In a study published in Nature Genetics, Cato Romero, Mats Nagel, and Sophie van der Sluis teamed up with colleagues from the Complex Trait Genetics department and the Million Veteran Program to scrutinize the genetic similarity of twelve psychiatric disorders. Besides identifying shared genes and biological substrates between pairs of disorders, they also elucidated the challenges jeopardizing the future success of cross-trait genetic research.
Often, people are not diagnosed with one, but with multiple psychiatric disorders, i.e., within the same individual, psychiatric disorders co-occur much more often than expected by chance: they are comorbid. This comorbidity could be due to different disorders sharing the same genetic risk factors. Previous research has estimated significant overlap in genetic signal between multiple psychiatric disorders, but it is still challenging to identify which genes or genetic variants, and which biological processes, comorbid disorders actually have in common. Pinpointing genetic variants and biological processes shared between psychiatric disorders is, however, essential to improve treatment of these debilitating disorders and can potentially even lead to genetically informed therapy (e.g., drug development, drug repurposing) and genetically informed adaptation of our diagnostic system. With novel methodologies and the ever-increasing, publicly available genetic datasets, we are beginning to see the potential of cross-trait genetic research.
Genetic overlap at different levels
In the paper by Romero et al., genetic overlap was studied on different biological levels; from genetic variants (SNPs), genes, and genomic regions, to overlapping biological processes, tissue- and cell-types. Considerable overlap was observed, but mostly between pairs of psychiatric disorders. The largest overlap was observed between schizophrenia and bipolar disorder, which shared 117 genetic variants, 28 associated genes, and genetic signal in 9 different genomic regions. Only genomic regions related to evolutionary conservation were associated to most (9 out of 12) psychiatric disorders, which suggests genetic variation in essential biological processes as a common feature of psychiatric disorders.
Lessons for future cross-trait genetic research
While the degree of genetic overlap is often taken as an important determinant of the potential success of cross-trait genetic research, it is not the only deciding factor. Romero et al. showed that the 12 psychiatric disorders vary considerably with respect to statistical power (e.g., how many patients donated genetic data), polygenicity (i.e., the proportion of the genome that is causally involved in the disorder), and discoverability (i.e., the effect sizes and penetrance of these causally associated genetic variants). Variation between traits on these three dimensions crucially determines the potential success of future cross-trait genetic research. As more and more genetic data is being collected and shared, it is a matter of time before circumstances for genetic comorbidity research improve. Until then, cross-trait genetic research on psychiatric disorders is a promising area of investigation that ultimately has the potential to improve treatments of cooccurring psychiatric conditions, improve diagnostic accuracy, and inform the next generation of diagnostic nosology.
Figure 1. Schematic overview of all annotations analyses conducted on 12 psychiatric disorders.
This figure summarises all the ways genetic association results for the 12 individual psychiatric disorders were annotated and the extent to which annotations shared between psychiatric disorders were identified. First, overlap in SNPs, mapped genes, and genomic loci was examined. Subsequently, gene set and gene property analyses were conducted, examining, for each of the 12 psychiatric disorders, whether the genetic signal was enriched in 9,508 gene sets, 565 cell types and 53 tissue types. Partitioned LDSC was used to test for enrichment of 61 functional genomic categories. All pictures used in this figure are illustrations and do not reflect the actual pathways identified in the analysis.
