The Virtual International Stroke Trials Archive (VISTA): Promulgation of a Clinical Trial Resource

Abstract Chapter 1 provides an introduction to stroke including its current prevalence both nationally and globally, aetiology, global importance and social & financial burden. We also describe here current acute stroke management practices, the role of clinical trials in the development of therapies, the richness of data within clinical trials and changes in regulatory thinking regarding data access. We provide recommendations for the use of trial data for novel exploratory investigations of clinical trial design and epidemiological studies. In Chapter 2 we describe the establishment of the Virtual International Stroke Trials Archive (VISTA) to address the need for reliable data on which to plan future clinical trials. This chapter details the methodology and logistics of establishing the resource, including details of regulatory policy for data collection and use, establishment of a Steering Committee and development of a constitution to safeguard data access and use. As of June 2008, VISTA contains 28 acute stroke clinical trials and one acute stroke registry. We collated data on over 27,500 patients with either ischaemic or haemorrhagic stroke. Patient age ranges from 18 to 103 years and outcome measures include Barthel Index, Scandinavian Stroke Scale, National Institutes of Health Stroke Scale, Orgogozo Scale, and modified Rankin Scale. Medical history and onset to treatment times are readily available and computed tomography (CT) lesion data are available for selected trials. We discuss the establishment and potential uses of this resource in the context of existing stroke resources. Chapter 3 demonstrates how we utilised VISTA to investigate natural history patterns in acute stroke. There are prominent differences in stroke incidence and outcome across different geographical locations; these are not confined to the Eastern- Western axis. We aimed to examine whether there were any differences in index stroke severity, stroke risk factors, and stroke outcome between geographical locations, after adjusting for case-mix. We found that patients who were enrolled in the USA and Canada had the worst index strokes, whilst patients enrolled in Austria and Switzerland had the mildest index stroke, and better functional (p=0.023) and neurological outcome (p=0.034) at 90 days. 90 –day survival was greater in patients who were enrolled in Spain and Portugal (p<0.0001). Chapter 4 demonstrates the use of VISTA to inform stroke clinical trial design by examining the impact of early follow up on adverse event and functional outcome profiles. We aimed to assess the contribution of adverse complications unrelated to stroke, to 30 and 90- day functional outcome. If fewer ‘stroke-unrelated’ adverse events were seen at later time points, and if the absence of these events appeared to influence functional outcome, then further investigation into shortening the follow up period of clinical trials with a view to minimizing complications may be warranted. We identified idiopathic post-stroke complications (deemed to be ‘stroke- unrelated’) but their absence did not beneficially alter outcome at either 30 days (p<0.0001, adjusted OR for good outcome =0.47, 95% CI [0.26, 0.67]), or 90 days (p=0.002, adjusted OR for good outcome =0.38, 95% CI [0.14, 0.61]). We concluded that shortening the follow up period with the aim of minimizing ‘stroke-unrelated’ complications did not benefit functional outcome, however further investigation is required. Chapter 5 illustrates the use of VISTA to investigate the natural history of complications after intracerebral haemorrhage (ICH). Treatments available for ICH remain limited. The use of haemostatic agents to promote local coagulation has had no significant benefit on outcome. However promising results from a subgroup analysis of patients from the FAST trial has raised the possibility of treatment with recombinant factor VIIa (rFVIIa) in patients with ICH. We sought to document the natural history of complications after ICH in order to inform safety in future trials of haemostatic agents for ICH. We found that the risk of thromboembolic complications after ICH was low (4 events affecting 2% of patients). The absence of these thromboembolic complications did not significantly affect the attainment of good functional outcome (p>0.05). The occurrence of haemorrhagic expansion was common, affecting 14% of patients, and significantly influenced attainment of good functional outcome at 90 days (p= p<0.0001, adjusted odds ratio for good functional outcome=21.9, 95% confidence interval [5.5, 88.3]). Although infection occurred in 11% of patients, this did not significantly influence attainment of good functional outcome at 90 days (p=0.8). The complications encountered in this investigation and their time to onset will serve to inform prophylaxis in future ICH clinical trials. Chapter 6 describes the processes involved in drug development from phase I, first- in – man studies to phase III efficacy trials and identifies a key area in the drug development process where use of VISTA as a historical comparator resource could be of benefit: phase II studies. We detail here the types of conventional comparator groups available for use in a phase II investigation, advantages and disadvantages of using each of these comparator groups, the potential for use of historical comparators in some scenarios where use of conventional comparator groups is infeasible, and possible solutions to address the limitations associated with use of historical comparators. Chapter 7 illustrates the use of VISTA as a resource for historical comparators in the context of an acute stroke device trial conducted by a small company with limited resources. BrainsGate, the manufacturers of the NeuroPath™ Device for treatment of ischaemic stroke, sought to collaborate with the VISTA group to examine initial efficacy of their device against outcomes derived from VISTA historical comparators. We discuss the example of this device in early phase testing, where VISTA was primed for use as a resource for historical comparators. We also describe the limitations associated with the use of historical comparators, how these limitations could be overcome in practice through use of matched patients, implementation of strict eligibility criteria and use of similar follow up periods and stroke scales, as well as the measures taken to ensure the validity of results. Chapter 8 describes a collaboration with the DESTINY trial group to investigate stroke outcomes after malignant middle cerebral artery occlusion (mMCAO). The DESTINY trial examined the impact of decompressive hemicraniectomy on outcome after mMCAO, compared with randomised controls. We compared the outcomes of operated patients from the DESTINY trial with historical comparators from VISTA to determine whether the findings could be replicated and if historical comparators could be used as an alternative in a situation where a randomised controlled trial (RCT) is infeasible or unethical. We found that fewer patients in the VISTA comparator group achieved a good functional outcome by mRS at final follow up (19%), when compared with the DESTINY surgical group (47%, Chi- Square test p=0.04). This difference persisted after adjusting for baseline NIHSS (logistic regression p=0.04). Analysis of Barthel Index at final follow up revealed no significant difference between the two groups and we also found no difference in 6 month survival rates between the surgical and VISTA comparator groups (Cox Proportional Hazards model p>0.05). We concluded that for effective replication of results, the database from which historical comparators are to be drawn should cover a similar or broader spectrum of patient prognostic factors. Chapter 9 discusses the implications of the investigations described in this thesis, outlines the scope for expanding the resource and proposes areas for future research.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.