 person years for non-fatal stroke, hemorrhagic stroke, non-fatal myocardial infarction, cardiovascular mortality, all-cause mortality and major gastrointestinal bleeding are reported. The incidence rates are reported for the aspirin group, control group and the difference between the two groups. The incidence rate was calculated by dividing the event totals by the person years of follow-up. CI: confidence interval; NNT: number needed to treat; NNH: number needed to harm.

Aspirin Control Difference Estimate 95% CI Estimate 95% CI Estimate 95% CI NNT/NNH Stroke Non-fatal stroke 2.59 1.84 to 3.63 2.75 2.01 to 3.77 0.16 −1.07 to 1.39 6250.00 Hemorrhagic stroke 0.51 0.36 to 0.72 0.39 0.28 to 0.54 −0.12 −0.50 to 0.26 −8333.33 Other outcomes Non-fatal myocardial  infarction 2.24 1.53 to 3.29 2.78 1.89 to 4.08 0.54 −0.83 to 1.91 1851.85 Cardiovascular mortality 2.01 1.25 to 3.23 2.33 1.53 to 3.56 0.32 −1.05 to 1.69 3125.00 All-cause mortality 7.67 5.64 to 10.42 8.02 5.94 to 10.84 0.35 −3.02 to 3.72 2857.14 Major gastrointestinal  bleeding 1.04 0.55 to 1.96 0.55 0.29 to 1.02 −0.49 −1.23 to 0.25 −2040.82 Note: The incidence rates for 1000 person years for non-fatal stroke, hemorrhagic stroke, non-fatal myocardial infarction, cardiovascular mortality, all-cause mortality and major gastrointestinal bleeding are reported. The incidence rates are reported for the aspirin group, control group and the difference between the two groups. The incidence rate was calculated by dividing the event totals by the person years of follow-up. CI: confidence interval; NNT: number needed to treat; NNH: number needed to harm. Discussion Main findings We performed a cumulative systematic review and meta-analysis of all randomized controlled trials of aspirin for primary prevention of cardiovascular disease to investigate the relationship between aspirin therapy and stroke. We did not find a statistically significant decreased risk for non-fatal stroke (odds ratio, 0.94, 95% CI, 0.85 to 1.04). We did find a significant increase in hemorrhagic stroke (odds ratio, 1.29; 95% CI, 1.06 to 1.56). There was no significant benefit for all-cause mortality (odds ratio, 0.97; 95% CI, 0.92 to 1.03) or cardiovascular mortality (odds ratio, 0.94; 95% CI, 0.85 to 1.03). We found a statistically significant decreased risk of non-fatal myocardial infarction (odds ratio, 0.80; 95% CI, 0.69 to 0.94) with aspirin use for primary prevention, but a commensurate increase in major gastrointestinal bleeding (odds ratio, 1.83; 95% CI, 1.43 to 2.35). Our net-benefit analysis showed no significant effect of aspirin on the composite of all-cause mortality, non-fatal stroke, non-fatal myocardial infarction and major bleeding.

ith aspirin use for primary prevention, but a commensurate increase in major gastrointestinal bleeding (odds ratio, 1.83; 95% CI, 1.43 to 2.35). Our net-benefit analysis showed no significant effect of aspirin on the composite of all-cause mortality, non-fatal stroke, non-fatal myocardial infarction and major bleeding. Our updated review reports similar treatment estimates for non-fatal stroke and non-fatal myocardial infarction to previous meta-analysis of primary prevention populations (completed prior to recent published RCTs, ARRIVE, ASCEND and ASPREE)12 but differ for all-cause mortality. Previous meta-analyses have reported a small significant reduction in all-cause mortality but our updated results show no significant reduction in this outcome. One potential explanation for the reduction in pooled benefit of aspirin is that the earlier trials were performed in a time of suboptimal modifiable risk factor control especially blood pressure, smoking and hyperlipidemia. For example, the ASCEND and ASPREE trials had rates of statin use of 75% and 34% respectively and had low rates of current smoking, 8.3% and 4% respectively. A recent meta-analysis28 that included ARRIVE, ASCEND and ASPREE, reported significant reductions in ischemic stroke, with an increase in ICH. However, that meta-analysis included clinical trials of populations with subclinical cardiovascular disease i.e. participants with sub-clinical peripheral vascular disease.17,18 As such, our meta-analysis specifically addresses the net effect of aspirin on stroke outcomes in general populations without clinical or subclinical cardiovascular disease. Finally, similar to Zheng et al., we repeated the analysis for all stroke (fatal and non-fatal ischemic and hemorrhagic) but excluded participants with previous cardiovascular disease,17,18 the pooled estimate for reduction in all stroke remained non-significant (odds ratio, 0.95; 95% CI, 0.88 to 1.03) for aspirin use in primary prevention.

eng et al., we repeated the analysis for all stroke (fatal and non-fatal ischemic and hemorrhagic) but excluded participants with previous cardiovascular disease,17,18 the pooled estimate for reduction in all stroke remained non-significant (odds ratio, 0.95; 95% CI, 0.88 to 1.03) for aspirin use in primary prevention. The current American Heart Association/American Stroke Association (AHA/ASA) guidelines give a class IIa recommendation to the use of aspirin for cardiovascular prevention (including but not specific to stroke) for patients at high risk (10-year risk > 10%). The results of this meta-analysis do not support this recommendation. There is now no overall net clinical benefit for aspirin in primary prevention.

give a class IIa recommendation to the use of aspirin for cardiovascular prevention (including but not specific to stroke) for patients at high risk (10-year risk > 10%). The results of this meta-analysis do not support this recommendation. There is now no overall net clinical benefit for aspirin in primary prevention. Strengths and limitations There is substantial heterogeneity in population characteristics between the studies included including sex, age, and baseline co-morbidities. Two studies did not report stroke outcome and had to be excluded from the analysis, introducing a possible reporting bias. Three studies21,26,27 did not require imaging for diagnosis of stroke and three studies did not report if imaging was or was not used for diagnosis.9,10,24 This could introduce a misclassification bias and does not account for small strokes which may have been missed if MRI imaging was not used. A sensitivity analysis including only studies with imaging requirement for diagnosis of stroke7,20,22,23,25 remained non-significant for aspirin benefit on non-fatal stroke in primary prevention We also excluded three of the trials included in previous meta-analysis were excluded due to prior cardiovascular disease, two trials with participants having peripheral vascular disease17,18 and one trial with nearly half having previous cardiovascular disease.19 Excluding these trials gave a more precise answer to our research question, primary prevention of stroke with aspirin. This meta-analysis expands on previous work by including three recent large randomized control trials recently completed in varied populations, diabetic participants without history of cardiovascular disease, elderly patents (>70) and non-diabetic patients at moderate cardiovascular risk.

ention of stroke with aspirin. This meta-analysis expands on previous work by including three recent large randomized control trials recently completed in varied populations, diabetic participants without history of cardiovascular disease, elderly patents (>70) and non-diabetic patients at moderate cardiovascular risk. Implications In conclusion, there is no evidence of a reduction in non-fatal stroke for aspirin in primary prevention of cardiovascular events. There appears to be a small modest reduction in non-fatal myocardial infarction. Balancing this with an increased risk of major gastrointestinal bleeding and hemorrhagic stroke, there appears to be an even smaller net clinical benefit. Our findings do not support routine use of aspirin for primary prevention of cardiovascular events including stroke. Supplemental Material Supplemental material for Aspirin for primary prevention of stroke in individuals without cardiovascular disease—A meta-analysis Click here for additional data file. Supplemental Material for Aspirin for primary prevention of stroke in individuals without cardiovascular disease—A meta-analysis by Conor Judge, Sarah Ruttledge, Robert Murphy, Elaine Loughlin, Sarah Gorey, Maria Costello, Aoife Nolan, John Ferguson, Martin O Halloran, Michelle O'Canavan and Martin J O'Donnell in International Journal of Stroke Author contributions CJ, SR and RM were responsible for data collection. CJ performed analysis. All authors contributed to data interpretation and critical revision of the report.

Supplemental Material for Aspirin for primary prevention of stroke in individuals without cardiovascular disease—A meta-analysis by Conor Judge, Sarah Ruttledge, Robert Murphy, Elaine Loughlin, Sarah Gorey, Maria Costello, Aoife Nolan, John Ferguson, Martin O Halloran, Michelle O'Canavan and Martin J O'Donnell in International Journal of Stroke Author contributions CJ, SR and RM were responsible for data collection. CJ performed analysis. All authors contributed to data interpretation and critical revision of the report. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was performed within the Irish Clinical Academic Training (ICAT) Programme, supported by the Wellcome Trust and the Health Research Board (grant number 203930/B/16/Z), the Health Service Executive, National Doctors Training and Planning and the Health and Social Care, Research and Development Division, Northern Ireland. MOD was supported by the European Research Council (COSIP grant, 640580). The funding source had no role in the study design, analysis or writing of report.

Introduction Cognitive deficits, particularly in the domains of attention, executive function and verbal memory,1 and emotional problems such as depression and anxiety2 are common post-stroke. The potential benefits of music in the treatment of mental disorders are long recognized.3 Music activates brain areas related to attention, memory, motor and affective processing4 and has potential as a safe, accessible, and low-cost intervention to alleviate and/or prevent psychological morbidity after stroke.5,6 Regular, active music listening (i.e. consciously attending to music rather than simply listening to music in the background whilst doing other activities) may also induce relaxation, positive mood change, and evoke memories and reflective thoughts about the past and the future.7,8 A parallel can be drawn with mindfulness training, which has been shown to reduce depressive symptoms and to improve performance on measures of working memory and sustained attention in healthy individuals.9 Mindfulness aims to promote wellbeing through learning to pay attention to the present moment in a non-judgemental way. Mindfulness has also been reported to enhance mood following brain injury10 and has potential applications after stroke.11

n measures of working memory and sustained attention in healthy individuals.9 Mindfulness aims to promote wellbeing through learning to pay attention to the present moment in a non-judgemental way. Mindfulness has also been reported to enhance mood following brain injury10 and has potential applications after stroke.11 It is possible that music listening and mindfulness share a common mechanism of action via attentional control,12 through which they operate by reducing self-focus and increasing metacognitive control and attention to present experiences thus limiting rumination on negative thoughts and feelings. The effects of music listening may therefore be enhanced by incorporating components of mindfulness – music listening could be a vehicle by which attentional control skills associated with mindfulness practice are developed. We performed a pilot feasibility randomized controlled trial (RCT) of a novel intervention combining music listening with brief mindfulness practice aimed at reducing low mood and promoting cognitive recovery in the initial months post-stroke. Aims To assess the feasibility and acceptability of incorporating brief mindfulness training into a music listening intervention in an RCT context post-stroke. Secondary aims to determine recruitment and retention rates over 18 months, to determine estimated effect sizes (with 95% confidence intervals) for change on key outcome measures of mood and cognition.

Aims To assess the feasibility and acceptability of incorporating brief mindfulness training into a music listening intervention in an RCT context post-stroke. Secondary aims to determine recruitment and retention rates over 18 months, to determine estimated effect sizes (with 95% confidence intervals) for change on key outcome measures of mood and cognition. Methods A three-arm, parallel group, single-blind pilot RCT comparing (1) music listening with brief mindfulness training (referred to as mindful-music listening from here on), (2) music listening alone, and (3) audiobook listening (control arm). During the study, all participants received treatment as usual, including multidisciplinary rehabilitation. Figure 1 shows participant flow through the study. Reporting follows CONSORT extension for pilot and feasibility trials.13 The study was approved by the West of Scotland Research Ethics Committee (14/WS/1089). All participants gave written informed consent. The study was registered with clinicaltrials.gov (NCT02259062) and the UK Clinical Research Network (ID 18019). Figure 1. Participant flow. Participants Eligibility criteria Native English-speaking adults (aged ≥ 18, upper age limit of 80 for the first 11 months of recruitment) in the acute phase (≤14 days post-stroke) following clinically and/or radiologically (CT and/or MRI) confirmed diagnosis of ischemic stroke.

Methods A three-arm, parallel group, single-blind pilot RCT comparing (1) music listening with brief mindfulness training (referred to as mindful-music listening from here on), (2) music listening alone, and (3) audiobook listening (control arm). During the study, all participants received treatment as usual, including multidisciplinary rehabilitation. Figure 1 shows participant flow through the study. Reporting follows CONSORT extension for pilot and feasibility trials.13 The study was approved by the West of Scotland Research Ethics Committee (14/WS/1089). All participants gave written informed consent. The study was registered with clinicaltrials.gov (NCT02259062) and the UK Clinical Research Network (ID 18019). Figure 1. Participant flow. Participants Eligibility criteria Native English-speaking adults (aged ≥ 18, upper age limit of 80 for the first 11 months of recruitment) in the acute phase (≤14 days post-stroke) following clinically and/or radiologically (CT and/or MRI) confirmed diagnosis of ischemic stroke. Exclusion criteria Comorbid progressive neurological or neurodegenerative condition, major psychiatric disorder (pre-stroke history of mood disorder or stable antidepressant medication did not lead to exclusion), history of major substance abuse problems, clinically unstable, unable to give informed consent or unable to cooperate with the study protocol (e.g. due to severe aphasia, uncorrected impairment of hearing or vision). Co-recruitment with intervention studies with potential impact on mood/cognition was not allowed.

on), history of major substance abuse problems, clinically unstable, unable to give informed consent or unable to cooperate with the study protocol (e.g. due to severe aphasia, uncorrected impairment of hearing or vision). Co-recruitment with intervention studies with potential impact on mood/cognition was not allowed. In addition, participants were asked to nominate an informant (optional). Recruitment Participants were recruited from acute stroke units within NHS Greater Glasgow and Clyde to have baseline assessments within four weeks of stroke onset. The Scottish Stroke Research Network nurses approached potential participants at the ward who were enrolled between 12 January 2015 and 28 January 2016. Participants were advised the study investigated listening-based leisure activities to avoid emphasizing music as an active ingredient. Participants received no financial support for their participation in the study. Assessments and randomization Baseline cognitive status and mood were assessed using standardized assessments (Table 1). The order of presentation was standardized to prioritize a core set of tests while getting participants to complete as many tests from the full assessment battery as feasible. Prior to unblinding and data analysis, we selected primary outcome measures for cognition (Delayed Story Recall) and mood (HADS) based on previous findings,5 clinical importance, expected sensitivity to the intervention and feasibility of administration in a full trial. Table 1. Assessments

essment battery as feasible. Prior to unblinding and data analysis, we selected primary outcome measures for cognition (Delayed Story Recall) and mood (HADS) based on previous findings,5 clinical importance, expected sensitivity to the intervention and feasibility of administration in a full trial. Table 1. Assessments Cognitive domain Global Cognition Montreal Cognitive Assessment (MoCA)14 Attention Test of Everyday Attention (TEA)15: Map search; Elevator counting; Elevator counting with distraction; Visual Elevator; Telephone search; Telephone search with counting. Sustained attention Sustained Attention to Response Task (SART)16; TEA Lottery15 Verbal memory BIRT Memory and Information Processing Battery (BMIPB)17 subtests: (i) Story recall (immediate and delayed); (ii) List learning Speed of processing BMIPB17 Speed of information processing Reaction time CANTAB18 5-choice reaction time subtest Verbal working memory Wechsler Adult Intelligence Scale – Fourth Edition (WAIS-IV)19; Digit Span Visual working memory Wechsler Memory Scale – Fourth Edition (WMS-IV)20 Symbol Span Phonemic fluency Controlled Oral Word Association Test21 Questionnaire Mood Hospital Anxiety and Depression Scale (HADS)22 Attentional control Metacognitions Questionnaire short form (MCQ-30)23 Mindfulness Five Facet Mindfulness Questionnaire short form (FFMQ-SF)24 Emotion regulation Brain Injury Rehabilitation Trust Regulation of Emotions Questionnaire (BREQ)25 Functional independence Mayo-Portland Adaptability Inventory-4 (MPAI-4)26 (Baseline + 6-months only)

tacognitions Questionnaire short form (MCQ-30)23 Mindfulness Five Facet Mindfulness Questionnaire short form (FFMQ-SF)24 Emotion regulation Brain Injury Rehabilitation Trust Regulation of Emotions Questionnaire (BREQ)25 Functional independence Mayo-Portland Adaptability Inventory-4 (MPAI-4)26 (Baseline + 6-months only) Following completion of the baseline assessments, stratified randomization with blocking (block size 3, allocation ratio 1:1:1) was used to allocate participants to groups based on: (i) recruitment location, (ii) type of stroke (cortical vs. subcortical), and (iii) recurrence (first vs. recurrent stroke). Randomization was via an automated telephone system. Being a single blind study, participants were not blind to intervention group allocation.

as used to allocate participants to groups based on: (i) recruitment location, (ii) type of stroke (cortical vs. subcortical), and (iii) recurrence (first vs. recurrent stroke). Randomization was via an automated telephone system. Being a single blind study, participants were not blind to intervention group allocation. Interventions All interventions followed a manualized eight-week program developed by the research team and delivered by an assistant psychologist on an individual basis. Participants were given an iPod Nano (7th Generation, Apple Inc.) and asked to listen to their material daily on their own for at least an hour during the intervention phase (target, 56 h over eight consecutive weeks). They were also asked to keep a written daily record of listening though which adherence to the interventions was measured. Written instructions were provided. Participants selected their preferred music/audiobooks from any genre with the therapist keeping a record of the genres given at each visit. The therapist also rated their perceived feasibility of intervention delivery at each session as ’fully feasible’, ‘partially feasible’ or ‘not feasible’.

tructions were provided. Participants selected their preferred music/audiobooks from any genre with the therapist keeping a record of the genres given at each visit. The therapist also rated their perceived feasibility of intervention delivery at each session as ’fully feasible’, ‘partially feasible’ or ‘not feasible’. The mindful-music participants were introduced to the concept of mindfulness at the first visit and given a recording containing a brief mindfulness exercise (Body scan) to complete daily prior to music listening (weeks 1–3). The brief (∼5 min) mindfulness exercises8 focused on key elements of mindfulness (e.g. paying attention to the present moment). If participants were to notice any thoughts or sensations arising either during the brief exercise or during subsequent music listening, they were to allow them to pass and to gently bring their attention back to the exercise/music. Seven more weekly visits followed during which progress was monitored, listening records collected and additional material provided as required. Adherence to the daily mindfulness exercises was measured via participants’ diary records. At the fourth visit, another brief exercise (Following the breath) was provided for use over the following three weeks (weeks 4–6). For the last two weeks, participants could choose which exercise to complete. The two music groups therefore differed by the virtue of attentional control with the mindful-music group asked to return their attention back to the music whenever their mind had wandered with no specific listening instructions given to the music only group.

weeks, participants could choose which exercise to complete. The two music groups therefore differed by the virtue of attentional control with the mindful-music group asked to return their attention back to the music whenever their mind had wandered with no specific listening instructions given to the music only group. At the final visit, plans for listening post-intervention were discussed and CD/recording of the mindfulness exercises given to the mindful-music group. Post-intervention, participants were interviewed about their experience of engaging in the trial, including participant perceived feasibility and acceptability of the interventions. These findings are reported elsewhere.8 Follow-up Follow-up assessments of mood and cognition using parallel versions where available (Table 1) were completed post-intervention and at six months post-stroke by an assessor blind to group allocation. Information on medication and other interventions were also collected. Informants provided ratings of emotion regulation and functional independence at six months post-stroke.

llel versions where available (Table 1) were completed post-intervention and at six months post-stroke by an assessor blind to group allocation. Information on medication and other interventions were also collected. Informants provided ratings of emotion regulation and functional independence at six months post-stroke. Treatment fidelity All intervention sessions were audio-recorded. A random sub-sample (n = 18, 5%) were checked for treatment fidelity by the principal investigator (JE) who was not involved in intervention delivery or outcome assessments. The selection was done using an Excel random number generator with approximately equal numbers selected from the beginning (sessions 1–3), middle (sessions 4–5), and end (sessions 6–8) of the intervention. Adherence was rated using a three-point scale (content consistent with protocol for stage of treatment; partially consistent but evidence of deviation into other unrelated areas or treatment methods; largely inconsistent with protocol for stage of treatment).

(sessions 4–5), and end (sessions 6–8) of the intervention. Adherence was rated using a three-point scale (content consistent with protocol for stage of treatment; partially consistent but evidence of deviation into other unrelated areas or treatment methods; largely inconsistent with protocol for stage of treatment). Statistical methods Data were analyzed using SAS software (v9.3), following a statistical analysis plan, on an intention-to-treat basis. Missing data were not imputed. Continuous variables were summarized as mean and standard deviation (SD) or median and interquartile range (IQR), depending on distribution. Categorical variables were summarized as number and percentage (n(%)). Between-group differences were assessed using analysis of variance (ANOVA), Kruskal–Wallis, or Fisher’s tests, as appropriate. Effect sizes were calculated using linear regression models, adjusted for baseline measure and stratification factors (recruitment location/type of stroke/stroke recurrence). The study was not designed to have power to detect significant differences in outcomes between groups. Results Feasibility of recruitment Feasibility of recruitment is shown in Figure 2. Figure 2. CONSORT Flowchart. Reasons for withdrawal prior to randomization did not appear to relate to the intervention per se but related to feeling too unwell in general (n3), feeling overwhelmed following hospital discharge (n2), deciding to focus on speech and language therapy (n1), change of mind (n2) or change in living arrangements (n2). Some gave no specific reason (n7).

randomization did not appear to relate to the intervention per se but related to feeling too unwell in general (n3), feeling overwhelmed following hospital discharge (n2), deciding to focus on speech and language therapy (n1), change of mind (n2) or change in living arrangements (n2). Some gave no specific reason (n7). Sixty-two (86.1%) of those randomized nominated an informant to complete questionnaires at the six-month follow-up. Forty-eight (77.4%) consented to receiving the questionnaires, of which 34 (73.8%) were returned by a spouse/partner (41.5%), child (18.5%), friend (6.2%), parent (4.6%), sibling (1.5%), and other relative (1.5%). The average rate of recruitment was 1.6 participants per week (89 participants/55 weeks). Sample characteristics Sample characteristics at baseline are presented in Table 2. A summary of cognitive and mood status, medication and other interventions including engagement in listening and mindfulness-based leisure activities pre-stroke is provided in the Supplementary Tables 1 to 3. Table 2. Baseline participant characteristics

tics Sample characteristics at baseline are presented in Table 2. A summary of cognitive and mood status, medication and other interventions including engagement in listening and mindfulness-based leisure activities pre-stroke is provided in the Supplementary Tables 1 to 3. Table 2. Baseline participant characteristics Overall (N = 72) Mindfulness (N = 23) Music (N = 24) Audiobook (N = 25) Age (years) Mean (SD) 64.0 (11.60) 65.3 (11.13) 61.1 (10.36) 65.7 (12.97) Gender n (%) Male 45 (62.5%) 13 (56.5%) 17 (70.8%) 15 (60.0%) Education (years) Median (IQR) 11.0 (10.0, 15.0) 12.0 (10.0, 15.0) 11.2 (10.5, 15.0) 10.0 (10.0, 13.0) SIMD quintile 0–20% 32(44.4%) 6 (26.1%) 13 (54.2%) 13 (52.0%) 0% most deprived, 20–40% 14(19.4%) 7(30.4%) 5(20.8%) 2(8.0%) 100% least deprived 40–60% 9 (12.5%) 4 (17.4%) 2 (8.3%) 3 (12.0%) 60–80% 10 (13.9%) 2 (8.7%) 2 (8.3%) 6 (24.0%) 80–100% 7 (9.7%) 4(17.4%) 2 (8.3%) 1 (4.0%) Oxford stroke classification Cortical 46 (63.9%) 15 (65.2%) 15 (62.5%) 16 (64.0%) Subcortical 26 (36.1%) 8(34.8%) 9 (37.5%) 9 (36.0%) If cortical, type TACS 3 (6.5%) 1 (6.7%) 1 (6.7%) 1 (6.3%) PACS 21 (45.7%) 9 (60.0%) 5 (33.3%) 7 (43.8%) POCS 22 (47.8%) 5(33.3%) 9 (60.0%) 8 (50.0%) Hemisphere Right 27 (38.0%) 6 (27.3%) 10 (41.7%) 11 (44.0%) Left 40 (56.3%) 15 (68.2%) 14 (58.3%) 11 (44.0%) Other 4 (5.6%) 1 (4.5%) 0 3 (12.0%) Stroke recurrence First 59 (81.9%) 19 (82.6%) 20 (83.3%) 20 (80.0%) Second 12 (16.7%) 3 (13.0%) 4 (16.7%) 5 (20.0%) Third or more 1 (1.4%) 1 (4.3%) 0 0 Modified Rankin scale 0 (no symptoms) 5 (6.9%) 4 (17.4%) 0 1(4.0%) 1 22 (30.7%) 7 (30.4%) 9 (37.5%) 6 (24.0%) 2 33 (45.8%) 7 (30.4%) 14 (58.3%) 12 (48.0%) 3 6 (8.3%) 3 (13.0%) 0 3 (12.0%) 4 (Moderately severe disability) 6 (8.3%) 2 (8.7%) 1 (4.2%) 3 (12.0%) IQR: interquartile range; PACS: partial anterior circulation stroke; POCS: posterior circulation stroke; SD: standard deviation; SIMD: Scottish index of multiple deprivation; TACS: total anterior circulation stroke

%) 3 6 (8.3%) 3 (13.0%) 0 3 (12.0%) 4 (Moderately severe disability) 6 (8.3%) 2 (8.7%) 1 (4.2%) 3 (12.0%) IQR: interquartile range; PACS: partial anterior circulation stroke; POCS: posterior circulation stroke; SD: standard deviation; SIMD: Scottish index of multiple deprivation; TACS: total anterior circulation stroke The median time from stroke onset to baseline assessment was 19 days (range 5–28). Retention during intervention Of the 72 participants randomized, 65 (90.3%) commenced treatment. Of those who did not commence, four withdrew after randomization (mindful-music (n2), audiobooks (n2)), one was lost to contact, and two were unable to commence for other reasons. Median time from stroke to the first intervention session was 31 days (range 17–56). Seven individuals dropped out of the intervention but remained in the study. One dropped out of the mindful-music group (one session completed) due to difficulty concentrating in general, two from the music group (one after one session and one after two sessions) due to group allocation, high number of other treatments, and four from the audiobooks (no, two, three and four sessions completed) due to group allocation, low mood, and issues with visit scheduling (e.g. staying with family in another part of the country). Data from these individuals have been included in statistical analyses. Two withdrew from the study during the intervention phase due to being unwell.

o, two, three and four sessions completed) due to group allocation, low mood, and issues with visit scheduling (e.g. staying with family in another part of the country). Data from these individuals have been included in statistical analyses. Two withdrew from the study during the intervention phase due to being unwell. Feasibility and acceptability of interventions Intervention feasibility was assessed via therapist ratings with 96.9% of treatment sessions rated ‘fully feasible’ by the therapist. The remaining 3.1% were rated ‘partially feasible’ due to fatigue, hand function, mood, comprehension, verbal expression, cognition or session content. The median treatment session completion was 7 for mindful-music, 6 for music, and 8 for audiobooks with 68.1% completing at least six intervention sessions and 37.5% completing all eight sessions. The majority (86.6%) of sessions took place at the participant’s home, followed by work (8.4%), telephone (2.5%), hospital (2.2%), and e-mail (0.2%). The mean first intervention session length was 75 min (SD 29) for mindful-music, 63 min (SD 17) for music, and 59 min (SD 21) for the audiobook group (p = 0.24). Subsequent sessions were approximately 20 min. The mean listening diary completion rate over eight weeks for those randomized was 68% (0–100%) and 80% (0–100%) for those completing the study.

on length was 75 min (SD 29) for mindful-music, 63 min (SD 17) for music, and 59 min (SD 21) for the audiobook group (p = 0.24). Subsequent sessions were approximately 20 min. The mean listening diary completion rate over eight weeks for those randomized was 68% (0–100%) and 80% (0–100%) for those completing the study. Adherence to music/audiobook listening and mindfulness exercises was measured via participants’ diary records against the recommended dose (1 h × 7 days × 8 weeks = 56 h). Median listening time in the mindful-music group was 45 h (80.4% of target listening), 60 h in the music group (107% of target listening), and 49 h (87.5% of target listening) in the audiobook group which did not differ statistically (p = 0.15). The mean adherence to the mindfulness exercises in the mindful-music group was 40% (0–100%) for those randomized and 60% (0–100%) for those completing the six-month follow-up. New listening material was provided in 59% of sessions, with fiction/crime being most popular for audiobooks, pop/rock in the music group, and classical/easy listening in the mindful-music group. As described elsewhere,8 94.6% of participants found appointments convenient and 73.3% reported it was feasible to listen daily for up to an hour. The music and audiobook groups were more likely to engage in daily music listening post-intervention compared to the mindful-music group (p = 0.012). Similarly, the audiobook group was more likely to continue audiobook listening post-intervention compared to the other two groups (p = 0.0052). No significant group differences (p = 0.31) were found in engagement in relaxation, mindfulness or meditation-based leisure activities (Supplementary Table 4).

oup (p = 0.012). Similarly, the audiobook group was more likely to continue audiobook listening post-intervention compared to the other two groups (p = 0.0052). No significant group differences (p = 0.31) were found in engagement in relaxation, mindfulness or meditation-based leisure activities (Supplementary Table 4). Retention to six-month end-point Retention to the six-month end-point is summarized in Figure 2 and was similar across groups (p = 0.13). For the mindful-music group, seven were lost (four withdrew due to ill-health [prior intervention (n2), after session 4 (n1) and at qualitative interview (n1)] and three were lost to contact [prior intervention (n1) and during the intervention (n2)]). For the music group, two were lost (one withdrew due to ill-health and one was lost to contact during the intervention). For the audiobook group, three were lost (one withdrew prior intervention due to ill-health, one immediately after randomization due to unwillingness to listen to audiobooks and one was lost to contact at three-month follow-up). Withdrawals due to ill-health did not appear to be related to the interventions but to other health problems or the effects of stroke (e.g. feeling overwhelmed after returning home from hospital). Treatment fidelity Treatment fidelity was judged to be high with 17/18 (94.4%) sessions being fully consistent with the study protocol.

Retention to six-month end-point Retention to the six-month end-point is summarized in Figure 2 and was similar across groups (p = 0.13). For the mindful-music group, seven were lost (four withdrew due to ill-health [prior intervention (n2), after session 4 (n1) and at qualitative interview (n1)] and three were lost to contact [prior intervention (n1) and during the intervention (n2)]). For the music group, two were lost (one withdrew due to ill-health and one was lost to contact during the intervention). For the audiobook group, three were lost (one withdrew prior intervention due to ill-health, one immediately after randomization due to unwillingness to listen to audiobooks and one was lost to contact at three-month follow-up). Withdrawals due to ill-health did not appear to be related to the interventions but to other health problems or the effects of stroke (e.g. feeling overwhelmed after returning home from hospital). Treatment fidelity Treatment fidelity was judged to be high with 17/18 (94.4%) sessions being fully consistent with the study protocol. Effect sizes and sample size calculations for full-scale trial The study was not designed to evaluate differences in outcomes between groups; however, intention to treat analysis was carried out to obtain an estimate of treatment effect (Cohen’s d) on outcomes at three and six months post-stroke. Cohen27 suggested that effect sizes are classed as small (d = 0.2), medium (d = 0.5), and large (d = 0.8). Results of this analysis are provided in the Supplementary Tables 5 and 6. With regard to the outcome measures chosen a priori for a future full-scale trial, it is noteworthy that Delayed Story Recall produced one of the largest effect sizes (standardized mean difference) at six months, favoring the music (d = 0.42 [−0.07, 0.92]) and the mindful-music (d = 0.44 [−0.11, 1.00]) groups (small-medium effect), compared with audiobooks. Similar findings were apparent for Immediate Story Recall favoring the music (d = 0.58 [0.06, 1.11]) and mindful-music (d = 0.51 [−0.07, 1.09]) groups (medium-large effect) over audiobooks; and for the measure of attentional switching (visual elevator accuracy score), which also favored the music (d = 0.67 [0.12, 1.22]) and mindful-music (d = 0.77 [0.16, 1.38]) groups (medium-large effect) over audiobooks. The mean HADS mood scores were within the normal range across groups at all time points. Many of the other measures of cognition and the self-report questionnaires showed very little difference between groups.

d = 0.67 [0.12, 1.22]) and mindful-music (d = 0.77 [0.16, 1.38]) groups (medium-large effect) over audiobooks. The mean HADS mood scores were within the normal range across groups at all time points. Many of the other measures of cognition and the self-report questionnaires showed very little difference between groups. Measures of variance in change scores were used to estimate sample size for a full-scale trial. Based on Delayed Story Recall at six months, a full-scale three-arm trial would require the randomization of 306 participants to detect a clinically substantial difference in improvement (z-score difference = 0.66, p = 0.017 [Bonferroni-correction], two-tailed, 80% power, with 16.7% attrition). This would also provide 91% power to detect a modest (2-point) difference in change in depression and 84% power to detect a 2-point difference in change in anxiety measured using the HADS.

ference in improvement (z-score difference = 0.66, p = 0.017 [Bonferroni-correction], two-tailed, 80% power, with 16.7% attrition). This would also provide 91% power to detect a modest (2-point) difference in change in depression and 84% power to detect a 2-point difference in change in anxiety measured using the HADS. Discussion Recruitment to MELLO was feasible, though slower than anticipated. The interventions were feasible to deliver and acceptable to participants with high treatment fidelity. Listening adherence was found to be excellent with the music listening group exceeding their listening target and the other two arms achieving at least 80% of their target listening time with the both music only and audiobook groups found to be significantly more likely to continue listening to their allocated material during the three month follow-up phase compared to the mindful music listening group. All treatment arms had attrition rates equal to or better than what might be typically expected for psychological therapies post-stroke.28 Our recruitment and adherence results are also aligned with previous studies of post-stroke music listening.5,29 The study was not powered as an efficacy trial but treatment effect sizes for both music listening groups compared to audiobook listening at six-months on measures identified a-priori as key outcomes (verbal memory and attentional switching), were consistent with previous research.5 One must be cautious, however, in interpreting effect sizes, particularly since some measures had missing data at one or more time points. We have previously reported positive self-reported cognitive impacts in the domains of memory and attention based on participants’ qualitative experiences of the interventions.8 Both music groups but not the audiobook group reported memory reminiscence, while the mindful music group referred to being better able to refocus their mind following mind wandering consistent with the idea of improved attentional control or attentional switching.

ipants’ qualitative experiences of the interventions.8 Both music groups but not the audiobook group reported memory reminiscence, while the mindful music group referred to being better able to refocus their mind following mind wandering consistent with the idea of improved attentional control or attentional switching. Our sample had low levels of depression and anxiety throughout the study. It is possible that participants with, or vulnerable to, mood disorder chose not to participate (sampling bias), or that all groups benefitted from a listening-based activity and weekly contact. The latter is partially supported by feedback from post-intervention interviews8 suggested that participants in the music listening groups more frequently referred to listening being enjoyable and uplifting, and promoting memory reminiscence compared to audiobooks. But there were differences between the music groups, with the mindful-music group more frequently referring to listening aiding relaxation, attentional control, and emotion regulation. A strength of our study was inclusion of a clinically and demographically diverse population compared to the previous music listening study in stroke.5 As with other acute stroke, interventional studies numbers with moderate to severe stroke were low. In contrast to Särkämö et al.,5 we did not use a qualified music therapist to deliver the interventions, given the limited availability of this expertise within the UK National Health Service. Similar findings between the studies suggest this may not be critical.

studies numbers with moderate to severe stroke were low. In contrast to Särkämö et al.,5 we did not use a qualified music therapist to deliver the interventions, given the limited availability of this expertise within the UK National Health Service. Similar findings between the studies suggest this may not be critical. We recognize limitations in our study but can use these to improve design of a future trial. Given the nature of cognitive, physical, and psychological difficulties in our participants, it was not possible to administer all tests to every participant at each time point. We therefore prioritized a core set of tests, albeit attempted to administer as many of the planned assessments as was feasible without over-burdening our participants. Future trials should limit the number of tests administered and the finding that medium-large effect sizes for measures of verbal memory (Story Recall) and attentional switching were consistent with previous research suggests these would be appropriate for future studies.

asible without over-burdening our participants. Future trials should limit the number of tests administered and the finding that medium-large effect sizes for measures of verbal memory (Story Recall) and attentional switching were consistent with previous research suggests these would be appropriate for future studies. Other limitations were the use of self-reported listening diaries. Although the diaries were completed daily and collected on a weekly basis, estimation of listening duration is subject to bias. A future trial would benefit from using an objective measure of listening duration. Future trials could also explore options for using technology or existing services to support listening and examine whether using music in different ways (for relaxation, to improve concentration) affects outcomes or improves engagement in other therapies (e.g. physiotherapy). Research should also investigate potential mechanisms of action. One possibility may be improved attentional control, but there are other potential psychological and neurobiological mechanisms such as emotion regulation, reward and stress reduction, which may mediate the cognitive effects of music30 and mindfulness-based interventions.31

o investigate potential mechanisms of action. One possibility may be improved attentional control, but there are other potential psychological and neurobiological mechanisms such as emotion regulation, reward and stress reduction, which may mediate the cognitive effects of music30 and mindfulness-based interventions.31 In summary, we found that it was feasible and acceptable to incorporate brief mindfulness training into a music listening intervention in an RCT context post-stroke. Although effect sizes for key cognitive (memory and attention) variables were consistent with previous research in favoring the music listening interventions over the audiobook control, a full scale definitive RCT is needed before treatment recommendations could be made. Recruitment and retention data, in combination with power analysis, suggest that progression to a full-scale trial seems feasible and justified. Supplemental Material Supplemental material for Measuring the effects of listening for leisure on outcome after stroke (MELLO): A pilot randomized controlled trial of mindful music listening Click here for additional data file. Supplemental Material for Measuring the effects of listening for leisure on outcome after stroke (MELLO): A pilot randomized controlled trial of mindful music listening by Satu Baylan, Caroline Haig, Maxine MacDonald, Ciara Stiles, Jake Easto, Meigan Thomson, Breda Cullen, Terence J Quinn, David Stott, Stewart W Mercer, Niall M Broomfield, Heather Murray and Jonathan J Evans in International Journal of Stroke

roke (MELLO): A pilot randomized controlled trial of mindful music listening by Satu Baylan, Caroline Haig, Maxine MacDonald, Ciara Stiles, Jake Easto, Meigan Thomson, Breda Cullen, Terence J Quinn, David Stott, Stewart W Mercer, Niall M Broomfield, Heather Murray and Jonathan J Evans in International Journal of Stroke Acknowledgements The authors would like to thank Dr Sharon Byrne for devising and recording of the brief mindfulness exercises; Scottish Stroke Research Network for participant recruitment; Dr Teppo Särkämö for advice. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by the Dunhill Medical Trust, grant R432/0214. Additional support from Scottish Executive Chief Scientist Office (TQ/BC), Stroke Association (TQ) and The Dr Mortimer and Theresa Sackler Foundation (BC/SB).

