ABSTRACT Background Observational studies link insomnia to hippocampal atrophy, yet causal inference remains limited by confounding and reverse causality. While Mendelian randomization (MR) studies have explored sleep‐brain associations, bidirectional causality between insomnia and hippocampal subfield volumes remains unaddressed. Methods We conducted bidirectional two‐sample MR using genome‐wide association study (GWAS) summary data. Genetic instruments for insomnia ( N = 462,341) and eight hippocampal volume imaging‐derived phenotypes (IDPs; N ≈ 31,968–33,219) were selected ( p 10 ensuring robustness. Primary analyses employed inverse‐variance weighted regression, supplemented by sensitivity analyses (MR‐Egger, weighted median, MR‐PRESSO) to assess pleiotropy and heterogeneity. Multiple testing was controlled using Benjamini–Hochberg FDR. Results Forward MR revealed inverse associations between genetically proxied insomnia and six of eight hippocampal phenotypes: T1 left hippocampus volume (OR = 0.62, 95% CI = 0.43–0.89; P FDR = 0.014), left hippocampus gray matter volume (OR = 0.63, 95% CI = 0.47–0.84; P FDR = 0.008), right hippocampus gray matter volume (OR = 0.69, 95% CI = 0.49–0.98; P FDR = 0.035), left and right hippocampal volumes from automated subcortical segmentation (ORs = 0.60, 95% CI = 0.42–0.86, P FDR = 0.010; and 0.71, 0.51–0.99, P FDR = 0.037), and right whole‐hippocampus volume from subfield segmentation (OR = 0.61, 95% CI = 0.43–0.86; P FDR = 0.010). Sensitivity analyses supported robustness; reverse MR detected no evidence of effects of hippocampal volumes on insomnia ( P FDR = 1). Conclusions Genetically proxied insomnia shows strong genetic evidence consistent with a causal reduction in hippocampal volumes. Clinically, early intervention in insomnia, particularly through cognitive‐behavioral therapy for insomnia, may help preserve hippocampal health. This study highlights the importance of addressing insomnia as a public health priority. Future research should investigate the underlying biological mechanisms and incorporate more diverse cohorts to enhance the generalizability of these findings.
Han et al. (Sun,) studied this question.