A multi-site comparison of the multimodal response to an acute social stressor in the MRI environment

Abstract

Understanding the mechanisms behind (mal)adaptive stress responses is crucial for addressing stress-related mental disorders, which remain leading contributors to global disability and mortality. However, individual differences in stress responses present a challenge for single studies due to limited sample sizes. Multi-site studies can overcome this by increasing statistical power and generalizability, but it remains unclear whether even optimally harmonized procedures can ensure cross-site comparability. To investigate the impact of study site on the multimodal response to an acute social stressor, we analyzed data from the Dynamic Modeling of Resilience Observational (DynaM-OBS) Study, encompassing five study sites across Europe and Israel. By employing harmonized procedures for stress induction through the adapted ScanSTRESS-C MRI paradigm, along with consistent protocols for data acquisition and processing, we assessed the following markers: subjective stress ratings, heart rate, salivary cortisol, salivary alpha-amylase levels, and fMRI BOLD response. Bayesian inference allowed us to evaluate the evidence for and against the presence of site effects on stress markers. Results indicate successful stress induction, as evidenced by subjective, cardiac, and neural measures, though the salivary stress markers did, on average, not show a typical increase. Comparable stress responses were observed across most sites, highlighting the potential of rigorous procedural harmonization. However, the notable differences at the geographically most distant site may partially reflect variations in stressor exposure, as well as potential cultural differences. These findings highlight the importance of considering demographic and geo-cultural factors in multi-site stress research. Additionally, we emphasize the value of employing Bayesian approaches to integrate and evaluate data from diverse sources. Overall, while such studies enhance statistical power and generalizability, careful interpretation of site-specific effects is essential for advancing our understanding of stress-related mental health.

Keywords: Bayesian; FMRI; Multi-site; Resilience; ScanSTRESS; Stress.