Introduction Stroke is a considerable cause of mortality and the leading cause of neurological disability, with post-stroke sequelae negatively impacting long-term mental and physical health.1 Despite a high prevalence and growing incidence, current stroke treatments are limited to reperfusion strategies that, though effective in some patients, are not widely applicable. Lack of alternative stroke treatments is due to a number of reasons, one of which is an under-appreciation of the role that co-morbidities play in stroke etiology and prognosis.2 It is clear that pre-existing conditions, often with inflammatory pathogeneses, are major determinants of stroke-induced damage in both humans and rodent models. Indeed, obesity,3–5 infection,6,7 atherosclerosis,8 and old age9,10 can all negatively affect ischemic brain damage and overall recovery in animal models. Although there is an emerging focus on stroke co-morbidities, the common oral disease periodontitis (PD) is under-studied in this respect, despite being frequently associated with increased risk of ischemic stroke in humans.11–20

,10 can all negatively affect ischemic brain damage and overall recovery in animal models. Although there is an emerging focus on stroke co-morbidities, the common oral disease periodontitis (PD) is under-studied in this respect, despite being frequently associated with increased risk of ischemic stroke in humans.11–20 PD is a chronic inflammatory condition that affects the supporting structures of the dentition, and is considered one of the most prevalent inflammatory diseases in humans.21,22 PD involves formation of a deep ulcerated pocket that harbors proliferating anaerobic bacteria, which leads to chronic inflammation and destruction of the supporting bone. Up to 50% of the global population is affected by some form of periodontal disease and as much as 10% affected by severe PD, which can adversely affect systemic health.23,24 Indeed, PD is emerging as a possible risk factor for many systemic diseases,25 including diabetes,26 rheumatoid arthritis,27 Alzheimer's disease,28–31 and cardiovascular disease.32–34 Pre-clinical studies have directly implicated PD in the development and progression of atherosclerosis,35–39 which is intimately linked to increased incidence of stroke.

ble risk factor for many systemic diseases,25 including diabetes,26 rheumatoid arthritis,27 Alzheimer's disease,28–31 and cardiovascular disease.32–34 Pre-clinical studies have directly implicated PD in the development and progression of atherosclerosis,35–39 which is intimately linked to increased incidence of stroke. There are a number of proposed mechanisms that explain how pathophysiological events in the oral cavity affect the cardiac vasculature and distant tissue sites. During PD, ulceration of the epithelium and lysis of deeper periodontal tissues facilitates greater translocation of oral bacteria or their products (e.g. lipopolysaccharide (LPS)) into the bloodstream which can induce systemic inflammation or endotoxemia.40,41 Similarly, pro-inflammatory mediators from the periodontium can elicit systemic effects if they “spill over” into the bloodstream.11,21 Chronic PD is known to result in increased and sustained levels of pro-inflammatory mediators in the systemic circulation.42 Further, the healthy brain with an intact blood–brain barrier (BBB) is susceptible to systemic inflammatory challenge.43 A compromised BBB, as occurs in stroke, may facilitate entry of pro-inflammatory molecules and/or bacterial products and propagate damage within the brain. LPS from the prominent periopathogen Porphyromonas gingivalis (P. gingivalis) has been found in the demented brain.30 Furthermore, periodontal pathogens themselves are highly invasive; P. gingivalis can compromise and cross the BBB into the brain,44 and others (Treponema spp.) have been speculated to enter the brain directly via peripheral trigeminal nerves.31 Taken together, this indicates that chronic periodontal injury may modulate events in the brain post-stroke and affect outcome.

ghly invasive; P. gingivalis can compromise and cross the BBB into the brain,44 and others (Treponema spp.) have been speculated to enter the brain directly via peripheral trigeminal nerves.31 Taken together, this indicates that chronic periodontal injury may modulate events in the brain post-stroke and affect outcome. PD and stroke share similar risk factors, such as age, smoking, hypertension, and obesity,45 and both conditions share inflammatory pathways. Further, a systemic pro-inflammatory profile pre- and post-ischemia is an important determinant of poorer outcome.7,46–48 Although epidemiological studies suggest that PD is associated with increased stroke risk, there is a lack of pre-clinical research on whether PD worsens outcome after stroke. Aims The aim of the present study was therefore to determine the contribution of the prevalent oral disease PD to ischemic brain damage. We employed a clinically relevant model of PD in mice and determined its effect on peripheral and central inflammation and acute stroke outcome by using two different models of cerebral ischemia.

of the present study was therefore to determine the contribution of the prevalent oral disease PD to ischemic brain damage. We employed a clinically relevant model of PD in mice and determined its effect on peripheral and central inflammation and acute stroke outcome by using two different models of cerebral ischemia. Materials and methods Animals Male 8–12-week-old C57BL/6 mice (Envigo, UK) were used for all studies. Animals were housed in individually ventilated cages (temperature 21 ± 2℃; humidity 55% ± 5%; 12-h light/12-h dark cycle) and given access to food and water ad libitum under specific pathogen-free conditions. Animals were allocated treatments and surgical fates in a randomized order across cages. The order of surgeries was also randomized. All scientific procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986 under relevant UK Home Office licences and approved by the local Animal Welfare and Ethical Review Board (University of Manchester, UK).

n a randomized order across cages. The order of surgeries was also randomized. All scientific procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986 under relevant UK Home Office licences and approved by the local Animal Welfare and Ethical Review Board (University of Manchester, UK). Ligature-induced PD and LPS challenge Procedures described here are adapted from Abe and Hajishengallis.49 Briefly, mice were anaesthetized via the intraperitoneal route with a mix of Narketan (50 mg/kg; ketamine) and Domitor (650 µg/kg; medetomidine) (both Vétoquinol, UK). Silk ligatures (5–0, Fine Science Tools, Germany) were tied around both second maxillary molars. Mice were revived with subcutaneous administration of Antisedan (1 mg/kg; atipamezole) (Vétoquinol) and monitored until recovery. “Control” animals were anesthetized only and not subjected to creation of interdental spaces or ligature placement (as to do so would cause damage that is associated with PD). Bone loss in this model occurs at 10 days post-ligature placement. Ligatures in place longer would risk teeth falling out which is a clear exclusion criterion. There was no mortality during or post-surgery. Where applicable, LPS from P. gingivalis (Pg-LPS) (1 mg/kg; InvivoGen, France) was administered intravenously at days 1, 3, 5, 7, 9 post-ligature placement.

ost-ligature placement. Ligatures in place longer would risk teeth falling out which is a clear exclusion criterion. There was no mortality during or post-surgery. Where applicable, LPS from P. gingivalis (Pg-LPS) (1 mg/kg; InvivoGen, France) was administered intravenously at days 1, 3, 5, 7, 9 post-ligature placement. Bone loss measurements For bone loss, mice were euthanized via CO2. Heads were then removed and defleshed, and maxillae were isolated and stained with 1% methylene blue. Images of the palatal side of the maxillae were taken using a Leica Stereo Fluorescence M205 FA (Leica Microsystems, UK). Quantification of bone loss was then calculated across six different molar sites by measuring the distance from cemento-enamel junction (CEJ) and alveolar bone crest (ABC) using Image J (NIH, USA). Bacterial colony-forming units Swabs of the oral cavity were taken at time of sacrifice and placed into sterile phosphate buffered saline (PBS). Dilutions were plated on trypticase soy agar (Oxoid, UK) and incubated at 37℃ for 24 h to detect non-selective growth of aerobic bacterial species. To detect anaerobic growth, oral swab suspensions were plated on Wilkins-Chalgrens agar (Oxoid, UK) and placed in a hypoxic chamber at 37℃ for 72 h. Colony-forming units (CFUs) were determined by manually counting bacterial colonies.

incubated at 37℃ for 24 h to detect non-selective growth of aerobic bacterial species. To detect anaerobic growth, oral swab suspensions were plated on Wilkins-Chalgrens agar (Oxoid, UK) and placed in a hypoxic chamber at 37℃ for 72 h. Colony-forming units (CFUs) were determined by manually counting bacterial colonies. Filament middle cerebral artery occlusion Focal cerebral ischemia was induced via transient occlusion of the middle cerebral artery by filament (fMCAo) as described previously.50 Briefly, under isoflurane anesthesia (4% induction, 1.5–2% maintenance, in a 70:30 mix of N2O:O2), the carotid arteries were exposed and a 6/0 silicon-coated nylon filament (Doccol, USA) was introduced into the internal carotid artery and advanced to occlude the middle cerebral artery (MCA) at its origin. A laser Doppler probe (Moor Instruments, UK) was used to monitor cerebral blood flow (CBF). Successful occlusion was confirmed by reduction in CBF of approximately 80% or greater. The filament was withdrawn after 20 min to allow reperfusion and the wound sutured. Mice were normothermic prior to surgery and during surgery core body temperature was maintained at 37 ± 0.5℃. Post-surgery, buprenorphine (0.05 mg/kg) for analgesia and 0.5 ml saline were administered subcutaneously and mice were placed in a warm cabinet (27–28℃) to recover. Animals were excluded from analyses if there was absence of stroke (n = 1) or mortality during surgery (n = 1).

y temperature was maintained at 37 ± 0.5℃. Post-surgery, buprenorphine (0.05 mg/kg) for analgesia and 0.5 ml saline were administered subcutaneously and mice were placed in a warm cabinet (27–28℃) to recover. Animals were excluded from analyses if there was absence of stroke (n = 1) or mortality during surgery (n = 1). Distal middle cerebral artery occlusion Permanent occlusion of the distal MCA (dMCAo) was induced as described previously.51 Briefly, mice were anaesthetized with isoflurane (5% induction, 2–2.5% maintenance, in a 70:30 mix of N2O:O2) and mounted on a stereotactic frame. The temporal muscle was detached from the skull and a small cranial window drilled directly above the MCA. A triangle of filter paper soaked in freshly prepared 30% ferric chloride (FeCl3) was applied to the distal MCA at a bifurcation above the zygomatic arch and left in place for 5 min. A platelet-rich thrombus formed in situ and was allowed to develop for 45 min until it fully occluded the vessel. A laser Doppler probe (Moor Instruments) was used to monitor CBF. Successful occlusion was confirmed by a stable CBF reduction of approximately 70% or greater. During surgery, core body temperature was maintained at 37 ± 0.5℃. Buprenorphine (0.05 mg/kg) for analgesia and 0.3 ml saline were administered subcutaneously post-surgery. Animals were excluded from analyses if there was absence of stroke (n = 4).

as confirmed by a stable CBF reduction of approximately 70% or greater. During surgery, core body temperature was maintained at 37 ± 0.5℃. Buprenorphine (0.05 mg/kg) for analgesia and 0.3 ml saline were administered subcutaneously post-surgery. Animals were excluded from analyses if there was absence of stroke (n = 4). Infarct volume Under terminal isoflurane anesthesia, animals were perfused with PBS and subsequently perfuse-fixed with 4% paraformaldehyde (PFA). Brains were excised, post-fixed in 4% PFA, and then immersed in 30% sucrose for cryoprotection before freezing at −80℃. Coronal brain slices (30 µm) were sectioned on a freezing sledge microtome and mounted on gelatin-coated slides, and then stained with 1% cresyl violet. For dMCAo, infarcted areas, denoted by pale staining, were calculated by total area × thickness of section. For fMCAo, infarcted areas were identified and were drawn onto corresponding brain maps at eight pre-determined coronal levels (2.22, 1.54, 0.86, 0.14, −0.58, −1.22, −1.94, −2.7 mm according to Bregma).52 Total infarct volume was determined by calculating area under the curve using GraphPad Prism v7.

ckness of section. For fMCAo, infarcted areas were identified and were drawn onto corresponding brain maps at eight pre-determined coronal levels (2.22, 1.54, 0.86, 0.14, −0.58, −1.22, −1.94, −2.7 mm according to Bregma).52 Total infarct volume was determined by calculating area under the curve using GraphPad Prism v7. Immunohistochemistry Coronal brain sections (30 µm) were used for all subsequent immunohistochemistry procedures. Endogenous peroxidase activity was quenched with 1% H2O2, followed by 5% goat serum to prevent non-specific antibody binding. Anti-neutrophil serum SJC-4 (1:10,000) (kindly provided by Professor D. Anthony, University of Oxford, UK) was applied overnight at 4℃, followed by a biotinylated goat anti-rabbit secondary antibody (1:500; Vector Laboratories, UK). Amplification of signal was achieved by a Vectastain ABC-HRP kit (Vector Laboratories) and visualization of positive staining was achieved with 3,3′-diaminobenzidine (DAB) (Sigma-Aldrich, UK). For IgG staining, immunohistochemistry was performed as above but omitting the primary antibody step. Images were scanned on a Pannoramic 250 Flash III slide scanner (3DHISTECH, Hungary) and SJC-4-positive neutrophils were quantified at 5 × magnification in the entire area of striatum and/or cortex with Image J/Fiji. Counts were averaged over three coronal levels (0.86, 0.14, 0.58 mm according to Bregma)52 and expressed as neutrophils/section. IgG staining intensity in the infarcted hemisphere was calculated as the change in pixel contrast from the contralateral side and averaged across three coronal levels (0.14, −0.58, −1.94 mm according to Bregma).52

ver three coronal levels (0.86, 0.14, 0.58 mm according to Bregma)52 and expressed as neutrophils/section. IgG staining intensity in the infarcted hemisphere was calculated as the change in pixel contrast from the contralateral side and averaged across three coronal levels (0.14, −0.58, −1.94 mm according to Bregma).52 Plasma cytokine determination Blood was centrifuged at 1500 × g for 10 min to isolate plasma. Multiplex analysis of cytokine production was determined using LEGENDplex (Biolegend), according to the manufacturer's instructions. Samples were acquired on a FACSVerse flow cytometer (BD Biosciences) and analyzed using the LEGENDplex v8 software. In a minority of samples that did not yield detectable cytokine levels, these samples were assigned the lowest limit of detection. Preparation of single cell suspensions Blood, bone marrow, spleen, and sub-mandibular lymph nodes (SMLNs) were collected in RPMI 1640 medium (Sigma) supplemented with HEPES (1M; Sigma), Penicillin–Streptomycin (Sigma), and 3% fetal bovine serum (FBS) (Life Technologies). Spleens and SMLNs were mashed through a 70 µm filter (Fisherbrand, UK) to yield a single cell suspension, followed by red blood cell lysis (Lonza, UK). Blood was subjected to two red blood cell lyses. Bones were opened at the knee joint and centrifuged to collect the marrow, which subsequently underwent red blood cell lysis.

s and SMLNs were mashed through a 70 µm filter (Fisherbrand, UK) to yield a single cell suspension, followed by red blood cell lysis (Lonza, UK). Blood was subjected to two red blood cell lyses. Bones were opened at the knee joint and centrifuged to collect the marrow, which subsequently underwent red blood cell lysis. Flow cytometry Single cell suspensions were stained with Fc receptor block (TruStain fcX, Biolegend) along with the appropriate anti-mouse antibodies: CD45 (AF700; clone 30-F11, Biolegend), TNFα (AF700; clone MP6-XT22, Biolegend), CD11b (BV605 or BV650; clone M1/70, Biolegend), Ly6C (BV711; clone HK1.4; Biolegend), Ly6G (APC-Cy7; clone 1A8), CD115 (APC; clone AFS98, Biolegend) or PE-Cy7; AFS98, eBioscience), CD68 (PE; clone FA-11, Biolegend). A lineage gate was used to exclude other cell populations and consisted of the following biotinylated antibodies: CD3 (clone 145-2C11, Biolegend), CD19 (clone 6D5, Biolegend), NK1.1 (clone PK136, Biolegend), Ter119 (clone TER-119, Biolegend), and SiglecF (PE-CF594; E50-2440, BD Biosciences). A fixable stain (LIVE/DEAD Blue, Life Technologies) was used to exclude dead cells. For intracellular staining, cells were first stained for surface markers, fixed in 2% PFA, then permeabilized in 0.5% Saponin (Sigma), and stained for intracellular markers. Fluorescence-minus-one controls were used to validate results. Samples were acquired on a Fortessa flow cytometer (BD Biosciences) and analyzed using FlowJo software v10 (FlowJo LLC, USA).

st stained for surface markers, fixed in 2% PFA, then permeabilized in 0.5% Saponin (Sigma), and stained for intracellular markers. Fluorescence-minus-one controls were used to validate results. Samples were acquired on a Fortessa flow cytometer (BD Biosciences) and analyzed using FlowJo software v10 (FlowJo LLC, USA). Ex vivo re-stimulation for cytokine induction Bone marrow cells were stimulated with or without LPS (10 ng; O111:B4 Ultrapure, InvivoGen) in 10% FBS-containing media in the presence of Brefeldin A (5 µg/ml; BD Biosciences) for 2.5 h at 37℃ and then stained for flow cytometry. Data and statistical analyses All statistical analyses were performed using GraphPad Prism v7 using the appropriate test (stated in the figure legends). Data were transformed where appropriate. All data are presented as SEM. p < 0.05 was considered statistically significant. Sample sizes were determined by power calculation (α = 0.05, β = 0.2) of our previous data. All ex vivo quantifications were performed in a blinded manner. All reporting of animal experiments complied with the ARRIVE guidelines (Animal Research: Reporting In Vivo Experiments).

was considered statistically significant. Sample sizes were determined by power calculation (α = 0.05, β = 0.2) of our previous data. All ex vivo quantifications were performed in a blinded manner. All reporting of animal experiments complied with the ARRIVE guidelines (Animal Research: Reporting In Vivo Experiments). Results Ligature-induced PD causes robust pathology in the oral cavity Placement of ligatures around the second maxillary molars provides a reservoir for bacterial growth and also prevents habitual cleaning, culminating in profound bone loss within an acute timeframe. Indeed, within 10 days of ligature placement, there was a significant outgrowth of aerobic and anaerobic bacteria in the PD group (Figure 1(a)). Robust bone loss was also observed in the PD mice predominantly at sites around the second molar (Figure 1(b)). In the draining lymph nodes for the oral cavity, the sub-mandibular lymph nodes, there was a 6-fold increase in overall cellularity, and a significant influx of neutrophils and monocytes (Figure 1(c)). Combined, ligature placement was an effective model of PD, activating bacterial and host pathways that drive predictable periodontal bone loss in an acute timeframe. Figure 1. Ligature-induced periodontitis causes robust pathology in the oral cavity. To induce periodontitis (PD), mice were subjected to ligatures placed bilaterally around the second maxillary molars for 10 days or control surgery. (a) Oral swabs from periodontitis (PD) and control mice taken at day 10 were plated under aerobic conditions for 24 h or anaerobic conditions for 72 h. Scale bar, 1 cm. (b) i. Representative images of maxillae showing bone loss at day 10 post-ligature placement. Green line: alveolar bone crest (ABC). Red line: cemento-enamel junction (CEJ). Scale bar, 0.5 mm. (b) ii. Individual sites of bone loss (b) iii. Total bone loss was measured by sum of the CEJ-ABC distances across the six molar sites. (c) i. Sub-mandibular lymph nodes were excised and overall cellularity determined. (c) ii. Total numbers of neutrophils and monocytes were determined by flow cytometry (for gating strategy, see Figure 2(b)). (a–c): **p < 0.01, ***p < 0.001, by unpaired Student's t-test. (b) ii. *p < 0.05, **p < 0.01 by two-way ANOVA with Sidak's post hoc test). n = 4–5 per group. N.B.: one aerobic PD plate was overgrown and not able to be counted and thus omitted.

es were determined by flow cytometry (for gating strategy, see Figure 2(b)). (a–c): **p < 0.01, ***p < 0.001, by unpaired Student's t-test. (b) ii. *p < 0.05, **p < 0.01 by two-way ANOVA with Sidak's post hoc test). n = 4–5 per group. N.B.: one aerobic PD plate was overgrown and not able to be counted and thus omitted. Ligature-induced PD causes systemic inflammatory changes Clinically, PD is proposed to impact systemic health.53 However, there is limited evidence that ligature models in rodents induce inflammation outside the oral cavity. Here we show that there are systemic inflammatory consequences at 10 days after PD induction. The pro-inflammatory cytokine interleukin-1beta (IL-1β) was significantly elevated in the plasma of animals with experimental PD (Figure 2(a)). There was also a significant increase in both plasma IL-17A and granulocyte-macrophage colony-stimulating factor (GM-CSF) levels in the PD group, cytokines that mediate granulocyte and monocyte responses. Despite increases in GM-CSF and IL-17A, there was no difference in the overall number of bone marrow neutrophils, indicating that production of these cells was unaltered (Figure 2(c)). However, there was an increase in monocyte production in the bone marrow of PD animals (Figure 2(d)), and these monocytes were primed to produce more TNF-α (Figure 2(e)). Together, these data show that ligature-induced PD mediates specific inflammatory effects at peripheral sites distal from the oral cavity. Figure 2. Ligature-induced periodontitis causes systemic inflammatory changes. To induce periodontitis (PD), mice were subjected to ligatures placed bilaterally around the second maxillary molars for 10 days or control surgery. (a) Plasma cytokine levels were determined by LEGENDplex assay. Grey dotted lines indicate the limit of detection for this assay. (b) General flow cytometry gating strategy for bone marrow monocytes# (Live, CD45+, Lineage−, CD11b+, CD115+) and neutrophils (Live, CD45+, Lineage−, CD11b+, Ly6G+). (c,d) Frequency and numbers of neutrophils and monocytes in the bone marrow. (e) Ex vivo TNF-α production by bone marrow monocytes after 2.5 h stimulation with 10 ng LPS. ((a)–(d): *p < 0.05, **p < 0.01 by unpaired Student's t-test. (e) ***p < 0.001, by two-way ANOVA with Sidak's post hoc test). n = 4 per group, except (a) n = 9–10 per group. #For cytokine production, monocytes were gated CD68+ instead of CD115+.

roduction by bone marrow monocytes after 2.5 h stimulation with 10 ng LPS. ((a)–(d): *p < 0.05, **p < 0.01 by unpaired Student's t-test. (e) ***p < 0.001, by two-way ANOVA with Sidak's post hoc test). n = 4 per group, except (a) n = 9–10 per group. #For cytokine production, monocytes were gated CD68+ instead of CD115+. Ligature-induced PD does not alter ischemic brain damage after filament MCAo In order to evaluate the impact of PD on stroke outcome, mice were given ligature-induced PD for 10 days followed by fMCAo (Figure 3(a)). As expected, fMCAo resulted in infarction of the striatum, and extended to the cerebral cortex in some animals. However, there was no difference in infarct volume between mice with or without PD assessed at 48 h (Figure 3(b)). In addition, there was also no difference in sensory-motor functional assessment between stroked mice with or without PD (Figure 3(c)). BBB breakdown occurred after stroke yet IgG staining intensity in the ipsilateral hemisphere was not significantly affected by PD (Figure 3(d)). Despite alterations in myeloid trafficking in the periphery in PD-only animals (Figures 1 and 2), neutrophil infiltration in the ipsilateral hemisphere post-stroke was similar between the groups (Figure 3(e)). In the peripheral compartment, PD did not significantly modulate myeloid trafficking compared to the levels observed in the stroke-only group (Figure 3(f)), nor were plasma concentrations of IL-1β, IL-17A, and GM-CSF altered (Figure 3(g)). Further, the levels of B cells and T cells in the peripheral sites shown were also unchanged (data not shown). Taken together, these data show that PD does not modulate any of the assessed acute outcomes in a transient model of focal cerebral ischemia. Figure 3. Ligature-induced periodontitis does not alter ischemic brain damage after filament middle cerebral artery occlusion. Mice were subjected to either ligature-induced periodontitis (PD) or control surgery for 10 days upon which they were given a 20-min filament middle cerebral artery occlusion (fMCAo) and sacrificed 48 h later. (a) Experimental setup. (b) Total infarct volume was determined via cresyl violet staining. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (c) Neurological impairment at 48 h was evaluated by 28-point neurological score. (d) Blood–brain barrier breakdown assessed by IgG immunohistochemistry. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm.

ection with dotted line representing area of infarct. Scale bar, 1 mm. (c) Neurological impairment at 48 h was evaluated by 28-point neurological score. (d) Blood–brain barrier breakdown assessed by IgG immunohistochemistry. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (e) SJC-4 + neutrophils were counted in the striatum (Str.) and cortex (Ctx.). Scale bar, 500 µm. (f) Peripheral monocytes and neutrophil frequencies were assessed by flow cytometry. (g) LEGENDplex analysis of plasma cytokine levels. n = 10–11 per group. SMLNs: sub-mandibular lymph nodes.

ection with dotted line representing area of infarct. Scale bar, 1 mm. (c) Neurological impairment at 48 h was evaluated by 28-point neurological score. (d) Blood–brain barrier breakdown assessed by IgG immunohistochemistry. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (e) SJC-4 + neutrophils were counted in the striatum (Str.) and cortex (Ctx.). Scale bar, 500 µm. (f) Peripheral monocytes and neutrophil frequencies were assessed by flow cytometry. (g) LEGENDplex analysis of plasma cytokine levels. n = 10–11 per group. SMLNs: sub-mandibular lymph nodes. Ligature-induced PD does not alter ischemic brain damage after distal MCAo As ligature-induced PD did not affect acute outcome in a transient model of focal cerebral ischemia, we employed a permanent model to determine if experimental PD could affect stroke pathology in a permanent model of stroke (dMCAo) and consolidate the previous finding. We also intravenously administered repeated doses of LPS from P. gingivalis (Pg-LPS) concomitantly with ligature-induced PD to better reflect the systemic inflammatory changes reported in PD patients.54 After dMCAo, however, PD had no effect on ischemic brain damage or on BBB breakdown (Figure 4(b) and (c)). Administration of Pg-LPS also had no effect on infarct volume or BBB permeability in either control or PD mice. Neutrophil infiltration into the ipsilateral cortex was also unaffected by Pg-LPS and/or PD (Figure 4(d)). Moreover, in the blood, bone marrow, and spleen, neutrophils and monocytes were not significantly altered by either Pg-LPS and/or PD following dMCAo (Figure 4(e)). B and T cell levels in these peripheral sites were also unchanged (data not shown). Together, these data show that experimental PD does not alter acute outcome after dMCAo. Figure 4. Ligature-induced periodontitis does not alter ischemic brain damage after distal middle cerebral artery occlusion. Mice were subjected to either ligature-induced periodontitis (PD) or control surgery in tandem with intravenously administered repeated low-dose (1 mg/kg) Porphyromonas gingivalis LPS (Pg-LPS) or PBS for 10 days, upon which they were given a distal middle cerebral artery occlusion (dMCAo) and sacrificed 48 h later. (a) Experimental setup and dosing strategy. (b) Total infarct volume was determined via cresyl violet staining. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (c) Blood–brain barrier breakdown assessed by IgG immunohistochemistry. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (d) SJC-4 + neutrophils were counted in the cortex (Ctx.). Scale bar, 200 µm. (e) Peripheral monocytes and neutrophil frequencies were assessed by flow cytometry.

brain barrier breakdown assessed by IgG immunohistochemistry. Representative image of section with dotted line representing area of infarct. Scale bar, 1 mm. (d) SJC-4 + neutrophils were counted in the cortex (Ctx.). Scale bar, 200 µm. (e) Peripheral monocytes and neutrophil frequencies were assessed by flow cytometry. n = 5–9 per group, except (e) n = 3–5 per group. Discussion PD has been implicated as a factor for increased incidence of stroke but whether PD can modify ischemic brain damage is not fully understood. The present study shows that ligature-induced PD induced robust bone loss in an acute timespan accompanied by local and systemic inflammation that is comparable to the clinical condition. However, in both permanent (dMCAo) and transient (fMCAo) models of focal cerebral ischemia, PD did not alter infarct volume, BBB breakdown, or acute inflammatory processes.

duced PD induced robust bone loss in an acute timespan accompanied by local and systemic inflammation that is comparable to the clinical condition. However, in both permanent (dMCAo) and transient (fMCAo) models of focal cerebral ischemia, PD did not alter infarct volume, BBB breakdown, or acute inflammatory processes. The double ligature model of PD is an attractive tool for experimental periodontal research.49 Bone loss is the hallmark feature of PD in humans. These changes are also accompanied by bacterial outgrowth and local inflammation. We found that ligature-induced PD successfully mimics these clinical characteristics; bone loss occurred in an acute timeframe, in agreement with other reports,49,55,56 and this was accompanied by profound bacterial expansion and changes in local inflammatory cell trafficking. This model is thus more favorable than other established models of PD, such as oral inoculation with P. gingivalis, which does not always yield robust bone loss and require antibiotic pre-treatment in order for P. gingivalis successfully colonize.57–59 A predictable timescale with human-relevant pathology makes ligature-induced PD a tractable model for studying the pathophysiology of PD.

such as oral inoculation with P. gingivalis, which does not always yield robust bone loss and require antibiotic pre-treatment in order for P. gingivalis successfully colonize.57–59 A predictable timescale with human-relevant pathology makes ligature-induced PD a tractable model for studying the pathophysiology of PD. PD is attributed to increased risk of inflammatory diseases but direct causal evidence is lacking. In-depth analyses of peripheral tissues and immune populations have not been given much focus in ligature-induced rodent models. Here, we report that ligature-induced PD caused a specific increase in plasma IL-1β, IL-17A, and GM-CSF levels. These cytokines have been implicated as key drivers of PD pathogenesis,60 but IL-17A and GM-CSF have not been previously reported in the circulation in animal models. These changes during PD have potentially important implications for stroke as systemic inflammation unequivocally worsens ischemic damage,46,47 and IL-1β has a well-documented central role in the acute phases post-stroke.43,47,61 IL-17A has also been implicated in perpetuating damage after stroke; trafficking of IL-17+ γδ T cells to the ischemic brain is associated with worse outcome.62–64 In contrast, GM-CSF is reportedly neuroprotective after stroke,65 indicating a potentially complex role for the IL-17A:GM-CSF axis in ischemic brain damage.

IL-17A has also been implicated in perpetuating damage after stroke; trafficking of IL-17+ γδ T cells to the ischemic brain is associated with worse outcome.62–64 In contrast, GM-CSF is reportedly neuroprotective after stroke,65 indicating a potentially complex role for the IL-17A:GM-CSF axis in ischemic brain damage. Many studies have also highlighted the contribution of neutrophil and monocyte trafficking to the ischemic brain; acute neutrophil infiltration is associated with increased brain damage, whereas monocyte recruitment can be detrimental or beneficial.47,48,66,67 We found that neutrophils and monocytes were mobilized to local lymph nodes during PD, and monocytes in the bone marrow increased in number and were primed to produce elevated TNF-α. This suggests that one way in which PD could mediate peripheral inflammation is by affecting monocyte phenotype as monocytes can be primed prior to bone marrow egress.68 This is in agreement with a study by Miyajima et al.,69 in which ligature-induced PD promoted monocyte/macrophage activation and adherence to aortic walls. Interestingly, the authors also found elevated TNF-α signalling components in the aortas of PD mice,69 but no change in serum TNF-α levels, which agrees with our finding that plasma TNF-α levels did not change in PD mice, despite a change in monocyte phenotype. This indicates that the peripheral effects of experimental PD are “low-grade” and only affect certain cell populations or certain tissue sites and do not induce multi-site “high-grade” pro-inflammatory responses.

ith our finding that plasma TNF-α levels did not change in PD mice, despite a change in monocyte phenotype. This indicates that the peripheral effects of experimental PD are “low-grade” and only affect certain cell populations or certain tissue sites and do not induce multi-site “high-grade” pro-inflammatory responses. Though high-grade systemic inflammation is deleterious in the context of stroke,46–48 there are contrasting reports on whether low-grade inflammation is harmful or protective after cerebral ischemia. Obesity in mice (which is associated with low-grade inflammation) increases ischemic brain damage, BBB breakdown, and also increases neutrophil infiltration into the ischemic territory.3–5 However, in rats, mild systemic inflammation induced by ligature-induced PD has been reported to have neuroprotective effects, supressing macrophage accumulation in the brain after stroke.70 The latter study conflicts with our data where we show PD did not alter ischemic brain damage. We did not find elevated levels of plasma IL-10 and interferon (IFN)-γ in PD animals, in contrast to Petcu et al.70 The use of a different species (rat), longer PD timeframe (15–21 days), and a different stroke model as well as longer reperfusion time (7 days) could account for the differing results. However, our study has the advantage of using two different stroke models and a clinically relevant LPS to consolidate our findings.

0 The use of a different species (rat), longer PD timeframe (15–21 days), and a different stroke model as well as longer reperfusion time (7 days) could account for the differing results. However, our study has the advantage of using two different stroke models and a clinically relevant LPS to consolidate our findings. Transient MCAo by intraluminal filament is a well-characterized and well-established experimental murine model of stroke that mimics severe cerebral ischemia in humans.71 Importantly, ischemic damage and the inflammatory response can be modulated by the presence of co-morbidities, allowing dissection of the pathophysiological mechanisms after stroke.3,6,7 However, in the present study, PD did not drive ischemic damage above control levels; this is likely due to the fact that other studied co-morbidities involve major infection,6,7 or major metabolic shift (i.e. obesity),3,72 which may act as severe systemic stressors in the context of stroke in a manner that experimental PD does not. With this in mind, we sought to enhance the systemic inflammatory response associated with PD by administering the oral-specific Pg-LPS, as over 50% of PD patients have Pg-LPS in the bloodstream.54 Additionally, a permanent occlusion model (dMCAo) was employed, wherein the infarcts are smaller and confined to the cortex.73 However, neither ligature placement nor Pg-LPS affected ischemic brain damage in this model. Other in vivo studies have used lower efficacious Pg-LPS dosing regimes than that of the present study,74 which would indicate that biological efficacy of the Pg-LPS dose is not in question. However, it is plausible that tolerance to Pg-LPS, which can occur with repeated low doses,75,76 could explain the lack of significant change in stroke severity.

lower efficacious Pg-LPS dosing regimes than that of the present study,74 which would indicate that biological efficacy of the Pg-LPS dose is not in question. However, it is plausible that tolerance to Pg-LPS, which can occur with repeated low doses,75,76 could explain the lack of significant change in stroke severity. The acute timeframe of the PD model must be considered, as this may not lead to sufficiently chronic inflammatory changes to affect stroke pathophysiology. Certainly, human PD is a chronic condition associated with transient bacteremia and low-level inflammation that occurs over a long period.40 However, ligatures left in place for longer would risk tooth loss and thus a more suitable PD model may need to be developed that encapsulates local and systemic pathology in a chronic manner. Importantly, the mice used in this study were young and otherwise healthy, and most individuals with PD have other confounding factors (smoking, hypertension, old age, obesity) which can independently account for increased stroke risk.2 As experimental PD did not alter outcome in the present study, it indicates that PD alone is not sufficient to modulate the pathological changes associated with stroke. It remains to be determined, however, if PD increases the likelihood of stroke, if PD affects long-term recovery after stroke, and also, if in tandem with other co-morbidities PD worsens prognosis post-stroke. Such questions are difficult to address in experimental animal models, but given the high prevalence of PD and the growing incidence of stroke, they are important avenues that require future study.

ffects long-term recovery after stroke, and also, if in tandem with other co-morbidities PD worsens prognosis post-stroke. Such questions are difficult to address in experimental animal models, but given the high prevalence of PD and the growing incidence of stroke, they are important avenues that require future study. Acknowledgements This work was also made possible through use of the University of Manchester Flow Cytometry Core Facility, Bioimaging Facility, and Biological Services Facility. We thank Hayley Bridgeman and Siddharth Krishnan for assistance with experimental workloads. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Medical Research Council, Biotechnology and Biological Sciences Research Council (grant BB/M025977/1) and the Kohn Foundation.

Penetrating artery infarcts that are predominantly caused by occlusion at the vessel orifices of larger caliber penetrating arteries by atheromatous plaque can represent a distinctive stroke entity as intracranial branch atheromatous disease (BAD) 1. BAD often shows progressive motor deficits leading to severe disability 2. Thrombolytic therapy by tissue plasminogen activator has not been demonstrated to be effective for BAD 3. We designed a combined use of antiplatelet agents for BAD. During 12 years, 313 consecutive patients with BAD located within the territories of lenticulostriate arteries (LSAs) and anterior pontine arteries (APAs) were prospectively collected. Treatment protocols are as follows: phase 1 (2001–2005, n = 105), the medical treatment that was considered best; phase 2 (2005–2009, n = 104), a combined treatment of argatroban, cilostazol, and edaravone; and phase 3 (2009–20012, n = 104), additional clopidogrel on top of the phase 2 protocol. Functional outcome was assessed by the modified Rankin scale (mRS) at one-month after stroke onset. As a result, in the total population, the phase 2 and the phase 3 showed better outcome than the phase 1 (P = 0·0004 and P < 0·0001). In the LSA infarct group, the phase 2 and the phase 3 showed better outcome than the phase 1 (P = 0·046 and P = 0·0001). The phase 3 showed better outcome than the phase 2 (P = 0·018). In the APA infarct group, the phase 2 and the phase 3 showed better outcome than the phase 1 (P = 0·0004 and P = 0·0006) (see Fig. 1). Cilostazol appeared to be more effective for the infarcts of the APA that branch off from the basilar artery with a diameter of 200–300 micrometers, whereas clopidogrel is more effective for the infarcts of the LSA of 700–800 micrometers that turns sharply and forms a curve or a real loop. The actions of vasodilatation and endothelial protection in cilostazol and inhibition of shear-induced platelet activation in clopidogrel might work effectively 4,5.

micrometers, whereas clopidogrel is more effective for the infarcts of the LSA of 700–800 micrometers that turns sharply and forms a curve or a real loop. The actions of vasodilatation and endothelial protection in cilostazol and inhibition of shear-induced platelet activation in clopidogrel might work effectively 4,5. Figure 1 Upper LSA group, lower APA group.

Background Stroke is one of the leading causes of death and the most frequent reason for permanent disability (1). Thrombolysis with recombinant tissue plasminogen activator (rt-PA) is the only approved and causal therapy for acute ischemic stroke (2). The benefit of this therapy is, however, extremely time sensitive: The number needed to treat to achieve a good outcome is 4·5 if treatment starts within 1·5 h. This number doubles to 9 if treatment is initiated within 1·5 to 3·0 h and reaches 14·1 if treatment occurs within the temporal window of 3·0 to 4·5 h (3). The ‘time is brain’ concept that has been derived from such observations is also supported by earlier experimental animal research (4,5) and by calculations indicating that for each minute in which a stroke remains untreated, as many as 1·9 million neurons and 14 billion synapses may die (6). However, before rt-PA can be administered, a complex diagnostic workup, including neurological examination, imaging studies, and laboratory tests, is necessary to exclude hemorrhage or other contraindications to rt-PA therapy. For this reason, treatment within the narrow temporal window of a few hours is difficult to achieve in clinical reality, and, in the end, no more than 2% to 7% of all acute stroke patients currently receive treatment (7).

laboratory tests, is necessary to exclude hemorrhage or other contraindications to rt-PA therapy. For this reason, treatment within the narrow temporal window of a few hours is difficult to achieve in clinical reality, and, in the end, no more than 2% to 7% of all acute stroke patients currently receive treatment (7). The blame for this problem of undertreatment with thrombolysis can be placed primarily on activities that occur before the patient reaches the doors of the hospital. In Germany, for example, the median prehospital time is 151 mins, and only 45% of patients reach the hospital within three-hours (8). Data from the American Get With the Guidelines Stroke Registry clearly show that, despite considerable attempts to improve stroke management (e.g., by public education programs), delays in the time to hospital admission did not improve in recent years (9) [for a detailed systematic literature review on prehospital delays, see Evenson et al. (10)]. Methods This review examined reports published since 1980 and found by searching PubMed for articles containing the terms ‘stroke management’; ‘prehospital’ and ‘stroke’; ‘stroke’ and ‘educational campaign(s)’; ‘stroke’ and ‘public awareness’; and ‘emergency medical service’ and ‘stroke’. Articles were selected on the basis of their originality and their relevance to the topic of prehospital stroke management.

rticles containing the terms ‘stroke management’; ‘prehospital’ and ‘stroke’; ‘stroke’ and ‘educational campaign(s)’; ‘stroke’ and ‘public awareness’; and ‘emergency medical service’ and ‘stroke’. Articles were selected on the basis of their originality and their relevance to the topic of prehospital stroke management. Results and discussion Role of patients and relatives Problems in prehospital stroke management can be attributed to two groups: the patients and their families, and the emergency medical services (EMS) team. With regard to the role of the patients and their relatives, it is important to consider that 24% to 55% of acute stroke patients or their relatives do not notify the EMS within one-hour, but rather use a private vehicle to transport the patient to the hospital, visit their family doctor, wait too long, or do not notify anyone (11–13). Many reasons for this inadequate response to stroke symptoms have been described and may be demographic, social, medical, or psychological in nature [(11–19), Table 1]. Table 1 Determinants of care-seeking behavior in acute stroke

Results and discussion Role of patients and relatives Problems in prehospital stroke management can be attributed to two groups: the patients and their families, and the emergency medical services (EMS) team. With regard to the role of the patients and their relatives, it is important to consider that 24% to 55% of acute stroke patients or their relatives do not notify the EMS within one-hour, but rather use a private vehicle to transport the patient to the hospital, visit their family doctor, wait too long, or do not notify anyone (11–13). Many reasons for this inadequate response to stroke symptoms have been described and may be demographic, social, medical, or psychological in nature [(11–19), Table 1]. Table 1 Determinants of care-seeking behavior in acute stroke Factors Early alarm Late alarm References Demographic Women Men (11) High level of education Low level of education (12,14,15) High income Low income Ethnic minorities (16) Social Presence of bystanders Being alone (12,17,18) Medical Family history of stroke No family history (13,18) Severe symptoms Mild symptoms Acute onset Delayed onset Psychological Fear of disease and hospital (11,19) Stroke educational campaigns could, therefore, be a solution for improving patients' and families' awareness of the correct response in case of stroke. Such campaigns have been shown to have a short-term impact on stroke symptom knowledge. However, consistent evidence confirms an existing gap between the knowledge and recognition of stroke symptoms and the appropriate urgent response to such symptoms (for detailed systematic literature reviews, see Teuschl & Brainin, Jones et al., and Lecouturier et al. (20–22)).

