Improving anxiety research: novel approach to reveal trait anxiety through summary measures of multiple states

Abstract The reliability and validity of preclinical anxiety testing is essential for translational research outcomes. However, widely-used anxiety tests lack inter-test correlations and have repeatability difficulties that need clarification. Translational research seeks to capture individual variability and advance personalized medicine, which demands trait-like features reflecting the underlying neural characteristics. Here, we show that detailed sampling across multiple time-points and contexts covers various states of the individuals, which is needed to reliably capture trait anxiety (TA). We also propose a validated, optimized test battery to reveal TA in rats and mice. Instead of developing novel tests, we combined widely-used tests (elevated plus-maze, open field, light-dark box) to clarify current inter-test and repeatability issues and provide instantly applicable adjustments for better predictive validity. We repeated tests three times to capture multiple anxiety states in various paradigms that we combined to generate summary measures (SuMs). Using correlations and machine learning, we found that our approach resolves correlation issues and provides better predictions for subsequent outcomes under anxiogenic conditions or fear conditioning. Moreover, SuMs were more sensitive to detect anxiety differences in an etiological model of social isolation. Finally, we tested our sampling method’s efficiency in discovering anxiety-related molecular pathways through RNA sequencing of the medial prefrontal cortex. We identified four times more molecular correlates of anxiety using SuMs, which pointed out functional gene clusters that had not emerged using single measurements applied by most studies. Overall, temporally stable SuMs are necessary to capture trait-like anxiety in rodents, providing better predictions for potential therapeutic targets.