Polygenic Risk Scores and Genetics in Psychiatry.

Abstract

Back to table of contents Next article Editor’s NoteFull AccessPolygenic Risk Scores and Genetics in PsychiatryNed H. Kalin, M.D.Ned H. KalinSearch for more papers by this author, M.D.Published Online:1 Nov 2022https://doi.org/10.1176/appi.ajp.20220789AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack Citations ShareShare onFacebookTwitterLinked InEmail Psychiatric disorders are in part heritable and polygenic in nature, and advances in psychiatric genetics have helped to identify the involvement of numerous genetic variants that underlie the risk to develop psychiatric illnesses. Key issues the field is dealing with include how to understand the complex interactions among the numerous genetic variants implicated in an illness with critical environmental events, as well as the meaningfulness of the small effect sizes that have been determined from studies linking genetic variation to phenotypes. This issue of the Journal is focused on recent advances that further explore the genetic underpinnings of psychiatric illnesses. We begin the issue with an overview authored by Drs. Cathryn Lewis and Evangelos Vassos, from King’s College London, that focuses on the use of polygenic risk scores in psychiatry (1). This overview provides basic information on how polygenic risk scores are determined, summarizes current findings related to various psychiatric illnesses, and discusses the limitations of using polygenic risk scores for clinical decision making. While not on the topic of psychiatric genetics, we also include an overview by Hagman and colleagues that reports on the development of a new NIAAA definition of recovery from alcohol use disorder (2). This paper is intended to provide a framework for researchers and clinicians as they work to facilitate advances in helping individuals recover from alcohol dependency.The original research papers and accompanying editorials in this issue address a variety of psychiatric genetics-related topics. One theme is related to understanding the genetics underlying the comorbidities between psychiatric and medical illnesses. In this regard, one paper considers the shared genetics between PTSD and cardiovascular disease, and another paper characterizes the shared heritability relevant to the comorbidity between depression and endocrine/metabolic disorders. Another paper uses novel statistical methods to deepen the understanding of the shared genetics across different psychiatric disorders, while yet an additional paper examines the use of polygenic risk scores to predict ECT treatment responses. Finally, we conclude with a paper that adds to the understanding of the genetic vulnerabilities underlying early childhood psychopathology by characterizing deleterious copy number variants in children and adolescents with early onset psychosis.Understanding the Shared Genetic Risk for PTSD, Depression, and Cardiovascular DiseasePTSD and cardiovascular disease are illnesses that have been linked to stress exposure and considerable evidence also links PTSD to the development of cardiovascular disease. To further understand mechanisms underlying the relation between PTSD and cardiovascular disease, Seligowski and colleagues (3) use genomic data to examine the extent to which these illnesses have common genetic underpinnings. The researchers used genomic data from 36,412 participants from the Mass General Brigham Biobank, along with genomic data on PTSD, depression, and cardiovascular disease from large publicly available databases. A meta-analysis across the databases revealed that PTSD and depression are significantly genetically correlated with hypertension and coronary artery disease. More specifically for PTSD, the genetic correlations with hypertension were 0.35 and with coronary artery disease were 0.24. Of note, and not unexpected, the genetic correlation between PTSD and depression was high (0.81). Next, by using polygenic risk scores, the researchers demonstrated that in addition to polygenic risk scores for PTSD, genetic data related to major depression and cardiovascular disease significantly improved the ability to predict PTSD. And, using a Mendelian randomization approach, the researchers provide data to support a possible causal link from PTSD and depression to hypertension and coronary artery disease. Finally, using pathway enrichment analytic methods, it was found that genetic variants common to the psychiatric and cardiovascular conditions were enriched in pathways involving synaptic structure and inflammation (i.e., interleukin-7). In their editorial (4), Dr. Robert Ursano from the Uniformed Services University of Health Sciences and Dr. Murray Stein from the University of California at San Diego elaborate on the brain-heart connection in relation to PTSD and provide a general discussion of how PTSD can be currently conceptualized.Factors Underlying the Comorbidity Between Endocrine and Metabolic Disorders with DepressionIn clinical practice it is not uncommon to treat patients that have psychiatric illnesses that are comorbid with endocrine or metabolic disorders, and in assessing patients with depression it is important to rule out underlying endocrine alterations. Leone et al. (5) examine the factors underlying the comorbidity between depression and “autoimmune disorders” (i.e., autoimmune hypothyroidism, Graves’ disease, and type 1 diabetes) and between depression and “nonautoimmune disorders” (i.e., type 2 diabetes, obesity, and polycystic ovarian syndrome). The researchers use an extremely large sample from the Swedish registry, over 2 million individuals followed on average for 27 years, to explore the role of heritable and environmental factors in mediating the comorbidity between these conditions. Over the period of assessment, 5.2% of the individuals were deemed to be depressed. Having an endocrine/metabolic disorder significantly increased the likelihood of also having depression (odds ratios of 2.05 for autoimmune diseases and 2.33 for nonautoimmune disease). The association between type 2 diabetes and depression was greatest (odds ratio of 3.48). Through the use of familial cosegregation