Psychiatric disorders: While many factors contribute to risk, epidemiological evidence suggests that the genetic contribution carries the highest risk burden; the increased rates are nonspecific

Risk in Relatives, Heritability, SNP-Based Heritability, and Genetic Correlations in Psychiatric Disorders: A Review. Bart M.L. Baselmans et al. Biological Psychiatry, Volume 89, Issue 1, January 1 2021, Pages 11-19. https://doi.org/10.1016/j.biopsych.2020.05.034

Rolf Degen's take: https://twitter.com/DegenRolf/status/1333344147185541120

Abstract: The genetic contribution to psychiatric disorders is observed through the increased rates of disorders in the relatives of those diagnosed with disorders. These increased rates are observed to be nonspecific; for example, children of those with schizophrenia have increased rates of schizophrenia but also a broad range of other psychiatric diagnoses. While many factors contribute to risk, epidemiological evidence suggests that the genetic contribution carries the highest risk burden. The patterns of inheritance are consistent with a polygenic architecture of many contributing risk loci. The genetic studies of the past decade have provided empirical evidence identifying thousands of DNA variants associated with psychiatric disorders. Here, we describe how these latest results are consistent with observations from epidemiology. We provide an R tool (CHARRGe) to calculate genetic parameters from epidemiological parameters and vice versa. We discuss how the single nucleotide polymorphism–based estimates of heritability and genetic correlation relate to those estimated from family records.

Keywords: Family register dataGenetic correlationGWASHeritabilityPsychiatric geneticsRisk in relatives

Conclusions

In this capstone narrative, we bring together the methods and results that summarize the genetic contribution to psychiatric disorders and the genetic relationship between them. We note that we use the common assumption that psychiatric disorder diagnosis definitions are underpinned by a consistent polygenic biology. If this is not true—for example, if a single clinical diagnosis is allocated to one or more independent or correlated biological diseases—then further thought is needed to interpret the estimates of heritability and genetic correlation. Such a scenario could explain (41), in part, the large difference between heritability and SNP-based heritability (Figure 1) in addition to contributions from rare variants and low LD between genotyped and causal variants. Previously, we concluded that only with large GWAS sample sizes and extensive clinical data (40,41) would we have the information needed to examine this interesting question. Despite this caveat, multiple results from GWAS data confirm that individuals allocated a specific diagnosis are genetically more similar, on average, than those allocated other diagnoses (i.e., heritabilities of individual disorders are greater than co-heritabilities between disorders) (Figure S2 in Supplement 2).

Understanding the genetic contribution to common disease is a foundation for many other research directions. It is outside the scope of this review to focus on the utility of the estimates of heritability and genetic correlation in detail. Estimates of SNP-based heritability help to guide whether efforts to increase GWAS sample sizes should continue, as they provide an upper limit on the combined effects of individual associated loci. Estimates of heritability and SNP-based heritability provide guidelines of maximum future accuracy of risk prediction applied to people whose disease status is not yet known. Genetic correlations can be used to determine how much the accuracy of the risk prediction can be improved by drawing on information from correlated traits, which perhaps are available in much larger samples than for the primary disorder itself (68). Here, we have focused on genetic correlations between psychiatric disorders, an approach that is likely to reflect pleiotropy (same causal variants affecting more than one disorder). However, genetic correlations can also be estimated between psychiatric disorders and other common diseases, or between psychiatric disorders and traits measurable in the population (such as educational attainment or smoking status), and these estimates could reflect causal relationships, which have been long-discussed in the psychiatric epidemiology literature (69). In the past 5 years, results from GWASs have allowed causal relationships using putative exposure traits and psychiatric disorders to be explored, as well as those between psychiatric disorders and subsequent metabolic disease, using the Mendelian randomization approach. The application of Mendelian randomization to psychiatric disorders has been discussed elsewhere (70) and is an exciting tool in psychiatry (as long as studies are well powered) to investigate putative causal relationships that are impossible or unethical to address through clinical trials. As an example, we recently showed that although there is considerable pleiotropy between genetic variants for vitamin D and psychiatric disorders, there is no evidence of a causal relationship (71). Such analyses contribute hard data to a long discussion in psychiatric epidemiology (72,73). Finally, we hope that our Supplementary materials, including Rmarkdown script and CHARRGe Shiny application (https://shiny.cnsgenomics.com/CHARRGe/), are useful to others both in research and as teaching and learning aids.