Systematic review and meta-analysis of animal models of acute ischaemic stroke

Abstract

Ischaemic stroke is responsible for substantial death and disability and creates a huge financial burden for healthcare budgets worldwide. At present there are few effective treatments for acute stroke and these are urgently required. Increased understanding of the ischaemic cascade has generated interest in neuroprotection for focal cerebral ischaemia. However, treatment effects observed in of over 500 interventions in animal models have yet to be translated to the clinic. Systematic review and meta-analysis allows unbiased identification of all relevant data for a given intervention, gives a clearer view of its true efficacy and the limitations to its therapeutic potential. Understanding the reasons for this bench-to-bedside failure and providing quantitative explanations may help to address these discrepancies. Random effects weighted mean difference meta-analysis of six interventions (tirilazad, tPA, NXY-059, Hypothermia, Piracetam and IL1-RA) reported study quality to be consistently low. In some instances, potential sources of bias were associated with overestimations of efficacy. Likewise, clinical trials have tested interventions in conditions where efficacy was not observed in animals. Cumulative meta-analysis suggests that for tPA the estimate of efficacy is stable after the inclusion of data from 1500 animals; hypothermia and FK506 are the only other interventions to have been tested in at least 1500 animals. Meta-regression suggests biological rather methodical factors are better predictors of outcome; a major limitation of these data is the impact of publication bias, and this work suggests effect sizes from met-analyses are inflated by about 31% because 16% of studies remain unpublished. The systematic review and meta-analysis of hypothermia was used to plan experiments investigating the possible impact of pethidine, a drug used to prevent shivering. This in vivo experiment, in which potential sources of bias were minimised, suggests that pethidine does not influence the observed efficacy of hypothermia in an animal model of ischaemic stroke. This thesis reports that animal studies of ischaemic stroke are often not conducted with sufficient rigour. Both minimising potential sources of bias in individual experiments and using meta-analysis to summarise data from a number of experiments may be helpful in improving the translation of neuroprotective efficacy in ischaemic stroke.