Repository logo
 

Functional neuroimaging studies of peripheral inflammation-related depression


Type

Thesis

Change log

Authors

Abstract

Depression is recognized as a leading cause of disability worldwide although it lacks one of the typical characteristics of a disorder, disease or disability in the traditional sense, i.e., biological markers. Depression is difficult to characterize in that it is highly heterogeneous epidemiologically and symptomatically. Depression is often a comorbidity, associated with other physical or mental health disorders, and depressive symptoms occur on a spectrum of severity ranging from mild dysphoria to major depressive disorder (MDD). Comorbid depression is associated with an array of neurological disorders with neuroimmunological or neuroinflammatory mechanisms, as well as peripheral systemic inflammatory disorders such as rheumatoid arthritis, psoriasis, and cancers. Recent research on the immune mechanisms of depression has reproducibly demonstrated a robust association between biological markers of peripheral inflammation, e.g., blood levels of C-reactive protein, and depressive symptoms. However, there have been fewer brain imaging studies of inflammation-related depression, which are important for bridging the explanatory gap between peripheral immune states, like inflammation, and mental states, like depression, in humans.

Chapter 1 introduces a dichotomy in thinking about depression, discussing briefly historical concepts that on one hand supported a monolithic view of the disorder, and on the other hand, alluded to greater complexity based on how the body might interact with the brain and mind. On this basis, I highlight areas of contemporary immunopsychiatric research that both support and challenge the hypothesis central to this thesis, that is, peripheral (bodily) inflammation may elicit depressive symptoms via brain functional abnormalities.

In Chapter 2 I present a more focused discussion in the form of a literature review of current brain functional neuroimaging studies on inflammation-linked depression. I noted through this endeavor that the body of knowledge addressing fMRI abnormalities in inflammation-linked depression is presently limited, and it is further complicated by considerable variability in study setting, methodology, sample and analytic approaches. Peripheral inflammation has often been measured simply by blood concentration of C-reactive protein (CRP) and brain phenotypes have often been measured in a few, selected regions of interest (ROIs) rather than across the whole brain. Nonetheless, the extant literature very broadly converged to suggest the concept that dysregulation of the peripheral immune system could indeed be associated with brain functional abnormalities in depression.

In Chapter 3, I describe the design and assessments used in an observational case-control study (BioDep) of three groups of participants, i.e., healthy controls (HC), low-CRP depression cases (CRP ≤ 3 mg/L), and high-CRP depression cases (CRP > 3 mg/L). All participants completed assessments of peripheral blood immune markers, behavioral questionnaires, resting-state fMRI, and a probabilistic reinforcement-learning task-based fMRI paradigm. On this basis, I tested the following key hypotheses which were the focus of subsequent chapters: (i) increased concentrations of innate and adaptive immune markers is characteristic of inflammation-linked depression, (ii) inflammation-linked depression is associated with diffuse functional connectivity abnormalities, (iii) increased peripheral inflammation attenuates functional connectivity in depression, and (iv) peripheral inflammation reorients response to affective experience in depression.

In Chapter 4, I first investigated sociodemographic, clinical, behavioral and immune variability in the analyzable cohort (N = 129; Ndepression = 83). BMI and sex differed significantly between the low CRP (N = 50) and high CRP depression cases (N = 33), with both BMI and proportion of females being greater in the high CRP depression cases. Sex and BMI were therefore noted as potentially confounding variables that would require careful consideration in subsequent analyses. With regards to immune markers, I first performed a simple pair-wise correlational analysis on CRP and 16 other inflammatory proteins, i.e., cytokines and chemokines. Weighted network visualization, coupled with a literature search-based functional assignment of each biomarker, indicated that functionally-related inflammatory proteins were more strongly positively correlated with each other. Thus, concentrations of these clusters of immune markers were similarly increased or decreased in the blood. Next, I reduced the dimensionality of this multivariate biomarker dataset using principal component analysis (PCA) on 15 inflammatory proteins (excluding IL-6). This resulted in 5 selected principal components. I interpreted the first principal component (PC1) to be a weighted average of all inflammatory proteins or a global index of immune state. PC1 score was also positively correlated with neutrophil count.

