Investigating Bidirectional Causal Relationships Between Imaging‐Derived Brain Phenotypes and Sedative‐Hypnotic Use Disorder: A Mendelian Randomization Study

ABSTRACT Despite sedative‐hypnotic use disorder (SHUD) being a major public health priority due to its high abuse liability, the underlying causal neurobiological mechanisms of individual susceptibility remain largely unknown. We conducted a large‐scale, bidirectional two‐sample Mendelian randomization (MR) study utilizing genome‐wide association study (GWAS) summary statistics for 3935 brain imaging‐derived phenotypes (IDPs) from the UK Biobank ( N = 33 000 participants) and SHUD data from the FinnGen R10 release ( N = 2258 cases). Causal effects were primarily estimated using the inverse‐variance weighted (IVW) method, which was supplemented by a suite of sensitivity analyses including MR‐Egger, weighted median and MR‐PRESSO. Multiple testing was controlled via Benjamini–Hochberg false discovery rate (FDR) correction (). Forward MR identified 34 brain IDPs with significant causal associations with SHUD risk. Specifically, increased susceptibility was linked to structural metrics in the temporal and limbic regions, including higher cortical thickness in the left transverse temporal sulcus (OR = 1.35, 95% CI 1.09–1.67, FDR = 0.04) and right middle temporal gyrus (OR = 1.30, 95% CI 1.07–1.58, FDR = 0.04), as well as elevated white‐to‐grey matter intensity contrast in the right insula (OR = 1.31, 95% CI 1.13–1.51, FDR = 0.02) and left parahippocampal gyrus (OR = 1.29, 95% CI 1.08–1.54, FDR = 0.04). Conversely, protective effects were observed for brainstem volume (OR = 0.74, 95% CI 0.61–0.91, FDR = 0.04) and white matter integrity in the left cingulum (OR = 0.73, 95% CI 0.57–0.92, FDR = 0.04). Functional connectivity analyses revealed that higher node activity in the salience network (OR = 0.70, 95% CI 0.53–0.92, FDR = 0.04) and strengthened connectivity between the visual and somatomotor networks (OR range: 0.54–0.74) predicted reduced risk. In contrast, specific edges within the default mode network and visual network were positively associated with SHUD susceptibility (OR = 1.56 and 1.48, respectively). Reverse MR analysis provided limited evidence for a causal effect of genetically predicted SHUD on right middle temporal gyrus thickness. Although a nominal association was observed (OR = 1.03, 95% CI 1.00–1.05, p = 0.02), this finding did not survive multiple testing correction (FDR = 0.56) and lacked robustness across sensitivity models. This research establishes a causal link between SHUD susceptibility and specific cortical morphologies, subcortical volumes and functional network topographies, emphasizing the pivotal roles of the temporal–limbic and salience systems. These findings provide novel mechanistic insights into the neurobiological foundations of SHUD while identifying potential imaging biomarkers for risk stratification and the advancement of therapeutic targets.