Abstract
Depression and autoimmune diseases are frequently comorbid, and longitudinal studies indicate a bi-directional relationship. The mechanisms driving the bi-directionality are poorly understood, but shared genetic factors may contribute. The aim of this thesis was to investigate pleiotropy, the sharing of risk alleles, between depression and autoimmune diseases. Using data from the Psychiatric Genomics Consortium (PGC) and the UK Biobank (UKB), we identified affected individuals and applied a range of statistical genetics techniques to test for shared genetic influences.First, we investigated genetic variation within the Major Histocompatibility Complex, a region robustly associated with autoimmune diseases. We imputed Human Leukocyte Antigen (HLA) alleles, and Complement Component 4 (C4) haplotypes and tested the imputed variants for association with depression in affected individuals from the PGC and UKB. We found no strong evidence for association, indicating that any HLA or C4 variants associated with depression either are rare or have very modest effect sizes.
With the objective of increasing the sample size for our subsequent studies, we next developed a genetically-informed framework for identifying individuals affected by depression in the UKB. We calculated the number of depression measures endorsed, including hospital episode statistics, interview data, and the gold-standard Composite International Diagnostic Interview (CIDI) completed by a third of participants. SNP-based heritability and variance explained by depression polygenic risk scores (PRS) increased with degree of endorsement. The genetic contribution to cases defined by at least two non-CIDI measures of depression approximated that for CIDI-defined cases. We conclude that multiple endorsements can serve as a reliable approximation where the CIDI is not available, and that our framework can be extended to identify other complex traits with increasing validity.
Adopting our genetically informed framework, we next identified individuals affected by any of fourteen autoimmune diseases or depression in the UKB. We confirmed that depression was more common in autoimmune diseases, and vice-versa. Cross-trait PRS analyses were performed to test for pleiotropy, i.e. we tested whether PRS for depression could predict autoimmune disease status, and vice-versa. There was some evidence for association, but analyses showed weak effects, indicating that shared genetic factors do not play a major role in the observed comorbidity.
Finally, we extended the project scope to apply PRS in an industrial setting. Using clinical trial data, we tested whether PRS for schizophrenia and correlated psychiatric traits were associated with treatment response to a novel compound developed for treatment-resistant schizophrenia. There was evidence for an association between treatment response and PRS for schizophrenia and bipolar disorder in the discovery sample, although the results were not replicated. Continued collaboration and data sharing will be required to achieve the sample sizes required to elucidate the common genetic basis of pharmaceutical treatment response.
This work has demonstrated how the scale and breadth of large-scale biobank resources can be used to optimise the sample size and validity of complex traits, including depression and autoimmune diseases. Our findings indicate that shared genetic factors are not a primary driver of the bi-directional relationship, and future research should focus on identifying alternative risk factors involved in comorbid depression and autoimmune diseases.
| Date of Award | 1 Apr 2021 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Cathryn Lewis (Supervisor), Paul O'Reilly (Supervisor) & James Galloway (Supervisor) |