Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

Genetic Underpinnings of Personality Show Overlap With Psychiatric Disorders

9 Dec 2016

by Pat McCaffrey

ExtravertedWordCloudThe largest genomewide association study (GWAS) to date of the “big five” personality traits has revealed several loci significantly associated with neuroticism and extraversion. The study also found new genetic correlations between personality traits and psychiatric disorders, including schizophrenia, supporting the idea of a shared genetic basis for complex personality traits and psychiatric disease.

The work was published online December 5 in Nature Genetics.

In the study and associated meta-analysis, Chi-Hua Chen at the University of California, San Diego, and collaborators looked at the personality dimensions of extraversion, neuroticism, agreeableness, conscientiousness, and openness to experience in new and previously published cohorts. New data from 98,725 customers of the consumer genetics company 23andMe were combined with data from the Genetics of Personality Consortium (de Moor et al, 2012; GPC et al., 2015), the UK Biobank (Smith et al., 2016), and deCODE Genetics, an Iceland-based human genetics company. In the final meta-analysis, between 123,132 and 260,861 subjects were included per trait.

Previously, a handful of SNPs had been identified in personality studies, but they had been hard to replicate. The new work identified eight new single nucleotide polymorphisms (SNPs) for three personality traits—six held up in a replication cohort, and two just missed significance. The strongest association was between neuroticism (a trait characterized by negative emotions such as anxiety and fear) and an SNP at 8p23.1, which replicated a previous report from the UK study.

Some schizophrenia-related genes popped out in the analysis: L3MBTL, one of hundreds of genes implicated in the latest schizophrenia GWAS (see SRF related story), was significantly associated with neuroticism, whileMTMR9, a potential player in adverse antipsychotic effects (Aberg et al., 2010), was linked to extraversion. MTMR9 represents the first gene to be found for that personality trait.

“People have doubted whether genes for personality could be identified, and this study confirms (yet again) that once sample size crosses a threshold (where the threshold might be >100,000 subjects), we can find SNPs that are causal or in linkage disequilibrium with causal variants,” commented Dorret Boomsma of Vrije Universiteit in Amsterdam in an email to SRF. Boomsma was not involved in the current analysis, but the study made use of data from 80,000 subjects in the Genetics of Personality Consortium, which she heads.

The study “shows the value of the earlier studies of the genetics of personality consortia that made their complete sets of results available for others (I hope that will also be the case for the current paper). The meta-analysis approaches developed in the genetics community can make optimal use [of] new and existing GWA studies, leading to real progress,” Boomsma wrote.

Parsing personality
The five personality domains are not independent traits—in people, phenotypic measures of openness, agreeableness, conscientiousness, and extraversion positively correlate with each other and negatively correlate with neuroticism. Chen and colleagues found that the genetic correlations followed the same pattern but appeared even stronger, reflecting the shared genetic underpinnings of the traits.

To expand the analysis to psychiatric disorders, the researchers pooled their personality data with genetic information about six conditions: schizophrenia, bipolar disorder, major depression, ADHD, anorexia nervosa, and autism. They calculated genetic correlations and performed a principal component analysis of the entire data set.

As expected, the analysis picked up significant genetic correlation between depression and neuroticism. The correlation, the strongest one in the study, confirms genetic data from recent GPC studies and agrees with clinical data showing links between neuroticism and major depression and anxiety.

The study also found a new correlation between extraversion and ADHD, which Chen thinks may go along with overlapping features of the phenotypes. “One key feature of extraversion personality is a high level of activity, and while there are different subtypes of ADHD, some also show high activity,” she told SRF.

A third relationship, one that Chen said at first she found surprising, linked openness to experience with schizophrenia and bipolar disorder. However, both have been linked to higher dopamine levels, she says, which might provide a common basis for the two.

“These results show that there are a number of genetic ‘dimensions’ in the population with a continuous distribution that cause variation in traits that were considered dichotomous and in the medical domain of psychiatry, as well as in traits that were considered as belonging to the psychology domain,” Boomsma commented. She compared the current study to work showing a genetic dimension that influences risk of schizophrenia, as well as creativity (Power et al., 2015).

Exactly which genes underlie these domains and how they might be linked causally to personality traits and psychiatric disorders remains to be seen. In addition, the SNPs uncovered to date account for a tiny fraction of the heritability of these traits, indicating that there are many other genes waiting to be identified. “This is only the beginning of discovering genes related to personality traits or these complex psychiatric phenotypes,” said Chen.

References: 

Comments

Submitted by Dorret Boomsma on

This is currently the largest GWAS to look at all five major personality traits. People have doubted whether genes for personality could be identified, and the study by Lo and colleagues confirms (yet again) that once sample size crosses a threshold (where the threshold might be >100,000 patients), we can find SNPs that are causal or in LD with causal variants. Significantly, it also shows the value of the earlier studies of the genetics of personality consortia that made their complete sets of results available for others. (I hope that will also be the case for the current paper.) The meta-analysis approaches developed in the genetics community can make optimal use of new and existing GWAS, leading to real progress.

