The American Journal of Medical Genetics has dedicated its March issue to the study of anxiety, including a contribution from our own lab! Here, I summarise some of the exciting developments it describes.
Harvard’s Jordan Smoller (whose contributions feature heavily in the issue) sets the scene for this foray into “the wilderness of anxiety disorder genetics”. Championing the innovative work represented in the issue, Smoller highlights the difficulties of researching this complex phenotype, and urges funders and investigators to commit to larger and systematic investigations (a call that is not falling on deaf ears in Camberwell…!).
The first report comes from the Army STARRS program, a longitudinal study of 72,000 US Army personnel that promises to contribute considerably to our increasing understanding of mental health through several ongoing longitudinal projects. Army STARRS has already performed multiple genomic studies of post-traumatic stress disorder, and in this issue they present the first genome-wide association study (GWAS) of social anxiety disorder, in 14,592 participants. The studied individuals were drawn from European, Latino and African backgrounds, offering a much-needed increase in diversity to the psychiatric genetic literature (although there is still a long way to go).
“Smoller highlights the difficulties of researching this complex phenotype, and urges funders and investigators to commit to larger and systematic investigations”
What then do we learn of the genetics of social anxiety from this effort? Firstly, that studying common genetics is valuable – the authors use a technique known as GREML to estimate the contribution of common, additive genetic variants to variance in social anxiety (SNP-heritability), and obtain a significant estimate of 12% in the European participants, and 21% in the Latino cohort. Similar-sized estimates were identified in the African cohort, but these were non-significant – this is likely to be a limitation of power, rather than an absence of true signal.
From an individual-variant perspective, the power of the cohort was again limited by its size – it is likely that tens of thousands of participants will be necessary to robustly identify variants. Nonetheless, studies in both the European and the African cohorts identified a region at genome-wide significance (although not the same region). Functional annotation in GWAS is a difficult task that many investigators are actively involved in solving, and the success of science comes in the replication of results, so these regions should be considered of preliminary interest, rather than definitive associations with social anxiety. Nonetheless, they represent a crucial step forward in our understanding of the common and impairing illness.
“the success of science comes in the replication of results, so these regions should be considered of preliminary interest, rather than definitive associations with social anxiety”
From one GWAS to another, this time a study of Generalised Anxiety Disorder (GAD) from 12,282 members of the Hispanic Community Health Study/Study of Latinos. If you include PTSD in the umbrella, it’s reasonable to say the anxiety disorders are punching above their weight in providing genomic data on ancestry groups beyond Europeans. Again, evidence was shown of SNP-heritability, an estimate of 7%. However in this instance, no individual variants were identified. Yet it is extremely promising to see this study published – the approach taken to analyse the data is rigorous and sensible, the question asked is important, and the data it puts into the scientific sphere will be vital in developing our understanding of GAD, especially given the ethnicity of the cohort.
Our contribution to the issue was not a GWAS. In fact (whisper it quietly) it was a candidate gene study. I have previously been a bit negative about this method (it tends to use small sample sizes, fails to systematically interrogate variants in the regions of interest, and the results yielded have proved difficult to replicate). However, we tried to address some of these issues.
We used 1,309 of the participants in our cohort of children with anxiety treated with cognitive behavioural therapy to investigate the association between response to therapy and genetic variants in the endocannabinoid system genes. The endocannabinoid system has previously been implicated in fear learning (which has clear parallels to cognitive behavioural therapy). Using data from our GWAS in the same cohort, combined with targeted genotyping of variants around these genes, we were able to conduct a systematic study, covering six hundred variants across thirty regions (defined from GWAS data) that span three genes (CNR1, CNR2, and FAAH).
However, statistical rigour does not relieve you from results that land on the edge of significance, and that was much the case in this work – several variants were close to the multiple-testing threshold, and one snuck over to be declared statistically significant. Nonetheless, we could conclude only that we found very limited evidence in favour of a role for the endocannabinoids. A number of genes previously hypothesised to be relevant in cognitive behavioural therapy have shown extremely weak evidence of association with response in our cohort. That was not the case here – we cannot highlight the endocannabinoids as a clear system of interest, but equally we cannot rule them out. Once again, replication (preferably in larger, more homogenous cohorts) will be the key.
So, where are we? I think the answer is “not there yet” – we still have only a few variants reliably associated with any anxiety disorders, but much of the initial work has been done. Smoller, in his introduction, highlights the new Anxiety Disorders Working Group of the PGC as an exciting development – I (unsurprisingly) agree. The PGC has a successful history of being a catalyst for developing genomic studies in psychiatric genetics, and it can enable the fantastic efforts that have been made by the groups represented in this issue, and elsewhere, to be brought together to yield robust results. Anxiety genomics is not there yet – but it is coming.