methods, the evidence for heritable influences was inferred by examining the findings in relation to sibling status. For example, for type 2 diabetes, obesity, autoimmune hypothyroidism, and polycystic ovary disease, the association with depression was highest in full siblings followed by maternal half siblings. This was not found to be the case for the association between type 1 diabetes and depression. Using quantitative genetic analyses, phenotypic correlations were determined. As an example, the phenotypic correlation between depression and type 2 diabetes was found to be 0.22. Further analyses revealed that genetic factors significantly contributed to the phenotypic correlations between nonautoimmune conditions and depression; for example, 74% of the depression-type 2 diabetes phenotypic correlation was attributed to genetic factors. In contrast, nonshared environmental factors were found to underlie the phenotypic correlations between depression and autoimmune disorders, such as type 1 diabetes. In their editorial (6), Drs. Jespersen, Yilmaz, and Vihjalmsson from Aarhus University discuss the importance of the findings from this paper, emphasizing the value of large population registries to understand risk factors underlying psychiatric illnesses as well as their comorbidities with medical illnesses.New Insights Into Shared Genetic Alterations Across Various Psychiatric Illnesses and Associated TraitsWhile it has been well established that different psychiatric illnesses share genetic alterations, Hindley et al. (7) use statistical methods novel to psychiatric research that further the understanding of the overlapping genetics across disorders. Typically, researchers use the measure of genetic correlation (rG) to estimate the degree to which different disorders have shared genetics. With this method, if effects of the same genetic variant are opposite across two disorders, they would in a sense cancel each other out when calculating shared genetic effects across these disorders. In the current study, the researchers use a method called MiXeR that, unlike rG, accounts for the shared genetics across disorders regardless of the direction of the relation between a particular genetic variant and a specific disorder. Using large GWAS samples, the researchers investigated the genetic overlaps among ADHD, bipolar disorder, schizophrenia, and major depression as well as among traits associated with these disorders. Results confirmed the polygenic nature of these illnesses, with major depression being associated with the greatest number of genetic variants (an estimated 14,500 variants) and ADHD the least (an estimated 5,600 variants). Analyses revealed extensive overlaps across these disorders. For example, when considering bipolar disorder and major depression, 7,500 variants were estimated to overlap with 1,100 variants selectively related to bipolar disorder and 7,000 variants selectively associated with major depression. An estimation of the overlapping variants between bipolar disorder and schizophrenia revealed that of the 9,500 variants associated with schizophrenia and the 8,600 variants associated with bipolar disorder, 8,500 variants overlapped. Also, the genetic variants associated with these illnesses had considerable overlap with traits such as intelligence, educational status, neuroticism, and well-being. As a comparator, the researchers also examined genetic overlaps between psychiatric disorders and physical traits such as height, which revealed only a small number of overlapping variants. Taken together, these results demonstrate that in some cases the more traditional method of estimating shared genetics may have underestimated genetic overlaps across psychiatric disorders. In attempting to explain how these shared variants may be related to the development of different illnesses and traits, the authors suggest that “genetic risk for mental disorders is predominantly differentiated by divergent effect distributions of pleiotropic genetic variants rather than disorder specific variants.” In their editorial, Drs. Angelica Ronald from Birkbeck University of London and Oliver Pain from King’s College London (8) comment on the new insights this paper brings to psychiatric genomics and how these types of findings can be used to conceptualize personalized treatment approaches.Polygenic Risk Scores Associated With Effective ECT Treatment for Major DepressionUsing genomics to predict treatment outcomes is of considerable interest as psychiatry moves closer to personalized treatment approaches. In this regard, Sigström et al. (9) examine the extent to which polygenic risks scores can be used to predict ECT treatment outcomes. Using clinical practice data from the unique Swedish National Quality Register for ECT, 2,320 patients were identified who underwent at least one course of ECT for a major depressive episode. Of these individuals, 77.1% suffered from major unipolar depression, whereas the remainder had depressive episodes in the context of other illnesses such as bipolar and schizoaffective disorder. Using clinical global impression improvement ratings (CGI-I) as the primary outcome measure, the results demonstrated a negative relation between polygenic risk scores for major depressive disorder and ECT outcomes. For example, individuals with depression polygenic risk scores in the top quintile were 31% less likely to improve when compared with depressed individuals in the lowest quintile. This finding is somewhat surprising as ECT is generally thought to be effective for the most severely depressed patients. In contrast, polygenic risk scores for bipolar disorder were positively associated with ECT response, with individuals who had scores in the highest quintile 37% more likely to respond when compared with those in the lowest quintile. It is notable that similar findings were observed when the analysis was restricted to only those individuals with unipolar depression. Finally, the researchers determined that ECT treatment response as assessed with the CGI-I was not associated with polygenic risk scores for schizophrenia. The authors acknowledge that the effect sizes are small for the