-term impact on stroke symptom knowledge. However, consistent evidence confirms an existing gap between the knowledge and recognition of stroke symptoms and the appropriate urgent response to such symptoms (for detailed systematic literature reviews, see Teuschl & Brainin, Jones et al., and Lecouturier et al. (20–22)). A very recent study consisting of individual semi-structured interviews with stroke patients, stroke witnesses, and primary care clinicians examined the perceived impact of and views about the United Kingdom's mass media campaign Act FAST. Most participants were aware of the Act FAST campaign, and some patients and witnesses reported that the campaign affected their stroke recognition and response, but most reported no effect. Clinicians were positive about the campaign and believed that it had affected stroke awareness and recognition, but doubted its impact on response behavior (23).

the Act FAST campaign, and some patients and witnesses reported that the campaign affected their stroke recognition and response, but most reported no effect. Clinicians were positive about the campaign and believed that it had affected stroke awareness and recognition, but doubted its impact on response behavior (23). Only a few of these existing studies, however, analyzed the effects on clinically relevant end-points of stroke management, such as time to hospital admission or thrombolysis rates [(24–31), Table 2]. These variables are examined primarily in interventional studies with a noncontrolled ‘before and after intervention’ design. Only two of the currently existing studies applied a controlled design, including control groups of patients from catchment areas without such interventions (Table 2). Most of these studies revealed the effects of public awareness campaigns on patients' behavior (Table 2); however, these effects probably last for no more than five-months. Because of studies showing that the nature of the effects of such campaigns is rather transient (26), constant repetition of their message is a central precondition for their success (26). Table 2 Studies on the effect of public awareness campaigns on indicators of stroke management quality

Only a few of these existing studies, however, analyzed the effects on clinically relevant end-points of stroke management, such as time to hospital admission or thrombolysis rates [(24–31), Table 2]. These variables are examined primarily in interventional studies with a noncontrolled ‘before and after intervention’ design. Only two of the currently existing studies applied a controlled design, including control groups of patients from catchment areas without such interventions (Table 2). Most of these studies revealed the effects of public awareness campaigns on patients' behavior (Table 2); however, these effects probably last for no more than five-months. Because of studies showing that the nature of the effects of such campaigns is rather transient (26), constant repetition of their message is a central precondition for their success (26). Table 2 Studies on the effect of public awareness campaigns on indicators of stroke management quality References Site (number of stroke centers) Study design Study duration Number of stroke patients with and without intervention Target group Time until hospital admission with and without intervention Thrombolysis rates with and without intervention General public EMS Alberts et al. (24) Durham, United States (1) Before vs. after implementation Three-years 189 vs. 290 + + Within 24 h: 86% vs. 37%, P < 0·00001 – Wojner-Alexandrov et al. (25) Houston, TX, United States (6) Before vs. after implementation Three-years 1072 vs. 446 + + Within two-hours: 62% vs. 58%, P = 0·002 Increase in 4/6 centers, decrease in 2/6 Hodgson et al. (26) Ontario, Canada (11) Longitudinal observation 31 months 12534 + – within 2·5 h: continuous increase between 2003 and 2005, R2 = 0·60, P < 0·001 – Morgenstern et al. (27) Texas, United States (10) Controlled observation 15 months interventional region: 400 vs. 218; control region: 365 vs. 206 + + Within two-hours: 36·5% vs. 26·5%, P < 0·05 in the interventional region and 30·4% vs. 21·4%, P = 0·07 in the control region 5·8% vs. 1·4%, P < 0·05 in the interventional region vs. 0·6 vs. 0·5%, n. s. in the control region Barsan et al. (28) United States (12) Before vs. after implementation Three-years 487 vs. 487 + + 1·5 h vs. 3·2 h (means, P < 0·05) – Müller-Nordhorn et al. (29) Berlin, Germany (3) Controlled observation One-year 647 vs. 741 + – Within three-hours: 34% vs. 28% (significant only for women, acceleration factor 0·73; 95% CI: 0·58–0·94; P < 0·05) – Addo et al. (16) London, UK (1) Before vs. after implementation Two-years 154 vs. 195 + + Within three-hours: 44·9% vs. 40·7%, n.s. 16·4% vs. 16·9%, n.s. Behrens et al. (30) Mannheim, Germany (1) Before vs. after implementation Three-months 113 vs. 83 + + 3·28 h ± 40 min vs. 5·22 h ± 84 min (means ± SD, P < 0·05) 10·5% vs. 2%, P < 0·01 Rau et al. (31) Wesel, Germany (8) Before vs. after implementation Two-years 375 vs. 326 + – Within three-hours: 27·5% vs. 27·3%, n.s. – EMS, emergency medical services; n.s., not significant.

er implementation Three-months 113 vs. 83 + + 3·28 h ± 40 min vs. 5·22 h ± 84 min (means ± SD, P < 0·05) 10·5% vs. 2%, P < 0·01 Rau et al. (31) Wesel, Germany (8) Before vs. after implementation Two-years 375 vs. 326 + – Within three-hours: 27·5% vs. 27·3%, n.s. – EMS, emergency medical services; n.s., not significant. These findings suggest that public education efforts are worthwhile, and future efforts should focus more strongly on specific target groups, such as the elderly, minorities, neighbors of stroke survivors, medical students, and even children (who may be future relatives, patients, or physicians) (32–36). Importantly, educational campaigns should present in a very simple message the action that should be undertaken in case of emergency (e.g., ‘call EMS immediately’). However, rather than relying on a fear appeal alone, which is more likely to cause people to stop a behavior rather than to perform an action, such a message should also be encouraging. Thus, a still-ongoing campaign, the National Stroke Association's Faces of Stroke multimedia public awareness campaign (37) presents the personal and emotional side of stroke, including stories of stroke survivors and caregivers, thereby aiming to generate empathic feelings and to transmit the positive message of the existence of treatment. In general, the major problem of undertreatment of stroke and change of behavior is a highly interdisciplinary task, including, for example, the contribution of health behavior scientists in assessing concepts of behavior change.

rate empathic feelings and to transmit the positive message of the existence of treatment. In general, the major problem of undertreatment of stroke and change of behavior is a highly interdisciplinary task, including, for example, the contribution of health behavior scientists in assessing concepts of behavior change. Role of the EMS Structures of EMS systems differ not only between countries but also within single countries. This is especially the case with regard to the variable disposition of an emergency physician when acute stroke is suspected. However, a large body of evidence already exists for some measures of acute stroke management, and, therefore, clear recommendations have been given by national and international stroke management guidelines (38,39). These guidelines include the recommendation of continued education of the EMS team regarding the use of instruments for the recognition of stroke symptoms, the emergency transport of patients to a hospital with stroke expertise, and, finally, the prenotification of the receiving hospital.

management guidelines (38,39). These guidelines include the recommendation of continued education of the EMS team regarding the use of instruments for the recognition of stroke symptoms, the emergency transport of patients to a hospital with stroke expertise, and, finally, the prenotification of the receiving hospital. Use of instruments for symptom recognition It has been shown that the accuracy with which EMS dispatchers identify stroke symptoms is highly variable, ranging from 30% to 83% (40,41); this finding highlights the need for continued further training. Most national and international stroke guidelines recommend, apart from medical training, the use of structured interviews by the dispatcher and the application of instruments designed for recognition of stroke symptoms by the EMS team in the field. For example, the Cincinnati Prehospital Stroke Scale (sensitivity of 90% and specificity of 66% for the presence of acute stroke) is based on the presence of facial paresis, one-sided paresis of an upper extremity, and speech disorder (by asking the patients to repeat specific sentences) (42). The Los Angeles Prehospital Stroke Screen includes, in addition to these items, four additional questions about history and the results of a glucose test (43). The sensitivity (91%) and specificity (97%) of this scale are very high (43); however, this scale is quite complex for routine daily use and is also time consuming. The Face Arm Speech Time (FAST) Stroke Assessment is based on the three elements of the Cincinnati Prehospital Stroke Scale, but it assesses possible speech disorders during normal conservation (44). This scale has a sensitivity of 79% and is so easy to perform that it can even be used by the general public.

consuming. The Face Arm Speech Time (FAST) Stroke Assessment is based on the three elements of the Cincinnati Prehospital Stroke Scale, but it assesses possible speech disorders during normal conservation (44). This scale has a sensitivity of 79% and is so easy to perform that it can even be used by the general public. Whereas in the United States either the Los Angeles Prehospital Stroke Screen or the Cincinnati Prehospital Stroke Scale is most frequently used, in Europe the FAST Scale is most widely distributed; and in Australia the Melbourne Stroke Screen (45) is most often used. In general, however, the daily routine of the dispatcher and of the EMS team is still characterized by the mostly nonsystematic and undocumented use of such instruments; this finding underscores the importance of further training efforts.

istributed; and in Australia the Melbourne Stroke Screen (45) is most often used. In general, however, the daily routine of the dispatcher and of the EMS team is still characterized by the mostly nonsystematic and undocumented use of such instruments; this finding underscores the importance of further training efforts. Prioritized transport to hospitals with stroke expertise Prioritized transport to hospitals with stroke expertise and with the option for thrombolytic treatment is crucial in optimized prehospital stroke management. Such measure has been shown to reduce time to treatment and to increase thrombolysis rates without negatively influencing the quality of treatment of other emergencies (46). An additional reason for this recommendation is the strong evidence of a general benefit for treatment in a specialized stroke unit. In specific settings, the use of helicopters for transport to more distant stroke centers can also save critical time to treatment (47). A recent prospective multicentric study showed that treatment rates increase from 14·1% to 21·9% (OR, 1·72; 95% CI, 1·22–2·43) if patients are transported to stroke centers rather than to nonspecialized institutions (48). This is even the case if the distance to a stroke center is considerably greater than that to a nonspecialized hospital.

ulticentric study showed that treatment rates increase from 14·1% to 21·9% (OR, 1·72; 95% CI, 1·22–2·43) if patients are transported to stroke centers rather than to nonspecialized institutions (48). This is even the case if the distance to a stroke center is considerably greater than that to a nonspecialized hospital. Most advanced stroke management protocols, such as a city-wide protocol implemented in Toronto, Ontario, Canada (49), also include, apart from the use of standardized screening systems by the EMS team, the implementation of protocols for bypassing hospitals without stroke expertise. Such a protocol can be achieved by innovative regional cooperation, with contracts between hospitals regarding the later repatriation of the patients.

9), also include, apart from the use of standardized screening systems by the EMS team, the implementation of protocols for bypassing hospitals without stroke expertise. Such a protocol can be achieved by innovative regional cooperation, with contracts between hospitals regarding the later repatriation of the patients. Key role of prenotification Apart from the fact that the transmission of information regarding the onset of symptoms or thrombolysis contraindications is an integral component of the initial interaction between the EMS team and the hospital stroke team, stroke management guidelines (38,39) additionally recommend prenotification of the receiving hospital about the arriving patient. This prenotification allows the fastest possible activation of the stroke team and, especially, the reservation of computed tomography (CT) scanners. Previous interventional studies of the effects of prenotification of the hospital team alone or in combination with further restructuring of stroke management plans showed that crucial time to therapy can be saved and thrombolysis rates can significantly be increased [(50–57), Table 3]. The existing studies compared findings regarding the effects of a prenotification intervention either with findings from a historical control group or with findings from a parallel observation of patients for whom no such prenotification intervention was used (Table 3). So far, no data from randomized studies are available, and, in light of ethical aspects and because the existing studies already consistently show a considerable acceleration of in-hospital treatment, it is unlikely that such studies will ever be performed in the future. Regarding methodology, the transfer of structured clinical data between the EMS team at the emergency site and the hospital stroke team could also be optimized by the use of personal digital assistants (58) or smart phones (59).

-hospital treatment, it is unlikely that such studies will ever be performed in the future. Regarding methodology, the transfer of structured clinical data between the EMS team at the emergency site and the hospital stroke team could also be optimized by the use of personal digital assistants (58) or smart phones (59). Table 3 Studies on the effect of prenotification on stroke management quality

-hospital treatment, it is unlikely that such studies will ever be performed in the future. Regarding methodology, the transfer of structured clinical data between the EMS team at the emergency site and the hospital stroke team could also be optimized by the use of personal digital assistants (58) or smart phones (59). Table 3 Studies on the effect of prenotification on stroke management quality References Site (number of stroke centers) Study design Year Number of stroke patients with and without intervention Intervention Onset-to-door time (min) with and without intervention Thrombolysis rates (%) with and without intervention Belvís et al. (50) Barcelona, Spain (1) Parallel observation 2001–2002 39 vs. 181 Prenotification Mean (SD): 64·6 (37·8) vs. 69·4 (44·6), P = 0·542 19 vs. 4·5, P = 0·003 Abdullah et al. (51) Bοston, United States (1) Before vs. after implementation 2004–2005 44 vs. 74 Prenotification Median (IQR): 66 (42–126) vs. 90 (42–174), P = 0·42 41 vs. 21, P = 0·04 Quain et al. (52) Newcastle, Australia (1) Before vs. after implementation 2005–2007 232 vs. 205 Prenotification plus bypass protocol Median (IQR): 90·5 (63–185) vs. 150 (93–339), P = 0·004 21·4 vs. 4·7, P < 0·001 Kim et al. (53) Busan, Korea (1) Before vs. after implementation 2006–2007 328 vs. 678 Prenotification plus in-hospital reorganization Mean (SD): 121·5 (34·8) vs. 74·7 (38·5), P < 0·01* 14·3 vs. 6·5, no P values indicated Köhrmann et al. (54) Erlangen, Germany (1) Longitudinal observation 2006–2009 246* (without control group) Prenotification, plus EMS education, in-hospital reorganization Median (IQR): 72·5 (52–99) – Gladstone et al. (49) Toronto, Canada (3) Before vs. after implementation 2004–2005 290 vs. 217 Prenotification plus EMS screening tool, ambulance destination decision rule with bypass protocol Median (IQR): 63 (30) vs. 46 (7), P = 0·83 23·4 vs. 9·5, P = 0·01 O'Brien et al. (55) Gosford, Australia (1) Before vs. after implementation 2006–2008 115 vs. 67 Prenotification plus prehospital assessment tool, bypass protocol, in-hospital reorganization Mean†: 76 vs. 59, P = 0·18* 19 vs. 7, P = 0·03 Meretoja et al. (56) Helsinki, Finland (1) Before vs. after implementation 1998–2011 167 in 2011 vs. 7 in 1998* Prenotification plus EMS education, use of stroke recognition tools, in-hospital reorganization Median (IQR): 89 (62–138) vs. 75 (45–145)* 31 in 2011, no earlier data Casolla et al. (57) Lille, France (1) Parallel observation 2008–2011 191 vs. 56 Prenotification Median (IQR): 81 (61–120) vs.

167 in 2011 vs. 7 in 1998* Prenotification plus EMS education, use of stroke recognition tools, in-hospital reorganization Median (IQR): 89 (62–138) vs. 75 (45–145)* 31 in 2011, no earlier data Casolla et al. (57) Lille, France (1) Parallel observation 2008–2011 191 vs. 56 Prenotification Median (IQR): 81 (61–120) vs. 97 (49–144), P = 0·628 – * Only rt-PA-treated patients were included. † No SD displayed. EMS, emergency medical services; IQR, interquartile range; rt-PA, recombinant tissue plasminogen activator. Interestingly, one study found that combining the concept of prenotification with additional improvements in in-hospital stroke management resulted in a ‘door-to-needle’ time of only 20 mins (56). The philosophy behind these very short door-to-needle times is, according to the Finnish authors, ‘to do as little as possible after the patient has been arrived in the clinic, and as much as possible when the patient is on the way to the clinic’ (56). Telemedicine interaction between regional hospital and stroke center and perspectives for communication between emergency site and stroke center By using systems for bidirectional audiovisual videoconferencing and exchange of videos of the examination of the patient and of CT scans, nonspecialized regional hospitals can already obtain guidance in stroke treatment from hospitals designated as stroke centers [(60), Fig. 1]. Previous studies not only showed that such telemedicine interaction between two hospitals is reliable and safe (61–64), but also that it exerts positive effects on thrombolysis rates and clinical outcome (65).

ls can already obtain guidance in stroke treatment from hospitals designated as stroke centers [(60), Fig. 1]. Previous studies not only showed that such telemedicine interaction between two hospitals is reliable and safe (61–64), but also that it exerts positive effects on thrombolysis rates and clinical outcome (65). Figure 1 Strategies of acute stroke management. (a) Conventional stroke management; (b) optimized stroke management with prenotification of the receiving hospital and in-hospital reorganization with diagnostic workup and treatment at one site; (c) telemedicine interaction between regional hospital and stroke center, allowing the stroke center to provide guidance in acute treatment; (d) prehospital stroke treatment in an ambulance equipped with a CT scanner, point-of-care laboratory, and telemedicine interaction with the stroke center. CT, computed tomography; EMS, emergency medical services; MSU, mobile stroke unit; POC, point-of-care. Importantly, such telemedicine technologies could in principle also allow bidirectional communication between the EMS team at the emergency site and the stroke center. Such strategies have been investigated for many years (66,67); however, technical problems such as the temporary loss of signals still today impair reliable interaction between ambulance and hospital (68,69).

nciple also allow bidirectional communication between the EMS team at the emergency site and the stroke center. Such strategies have been investigated for many years (66,67); however, technical problems such as the temporary loss of signals still today impair reliable interaction between ambulance and hospital (68,69). Prehospital stroke treatment: an alternative strategy for reducing delay to treatment? To achieve the goal of enabling more than a minority of stroke patients to profit from recanalization therapies in the future, the concept of prehospital stroke treatment has been elaborated upon in the last decade [(70), Fig. 1]. This concept is based on the use of an otherwise conventional ambulance that, additionally, contains a small CT scanner and a point-of-care laboratory {mobile stroke unit [(71), Fig. 2]}. This ambulance also contains equipment that allows telemedicine interaction with the hospital, thereby making possible bidirectional communication and transfer of videos or CT scans of patients from the emergency site to the hospital. Figure 2 Ambulance for prehospital stroke treatment. (a) External view of the second-generation mobile stroke unit (in front) with integrated multimodal CT scanner, point-of-care laboratory, and telemedicine capability; the unit is sized as a conventional ambulance (in back). (b) Interior view. (c) Exemplary computed tomography scan obtained at the emergency site.

tment. (a) External view of the second-generation mobile stroke unit (in front) with integrated multimodal CT scanner, point-of-care laboratory, and telemedicine capability; the unit is sized as a conventional ambulance (in back). (b) Interior view. (c) Exemplary computed tomography scan obtained at the emergency site. A first randomized trial involving 100 patients recently showed that, compared with optimized conventional stroke management, prehospital stroke treatment reduces the time between alarm and therapy decision from 76 to 35 mins (72). A therapy decision was made within 60 mins after symptom onset (the ‘golden hour’) for 57% of the patients treated in the mobile stroke unit but for only 4% of the conventionally treated patients. This and other studies of stroke treatment directly at the emergency site already document a broad spectrum of additional novel medical options, especially the option of triaging patients to the most appropriate hospital on the basis of a diagnosis clarified before the patients are transported. For example, patients with large-vessel occlusion demonstrated by prehospital CT angiography could specifically be triaged to specialized stroke centers that offer endovascular treatment (73). Moreover, this strategy allows organization of further specialized treatments and etiology-specific blood pressure management already in the prehospital phase of stroke management (73–75). The latter could be specifically clinically relevant because there are indications that differential adjustment of blood pressure can be beneficial for patients with ischemic stroke (tolerating higher blood pressure values) or hemorrhagic stroke (reducing elevated blood pressure) (76).

ase of stroke management (73–75). The latter could be specifically clinically relevant because there are indications that differential adjustment of blood pressure can be beneficial for patients with ischemic stroke (tolerating higher blood pressure values) or hemorrhagic stroke (reducing elevated blood pressure) (76). In the future, the concept of prehospital stroke treatment could be complemented by the inclusion of other diagnostic and therapeutic strategies, such as further imaging procedures, neuroprotective strategies, or future hemorrhage therapies. However, currently, prehospital stroke treatment is still a potential perspective rather than clinical reality, and further research is still needed with regard to the medical efficacy of and the best setting for this concept. Finally, although many arguments suggest that increased allocation of resources in the golden hour of stroke could save the much higher costs of long-term care of disabled patients, the cost-effectiveness of prehospital stroke treatment remains to be demonstrated. As limitation, this review on the different links of the prehospital stroke rescue chain did not completely follow a systematic approach as articles found by PubMed search were selected and weighted based on their originality and their relevance to the topic of prehospital stroke management.

In the future, the concept of prehospital stroke treatment could be complemented by the inclusion of other diagnostic and therapeutic strategies, such as further imaging procedures, neuroprotective strategies, or future hemorrhage therapies. However, currently, prehospital stroke treatment is still a potential perspective rather than clinical reality, and further research is still needed with regard to the medical efficacy of and the best setting for this concept. Finally, although many arguments suggest that increased allocation of resources in the golden hour of stroke could save the much higher costs of long-term care of disabled patients, the cost-effectiveness of prehospital stroke treatment remains to be demonstrated. As limitation, this review on the different links of the prehospital stroke rescue chain did not completely follow a systematic approach as articles found by PubMed search were selected and weighted based on their originality and their relevance to the topic of prehospital stroke management. In summary, this review indicates that prehospital delay is a major reason that only a minority of patients obtain recanalizing therapy today. Options for the improvement of prehospital stroke management include, on the patient's side, target-specific and continual public awareness campaigns. On the EMS side, frequently repeated training in the use of tools for symptom recognition, prioritized transport to experienced stroke centers, and, probably most important, prenotification of the receiving hospital can improve time to treatment and treatment rates. In the future, even prehospital stroke treatment could contribute to better stroke management.

Introduction and rationale Currently, intravenous (i.v.) thrombolysis with recombinant tissue-type plasminogen activator (rt-PA) is the only evidence-based effective treatment for acute ischemic stroke within 4·5 h after symptom onset (1,2). Because the therapeutic time range for effective i.v. rt-PA therapy is limited, onset-to-treatment time is vital. Delayed thrombolysis may cause a life-threatening condition such as symptomatic intracranial hemorrhage (sICH). Because accurate onset-to-treatment time is not determined for patients with unclear-onset time of stroke symptoms or those who wake up with stroke symptoms, i.v. rt-PA therapy is usually contraindicated for these patients. According to previous reports, about one-fourth of acute stroke patients suffer from stroke symptoms with unclear-onset time or onset during sleep (3–5). These patients were reported to have similar early ischemic findings on initial computed tomography (CT) or magnetic resonance imaging (MRI) compared with those presenting within three-hours (6) or six-hours of symptom recognition (7). Because onset of all subtypes of stroke has an early morning peak (8), a large number of patients who wake up with stroke symptoms may still be within the time window for i.v. thrombolysis when they arrive at the hospital.

I) compared with those presenting within three-hours (6) or six-hours of symptom recognition (7). Because onset of all subtypes of stroke has an early morning peak (8), a large number of patients who wake up with stroke symptoms may still be within the time window for i.v. thrombolysis when they arrive at the hospital. Recently, stroke MRI findings with positive diffusion-weighted imaging (DWI) and no marked parenchymal hyperintensity on fluid-attenuated inversion recovery (FLAIR) (negative FLAIR pattern) was proposed to act as a ‘brain clock’ that indicates the stroke occurred within 3–4·5 h from stroke symptom onset (9–11). DWI depicts a reduced apparent diffusion coefficient indicating cytotoxic edema caused by ischemia within minutes of stroke (12). FLAIR is characterized by strong T2 weighting together with suppression of cerebrospinal fluid signal and is often positive 3–4·5 h after stroke onset (9–11,13).

stroke symptom onset (9–11). DWI depicts a reduced apparent diffusion coefficient indicating cytotoxic edema caused by ischemia within minutes of stroke (12). FLAIR is characterized by strong T2 weighting together with suppression of cerebrospinal fluid signal and is often positive 3–4·5 h after stroke onset (9–11,13). There are several reports of nonrandomized patients on thrombolysis based on CT or MRI findings in acute stroke patients with unclear-onset time. Most reports had retrospective study designs and suggested the safety and feasibility of thrombolysis for this subgroup of stroke patients (14–19). In Japan, a single-center, nonrandomized, prospective study involving 20 patients with unclear-onset time and negative FLAIR pattern on initial MRI showed no occurrence of sICH after i.v. rt-PA at 0·6 mg/kg alteplase; at 90 days, 47% of patients showed modified Rankin Scale (mRS) scores of 0–2 (15). However, there is no established data from randomized, controlled trials of i.v. rt-PA in patients with unclear-onset time. To our knowledge, there are six ongoing clinical trials to test the safety and efficacy of thrombolysis in acute stroke patients with unclear-onset time worldwide: the Safety of Intravenous Thrombolytics in Stroke on Awakening (SAIL-ON) (ClinicalTrials.gov Identifier NCT01643902), the Safety of Intravenous Thrombolysis for Wake-up Stroke (Wake-Up Stroke) (ClinicalTrials.gov Identifier NCT01183533), the Efficacy and Safety of MRI-based Thrombolysis in Wake-up Stroke (WAKE-UP) (ClinicalTrials.gov Identifier NCT01525290) (20), the Wake up Symptomatic Stroke in Acute Brain Ischemia (WASSABI) Trial (ClinicalTrials.gov Identifier NCT01455935), and WUS-rTPA (EudraCT no. 2010–019359-23). All of the trials use alteplase at 0·9 mg/kg. We hypothesized that stroke patients with unclear-onset time and a negative FLAIR pattern on MRI will improve more by i.v. thrombolysis than by standard treatment using alteplase at 0·6 mg/kg, a dose that is unique to Japanese patients. Thus, we planned a multicenter trial to test this hypothesis.

teplase at 0·9 mg/kg. We hypothesized that stroke patients with unclear-onset time and a negative FLAIR pattern on MRI will improve more by i.v. thrombolysis than by standard treatment using alteplase at 0·6 mg/kg, a dose that is unique to Japanese patients. Thus, we planned a multicenter trial to test this hypothesis. Methods Design The THrombolysis for Acute Wake-up and unclear-onset Strokes with alteplase at 0·6 mg/kg (THAWS) Trial is an investigator-initiated, multicenter, prospective, randomized, open-treatment, blinded-end-point clinical trial comparing i.v. rt-PA (alteplase) and standard treatment in unclear-onset stroke. This trial is registered with the ClinicalTrials.gov (Clinical Trials.gov Identifier NCT02002325) and the UMIN clinical trial (ID: UMIN000011630). Figure 1 shows a flowchart of the trial design. Table 1 shows a schedule of the trial. The design is similar to the WAKE-UP trial (20). Fig. 1 Thrombolysis for Acute Wake-up and unclear-onset Strokes with alteplase at 0·6 mg/kg (THAWS) trial flow chart. AEs, adverse events; ASPECTS, the Alberta Stroke Program Early CT Score; CT, computed tomography; DWI, diffusion-weighted imaging; FLAIR, fluid-attenuated inversion recovery; i.v., intravenous; MRI, magnetic resonance imaging; NIHSS, National Institutes of Health Stroke Scale; mRS, modified Rankin Scale; rt-PA, recombinant tissue-type plasminogen activator. Table 1 Schedule of assessments for THAWS trial

Fig. 1 Thrombolysis for Acute Wake-up and unclear-onset Strokes with alteplase at 0·6 mg/kg (THAWS) trial flow chart. AEs, adverse events; ASPECTS, the Alberta Stroke Program Early CT Score; CT, computed tomography; DWI, diffusion-weighted imaging; FLAIR, fluid-attenuated inversion recovery; i.v., intravenous; MRI, magnetic resonance imaging; NIHSS, National Institutes of Health Stroke Scale; mRS, modified Rankin Scale; rt-PA, recombinant tissue-type plasminogen activator. Table 1 Schedule of assessments for THAWS trial Baseline Administration of treatment Observational period Timing Enrollment 0 h 24 h* ± 3 h Day 7 or discharge ± 1 day† Day 90 ± 14 days Consent ○ Demographics/Baseline information/Medical history/Prior medication ○ Screening/Eligibility ○ Randomization ○ Physical examination NIHSS ○ ○ ○ mRS Premorbid ○ ○ Height/Weight ○ Blood pressure/Pulse rate ○ ○ ○ Body temperature ○ Laboratory tests/Imaging Blood test ○ ○ Urine test ○ Electrocardiogrphy ○ MRI ○ ○* ○† Adverse events ○ ○ ○ ○ Alteplase administration ○ * Follow-up MRI at 24 h is allowed to be obtained between 22 and 36 h after the treatment administration. † Follow-up MRI at day 7 is allowed to be obtained between 7 and 14 days after the treatment administration or at discharge (whichever is first). MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale.

Baseline Administration of treatment Observational period Timing Enrollment 0 h 24 h* ± 3 h Day 7 or discharge ± 1 day† Day 90 ± 14 days Consent ○ Demographics/Baseline information/Medical history/Prior medication ○ Screening/Eligibility ○ Randomization ○ Physical examination NIHSS ○ ○ ○ mRS Premorbid ○ ○ Height/Weight ○ Blood pressure/Pulse rate ○ ○ ○ Body temperature ○ Laboratory tests/Imaging Blood test ○ ○ Urine test ○ Electrocardiogrphy ○ MRI ○ ○* ○† Adverse events ○ ○ ○ ○ Alteplase administration ○ * Follow-up MRI at 24 h is allowed to be obtained between 22 and 36 h after the treatment administration. † Follow-up MRI at day 7 is allowed to be obtained between 7 and 14 days after the treatment administration or at discharge (whichever is first). MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale. Patient population Patients with acute ischemic stroke whose onset time of stroke symptoms cannot be determined represent the target population for the THAWS. Inclusion and exclusion criteria are listed in Table 2. As clinical inclusion criteria for timing of treatment initiation, time from last-known-well period without neurological symptoms to treatment initiation should be between 4·5 and 12 h, and time from symptom recognition to treatment initiation should be within 4·5 h. Patients with posterior circulation stroke including brain stem stroke are not excluded because they are reported to potentially have a longer therapeutic time window than those with anterior stroke (21). The negative FLAIR pattern as an acute ischemic lesion visible on DWI, but no marked parenchymal hyperintensity visible on FLAIR is a vital imaging inclusion (Fig. 2). Exclusion criteria mainly follow the prescribing information for alteplase in Japan. A detailed imaging guidebook kindly provided by the WAKE-UP steering committee will confer extensive examples illustrating inclusion and exclusion criteria on MRI (20).

rintensity visible on FLAIR is a vital imaging inclusion (Fig. 2). Exclusion criteria mainly follow the prescribing information for alteplase in Japan. A detailed imaging guidebook kindly provided by the WAKE-UP steering committee will confer extensive examples illustrating inclusion and exclusion criteria on MRI (20). Table 2 Inclusion and exclusion criteria Clinical inclusion criteria • Clinical diagnosis of acute ischemic stroke with unknown symptom onset (e.g., acute wake-up ischemic stroke and acute ischemic stroke with unknown time of symptom onset) • Age 20 years or older • Last-known-well period without neurological symptoms > 4·5 h and <12 h of treatment initiation • Treatment can be started within 4·5 h of symptom recognition (e.g., awakening) • Initial NIHSS ≥ 5 and ≤25 • Written informed consent by patient or next of kin Imaging inclusion criteria • Acute stroke MRI including DWI and FLAIR completed • ASPECTS on initial DWI ≥ 5 • Pretreatment MRI showing a pattern of ‘negative FLAIR’, that is, acute ischemic lesion visible (or normally visible) on DWI but no marked parenchymal hyperintensity visible on FLAIR indicative of an acute ischemic lesion ≤ 4·5 h of age

• Acute stroke MRI including DWI and FLAIR completed • ASPECTS on initial DWI ≥ 5 • Pretreatment MRI showing a pattern of ‘negative FLAIR’, that is, acute ischemic lesion visible (or normally visible) on DWI but no marked parenchymal hyperintensity visible on FLAIR indicative of an acute ischemic lesion ≤ 4·5 h of age Clinical exclusion criteria • Prestroke mRS > 1 (patients who have inability to carry out all daily activities and require some help or supervision) • Contraindications in the Japanese guidelines for the intravenous application of recombinant tissue-type plasminogen activator (alteplase) √ History of nontraumatic intracranial hemorrhage √ History of stroke within the last one-month (excluding transient ischemic attack) √ History of significant head/spinal injury or surgery within the last three-months √ History of gastrointestinal or urinary tract bleeding within the last 21 days √ History of major surgery or significant trauma other than head injury within the last 14 days √ Hypersensitivity to alteplase or any of the excipients √ Suspected subarachnoid hemorrhage √ Concurrent acute aortic dissection √ Concurrent hemorrhage (e.g., intracranial, gastrointestinal, urinary tract, or retroperitoneal) √ Systolic blood pressure ≥ 185 mmHg despite antihypertensive therapy √ Diastolic blood pressure ≥ 110 mmHg despite antihypertensive therapy √ Significant hepatic disorder √ Acute pancreatitis √ Blood glucose < 50 or >400 mg/dL (<2·8 or >22·2 mmol/L) √ Platelet count ≤ 100 000/mm3 √ PT-INR > 1·7 or prolonged aPTT [>1·5 times the baseline value ( > approximately 40 s only as a guide)] for patients on anticoagulation therapy or those with abnormal coagulation • Any contraindication to MRI (e.g., cardiac pacemaker) • Planned or anticipated treatment with surgery or endovascular reperfusion strategies (e.g., intra-arterial thrombolysis, mechanical recanalization techniques) • Pregnant, lactating, or potentially pregnant • Life expectancy six-months or less by judgment of the investigator • Inappropriate for study enrollment by judgment of the investigator

treatment with surgery or endovascular reperfusion strategies (e.g., intra-arterial thrombolysis, mechanical recanalization techniques) • Pregnant, lactating, or potentially pregnant • Life expectancy six-months or less by judgment of the investigator • Inappropriate for study enrollment by judgment of the investigator Imaging exclusion criteria • Poor MRI quality precluding interpretation according to the study protocol • Large DWI lesion volume > 50% of the anterior cerebral artery or posterior cerebral artery territory (visual inspection) • Large DWI lesion in brain stem or cerebellum (e.g., more than half of brain stem or more than half of unilateral cerebellar hemisphere) • Any sign of intracranial hemorrhage on baseline MRI • FLAIR showing marked parenchymal hyperintensity corresponding to the acute DWI lesion indicative of an acute ischemic lesion with a high likelihood of being >4·5 h old (‘positive FLAIR’) • Any MRI findings indicative of a high risk of symptomatic intracranial hemorrhage related to potential intravenous alteplase treatment in the judgment of the investigator aPTT, activated partial thromboplastin time; ASPECTS, Alberta Stroke Program Early CT score; DWI, diffusion weighted imaging; FLAIR, fluid attenuated inversion recovery; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; PT-INR, prothrombin time international normalized ratio.

rtial thromboplastin time; ASPECTS, Alberta Stroke Program Early CT score; DWI, diffusion weighted imaging; FLAIR, fluid attenuated inversion recovery; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; PT-INR, prothrombin time international normalized ratio. Fig. 2 Examples of magnetic resonance imaging (MRI) inclusion and exclusion criteria. (a) A negative fluid-attenuated inversion recovery (FLAIR) pattern shows an acute ischemic lesion clearly visible on diffusion-weighted imaging (DWI), but no marked parenchymal hyperintensity visible on fluid-attenuated inversion recovery (FLAIR) corresponding to the DWI lesion (yellow circles). (b) A positive FLAIR pattern shows an acute ischemic lesion clearly visible on DWI and clear parenchymal hyperintensity on FLAIR corresponding to the acute DWI lesion (yellow circle).

rked parenchymal hyperintensity visible on fluid-attenuated inversion recovery (FLAIR) corresponding to the DWI lesion (yellow circles). (b) A positive FLAIR pattern shows an acute ischemic lesion clearly visible on DWI and clear parenchymal hyperintensity on FLAIR corresponding to the acute DWI lesion (yellow circle). MRI sequence and baseline assessment DWI (spin-echo echo planar imaging), FLAIR (fast spin echo), T2* and magnetic resonance angiography (MRA) sequences are mandatory acquired. Perfusion imaging with dynamic susceptibility contrast is optionally added for baseline imaging. The details of DWI and FLAIR parameters are as follows: DWI with field of view (FOV) of 240 mm, acquisition matrix of 128 × 128, slice thickness of ≈5 mm, gap of 0–1 mm, repetition time (TR) of ≈8000 ms, and echo time (TE) of ≤100 ms; FLAIR on 1·5 Tesla MRI with FOV of 240 mm, acquisition matrix of 256 × 256, slice thickness of ≈5 mm, gap of 0–1 mm, TR of ≥8000 ms, TE of 100–140 ms, and inversion time (TI) of ≈2300 ms; and FLAIR on 3 Tesla MRI with FOV of 240 mm, acquisition matrix of 256 × 256, slice thickness of ≈5 mm, gap of 0–1 mm, TR of ≥10000 ms, TE of 95–125 ms, and TI of ≈2600 ms.

m, acquisition matrix of 256 × 256, slice thickness of ≈5 mm, gap of 0–1 mm, TR of ≥8000 ms, TE of 100–140 ms, and inversion time (TI) of ≈2300 ms; and FLAIR on 3 Tesla MRI with FOV of 240 mm, acquisition matrix of 256 × 256, slice thickness of ≈5 mm, gap of 0–1 mm, TR of ≥10000 ms, TE of 95–125 ms, and TI of ≈2600 ms. Basically, DWI and FLAIR lesions are visually assessed according to the WAKE-UP imaging guidebook. For DWI assessment, apparent diffusion coefficient map can be used to exclude a T2 shine through effect. If DWI lesion is extensively overlapping with a previous stroke lesion or extensive white matter change, such patient needs to be excluded. The guidebook optionally provides the objective guidance to include patients with relative signal intensity of <1·2 on FLAIR lesion corresponding to acute DWI lesion as compared with contralateral normal signal intensity. Randomization and data management Eligible patients are randomized 1:1 to either i.v. rt-PA (alteplase, the rt-PA group) or standard treatment (the control group). Both patients and investigators are open to treatment allocation. However, primary and secondary outcomes are assessed without information regarding treatment allocation by independent neurologists, neurosurgeons, or nurses. The Research Electronic Data Capture (REDCap) system is used for data entering and management.

Both patients and investigators are open to treatment allocation. However, primary and secondary outcomes are assessed without information regarding treatment allocation by independent neurologists, neurosurgeons, or nurses. The Research Electronic Data Capture (REDCap) system is used for data entering and management. Treatment Alteplase is supplied in glass vials. Labeling and packaging of study medication are conducted according to good manufacturing practice, good clinical practice, and local and national regulatory requirements. Patients randomized to the rt-PA group receive alteplase 0·6 mg/kg body weight i.v. up to a maximum of 60 mg, 10% as bolus, and 90% as continuous infusion over one-hour. Patients randomized to the control group do not receive i.v. rt-PA but are treated with one to three antithrombotic drugs, including aspirin (160–300 mg/day), clopidogrel (75 mg/day), argatroban, and unfractionated heparin, except for the combination of argatroban and heparin according to attending physician's decisions. Such antithrombotics are prohibited for use in the rt-PA group within the initial 25 h. Treatment has to be initiated as soon as possible within 60 min of the end of the MRI examination. In addition, i.v. edaravone (free radical scavenger) is routinely given before or soon after trial enrollment for both groups, for a maximum of 14 days, unless contraindicated or inappropriate. Other antiplatelets (e.g., cilostazol) and anticoagulants (e.g., warfarin) are prohibited within the initial 25 h, and thrombolytic agents such as urokinase and monteplase are prohibited during the 90-day study period in both groups.

nt for both groups, for a maximum of 14 days, unless contraindicated or inappropriate. Other antiplatelets (e.g., cilostazol) and anticoagulants (e.g., warfarin) are prohibited within the initial 25 h, and thrombolytic agents such as urokinase and monteplase are prohibited during the 90-day study period in both groups. Clinical and radiological assessments Trained neurologists, neurosurgeons, or nurses will perform clinical assessment at baseline, at 22–36 h, at 7–14 days or at hospital discharge, and at 90 days after stroke onset. Neurological severity is evaluated using the National Institutes of Health Stroke Scale (NIHSS). At 90 days, a physician, nurse, or clinical research coordinator who is not aware of treatment assignment assesses mRS and adverse events at the clinic or by telephone interview. Investigators are recommended to complete a training and certification program for NIHSS and mRS. In addition to the initial MRI prior to randomization, follow-up MRI is performed after 22–36 h to identify ICH, and after 7–14 days to delineate final infarct volume. All images are judged at the MRI examination scanner or the display for reading or viewing purposes by the local investigators. All investigators are recommended to pass a standardized training program for image judgment provided by the WAKE-UP committee. A central image reading board continuously monitors the fulfillment of the prespecified MRI standards in each participating center and the compliance of patients randomized with the imaging inclusion and exclusion criteria.

ed to pass a standardized training program for image judgment provided by the WAKE-UP committee. A central image reading board continuously monitors the fulfillment of the prespecified MRI standards in each participating center and the compliance of patients randomized with the imaging inclusion and exclusion criteria. Primary and secondary outcomes Efficacy and safety end-points are listed in Table 3. Primary efficacy end-point is favorable outcome defined by mRS score 0–1 at 90 days after stroke onset. The safety end-points are sICH at 22–36 h and major bleeding (22), and death due to any cause at 90 days. Table 3 Efficacy and safety assessment Primary efficacy end-point • Favorable outcome defined by mRS score 0–1 at 90 days after stroke onset Secondary efficacy end-points • Categorical shift in NIHSS score at 24 h after the initiation of treatment • Categorical shift in NIHSS score at seven-days after the initiation of treatment • mRS score 0–2 at 90 days after stroke onset • Categorical shift in mRS score at 90 days after stroke onset • Recanalization of culprit artery on MRA 22–36 h after the initiation of treatment • Infarct volume on FLAIR 7–14 days after the initiation of treatment Safety end-points

• Categorical shift in NIHSS score at 24 h after the initiation of treatment • Categorical shift in NIHSS score at seven-days after the initiation of treatment • mRS score 0–2 at 90 days after stroke onset • Categorical shift in mRS score at 90 days after stroke onset • Recanalization of culprit artery on MRA 22–36 h after the initiation of treatment • Infarct volume on FLAIR 7–14 days after the initiation of treatment Safety end-points • sICH as defined by an increase of NIHSS score of ≥4 from baseline and parenchymal hematoma type II (PH-2) on MRI 22–36 h after the initiation of treatment • Major bleeding as defined by fatal bleeding, symptomatic bleeding in a critical area or organ, such as intraspinal, intraocular, retroperitoneal, intraarticular, or pericardial, or intramuscular with compartment syndrome, or bleeding causing a fall in hemoglobin level ≥ 2 g/dL, or leading to transfusion of ≥4·5 units (≈1125 mL) of whole blood or red cells according to the definition of the International Society on Thrombosis and Haemostasis (22) within 90 days after stroke onset • Death (mRS 6) due to any cause at 90 days after stroke onset FLAIR, fluid attenuated inversion recovery; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; sICH, symptomatic intracerebral hemorrhage.

s after stroke onset • Death (mRS 6) due to any cause at 90 days after stroke onset FLAIR, fluid attenuated inversion recovery; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; sICH, symptomatic intracerebral hemorrhage. Data monitoring body Data monitoring is centrally conducted by the members of Department of Advanced Medical Technology Development, National Cerebral and Cardiovascular Center. According to the order from the steering committee, the members occasionally visit collaborating hospitals to review source materials, including related medical records and documents for consent forms. Responsible authorities, related ethics committees, or directors of collaborating hospitals have the right to review source material if necessary.

the steering committee, the members occasionally visit collaborating hospitals to review source materials, including related medical records and documents for consent forms. Responsible authorities, related ethics committees, or directors of collaborating hospitals have the right to review source material if necessary. Data and safety monitoring board An independent data and safety monitoring board (DSMB) oversees the conduct of the trial. The occurrence of any safety end-point is immediately reported to the DSMB by the responsible physician via the central office, along with the accumulated number of safety end-points and total patient enrollment numbers. All safety end-points are mandatorily analyzed after inclusion of 150 and 300 patients. The proportion of sICH in the i.v. rt-PA group is compared with those of previous reports (23–27), and the proportions of major bleeding and death in the i.v. rt-PA group are compared with those in the control group. In any case of concern about the safety of participants, the DSMB makes a recommendation to the steering committee about continuing, stopping, or modifying the trial.

those of previous reports (23–27), and the proportions of major bleeding and death in the i.v. rt-PA group are compared with those in the control group. In any case of concern about the safety of participants, the DSMB makes a recommendation to the steering committee about continuing, stopping, or modifying the trial. Sample size calculation A total of 278 patients (139 per group) is required to ensure 1 − β = 90% probability to demonstrate that the relative effect of i.v. rt-PA to standard treatment for ischemic stroke patients is more than a fraction of 0·5 of the combined relative effect of i.v. rt-PA across the stroke thrombolysis studies (15–17,28), by using one-sided chi-square test of significant level of 2·5%, where the effects of i.v. rt-PA and standard treatment are assumed 30% and 20% commonly for Japanese patients and comparable combined studies. Accounting for possible treatment failures, protocol violations, and dropouts, a total of 300 patients (150 per treatment group) will be recruited.

