Determining the source of variability within experimental stroke models

In the UK, cerebral stroke is the 3rd cause of death, the leading cause of severe adult disability in adults and costs the NHS an estimated £3 billion per year. The generation of novel and rational therapeutics is essential. Thus, every year thousands of animals are used experimentally to model stroke and investigate potential therapies. Mice are commonly used in experimental studies aimed at modelling stroke not least because they are useful for genetic manipulation studies. However, studies using eperimental stroke models tend to show a large variabiity in the data produced. This usually results in researchers using increased number of animals per study in order to try and detect a significant difference. We propose that a large portion of this variability may be due to differences in the anatomy of blood vessels in the brain. Such variability has been well documented in anatomical studies and we propose here the introduction of a non-invasive screening process whereby those animal with the highest degree of variability can be excluded from studies prior to them undergoing experimental stroke. Thus, this should reduce the variability in data obtained from using experimetnal stroke models and therefore reduce the number of animals undergoing such procedures. In addition, we intend to deternine with a variation in the established protocol for inducing experimental stroke can also result in a reduction in the variability of data and therefore reduce the number of animals used in such experiments.

Efficacy of stroke interventions are tested using rodent middle cerebral artery occlusion (MCAO) models of focal brain ischemia but these have a poor record of translating into clinically
effective treatments. The majority of experimental stroke studies use lesion volume as the primary outcome measure. However, variability in MCAO lesion volume is high with
significant differences generally tested using parametric analyses requiring a normal data distribution. Data from my laboratory over the last 10 years (and confirmed by
others) shows that MCAO lesion volume data is not normally distributed and results in a bimodal distribution of small (purely striatal) and large (striato-cortical) lesions. We aim to determine (i) whether variability in MCAO lesion data is due to differences in cerebrovascular anatomy and, (ii) whether refinement of the surgical technique can improve outcome reproducibility. Rodents show a large variability in the Circle of Willis and studies have shown this to produce variability in lesion volume thus, requiring increased animal numbers to achieve statistical significance. However, traditional methods to assess the Circle of Willis anatomy are conducted post-mortem following the induction of MCAO. Firstly, we will assess, using MRI, whether the reported variability in outcome following MCAO is dependent upon the anatomy of the Circle of Willis and if so, can we develop a screening technique to identify 'suitable' mice to undergo MCAO and therefore reduce the number of animals undergoing MCAO. In addition, we will assess whether a refinement in the surgical approach used to permit reperfusion followng MCAO can be applied in mice. This will allow reperfusion to occur irrespective of the completeness of the Circle of Willis which is currently the main limitation in determining the effectiveness of reperfusion. This should reduce the variability in reperfusion associated experimental stroke models and reduce the number of animals used.

Since 2005, approximately 4,800 research papers (Web of Science, January 2014) have reported original research findings involving the induction of experimental stroke in mice. Mice are the most common species exposed to experimental stroke due to their genetic utility. As a conservative estimate, each publication contains data from ~30 mice, thus, since 2005 at least 14,400 mice have undergone MCAO. Of course, this does not take into account those experiments taking place where data is not publised thus, 14,400 is probably a gross underestimation. The first objective, if successful, will help determine the cause of the inherent variabiity associated with these models. If our hypothesis is correct and this variability is due to cerebrovascular anatomy then we will propose a screening technique for mice undergoing experimental stroke. The screening process will involve mice intended for experimental stroke to undergo a brief MRI procedure in order to obtain angiography data and only those animals deemed suitable will then be permitted to undergo experimental stroke. Thus, this will reduce the variability within groups and therefore reduce the number animals undergoing experimental stroke which has a substantial impact on animal welfare. In terms of the practical utility of such screening it is worth noting that all the major experimental stroke labs in the UK (i.e. Mancheser, Glasgow, Oxford, UCL, Nottingham, Leicester) have access to on-site MRI facilities and could easily adopt such a technique leading to a significant reduction in the number of mice undergoing experimental stroke in the UK. Of those 4,8000 research papes mentioned above approximately half of those use a transient model which is dependent upon reperfusion. Current surgical techniques used in mice are reliant upon the cerebrovascular anatomy in order to achieve reperfusion. Variations in reperfusion increase the variability of outcome and therefore increased numbers of animals are used to achieve significance. Thus, our 2nd objective will evalaute a surgical refinement which should reduce the variability in reperfusion and therefore reduce the number of animals used in MCAO studies. In addition, if consistent patterns of reperfusion are obtained then it is likely that animals undergoing experimental stroke will recover better from the procedure and the impact upon animal welfare will be reduced. For short periods of occlusion this may result in a reduction in the severity banding of the procedure. We estimate that each objective can reduce the number of animals undergoing experimental group by 25-30%. This is based on anatomical evidence which shows that 25-30% mice have a high degree of heterogeneity in their cerebrovascular anatomy and therefore if those animals can be removed from studies (via the screening process) or reperfusion established independently of the cerebrovascular anatomy then this should result in such a reduction.