Abstract
Despite emerging evidence of altered brain region volumes and functional networks in major depression (MD), causal relationships remain unexplored. Here, using genetic variants related to human brain morphometry (n = 36778) and functional networks (n = 47276), bidirectional two-sample and multivariable Mendelian randomization (MR) is performed to systematically investigate causal relationships between 83 brain-wide volumes, 191 resting-state functional magnetic resonance imaging phenotypes, and MD to identify core brain region volumes and functional networks associated with MD onset. Unlike other severe psychiatric disorders, reductions in left and right ventral diencephalon and thalamus volumes, together with decreased functional connectivity between the precuneus, cuneus, or cingulate gyri and frontal regions within the triple network (default mode network, central executive network, and salience network) uniquely predict MD onset. Using the identified markers as mediators in two-step MR recommends quitting smoking, moderately increasing alcohol frequency without developing alcohol use disorder, engaging in walking for leisure, and participating in exercises (e.g., swimming, cycling, keep fit, bowling) to positively influence imaging phenotypes and reduce MD risk. These findings enhance the understanding of predictive outcomes in brain imaging and support proactive prevention of severe mental health disorders through behavioral modifications.