of significant level of 2·5%, where the effects of i.v. rt-PA and standard treatment are assumed 30% and 20% commonly for Japanese patients and comparable combined studies. Accounting for possible treatment failures, protocol violations, and dropouts, a total of 300 patients (150 per treatment group) will be recruited. Statistical analysis Analyses will be done according to an intention-to-treat (ITT) principle. A secondary per-protocol sensitivity analysis will be done to assess the robustness of conclusion from ITT basis analysis. Patient demographic data will be analyzed descriptively; categorical variables will be assessed with the chi-square test or Fisher's exact test, whereas continuous variables will be assessed with the Student's t-test or the Wilcoxon rank-sum test, as appropriate. The primary outcome is the proportion of patients with mRS 0–1 (i.e., primary end-point) at 90 days between the i.v. rt-PA and control groups, analyzed by the chi-square test. Relative risk (RR) for the primary outcome will be calculated with the corresponding 95% confidence interval. The secondary outcome is the change in NIHSS from baseline to at 24 h or 7 days, analyzed by analysis of covariance (ANCOVA), where the model includes treatment group as factor and NIHSS at baseline as a covariate. Safety data will be analyzed descriptively for the treated set, which consists of all randomized patients who receive at least one study treatment. The statistical analysis plan, which includes more technical and detailed elaboration of the principal features stated in the protocol, will separately be prepared and be finalized before breaking the blind.

ptively for the treated set, which consists of all randomized patients who receive at least one study treatment. The statistical analysis plan, which includes more technical and detailed elaboration of the principal features stated in the protocol, will separately be prepared and be finalized before breaking the blind. Study organization and funding The THAWS is organized by a central coordinating center located at the National Cerebral and Cardiovascular Center, and conducted in approximately 35 centers in Japan after the approval of Advanced Medical Technology Development authentication system by the Ministry of Health, Labour and Welfare. The steering committee manages the trial. The THAWS receives funding support from the Award from Charitable Trust Mihara Cerebrovascular Disorder Research Promotion Fund (to Minematsu) and the Intramural Research Fund for Cardiovascular Diseases of National Cerebral and Cardiovascular Center (H23-4-3).

our and Welfare. The steering committee manages the trial. The THAWS receives funding support from the Award from Charitable Trust Mihara Cerebrovascular Disorder Research Promotion Fund (to Minematsu) and the Intramural Research Fund for Cardiovascular Diseases of National Cerebral and Cardiovascular Center (H23-4-3). Discussion and conclusion The THAWS is a randomized, controlled trial of stroke thrombolysis with alteplase 0·6 mg/kg based on the presence of negative FLAIR pattern on initial MRI in patients with unclear-onset time of stroke symptom, that is, wake-up stroke. The negative FLAIR will ensure the enrollment of patients with ischemic lesions likely to be less than 4·5 h after stroke onset who are likely to benefit from thrombolysis. The THAWS may trigger approval of low-dose i.v. thrombolysis using MRI-based selection as a routine clinical practice for ischemic stroke patients with unclear-onset time. Furthermore, 0·6 mg/kg of alteplase is expected to have similar efficacy and higher safety than 0·9 mg/kg in Asian countries (29), and it is now investigated in the Enhanced Control of Hypertension and Thrombolysis Stroke Study (ENCHANTED) (Clinical Trials.gov Identifier NCT01422616).

patients with unclear-onset time. Furthermore, 0·6 mg/kg of alteplase is expected to have similar efficacy and higher safety than 0·9 mg/kg in Asian countries (29), and it is now investigated in the Enhanced Control of Hypertension and Thrombolysis Stroke Study (ENCHANTED) (Clinical Trials.gov Identifier NCT01422616). Of several ongoing clinical trials on thrombolysis in acute stroke patients with unclear-onset time in worldwide, the WAKE-UP trial is the largest. We are in contact with the WAKE-UP group and they kindly provided us their detailed imaging guidebook and standardized training program for image judgment to share the same imaging inclusion criteria. We may conduct a meta-analysis with the ongoing trials for unclear-onset stroke, including the WAKE-UP trial. THAWS boards and institutions Senior advisor: Kazuo Minematsu Steering committee: Kazunori Toyoda (Chair), Kazumi Kimura (vice-chair), Haruko Yamamoto (data monitoring), Masatoshi Koga (central office) Protocol committee: Junya Aoki, Toshimitsu Hamasaki, Kazumi Kimura, Masatoshi Koga, Shoichiro Sato, Kazunori Toyoda Statistician: Toshimitsu Hamasaki Independent safety monitoring board: Takanari Kitazono (Chair), Toshisuke Otsuki, Wataru Shimizu, Takashi Sozu Central imaging reading board: Makoto Sasaki (Chair), Teruyuki Hirano, Kohsuke Kudo, Naomi Morita Central pharmacy: Ken Kuwahara Coordinating investigators: Sohei Yoshimura, Shoichiro Sato, Kazunari Homma, Kenta Seki Secretariats: Haruka Kanai, Azusa Tokunaga

Independent safety monitoring board: Takanari Kitazono (Chair), Toshisuke Otsuki, Wataru Shimizu, Takashi Sozu Central imaging reading board: Makoto Sasaki (Chair), Teruyuki Hirano, Kohsuke Kudo, Naomi Morita Central pharmacy: Ken Kuwahara Coordinating investigators: Sohei Yoshimura, Shoichiro Sato, Kazunari Homma, Kenta Seki Secretariats: Haruka Kanai, Azusa Tokunaga THAWS consortium National Cerebral and Cardiovascular Center, Suita, Osaka – Kazunori Toyoda, Kazuyuki Nagatsuka, Haruko Yamamoto, Masatoshi Koga, Hiroshi Yamagami, Shoichiro Sato, Kazunari Homma, Kenta Seki Kawasaki Medical School, Kurashiki – Kazumi Kimura, Junya Aoki Hokkaido P.W.F.A.C. Obihiro-Kosei General Hospital, Obihiro – Masafumi Ohtaki Nakamura Memorial Hospital, Sapporo – Kamiyama Kenji Hirosaki Stroke and Rehabilitation Center, Hirosaki – Metoki Norifumi Kohnan Hospital, Sendai – Eisuke Furui Yamagata City Hospital SAISEIKAN, Yamagata – Rei Kondo South Miyagi Medical Center, Shibata – Hiroshi Mochizuki Niigata City General Hospital, Niigata – Shuichi Igarashi Mihara Memorial Hospital, Isesaki – Ban Mihara Jichi Medical University, Shimotsuke – Tomoaki Kameda Saitama Medical University International Medical Center, Hidaka – Norio Tanahashi, Ichiro Deguchi Juntendo University Urayasu Hospital, Urayasu – Takao Urabe Juntendo University Hospital, Tokyo – Nobutaka Hattori Jikei University School of Medicine, Tokyo – Yasuyuki Iguchi Toranomon Hospital, Tokyo – Yoshikazu Uesaka Kyorin University, Mitaka – Yoshiaki Shiokawa, Denbo Norihisa St. Marianna University School of Medicine, Kawasaki – Yasuhiro Hasegawa Tokai University, Isehara – Shunya Takizawa

Juntendo University Hospital, Tokyo – Nobutaka Hattori Jikei University School of Medicine, Tokyo – Yasuyuki Iguchi Toranomon Hospital, Tokyo – Yoshikazu Uesaka Kyorin University, Mitaka – Yoshiaki Shiokawa, Denbo Norihisa St. Marianna University School of Medicine, Kawasaki – Yasuhiro Hasegawa Tokai University, Isehara – Shunya Takizawa Shizuoka Saiseikai General Hospital, Shizuoka – Jin Yoshii Toyota Memorial Hospital, Toyota – Yasuhiro Ito National Hospital Organization Nagoya Medical Center, Nagoya – Satoshi Okuda Japanese Red Cross Kyoto Daini Hospital, Kyoto – Yoshinari Nagakane Hyogo College of Medicine, Nishinomiya – Shinichi Yoshimura Kobe City Medical Center Central Hospital, Kobe – Nobuyuki Sakai, Kenichi Todo. Ohnishi Neurological Center, Akashi – Hideyuki Ohnishi Kawasaki Hospital, Okayama – Tsuyoshi Inoue Osaka Neurosurgical Hospital, Takamatsu – Hideo Ohyama Kokura Memorial Hospital, Kokura – Ichiro Nakahara Steel Memorial Yawata Hospital, Kitakyushu – Shigeru Fujimoto Japanese Red Cross Fukuoka Hospital, Fukuoka – Kenichiro Fujii, Sohei Yoshimura National Hospital Organization Kyushu Medical Center, Fukuoka – Yasushi Okada, Takeshi Uwatoko Nagasaki University Hospital, Nagasaki – Akira Tsujino Saiseikai Kumamoto Hospital, Kumamoto – Toshiro Yonehara Japanese Red Cross Kumamoto Hospital, Kumamoto – Tadashi Terasaki

Introduction Background Stroke is the leading cause of death and disability in industrialized countries, imposing immeasurable burdens on the society. The reduction of social burden may be most efficiently achieved by preventing first strokes. However, once stroke has occurred, the therapeutic target shifts to the functional improvement and prevention of recurrent stroke. Although common strategies for such secondary prevention include antiplatelet, anticoagulant, and antihypertensive agents 1,2, the prevention strategy must reflect the sub-types of occurred stroke and should be optimized based on the respective etiologies.

the functional improvement and prevention of recurrent stroke. Although common strategies for such secondary prevention include antiplatelet, anticoagulant, and antihypertensive agents 1,2, the prevention strategy must reflect the sub-types of occurred stroke and should be optimized based on the respective etiologies. 3-Hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors, called statins, are widely used to improve serum lipid profile. Because of the potent lipid-lowering effects, statins are commonly used to prevent coronary artery disease. In addition to their established value for coronary protection, statins are beneficial for stroke prevention. Indeed, the use of statins has been associated with 20–30% reduction in the risk of stroke in patients with coronary artery disease 3–5. Additionally, statin usage was shown to reduce the risk of stroke by 27% in nearly 20 000 patients at cardiovascular risk 6. Also, in the Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese study, statin usage was associated with a 46% reduction in the risk of stroke in hypertensive patients 7. These findings were derived from patients without prior stroke, suggesting a certain benefit of statins for the primary prevention of stroke. However, such a preventive effect has been less robust for the recurrence of stroke. For instance, in the Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) trial, including 4731 patients with stroke or transient ischemic attack (TIA), use of a statin was associated with 16% risk reduction for recurrent stroke 8. Also, a meta-analysis involving eight studies and 10 000 patients demonstrated that statin therapy had only a marginal effect in reducing the occurrence of subsequent stroke in patients with prior stroke or TIA 9. Additionally, these findings were obtained from the western people, and whether they apply for the Japanese patients remains to be examined.

ght studies and 10 000 patients demonstrated that statin therapy had only a marginal effect in reducing the occurrence of subsequent stroke in patients with prior stroke or TIA 9. Additionally, these findings were obtained from the western people, and whether they apply for the Japanese patients remains to be examined. Indeed, stroke is a heterogeneous disease with different etiologies, with or without underlying arterial pathologies. Thus, the benefits of statins should reflect the respective stroke sub-types. Particularly, statins are not likely to exert preventive effects in patients with cardioembolic infarction, in which involvement of lipids or atherosclerosis is limited. Given the structural difference between major cerebral arteries and the perforating branches, the effects of statins can differ between atherothrombotic and lacunar infarctions. Moreover, the use of statins might increase the risk of hemorrhagic stroke 10. Nevertheless, the majority of prior studies defined stroke as a whole, with no distinction between sub-types. Consequently, the risk reduction reported in prior studies may have been diluted as a consequence of the combination of all stroke sub-types. In fact, when limited to patients with noncardioembolic infarction, use of statins significantly reduced the risk of recurrent stroke 11. Furthermore, statins have pleiotropic favorable effects on arteries 12–15, including suppression of inflammation 16–18, regression of atherosclerosis 19,20, and improvement of endothelial function 21,22. In addition, Briel et al. demonstrated a stronger association of stroke reduction with statin use than with the extent of lipid reduction 23. Accordingly, pleiotropic vascular protective effects may contribute to the reduction of certain stroke sub-types, requiring further studies to clarify the role of statins in the secondary prevention of stroke. Particularly, elucidation of the effects of statins on respective stroke sub-types can help identify patients who would benefit from these treatments, thereby facilitating a more refined strategy for stroke prevention. Additionally, an examination of these effects in patients with mild hyperlipidemia could reveal whether stroke suppression is induced by lipid reduction or through other pleiotropic effects of statins.

patients who would benefit from these treatments, thereby facilitating a more refined strategy for stroke prevention. Additionally, an examination of these effects in patients with mild hyperlipidemia could reveal whether stroke suppression is induced by lipid reduction or through other pleiotropic effects of statins. Objectives Pravastatin, a traditional statin widely prescribed in the clinic, was selected for use in this study to determine whether this drug could reduce the recurrence of stroke with presumed arterial damage in patients with mild hyperlipidemia. In addition, this study evaluates the effects of pravastatin on the onset of respective stroke sub-types and explores the impact of this treatment on the functional outcomes related to stroke. Additionally, concurrent sub-studies are being conducted to assess whether statin usage suppresses chronic inflammation and/or carotid atherosclerosis in the patients enrolled in this study.

the onset of respective stroke sub-types and explores the impact of this treatment on the functional outcomes related to stroke. Additionally, concurrent sub-studies are being conducted to assess whether statin usage suppresses chronic inflammation and/or carotid atherosclerosis in the patients enrolled in this study. Methods Trial design This is a multicenter, randomized (1:1), open-label, parallel group study conducted on outpatients with a prior history of stroke (NCT00221104). The study protocol and informed consent form were approved by the institutional review board of each center. All patients received information on the purpose and nature of this study as well as the potential risks and benefits. Thereafter, written informed consent for participation was obtained from each patient. In addition, this study is being conducted under the health insurance system of Japan, in accordance with the Declaration of Helsinki and the Ethical Guideline on Clinical Studies of the Ministry of Health, Labour and Welfare of Japan. Participants The current study has enrolled patients aged 45–80 years with a history of noncardioembolic ischemic stroke (atherothrombotic infarction, lacunar infarction, and infarction of undetermined etiology) within the preceding month to three-years. All patients had been previously diagnosed with hyperlipidemia, with serum total cholesterol levels maintained between 180 and 240 mg/dl by treatments other than statins.

ischemic stroke (atherothrombotic infarction, lacunar infarction, and infarction of undetermined etiology) within the preceding month to three-years. All patients had been previously diagnosed with hyperlipidemia, with serum total cholesterol levels maintained between 180 and 240 mg/dl by treatments other than statins. Patients were excluded if they had a cerebral infarction of determined rare etiology, (e.g., vertebral artery dissection, fibromuscular dysplasia, and moyamoya disease) or presented an infarction associated with catheterization or surgery. In addition, patients were excluded when use of statins was preferred for the care of coexisting coronary artery disease. Moreover, patients with the following conditions were excluded: hemorrhagic diathesis, coagulopathy, hemorrhagic diseases (e.g., intracerebral hemorrhage, sub-arachnoid hemorrhage, and active peptic ulcer), thrombocytopenia (platelet count ≤100 000/mm3), hepatic dysfunction (aspartate transaminase or alanine transaminase ≥ 100 IU/l), renal dysfunction (serum creatinine ≥2·0 mg/dl), scheduled surgery, and cancer treatment. Study settings Patients were enrolled for five-years (from March 2004 to February 2009) from 123 centers across the country. Initially, the enrollment period was three-years, but this period was extended twice due to insufficient accrual of patients. All centers were regional core hospitals with multiple stroke neurologists that provide comprehensive medical services for stroke. Patient follow-up is ongoing and will end in February 2014.

ry. Initially, the enrollment period was three-years, but this period was extended twice due to insufficient accrual of patients. All centers were regional core hospitals with multiple stroke neurologists that provide comprehensive medical services for stroke. Patient follow-up is ongoing and will end in February 2014. Sub-studies To explore the pleiotropic effects of the statin, sub-studies focusing on high-sensitivity C-reactive protein (NCT00361699) and carotid artery intima-media thickness (NCT00361530) are concurrently in progress; in these sub-studies, the target measures are centrally evaluated by standardized methods. In addition, for carotid artery evaluation, qualified and accredited sonographers have been employed. Moreover, because of the purported link between statin effects and certain hereditary backgrounds, the evaluation of single-nucleotide polymorphisms (SNPs) was subsequently added to the current study; 396 SNPs are being analyzed in association with stroke and statin effects. The details of these sub-studies will be reported elsewhere.

e of the purported link between statin effects and certain hereditary backgrounds, the evaluation of single-nucleotide polymorphisms (SNPs) was subsequently added to the current study; 396 SNPs are being analyzed in association with stroke and statin effects. The details of these sub-studies will be reported elsewhere. Interventions Patients assigned to the pravastatin group receive a daily oral dose of pravastatin of 10 mg. The administration was initiated within one-month after randomization and continued until the final observation or patient's death. When cholesterol levels consistently exceed 240 mg/dl at routine clinical visits, diet and exercise therapies are reinforced. Only when the reinforcement is insufficient, treatments with pravastatin are intensified or other drugs are added, based on decision by the primary physician. However, the use of statins other than pravastatin is prohibited. Patients assigned to the control group receive no statin treatment, although other drugs are administered when necessary. In both groups, hypertension and diabetes mellitus are treated in accordance with clinical practice, without restriction of drug types. Moreover, compliance to the treatment is monitored at every clinical visit.

gned to the control group receive no statin treatment, although other drugs are administered when necessary. In both groups, hypertension and diabetes mellitus are treated in accordance with clinical practice, without restriction of drug types. Moreover, compliance to the treatment is monitored at every clinical visit. Outcomes The primary end-point is the onset of recurrent stroke of any sub-type or TIA. The secondary end-points include the onset of each stroke sub-type, myocardial infarction, vascular accident, death, hospitalization, degree of disability or dependence in daily activities, onset of dementia, and severity of cognitive impairment. Stroke sub-types are diagnosed in accordance with the Treatment of Acute Stroke Trial criteria 24 and defined in accordance with the National Institute of Neurological Disorders and Stroke classification 25. The diagnostic criteria for each stroke sub-type, TIA, myocardial infarction, and vascular accidents are shown in Appendix 1. The degree of disability and dependence in daily activities are evaluated by the Barthel Index (BI) and modified Rankin Scale (mRS). Dementia is diagnosed by the Diagnostic and Statistical Manual of Mental Disorders IIIR criteria. Impairment of cognitive function is assessed in accordance with the Clinical Dementia Rating (CDR) and Mini Mental State Examination (MMSE).

tivities are evaluated by the Barthel Index (BI) and modified Rankin Scale (mRS). Dementia is diagnosed by the Diagnostic and Statistical Manual of Mental Disorders IIIR criteria. Impairment of cognitive function is assessed in accordance with the Clinical Dementia Rating (CDR) and Mini Mental State Examination (MMSE). The onset of recurrent stroke, myocardial infarction, vascular accident, death, and hospitalization are monitored at two- and six-months; one-, two, three-, four-, and five-years; and additionally at the completion of this study. Brain magnetic resonance imaging or computed tomography imaging are performed at baseline, at two-years, at the study completion, and when recurrent stroke occurs. In addition, lipid levels are evaluated at the Special Reference Laboratory (SRL, Inc., Tokyo, Japan), which is certified for major lipid measurements in accordance with the Centers for Disease Control and Prevention (Atlanta, GA, USA). Occasionally, intrahospital lipid measurements are conducted at certified study centers. The primary physician at each study center evaluates each end-point, and a different physician at the same center confirms the evaluation. The central committee annually reviews the data concerning the occurrence of stroke and myocardial infarction, based on information submitted from each center, and judgments are corrected if necessary. Additionally, a source document verification is performed for arbitrarily selected centers.

r confirms the evaluation. The central committee annually reviews the data concerning the occurrence of stroke and myocardial infarction, based on information submitted from each center, and judgments are corrected if necessary. Additionally, a source document verification is performed for arbitrarily selected centers. Hypertension is defined as a blood pressure ≥140/90 mmHg or the use of antihypertensive agents. Diabetes mellitus is diagnosed when any of the following criteria are satisfied: fasting blood glucose ≥126 mg/dl or casual blood glucose ≥200 mg/dl during the past three-months; blood glucose ≥200 mg/dl at two-hours after 75 g OGTT; taking antidiabetic agents; or a previous diagnosis of diabetes mellitus. Hyperlipidemia is defined as total cholesterol ≥220 mg/dl, low density lipoprotein cholesterol ≥140 mg/dl or triglyceride ≥150 mg/dl or by the use of antihyperlipidemic agents. Sample size Based on an annual 5% recurrence of stroke (including TIAs) in the control group 26, 25% risk reduction in the pravastatin group and five-years of follow-up, the number of patients required to detect differences between the pravastatin group and the control group was calculated as 1292 for each group (two-tailed 5% significance level, power level of 90%). Assuming that 14% of patients are lost during follow-up, the sample size was set at 1500 in each group and 3000 for the two groups combined.

ts required to detect differences between the pravastatin group and the control group was calculated as 1292 for each group (two-tailed 5% significance level, power level of 90%). Assuming that 14% of patients are lost during follow-up, the sample size was set at 1500 in each group and 3000 for the two groups combined. Interim analyses The independent data monitoring committee annually reviews the safety data to evaluate the appropriateness of the study continuation. An interim analysis was performed using the Haybittle-Peto 3 SD method 27 in September 2011. Based on the results of the interim analysis, the independent data monitoring committee recommended the continuation of the study. Randomization The patients were enrolled through a web-based registration and follow-up system provided by the data center at the Translational Research Informatics Center, Kobe, Japan. After the primary physician obtained patient consent, generally through the support of the clinical research coordinators, access to the system was granted and the physician sent the information required for enrollment. The system automatically evaluated the eligibility of each patient and randomly assigned participants to either the pravastatin or control group (1:1 allocation). When the allocation was performed, the prevalence of stroke sub-types (atherothrombotic infarction vs. other), elevated blood pressure (≥150/90 mmHg vs. not), and diabetes (absence vs. presence) was dynamically balanced between the two groups.

ed participants to either the pravastatin or control group (1:1 allocation). When the allocation was performed, the prevalence of stroke sub-types (atherothrombotic infarction vs. other), elevated blood pressure (≥150/90 mmHg vs. not), and diabetes (absence vs. presence) was dynamically balanced between the two groups. Statistical analysis Occurrences of predefined events, including stroke, myocardial infarction, and death, are analyzed in accordance with the intention-to-treat principle. The neurological and psychological outcomes are measured in patients for whom such data are available, and last observation carried forward data are used when necessary. The safety analysis set comprises eligible patients who received at least one dose of the study drug. The cumulative incidences of events are estimated using the Kaplan–Meier method. The cumulative incidence curves for the two groups are compared using a stratified log-rank test. The Cox proportional hazard model is also used to estimate the relative risk (hazard ratio) and the 95% confidence interval, in which risk reduction is expressed as (1 – hazard ratio) × 100%. When events occur two or more times in a patient, the person-years method is used to estimate the incidence and 95% confidence interval.

proportional hazard model is also used to estimate the relative risk (hazard ratio) and the 95% confidence interval, in which risk reduction is expressed as (1 – hazard ratio) × 100%. When events occur two or more times in a patient, the person-years method is used to estimate the incidence and 95% confidence interval. The BI and mRS are compared between the two groups using the Wilcoxon rank-sum test. For MMSE, a decrease of 5 points or more is defined as cognitive function decline, and the percentage is compared between the two groups using the χ2 test. In patients without dementia at enrollment, the occurrence of dementia and the CDR score are compared between the two groups. Sub-group analyses are based on baseline lipids, blood pressure, the presence of hypertension or diabetes mellitus, and the use of antiplatelet agents. For lipids and blood pressure, the sub-groups were divided into five groups using the baseline data, and the trend for the stroke recurrence risk is being explored. Incidences of treatment discontinuation and serious adverse events are compared using the χ2 test. All analyses are performed using sas software (Cary, NC, USA), and the level of significance is P < 0·05 (two-tailed).

o five groups using the baseline data, and the trend for the stroke recurrence risk is being explored. Incidences of treatment discontinuation and serious adverse events are compared using the χ2 test. All analyses are performed using sas software (Cary, NC, USA), and the level of significance is P < 0·05 (two-tailed). Results The current study included 1579 patients from 123 participating centers from March 1, 2004 to February 28, 2009. One patient proved to be ineligible after randomization, reducing the total number of eligible patients to 1578. Of these, 1095 and 864 patients, respectively, were enrolled for the concurrent sub-studies on high-sensitivity C-reactive protein and carotid artery intima-media thickness. The average age of the patients was 66·2 ± 8·5 years old, including 68·8% men (Table 1). In addition, the study sample exhibited a moderate prevalence of hypertension, diabetes, and current smoking. Although all patients had been diagnosed with hyperlipidemia, lipid levels were generally well controlled at baseline. Similarly, blood pressure and fasting blood glucose levels were within the normal ranges, largely due to medication provided at the hospitals. In addition, because all patients had a history of stroke, the majority of the participants were taking antiplatelet agents for the secondary prevention. Moreover, the prevalence of coronary artery disease was relatively low. Table 1 Baseline profiles of cardiovascular risk factors

The average age of the patients was 66·2 ± 8·5 years old, including 68·8% men (Table 1). In addition, the study sample exhibited a moderate prevalence of hypertension, diabetes, and current smoking. Although all patients had been diagnosed with hyperlipidemia, lipid levels were generally well controlled at baseline. Similarly, blood pressure and fasting blood glucose levels were within the normal ranges, largely due to medication provided at the hospitals. In addition, because all patients had a history of stroke, the majority of the participants were taking antiplatelet agents for the secondary prevention. Moreover, the prevalence of coronary artery disease was relatively low. Table 1 Baseline profiles of cardiovascular risk factors n 1578 Age (years) 66·2 ± 8·5 Gender (male%) 68·8 Height (cm)/weight (kg) 160·3 ± 8·7/61·1 ± 10·2 Hypertension (%) 75·9 Diabetes mellitus (%) 23·3 Smoking habit (%) Current smoker/past smoker/nonsmoker/unknown 16·5/36·9/44·8/0·9 T. Chol/HDL-C/LDL-C (mg/dl) 210·0 ± 24·7/53·5 ± 15·8/129·5 ± 24·5 Triglyceride (mg/dl) 142·2 ± 74·2 Systolic/diastolic blood pressure (mmHg) 137·1 ± 17·8/79·3 ± 11·3 Levels of blood pressure ≥150/90 mmHg (%) 39·0 Fasting blood glucose (mg/dl) 117·6 ± 41·0 Use of antiplatelet agents (%) 90·7 Coronary artery disease (%) 5·1 The data are the mean ± SD for continuous variables. HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; T. Chol, total cholesterol.

of blood pressure ≥150/90 mmHg (%) 39·0 Fasting blood glucose (mg/dl) 117·6 ± 41·0 Use of antiplatelet agents (%) 90·7 Coronary artery disease (%) 5·1 The data are the mean ± SD for continuous variables. HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; T. Chol, total cholesterol. Because we enrolled selected stroke patients, the sub-types of stroke comprised 25·4% atherothrombotic infarction, 64·2% lacunar infarction, and 10·4% infarction of undetermined etiology (Table 2). In accordance with the higher prevalence of lacunar infarction, the size of the infarction was small in more than two-thirds (73·1%) of patients. In addition, the infarctions were predominantly located in the perforating artery region and were most often detected in the middle cerebral artery territory (59·0%), followed by the vertebrobasilar artery territory (23·6%). Notably, large infarctions were observed in only 1·2% of patients. Consistent with these results, the disability of each patient was generally mild according to the National Institutes of Health Stroke Scale, BI, and mRS (Fig. 1). Although dementia was diagnosed in 3·2% of patients, cognitive function was preserved in the majority of patients, based on CDR and MMSE. Fig 1 Baseline profiles of stroke-related measures −2.NIHSS, National Institutes of Health Stroke Scale; BI, Barthel Index; mRS, modified Ranking Scale; CDR, Clinical Dementia Rating; MMSE, Mini Mental State Examination. The vertical axes represent the number of patients in respective value ranges.

Because we enrolled selected stroke patients, the sub-types of stroke comprised 25·4% atherothrombotic infarction, 64·2% lacunar infarction, and 10·4% infarction of undetermined etiology (Table 2). In accordance with the higher prevalence of lacunar infarction, the size of the infarction was small in more than two-thirds (73·1%) of patients. In addition, the infarctions were predominantly located in the perforating artery region and were most often detected in the middle cerebral artery territory (59·0%), followed by the vertebrobasilar artery territory (23·6%). Notably, large infarctions were observed in only 1·2% of patients. Consistent with these results, the disability of each patient was generally mild according to the National Institutes of Health Stroke Scale, BI, and mRS (Fig. 1). Although dementia was diagnosed in 3·2% of patients, cognitive function was preserved in the majority of patients, based on CDR and MMSE. Fig 1 Baseline profiles of stroke-related measures −2.NIHSS, National Institutes of Health Stroke Scale; BI, Barthel Index; mRS, modified Ranking Scale; CDR, Clinical Dementia Rating; MMSE, Mini Mental State Examination. The vertical axes represent the number of patients in respective value ranges. Table 2 Baseline profiles of stroke related measures −1

Fig 1 Baseline profiles of stroke-related measures −2.NIHSS, National Institutes of Health Stroke Scale; BI, Barthel Index; mRS, modified Ranking Scale; CDR, Clinical Dementia Rating; MMSE, Mini Mental State Examination. The vertical axes represent the number of patients in respective value ranges. Table 2 Baseline profiles of stroke related measures −1 n 1578 Sub-types of ischemic stroke (%) Atherothrombotic/lacunar/undetermined etiology 25·4/64·2/10·4 Size of infarction* (%) Small/medium/large 73·1/20·5/1·2 Location of infarction (%) Cortical/perforating/both 18·1/72·3/4·5 Responsible arteries† (%) ACA/MCA/PCA/VB/BZ 2·0/59·0/7·5/23·6/2·7 Dementia (%) 3·2 * Small, <1·5 cm in diameter; medium, between small and large; large, half or more of the cerebral lobe. † ACA, anterior cerebral artery; BZ, border zone; MCA, middle cerebral artery; PCA, posterior cerebral artery; VB, vertebrobasilar arteries. Discussion The aim of this study was to evaluate the effects of an HMG-CoA reductase inhibitor for the prevention of recurrent stroke. Patient enrollment was closed at the end of February 2009, and follow-up studies are ongoing. This manuscript reports the study rationale and design as well as baseline features of the patients.

n The aim of this study was to evaluate the effects of an HMG-CoA reductase inhibitor for the prevention of recurrent stroke. Patient enrollment was closed at the end of February 2009, and follow-up studies are ongoing. This manuscript reports the study rationale and design as well as baseline features of the patients. Because pravastatin was used, this study was designed and conducted under the health insurance system of the country. Because use of placebo is not permitted in such a setting, we employed an open-label approach, utilizing clinically prescribed drugs. In addition, this study is conducted as part of the clinical practice, and thus the benefits of individual patients take precedence over the study purposes. For instance, lipid levels must be maintained in clinically acceptable ranges in both the treatment and control groups to reduce cardiovascular risk in individual patients. Similarly, if statins are considered beneficial for the treatment of coexisting coronary stenosis, the usage cannot be prohibited in such patients. Thus, this study was designed under the conditions of the real world of medicine rather than under ideal world conditions.

educe cardiovascular risk in individual patients. Similarly, if statins are considered beneficial for the treatment of coexisting coronary stenosis, the usage cannot be prohibited in such patients. Thus, this study was designed under the conditions of the real world of medicine rather than under ideal world conditions. In the current study, we focused on the vascular protective effects of statins and enrolled stroke patients with presumed arterial damage. Thus, only patients with atherothrombotic infarction, lacunar infarction, or infarction of undetermined etiology were included, where every effort was made to accurately diagnose these stroke sub-types. In addition, we enrolled patients who were previously diagnosed with hyperlipidemia and whose serum total cholesterol levels were maintained between 180 and 240 mg/dl at enrollment because the use or disuse of statins was less likely to jeopardize these patients, allowing for randomization in the clinical practice. Additionally, we excluded patients for whom statins would be prescribed for coronary protection because the inclusion of such patients might increase protocol violation or patient loss in the control group. These factors have substantially narrowed the window of patient enrollments, decreasing the potential candidates for this study.

e excluded patients for whom statins would be prescribed for coronary protection because the inclusion of such patients might increase protocol violation or patient loss in the control group. These factors have substantially narrowed the window of patient enrollments, decreasing the potential candidates for this study. To treat the patients, we set the dose of pravastatin at 10 mg/day, which is less than that commonly used in the United States and Europe but is the standard in Japan, where lifestyles and hereditary backgrounds are different. In addition, because the study is being conducted as part of the clinical practice, the physicians make every effort to prevent the elevation of lipids in both groups. When serum total cholesterol levels consistently exceed 240 mg/dl during regular patient visits to the clinic, despite the reinforcement of diet and exercise therapies, the dose of pravastatin is increased if the patient is in the treatment group, whereas other types of drugs (except statins) are administered in the control group. Under these conditions, it is reasonable to say that this study is more focused on the pleiotropic effects of statins than on lipid reduction. The use of this approach is consistent with the stronger association of stroke reduction with statin treatment than with the extent of lipid reduction, as reported in previous studies 23,28.

onditions, it is reasonable to say that this study is more focused on the pleiotropic effects of statins than on lipid reduction. The use of this approach is consistent with the stronger association of stroke reduction with statin treatment than with the extent of lipid reduction, as reported in previous studies 23,28. The current study uses stroke recurrence as the primary end-point, including TIA, because the recent advent of thrombolytic or antithrombotic therapies often obscures the separation between stroke and TIA. Among the secondary end-points adopted, the onset of each stroke sub-type is of particular interest, and diagnostic criteria have been clearly defined (Appendix 1). Particularly, determination of the associations between statin use and the onset of respective stroke sub-types could help identify patients that might benefit from this treatment, potentially facilitating a more efficient preventive strategy. However, we must be aware of confounding opinions; the post hoc analysis of SPARCL trial suggested that statin usages were similarly efficacious in preventing recurrent stroke irrespective of the baseline sub-types 29. Thus, this issue is prospectively addressed in the current study, even though the statistical power may not be enough. Other secondary end-points include myocardial infarction, death and hospitalization, consistent with other studies. In addition, several stroke-related functional measures (e.g., BI, mRS, CDR, and MMSE) are included in the secondary end-points. Analyses of the changes in these measures could help determine whether statins are protective against functional deterioration over time. Notably, correlative sub-studies focusing on high-sensitivity C-reactive protein and carotid artery intima-media thickness are being performed concurrently, allowing for further studies on the pleiotropic effects of statins in patients with relatively low lipid levels.

ective against functional deterioration over time. Notably, correlative sub-studies focusing on high-sensitivity C-reactive protein and carotid artery intima-media thickness are being performed concurrently, allowing for further studies on the pleiotropic effects of statins in patients with relatively low lipid levels. Based on the risk reduction reported in previous studies, we set the sample size at 3000, but due to the narrower window of patient enrollment, this number was not realized. Consequently, 1579 patients were recruited, and 1578 patients were eligible for the current study, roughly corresponding to half of the initial target and decreasing the probability of detecting intergroup differences in stroke recurrence. However, we excluded stroke patients unlikely to benefit from statin treatment, which could enrich the putative risk reduction in the study sample, mitigating the shortage of the statistical power. Indeed, after carefully reviewing the results of interim analysis, the independent data monitoring committee recommended continuation of the study to potentially answer the research questions of this study and concurrent sub-studies. Accordingly, patient follow-ups and evaluations are in progress.

he statistical power. Indeed, after carefully reviewing the results of interim analysis, the independent data monitoring committee recommended continuation of the study to potentially answer the research questions of this study and concurrent sub-studies. Accordingly, patient follow-ups and evaluations are in progress. Because all patients previously experienced stroke, the study sample had a moderate prevalence of traditional cardiovascular risk factors (Table 1). Notably, although all patients had been previously diagnosed with hyperlipidemia, the lipid levels were generally well controlled at baseline. Similarly, the levels of blood pressure and fasting blood glucose were within normal ranges, largely due to medications provided at the hospitals. Additionally, the majority of patients were taking antiplatelet agents to prevent stroke recurrence, which should be considered when interpreting the results. The lower prevalence of coronary artery disease may reflect the exclusion of patients requiring statins for coronary protection.

tions provided at the hospitals. Additionally, the majority of patients were taking antiplatelet agents to prevent stroke recurrence, which should be considered when interpreting the results. The lower prevalence of coronary artery disease may reflect the exclusion of patients requiring statins for coronary protection. Among the stroke types targeted, lacunar infarction was most prevalent, accounting for two-thirds of patients (Table 2), consistent with the prevalence reported for the Japanese population 30. The lesions were generally small and predominantly located in the perforating branch area of the middle cerebral artery. Atherothrombotic infarction was second most frequent, representing one-fourth of patients. Given the purported arterial protective effects of statins, whether pravastatin suppresses the onset of this stroke sub-type may be of interest. The remaining patients were diagnosed with infarction of undetermined etiology, in which involvement of arterial damage was presumed after comprehensive investigation for atherosclerosis and occult emboli in the systemic vessels. In fact, we made every effort to specify the cause of infarctions and cautiously excluded infarctions of determined etiology. Moreover, large infarctions involving the cerebral cortex were detected in 1·2% of patients, reflecting the outpatient nature of the study sample. Furthermore, the disability of patients was generally mild, which was anticipated because of the high prevalence of lacunar infarction. Although dementia was diagnosed in a small portion of patients, cognitive function was preserved in the majority of patients. Analyses on the relationships between statin use and changes of cognitive function are needed to evaluate the antidementia effects of statins, as previously reported 31.

e of lacunar infarction. Although dementia was diagnosed in a small portion of patients, cognitive function was preserved in the majority of patients. Analyses on the relationships between statin use and changes of cognitive function are needed to evaluate the antidementia effects of statins, as previously reported 31. The current study has certain limitations. First, because it is being conducted as part of the clinical practice, we cannot strictly prohibit the use of statins in the control group, which could potentially increase protocol violation or patient loss. Second, TIA is included in the primary end-point, requiring a caution when interpreting the results of this study. Thus, to allow for a reasonable interpretation, we have clearly defined the diagnostic criteria for respective stroke sub-types and TIA (Appendix 1). Third, the shortage of patient accrual reduces the chance of detecting intergroup differences in stroke recurrence. Indeed, our sample size is substantially smaller than the SPARCL trial (4731 patients with stroke or TIA), which demonstrated the benefit of statin for reducing recurrent stroke (hazard ratio, 0·84; 95% confidence interval, 0·71–0·99) 8. Nonetheless, we are carefully monitoring the onset of each stroke sub-type and evaluating functional measures related to stroke, facilitating extensive analyses regarding the relationships of statin use with stroke sub-types and functional outcomes. Moreover, concurrent sub-studies are ongoing that will facilitate an investigation of the link between pleiotropic effects of statins and stroke recurrence and the role of certain SNPs in these linkages.

litating extensive analyses regarding the relationships of statin use with stroke sub-types and functional outcomes. Moreover, concurrent sub-studies are ongoing that will facilitate an investigation of the link between pleiotropic effects of statins and stroke recurrence and the role of certain SNPs in these linkages. Conclusions This article reported the rationale, design, and baseline features of the Japanese Statin Treatment against Recurrent Stroke (J-STARS). Patient enrollment is closed, and follow-up studies are in progress. The results derived from this study could refine the role of statins in the secondary prevention of stroke. The authors would like to thank all study participants, physicians, comedical staff and co-workers for their assistance in the preparation and execution of this study. In addition, the authors would also like to acknowledge the late Dr Hideo Tohgi for invaluable advice on conceptualization of the study protocol and the late Dr Takeshi Shima for his great help as a member of regional promotion committee. Registration This study is registered in ClinicalTrials.gov under NCT00221104. Organization The organizational elements of J-STARS are listed in Appendix 2. Sub-studies Concurrent with this study, correlative sub-studies on high sensitivity C-reactive protein (1095 patients enrolled) and carotid artery intima-media thickness (864 patients enrolled) are in progress. In addition, the analysis of 396 SNPs was conducted in the current study, and the results will be analyzed in association with statin effects and stroke recurrence.