associations between polygenic risk score and ECT treatment response, and at this time are not clinically relevant. Drs. Bart Rutten and Suzanne Bronswijk from Maastricht University provide an editorial (10) that places the findings from this study in the context of other studies and discusses the implications of the small but significant polygenic risk score effect sizes for personalized medicine approaches.Copy Number Variants in Neurodevelopmental Disorders: Early-Onset Psychosis and Autism Spectrum DisorderCopy number variants (CNVs), which commonly occur and usually do not have deleterious effects, are structural genetic alterations that are characterized by duplications or deletions of genetic sequences. Recurrent CNVs are of particular interest since they are usually identical genetic alterations shared across individuals and have been associated with various psychiatric illness including childhood-onset schizophrenia and other neurodevelopmental disorders such as autism spectrum disorder (11, 12). In contrast to recurrent CNVs, nonrecurrent CNVs are structural variations that typically differ in size and composition across individuals, making them more difficult to study at the population level. Brownstein et al. (13) attempt to characterize genetic alterations that are associated with early-onset psychosis by first examining the prevalence of recurrent CNVs in a cohort of 137 children and adolescents with early-onset psychosis. Early-onset psychosis, defined as significant psychotic symptoms emerging before 18 years of age in the presence or absence of other psychopathology, is important to understand as it can be a forerunner of schizophrenia but is also associated with other psychiatric diagnoses. Data from 5,540 participants with autism spectrum disorder and 16,504 controls were also included, which afforded the opportunity to compare the genetic alterations detected in early-onset psychosis with that of a well-established neurodevelopmental disorder known to be associated with CNVs (12). The analyses were focused on 47 CNVs that are associated with neuropsychiatric disorders. It is interesting that the prevalence of recurrent CNVs in the early-onset psychosis individuals was significantly greater than that found in the autism spectrum disorder and control individuals (odds ratios of 2.42 and 5.19, respectively). However, when analyses were performed excluding individuals with early-onset psychosis that also had either comorbid autism spectrum disorder or intellectual disabilities, the differences between individuals with early-onset psychosis and autism spectrum disorder were not significantly different. Among the CNVs identified, it is of interest that the 22q11.2 deletion was found to be associated with early-onset psychosis, as this mutation has also previously been associated with schizophrenia, other psychotic disorders, and autism spectrum disorder. To understand the impact of the total number of CNVs associated with early-onset psychosis and autism spectrum disorder, the researchers used what they term “copy number variant risk scores,” which take into account all of the CNVs within an individual’s genome. Significantly higher CNV risk scores were found for early-onset psychosis and autism spectrum individuals when compared to controls. It is important to note that when excluding participants with the diagnosis of schizophrenia, the finding for the early-onset psychosis group remained significant. The CNV risk scores did not significantly differ when comparing individuals with early-onset psychosis to those with autism spectrum disorder. Based on the findings from this study, and in relation to the use of genetic screening in autism spectrum disorder, the authors suggest that the data from the current study support the use of genetic screening in youth that present with early-onset psychosis. Dr. David Skuse from University College London contributes an editorial (14) that elaborates on the findings from this paper and based on the data also calls for genetic screening of children who have early-onset psychosis.ConclusionsConsiderable progress has been achieved in clarifying the complexity and polygenic nature of psychiatric illnesses and the shared genetic variation across different illnesses and their medical and psychiatric comorbidities. Major findings from the papers published in this issue of the Journal include: 1) significant genetic correlations between PTSD and parameters of cardiovascular disease, with a suggestion that PTSD maybe a causal determinant of cardiovascular disease; 2) the phenotypic correlation between depression and type 2 diabetes can in large part be attributed to shared genetic factors, whereas this does not appear to be the case for the relation between depression and type 1 diabetes; 3) a new statistical approach reveals that the shared genetics across psychiatric disorders may be considerably greater than previously conceived, and highlights the importance of pleiotropy involving genetic variants shared across disorders in determining different phenotypes; 4) polygenic risk scores for bipolar disorder are positively associated with ECT response, whereas polygenic risk scores for major depression appear to be negatively associated with ECT treatment response; and 5) early-life onset psychosis is associated with a relatively high level of deleterious CNVs, which could support the value of genetic screening in this population.While the overall use of genetic information in routine clinical practice is premature, the scientific insights gleaned from current psychiatric genetic studies are important and their value should not be underestimated. These findings, some of which are presented in this issue, are providing insights into the genetic architecture and molecular pathways that may be involved in early-life risk, neural circuit alterations, comorbidities, and treatment responses associated with psychiatric illnesses.Department of Psychiatry, University of Wisconsin School of Medicine and Public Health, Madison.Send correspondence to Dr. Kalin ([email protected]).Disclosures of Editors’ financial relationships appear in the April 2022 issue of the Journal.