Neuropsychiatric disorders exhibit substantial heritability, supported by twin and family studies. Genome-wide association studies (GWAS) have revealed that common genetic variants contribute to this heritability, and that a portion of this heritability is shared across different neuropsychiatric disorders. This genetic overlap may explain the co-occurrence of these disorders. Structural magnetic resonance imaging (sMRI) phenotypes, such as subcortical brain volumes and cortical thickness/surface area, dynamically change throughout life and have heritability ranging from 5% to 42%. Prior ENIGMA studies have identified significant case-control differences in sMRI phenotypes for several neuropsychiatric disorders. This study aimed to quantify the similarity in sMRI phenotypes across these disorders and investigate the extent to which these similarities are attributable to shared genetic architectures.

The introduction cites numerous studies demonstrating the heritability of neuropsychiatric disorders and the genetic overlap between them as shown by GWAS. The authors also reference ENIGMA studies that have previously characterized MRI-derived phenotypes and their heritability, as well as case-control differences in sMRI phenotypes for specific disorders (ADHD, ASD, BD, epilepsy, MDD, OCD, and SCZ). This review sets the stage for the current study by highlighting the existing knowledge of shared genetic influences and brain structural abnormalities in these disorders, laying the groundwork for investigating the relationship between the two.

Summary statistics from twelve multisite ENIGMA analyses were collected for seven neuropsychiatric disorders: ADHD, ASD, BD, epilepsy, MDD, OCD, and SCZ. These data included covariate-adjusted Cohen's d standardized mean differences (SMDs) for 41 brain regions (7 subcortical and 34 cortical). Publicly available summary statistics from GWAS (primarily from the Psychiatric Genomics Consortium (PGC)) were also used. Genetic correlations (r) between pairs of disorders were quantified using LD-score regression. Pairwise Spearman's rank correlations were computed between the Cohen's d SMDs for each pair of disorders to assess the similarity in sMRI phenotypes. Pearson's correlation was used to assess the relationship between genetic correlations and sMRI phenotype correlations. A permutation framework was used to generate a null distribution of sMRI phenotype correlations, although limitations prevented a fully robust p-value calculation due to sample overlap. Leave-one-out analyses were also performed. Cochran's Q test and binomial sign tests were employed to assess heterogeneity and directionality of effects, respectively.

Significant correlations in sMRI phenotypes were observed between several pairs of disorders. The strongest positive correlation was between SCZ and BD (r=0.81, p<1.3×10⁻¹⁸). Other significant correlations included BD and MDD, OCD and SCZ, MDD and SCZ, and several negative correlations, though some did not survive multiple testing correction. Importantly, the cross-disorder correlations in sMRI phenotypes were positively correlated with the corresponding genetic correlations from GWAS (Spearman's ρ=0.44, p=0.049). This correlation remained consistent in leave-one-out analyses, except when excluding the SCZ/BD pair. Analysis of case-control differences revealed that the most prominent reductions in brain regions were observed in SCZ, epilepsy, and BD, while MDD showed the smallest changes.

The study's key finding—the substantial correlation between sMRI phenotype correlations and genetic correlations across neuropsychiatric disorders—suggests that shared genetic architecture contributes significantly to similar brain structural alterations observed in these conditions. This strengthens the hypothesis of a shared genetic etiology underlying various neuropsychiatric disorders. The findings are consistent with recent research showing shared brain structural abnormalities across disorders. The observed heterogeneity in effect sizes across disorders and brain regions highlights the complexity of these relationships and indicates the need for further investigation to clarify specific genetic variants and biological mechanisms involved.

This study provides strong evidence for substantial similarities in sMRI phenotypes across multiple neuropsychiatric disorders, and importantly, links these similarities to shared genetic factors. The positive correlation between sMRI phenotype correlations and genetic correlations highlights the significant contribution of shared genetic architecture to the observed brain structural alterations. Further research should focus on identifying specific genetic variants and pathways that underlie these shared patterns of brain abnormalities to better understand the etiology and potential for treatment of neuropsychiatric disorders.

The study's reliance on summary statistics from existing ENIGMA studies limited the ability to fully adjust for sample overlap and spatial autocorrelation in the brain images. The absence of individual-level data also prevented more sophisticated analyses to address these limitations. Additionally, the limited availability of fully complete GWAS data for MDD and epilepsy may also have influenced the findings.