Animal Models of Stroke Versus Clinical Stroke: Comparison of Infarct Size, Cause, Location, Study Design, and Efficacy of Experimental Therapies

A quantitative and qualitative comparison of contemporary neuroprotection and thrombolytic stroke trials and their preclinical animal counterparts has been undertaken, with meta-analysis (DerSimonian and Laird, 1986) used to evaluate imaging and histological outcomes.Results from 35 clinical trials including 5,532 patients were compared with data from 3,145 pre-clinical acute-stroke experiments in 45,476 animals. While clinical trials tended to be of higher methodological quality and have larger sample sizes than animal experiments (71 patients vs. 7 animals per group), both were similarly underpowered owing to the greater variability in human stroke (average standard deviation of mean in humans 99% v 30% in animals). Proportionally, animal infarcts were almost four times larger than human infarcts in untreated control groups (27% v 8% of the hemisphere) although there was considerable variability in size owing to comorbidities and stroke type. Eighty-six percent of animal studies and 54% of clinical trials reported smaller infarcts in groups receiving treatment, with 41% of clinical trials reporting an improvement in the pre-specified hypothesis. Animal experiments were not effective in predicting individual trial results, nor the level of neuroprotection, however, there was a fair agreement between the direction of the animal and clinical outcomes when looking at the overall direction of drug outcome. As a drug screening tool, experimental stroke studies need refinement. Rational frameworks for translational research will help.