I see figure 3 in the paper as providing important information regarding the relationship of personality and psychiatric disorders: It tells us that 1) personality dimensions are not independent from each other, and 2) personality and psychiatric disorders share a genetic basis. With respect to 1, these findings confirm what we knew, for example, from a large genetic twin analysis of 60 NEO items in which all relations among the items that make up the five personality scales were modeled (Franic et al., 2014). Thus these five major dimensions are not independent as is sometimes said, and part of their inter-correlations derive from genetic correlations. With regard to 2, some of these genetic correlations are reported here for the first time. We knew about the strong genetic correlation of depression and neuroticism, but for the other personality traits, these are new findings.

We should not forget that genetic correlations are correlations (not causes) and that they can arise through a multitude of mechanisms: For example, if one trait causes another trait, and the first trait is influenced by genes, then the second trait also will be heritable and show a genetic correlation with the first trait. This is a very different mechanism than that of genetic pleiotropy, and other processes may also lead to genetic correlations, including statistical or methodological artifacts, or a shared etiology between variables (Nivard and Boomsma, 2016).

Finally, these results show that there are a number of genetic "dimensions" in the population with a continuous distribution that cause variation in traits that were considered as dichotomous and in the medical domain of psychiatry, as well as in traits that were considered as belonging to the psychology domain. Compare also, for example, to the 2015 paper of Power and colleagues (Power et al., 2015), indicating a genetic dimension in the population that influences risk of schizophrenia and also influences creativity.

Submitted by Michael Pluess on

Although personality traits are known to be heritable given consistent and robust twin study results, it has proven very difficult to identify replicable gene variants that underlie this heritability. One of the reasons for the failure to identify genes for personality is that complex traits are the function of many thousands of gene variants of very small effect rather than a few gene variants with large effects. Reliable detection of these small effects requires very large samples which have only become available recently. With increased access to large samples (N >100,000), genomewide association studies (GWAS) have become more successful at identifying gene variants associated with common personality traits. But even large samples, such as those featured by Lo et al. (2016), often yield only a handful of gene variants at genomewide significance, each explaining only a tiny proportion of the variance (usually <.05 percent). Over the last years, the field has moved toward statistical approaches that take into account information across the whole genome. These analyses allow for the estimation of genetic heritability as well as for the testing of genetic overlap across different phenotypes. Furthermore, polygenic scores which reflect summed-up weights across thousands of gene variants have become a promising approach when trying to understand the genetic architecture of complex traits. Lo et al. (2016) are making use of several of these more recent approaches.

Applying a GWAS meta-analysis across their large samples, Lo et al. detected a total of six gene variants for three of the five personality traits. When estimating the genetic heritability, all five personality traits yielded significant heritability estimates, explaining 9-18 percent of the variance, which suggests that although the sum of all genotyped variants explains substantial variance in all five personality traits, the individual contribution of gene variants is so small that even very large samples like the ones used in this study are not sufficiently powered to detect them. Interestingly, the genetic associations for the different personality traits were significantly correlated with each other except for agreeableness and openness. When correlating the genetic scores of personality with those of psychiatric disorders, a range of significant correlations emerged. Most notably, genes for openness correlated with genes for schizophrenia and bipolar disorder, genes for neuroticism with genes for major depression, and genes for extraversion with genes for attention deficit hyperactivity disorder (correlations ranging from r = .30-.56). In a final analysis, Lo et al. conducted a principal component analysis to investigate genetic variation across all included phenotypes. Results suggested that most psychiatric disorders fall into a different cluster than personality traits.

What the findings of Lo et al. confirm is that even very large samples of N >250,000 are probably too small to identify a large number of the numerous gene variants associated with complex traits (beyond a handful of variants accounting for a very small proportion of the variance). Complex traits such as personality traits and psychiatric disorders are most likely made up of thousands—or rather hundreds of thousands—of gene variants, many making contributions too small to be detected with traditional GWAS. Consequently, data-reducing approaches such as polygenic scores and principal component analyses may be more helpful in understanding the genetic architecture of complex traits rather than trying to identify biological processes based on genomewide significant hits.

A further lesson is that genetic influences for complex traits are less specific than expected, given the significant genetic correlations among the five personality traits and with several psychiatric disorders. Important to consider is that the traits included in the current study by Lo et al. are also correlated with each other on the phenotypic level. In light of this shared variance across the different phenotypes, it should not be surprising that the polygenic scores associated with these overlapping phenotypes also share variance with each other. In other words, if the phenotypes themselves are not specific, GWAS analyses using these unspecific phenotypes as outcomes are likely also to yield unspecific genetic solutions underlying these traits. Consequently, GWAS-based analyses are always limited by the quality and specificity of the target phenotype.

In sum, the recent paper by Lo et al. represents an important contribution to our understanding of the genetics of common personality traits overcoming limitations of previous efforts in the field. However, given the overlap between phenotypic measures of personality and psychiatry disorders, as well as the fundamentally correlational nature of GWAS approaches, caution is advised regarding the interpretation of the detected association between personality genes and psychiatric disorders.