Psychiatric disorders are highly heterogeneous. It is clinically valuable to distinguish psychiatric disorders by the presence or absence of a specific comorbid condition.

We employed a novel algorithm (CombGWAS) to decipher the genetic basis of psychiatric disorder combinations using genome-wide association studies summary statistics. We focused on comorbidities and combinations of diseases, such as schizophrenia (SCZ) with and without depression, which can be considered as two ‘subtypes’ of SCZ. We also studied psychiatric disorders comorbid with obesity as disease subtypes.

We compared the genetic architectures of psychiatric disorders with and without specific comorbidities, identifying both shared and unique susceptibility genes/variants across 8 subtype pairs (16 entities). Despite high genetic correlations between subtypes, most subtype pairs exhibited distinct genetic correlations with the same cardiovascular disease (CVD). Some pairs even displayed opposite genetic correlations, especially those involving obesity. For instance, the genetic correlation (rg) between SCZ with obesity and type 2 diabetes (T2DM) was 0.248 (p = 4.42E−28), while the rg between SCZ without obesity and T2DM was −0.154 (p = 6.79E−12). Mendelian randomization analyses revealed that comorbid psychiatric disorders often have stronger causal effects on cardiovascular risks compared to single disorders, but the effects vary across psychiatric subtypes. Notably, obese and nonobese major depressive disorder/SCZ showed opposite causal effects on the risks of T2DM.

Our study provides novel insights into the genetic basis of psychiatric disorder heterogeneity, revealing unique genetic signatures across various disorder combinations. Notably, comorbid psychiatric disorders often showed different causal relationships with CVD compared to single disorders.