Exploring the Genetic Similarities and Differences Between Major Depressive Disorder and Depression as a Normal Human Experience

Student thesis: Doctoral ThesisDoctor of Philosophy


The boundary between normal sadness and Major Depressive Disorder (MDD) is difficult to identify due to the similarity in symptoms and the subjective decision-making required from clinicians. Both academics and clinicians suggest approaches to separate between the two phenotypes, however, objective evidence for the similarities and differences is limited. It is important to determine if such a boundary exists as it could help refine the substantial heterogeneity in MDD. In turn, this could unpick risk factors which may be used to develop a deeper aetiological understanding and therefore improve treatment for both normal sadness and MDD.

Using modern methods from statistical genetics, this thesis aimed to define various definitions of normal sadness and identify similarities and differences in the genetics relative to the current clinical definitions of MDD. Each empirical chapter tackles a different conceptualisation of normal sadness, defined as ‘Statistical Normality’ (Chapter 2), ‘Diagnostic Normality’ (Chapter 3) and ‘Contextual Normality’ (Chapter 4).

In Chapter 2, I asked whether defining MDD as a continuous trait holds information over and above a binary lifetime diagnosis. I used factor analysis on responses to the PHQ-9, GAD-7, and items relating to subjective well-being to define a series of MDD dimensions. I find MDD cannot be modelled as a single continuous dimension and demonstrate an association between MDD Polygenic Risk Scores (PRS) and each dimension of MDD. The association with MDD PRS remained when the dimensions are stratified according to case and control status of lifetime MDD.

In Chapter 3, I extended the phenotype beyond the symptoms of MDD to account for additional diagnostic components, namely - episode duration, episode recurrence, number of symptoms, functional impairment, and persistence of the depression during the episode. I used data from the UK Biobank to systematically account for these components by defining 32 MDD phenotypes to explore the influence on single nucleotide polymorphism-based (SNP-based) heritability and genetic correlation with both a gold-standard MDD phenotype (Psychiatric Genomics Consortium defined MDD) and two minimal phenotypes (Broad Depression and 23andMe defined depression). Requiring 5 or more symptoms consistently increased the SNP-based heritability to the greatest degree whereas episode duration showed the greatest decrease. In all instances, statistical significance in the differences was limited. Similarly, there was limited evidence of an impact on genetic correlations with previously defined MDD phenotypes and no evidence of any difference in trend when comparing between minimal and gold standard phenotypes.

In Chapter 4, I defined normal sadness according to the context in which the major depressive episode presents. I used a combination of self-report and administrative data from the UK Biobank to identify individuals with a first record of a major depressive episode within a year of a diagnosis of a subset of chronic diseases: cancer, stroke, myocardial infarction, autoimmune disorders, epilepsy, diabetes (type 1 and 2) and motor neurone disease. I used a similar methodology to identify cases of postpartum depression. We show differences in the environmental and genetic risk factor profiles of postpartum depression and depression following a chronic disease definition relative to a typical, uncontextualized definition of MDD. MDD in response to a diagnosis of a chronic disease had weaker associations with neuroticism scores and reports of childhood trauma. Postpartum depression had stronger associations with reports of adulthood and childhood trauma, neuroticism scores, MDD PRS and family history of severe depression. In contrast postpartum depression cases were less likely to live in a deprived neighbourhood.

Using definitions of normal sadness as our inspiration, this thesis considered a series of depression phenotypes that may be useful to disentangling various forms of MDD. I demonstrate the utility of a continuous form of MDD, highlighting the extra information contained within a dimensional depression trait as opposed to a categorical disorder. I find genetic differences between phenotypes defined according to current diagnostic guidelines to be limited, suggesting comparable genetic insights can be gleamed from more minimally defined phenotypes. Finally, I demonstrate the importance of considering the context of an individual’s depressive episode to refining estimates of effect sizes between groups.
Date of Award1 Nov 2021
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorCathryn Lewis (Supervisor), Evangelos Vassos (Supervisor) & Saskia Hagenaars (Supervisor)

Cite this