-studies on high sensitivity C-reactive protein (1095 patients enrolled) and carotid artery intima-media thickness (864 patients enrolled) are in progress. In addition, the analysis of 396 SNPs was conducted in the current study, and the results will be analyzed in association with statin effects and stroke recurrence. Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Appendix 1 Definition of endpoints Appendix 2 J-STARS Group: organizational structure and participants

The most recent Global Burden of Disease 2010 estimates (1) showed that the global burden of stroke continues to increase, with 16·9 million of people being affected by stroke annually, resulting in over 100 million disability-adjusted life years lost. There is also a worrisome global trend showing an increase in the number of strokes in young and middle-aged adults by 25% between 1990 and 2010 (2). This epidemic of stroke can and should be stopped and reversed (3), as over 90% of strokes are potentially avoidable (4). While management strategies for primary stroke prevention in high cardiovascular disease (CVD) risk individuals [including stroke and transient ischemic attack (TIA)] are well established (5), they are underutilized (5–7), and existing methods of primary stroke prevention are not sufficiently effective (8,9). The health care system has so far been largely unsuccessful in providing meaningful information to assist people to adhere to recommended lifestyle and medications (5,9,10). Uptake of this information is particularly low in people with moderately increased risk of stroke who would benefit from lifestyle changes (9,11). Inadequate CVD risk factor management (9) is implicated in underutilization of evidence-based primary stroke prevention strategies in those with moderately increased risk of stroke (12).

s information is particularly low in people with moderately increased risk of stroke who would benefit from lifestyle changes (9,11). Inadequate CVD risk factor management (9) is implicated in underutilization of evidence-based primary stroke prevention strategies in those with moderately increased risk of stroke (12). It is generally accepted that to be effective, primary stroke prevention should include both community-wide and high-risk preventative strategies. While community-wide efforts to reduce the prevalence and improve control of major modifiable risk factors for stroke (e.g. salt intake reduction, smoking cessation, blood pressure control) are the most cost-effective strategies for stroke prevention at a population level (3,13), they often require legislative changes and continuous educational campaigns. Currently used by health professionals, high-risk preventative strategies are aimed at the identification and management of people with high risk of stroke (e.g. people with elevated blood pressure, dyslipidemia, carotid artery stenosis). However, one of the main problems with this high-risk stroke prevention strategy is that it misses out most of the people who later develop a stroke because in reality most strokes are happening in people with only a mild increased risk of CVD (14,15). In addition, people with only a mildly elevated risk of CVD often do not seek medical attention and, therefore, it is difficult to include them in stroke prevention interventions. Currently available CVD risk assessment algorithms (16–18) allow calculation of absolute risk of CVD (including stroke), but they are designed for use by health professionals. Yet, one of the main challenges in effective stroke prevention on an individual level is the lack of awareness about stroke symptoms and risk factors as well as self-managing strategies to reduce their risk of stroke.

ion of absolute risk of CVD (including stroke), but they are designed for use by health professionals. Yet, one of the main challenges in effective stroke prevention on an individual level is the lack of awareness about stroke symptoms and risk factors as well as self-managing strategies to reduce their risk of stroke. The question is how to improve stroke prevention in individuals with an increased CVD risk in the most efficient way? We recently developed an app called Stroke Riskometer (19) that has the potential to significantly improve stroke and CVD prevention on an individual level. Based on the Framingham Heart Study stroke prediction algorithm (17) and enhanced to include seven additional major risk factors important for stroke (diet, physical activity, waist-to-hip ratio, alcohol, psychosocial stress, family history of stroke or heart attack, race/ethnicity), this user-friendly Stroke Riskometer is able to provide an estimate of the individual's absolute risk of stroke within the next 5 and 10 years for anyone from the age of 20 up to 90+ years old. The importance of adding those seven risk factors for better stroke prediction has been demonstrated in the recent INTERSTROKE study (4). Importantly, the Stroke Riskometer user can find out not only their absolute risk of stroke development but also a baseline risk to compare their risk against, thus allowing them to know their risk of stroke compared with someone of their age and gender who has no risk factors. The former represents a new paradigm for high-risk stroke prevention strategy, which is distinctly different from the traditional threshold-based approach in which people are categorized into low, moderate, and high-risk groups (20). However, the threshold-based high-risk approach is for use by health professionals and is less appealing to the individuals concerned because it does not tell them at what particular risk of CVD/stroke they are compared with people of the same age and gender but without any additional risk factors. For example, a woman of 35 years old who has a family history of stroke (her father had a stroke at the age of 64),eats less than six servings of fruits and/or vegetables a day, has experienced significant mental or emotional stress (permanent or several periods) over the last year, has a systolic blood pressure reading of 128 mmHg, and has an absolute risk of stroke within the next 5 years of 0·15%.

had a stroke at the age of 64),eats less than six servings of fruits and/or vegetables a day, has experienced significant mental or emotional stress (permanent or several periods) over the last year, has a systolic blood pressure reading of 128 mmHg, and has an absolute risk of stroke within the next 5 years of 0·15%. Although her absolute risk is relatively low, her relative risk of stroke is roughly 1·7 times greater than someone of her age and gender who has no contributing risk factors (0·09%), and this is likely to be perceived by the woman as the significant stimulus to reduce her risk of having a stroke. However, if we base our estimates on one of the threshold-based CVD risk charts (21), the woman's risk falls into a mild CVD risk category, which seems to be far less of a motivation for her to reduce her risk of stroke. This represents a new application of the basic underlying idea of targeting the whole population instead of a high-risk group for primary prevention (so-called ‘prevention paradox’ concept) first introduced by the epidemiologist Geoffrey Rose in 1981 (14). The key novelty of this approach is twofold. First, it utilizes modern far-reaching technologies [global mobile Internet subscriptions is expected to reach 4·5 billion by the end of 2018, with the mobile phone remaining the most frequently used access device (http://www.ericsson.com/ericsson-mobility-report accessed date 17 March 2014)], allowing huge number of individuals across the globe to calculate their absolute risk of stroke within the next 5 to 10 years and to compare their risk with those of the same age and gender without risk factors. Second, it employs self-management strategies to engage the person concerned in stroke/CVD prevention, which is tailored to the person's individual risk profile. However, whether this smartphone-based high-risk strategy on a population level is effective in changing people's behavior and improving stroke prevention remains to be proven, and we are planning an international clinical trial to test this hypothesis. Current feedback from over 5300 Stroke Riskometer app users is encouraging, and the app is being validated against the Framingham stroke prediction algorithm.

in changing people's behavior and improving stroke prevention remains to be proven, and we are planning an international clinical trial to test this hypothesis. Current feedback from over 5300 Stroke Riskometer app users is encouraging, and the app is being validated against the Framingham stroke prediction algorithm. There are also other important features of the Stroke Riskometer to improve stroke prevention. For example, people can change details they enter in the Stroke Riskometer to visually observe the effect of changes in their risk of having a stroke. The Stroke Riskometer can also be used to estimate a risk of recurrent stroke by people who have already experienced a stroke or TIA and to calculate their risk of having a heart attack within the next 5 and 10 years. It can also be used for monitoring progress of stroke and heart attack prevention. The app (version 1) is available as a free download for iPhone, iPad [https://itunes.apple.com/nz/app/stroke-riskometer/id725335272?mt=8 (access date 17 March 2014)], Android tablets, and other smartphone users [https://play.google.com/store/search?q=stroke%20riskometer&hl=en (access date 17 March 2014)] (Fig. 1). The World Stroke Organization has endorsed the free version of the Stroke Riskometer [http://www.world-stroke.org/education/stroke-riskometer (access date 17 March 2014)]. In addition, a premium version of the Stroke Riskometer app (Stroke Riskometer Pro) contains audio-visual educational information about stroke and stroke prevention and provides recommendations from experts on how to manage risk factors based on a person's stroke risk profile to reduce the risk of having a stroke or heart attack. The self-management of stroke and CVD risk factors is also a new and appealing approach in preventative medicine (22,23). All the education information and recommendations are based on internationally recognized primary stroke prevention guidelines (5). Tailoring the recommendations to a particular person and presenting them in a self-management format is likely to improve the uptake of this educational information. It is expected that the self-management, individual risk profile-oriented recommendations together with the educational and stroke risk self-assessment tool will empower a user to take control of and improve their stroke prevention.

self-management format is likely to improve the uptake of this educational information. It is expected that the self-management, individual risk profile-oriented recommendations together with the educational and stroke risk self-assessment tool will empower a user to take control of and improve their stroke prevention. The increased utilization of modern mobile technologies such as smartphones provides a unique opportunity to introduce new prevention strategies to those at elevated risk of stroke or CVD (there are currently 1 billion smartphone users in the world and this number is projected to increase to 1·4 billion by the end of 2013). Although this is a new ‘mass-elevated risk stroke/CVD prevention’ approach to stroke/CVD prevention at an individual level, it is important to note that the Stroke Riskometer is not designed to be a replacement for seeking professional medical advice, but rather an educational self-management tool that allows users to identify if they fall into a risk category and take preventive measures to protect against future strokes. Therefore, the Stroke Riskometer app should be considered as an addition to the currently adopted absolute stroke/CVD prevention approach. It should also be noted that estimates based on the Stroke Riskometer app may not be generalizable to a global audience, and further research should be carried out to validate and improve accuracy of the stroke prediction algorithm for different populations in different countries. Fig. 1 Some snapshots of the Stroke Riskometer App screens from some smartphones and iPad.

Introduction and rationale Following stroke or transient ischemic attack (TIA), the risk of recurrence is very high over the first few hours and days, reaching 10·3% by three‐months 1, 2. Risk then declines and totals about 40% by five‐years. Importantly, recurrent strokes are usually more severe than first events and so are more likely to lead to dependency, cognitive impairment and dementia, depression, poor quality of life, and need for long‐term institutional care 3, 4. The long‐term risk of recurrence can be reduced, but not abolished, with lifestyle changes (reducing weight, saturated fat, salt and high alcohol intake, and stopping smoking) and evidence‐based and cost‐effective clinical interventions including lowering blood pressure (BP) (all stroke and TIA) and lipids (ischemic stroke and TIA), and carotid endarterectomy (large artery ischemic stroke and TIA) 1, 2, 5, 6, 7, 8, 9, 10. While oral anticoagulants are established therapy after cardioembolic stroke and TIA 11, 12, 13, 14, the majority of patients with acute and chronic ischemic stroke or TIA need antiplatelets 15, 16, 17, 18, 19, 20, 21, 22, 23.

), and carotid endarterectomy (large artery ischemic stroke and TIA) 1, 2, 5, 6, 7, 8, 9, 10. While oral anticoagulants are established therapy after cardioembolic stroke and TIA 11, 12, 13, 14, the majority of patients with acute and chronic ischemic stroke or TIA need antiplatelets 15, 16, 17, 18, 19, 20, 21, 22, 23. Antiplatelet therapy for acute ischemic stroke is based on aspirin alone as a result of the IST‐1 and CAST mega‐trials 17, 18, although the effect size for improving functional outcome was small (absolute risk reduction ∼1·1%), mostly explained by aspirin reducing early recurrence. Until recently, the acute treatment of TIA had not been investigated. Early and short‐term use of two agents appears to be superior to monotherapy, as suggested by observational studies (EXPRESS, SOS 1, 2), small trials (FASTER, EARLY) 24, 25, and a post hoc subgroup analysis of the PRoFESS mega‐trial 26. These findings were strengthened by the large CHANCE trial that showed that the combination of aspirin + clopidogrel was superior to aspirin alone in reducing stroke recurrence 23. Indeed, it appears in meta‐analyses that any pair of antiplatelets is superior to any single agent 27, 28. A potential advantage of multi‐antiplatelet therapy is that it will help cover treatment resistance seen with monotherapy for either aspirin or clopidogrel 29, 30, 31.

pirin alone in reducing stroke recurrence 23. Indeed, it appears in meta‐analyses that any pair of antiplatelets is superior to any single agent 27, 28. A potential advantage of multi‐antiplatelet therapy is that it will help cover treatment resistance seen with monotherapy for either aspirin or clopidogrel 29, 30, 31. The situation in acute stroke and TIA differs from chronic stroke (long‐term secondary prophylaxis) where dual therapy with aspirin + dipyridamole reduced events by 23% in comparison with aspirin or dipyridamole alone without increasing the risk of bleeding (ESPS‐2, ESPRIT) 19, 21. However, dual therapy based on aspirin + clopidogrel was not superior to monotherapy with either agent alone (CHARISMA, MATCH) 32, 33. In MATCH, dual therapy caused more bleeding than clopidogrel alone but without a significant reduction in recurrence 33, 34.

out increasing the risk of bleeding (ESPS‐2, ESPRIT) 19, 21. However, dual therapy based on aspirin + clopidogrel was not superior to monotherapy with either agent alone (CHARISMA, MATCH) 32, 33. In MATCH, dual therapy caused more bleeding than clopidogrel alone but without a significant reduction in recurrence 33, 34. If dual therapy is more effective at preventing recurrence than monotherapy for acute prophylaxis, then intensive therapy with triple antiplatelets (combined aspirin + clopidogrel + dipyridamole) may be better still, provided the risk of recurrence is high and bleeding does not become excessive. We have performed a series of ‘proof‐of‐mechanism’ and ‘proof‐of‐concept’ laboratory studies and clinical trials investigating this approach 35, 36, 37, 38, 39. In vitro studies starting in 2000 found that triple therapy was most effective in inhibiting platelet aggregation, platelet–leucocyte conjugation, and leucocyte activation 35, 36, 37. In multiway crossover phase I and II trials, short‐term administrations of mono (aspirin, clopidogrel, or dipyridamole), dual (combinations of aspirin and clopidogrel, aspirin and dipyridamole, or clopidogrel and dipyridamole), and triple (combined aspirin, clopidogrel, and dipyridamole) antiplatelet therapy were compared; the combination of aspirin and clopidogrel, with or without dipyridamole, was most potent in inhibiting platelet function ex vivo in both normal volunteers and participants with previous stroke/TIA 38, 39. (Of note, the platelet function tests used in these studies are relatively insensitive to the intracellular effects of dipyridamole.) In the only parallel group trial of intensive/triple therapy in participants with stroke, we found that combined aspirin, clopidogrel, and dipyridamole (vs. aspirin alone, chosen because it was the UK standard of care at the time) was feasible to administer in a pilot trial for up to 24 months 40. However, the trial was stopped early on publication of ESPRIT 21 confirming the superiority of combined aspirin and dipyridamole over aspirin alone. There was a non‐significant trend to increased bleeding with triple antiplatelet therapy vs. aspirin alone. Although unintended, the participants were at low risk of recurrence (young/recruited months after the event/many lacunar strokes) 40, a problem also seen in MATCH and CHARISMA 32, 33.

idamole over aspirin alone. There was a non‐significant trend to increased bleeding with triple antiplatelet therapy vs. aspirin alone. Although unintended, the participants were at low risk of recurrence (young/recruited months after the event/many lacunar strokes) 40, a problem also seen in MATCH and CHARISMA 32, 33. The study concluded that future trials of combined aspirin, clopidogrel, and dipyridamole needed to target participants at high risk of recurrence and for a short treatment duration to minimize bleeding, so that benefit is likely to outweigh hazard. Clinical use of triple antiplatelet therapy has also been reported in a case series 41. The TARDIS trial was designed to build on these laboratory and clinical studies and aims to test the overall safety and efficacy of intensive antiplatelet therapy with three agents in comparison with guideline treatment. Primary research question Is intensive antiplatelet therapy (combined aspirin, clopidogrel, and dipyridamole) safe and effective in reducing recurrence and its severity at three‐months, as compared with guideline antiplatelet therapy (clopidogrel, or combined aspirin and dipyridamole), when given acutely after stroke or TIA for one‐month? Methods ¶ refers to a change from the current Protocol version 1·5 (downloadable from http://www.nets.nihr.ac.uk/projects/hta/1010424). Design TARDIS is an international collaborative multicenter parallel‐group prospective randomized open‐label blinded‐end‐point phase III controlled trial. Patient population Inclusion criteria:Age ≥50 years Event to randomization ≤48 h (24–48 h if thrombolysed)

Methods ¶ refers to a change from the current Protocol version 1·5 (downloadable from http://www.nets.nihr.ac.uk/projects/hta/1010424). Design TARDIS is an international collaborative multicenter parallel‐group prospective randomized open‐label blinded‐end‐point phase III controlled trial. Patient population Inclusion criteria:Age ≥50 years Event to randomization ≤48 h (24–48 h if thrombolysed) Index event is a TIA (defined in supplement of Statistical Analysis Plan 42) with: ○  Resolved limb weakness and/or dysphasia ○  Duration 10 min to <24 h ○  ABCD2 score ≥4; AND/OR crescendo TIA; AND/OR already on dual antiplatelet therapy Index event is a non‐cardioembolic ischemic stroke with: ○  Ongoing limb weakness OR ongoing facial weakness with resolved limb weakness; AND/OR dysphasia; AND/OR ongoing isolated hemianopia (with positive neuroimaging evidence showing ischemic stroke in occipital lobe); AND duration ≥one‐hour ○  Resolved limb weakness; AND/OR dysphasia; AND duration >24 h after onset (i.e. resolution between 24 h and randomization) Willing and able to provide written informed consent; proxy consent is acceptable if patients are dysphasic or confused, in accordance with the practice of the local site Exclusion criteria:Isolated sensory symptoms, facial weakness, or vertigo/dizziness Isolated hemianopia without positive neuroimaging evidence Intracranial haemorrhage Baseline neuroimaging shows intracranial haemorrhage or parenchymal haemorrhagic transformation (PH 1 or 2) of infarct, subarachnoid haemorrhage, or other non‐ischemic cause for symptoms

Exclusion criteria:Isolated sensory symptoms, facial weakness, or vertigo/dizziness Isolated hemianopia without positive neuroimaging evidence Intracranial haemorrhage Baseline neuroimaging shows intracranial haemorrhage or parenchymal haemorrhagic transformation (PH 1 or 2) of infarct, subarachnoid haemorrhage, or other non‐ischemic cause for symptoms Presumed cardioembolic stroke (e.g. history of current atrial fibrillation (AF), myocardial infarction <three‐months) Contraindications to, or intolerance of, aspirin, clopidogrel, or dipyridamole Definite need for aspirin, clopidogrel, or dipyridamole individually or in combination (e.g. aspirin and clopidogrel for recent myocardial infarction (MI)/acute coronary syndrome) Definite need for full dose oral (e.g. apixaban, dabigatran, rivaroxaban, warfarin) or medium to high dose parenteral (e.g. heparin) anticoagulation Definite need for glycoprotein IIb/IIIa inhibitor No enteral access Pre‐morbid dependency [modified Rankin Scale (mRS) > 2] Severe high BP (BP > 185/110 mmHg) Haemoglobin <100 g/l Platelet count <100 × 109/l or >600 × 109/l White cell count <3·5 × 109/l or >30 × 109/l Major bleeding within one‐year (e.g. peptic ulcer, intracerebral haemorrhage) Planned surgery in next three‐months (e.g. known need for carotid endarterectomy) Concomitant acute coronary syndrome [e.g. ST segment elevation myocardial infarction (STEMI) or non‐STEMI (NSTEMI)] Stroke secondary to a procedure (e.g. carotid or coronary intervention) Coma [Glasgow Coma Scale (GCS) < 8] Non‐stroke life expectancy <six‐months Known dementia Women of childbearing potential, pregnant, or breastfeeding

Planned surgery in next three‐months (e.g. known need for carotid endarterectomy) Concomitant acute coronary syndrome [e.g. ST segment elevation myocardial infarction (STEMI) or non‐STEMI (NSTEMI)] Stroke secondary to a procedure (e.g. carotid or coronary intervention) Coma [Glasgow Coma Scale (GCS) < 8] Non‐stroke life expectancy <six‐months Known dementia Women of childbearing potential, pregnant, or breastfeeding Geographical or other factors that may interfere with follow‐up Patients who have not had post‐thrombolysis neuroimaging Patients may be enrolled concurrently into observational studies or non‐drug/device trials. Baseline measures Baseline demographic details (age, gender, race‐ethnicity), pre‐morbid mRS, clinical details (syndrome 43), stroke severity [National Institutes of Health Stroke Scale, (NIHSS) 44 ], BP, full blood count, and electrocardiogram (ECG) are determined after consent/assent and before randomization.

Patients may be enrolled concurrently into observational studies or non‐drug/device trials. Baseline measures Baseline demographic details (age, gender, race‐ethnicity), pre‐morbid mRS, clinical details (syndrome 43), stroke severity [National Institutes of Health Stroke Scale, (NIHSS) 44 ], BP, full blood count, and electrocardiogram (ECG) are determined after consent/assent and before randomization. Neuroimaging – computerized tomography (CT) or magnetic resonance (MR) scanning – is performed for patients with ischemic stroke to exclude intracranial haemorrhage and non‐stroke diagnoses. If thrombolysis is performed, CT/MR must be undertaken afterward and prior to randomization to exclude haemorrhagic transformation. Patients presenting with a TIA do not have to have a CT/MR as this reflects routine practice at many stroke centers. Patients with cerebral events that occur during treatment must also be re‐scanned to identify potential secondary bleeding. Local site reporting of scans is recorded; all scans are also uploaded over the Internet for independent adjudication using a validated structured classification system [as used in IST‐3 and Efficacy of Nitric Oxide in Stroke trial (ENOS)] 45, 46, 47, 48 and masked to treatment. At baseline and day 7 ± 1, optional research blood samples may be taken for substudies involving biomarkers and genetics; some samples are centrifuged to collect plasma and serum, and then frozen.

Neuroimaging – computerized tomography (CT) or magnetic resonance (MR) scanning – is performed for patients with ischemic stroke to exclude intracranial haemorrhage and non‐stroke diagnoses. If thrombolysis is performed, CT/MR must be undertaken afterward and prior to randomization to exclude haemorrhagic transformation. Patients presenting with a TIA do not have to have a CT/MR as this reflects routine practice at many stroke centers. Patients with cerebral events that occur during treatment must also be re‐scanned to identify potential secondary bleeding. Local site reporting of scans is recorded; all scans are also uploaded over the Internet for independent adjudication using a validated structured classification system [as used in IST‐3 and Efficacy of Nitric Oxide in Stroke trial (ENOS)] 45, 46, 47, 48 and masked to treatment. At baseline and day 7 ± 1, optional research blood samples may be taken for substudies involving biomarkers and genetics; some samples are centrifuged to collect plasma and serum, and then frozen. Randomization All randomization, data collection, and serious adverse event (SAE) and CT adjudication are performed over a secure password‐protected and data‐encrypted Internet website: www.tardistrial.org. Patients are randomized in real time with: Stratification on:Index event: stroke/TIA Country Minimization on key prognostic baseline factors:Age: ≤70 vs. >70 years Gender: female, male Pre‐morbid mRS: 0, >0 Time, stroke/TIA to randomization: 24 vs. <24 h Number of antiplatelets before index event: 0/1, 2 Clinical syndrome: lacunar (LACS/POCS), cortical (PACS/TACS) 49

Randomization All randomization, data collection, and serious adverse event (SAE) and CT adjudication are performed over a secure password‐protected and data‐encrypted Internet website: www.tardistrial.org. Patients are randomized in real time with: Stratification on:Index event: stroke/TIA Country Minimization on key prognostic baseline factors:Age: ≤70 vs. >70 years Gender: female, male Pre‐morbid mRS: 0, >0 Time, stroke/TIA to randomization: 24 vs. <24 h Number of antiplatelets before index event: 0/1, 2 Clinical syndrome: lacunar (LACS/POCS), cortical (PACS/TACS) 49 Systolic BP: ≤160, >160 mmHg Gastro‐protection: yes, no Use of low dose heparin: no, yes Additional minimization is performed if the index event is an ischemic stroke:NIHSS: 0–3, >3 Treatment with alteplase: yes, no Additional minimization is performed if the index event is a TIA:ABCD2 score: 0–5, >5 Number of TIAs in last week: 0/1, >1 Simple randomization:In 5% of patients Stratification and minimization allow for improved matching at baseline, stratification allows variable categories to be treated as trials in their own right, minimization increases statistical power 50, and simple randomization reduces predictability. Investigational medicinal products Trial interventions are given open label for one‐month (28 or 30 days depending on treatment pack size, to cover the period of maximum risk of recurrence but minimize bleeding) and comprise (Fig. 1):Aspirin: loading dose 300 mg 51 then 50–150 mg daily; by oral, nasogastric tube (NGT), or rectal route Clopidogrel: loading dose 300 mg 52 then 75 mg daily; by oral or NGT route

Investigational medicinal products Trial interventions are given open label for one‐month (28 or 30 days depending on treatment pack size, to cover the period of maximum risk of recurrence but minimize bleeding) and comprise (Fig. 1):Aspirin: loading dose 300 mg 51 then 50–150 mg daily; by oral, nasogastric tube (NGT), or rectal route Clopidogrel: loading dose 300 mg 52 then 75 mg daily; by oral or NGT route Dipyridamole: 225–450 mg in divided doses; by oral (as 200 mg extended release capsules twice daily 53, or as tablets three to four times daily) or NGT (as suspension or crushed tablets three to four times daily) route Figure 1 Treatment (day 0–30) and follow‐up (to day 90). A, aspirin; C, clopidogrel; D, dipyridamole. figure2015 World Stroke OrganizationTreatment groups comprise:Intensive/triple antiplatelets (active): Combined aspirin, clopidogrel, and dipyridamole Guideline/dual or mono antiplatelet(s): Combined aspirin and dipyridamole, or clopidogrel alone [monotherapy was added following a change in UK National Institute for Health and Care Excellence (NICE) guidance 54] Sites choose in advance which guideline treatment regimen they wish to use, a choice that is made separately for ischemic stroke (IS) and TIA:Aspirin + dipyridamole, or clopidogrel (1:1) Aspirin + dipyridamole Clopidogrel The principal investigator (PI) may change the choice of comparator at any stage during the trial with 48‐h notice, that is, a change cannot influence treatment for a patient who is in the process of being enrolled.

Sites choose in advance which guideline treatment regimen they wish to use, a choice that is made separately for ischemic stroke (IS) and TIA:Aspirin + dipyridamole, or clopidogrel (1:1) Aspirin + dipyridamole Clopidogrel The principal investigator (PI) may change the choice of comparator at any stage during the trial with 48‐h notice, that is, a change cannot influence treatment for a patient who is in the process of being enrolled. Antiplatelet agents may be administered after the index stroke/TIA event and before randomization as follows:Aspirin may be given after the index event and prior to randomization in any potential trial participant. Clopidogrel is only allowed if the patient may receive it as part of the trial according to the PI's choice of comparator, that is, the patient can only be randomized to intensive antiplatelets vs. clopidogrel alone. Dipyridamole (in combination with aspirin) is only allowed if the patient may receive it as part of the trial according to the PI's choice of comparator, that is, the patient can only be randomized to intensive antiplatelets vs. combined aspirin and dipyridamole. If the patient is given a ‘confounding’ antiplatelet after their event and before randomization, the patient may still be included but randomization will then only involve the appropriate comparison to prevent confounding of treatment. Patients who have received combined clopidogrel and aspirin, clopidogrel and dipyridamole, cilostazol (whether singly or in combination), or triflusal (whether singly or in combination) are excluded from the trial.

If the patient is given a ‘confounding’ antiplatelet after their event and before randomization, the patient may still be included but randomization will then only involve the appropriate comparison to prevent confounding of treatment. Patients who have received combined clopidogrel and aspirin, clopidogrel and dipyridamole, cilostazol (whether singly or in combination), or triflusal (whether singly or in combination) are excluded from the trial. Study drugs may be stopped around procedures that become necessary after enrollment; trial drugs should be re‐started as soon as possible after the procedure once clinically appropriate. Participants can be withdrawn from therapy either at their own request, for safety reasons, or if unacceptable adverse events develop. After the 28–30‐day treatment period, participants are expected to return to guideline antiplatelet therapy as recommended by local, national, and international guidelines. Patients are also offered standard ‘best care’ prophylaxis, including lifestyle advice, BP and lipid‐lowering drugs, and carotid endarterectomy (as necessary). Primary outcome The primary outcome is the frequency and severity of recurrent strokes and TIA in participants who have a recurrent event, with assessment at day 90. Severity is measured using a six‐level ordered categorical scale that incorporates the mRS:Fatal stroke/severe non‐fatal stroke (mRS 4 or 5)/moderate stroke (mRS 2 or 3)/mild stroke (mRS 0 or 1)/TIA/no stroke‐TIA¶

verity of recurrent strokes and TIA in participants who have a recurrent event, with assessment at day 90. Severity is measured using a six‐level ordered categorical scale that incorporates the mRS:Fatal stroke/severe non‐fatal stroke (mRS 4 or 5)/moderate stroke (mRS 2 or 3)/mild stroke (mRS 0 or 1)/TIA/no stroke‐TIA¶ Ascertainment of recurrent events, and mRS, is determined centrally by telephone by a trained assessor who is masked to outcome at day 90 ± 7. To ensure recurrent events are identified, corroborating information is sought from the general practitioner and recruiting hospital site. The effect of the intervention on the primary outcome will be performed within the following subgroups:(a)  Geographical region: UK, other ¶ (b)  Age: ≤70 years, >70 years ¶ (c)  Gender: female, male (d)  mRS: 0, >0 ¶ (e)  Index event: ischemic stroke, TIA (f)  Stroke/TIA syndrome: LACS, POCS, PACS, TACS 49 ¶ (g)  Stroke/TIA etiology: small vessel disease (SVD), large artery disease (LAD), other ¶ (h)  NIHSS (stroke only): 0–3, >3 ¶ (i)  ABCD2 score (TIA only): 0–5, >5 ¶ (j)  Crescendo TIA (TIA only): no, yes ¶ (k)  Number of antiplatelet agents at baseline: 0, 1, 2 ¶ (l)  Type of comparator: AD, C, either ¶ (m)  Systolic BP: ≤140, 141–160, >160 mmHg ¶ (n)  Time, event to randomization >24, 12·1–24, ≤12 h ¶ (o)  Use of low dose heparin: no, yes ¶ (p)  Treated with alteplase prior to randomization (stroke only): yes, no (q)  Gastroprotection: yes, no ¶ (r)  Carotid stenosis (ipsilateral ≥50%): no, yes ¶ (s)  Old lesion on baseline neuroimaging: no, yes ¶

(m)  Systolic BP: ≤140, 141–160, >160 mmHg ¶ (n)  Time, event to randomization >24, 12·1–24, ≤12 h ¶ (o)  Use of low dose heparin: no, yes ¶ (p)  Treated with alteplase prior to randomization (stroke only): yes, no (q)  Gastroprotection: yes, no ¶ (r)  Carotid stenosis (ipsilateral ≥50%): no, yes ¶ (s)  Old lesion on baseline neuroimaging: no, yes ¶ Secondary, bleeding, and safety outcomes Investigators assess secondary outcomes at days 7 ± 1 and 35 ± 3, and on discharge from hospital (if admitted). The National Coordinating Centre assesses secondary outcomes at day 90 ± 7 by a telephone call between the patient (or carer) and an assessor blinded to treatment. Reported outcomes (stroke, MI) and SAEs are adjudicated by a member of the independent adjudicator panel who is blinded to treatment. Day 7 ¶(a)  Headache that required treatment or led to discontinuation (b)  Recurrent stroke or TIA (c)  Impairment (NIHSS, including death) (d)  Neurological deterioration (increase in NIHSS by four points or more) (e)  Composite vascular event (f)  Venous thromboembolism (g)  Haemoglobin (h)  Bleeding (i)  SAEs Day 35 (end of treatment)(a)  Headache that required treatment or led to discontinuation (b)  Recurrent stroke or TIA (c)  Impairment (NIHSS, including death) (d)  Neurological deterioration (increase in NIHSS by four points or more) (e)  Composite vascular (f)  Myocardial infarction (g)  Venous thromboembolism (h)  Haemoglobin (i)  Bleeding (j)  SAEs Hospital discharge (collected at discharge or if death in hospital)(a)  Length of stay in hospital (b)  Discharge disposition (death/institution/home)

(d)  Neurological deterioration (increase in NIHSS by four points or more) (e)  Composite vascular (f)  Myocardial infarction (g)  Venous thromboembolism (h)  Haemoglobin (i)  Bleeding (j)  SAEs Hospital discharge (collected at discharge or if death in hospital)(a)  Length of stay in hospital (b)  Discharge disposition (death/institution/home) Day 90 (end of follow‐up)(a)  Death: time to death (censored at 110 days) and by what cause (b)  Composite vascular (c)  Myocardial infarction 55 (d)  Venous thromboembolism (e)  Barthel Index (BI) (f)  Dead or disabled (BI < 60) (g)  Quality of life/Health Utility Score [derived from European Quality of Life‐5 Dimensions (EQ‐5D)] 56 Δ (h)  Quality of life [European Quality of Life Visual Analog Scale (EQ‐VAS)] Δ (i)  Telephone‐Mini‐Mental State Examination Δ (j)  Telephone Interview Cognition Scale‐Modified Δ (k)  Verbal fluency (animal naming over one‐minute) Δ (l)  Zung Depression Scale (mood) 57 Δ (m)  Disposition (death/institution/home) (n)  Bleeding – by site and severity 58 (o)  SAEs – by time, type, site, and severity 21 Δ will not be collected if carer answers questions without recourse to participant. Sample size TARDIS was designed with a start‐up phase (funded by British Heart Foundation, and assessing safety, feasibility, and tolerability) and a main phase (funded by Health Technology Assessment, and assessing safety and efficacy).

(o)  SAEs – by time, type, site, and severity 21 Δ will not be collected if carer answers questions without recourse to participant. Sample size TARDIS was designed with a start‐up phase (funded by British Heart Foundation, and assessing safety, feasibility, and tolerability) and a main phase (funded by Health Technology Assessment, and assessing safety and efficacy). The null hypothesis (H0) is that intensive antiplatelets will not alter the frequency and severity of stroke/TIA in participants with previous ischemic stroke or TIA. The alternative hypothesis is that the frequency and severity of stroke/TIA differ between those participants randomized to intensive vs. guideline antiplatelets. A total sample size 59, 60 of 4100 (2050 per group) participants with ischemic stroke or TIA is required, assuming:Overall significance (alpha) = 0·05 Power (1‐beta) = 0·90 Odds ratio of 0·68 (equivalent to an odds ratio of 0·57 and relative risk reduction = 0·31 for binary stroke) Distribution in outcome based on recurrent stroke and its severity using mRS (based on data from n = 1460 participants with a final outcome): ○  Fatal stroke, 0·55%/mRS 4 or 5, 0·55%/mRS 2 or 3, 1·30%/mRS 0 or 1, 1·23%/TIA, 3·22%/no event, 93·15% Treatment crossovers = 5·0% Losses to follow‐up = 2% Reduction of 20% for baseline covariate adjustment 61

Distribution in outcome based on recurrent stroke and its severity using mRS (based on data from n = 1460 participants with a final outcome): ○  Fatal stroke, 0·55%/mRS 4 or 5, 0·55%/mRS 2 or 3, 1·30%/mRS 0 or 1, 1·23%/TIA, 3·22%/no event, 93·15% Treatment crossovers = 5·0% Losses to follow‐up = 2% Reduction of 20% for baseline covariate adjustment 61 Statistical analyses Analyses will be performed by intention‐to‐treat using binary logistic regression for binary outcomes, ordinal logistic regression for ordered categorical variables (including the primary outcome), multiple regression for continuous variables, and Cox proportional hazards regression for time‐to‐event data. Analyses will be adjusted for stratification and minimization factors. Detailed analysis plans are given in the Statistical Analysis Plan. Study organization and oversight TARDIS is an independent academic trial performed by an international collaborative group. The Trial Steering Committee provides oversight and strategic input, and comprises independent members, grant applicants, and patient, sponsor, and funder representatives; it meets twice yearly. An International Advisory Committee meets annually and provides advice on national issues including recruitment and follow‐up. The Trial Management Committee runs the trial on a day‐to‐day basis and is based at the TARDIS Trial Coordinating Centre in Nottingham. A National Coordinating Centre and national coordinator are based in each participating country. Outcomes, SAEs, and brain imaging are adjudicated by trained assessors masked to treatment assignment.

ement Committee runs the trial on a day‐to‐day basis and is based at the TARDIS Trial Coordinating Centre in Nottingham. A National Coordinating Centre and national coordinator are based in each participating country. Outcomes, SAEs, and brain imaging are adjudicated by trained assessors masked to treatment assignment. The independent Data Monitoring Committee reviews unblinded data twice yearly in respect of safety and efficacy, and review recruitment, baseline data, balance in baseline factors between the treatment groups, completeness of data, compliance to treatment, co‐administered treatments, outcome by subgroups, SAEs (both adjudicated and unadjudicated), and protocol violations. They also take findings in the context of other published evidence.

w recruitment, baseline data, balance in baseline factors between the treatment groups, completeness of data, compliance to treatment, co‐administered treatments, outcome by subgroups, SAEs (both adjudicated and unadjudicated), and protocol violations. They also take findings in the context of other published evidence. Research governance TARDIS is conducted in accordance with the ethics and principles enshrined in the Declaration of Helsinki and good clinical practice, and is run in accordance with the UK Medicines for Human Use Regulations and Health Research Governance Framework. The management of personal data adheres to the UK Data Protection Act (1998). The trial has approval from the Medicines and Healthcare Products Regulatory Agency (reference 03057/0027/001‐0001, date 17/10/2008), Eudract number 2007‐006749‐21, and National Research Ethics Committee (Reference 08/H1102/112, date 9/1/2009). All sites have local Research Ethics Committee (REC) and NHS Research and Development (R&D) approvals. The trial is registered with Current Controlled Trials (ISRCTN47823388) and has been adopted by the UK NIHR Stroke Research Network, and endorsed by the Australasian Stroke Trials Network.

8/H1102/112, date 9/1/2009). All sites have local Research Ethics Committee (REC) and NHS Research and Development (R&D) approvals. The trial is registered with Current Controlled Trials (ISRCTN47823388) and has been adopted by the UK NIHR Stroke Research Network, and endorsed by the Australasian Stroke Trials Network. Summary and conclusions TARDIS is addressing a key issue in the management of patients with acute stroke and TIA, namely the safety and efficacy of short‐term intensive (combined aspirin, clopidogrel, and dipyridamole) vs. guideline antiplatelet therapy. The primary outcome is the frequency and severity of stroke recurrence; TARDIS is the first trial to use this novel end‐point. The sample size of 4100 patients means that a modest but worthwhile clinical effect can be detected with high statistical power (90%); to date, 2399 patients have been recruited from 104 sites in 4 countries, with one‐third presenting with a TIA. A positive trial would mean that triple antiplatelet therapy could be introduced rapidly into clinical practice as the drugs are already licensed, readily available, and inexpensive. We invite centers from around the world to join this important collaborative international venture. Supporting information Appendix S1. Full list of acknowledgements. Click here for additional data file. Acknowledgements Please see Appendix S1 for the full list.

Introduction and rationale Thrombectomy in acute ischemic stroke (AIS) patients with intracranial large vessel occlusion in the anterior circulation results in clear clinical benefits without increased adverse events. However, previous randomized clinical trials (RCTs) have limited this treatment option to patients with good prognostic factors (i.e. short time interval between stroke onset and endovascular treatment as well as small size of ischemic lesion prior to treatment).1–5 Of 1287 patients randomized in five large RCTs less than 5% were treated beyond 6 h of symptom onset. In four of these five trials, patients with early ischemic signs seen as an Alberta Stroke Program Early Computed Tomography Scan score (ASPECTS) below 6 or 7 were excluded. Only MR CLEAN did not specify pretreatment ASPECTS values as an exclusion criterion, but even in this study only small numbers of patients with low ASPECTS values were included. In 28 patients with ASPECTS 0–4 no treatment effect was evident with an OR of 1.1 (95% CI 0.1–8.4) for benefit in the thrombectomy group, while in 92 patients with ASPECTS 5–7 with an OR of 2.0 (0.9–4.4) a clear trend for a treatment benefit was observed.1 In a meta-analysis of individual patient data from the five positive thrombectomy trials, no treatment effect was observed for patients with ASPECTS 0–5 with an OR of 1.24 (95% CI 0.62–2.49).6 A recently published bicentric registry study which included 218 patients with a diffusion-weighted imaging (DWI)-ASPECTS ≤ 6 found an increased rate of favorable outcomes (modified Rankin Scale (mRS)≤2 at 90 days) and a decreased rate of mortality in reperfused patients (TICI score ≥2b) compared to nonreperfused patients.7 However, the rate of favorable outcomes did not differ significantly and mortality increased in patients with a DWI-APECTS <5. Additionally, this study did not evaluate outcomes in patients with a low ASPECTS who did not undergo any endovascular procedure. Thus, insufficient evidence is available to judge whether mechanical thrombectomy is also safe and effective in patients with an extended time window or signs of early ischemia, or a combination of both.

this study did not evaluate outcomes in patients with a low ASPECTS who did not undergo any endovascular procedure. Thus, insufficient evidence is available to judge whether mechanical thrombectomy is also safe and effective in patients with an extended time window or signs of early ischemia, or a combination of both. This was also acknowledged by experts from the European Stroke Organisation, the European Society for Minimally Invasive Neurological Therapy (ESMINT), and the European Society of Neuroradiology in a consensus statement on thrombectomy.8 In this statement the experts recommended further RCTs addressing open issues such as “treatment in a late and unknown time windows, treating patients with imaging findings not sufficiently covered in recent trials.…” Recently, the efficacy of thrombectomy in an extended time window has been shown in the DAWN and DEFUSE 3 trials.9,10 However, these studies only included patients with an initial small infarct core. The question remains if treatment is still beneficial and safe with an extended ischemic lesion prior to treatment. The TENSION (efficacy and safety of ThrombEctomy iN Stroke with extended leSION and extended time window: a randomized, controlled trial) trial aims to close this gap. As it is estimated that about 25% of acute stroke patients with LVO present with an ASPECTS of 0–5 this issue is highly relevant in clinical practice. The primary objective of TENSION is to test efficacy and safety of thrombectomy in acute stroke patients with large vessel occlusion and extended ischemic lesion size (ASPECTS 3–5) in an extended time window (up to 12 h or unknown onset).11

sent with an ASPECTS of 0–5 this issue is highly relevant in clinical practice. The primary objective of TENSION is to test efficacy and safety of thrombectomy in acute stroke patients with large vessel occlusion and extended ischemic lesion size (ASPECTS 3–5) in an extended time window (up to 12 h or unknown onset).11 Methods Design TENSION is an European investigator-initiated, randomized controlled, open label, blinded endpoint (PROBE), two-arm, randomized, postmarket study to compare the safety and effectiveness of endovascular thrombectomy as compared to best medical care alone in the treatment of AIS in patients with extended stroke lesions defined by an ASPECTS of 3–5 and in an extended time window (up to 12 h from onset or last known well). The study will apply an adaptive design study with interim analyses with prespecified stopping rules allowing for the possibility of early termination based on either a determination of study success or futility. As of January 2018, 40 centers in eight European countries (Austria, Czech Republic, Denmark, France, Germany, Norway, Slovakia, Sweden) have agreed to participate.

