Preprint Genome-wide study of somatic symptom and related disorders identifies novel genomic loci and map genetic architecture

forestglip

Moderator
Staff member
Genome-wide study of somatic symptom and related disorders identifies novel genomic loci and map genetic architecture

Vera Fominykh, Piotr Jaholkowski, Alexey A. Shadrin, Elise Koch, Laura B. Luitva, Dorte H. Mikkelsen, Mischa Lundberg, Ole Birger Pedersen, Sisse Rye Ostrowski, Christian Erikstrup, Maria Didriksen, Christina Mikkelsen, Erik Sørensen, Henrik Ullum, Mie Topholm Bruun, Bitten Aagaard, Kaarina Kowalec, Robert Karlsson, Håkan Karlsson, Christina Dalman, Pravesh Parekh, Viktoria Birkenæs, Yury Seliverstov, Bernhard Landwehrmeyer, Ida E. Sønderby, DBDS genetic consortium, Estonian Biobank research team, Olav B. Smeland, Kevin S. O’Connell, Lu Yi, Patrick F. Sullivan, Thomas M. Werge, Lili Milani, Ole A. Andreassen

[Line breaks added]

Abstract
Somatic symptom and related disorders (SSRD) are characterized by a mixture of neurological and psychiatric features and include functional neurological (FND) and somatic symptom disorders (SomD). While these complex neuropsychiatric disorders show evidence of genetic susceptibility, there are no genome-wide association studies (GWAS) of SSRD, and the heritability is unknown.

We did a GWAS of a total of 22,203 patients with SSRD, and 1,831,107 controls of European ancestry. We identified one genome-wide significant locus (chromosome 8:65565084) in SSRD, and one additional locus (chromosome 16:49074278) in the SomD subgroup (n cases = 18,536).

The observed-scale SNP heritability was estimated to be 7.3 % for SSRD, 15.7 % for FND and 7.7 % for SomD. FND and SomD were strongly genetically correlated (rg=0.94, SE=0.11, p=3.9E-18). SSRD showed significant genetic correlation with psychiatric disorders (highest with anxiety, post-traumatic stress disorders, depression, rg=0.3- 0.8), neurological disorders (migraine, chronic pain, rg=0.4-0.6) and immune-related diseases (rg=0.2-0.3).

Functional follow-up analysis of SSRD loci implicated the genes CYP7B1, BHLHE22, and CBLN1, which are involved in metabolic and brain-related processes, suggesting common underlying pathways.

We identified genomic loci associations with SSRD and showed strong genetic correlation between FND and SomD and with neurological and psychiatric disorders, as well as immune-related diseases. The current findings highlight shared underlying pathophysiological processes between SSRD diagnostic categories.

Web | PDF | Preprint: MedRxiv | Open Access
 
GeneCards:
CYP7B1
BHLHE22
CBLN1

CYP7B1 (the main SNP from this study maps to this gene) was previously flagged by @mariovitali:
Referencing the following post, please see below a Network Analysis graph (some more nodes are disclosed here (this is the graph version sent to Ron Davis in October 2017) . Interestingly, Vitamin K, Liver issues (NAFLD), Bile acid-related Metabolism Genes (CYP27A1, CYP7B1) are there.
The figure below depicts the network analysis results, identifying Bile Acid Metabolism issues (CYP7B1, CYP27A1, Oxysterols), Liver Disease, Vitamin K Metabolism and LXR/PPARs :

One person out of twenty from this study had a likely pathogenic variant in the same gene (Supplementary Table 1):

A network medicine approach to investigating ME/CFS pathogenesis in severely ill patients: a pilot study
 
SSRD cases were defined as having a lifetime main diagnosis according to ICD-10 criteria of FND (F44) or SomD (F45) (https://www.cdc.gov/nchs/icd/icd-10/, MOBA, UKB, Danish sample). If it was not possible to specify whether the patient had a primary or secondary diagnosis, we used the criteria that the diagnosis should have been made at a specialized healthcare facility (Estonia) or used available data for diagnosis of interest elsewhere (FinnGene, deCODE, All of Us). Participants from the same cohorts who did not have a primary or secondary diagnosis of FND or SomD were designated as control subjects. Detailed descriptions of the samples are provided in Table 1 and the Suppl. material. We also included information about comorbidity in Suppl. Table 1. For the SSRD sample GWAS analysis, cases were defined as having an FND AND/OR SomD diagnosis, accounting for overlapping comorbidity (except FinnGen, where this information was not available).
 
What specifically is the concern? You'd agree that the FND group is likely sicker on average than the healthy group, right? If a gene was significantly associated with being in the unhealthy group, then there's a good chance it's causal for being in that group.
The diagnostic groups they used are so arbitrary, loosely applied and biased based on the practitioner the patient saw, that it won’t tell us anything meaningful about any of the conditions.

All this tells us is that people that are predisposed to having neurological, psychiatric or immunological conditions are more likely to get an FND or SomD diagnosis.
 
If a gene was significantly associated with being in the unhealthy group, then there's a good chance it's causal for being in that group.

But they might be in the unhealthy group because they are unhealthy for a whole lot of reasons unrelated to the diagnosis attached to that group. Confounders like that seem very likely to turn up in this sort of study.
 
But they might be in the unhealthy group because they are unhealthy for a whole lot of reasons unrelated to the diagnosis attached to that group. Confounders like that seem very likely to turn up in this sort of study.
But they got a genome wide significant hit. Maybe it's because the gene makes people seek out FND practitioners or participate in studies or some other non-interesting reason. Or maybe the people who get FND/SSRD diagnoses do have some biological causal factor in common, even with the diverse presentations.
 
Or maybe the people who get FND/SSRD diagnoses do have some biological causal factor in common, even with the diverse presentations.
Or maybe there is a lot of heterogeneity, but they identified a gene responsible for only a subgroup. With such a large sample size (22k), that might have been enough to get significance for a gene related to only a portion of these people's diagnosis.
 
But they got a genome wide significant hit. Maybe it's because the gene makes people seek out FND practitioners or participate in studies or some other non-interesting reason. Or maybe the people who get FND/SSRD diagnoses do have some biological causal factor in common, even with the diverse presentations.

I raised this issue with the Edinburgh study of phenotypic markers in the UK Biobank cohort.
There are all sorts of reasons why people who are less well for the commonest of reasons (insulin resistance, low grade inflammation maybe) are more likely to end up seeing more practitioners and therefore having a higher chance of gravitating towards BPS style clinics.

You might easily pick up gene associations in these people. One of the problems with doing big number studies is that small systematic confounding biases then turn up as significant.
 
They mention that in the validation cohort, the identified SNPs were not statistically significant.

It seems contradictory to think of these disorders as having no organic basis and then identifying genetic causative factors. The interpretation of the results seems to be keen on reassuring readers that it's all in line with current dogma.
 
Last edited:
Validation analysis
In the validation cohort (Suppl. Table 4) two of the four genome-wide significant SNPs were in the same effect direction as the primary analyses. However, none of the identified loci were statistically significant, likely due to the small number of cases and controls in this cohort.
This is all they write about the validation. It seems like they just brush it off as not important.
 
Back
Top Bottom