Chapter 5 marks the start of my investigation of the links between peripheral inflammation and the brain, using whole-brain functional connectivity (FC) measures derived from resting-state functional magnetic resonance imaging (fMRI) data. I conducted a two-fold exploratory analysis: (i) multi-granular decomposition of the functional connectome at coarse- and fine- grained anatomical resolutions, and (ii) a data-driven subnetwork-level decomposition of the functional connectome using network-based statistics (NBS). This non-parametric method of significance testing for high CRP depression case-control differences yielded a set of interconnected brain regions - in other words, a network - that showed progressively increased abnormalities, denoted by weaker and more negative functional connections, in high CRP depression cases, followed by low CRP cases, compared to controls. The attenuated functional connections were mainly anatomically located between the left insula/frontal operculum and posterior cingulate cortices. Meta-analytic search of an independent fMRI database suggested that this network was functionally specialised for interoception, i.e., brain sensing of internal bodily states for emotion modulation.

In Chapter 6, the putative interoceptive network discovered by the case-control analysis of high CRP depression cases versus controls in Chapter 5 was used as a 'mask' to test the continuous association between brain functional connectivity and peripheral inflammation in all depression cases (including low CRP cases but excluding controls). There was robust negative scaling between average network connectivity (or 'within-network' connectivity) and CRP (N = 83), IL-6 (N = 72), and PC1 scores (N = 72) from the principal component analysis of 15 cytokines and chemokines in Chapter 4. Corroborating this association between functional connectivity and inflammatory proteins, neutrophil count (N = 36) also showed significant negative scaling with average network connectivity. These results are interpreted as evidence suggesting that inflammation-linked depression could be underpinned by abnormalities of interoceptive processing of afferent peripheral immune signals, and/or signaling in other motivational or reward-related circuits, which could be clinically manifest as dysregulation or misrepresentation of emotional states, i.e., feelings of depression.

In Chapter 7, I investigated abnormalities of affective traits, e.g., anhedonia and pessimism, using task-based fMRI in inflammation-linked depression. Through a probabilistic reinforcement learning paradigm, I tested for evidence of hyposensitivity to reward, and hypersensitivity to punishment, with increasing inflammation. Voxel-wise activation was observed in key brain regions sensitive to monetary reward (ventromedial prefrontal cortex, vmPFC; nucleus accumbens, NAc) and punishment (insula) outcomes in all three groups (HC, low CRP depression and high CRP depression). However, there was no significant difference in activation between any two groups. Within depression cases, increasing CRP scaled negatively with activation in the right vmPFC and left NAc. However, there was no significant association between regional activation and severity of anhedonic or negative attitudes measured by Beck’s Depression Inventory (version II).

Finally, in Chapter 8, I conclude by reviewing the initially proposed central hypothesis, noting that whilst the novel evidence generated by these studies provide some support for the hypothesis that peripheral inflammation is associated with functional abnormalities in depression-related brain networks, these data also prompt further refinement and evolution of this model. In particular, these results indicate that the brain circuits most sensitive to inflammatory states in depression may be functionally specialised for interoceptive sensing and processing of peripheral immune signals. These results have also shown that immune dysregulation in inflammation-linked depression is at the level of the 'immune interactome' - as opposed to a single immune marker in the periphery - warranting particular look into chemokines and immune cells, beyond CRP and IL-6. In view of these findings, I finally highlight the need for future work focused on interoceptive representation of peripheral immune signals within the brain, and CNS vascular physiology, that can together better delineate mechanisms of interaction between the brain and the body as demonstrated here in inflammation-related depression.

Description

Date

2022-03-31

Advisors

Bullmore, Edward Thomas

Keywords

Psychoneuroimmunology, Immunopsychiatry, Depression, Interoception, fMRI, Brain networks

Qualification

Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
Sponsorship
Wellcome Trust (104025/Z/14/Z)
Alzheimer's Research UK (ARUK-NIMA2019-001)
Mr Aruldass William and Mrs Regina Aruldass