Methods Design TENSION is an European investigator-initiated, randomized controlled, open label, blinded endpoint (PROBE), two-arm, randomized, postmarket study to compare the safety and effectiveness of endovascular thrombectomy as compared to best medical care alone in the treatment of AIS in patients with extended stroke lesions defined by an ASPECTS of 3–5 and in an extended time window (up to 12 h from onset or last known well). The study will apply an adaptive design study with interim analyses with prespecified stopping rules allowing for the possibility of early termination based on either a determination of study success or futility. As of January 2018, 40 centers in eight European countries (Austria, Czech Republic, Denmark, France, Germany, Norway, Slovakia, Sweden) have agreed to participate. Patient population Patients presenting with AIS based on focal occlusion in the M1 segment of the middle cerebral artery, and/or the intracranial segment of the distal internal carotid artery (ICA), determined by magnetic resonance angiography (MRA) or computed tomography angiography (CTA), and who meet all eligibility criteria will be considered for study enrolment. Table 1 lists the inclusion and exclusion criteria. Up to 665 subjects, 333 per treatment group, will be enrolled and randomized for the intent to treat (ITT) analysis. A screening log of all potential patients will be kept locally at each center. Table 1. TENSION inclusion and exclusion criteria

d for study enrolment. Table 1 lists the inclusion and exclusion criteria. Up to 665 subjects, 333 per treatment group, will be enrolled and randomized for the intent to treat (ITT) analysis. A screening log of all potential patients will be kept locally at each center. Table 1. TENSION inclusion and exclusion criteria Clinical inclusion criteria Clinical exclusion criteria Moderate to severe stroke (NIHSS score <26) Patient is an active participant in another drug or device treatment trial Premorbid modified Rankin Scale (mRS) score 0–2 Patient has preexisting neurological or psychiatric disease that could impede the study results or would confound the neurological or functional evaluations Life expectancy >6 months Patient has vascular disease preventing endovascular treatment (e.g. aortic dissection or aneurysm, no arterial transfemoral access) Age >18–80 years Patient has history of contraindication for contrast medium Treatment can be accomplished within 12 h after stroke onset (if known), i.e. randomization within 11 h after ictus Patient is known to have infective endocarditis Informed consent by the patient, legal guardian, or inclusion of patient under presumptive will, in accordance with national regulations after consultation of an independent physician and statement of investigator Patient's anticipated life expectancy is less than six months Imaging inclusion criteria Imaging exclusion criteria

o our question. These criteria were adapted from various quality assessment tools available for qualitative research and selected after discussion between three reviewers.22,23 Two reviewers performed the ratings independently and evaluations were compared. Disagreements were settled through consensus-based discussion. Results Search results The database search identified 1985 citations (Figure 1). An additional 44 were later identified by hand-searching citation lists. With duplicates removed, there were a total of 1666 references. After title and abstract screening, 1456 records were excluded and the remaining 210 articles were assessed for eligibility with full-text review. Ultimately 80 records were included in this review (see Supplementary References). Reasons for exclusion were categorized as: (1) outcome not of interest; (2) abstract only, without enough information provided; (3) language other than English; (4) wrong population; (5) wrong setting; and (6) previous publication with duplicate material. A reference list of the excluded studies with reasons for exclusion is presented in the Supplementary Methods. Figure 1. PRISMA flow diagram.

Clinical inclusion criteria Clinical exclusion criteria Moderate to severe stroke (NIHSS score <26) Patient is an active participant in another drug or device treatment trial Premorbid modified Rankin Scale (mRS) score 0–2 Patient has preexisting neurological or psychiatric disease that could impede the study results or would confound the neurological or functional evaluations Life expectancy >6 months Patient has vascular disease preventing endovascular treatment (e.g. aortic dissection or aneurysm, no arterial transfemoral access) Age >18–80 years Patient has history of contraindication for contrast medium Treatment can be accomplished within 12 h after stroke onset (if known), i.e. randomization within 11 h after ictus Patient is known to have infective endocarditis Informed consent by the patient, legal guardian, or inclusion of patient under presumptive will, in accordance with national regulations after consultation of an independent physician and statement of investigator Patient's anticipated life expectancy is less than six months Imaging inclusion criteria Imaging exclusion criteria Occlusion of the M1 segment of the middle cerebral artery (MCA) and/or the intracranial segment of the distal internal carotid artery (ICA), determined by MRA or CTA CT scan or MRI with evidence of mass effect or intracranial tumor, or hypodensity on unenhanced CT and cerebral blood volume (CBV) drop on CBV maps on CT perfusion, or, alternatively as per institutional standard, restricted diffusion on DWI with an ASPECT score of 0–2, or above 5 CT (noncontrast CT) or DWI with an ASPECT score of 3–5 Any other finding on brain CT or MRI considered as indicative of a high risk of SICH related to potential thrombectomy treatment in the judgment of the investigator ASPECTS: Alberta Stroke Program Early CT Score; CT: computed tomography; CTA: computed tomography angiography; DWI: diffusion-weighted imaging; MRA: magnetic resonance angiography; MRI: magnetic resonance imaging; NIHSS: National Institutes of Health Stroke Scale; SICH: symptomatic intracranial hemorrhage; TENSION: efficacy and safety of ThrombEctomy iN Stroke with extended leSION and extended time window: a randomized, controlled trial.

sion-weighted imaging; MRA: magnetic resonance angiography; MRI: magnetic resonance imaging; NIHSS: National Institutes of Health Stroke Scale; SICH: symptomatic intracranial hemorrhage; TENSION: efficacy and safety of ThrombEctomy iN Stroke with extended leSION and extended time window: a randomized, controlled trial. Randomization Patients are randomized in the two treatment arms using a web-based system with a 1:1 ratio. Stratified randomization by time from symptom onset/last known well (0–6 and >6 h) and stroke severity (National Institutes of Health Stroke Scale (NIHSS) ≤18, NIHSS >18) may minimize imbalances that might affect outcome and bias results. Due to the differences between treatment regimens blinding is not possible, but final assessment will be blinded to treatment. Treatment or intervention Treatment Arm A: Best medical care: Medical treatment will be performed as detailed in established Standard Operating Procedures, following regional guidelines (AHA, EROICAS, DSG, local country, etc.). The reason for iv tPA ineligibility will be documented on the eCRF. Treatment Arm B: Endovascular thrombectomy and best medical care: In the TENSION trial, CE-marked devices for thrombectomy will be used within their intended use according to their instruction for use. If a subject is randomized to thrombectomy and subsequently fails the angiographic screening or is not treated due to rapidly improving neurologic symptoms prior to procedure, the subject will remain in the ITT population.

s for thrombectomy will be used within their intended use according to their instruction for use. If a subject is randomized to thrombectomy and subsequently fails the angiographic screening or is not treated due to rapidly improving neurologic symptoms prior to procedure, the subject will remain in the ITT population. Clinical assessment Baseline disease characteristics include prestroke mRS, presenting symptom(s) and results of pretreatment imaging with a description of the occluded vessel. Neurological deficit will be assessed using the NIHSS by certified investigators at baseline, at 24–36 h, at seven days or at hospital discharge, and at 90 ± 14 days. At 90 ± 14 days and 12 months (±14 days), outcome assessment will also comprise the mRS, health-related quality of life (EQ-5D, PROMIS-10), and poststroke depression (PHQ-4). Imaging protocol Baseline imaging, either MRA with DWI or CTA should demonstrate a new focal occlusion (in the M1 segment and/or the intracranial ICA) accessible to the thrombectomy device. Baseline imaging should depict all supra-aortic vessels (head and neck). Angiographic imaging before, during, and after the endovascular procedure as well as follow-up imaging at 30 (−6/+6) hours to assess for intracranial hemorrhage will be sent to the Imaging Core Lab. All investigators should be qualified to assess images according to ASPECTS. ASPECTS training for TENSION will be performed using a web-based “reading academy” consisting of two modules: a training module and a rating module. Only physicians who pass the test are allowed to enroll patients.

ent to the Imaging Core Lab. All investigators should be qualified to assess images according to ASPECTS. ASPECTS training for TENSION will be performed using a web-based “reading academy” consisting of two modules: a training module and a rating module. Only physicians who pass the test are allowed to enroll patients. Primary outcomes The primary endpoint of TENSION is the patient's mRS at 90-day poststroke analyzed with a shift analysis. Secondary outcomes Secondary endpoints will comprise independent functional outcome (mRS ≤ 2), health-related quality of life (EQ-5D. PROMIS-10), survival, symptomatic intracranial hemorrhage at 30 h, new ischemic stroke, space-occupying infarction (malignant brain edema), AE, SAE, infarct volume at 30 h, infarct growth, rates of hemicraniectomy, treatment effect by device, and poststroke depression (PHQ-4). Cost-utility assessment will include health-related quality of life assessment at 90 (±14) days and 12 months (±14 days) and assessment of costs from the time of randomization to the 12-month follow-up, including costs of hospitalization, institutionalized living, outpatient care, informal care provided by relatives, and cost of lost productivity.

include health-related quality of life assessment at 90 (±14) days and 12 months (±14 days) and assessment of costs from the time of randomization to the 12-month follow-up, including costs of hospitalization, institutionalized living, outpatient care, informal care provided by relatives, and cost of lost productivity. Data and Safety Monitoring Board (DSMB) To ensure that appropriate ethical consideration is given to the welfare of the patients enrolled in the study, an independent Ethics Advisory Board as well as a DSMB was formed. The members of the DSMB are not participants of the TENSION consortium and not involved in the clinical trial in any other way. The DSMB will meet every 12 months during the study to review trial group data in partially unblinded fashion on the baseline and safety parameters. In addition, the trial will undergo two interim analyses with possible premature stopping for futility or early success with control of the overall type one error rate using a Lan–Demets alpha-spending function. Interim data analysis is planned after the primary endpoint has been obtained for one-third and two-thirds of the patients. At each of these sample sizes, the available 90-day mRS data for each treatment arm will be evaluated. The tasks and operating procedures of the DSMB will be described in detail in a separate DSMB charter.

ata analysis is planned after the primary endpoint has been obtained for one-third and two-thirds of the patients. At each of these sample sizes, the available 90-day mRS data for each treatment arm will be evaluated. The tasks and operating procedures of the DSMB will be described in detail in a separate DSMB charter. Sample size Simulations were carried out using an assumption for the possible true distribution of mRS from the literature (Goyal et al.,5 Supplement: distribution of mRS in patients with an ASPECT score of 7 or less) and a proportional odds alternative with an odds ratio of 1.5 to be assessed using the primary endpoint mRS shift analysis. Under the assumption of these distributions, a total of 620 patients is required to achieve a power of 80% for a one-sided test at the 0.025 level. Assuming a 7% dropout rate of patients to assess the primary endpoint obtained three months after inclusion an effective sample size of 665 is necessary to obtain 620 complete observations. Also, as a consequence of the sequential monitoring of the trial, the total sample size needs to be increased according to the characteristics of the alpha-spending function that will be chosen. With a power function of parameter ρ = 2, up to a maximum of 714 patients may be required if the trial does not stop early.

Sample size Simulations were carried out using an assumption for the possible true distribution of mRS from the literature (Goyal et al.,5 Supplement: distribution of mRS in patients with an ASPECT score of 7 or less) and a proportional odds alternative with an odds ratio of 1.5 to be assessed using the primary endpoint mRS shift analysis. Under the assumption of these distributions, a total of 620 patients is required to achieve a power of 80% for a one-sided test at the 0.025 level. Assuming a 7% dropout rate of patients to assess the primary endpoint obtained three months after inclusion an effective sample size of 665 is necessary to obtain 620 complete observations. Also, as a consequence of the sequential monitoring of the trial, the total sample size needs to be increased according to the characteristics of the alpha-spending function that will be chosen. With a power function of parameter ρ = 2, up to a maximum of 714 patients may be required if the trial does not stop early. Statistical analyses Summary tables for subject demographics and baseline characteristics will be provided and comparisons will be made between study arms for the ITT and PP analysis sets. Procedural characteristics unique to usage of devices for thrombectomy will be described for subjects in the best medical care plus thrombectomy treatment group (within the ITT analysis set). The primary and secondary effectiveness endpoints will be summarized and compared between study groups for the ITT and PP analysis sets. Safety endpoints will be summarized and compared between study groups for the safety analysis set.

e best medical care plus thrombectomy treatment group (within the ITT analysis set). The primary and secondary effectiveness endpoints will be summarized and compared between study groups for the ITT and PP analysis sets. Safety endpoints will be summarized and compared between study groups for the safety analysis set. In general, summaries will be presented by treatment group pooled across occlusion location and investigators/sites, and by occlusion location within treatment group in the relevant analysis populations. Descriptive statistics for dichotomous/categorical variables will include number and percent of subjects in each category (including missing), by treatment group. Descriptive statistics for continuous variables will include number of nonmissing and missing subjects, minimum, lower quartile, median, upper quartile, maximum, mean, and standard deviation, stratified by treatment group. Regarding comparisons between treatment groups, Chi-squared test (Fisher's exact test where appropriate) will be utilized for the comparison of categorical variables and the t-test (or the Mann–Whitney U test when appropriate) will be utilized for the comparison of continuous variables. Data for the primary endpoint will be presented by treatment group (pooled over all sites) and by treatment group within study site. The primary effectiveness endpoint analysis will be carried out in an ordinal logistic regression model with the mRS ordinal scale as response variable and treatment group, stroke severity (NIHSS 8–18 and NIHSS >18), and time from symptom onset until randomization (0–6 h and 6–11 h) as explanatory variables. Aside from protocol-specified hypothesis tests, confidence intervals will be presented to facilitate clinical judgment of the secondary safety and effectiveness endpoints, and not to test hypotheses. Confidence intervals for dichotomous or ordinal endpoints will be reported on the odds ratio scale. Figure 1. Study flow chart. AE: adverse event; ASPECTS: Alberta Stroke Program Early CT Score; CT: computed tomography; CTA: computed tomography angiography; CTP: computed tomography perfusion; DWI: diffusion-weighted imaging; EuroQol 5D; MRA: magnetic resonance angiography; MRI: magnetic resonance imaging; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; PHQ-4: Patient Health Questionnaire-4; PROMIS-10: Patient-Reported Outcomes Measurement Information System-10; SICH: symptomatic intracranial hemorrhage.

ging; EuroQol 5D; MRA: magnetic resonance angiography; MRI: magnetic resonance imaging; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; PHQ-4: Patient Health Questionnaire-4; PROMIS-10: Patient-Reported Outcomes Measurement Information System-10; SICH: symptomatic intracranial hemorrhage. Figure 2. The concept of treatment response in stroke patients with extended stroke lesions. ASPECTS: Alberta Stroke Program Early CT Score.

ging; EuroQol 5D; MRA: magnetic resonance angiography; MRI: magnetic resonance imaging; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; PHQ-4: Patient Health Questionnaire-4; PROMIS-10: Patient-Reported Outcomes Measurement Information System-10; SICH: symptomatic intracranial hemorrhage. Figure 2. The concept of treatment response in stroke patients with extended stroke lesions. ASPECTS: Alberta Stroke Program Early CT Score. Study organization and funding TENSION is an EU-funded, investigator-initiated and conducted trial. Coordination and project management will be provided by Prof. Götz Thomalla (Department of Neurology, University Hospital Hamburg Eppendorf, Germany). The principal investigator Prof. Martin Bendszus (Department of Neuroradiology, University Hospital Heidelberg, Germany) will organize the trial together with input from International Consortium for Health Outcomes Measurement (ICHOM), SAFE (European patients organization), and the national principal investigators: E. Gizweski (AUT), A. Krajina (CZE), C. Simonsen (DEN), L. Pierot (FRA), E. Berge (NOR), K. Zeleňák (SVK), P. Brouwer (SWE). Central trial management at the Koordinierungszentrum Klinische Studien (Coordination Center for Clinical Trials) at the University Hospital Heidelberg and European Clinical Research Infrastructure Network (ECRIN-ERIC) will perform the submission to ethics committees and competent authorities, trial management, data management, monitoring, and pharmacovigilance. Prof. Jens Fiehler (Eppdata GmbH and Department of Neuroradiology, University Hospital Hamburg Eppendorf, Germany) will be responsible for the Imaging Core Lab.

(ECRIN-ERIC) will perform the submission to ethics committees and competent authorities, trial management, data management, monitoring, and pharmacovigilance. Prof. Jens Fiehler (Eppdata GmbH and Department of Neuroradiology, University Hospital Hamburg Eppendorf, Germany) will be responsible for the Imaging Core Lab. TENSION is registered at ClinicalTrials.gov (ClinicalTrials.gov Identifier NCT03094715). Conclusion TENSION is a European PROBE, European two-arm, postmarket study to compare the safety and effectiveness of endovascular thrombectomy as compared to best medical care alone in the treatment of AIS patients with extended stroke lesions defined by an ASPECT score of 3–5 and in an extended time window (up to 12 h or unknown time of symptom onset). Up to 714 subjects will be randomized. Primary endpoint will be functional outcome assessed by the mRS at 90-day poststroke (“mRS shift analysis”). By this, TENSION will provide evidence of efficacy and safety of thrombectomy in an acute stroke population with uncertain benefit of endovascular stroke treatment so far and may greatly increase the proportion of patients eligible for treatment. Acknowledgments TENSION boards and institutions: Data and Safety Monitoring Board, Ethics Advisory Board, Innovation and Exploitation Management Board, Steering Committee, ECRIN, ICHOM, ESMINT. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Acknowledgments TENSION boards and institutions: Data and Safety Monitoring Board, Ethics Advisory Board, Innovation and Exploitation Management Board, Steering Committee, ECRIN, ICHOM, ESMINT. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: TENSION receives funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 754640.

Dear Editor, We read the review by Rabinstein et al. with interest. The authors discussed factors related to poor functional outcomes despite good reperfusion in acute ischemic stroke patients treated with endovascular thrombectomy (EVT).1 On the subject of anesthetic techniques during the intervention, the authors conclude that equipoise exists between conscious sedation (CS) and general anesthesia (GA) and large multicenter randomized trials are needed to determine whether or not CS and GA are equally safe and effective.

dovascular thrombectomy (EVT).1 On the subject of anesthetic techniques during the intervention, the authors conclude that equipoise exists between conscious sedation (CS) and general anesthesia (GA) and large multicenter randomized trials are needed to determine whether or not CS and GA are equally safe and effective. We think that focusing solely on CS and GA does not do justice to a simple and potentially safer anesthetic strategy: local anesthesia at the groin puncture site only (LA). The review mentioned the well-known trials (GOLIATH, SIESTA, ANSTROKE) that randomized between CS or GA during EVT and showed contrasting results.2–4 In the HERMES meta-analysis non-GA was superior to GA. However, the non-GA group was defined as the composite of local anesthesia (LA) at the groin puncture site only and CS.5 Therefore, the better functional outcomes in the non-GA arm might well be the result of patients receiving LA only. Recently, we compared the effect of LA only during EVT to CS and we reported better functional outcomes in patients receiving LA.6 Several mechanisms, present in both GA and CS (e.g. blood pressure drops, impaired airway reflexes), could explain poorer outcomes in the CS group. We think that these results should be taken into account when considering what is the optimal anesthetic approach during EVT. In our opinion, future trials should consider LA as one of the initial anesthetic strategies during EVT.

pressure drops, impaired airway reflexes), could explain poorer outcomes in the CS group. We think that these results should be taken into account when considering what is the optimal anesthetic approach during EVT. In our opinion, future trials should consider LA as one of the initial anesthetic strategies during EVT. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) received no financial support for the research, authorship, and/or publication of this article.

Introduction In about 20% of cases, stroke is preceded by an episode of temporary symptoms called a transient ischemic attack (TIA).1 Studies have shown that identifying and treating patients with TIA is an effective means of preventing stroke.2,3 Since the highest risk for stroke is in the first 48 h following symptom onset,4–6 it is critical that diagnosis and assessment occur rapidly. Unfortunately, diagnosing TIA can be difficult, as it depends on detailed history-taking; by definition, patients' symptoms have resolved at the time of assessment, and there is no established biomarker for TIA. As a consequence, approximately 30–50% of patients referred to stroke prevention clinics (SPCs) with a provisional diagnosis of TIA are ultimately found not to have had a TIA.7–9 This situation is problematic as high volumes of referrals of patients with TIA–mimics are directly related to delays in the care of TIA patients.3 While a major focus of recent research has been on risk-stratifying patients with TIA in order to decrease wait times for the highest risk patients, the many proposed risk scores10–13 suffer from an important weakness: they are derived from, and applied to, an undifferentiated population of patients with transient neurological symptoms including both patients with TIAs and mimics (e.g. migraine or seizure).8,14 In other words, the risk scores themselves do not differentiate patients with TIA from other clinical syndromes.

t weakness: they are derived from, and applied to, an undifferentiated population of patients with transient neurological symptoms including both patients with TIAs and mimics (e.g. migraine or seizure).8,14 In other words, the risk scores themselves do not differentiate patients with TIA from other clinical syndromes. Multiple small studies15–17 have looked at TIA diagnosis post hoc using expert panels for adjudication, though none has studied decision-making in vivo and none has sought to describe the diagnostic process, i.e. why a certain diagnosis is made. As such, we performed a systematic review to assess how and why neurologists call a particular clinical event a TIA or a mimic. We chose to study neurologists because they are considered stroke experts in most countries and because expertise is currently the “gold standard” for TIA diagnosis. Ultimately our goal is to make the SPC referral process more efficient by developing a method of selecting patients with TIA as accurately as possible from all those presenting to emergency departments and ambulatory clinics with transient neurological symptoms.18 Methods A systematic review was performed to address the question: “How do neurologists diagnose TIA?” We adhered to the PRISMA 2009 statement and conformed to its checklist (Supplementary Figure I).19

Multiple small studies15–17 have looked at TIA diagnosis post hoc using expert panels for adjudication, though none has studied decision-making in vivo and none has sought to describe the diagnostic process, i.e. why a certain diagnosis is made. As such, we performed a systematic review to assess how and why neurologists call a particular clinical event a TIA or a mimic. We chose to study neurologists because they are considered stroke experts in most countries and because expertise is currently the “gold standard” for TIA diagnosis. Ultimately our goal is to make the SPC referral process more efficient by developing a method of selecting patients with TIA as accurately as possible from all those presenting to emergency departments and ambulatory clinics with transient neurological symptoms.18 Methods A systematic review was performed to address the question: “How do neurologists diagnose TIA?” We adhered to the PRISMA 2009 statement and conformed to its checklist (Supplementary Figure I).19 Search strategy and selection criteria Keywords were selected and submitted to a librarian who created an initial search strategy. The search strategy was then revised to ensure that key studies were not omitted. Databases searched included MEDLINE, Embase, and the Cochrane Library. Supplementary Figure II details the search strategy that was used for MEDLINE. Similar strategies were utilized for the Embase and the Cochrane Library. The searches were conducted from inception of each database until 23 February 2017 with no language or date restrictions. The reference lists of manuscripts selected for inclusion were hand-searched for any additional potentially relevant citations that were not captured with the electronic search strategy alone.

rary. The searches were conducted from inception of each database until 23 February 2017 with no language or date restrictions. The reference lists of manuscripts selected for inclusion were hand-searched for any additional potentially relevant citations that were not captured with the electronic search strategy alone. Manuscripts were included if they explicitly addressed TIA diagnosis or if their inclusion criteria directly informed our study. To reduce the risk of publication bias, both peer-reviewed publications and unpublished studies (e.g. conference abstracts without a subsequent publication) were included. Both quantitative and qualitative studies were eligible for inclusion. All observational studies, including cohort, case–control, and cross-sectional designs, and all interventional studies with primary or secondary outcomes aimed at answering our research question were included. Studies that did not directly focus on answering our question but indirectly revealed neurologists' diagnostic decision-making by way of key statements or study inclusion criteria were also included. To ensure this systematic review was comprehensive and reflective of expert practice, textbooks and reviews, including nonsystematic approaches such as opinion pieces, commentaries, and literature reviews, were eligible for inclusion if written by a neurologist. Because the diagnosis of TIA depends upon clinical judgment, we included manuscripts containing statements of expert opinion and experience-based reasoning, both of which are often best reflected in nonsystematic reviews.

es, commentaries, and literature reviews, were eligible for inclusion if written by a neurologist. Because the diagnosis of TIA depends upon clinical judgment, we included manuscripts containing statements of expert opinion and experience-based reasoning, both of which are often best reflected in nonsystematic reviews. Manuscripts reflective of a nonneurologist only were excluded, as our goal was to focus on neurologist diagnostic decision-making. These exclusions were classified as “wrong setting.” Unpublished studies were eligible for inclusion but were excluded if there was not enough information in the abstract alone to answer our research question and if a full publication did not follow. Although the initial search strategy, as well as the title/abstract screening stage did not have language restrictions, during full-text screening, studies were excluded if the full text was in a language other than English as translation services were unavailable. Language restrictions were not imposed earlier to allow us to keep track of the number of articles deemed ineligible simply due to language and therefore to allow us to assess the magnitude of any potential language bias. Studies of pediatric patients (under the age of 18) were similarly excluded during the full-text screening stage as the objective of our systematic review was to identify clinical features of TIA, and patients in this population may experience different symptoms, may be unable to recognize transient neurological deficits, or may be unable to express their symptoms.

18) were similarly excluded during the full-text screening stage as the objective of our systematic review was to identify clinical features of TIA, and patients in this population may experience different symptoms, may be unable to recognize transient neurological deficits, or may be unable to express their symptoms. Screening Search results were compiled using Covidence systematic review software.20 Duplicate references were automatically detected and removed. Two reviewers independently screened each citation based on title and abstract. Pilot screening was performed for the first 25 records to ensure reviewers were consistent and to decrease conflicts. All disagreements were resolved through consensus discussion between the two reviewers with input from a third reviewer. Two reviewers then independently reviewed the full text of each article included from title and abstract screening, assessing eligibility for inclusion. All disagreements were again resolved through consensus discussion with input from a third reviewer. Data extraction and synthesis An abstraction datasheet was created using Microsoft Excel (2010) and two reviewers independently extracted study-level characteristics (e.g. study design, country of conduct) and the relevant data from each publication. The focus was on the identification of factors associated with diagnosis of TIA or TIA–mimic, and the process utilized by neurologists in their diagnostic decision-making. Where available, the rates of diagnosis of TIA and TIA–mimic were also collected.

, country of conduct) and the relevant data from each publication. The focus was on the identification of factors associated with diagnosis of TIA or TIA–mimic, and the process utilized by neurologists in their diagnostic decision-making. Where available, the rates of diagnosis of TIA and TIA–mimic were also collected. The results from the two independent extractions were then compared to ensure accuracy and completeness. Any discrepancies were resolved through discussion between the two reviewers. The data were then compiled using QSR's International NVivo 11 qualitative data analysis software.21 We performed a multistep “thematic synthesis,” which began with coding of text in NVivo in a “line-by-line” fashion, ensuring all relevant information was captured by the two reviewers. From the codes that were created, common themes emerged and concepts were grouped into the following categories: (1) symptoms suggestive of TIA, (2) qualitative features suggestive of TIA, (3) symptoms suggestive of TIA–mimic, (4) qualitative features suggestive of TIA–mimic, (5) risk factors and demographic features more common in TIA, and (6) risk factors and demographic features more common in TIA–mimic. Finally, analytical themes were generated from these descriptive themes to answer our initial question—“How do neurologists diagnose TIA?” The only quantitative information collected, TIA–mimic rate, was analyzed with descriptive statistics.

We performed a multistep “thematic synthesis,” which began with coding of text in NVivo in a “line-by-line” fashion, ensuring all relevant information was captured by the two reviewers. From the codes that were created, common themes emerged and concepts were grouped into the following categories: (1) symptoms suggestive of TIA, (2) qualitative features suggestive of TIA, (3) symptoms suggestive of TIA–mimic, (4) qualitative features suggestive of TIA–mimic, (5) risk factors and demographic features more common in TIA, and (6) risk factors and demographic features more common in TIA–mimic. Finally, analytical themes were generated from these descriptive themes to answer our initial question—“How do neurologists diagnose TIA?” The only quantitative information collected, TIA–mimic rate, was analyzed with descriptive statistics. Critical appraisal For descriptive purposes, each article that was included was assessed for risk of bias and strength/quality of evidence. Records were critically appraised based on six criteria—clarity of statement of aims, appropriateness of methodology, reliability/validity of data collection tools, reliability/validity of methods of data analysis, clarity of statement of results, and overall relevance to our question. These criteria were adapted from various quality assessment tools available for qualitative research and selected after discussion between three reviewers.22,23 Two reviewers performed the ratings independently and evaluations were compared. Disagreements were settled through consensus-based discussion.

bstract only, without enough information provided; (3) language other than English; (4) wrong population; (5) wrong setting; and (6) previous publication with duplicate material. A reference list of the excluded studies with reasons for exclusion is presented in the Supplementary Methods. Figure 1. PRISMA flow diagram. Study characteristics Study characteristics are presented in Supplementary Table 1. A total of 21 different countries were represented, with the United States (31 publications), and the United Kingdom (20 publications) being most common. Seven unpublished cohort studies and one unpublished case–control study were included (conference abstracts without subsequent manuscript publications). Publications included were mostly cohort design (49%), followed by literature review (21%) (Table 1). Table 1. Comparison of included records by study design Study design Number (%) Cohort study 39 (49%) Literature review 17 (21%) Opinion 8 (10%) Case report 5 (6%) Case series 4 (5%) Cross-sectional study 3 (4%) Case–control 1 (1%) Systematic review 1 (1%) Survey 1 (1%) Textbook 1 (1%)

Study characteristics Study characteristics are presented in Supplementary Table 1. A total of 21 different countries were represented, with the United States (31 publications), and the United Kingdom (20 publications) being most common. Seven unpublished cohort studies and one unpublished case–control study were included (conference abstracts without subsequent manuscript publications). Publications included were mostly cohort design (49%), followed by literature review (21%) (Table 1). Table 1. Comparison of included records by study design Study design Number (%) Cohort study 39 (49%) Literature review 17 (21%) Opinion 8 (10%) Case report 5 (6%) Case series 4 (5%) Cross-sectional study 3 (4%) Case–control 1 (1%) Systematic review 1 (1%) Survey 1 (1%) Textbook 1 (1%) Critical appraisal Critical appraisal assessments are summarized in Table 2. Critical appraisal of each individual study is presented in Supplementary Table 2. Statement of aims and statement of results were clear in the vast majority of publications (94 and 90%, respectively). Data collection was performed appropriately in 88% of publications. Appraisal questions related to methodology and data analysis were not applicable to certain study designs including opinion pieces and literature reviews. Where applicable, however, most studies performed well on this quality measure. Many of the studies (60%) were ultimately found to be of low relevance, but were included in our analysis because they did contribute content to the data collected. Table 2. Summary of the critical appraisal for included studies

re reviews. Where applicable, however, most studies performed well on this quality measure. Many of the studies (60%) were ultimately found to be of low relevance, but were included in our analysis because they did contribute content to the data collected. Table 2. Summary of the critical appraisal for included studies Appraisal question Yes No Unclear N/A Is there a clear statement of aims or a clearly defined question? 75 (94%) 5 (6%) 0 (0%) 0 (0%) Was the methodology employed appropriate to the research question? 47 (59%) 0 (0%) 4 (5%) 29 (36%) Was the data collection performed appropriately? 70 (88%) 9 (11%) 1 (1%) 0 (0%) Was the data analysis rigorous? 34 (42%) 5 (6%) 6 (8%) 35 (44%) Was there a clear statement of results? 72 (90%) 8 (10%) 0 (0%) 0 (0%) Was the overall relevance to our research question high? 32 (40%) 48 (60%) 0 (0%) 0 (0%) N/A: not applicable. Descriptive themes The main themes that emerged are presented in the following subsections: 1) Symptoms suggestive of TIA The most common clinical features that neurologists noted were suggestive of TIA rather than a non-TIA diagnosis are presented in Figure 2 including the frequency of each reference. Figure 2. Commonly identified clinical features suggestive of TIA. Symptoms are depicted in light gray bubbles and qualitative features are depicted in dark gray bubbles. N = number of publications making reference to each differentiating feature. The percentage of included studies is shown in parentheses. TIA: transient ischemic attack.

identified clinical features suggestive of TIA. Symptoms are depicted in light gray bubbles and qualitative features are depicted in dark gray bubbles. N = number of publications making reference to each differentiating feature. The percentage of included studies is shown in parentheses. TIA: transient ischemic attack. Overall, “negative symptoms”—characterized by loss of function—were the most frequently described (69% of included studies). This broad category included motor, sensory, and visual symptoms. Terms utilized to describe negative motor symptoms included “hemiparesis,” “unilateral arm/face/leg weakness,” “loss of motor function,” and “loss of muscle power.” Negative sensory symptoms were described with the terms “sensory loss” and “numbness.” Both monocular and binocular negative visual symptoms were encompassed by the negative visual symptoms category. Terms grouped included “homonymous hemianopsia,” “visual field deficit,” “visual loss,” “cortical blindness,” “monocular blindness,” and “amaurosis fugax.” The second-most frequently described symptom neurologists noted to be in keeping with TIA was “speech disturbance” (55% of included studies). Although some authors used specific terms such as “aphasia” or “dysarthria,” others used more general terms such as “speech disturbance” or “impaired speech.” These nonspecific terms were difficult to interpret separately and therefore all references to speech and communication were grouped into one category.

studies). Although some authors used specific terms such as “aphasia” or “dysarthria,” others used more general terms such as “speech disturbance” or “impaired speech.” These nonspecific terms were difficult to interpret separately and therefore all references to speech and communication were grouped into one category. Other symptoms commonly considered by neurologists to be supportive of a diagnosis of TIA included “ataxia,” “diplopia,” and “vertigo with other posterior circulation symptoms.” These were described in 14–18% of records. “Dizziness” was not a helpful differentiating feature: while some authors identified it as a symptom suggestive of TIA, a similar number of articles stated that it was a symptom more suggestive of TIA–mimic. When specified as “isolated vertigo” neurologists were also much more likely to diagnose a TIA–mimic.

f records. “Dizziness” was not a helpful differentiating feature: while some authors identified it as a symptom suggestive of TIA, a similar number of articles stated that it was a symptom more suggestive of TIA–mimic. When specified as “isolated vertigo” neurologists were also much more likely to diagnose a TIA–mimic. 2) Qualitative features suggestive of TIA In addition to the clinical symptoms described above, the expert diagnostic process also identified pattern of onset, localizability, and duration as important elements considered in diagnostic decision-making. In almost one-third of the articles (n = 25), neurologists were more likely to diagnose TIA if the onset of symptoms was “sudden,” “maximal at onset,” “nonprogressive,” or “acute.” Another characteristic identified in one quarter of included articles (n = 20) was “localizability” of the symptoms; terms grouped together included “focal,” “corresponding to a vascular territory,” and “consistent with a known stroke syndrome.” The last characteristic that was commonly identified as a TIA feature was symptoms with a “duration less than 1 h.”

quarter of included articles (n = 20) was “localizability” of the symptoms; terms grouped together included “focal,” “corresponding to a vascular territory,” and “consistent with a known stroke syndrome.” The last characteristic that was commonly identified as a TIA feature was symptoms with a “duration less than 1 h.” 3) Symptoms suggestive of TIA–mimic Figure 3 displays the features of transient neurological disturbance that neurologists consider to be indicative of a TIA–mimic diagnosis. The most commonly identified symptoms were those that fell under the category of “positive symptoms,” including motor, sensory, or visual phenomena (48% of records). Positive motor symptoms were described with the following terms: “jerking,” “shaking,” “seizure-like activity,” and “involuntary movement.” Terms used to describe positive sensory symptoms included “tingling” and “paresthesias.” “Scintillating scotoma,” “flashing lights,” and “visual aura” were grouped under the positive visual phenomena subsection. Figure 3. Commonly identified clinical features suggestive of TIA–mimic. Symptoms are depicted in light gray bubbles and qualitative features are depicted in dark gray bubbles. N = number of publications making reference to each differentiating feature. The percentage of included studies is shown in parentheses. TIA: transient ischemic attack.

fied clinical features suggestive of TIA–mimic. Symptoms are depicted in light gray bubbles and qualitative features are depicted in dark gray bubbles. N = number of publications making reference to each differentiating feature. The percentage of included studies is shown in parentheses. TIA: transient ischemic attack. The next category of TIA–mimic symptoms identified by neurologists was “altered level of consciousness (LOC)” (46% of included studies). Within this category we grouped any disturbance in consciousness including “presyncope,” “loss of consciousness,” “decreased level of consciousness,” and “impaired consciousness.” “Confusion” was separated from “altered level of consciousness” and was grouped with “cognitive symptoms” and “amnesia.” The presence of any of these symptoms was frequently considered to be supportive of a TIA–mimic diagnosis (31% of records). Other recurrent TIA–mimic themes were “headache,” “bowel or bladder symptoms,” and “generalized weakness.” 4) Qualitative features suggestive of TIA–mimic Features of symptoms suggestive of TIA–mimic included the inverse of those seen with TIA, including “nonfocal” and “nonlateralizing.” “Slow onset” of symptoms typically swayed neurologists toward a non-TIA diagnosis, as did, “slow progression,” “slow spread,” or “march” of symptoms (n = 30). The presence of a “Jacksonian march” was specifically identified as a key mimic feature. Lastly, many authors considered TIA–mimic more likely if the patient had had “recurrent” or “stereotyped” episodes (n = 9).

sts toward a non-TIA diagnosis, as did, “slow progression,” “slow spread,” or “march” of symptoms (n = 30). The presence of a “Jacksonian march” was specifically identified as a key mimic feature. Lastly, many authors considered TIA–mimic more likely if the patient had had “recurrent” or “stereotyped” episodes (n = 9). 5) Risk factors and demographic features more common in TIA Risk factors and demographic features associated with the diagnosis of TIA included advanced age, atrial fibrillation, preexisting hypertension, previous stroke/TIA, or other vascular risk factors including dyslipidemia and type II diabetes (Figure 4). These features were mentioned relatively infrequently compared to the clinical characteristics described above. Figure 4. Risk factors and demographic features commonly identified as predictors of TIA diagnosis and TIA–mimic diagnosis. N = number of publications making reference to each risk factor. The percentage of included studies is shown in parentheses. TIA: transient ischemic attack. 6) Risk factors and demographic features more common in TIA–mimic The only demographic feature that was consistently associated with a diagnosis of TIA–mimic diagnosis was younger age (Figure 4). 7) TIA–mimic rate Twenty-seven (34%) of the included articles provided a TIA–mimic rate. The mimic rates ranged between 6 and 73%,24,25 with a median mimic rate of 36% (25th, 75th percentiles range: 26%, 50%).

6) Risk factors and demographic features more common in TIA–mimic The only demographic feature that was consistently associated with a diagnosis of TIA–mimic diagnosis was younger age (Figure 4). 7) TIA–mimic rate Twenty-seven (34%) of the included articles provided a TIA–mimic rate. The mimic rates ranged between 6 and 73%,24,25 with a median mimic rate of 36% (25th, 75th percentiles range: 26%, 50%). Discussion Diseases are often defined in relation to blood tests, imaging findings, or some combination thereof. In the absence of such markers, diseases are diagnosed through a process of decision-making by experts, and this is the state of TIA in contemporary medicine. Therefore, we sought to perform a systematic review of all relevant qualitative, quantitative, and mixed methods studies that would inform our understanding of the process by which neurologists diagnose TIA. While in some regions, nonneurologists (i.e. geriatricians) may provide stroke care, we chose to limit the scope of our search to neurologists for the sake of consistency and as a general reflection of practice in most regions. To the best of our knowledge, this is the first qualitative systematic review to assess how neurologists diagnose TIA.

neurologists (i.e. geriatricians) may provide stroke care, we chose to limit the scope of our search to neurologists for the sake of consistency and as a general reflection of practice in most regions. To the best of our knowledge, this is the first qualitative systematic review to assess how neurologists diagnose TIA. Our study has revealed that according to neurologists, the most consistent predictors for a diagnosis of TIA include negative symptoms (loss of motor, sensory, or visual function) and speech disturbance. The strongest predictors for TIA–mimic are positive symptoms (such as motor jerking, sensory tingling, or visual scotomas) and any alteration of consciousness. Certain characteristics including pattern of onset, localizability of symptoms, and symptom recurrence were also important discriminative diagnostic features. While these findings may appear obvious to those who are experts, that speaks to the accuracy of our results at capturing their decision-making process. Moreover, these findings are not obvious to nonexperts, suggesting the importance of work like this. We recognize that this study is a preliminary step to further characterizing the decision-making process surrounding TIA.

e experts, that speaks to the accuracy of our results at capturing their decision-making process. Moreover, these findings are not obvious to nonexperts, suggesting the importance of work like this. We recognize that this study is a preliminary step to further characterizing the decision-making process surrounding TIA. Diffusion-weighted (DWI) MRI is more sensitive than CT for detecting acute ischemia, and up to one-third of patients diagnosed with TIA are found to have an infarct on DWI MRI.26 Consequently, many organizations have moved away from the traditional “time-based” definition of TIA toward a new “tissue-based” definition.27 While MRI can be a very useful tool and certainly reduces the rate of false-negative diagnoses, it still cannot replace expert assessment, especially for those patients who are MRI-negative. Furthermore, MRI is not available in all healthcare settings. For these reasons, we chose to focus our study entirely on the clinical diagnosis of TIA.

y useful tool and certainly reduces the rate of false-negative diagnoses, it still cannot replace expert assessment, especially for those patients who are MRI-negative. Furthermore, MRI is not available in all healthcare settings. For these reasons, we chose to focus our study entirely on the clinical diagnosis of TIA. In the absence of a reliable tool for the diagnosis of TIA, frontline clinicians frequently apply risk-stratification instruments such as the ABCD2 score for diagnostic purposes.28 The ABCD and later ABCD2 scores were developed from populations of patients with a provisional diagnosis of TIA, many of whom where later given a final diagnosis of TIA–mimic by experts.2 When applied in a blanket fashion to any patient with transient neurological symptoms, these instruments can result in a large number of inappropriate urgent referrals to the SPC since TIA–mimics can very easily generate high ABCD2 scores.29 We believe that a more standardized decisional process should be established for TIA so that the inappropriate use of risk-stratification tools can be avoided.

toms, these instruments can result in a large number of inappropriate urgent referrals to the SPC since TIA–mimics can very easily generate high ABCD2 scores.29 We believe that a more standardized decisional process should be established for TIA so that the inappropriate use of risk-stratification tools can be avoided. To address this deficiency, two diagnostic algorithms have previously been developed for TIA—the Dawson Score and the Diagnosis of TIA Score (DOTS).30,31 The Dawson Score is a clinical scoring tool developed in a specialist setting that considers nine predictive variables and was found to be of limited utility in a primary care setting.32 It has been criticized for struggling with posterior circulation32 and retinal31 events. In contrast, the DOTS considers 17 variables, many of which reflected the factors we identified in our systematic review. It had a sensitivity of 89% (CI: 84–93%) and a specificity of 76% (70–81%)31 in an internal cohort, but has not yet been externally validated. Ultimately, these scores are seeking to approximate a diagnostic process that, until now, had not yet been described empirically.

factors we identified in our systematic review. It had a sensitivity of 89% (CI: 84–93%) and a specificity of 76% (70–81%)31 in an internal cohort, but has not yet been externally validated. Ultimately, these scores are seeking to approximate a diagnostic process that, until now, had not yet been described empirically. Most of the variables in the DOTS were identified by our systematic review; however, our systematic review has also identified several novel concepts, which are not reflected in any previously developed TIA diagnostic algorithms, including the pattern of onset/spread of symptoms and recurrence/stereotyped nature of episodes. We recognize that in the right clinical context, recurrent or stereotyped symptoms do not exclude vascular etiology altogether (e.g. capsular warning syndrome). This highlights the importance of considering the whole clinical picture rather than making decisions based on isolated features. Another important lesson from our study is that neurologists clearly rely on focal/lateralizable symptoms for the diagnosis of TIA. While we acknowledge that some populations, especially elderly women, may present with “nontraditional” stroke symptoms,33 evidence is conflicting and more research is needed on this subject.

ther important lesson from our study is that neurologists clearly rely on focal/lateralizable symptoms for the diagnosis of TIA. While we acknowledge that some populations, especially elderly women, may present with “nontraditional” stroke symptoms,33 evidence is conflicting and more research is needed on this subject. We intend to use the results of our systematic review to inform further in vivo studies on the expert diagnosis of TIA. Our goal is to identify reliable factors that will help frontline clinicians make a provisional diagnosis of TIA with more accuracy. Rather than creating a new TIA score, we hope to focus our efforts on education around the key elements used in the process of TIA diagnosis. The dissemination of knowledge to primary care and emergency room physicians could have a substantial impact on patient care, as it would decrease the number of patients falsely labeled with a TIA event. As such, the quality and volumes of referrals to SPCs could be improved, contributing to enhanced efficiency of stroke prevention interventions. High rates of TIA–mimics referred for assessment contributes to delays in care through bottlenecking. If we are able to improve wait times, particularly for high-risk TIA patients, this could ultimately reduce stroke rates. The implications on health services are also significant, as better referrals would lead to marked cost savings by decreasing the number of unnecessary tests ordered for referred patients, and ultimately reducing the costs associated with preventable strokes.

isk TIA patients, this could ultimately reduce stroke rates. The implications on health services are also significant, as better referrals would lead to marked cost savings by decreasing the number of unnecessary tests ordered for referred patients, and ultimately reducing the costs associated with preventable strokes. This systematic review is not without limitations. Given the nature of TIA diagnosis, a variety of qualitative research studies have informed our analysis. Furthermore, we chose to include literature reviews and opinion pieces since expert opinion is often best reflected using these approaches. Since there is no confirmatory test for TIA, we are relying on the assumption that neurologist opinion is the “gold standard.” This naturally introduces potential bias as there will always be an element of subjectivity when it comes to making a diagnosis based on a patient's history alone. Unfortunately, we do not see any way to avoid this since at the present time there are no blood biomarkers or imaging tests available to reliably distinguish TIAs from TIA–mimics. Finally, another limitation of our study is that our literature search was performed as of February 2017.

ased on a patient's history alone. Unfortunately, we do not see any way to avoid this since at the present time there are no blood biomarkers or imaging tests available to reliably distinguish TIAs from TIA–mimics. Finally, another limitation of our study is that our literature search was performed as of February 2017. In conclusion, our systematic review has identified the key clinical characteristics that neurologists consider when differentiating between TIAs and TIA–mimics. We intend to explore this distinction further by studying real-world decision-making for patients referred to our SPC. Educating frontline clinicians on the features identified could have a significant impact on patient care and our healthcare system. Supplemental Material Supplemental material for How do neurologists diagnose transient ischemic attack: A systematic review Click here for additional data file. Supplemental Material for How do neurologists diagnose transient ischemic attack: A systematic review by Tess Fitzpatrick, Sophia Gocan, Chu Q Wang, Candyce Hamel, Aline Bourgoin, Dar Dowlatshahi, Grant Stotts and Michel Shamy in International Journal of Stroke Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding provided by the Stroke Research Consortium, University of Ottawa Brain and Mind Research Institute.

