The development of a novel composite score to characterize effect size of behavior and histopathology changes after a repetitive mild traumatic brain injury

• Main Author: Conley, Ashley
• Other Authors: Franzbalu, Carl
• Language: en_US
• Published: 2019
• Subjects: Medicine; Concussion; rmTBI; Sports medicine; Traumatic brain injury
• Online Access: https://hdl.handle.net/2144/36169

In this paper, we investigate the potential for the development of a composite score investigating population-level phenotype changes in a mouse model of traumatic brain injury. Traumatic brain injuries (TBI) are a growing concern in the United States because the number of individuals impacted by TBI and associated symptoms is increasing, leading to a growing demand for research both in the clinical and preclinical setting. However, preclinical TBI modeling is complicated by the lack of inter and intra lab consistency in the assessment of behavioral and pathologic outcomes. Indeed, it remains unclear which behavior assessments are most useful in evaluating the effects of preclinical TBI. To investigate the relative contribution of various behavior tests in the assessment of preclinical TBI, three statistical models (simple linear regression, pairwise correlation, and factor analysis) were conducted on behavioral data from the Mannix-Meehan lab at Boston Children’s Hospital in Boston, Massachusetts, U.S.A. from 2012-2018. In this paper, a composite metric was created from the computation analysis of the three statistical methods. The score revealed MWM and EPM as the most potent behavioral tests. The Open Field and Rotarod test had a small impact on the outcome, but only in one of the three statistical models assessed. Thus, to effectively analyze treatment efficiencies, injury severity and long-term impairments, MWM and EPM are the best behavioral test for a mouse model. Furthermore, this method of analysis of entire populations of mice allows for more subtle phenotypic changes resultant from injury models to be revealed, and the generalizability of this model lends to widespread use.