A meta-analysis of efficacy in pre-clinical human stem cell therapies for traumatic brain injury

Abstract

ObjectivesEvaluate the preclinical evidence for human cell therapies for the treatment of traumatic brain injury (TBI), determine behavioral effect sizes for modified and non-modified cells, and identify variables that correlate with greater effect sizes. MethodsA literature search identified 58 animal studies of TBI using human stem cells. Each study received a Quality Index (QI) score based on existing guidelines. Effect sizes for cell therapies were determined for the most common behavioral endpoints: Morris Water Maze (MWM) latency/correct quadrant, and modified Neurological Severity Score (mNSS). Results50 studies reported significant behavioral and/or histological improvement. The mean effect size for MWM latency was −1.08 for non-modified cells and −3.35 for modified cells. The mean effect size for MWM percent time in the correct quadrant was 1.66 for non-modified cells and 4.36 for modified cells. The mean effect size on the mNSS was −1.56 for non-modified cells and −4.46 for modified cells. No significant associations were found between methodological variables and effect sizes other than route of administration, where intra-lesional delivery resulted in larger effect sizes than i.v. or ventricular delivery. Studies with higher QI had smaller effect sizes; studies with larger effect sizes had greater standard errors. QI was not associated with journal impact factor. ConclusionsAlthough human cell therapy studies report improved behavioral outcomes in the majority of preclinical literature, the methods are too heterogeneous to facilitate direct comparisons and bias was detected. Replication and standardization are needed to identify procedural variables to yield the best results. We encourage the use of quality criteria and rigor for future studies of human cell therapy in animal models of TBI.