Dear Editor, The optimal preventive screening strategy for intracranial aneurysms in first-degree relatives (FDRs) of patients with aneurysmal subarachnoid hemorrhage (aSAH) remains unclear. We evaluated the correlation of age at time of aSAH in FDRs to assess whether age at time of aSAH may be a factor to consider in determining the optimal screening strategy. We included 87 Dutch, 43 Finnish, and 16 French families with ≥2 FDRs with definite or probable aSAH (Table 1).1 We calculated intraclass correlation coefficients (ICCs) for age at time of aSAH and age differences at time of aSAH between FDRs. We performed subanalyses on (1) FDRs with definite aSAH, (2) siblings, and (3) Dutch and French families as different patient characteristics are reported for the Finnish.2 Table 1. Baseline characteristics. Characteristic All FDRs (n = 319) Dutch FDRs (n = 196) Finnish FDRs (n = 87) French FDRs (n = 36) Women, n (%) 192 (60) 130 (66) 44 (51) 19 (53) Definite aSAH, n (%) 278 (87) 155 (79) 87 (100) 36 (100) Mean age at time of aSAH, years (SD) 48.4 (12.7) 49.9 (12.7) 46.6 (12.6) 44.8 (11.7) FDR: first-degree relative; aSAH: aneurysmal subarachnoid hemorrhage; SD: standard deviation.

sh FDRs (n = 87) French FDRs (n = 36) Women, n (%) 192 (60) 130 (66) 44 (51) 19 (53) Definite aSAH, n (%) 278 (87) 155 (79) 87 (100) 36 (100) Mean age at time of aSAH, years (SD) 48.4 (12.7) 49.9 (12.7) 46.6 (12.6) 44.8 (11.7) FDR: first-degree relative; aSAH: aneurysmal subarachnoid hemorrhage; SD: standard deviation. The ICC for age at time of aSAH in all 146 families was 0.21 (p < 0.01). The correlation remained essentially the same in the subanalyses. An age difference at time of aSAH of 20 years or less was observed in 84% of all FDRs (Figure 1). This age difference remained comparable in the subanalyses. Figure 1. Cumulative percentage of families per difference of age at time of aSAH. Results for all FDRs (a) and for subanalyses (b). FDR: first-degree relative; aSAH: aneurysmal subarachnoid hemorrhage. In conclusion, our study showed a poor correlation of age at time of aSAH in FDRs. Therefore, we did not find evidence that age at time of aSAH is a contributing factor in determining the optimal screening strategy for intracranial aneurysm. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) received no financial support for the research, authorship, and/or publication of this article.

Introduction Stroke is a major cause of morbidity and disability throughout the world.1 Across Saudi Arabia, stroke prevalence rates are estimated at 0.67%.2 Two decades ago, the rate of first stroke incidence was reported at 29.8 per 100,000 person-years, which when updated to the current demographic distribution leads to an estimated rate of 50.9/100,000.3 Similar rates were reported in a recent study in the Aseer region (57.64/100,000).4 The combination of an aging population and increased risk of stroke with age is expected to lead to a growing stroke burden from 16,900 first strokes up to 28,400 in the next 10 years.3 Treatment guidelines recommend reperfusion therapy for ischemic stroke (IS). Intravenous tissue plasminogen activator (IV-tPA) is a treatment with moderate benefit that can be administered up to 4.5 h after symptom onset. Mechanical thrombectomy (MT) is a treatment with substantial benefit that can be administered up to 6–8 h after symptom onset, and up to 24 h in imaging-selected patients.5 A challenge for stroke care systems is to achieve optimal patient flow, quickly and correctly identifying patients eligible for reperfusion treatments and transferring them to the appropriate center.6

tantial benefit that can be administered up to 6–8 h after symptom onset, and up to 24 h in imaging-selected patients.5 A challenge for stroke care systems is to achieve optimal patient flow, quickly and correctly identifying patients eligible for reperfusion treatments and transferring them to the appropriate center.6 Stroke care in Saudi Arabia lags behind other developed countries.7 Approximately 95% of patients are cared for at non-specialized stroke hospitals and receive non-reperfusion supportive care for both large vessel occlusion (LVO) and non-large vessel occlusion (non-LVO). Around 5% of patients are admitted to stroke units, among whom approximately 3.6% receive IV-tPA; fewer than 100 patients receive MT annually. This is broadly due to a lack of organization of care and inconsistent allocation of resources.7 Furthermore, a lack of care standards and incentives for patient management strategies leads to sub-optimal patient outcomes. An urgent need for stroke care development was identified by the Saudi Arabian Ministry of Health's stroke committee, which coordinated efforts to analyze the implications of improving care, to support the development of a plan.

Stroke care in Saudi Arabia lags behind other developed countries.7 Approximately 95% of patients are cared for at non-specialized stroke hospitals and receive non-reperfusion supportive care for both large vessel occlusion (LVO) and non-large vessel occlusion (non-LVO). Around 5% of patients are admitted to stroke units, among whom approximately 3.6% receive IV-tPA; fewer than 100 patients receive MT annually. This is broadly due to a lack of organization of care and inconsistent allocation of resources.7 Furthermore, a lack of care standards and incentives for patient management strategies leads to sub-optimal patient outcomes. An urgent need for stroke care development was identified by the Saudi Arabian Ministry of Health's stroke committee, which coordinated efforts to analyze the implications of improving care, to support the development of a plan. Aim The study aimed to undertake a health-economic analysis, comparing current stroke care in Saudi Arabia with a proposed stroke care development program, and to assess the associated clinical and cost outcomes. This included modeling the establishment of specialized stroke units to increase utilization of IV-tPA and MT, and changing the current organization of care and patient flow to provide quicker access to treatment.

with a proposed stroke care development program, and to assess the associated clinical and cost outcomes. This included modeling the establishment of specialized stroke units to increase utilization of IV-tPA and MT, and changing the current organization of care and patient flow to provide quicker access to treatment. Method Extensive research was undertaken to determine current stroke care provision in Saudi Arabia, and international stroke services were examined to identify the optimal model for developing stroke care. The economic analysis was conducted from a societal perspective. The model was designed in collaboration with a panel of leading Saudi stroke consultants, which was consulted to confirm differences in patient pathways, treatment rates, length of stay and discharge destinations. Model structure The model structure was based on a published cost-effectiveness model comparing reperfusion treatments.8 The model uses a Markov structure split into an acute phase from stroke onset to 90 days, and a rest-of-life phase from 91 days until death. The modified Rankin Scale (mRS) was used to characterize seven health states,9 and all treatment effects were assumed to occur within the acute phase. In the rest-of-life phase, patients remained in the same mRS as at 90 days until either a recurrent stroke or death.8 This structure was modified to calculate mRS scores, costs and quality-adjusted life-years (QALYs) for each IS patient.

h states,9 and all treatment effects were assumed to occur within the acute phase. In the rest-of-life phase, patients remained in the same mRS as at 90 days until either a recurrent stroke or death.8 This structure was modified to calculate mRS scores, costs and quality-adjusted life-years (QALYs) for each IS patient. Ten cohorts of IS patients entered the model over 10 years and we estimated lifetime costs of these patients. The patient numbers in each new cohort were based on increasing incidence figures to account for the effect of an aging population. A decision tree structure (Figure 1) was used to separate each cohort into each program and stroke type (LVO or non-LVO). The population was also separated into Saudi and non-Saudi because one-third of the population is non-native. According to local specialists, many non-Saudis return to their home countries after a disabling stroke since their main purpose for being in Saudi Arabia was employment (all patients have equivalent access to care, so this aspect is unaffected). However, it can result in different long-term and societal costs, arising only from their non-permanence in the country, and not from the care that they would receive if they were to remain. Treatment options included MT+IV-tPA, IV-tPA alone, MT alone, and non-reperfusion treatment and were available depending on stroke type. A gradual uptake of reperfusion treatments was assumed in the development program, which was estimated based on international evidence.10–15 Figure 1. Stroke care programs overview.

. Treatment options included MT+IV-tPA, IV-tPA alone, MT alone, and non-reperfusion treatment and were available depending on stroke type. A gradual uptake of reperfusion treatments was assumed in the development program, which was estimated based on international evidence.10–15 Figure 1. Stroke care programs overview. Progressive investment costs and resource allocation were considered to model changes to the stroke care pathway for the development program. The base case considers a 15-year time horizon to allow some benefits to accrue for the final cohort entering the model in year 10. Future health outcomes and costs were discounted at 3% per year.16 All costs are presented in US dollars ($1 = 3.75 Saudi Riyals, January 2019). Patient information Total populations of 20,768,627 Saudi and 12,645,033 non-Saudi individuals were included and used age- and gender-specific stroke incidence figures to calculate the annual number of strokes over 10 years. Incidence figures ranged from 50.9/100,000 people to 75.4/100,000 of the projected population.3,17 Ischemic strokes and LVO strokes were assumed to make up 85% and 34% of all strokes, respectively.15 The mean age of stroke for the base-case scenario was 62.4 Current treatment utilization was estimated using local data validated by the consultant panel (Supplement Table I).

.4/100,000 of the projected population.3,17 Ischemic strokes and LVO strokes were assumed to make up 85% and 34% of all strokes, respectively.15 The mean age of stroke for the base-case scenario was 62.4 Current treatment utilization was estimated using local data validated by the consultant panel (Supplement Table I). Clinical effectiveness The impact of a learning curve is often overlooked in the economic evaluation of medical devices, and can affect short- and long-term patient outcomes and complication rates.18 MT is a highly specialized procedure, requiring operator skills developed over time. Given the current facility shortage, lack of organization of stroke care and patient access to IV-tPA and MT, two different data sources were used for short-term LVO patient outcomes for the current and development scenarios.

rates.18 MT is a highly specialized procedure, requiring operator skills developed over time. Given the current facility shortage, lack of organization of stroke care and patient access to IV-tPA and MT, two different data sources were used for short-term LVO patient outcomes for the current and development scenarios. For the current program, outcomes data were applied from MR CLEAN, a study undertaken in the Netherlands in centers with variable experience with MT and with less restrictive patient selection compared with other randomized controlled trials (RCTs).19 Patient outcomes were used as a baseline to reflect the current experience level in Saudi Arabia. For the development program, data from the HERMES collaboration meta-analysis were applied, representing studies involving centers with more homogeneous advanced experience using MT.20 Given the effort towards creating specialized stroke units in Saudi Arabia, these data may better represent future patient outcomes as experience and skill level increase. Clinical efficacy data for non-LVO strokes were taken from a subgroup analysis of a pooled analysis of nine RCTs comparing alteplase with placebo/open control (Supplement Table II).21 Adverse event rates and recurrent stroke probabilities are detailed in Supplement Table II.20,22,23 Mortality Age-specific other-cause mortality was applied in the rest-of-life phase to model deaths unrelated to stroke.24 Other-cause mortality was adjusted to reflect the higher mortality risk observed in stroke survivors.25

For the current program, outcomes data were applied from MR CLEAN, a study undertaken in the Netherlands in centers with variable experience with MT and with less restrictive patient selection compared with other randomized controlled trials (RCTs).19 Patient outcomes were used as a baseline to reflect the current experience level in Saudi Arabia. For the development program, data from the HERMES collaboration meta-analysis were applied, representing studies involving centers with more homogeneous advanced experience using MT.20 Given the effort towards creating specialized stroke units in Saudi Arabia, these data may better represent future patient outcomes as experience and skill level increase. Clinical efficacy data for non-LVO strokes were taken from a subgroup analysis of a pooled analysis of nine RCTs comparing alteplase with placebo/open control (Supplement Table II).21 Adverse event rates and recurrent stroke probabilities are detailed in Supplement Table II.20,22,23 Mortality Age-specific other-cause mortality was applied in the rest-of-life phase to model deaths unrelated to stroke.24 Other-cause mortality was adjusted to reflect the higher mortality risk observed in stroke survivors.25 Unit costs and resource use Treatment costs included devices, drugs, staff, diagnostics and triage. They were estimated by analyzing resource use from Ministry of Health (MOH) hospitals. (Supplement Table III).

Mortality Age-specific other-cause mortality was applied in the rest-of-life phase to model deaths unrelated to stroke.24 Other-cause mortality was adjusted to reflect the higher mortality risk observed in stroke survivors.25 Unit costs and resource use Treatment costs included devices, drugs, staff, diagnostics and triage. They were estimated by analyzing resource use from Ministry of Health (MOH) hospitals. (Supplement Table III). Acute costs were dependent on each program. Two types of hospitals were considered: non-specialized stroke hospital and stroke hospitals. In the current program, the majority of patients (95%) were admitted to non-specialized stroke hospitals. In the development program, a gradual increase of patients cared for at stroke hospitals was projected, up to a maximum of 70%. Additionally, stroke hospitals in the development program were assumed to have more efficient inpatient diagnostic and rehabilitation care, resulting in shorter length of stay (LOS) and earlier discharge.26 All acute and long-term costs were calculated based on LOS data for different wards and facility types. These data were based on local data from MOH hospitals and validated by the consultant panel. Cost per bed, tests, and hospital visits were based on the official price list from MOH 2018, adjusted for hospital specialization level. (Supplement Table III)

culated based on LOS data for different wards and facility types. These data were based on local data from MOH hospitals and validated by the consultant panel. Cost per bed, tests, and hospital visits were based on the official price list from MOH 2018, adjusted for hospital specialization level. (Supplement Table III) Three discharge destinations were included within acute costs: home with outpatient rehabilitation, inpatient rehabilitation, and long-term care facility (nursing home). Patients were routed to different discharge destinations and had differing LOS depending on their mRS score. The same discharge destinations were used for the post-acute phase costs (90 days to one year), but were dependent on the stroke care program, rather than the hospital type. After one year, patients with outpatient rehabilitation and inpatient rehabilitation were assumed return home and have neurologist follow-up appointments only. Longer term care was included as a yearly cost for the rest-of-life for patients with more disabling strokes. In the base-case, it was assumed that all non-Saudi patients with mRS≤2 would remain in Saudi Arabia and incur longer term care costs. However, 70% of mRS 3 patients, and 20% of mRS 4 to 5 patients would remain in the country. Recurrent stroke costs included treatment and acute care. Patients were assumed to receive the same treatment as for their previous stroke except those with an mRS ≥3, who were assumed to receive non-reperfusion management.5,27

In the base-case, it was assumed that all non-Saudi patients with mRS≤2 would remain in Saudi Arabia and incur longer term care costs. However, 70% of mRS 3 patients, and 20% of mRS 4 to 5 patients would remain in the country. Recurrent stroke costs included treatment and acute care. Patients were assumed to receive the same treatment as for their previous stroke except those with an mRS ≥3, who were assumed to receive non-reperfusion management.5,27 Productivity losses were included to reflect the burden on society using the human capital approach. The population was split by Saudi and non-Saudi, gender, and age to reflect differences in wage and labour participation (Supplement Tables IV to VII).17,28,29 Investment in an additional biplane angiosuite was included in the development program, for which the cost provided by the consultant panel reflected the official purchasing cost of this equipment by MOH hospitals (Supplement Table VIII). The full cost of each investment was only applied for the first 10 years; after this time, new patients stop entering the model, therefore it would benefit future patients whose outcomes are not captured within the model. After the 10-year period, the annual costs are weighted by the proportion of recurrent stroke patients, as they would continue to benefit from investments.

first 10 years; after this time, new patients stop entering the model, therefore it would benefit future patients whose outcomes are not captured within the model. After the 10-year period, the annual costs are weighted by the proportion of recurrent stroke patients, as they would continue to benefit from investments. Model outcomes The primary model outcomes are the incremental costs and QALYs with the development program. A deterministic sensitivity analysis was performed to explore uncertainty around input parameters. Probabilistic sensitivity analysis was also undertaken to assess overall model uncertainty, varying all model parameters simultaneously to generate 1,000 sampled sets of inputs and outputs.30 Internal validity was confirmed by a health economist not involved in the model development. Results With 192,474 IS patients modeled over 10 years, the development program is associated with overall cost savings of $602 million over a 15 year time horizon, with a higher estimate of around $629 million when lifetime costs are considered. An estimated $255 million is saved in direct costs to the healthcare system, with further savings of $348 million in indirect costs to society. The key reductions in cost result from decreased acute hospital and long-term nursing costs, which outweigh the cost of developing stroke care provision (Figure 2). A reduction is also predicted for post-90-day inpatient/outpatient rehabilitation and societal costs. Other categories show a cost increase (Supplement Table IX). Figure 2. Cost breakdown by stroke care program.

ospital and long-term nursing costs, which outweigh the cost of developing stroke care provision (Figure 2). A reduction is also predicted for post-90-day inpatient/outpatient rehabilitation and societal costs. Other categories show a cost increase (Supplement Table IX). Figure 2. Cost breakdown by stroke care program. The development program was associated with better health outcomes showing higher QALYs (0.29) and life years (LYs) (0.25) gained per patient at 15 years compared with the current program (Table 1). The effect was greater when only LVO strokes were considered, with a predicted incremental 0.79 QALYs and 0.71 LYs over a lifetime horizon. Table 1. Base case results Category Current stroke care program Stroke care development program Incremental Total costs $9,244,395,513 $8,641,826,427 −$602,569,086 Costs/patient $48,029 $44,899 $−3,131 QALYs/patient 2.08 2.36 0.29 Life-years/patient 3.50 3.75 0.25 ICER Dominant QALYs: quality-adjusted life-years. ICER: incremental cost-effectiveness ratio. Modified Rankin scores are estimated at 90 days for both programs and stroke types (Figure 3). Approximately 30% of LVO patients have an mRS ≤2 in the development program compared with 12% in the current program, equalling an increase of 18% in functionally independent patients and a decrease in severe disability. Figure 3. Number of patients at each mRS at 90 days by program.

ograms and stroke types (Figure 3). Approximately 30% of LVO patients have an mRS ≤2 in the development program compared with 12% in the current program, equalling an increase of 18% in functionally independent patients and a decrease in severe disability. Figure 3. Number of patients at each mRS at 90 days by program. Sensitivity analysis Deterministic sensitivity analysis indicates the development program remains cost-saving across changes in a large range of parameters. From 1,000 probabilistic simulations, the development program was associated with greater QALYs in all estimates and is predicted to have cost savings in 77.9% of cases (Figure 4). Figure 4. Probabilistic sensitivity analysis (a) Percentage of iterations for incremental cost. (b) Percentage of iterations for proportion of patients with mRS≤2 for each program.

development program was associated with greater QALYs in all estimates and is predicted to have cost savings in 77.9% of cases (Figure 4). Figure 4. Probabilistic sensitivity analysis (a) Percentage of iterations for incremental cost. (b) Percentage of iterations for proportion of patients with mRS≤2 for each program. Discussion This study goes beyond traditional cost-effectiveness and budget impact analysis to provide a framework for evaluating costs and outcomes in a region where the development of integrated and organized stroke care programs is critical to meet the need for quality care. The flexibility and novelty of the model allow different scenarios to be explored according to the gradual expected changes in treatment patterns for all IS patients, investments required, and the combination of features not previously observed. The development program is predicted to be cost-saving with estimated savings of US$602 million and a QALY gain of 0.29 (per patient) over 15 years compared to the current program. The key drivers as identified in sensitivity analyses are the mean age of stroke, total acute hospital costs, and total long-term nursing home costs. Varying these parameters individually did not change the direction of the results. Probabilistic sensitivity analysis also indicated that the model results are robust to changes in input parameters.

d in sensitivity analyses are the mean age of stroke, total acute hospital costs, and total long-term nursing home costs. Varying these parameters individually did not change the direction of the results. Probabilistic sensitivity analysis also indicated that the model results are robust to changes in input parameters. As with all economic models, various assumptions were made. As local published clinical data were lacking, the data used in the current program came from an RCT conducted in the Netherlands.19 This was chosen to represent current stroke care because it was undertaken in centers with limited prior experience of treating patients with MT. Data from the HERMES meta-analysis, used to represent clinical outcomes in the development scenario, were also based on trials from countries other than Saudi Arabia. Further, combining the two different data sources in this way assumes the differences in mRS outcomes were due to differences in medical practitioner experience and stroke care, whereas such differences could have occurred for many reasons. Therefore, factors unrelated to treatment could have caused additional variation in mRS scores.

ng the two different data sources in this way assumes the differences in mRS outcomes were due to differences in medical practitioner experience and stroke care, whereas such differences could have occurred for many reasons. Therefore, factors unrelated to treatment could have caused additional variation in mRS scores. Stroke incidence rates are based on a Saudi study from the 1990s as this was the only source to report separate values by age group, which was necessary to accurately reflect changes in stroke risk with age. The recent study in the Aseer region reported similar incidence rates; however, it showed an unusual pattern in the two most elderly groups of patients, where the stroke rate was considerably higher for males than females, which may not accurately reflect the stroke numbers amongst the projected population of women aged 65 years and over. Adapting the model from an existing model led to the assumption that all patients have the same age at first stroke. This was adjusted for the inclusion of societal costs, but for other parameters, an average starting age was assumed for all patients to calculate healthcare costs. Likewise, patients are assumed to stay at the same mRS after their first stroke until either recurrent stroke or death. Recurrent strokes are also assumed to occur after the first 90 days and can occur once per year.8

other parameters, an average starting age was assumed for all patients to calculate healthcare costs. Likewise, patients are assumed to stay at the same mRS after their first stroke until either recurrent stroke or death. Recurrent strokes are also assumed to occur after the first 90 days and can occur once per year.8 The organization of care considered in the model includes at least one comprehensive stroke hospital (CSH) in each region (13 in total), with more CSHs in more densely populated areas to cover higher stroke care demands. The CSH will be served by an integrated care stroke network with urgent referral systems from primary stroke hospitals (PSH) and acute stroke ready hospitals (ASRH). This organization has been included in the stroke standards developed by the Stroke Clinical Advisory Committee, chaired by the first author. The model considers most of the resources required to implement the changes in the organization of care, which in several cases will imply a reallocation of resources such as beds and ambulances. The inclusion of additional technologies such as telestroke, imaging or prenotification software could be possible due to the model flexibility; nonetheless, the costs of their inclusion are unlikely to alter the model results. Additional inpatient rehabilitation and long-term services required for the development program will be outsourced by the private sector, and will therefore not be an MOH investment.

ftware could be possible due to the model flexibility; nonetheless, the costs of their inclusion are unlikely to alter the model results. Additional inpatient rehabilitation and long-term services required for the development program will be outsourced by the private sector, and will therefore not be an MOH investment. A strength of the analysis was the extensive input from local stroke specialists, who provided all local data and validated the model. The model also reflects the impact of gradual implementation of changes in the development program until it reaches an optimal operation/functional level. Moreover, this model is a useful tool to estimate the future state of stroke care and can be replicated in other regions to assess stroke care organization alternatives to inform better resource allocation from healthcare decision-makers. Conclusion Ischemic stroke provides a suitable case study for economic models given the complexity of the care pathway, facilities, and expertise required to provide specific interventions. Saudi Arabia's demographic situation, coupled with the current sub-optimal care available for stroke, make it a prime candidate to explore future needs and the impact of infrastructure and patient management changes upon cost and patient outcomes. The development of stroke care could improve patient outcomes and lower overall costs compared with the current stroke care provision. The model outcomes enabled the decision-making process towards the improvement of stroke care in Saudi Arabia.

rastructure and patient management changes upon cost and patient outcomes. The development of stroke care could improve patient outcomes and lower overall costs compared with the current stroke care provision. The model outcomes enabled the decision-making process towards the improvement of stroke care in Saudi Arabia. Supplemental Material Supplemental material for A national economic and clinical model for ischemic stroke care development in Saudi Arabia: A call for change Click here for additional data file. Supplemental Material for A national economic and clinical model for ischemic stroke care development in Saudi Arabia: A call for change by Fahmi Al-Senani, Mohammed Al-Johani, Mohammad Salawati, Souda ElSheikh, Maha AlQahtani, Jamal Muthana, Saeed AlZahrani, Judith Shore, Matthew Taylor, Valeska S Ravest, Simon Eggington, Matthieu Cuche, Heather Davies, Kyriakos Lobotesis and Jeffrey L Saver in International Journal of Stroke

Arabia: A call for change by Fahmi Al-Senani, Mohammed Al-Johani, Mohammad Salawati, Souda ElSheikh, Maha AlQahtani, Jamal Muthana, Saeed AlZahrani, Judith Shore, Matthew Taylor, Valeska S Ravest, Simon Eggington, Matthieu Cuche, Heather Davies, Kyriakos Lobotesis and Jeffrey L Saver in International Journal of Stroke Declaration of conflicting interests The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: JLS is an employee of the University of California, which has patent rights in stroke retrieval devices. He serves as an unpaid consultant to Genentech advising on PRISMS trial. JLS has received contracted hourly payments and travel reimbursement for services as a scientific consultant advising on trial design and conduct to Medtronic/Covidien, Stryker, Neuravi/Cerenovus, BrainsGate, Boehringer Ingelheim, and Diffusion Medical. JLS has received contracted stock options for services as a scientific consultant advising on trial design and conduct to Rapid Medical. FA-S, MS and MA-J have received reimbursement from Medtronic for travel costs for a workshop related to this analysis. MT, JS, and HD are employed by YHEC. MC, SE, and VSR are employees of Medtronic. No other authors have any declarations. Funding The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: York Health Economics Consortium (YHEC) received funding from Medtronic for the model development.

Introduction and rationale Patients with acute cerebral ischemia are at high risk of recurrent ischemic events, particularly ischemic stroke1–6 and current international guidelines recommend antiplatelet therapy for secondary prevention in patients with acute stroke or transient ischemic attack (TIA) of non-cardioembolic origin. Aspirin is the only antiplatelet agent that has received a class 1A recommendation.7–9 Ticagrelor is a reversibly binding, direct-acting, oral P2Y12 receptor antagonist that prevents adenosine diphosphate-mediated P2Y12 dependent platelet activation and aggregation.10 Ticagrelor has a faster onset and achieves greater and more consistent platelet inhibition than clopidogrel,11 which requires metabolism to its active form through a pathway that is genetically determined.12 Poor responsiveness to clopidogrel is common, with a frequency that may be as high as 20–50% in some populations.13 The Acute Stroke or Transient Ischemic Attack Treated with Aspirin or Ticagrelor and Patient Outcomes (SOCRATES) trial (NCT01994720) investigated whether ticagrelor was superior to aspirin, when initiated within 24 h after symptom onset in patients with acute cerebral ischemia.14 The rate of stroke, myocardial infarction, or death during 90 days were numerically lower with ticagrelor as compared with aspirin (hazard ratio (HR) 0.89; 95% confidence interval (CI) 0.78 to 1.01; p = 0.07), with no increase in major hemorrhage.4,15

24 h after symptom onset in patients with acute cerebral ischemia.14 The rate of stroke, myocardial infarction, or death during 90 days were numerically lower with ticagrelor as compared with aspirin (hazard ratio (HR) 0.89; 95% confidence interval (CI) 0.78 to 1.01; p = 0.07), with no increase in major hemorrhage.4,15 These promising results prompted several secondary analyses of SOCRATES to guide the design of the present study. One planned secondary analysis indicated benefit in reducing recurrent stroke events; HR 0.86 (95% CI 0.75 to 0.99; nominal p = 0.03). Stroke constituted almost 90% of primary events in SOCRATES, and the highest risk for new stroke events was seen during the first 30 days of treatment.4 Regarding population, patients with an ABCD2 score of 4–5 had a lower event rate compared with an ABCD2 score of 6–7 or minor ischemic strokes, consistent with international registry data.6 Also, patients with acute stroke/TIA with ipsilateral large vessel stenosis seemed to benefit more from ticagrelor treatment compared with aspirin.16 In another subgroup analysis, the treatment effect of ticagrelor was more pronounced in patients who received aspirin within 7 days before randomization.17 Since the antiplatelet effect of aspirin persisted into the first week of the trial among this group, short-term dual antiplatelet therapy (DAPT) could account for the greater benefit of ticagrelor in patients taking aspirin prior to randomization. This observation is in line with other studies suggesting that DAPT with clopidogrel and aspirin may be more effective in reducing the high risk of stroke after an acute ischemic stroke or TIA compared with aspirin alone, including studies of microembolization from atherosclerotic cerebral arteries in patients with acute cerebral ischemic events18,19 and in trials of patients with minor stroke or TIA.3,5

d aspirin may be more effective in reducing the high risk of stroke after an acute ischemic stroke or TIA compared with aspirin alone, including studies of microembolization from atherosclerotic cerebral arteries in patients with acute cerebral ischemic events18,19 and in trials of patients with minor stroke or TIA.3,5 The pharmacological properties of ticagrelor support the hypothesis that combining ticagrelor with aspirin would be an even more effective DAPT combination, without non-responders and without a clinically unacceptable risk of severe bleeding events. The aim of the THALES trial is, therefore, to test whether DAPT with ticagrelor and aspirin results in clinical benefit to patients with acute cerebral ischemia. Methods Design The THALES trial (NCT03354429) is a randomized, placebo-controlled, double-blind, parallel-group, international, multicenter, phase III study to test the hypothesis that ticagrelor and aspirin is superior to placebo and aspirin in preventing stroke and death in patients with acute cerebral ischemia.

The pharmacological properties of ticagrelor support the hypothesis that combining ticagrelor with aspirin would be an even more effective DAPT combination, without non-responders and without a clinically unacceptable risk of severe bleeding events. The aim of the THALES trial is, therefore, to test whether DAPT with ticagrelor and aspirin results in clinical benefit to patients with acute cerebral ischemia. Methods Design The THALES trial (NCT03354429) is a randomized, placebo-controlled, double-blind, parallel-group, international, multicenter, phase III study to test the hypothesis that ticagrelor and aspirin is superior to placebo and aspirin in preventing stroke and death in patients with acute cerebral ischemia. Patients will be randomized within 24 h of symptom onset to 30 days of treatment with ticagrelor or placebo on top of standard-of-care therapy with aspirin and followed up for 30 days for efficacy and 60 days for safety (Figure 1). Key design features of the THALES trial compared with the SOCRATES trial are presented in Table 1. In SOCRATES, the estimated treatment effect was similar whether adjudicated events or investigator-reported events were used in the analysis;20 therefore, investigator-reported outcomes will be used in THALES. Figure 1. THALES study design. R: randomization. Ticagrelor 180 mg loading dose (day 1) then 90 mg twice daily (days 2–30). Aspirin 300–325 mg loading dose (day 1) then 75–100 mg daily (days 2–30). Table 1. Comparison of THALES and SOCRATES study designs

Patients will be randomized within 24 h of symptom onset to 30 days of treatment with ticagrelor or placebo on top of standard-of-care therapy with aspirin and followed up for 30 days for efficacy and 60 days for safety (Figure 1). Key design features of the THALES trial compared with the SOCRATES trial are presented in Table 1. In SOCRATES, the estimated treatment effect was similar whether adjudicated events or investigator-reported events were used in the analysis;20 therefore, investigator-reported outcomes will be used in THALES. Figure 1. THALES study design. R: randomization. Ticagrelor 180 mg loading dose (day 1) then 90 mg twice daily (days 2–30). Aspirin 300–325 mg loading dose (day 1) then 75–100 mg daily (days 2–30). Table 1. Comparison of THALES and SOCRATES study designs Comparison THALES n ∼13,000 SOCRATES n = 13,199 Dose regimen Ticagrelor vs. placebo on top of aspirin (DAPT) Ticagrelor vs. aspirin (single antiplatelet therapy) Study duration 30 days + 30 days follow-up 90 days + 30 days follow-up Population TIA with ABCD2 score ≥ 6 and/or ipsilateral stenosis and acute ischemic stroke NIHSS ≤5 TIA with ABCD2 score ≥ 4 and/or ipsilateral stenosis and acute ischemic stroke NIHSS ≤5 Endpoints Primary efficacy Secondary efficacy Safety Stroke + death Ischemic stroke, disability (mRS) Severe bleeding and AEs leading to discontinuation of study medication Stroke + myocardial infarction + death Ischemic stroke, net clinical outcome Major bleeding and AEs leading to discontinuation of study medication DAPT: dual antiplatelet therapy; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; TIA: transient ischemic attack.

tinuation of study medication Stroke + myocardial infarction + death Ischemic stroke, net clinical outcome Major bleeding and AEs leading to discontinuation of study medication DAPT: dual antiplatelet therapy; mRS: modified Rankin Scale; NIHSS: National Institutes of Health Stroke Scale; TIA: transient ischemic attack. The study is event-driven with approximately 13,000 patients expected to be randomized from approximately 450 sites in 28 countries worldwide to identify 764 outcome events. The first patient was recruited on 22 January 2018. Patient population Eligible are patients ≥ 40 years of age who have experienced a non-cardioembolic acute ischemic stroke with a National Institutes of Health Stroke Scale score ≤5 or high-risk TIA (defined as an ABCD2 score ≥ 6 or ipsilateral atherosclerotic stenosis ≥50% in an extra/intracranial artery) who can be randomized within 24 h of symptom onset or for wake-up strokes since last time known to be free of new ischemic symptoms. The complete inclusion and exclusion criteria are shown in Tables S1 and S2, respectively. Randomization Randomization codes are computer-generated by the AstraZeneca Global Randomization system and loaded into the Interactive Web Response System database. Patients are randomized as soon as possible after symptom onset, but within 24 h.

Patient population Eligible are patients ≥ 40 years of age who have experienced a non-cardioembolic acute ischemic stroke with a National Institutes of Health Stroke Scale score ≤5 or high-risk TIA (defined as an ABCD2 score ≥ 6 or ipsilateral atherosclerotic stenosis ≥50% in an extra/intracranial artery) who can be randomized within 24 h of symptom onset or for wake-up strokes since last time known to be free of new ischemic symptoms. The complete inclusion and exclusion criteria are shown in Tables S1 and S2, respectively. Randomization Randomization codes are computer-generated by the AstraZeneca Global Randomization system and loaded into the Interactive Web Response System database. Patients are randomized as soon as possible after symptom onset, but within 24 h. Treatments Randomized patients receive either ticagrelor 180 mg loading dose on day 1, then 90 mg twice daily during the study treatment period or matching placebo, in addition to receiving standard-of-care open-label aspirin 300–325 mg on day 1, then 75–100 mg once daily during the study treatment period (Figure 1). The loading dose of ticagrelor/placebo should be given immediately after randomization. After the 30 days of study treatment, patients are treated with standard-of-care therapy at the discretion of the investigator and followed up for an additional 30 days with continued collection of endpoints and safety events. Primary outcomes The primary outcome is the time from randomization to first subsequent investigator-reported stroke or all-cause death at 30 days.

Treatments Randomized patients receive either ticagrelor 180 mg loading dose on day 1, then 90 mg twice daily during the study treatment period or matching placebo, in addition to receiving standard-of-care open-label aspirin 300–325 mg on day 1, then 75–100 mg once daily during the study treatment period (Figure 1). The loading dose of ticagrelor/placebo should be given immediately after randomization. After the 30 days of study treatment, patients are treated with standard-of-care therapy at the discretion of the investigator and followed up for an additional 30 days with continued collection of endpoints and safety events. Primary outcomes The primary outcome is the time from randomization to first subsequent investigator-reported stroke or all-cause death at 30 days. Secondary outcomes Two secondary efficacy outcomes will be evaluated hierarchically after assessing the primary outcome. First, time from randomization to first subsequent ischemic stroke will be assessed, and then modified Rankin Scale (mRS) score > 1 at the end of treatment (visit 3).21,22

Primary outcomes The primary outcome is the time from randomization to first subsequent investigator-reported stroke or all-cause death at 30 days. Secondary outcomes Two secondary efficacy outcomes will be evaluated hierarchically after assessing the primary outcome. First, time from randomization to first subsequent ischemic stroke will be assessed, and then modified Rankin Scale (mRS) score > 1 at the end of treatment (visit 3).21,22 The main safety outcomes are: time from randomization to first bleeding event, categorized as severe based on criteria from the Global Utilization of Streptokinase and tissue-type plasminogen activator for Occluded Coronary Arteries (GUSTO) trial; time from randomization to first intracranial hemorrhage or fatal bleeding event; time from randomization to first bleeding event categorized as GUSTO moderate or severe; time from randomization to premature permanent discontinuation of study treatment due to bleeding.23 Asymptomatic hemorrhagic transformations of brain infarctions and microhemorrhages <10 mm evident only on gradient-echo magnetic resonance imaging are excluded from fulfilling intracranial hemorrhage criteria. This revision is an adaptation of the standard GUSTO definition to better distinguish clinically relevant events in the acute stroke population, and has been the common convention in recent large stroke trials.15 Serious adverse events (SAEs) and AEs leading to premature and permanent discontinuation of study medication will also be assessed.

tation of the standard GUSTO definition to better distinguish clinically relevant events in the acute stroke population, and has been the common convention in recent large stroke trials.15 Serious adverse events (SAEs) and AEs leading to premature and permanent discontinuation of study medication will also be assessed. Pre-defined exploratory outcomes are time from randomization to first subsequent stroke or death in patients with ipsilateral atherosclerotic stenosis; mRS score > 2 at visit 3 in patients with subsequent stroke; and generic health status (using the EQ-5D-5L questionnaire). Data monitoring committee An independent data monitoring committee (DMC) will review on an ongoing basis accumulating study data to safeguard the interests of the patients. The DMC will assess the benefit/risk profile of the intervention during the study, ensure the validity and integrity of the study, review the overall conduct of the study, and provide recommendations to the Executive Committee (EC) regarding the continued conduct of the study. The DMC will have access to the individual treatment codes and will be able to merge these with the collected study data while the study is ongoing. One interim analysis for efficacy and futility will be performed by the DMC following the accrual of approximately 60% of planned primary events. A DMC Charter details roles, responsibilities, and procedures to ensure maintenance of the blinding and integrity of the study.

Data monitoring committee An independent data monitoring committee (DMC) will review on an ongoing basis accumulating study data to safeguard the interests of the patients. The DMC will assess the benefit/risk profile of the intervention during the study, ensure the validity and integrity of the study, review the overall conduct of the study, and provide recommendations to the Executive Committee (EC) regarding the continued conduct of the study. The DMC will have access to the individual treatment codes and will be able to merge these with the collected study data while the study is ongoing. One interim analysis for efficacy and futility will be performed by the DMC following the accrual of approximately 60% of planned primary events. A DMC Charter details roles, responsibilities, and procedures to ensure maintenance of the blinding and integrity of the study. Sample size estimates The study is event-driven and the final number of randomized patients will be based on the blinded data review of overall primary endpoints. At least 764 primary endpoint events are needed to provide 85% power, assuming an HR of 0.805 (based on an HR of 0.8 for stroke and cardiovascular death, and an HR of 1.0 for non-cardiovascular death) in favor of ticagrelor at the significance level of 4.988%, adjusted for the planned interim analysis. Based on data from the SOCRATES study, a primary endpoint rate of 6.7% in the placebo group is assumed at 30 days following randomization. Hence, randomizing approximately 13,000 patients to ticagrelor or placebo, in a 1:1 ratio, is expected to yield the required 764 events.

, adjusted for the planned interim analysis. Based on data from the SOCRATES study, a primary endpoint rate of 6.7% in the placebo group is assumed at 30 days following randomization. Hence, randomizing approximately 13,000 patients to ticagrelor or placebo, in a 1:1 ratio, is expected to yield the required 764 events. Statistical analyses All efficacy and safety analyses will be based on the intention-to-treat principle. In time-to-event analyses, the treatment groups will be compared using a Cox proportional hazards model with a factor for treatment group, using the Efron method for ties. P-values and 95% CIs for the HRs will be based on the Wald statistic. The primary and secondary efficacy outcomes will be included in a confirmatory testing procedure. Only if the analysis of the primary outcome is significant at the 4.988% level (adjusted for the interim analysis) will the secondary outcomes be tested in a confirmatory sense in the specified hierarchical order.

Statistical analyses All efficacy and safety analyses will be based on the intention-to-treat principle. In time-to-event analyses, the treatment groups will be compared using a Cox proportional hazards model with a factor for treatment group, using the Efron method for ties. P-values and 95% CIs for the HRs will be based on the Wald statistic. The primary and secondary efficacy outcomes will be included in a confirmatory testing procedure. Only if the analysis of the primary outcome is significant at the 4.988% level (adjusted for the interim analysis) will the secondary outcomes be tested in a confirmatory sense in the specified hierarchical order. Study organization and funding The EC, in collaboration with the sponsor, is responsible for the overall design, study protocol and amendments, interpretation, supervision, and reporting of the study results at international congresses and publishing in peer-reviewed journals. AstraZeneca is responsible for the operational study conduct. The EC will make recommendations to AstraZeneca regarding early stopping or modifications of the study based on the recommendations received from the DMC. The EC is comprised of six designated international academic experts including the Steering Committee chair and three non-voting AstraZeneca representatives, and operates under a separate charter.

s to AstraZeneca regarding early stopping or modifications of the study based on the recommendations received from the DMC. The EC is comprised of six designated international academic experts including the Steering Committee chair and three non-voting AstraZeneca representatives, and operates under a separate charter. The International Steering Committee is comprised of national lead investigators from each country where the study is conducted (Table S3) and will be supervised by the EC. Members of the Steering Committee will be responsible for providing clinical guidance on study implementation and conduct in their respective countries. This study is sponsored by AstraZeneca. Discussion Acute cerebral ischemia most often presents as a non-severe stroke or TIA,24 and risk of recurrent ischemic events is very high, particularly in the first few days.1–6 Therefore, urgent assessment of these cases is required to initiate treatment to reduce the risk of subsequent severe, disabling stroke. In this setting, aspirin is the only antiplatelet agent with a class 1A recommendation in international guidelines7–9 and the only antiplatelet therapy that has been shown to reduce disabling strokes.25 However, the status of aspirin as the standard-of-care for all non-cardioembolic strokes has been challenged by two trials investigating DAPT containing clopidogrel.3,5

gent with a class 1A recommendation in international guidelines7–9 and the only antiplatelet therapy that has been shown to reduce disabling strokes.25 However, the status of aspirin as the standard-of-care for all non-cardioembolic strokes has been challenged by two trials investigating DAPT containing clopidogrel.3,5 The Platelet-Oriented Inhibition in New TIA and minor ischemic stroke (POINT) trial26 recently reported a benefit for clopidogrel and aspirin over aspirin alone.5 The POINT trial was stopped early by the Data and Safety Monitoring Board after an interim analysis with 84% of pre-planned events. POINT reported both a benefit in reduction of ischemic events and an increased risk of major bleeding events. However, the robustness of the POINT trial results suffers from a high rate of study-drug discontinuation and a large number of patients lost to follow-up without known vital status. The Clopidogrel in High-Risk Patients with Acute Nondisabling Cerebrovascular Events (CHANCE) trial reported similar efficacy for clopidogrel-aspirin compared with aspirin in reducing stroke risk but did not demonstrate an increase in major bleeding events.3 Concerns about differences in standards of care, stroke subtypes, duration of antiplatelet therapy, and genetics of clopidogrel metabolism limited generalization of CHANCE findings beyond China, where it was performed.

d with aspirin in reducing stroke risk but did not demonstrate an increase in major bleeding events.3 Concerns about differences in standards of care, stroke subtypes, duration of antiplatelet therapy, and genetics of clopidogrel metabolism limited generalization of CHANCE findings beyond China, where it was performed. Thus, despite encouraging results in the POINT and CHANCE trials, clinically important questions remain in this setting. Treatment with clopidogrel in combination with aspirin in acute cerebral ischemia has not been evaluated by health authorities. Furthermore, clopidogrel as single antiplatelet therapy has not been approved for secondary prevention of stroke within 7 days after an acute stroke/TIA. Internationally, health authorities have approved aspirin for the indication of reducing the risk of death and recurrent stroke in patients experiencing an ischemic stroke or TIA, noting that aspirin treatment can be started in the acute setting.27,28 Regulatory agencies in the US, Europe, and China were consulted during the planning of the THALES trial and all stated that placebo, rather than clopidogrel, was the appropriate comparator in patients receiving standard-of-care aspirin in acute minor stroke/TIA.

pirin treatment can be started in the acute setting.27,28 Regulatory agencies in the US, Europe, and China were consulted during the planning of the THALES trial and all stated that placebo, rather than clopidogrel, was the appropriate comparator in patients receiving standard-of-care aspirin in acute minor stroke/TIA. There are potential benefits of ticagrelor therapy that should be acknowledged. Ticagrelor has a faster onset of action and achieves greater, more consistent, and predictable platelet inhibition than clopidogrel.11–13 Nearly all poor responders to clopidogrel will have platelet reactivity below the cut-off points associated with ischemic risk when treated with ticagrelor.12

should be acknowledged. Ticagrelor has a faster onset of action and achieves greater, more consistent, and predictable platelet inhibition than clopidogrel.11–13 Nearly all poor responders to clopidogrel will have platelet reactivity below the cut-off points associated with ischemic risk when treated with ticagrelor.12 The rapid and consistent onset of action of ticagrelor is expected to be important in all patients with acute cerebral ischemia, whereas the rapid offset may be more important for patients who are likely to undergo surgical intervention, such as patients with carotid stenosis. A subgroup analysis from the SOCRATES trial suggests that patients with ischemic events attributable to ipsilateral atherosclerotic stenosis may have a greater treatment effect with ticagrelor compared with those without ipsilateral stenosis.16 Patients with carotid stenosis were largely excluded from POINT,29 and were uncommon in the Chinese population enrolled in CHANCE.3 THALES has the opportunity to improve outcomes for these high-risk patients since patients with carotid stenosis are eligible for THALES, and the fast onset, fast offset, and reversible binding of ticagrelor11,12 may be an advantage compared with the irreversible inhibitor, clopidogrel. Moreover, preliminary analysis of the ticagrelor with aspirin on Platelet Reactivity In acute Non-disabling Cerebrovascular Events (PRINCE) trial showed a benefit of ticagrelor with aspirin compared with clopidogrel plus aspirin on platelet reactivity, which was the primary endpoint.30 Moreover, 21 strokes in 335 patients (6.3%) were reported in the ticagrelor-aspirin arm vs. 30 strokes in 339 patients (8.8%) in the clopidogrel-aspirin arm.

Events (PRINCE) trial showed a benefit of ticagrelor with aspirin compared with clopidogrel plus aspirin on platelet reactivity, which was the primary endpoint.30 Moreover, 21 strokes in 335 patients (6.3%) were reported in the ticagrelor-aspirin arm vs. 30 strokes in 339 patients (8.8%) in the clopidogrel-aspirin arm. The ticagrelor dose in THALES is the same as that used in the target population of patients with ischemic stroke or TIA in the SOCRATES study, in which ticagrelor was well-tolerated and had a similar safety profile as aspirin with respect to major bleeding events.4,15 In patients with acute coronary syndrome, the same ticagrelor dose was used and treatment with ticagrelor compared with clopidogrel resulted not only in a higher degree of platelet inhibition31 but also in a reduced rate of myocardial infarction, stroke, or death from vascular causes, without an increase in the rate of overall major bleeding in the PLATO study.32 Importantly, the efficacy and bleeding results with ticagrelor in high-risk patients with a history of stroke or TIA were consistent with the overall PLATO trial population, with a favorable clinical net benefit and associated impact on mortality.33

n increase in the rate of overall major bleeding in the PLATO study.32 Importantly, the efficacy and bleeding results with ticagrelor in high-risk patients with a history of stroke or TIA were consistent with the overall PLATO trial population, with a favorable clinical net benefit and associated impact on mortality.33 Summary and conclusions Overall, the available data support the potential for DAPT with ticagrelor and aspirin to improve outcomes in patients with acute ischemic stroke or high-risk TIA as compared with aspirin alone, the current standard of care. Despite advances in the field, there are still clinically important questions remaining about the benefits and risks of DAPT. The THALES trial will address these questions with a design and study conduct that will meet the rigorous standards of regulatory authorities. Supplemental Material Supplemental material for The Acute Stroke or Transient Ischemic Attack Treated with Ticagrelor and Aspirin for Prevention of Stroke and Death (THALES) trial: Rationale and design Click here for additional data file. Supplemental Material for The Acute Stroke or Transient Ischemic Attack Treated with Ticagrelor and Aspirin for Prevention of Stroke and Death (THALES) trial: Rationale and design by S Claiborne Johnston, Pierre Amarenco, Hans Denison, Scott R Evans, Anders Himmelmann, Stefan James, Mikael Knutsson, Per Ladenvall, Carlos A Molina, Yongjun Wang and for the THALES Investigators in International Journal of Stroke

or and Aspirin for Prevention of Stroke and Death (THALES) trial: Rationale and design by S Claiborne Johnston, Pierre Amarenco, Hans Denison, Scott R Evans, Anders Himmelmann, Stefan James, Mikael Knutsson, Per Ladenvall, Carlos A Molina, Yongjun Wang and for the THALES Investigators in International Journal of Stroke Acknowledgments Editorial support (formatting tables and figures, co-ordinating reviews, and preparing the manuscript for submission) was provided by Jackie Phillipson (Zoetic Science, an Ashfield company, part of UDG Healthcare plc, Macclesfield, UK). Authors' contributions All authors contributed to the design of the THALES study and will contribute to its oversight. Clay Johnston wrote the first draft of the manuscript, which was edited by all other authors.

Acknowledgments Editorial support (formatting tables and figures, co-ordinating reviews, and preparing the manuscript for submission) was provided by Jackie Phillipson (Zoetic Science, an Ashfield company, part of UDG Healthcare plc, Macclesfield, UK). Authors' contributions All authors contributed to the design of the THALES study and will contribute to its oversight. Clay Johnston wrote the first draft of the manuscript, which was edited by all other authors. Declaration of conflicting interests The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: S Claiborne Johnston reports receiving research grants from the National Institutes of Neurological Disorders and Stroke (NINDS) and the National Institutes of Health (NIH), including for the POINT trial; Sanofi provided free drug and placebo to the NIH for patients in the POINT trial. His institution received research support (significant) for the SOCRATES and THALES trials. Pierre Amarenco reports receiving significant research grant support from AstraZeneca, Sanofi, and Bristol-Myers Squibb (for TIAregistry.org), the French Government and Pfizer (for the TST trial), and Boston Scientific (for the WATCH-AF registry). He has received modest consultant/advisory board fees from Amgen and Bristol-Myers Squibb. He has also received modest honoraria from Amgen (speaker activities), Pfizer (SPIRE Program Executive Committee), AstraZeneca (SOCRATES Trial Executive Committee), and Kowa (PROMINENT Executive Committee), and significant honoraria from Bayer (XANTUS Executive Committee), AstraZeneca (THALES Executive Committee), and Fibrogen (ALPINE program trials DSMB member). Hans Denison, Anders Himmelmann, Mikael Knutsson, and Per Ladenvall are employees of AstraZeneca (all significant). Scott R Evans is a statistical consultant to AstraZeneca (significant) for the SOCRATES and THALES trials. Stefan James has received institutional research grants and lecture fees from AstraZeneca. Carlos A Molina reports serving on the Steering Committee (significant) of the Combined lysis of thrombus with ultrasound and systemic tissue plasminogen activator for emergent revascularization in acute ischemic stroke trial (Cerevast); SOCRATES (AstraZeneca), Implant Augmenting Cerebral Blood Flow Trial 24 hours from stroke onset trial (Brainsgate), Endovascular Revascularization With Solitaire Device Versus Best Medical Therapy in Anterior Circulation Stroke Within 8 Hours trial (Fundació Ictus Malaltia Vascular).

e ischemic stroke trial (Cerevast); SOCRATES (AstraZeneca), Implant Augmenting Cerebral Blood Flow Trial 24 hours from stroke onset trial (Brainsgate), Endovascular Revascularization With Solitaire Device Versus Best Medical Therapy in Anterior Circulation Stroke Within 8 Hours trial (Fundació Ictus Malaltia Vascular). He has received honoraria for participation in clinical trials, and contribution to advisory boards or oral presentations from: AstraZeneca (modest), Boehringer Ingelheim, Daiichi Sankyo, Bristol-Myers Squibb, Covidien, Cerevast, and Brainsgate. He has no ownership interest and does not own stocks of any pharmaceutical or medical device company. Yongjun Wang reports research grant support from AstraZeneca. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The THALES study is funded by AstraZeneca.

Introduction Acute ischemic stroke patients should receive intravenous thrombolysis (IVT) with recombinant tissue plasminogen activator (rt-PA) as quickly as possible for optimal clinical efficacy.1–5 The most feared complication of IVT is symptomatic intracranial hemorrhage (sICH) occurring in 2.7–5.7% of patients.6 Dose finding trials for rt-PA indicate that 0.9 mg/kg body weight has an optimal safety and efficacy profile: a lower dose resulted in reduced efficacy and a higher dose in increased sICH risk.7-9 The patient’s weight is therefore essential, but exact measurement can be time consuming leading to increased door-to-needle times (DNTs) with less clinical IVT efficacy. Therefore, estimation of body weight (EBW), rather than exact measurement of body weight (MBW), is often used with potential under- or overestimation.10–14 Indeed, overestimation due to EBW was shown to result in increased rt-PA dose and increased sICH risk.11,12,15,16 In contrast, other studies did not confirm this observation, indicating that EBW is acceptable since dosing errors did not influence outcomes.13,17 However, the sample sizes of these studies so far are small (n = 222; n = 308) and were therefore underpowered to detect differences in sICH rate. Hence, based on available evidence, it is not possible to draw conclusions on the best weight modality.

BW is acceptable since dosing errors did not influence outcomes.13,17 However, the sample sizes of these studies so far are small (n = 222; n = 308) and were therefore underpowered to detect differences in sICH rate. Hence, based on available evidence, it is not possible to draw conclusions on the best weight modality. National and the American Stroke Association guidelines lack recommendations regarding weight modality, thus both EBW and MBW are being used in clinical practice.18,19 We used this disparity to assess if weight modality is associated with (i) sICH rate, (ii) clinical outcome, and (iii) DNT. Methods Study design and patient selection We derived data from prospective IVT registries of 14 centers and included consecutive adult patients with acute ischemic stroke (AIS) treated with IVT between January 2009 and December 2016. Patients were excluded if weight modality was unknown or if no clinical data were available. The ethical standards committee of the Leiden University Medical Centre approved the protocol and waived the need for written informed consent from individual patients.

Methods Study design and patient selection We derived data from prospective IVT registries of 14 centers and included consecutive adult patients with acute ischemic stroke (AIS) treated with IVT between January 2009 and December 2016. Patients were excluded if weight modality was unknown or if no clinical data were available. The ethical standards committee of the Leiden University Medical Centre approved the protocol and waived the need for written informed consent from individual patients. Patient data The following data were collected: patient characteristics including demographics, vascular risk factors and history, medication use, admission blood pressure, and baseline stroke severity assessed with the National Institute of Health Stroke Scale (NIHSS) score. In case data were missing, these were complemented from the medical records. In case NIHSS score was not noted, this was reconstructed from neurological examination at admission with a validated algorithm as described previously.20 Weight assessment Mode of weight assessment during the inclusion period was acquired by asking the stroke neurologist involved and by assessing local protocols of each participating center. In all centers, either estimation or exact MBW was done before the CT scan. None of the EBW centers measured body weight during (infusion of) IVT, so possible discrepancies between reported and estimated weight did not led to alteplase dose adjustments.

ved and by assessing local protocols of each participating center. In all centers, either estimation or exact MBW was done before the CT scan. None of the EBW centers measured body weight during (infusion of) IVT, so possible discrepancies between reported and estimated weight did not led to alteplase dose adjustments. In the EBW group, policy was similar in all centers: (i) weight was assessed first by asking the patient; (ii) in case this was not possible (e.g. due to aphasia) by asking a relative; and (iii) if this was not possible estimation was always done by the treating physician, but in case another health care worker had a different estimation, consensus was reached. In the MBW group, weight was measured: (i) by transferring the patient to a bed with an inbuilt weighing option or a stretcher standing on a ground scale or (ii) by using a patient lift scale, requiring to lift the patient in a sling.

In the EBW group, policy was similar in all centers: (i) weight was assessed first by asking the patient; (ii) in case this was not possible (e.g. due to aphasia) by asking a relative; and (iii) if this was not possible estimation was always done by the treating physician, but in case another health care worker had a different estimation, consensus was reached. In the MBW group, weight was measured: (i) by transferring the patient to a bed with an inbuilt weighing option or a stretcher standing on a ground scale or (ii) by using a patient lift scale, requiring to lift the patient in a sling. Outcome measures Our primary outcome measure was the sICH rate. We defined sICH according to the ECASS-III definition, i.e. any apparently extravascular blood in the brain or within the cranium that was associated with clinical deterioration, as defined by an increase of 4 or more points on NIHSS score, or that led to death and that was identified as the predominant cause of the neurological deterioration.3 In our study, we included all sICH within seven days after stroke onset. Secondary outcome measures included favorable outcome at 90 days (defined as a score of 0–2 on the modified Rankin Scale (mRS)) and DNT (which was defined as the time between patient arrival at the hospital and intravenous rt-PA initiation).21 In case of missing data on clinical outcome at 90 days, the mRS was derived using available follow-up data before 3 months and ≥1 month after hospital discharge. Both sICH and clinical outcome were retrieved from medical records, including neuro-imaging data by two independent reviewers (TTMN and AEG). Discrepancies were solved by discussion. Time of symptom onset, time of center arrival, and time of IVT initiation were extracted to calculate the DNT.

after hospital discharge. Both sICH and clinical outcome were retrieved from medical records, including neuro-imaging data by two independent reviewers (TTMN and AEG). Discrepancies were solved by discussion. Time of symptom onset, time of center arrival, and time of IVT initiation were extracted to calculate the DNT. Statistical analysis Descriptive statistics were used to compare patient characteristics. Categorical variables were compared with χ2 test. Continuous variables were compared using the t test or Mann–Whitney U test, and are presented as mean ± standard deviation (SD) or median and interquartile range (IQR).

after hospital discharge. Both sICH and clinical outcome were retrieved from medical records, including neuro-imaging data by two independent reviewers (TTMN and AEG). Discrepancies were solved by discussion. Time of symptom onset, time of center arrival, and time of IVT initiation were extracted to calculate the DNT. Statistical analysis Descriptive statistics were used to compare patient characteristics. Categorical variables were compared with χ2 test. Continuous variables were compared using the t test or Mann–Whitney U test, and are presented as mean ± standard deviation (SD) or median and interquartile range (IQR). We used logistic regression to assess the association of separate outcomes (sICH and clinical outcome) in relation to weight modality, expressed as odds ratios (ORs) or adjusted ORs (aOR) with corresponding 95% confidence interval (CI). Linear regression analysis was performed to assess the association between weight modality and DNT, presented as regression coefficient (B) and corresponding 95% CI. In secondary analysis, we adjusted for baseline characteristics associated with outcomes (P < 0.1) except for the analysis related to the outcome DNT where we adjusted for variables known to have an association with the DNT: availability of a CT in the emergency room (ER), blood pressure above the threshold for IVT (>185/110 mmHg), NIHSS score at baseline,22,23 onset-to-door time (defined as the time between stroke onset and patient arrival at the hospital) and for annual IVT-volume divided as follows: low-volume (≤24), medium-volume (25–49), or high-volume (≥50) as described previously, with low-volume as reference category.24 In subgroup analyses we investigated if differences in methods within the EBW or the MBW group could have affected the association between weight modality and the outcome measures.

as follows: low-volume (≤24), medium-volume (25–49), or high-volume (≥50) as described previously, with low-volume as reference category.24 In subgroup analyses we investigated if differences in methods within the EBW or the MBW group could have affected the association between weight modality and the outcome measures. Missing data For missing data we performed multiple imputations with the fully conditional specification method with five sets of imputations. The predictive mean matching model type was used for scale variables. Then, we compared the results of the analysis of the imputed dataset with the nonimputed dataset to assess if this leads to consistent parameter estimates. Additionally, we performed post hoc sensitivity analyses to assess the effect of missing data with regards to mRS score after 90 days, by recalculating the estimates while omitting patients with missing mRS score after 90 days. Statistical analysis was performed using SPSS software (version 23, IBM, New York, USA).

Missing data For missing data we performed multiple imputations with the fully conditional specification method with five sets of imputations. The predictive mean matching model type was used for scale variables. Then, we compared the results of the analysis of the imputed dataset with the nonimputed dataset to assess if this leads to consistent parameter estimates. Additionally, we performed post hoc sensitivity analyses to assess the effect of missing data with regards to mRS score after 90 days, by recalculating the estimates while omitting patients with missing mRS score after 90 days. Statistical analysis was performed using SPSS software (version 23, IBM, New York, USA). Results Baseline characteristics Data from 5066 patients with AIS were collected. A total of 4801 (95%) patients met the inclusion criteria (Figure 1). Five centers used MBW, six centers EBW, and three centers changed from EBW to MBW during our inclusion time window. In 2048 of the patients (43%), MBW was used and in 2753 patients (57%) EBW. EBW-patients were slightly older and they had more cardiovascular risk factors (atrial fibrillation, diabetes mellitus, hypertension, and hyperlipidemia) (Table 1). Other known predictors for sICH (sex, NIHSS score, blood pressure, and onset-to-door time) did not differ between the EBW and the MBW group. More EBW-patients were treated in high-volume centers (n = 2181; 79%) compared to MBW-patients (n = 1121; 55%) and a CT in the ER was present for 407 EBW-patients (15%) and for 927 MBW-patients (45%). Figure 1. Flowchart of the study. aTen patients had incomplete data and an unknown weight modality. AIS: acute ischemic stroke; IVT: intravenous thrombolysis.

volume centers (n = 2181; 79%) compared to MBW-patients (n = 1121; 55%) and a CT in the ER was present for 407 EBW-patients (15%) and for 927 MBW-patients (45%). Figure 1. Flowchart of the study. aTen patients had incomplete data and an unknown weight modality. AIS: acute ischemic stroke; IVT: intravenous thrombolysis. Table 1. Patient characteristics

volume centers (n = 2181; 79%) compared to MBW-patients (n = 1121; 55%) and a CT in the ER was present for 407 EBW-patients (15%) and for 927 MBW-patients (45%). Figure 1. Flowchart of the study. aTen patients had incomplete data and an unknown weight modality. AIS: acute ischemic stroke; IVT: intravenous thrombolysis. Table 1. Patient characteristics Variables Measured body weight (n = 2048) Missing data % Estimated body weight (n = 2753) Missing data % P-values Patient characteristics Age at stroke, years—mean (±SD) 70 (±14) 0 71 (±14) 0 <0.01 Male sex—n (%) 1122 (54.8) 0.1 1502 (54.6) 0 0.86 Vascular risk factors Atrial fibrillation—n (%) 175 (8.6) 1.2 376 (14.1) 2.9 <0.01 Diabetes mellitus—n (%) 309 (15.3) 1.1 469 (17.5) 2.8 0.04 Hypertension—n (%) 840 (41.5) 1.2 1355 (50.7) 2.9 <0.01 Hyperlipidemia—n (%) 238 (11.8) 1.3 841 (31.6) 3.2 <0.01 Coronary artery disease—n (%) 410 (20.3) 1.2 507 (19.0) 2.9 0.27 Peripheral vascular disease—n (%) 112 (5.6) 1.6 156 (5.8) 2.8 0.69 Prior TIA/stroke—n (%) 528 (26.2) 1.5 660 (24.7) 2.9 0.25 Medication Antiplatelets—n (%) 449 (37.9) 42.2 638 (37.0) 37.4 0.62 Anticoagulation—n (%) 35 (3.0) 42.2 65 (3.8) 37.3 0.24 Admittance Systolic BP, mmHg, mean (±SD) 156 (±25) 5.9 156 (±26) 15.4 0.43 Diastolic BP, mmHg, mean (±SD) 86 (±17) 5.9 85 (±27) 15.4 0.02 NIHSS, median [IQR] 7 [4–12] 1.3 6 [3–12] 0.7 0.08 ODT, min–median [IQR] 69 [45–115] 13.5 69 [45–112] 7.8 0.89 IVT-volume (IVT/year) High volume (≥50)—n (%) 1121 (54.7) 0 2181 (79.2) 0 <0.01 Medium volume (25–49)—n (%) 656 (32.0) 0 470 (17.1) 0 <0.01 Low volume (≤24)—n (%) 271 (13.2) 0 102 (3.7) 0 <0.01 CT available in the ER 927 (45.3) 0 407 (14.8) 0 <0.01 BP: blood pressure; CT: computed tomography scan; ER: emergency room; IQR: interquartile range; IVT: intravenous thrombolysis; NIHSS: National Institute of Health Stroke Scale; ODT: onset-to-door time; TIA: transient ischemic attack.

271 (13.2) 0 102 (3.7) 0 <0.01 CT available in the ER 927 (45.3) 0 407 (14.8) 0 <0.01 BP: blood pressure; CT: computed tomography scan; ER: emergency room; IQR: interquartile range; IVT: intravenous thrombolysis; NIHSS: National Institute of Health Stroke Scale; ODT: onset-to-door time; TIA: transient ischemic attack. Outcomes We found no significant differences for the primary or secondary outcomes between the EBW and the MBW group (Table 2). The rate of sICH was 4.4% in EBW versus 4.1% in the MBW group, clinical outcome was favorable in 60% of the EBW and 56% of the MBW group, and DNT was 33 min (IQR 24–50) in the EBW and 32 min (IQR 23–47) in the MBW group. We did find significant differences for the DNT, when the MBW group was divided into subgroups according to exact weight measurement method. The DNT was 28 min (IQR 20–40) for the MBW group with inbuilt weighing bed and 38 min (IQR 28–53) for the MBW group with a patient lift scale. Weight modality (in this case EBW versus MBW) was not significantly associated with increased risk of sICH (aOR = 1.16; 95% CI 0.83–1.62), favorable outcome (aOR = 0.99; 95% CI 0.82–1.21), or with DNT (adjusted B = 0.28; 95% CI − 1.69 to 2.25) (Table 3). We also did not find a significant association with EBW versus either of the MBW subgroups (inbuilt weighing bed and patient lift scale) with an increased risk of sICH or favorable outcome (supplementary data, Table 4). We did, however, find a significant association for the DNT. The DNT was longer in the EBW group compared to the MBW group with inbuilt weighing bed (adjusted B = 3.57; 95% CI 1.33–5.80) and the DNT was shorter in the EBW group compared to the MBW with patient scale sling (adjusted B = −3.96; 95% CI − 6.38 to −1.53) (Table 3). Table 2. Outcome measures

t association for the DNT. The DNT was longer in the EBW group compared to the MBW group with inbuilt weighing bed (adjusted B = 3.57; 95% CI 1.33–5.80) and the DNT was shorter in the EBW group compared to the MBW with patient scale sling (adjusted B = −3.96; 95% CI − 6.38 to −1.53) (Table 3). Table 2. Outcome measures Outcome Measured body weight (n = 2048) Missing data % Estimated body weight (n = 2753) Missing data % P-values sICH—n (%) 83 (4.1) 0 122 (4.4) 0 0.52 mRS 0–2 at 90 days—n (%) 599 (56.2) 48.0 920 (59.9) 44.2 0.06 DNT, min–median [IQR] 32 [23–47] 2.7 33 [24–50] 2.8 0.15 • Inbuilt weighing bed 28 [20–40] 1.6 <0.01a • Patient lift scale 38 [28–53] 3.9 <0.01b DNT: door-to-needle time; IQR: interquartile range; mRS: modified Rankin Scale; sICH: symptomatic intracranial hemorrhage. a DNT for EBW versus inbuilt weighing bed. b DNT for EBW versus patient lift scale. Table 3. Logistic and linear regression analysis for the association between weight modality (EBW versus MBW) and the outcome measures

Outcome Measured body weight (n = 2048) Missing data % Estimated body weight (n = 2753) Missing data % P-values sICH—n (%) 83 (4.1) 0 122 (4.4) 0 0.52 mRS 0–2 at 90 days—n (%) 599 (56.2) 48.0 920 (59.9) 44.2 0.06 DNT, min–median [IQR] 32 [23–47] 2.7 33 [24–50] 2.8 0.15 • Inbuilt weighing bed 28 [20–40] 1.6 <0.01a • Patient lift scale 38 [28–53] 3.9 <0.01b DNT: door-to-needle time; IQR: interquartile range; mRS: modified Rankin Scale; sICH: symptomatic intracranial hemorrhage. a DNT for EBW versus inbuilt weighing bed. b DNT for EBW versus patient lift scale. Table 3. Logistic and linear regression analysis for the association between weight modality (EBW versus MBW) and the outcome measures Logistic regression analyses Outcome OR (95% CI) aOR (95% CI)a sICH 1.09 (0.83–1.46) 1.16 (0.83–1.62) mRS 0–2 at 90 days 1.01 (0.88–1.16) 0.99 (0.82–1.21) Linear regression analysis B (95% CI) B (95% CI)b DNT in minutes • EBW versus MBW 0.06 (−1.59 to 1.71) 0.28 (−1.69 to 2.25) DNT in minutesc • EBW versus inbuilt weighing bed 4.01 (1.99–6.01) 3.57 (1.33–5.80) DNT in minutesd • EBW versus patient lift scale −4.47 (−6.58 to −2.36) −3.96 (−6.38 to −1.53) aOR: adjusted OR; B: unstandardized regression coefficient; DNT: door-to-needle time; EBW: estimated body weight; MBW: measurement of body weight; OR: odds ratio. a aOR, adjusted for age; atrial fibrillation; diabetes mellitus; hypertension; hyperlipidemia; admission NIHSS, CT in the ER, and IVT-volume. b B, adjusted for blood pressure exceeding threshold for IVT; admission NIHSS, CT in the ER, onset-to-door time, and IVT-volume.

Logistic regression analyses Outcome OR (95% CI) aOR (95% CI)a sICH 1.09 (0.83–1.46) 1.16 (0.83–1.62) mRS 0–2 at 90 days 1.01 (0.88–1.16) 0.99 (0.82–1.21) Linear regression analysis B (95% CI) B (95% CI)b DNT in minutes • EBW versus MBW 0.06 (−1.59 to 1.71) 0.28 (−1.69 to 2.25) DNT in minutesc • EBW versus inbuilt weighing bed 4.01 (1.99–6.01) 3.57 (1.33–5.80) DNT in minutesd • EBW versus patient lift scale −4.47 (−6.58 to −2.36) −3.96 (−6.38 to −1.53) aOR: adjusted OR; B: unstandardized regression coefficient; DNT: door-to-needle time; EBW: estimated body weight; MBW: measurement of body weight; OR: odds ratio. a aOR, adjusted for age; atrial fibrillation; diabetes mellitus; hypertension; hyperlipidemia; admission NIHSS, CT in the ER, and IVT-volume. b B, adjusted for blood pressure exceeding threshold for IVT; admission NIHSS, CT in the ER, onset-to-door time, and IVT-volume. c DNT in minutes for EBW versus MBW, inbuilt weighing bed. d DNT in minutes for EBW versus MBW, patient lift scale. Missing data Baseline characteristics did not show a relevant difference in patients with or without a known clinical outcome and missing outcome data were also evenly distributed between the groups. Results of the analysis of the imputed dataset were essentially the same as the results of the analysis without imputed data (supplementary data Table 5). Furthermore, post hoc sensitivity analysis excluding patients with an unknown clinical outcome yielded similar robustness of the primary analysis (supplementary Table 6).

ts of the analysis of the imputed dataset were essentially the same as the results of the analysis without imputed data (supplementary data Table 5). Furthermore, post hoc sensitivity analysis excluding patients with an unknown clinical outcome yielded similar robustness of the primary analysis (supplementary Table 6). Discussion Our findings did not demonstrate an association between weight modality and sICH rates or clinical outcome. While previous prospective studies have shown that EBW leads to dosing errors, our results showed that this does not translate into a different safety and efficacy profile of intravenous rt-PA in clinical practice. Interestingly, we found that EBW leads to a longer DNT compared to MBW using an inbuilt weighing bed, but to a shorter DNT compared to MBW using a patient lift scale. Our main results are in line with some previous studies.13,17 However, our study has a much larger study population and unlike the previous studies it concerns a multicenter study. Therefore, it is unlikely that we missed a difference in outcome related to weight modality rendering our results more generalizable to routine clinical practice.

line with some previous studies.13,17 However, our study has a much larger study population and unlike the previous studies it concerns a multicenter study. Therefore, it is unlikely that we missed a difference in outcome related to weight modality rendering our results more generalizable to routine clinical practice. In contrast, two studies showed a difference in clinical outcome related to weight modality. One retrospective mono-center study (n = 164) found that EBW led to rt-PA overdose in 13 (16%) patients. Of those 13 patients, four had an intracranial hemorrhage (however, it remained unclear whether these were symptomatic or not).12 Another prospective mono-center study (n = 128) found that EBW leads to rt-PA overdose in 52% of the patients with more sICH in the first 24 h.15 The overall sICH rate for the whole group was 7.8% in the first 24 h which is much higher than one would expect from previous studies with this sICH definition.25 This may have influenced the results limiting generalizability. A possible explanation for the high sICH rate is the predominantly Asian population in this study as Asian ethnicity is associated with increased risk of sICH.26–28 Furthermore, a follow-up brain CT scan was performed as part of standard clinical care at 24 h. Therefore, researchers could have been more prone to attribute clinical symptoms to a hemorrhage seen on these standard imaging protocols. Finally, in our study weight modality was not associated with DNT even after adjusting for factors such as IVT volume, CT availability on the ER, baseline NIHSS, and blood pressure above IVT threshold.24,29–31 Nevertheless, other unknown factors related to the DNT we could not adjust for could possibly explain this lack of an association.

y weight modality was not associated with DNT even after adjusting for factors such as IVT volume, CT availability on the ER, baseline NIHSS, and blood pressure above IVT threshold.24,29–31 Nevertheless, other unknown factors related to the DNT we could not adjust for could possibly explain this lack of an association. Somewhat surprisingly, the median DNT was shorter in the MBW group using an inbuilt weighing bed compared to the EBW group. An explanation for this could be that in practice weight estimation can require multiple steps (asking the patient or relative and estimation by the treating physician), whereas an inbuilt weighing bed scale only requires one step (transfer of the patient), which is also done in the EBW group (e.g. from ambulance stretcher to hospital bed). Of note, this difference in DNT does not affect the finding that weight modality is not associated with an increased risk of sICH or clinical outcome, since we adjusted for the DNT in these analyses.

Somewhat surprisingly, the median DNT was shorter in the MBW group using an inbuilt weighing bed compared to the EBW group. An explanation for this could be that in practice weight estimation can require multiple steps (asking the patient or relative and estimation by the treating physician), whereas an inbuilt weighing bed scale only requires one step (transfer of the patient), which is also done in the EBW group (e.g. from ambulance stretcher to hospital bed). Of note, this difference in DNT does not affect the finding that weight modality is not associated with an increased risk of sICH or clinical outcome, since we adjusted for the DNT in these analyses. Our study has several limitations. First, a cluster-randomized trial would be a more suitable design for our research question, but in practice this does not seem feasible since clinics using MBW are not likely to change this to EBW. Due to the retrospective nature of our design, extraction of (outcome) data could have led to bias. However, assessment of our primary outcome, sICH, was done according to strict definitions by two independent reviewers and sICH rates are similar to previous studies using the same definition criteria.3,6 Second, clinical outcome was missing for a substantial proportion of patients. We investigated the possible influence of missing data on our parameter estimates, by performing different methods of handling missing data in our cohort. Results of the primary analysis remained consistent after imputing missing data (supplementary data Table 5) and after post hoc sensitivity analysis (excluding patients with unknown clinical outcome), indicating that missing data were not of significant influence on our outcome parameters (supplementary data Table 6). Additionally, missing outcome data were evenly distributed between the groups and baseline patient characteristics did not show a relevant difference in patients with or without a known clinical outcome (data not shown). Of note, even when excluding patients with unknown clinical outcome our cohort still remains the largest so far investigating weight modality in IVT treated patients. As for the DNT, this is an obligatory practice parameter in all centers and is therefore not likely to be affected by retrospective assessment. Furthermore, data on actual rt-PA dosage and (measured) body weight were lacking. Although this would have enabled us to determine exactly in which patient body weight was over- or underestimated and whether this resulted in over- or under dosing rt-PA, it apparently does not translate into an increased risk of sICH or a clinically meaningful difference.

sage and (measured) body weight were lacking. Although this would have enabled us to determine exactly in which patient body weight was over- or underestimated and whether this resulted in over- or under dosing rt-PA, it apparently does not translate into an increased risk of sICH or a clinically meaningful difference. Our data indicate that EBW was not associated with increased risk of sICH (aOR = 1.16; 95% CI 0.83–1.62), therefore a possible effect of weight modality on sICH would be smaller than 1.62 with 95% certainty, independent of whether there is a difference between estimated or measure bodyweight. Finally, a limitation is that centers, with or without a certain weighing modality, could differ in local policies which could lead to a bias related to outcomes. However, all centers treat IVT patients according to the same national guidelines, including prehospital notification of potential IVT patients and since patients in each group came from at least five centers and the outcomes are evenly distributed in both groups, we consider this risk minimal.

ias related to outcomes. However, all centers treat IVT patients according to the same national guidelines, including prehospital notification of potential IVT patients and since patients in each group came from at least five centers and the outcomes are evenly distributed in both groups, we consider this risk minimal. Our study provides the largest multicenter cohort study to date assessing the association between weight modality (EBW or MBW) with sICH, clinical outcome, and DNT. We found that MBW with an inbuilt weighing bed leads to shorter DNTs compared to EBW, whereas the latter strategy leads to shorter DNTs compared to MBW with a patient lift scale. We did not find evidence that weight modality for rt-PA titration in IVT eligible patients leads to clinically relevant dosing errors, since it was not associated with an increased risk of sICH or favorable clinical outcome. Supplemental Material Supplemental material for Thrombolysis related symptomatic intracranial hemorrhage in estimated versus measured body weight Click here for additional data file. Supplemental Material for Thrombolysis related symptomatic intracranial hemorrhage in estimated versus measured body weight by T Truc My Nguyen, Stephanie IW van de Stadt, Adrien E Groot, Marieke JH Wermer, Heleen M den Hertog, Hanneke M Droste, Erik W van Zwet, Sander M van Schaik, Jonathan M Coutinho and Nyika D Kruyt: for the SOCRATES Steering Committee and Investigators in International Journal of Stroke

ated versus measured body weight by T Truc My Nguyen, Stephanie IW van de Stadt, Adrien E Groot, Marieke JH Wermer, Heleen M den Hertog, Hanneke M Droste, Erik W van Zwet, Sander M van Schaik, Jonathan M Coutinho and Nyika D Kruyt: for the SOCRATES Steering Committee and Investigators in International Journal of Stroke Acknowledgements We thank all participating hospitals and physicians for their help: Yvo Roos (Amsterdam University Medical Centre, locatie AMC), Marieke Visser (Amsterdam University Medical Centre, locatie VUmc), Nyika Kruyt (Leiden University Medical Centre), Heleen den Hertog (Isala hospital, Zwolle), Patricia Halkes (Noord West Ziekenhuisgroep, locatie Alkmaar), Lahcen Hani (Noord West Ziekenhuisgroep, locatie Den Helder), Vincent Kwa (Onze Lieve Vrouwen Gasthuis, locatie Oost), Sander van Schaik (Onze Lieve Vrouwen Gasthuis, locatie West), Willem van der Meulen (Rode Kruis Ziekenhuis), Marieke de Graaf (Medical Centre Slotervaart), Frank de Beer (Spaarne Gasthuis), Jelle de Kruijk (Tergooi Ziekenhuis), Caspar Zwetsloot (Waterlandziekenhuis), and Taco van der Ree (West Fries Gasthuis).

Oost), Sander van Schaik (Onze Lieve Vrouwen Gasthuis, locatie West), Willem van der Meulen (Rode Kruis Ziekenhuis), Marieke de Graaf (Medical Centre Slotervaart), Frank de Beer (Spaarne Gasthuis), Jelle de Kruijk (Tergooi Ziekenhuis), Caspar Zwetsloot (Waterlandziekenhuis), and Taco van der Ree (West Fries Gasthuis). Authors’ contributions TTMN: study concept and design, acquisition, analysis, and interpretation of data, statistical analysis, and drafting and revising the manuscript. SIWS: acquisition and analysis of data and revising the manuscript for scientific content. AEG: acquisition of data and revising the manuscript for scientific content. MJHW, HMH, and SMS: critical revision of the manuscript for important intellectual content. HMD: acquisition of data. EWZ: statistic analytical support, revising the manuscript for scientific content. JMC: study concept and design and critical revision of the manuscript for important intellectual content. NDK: study concept and design, interpretation of data, critical revision of the manuscript for important intellectual content, and study supervision. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) received no financial support for the research, authorship, and/or publication of this article.