White-matter free-water diffusion MRI in schizophrenia

Student thesis: Doctoral ThesisDoctor of Philosophy



Schizophrenia is well recognised as a heritable, disabling and often chronic disorder. Even though its aetiology is still unknown, evidence has accumulated to suggest that an increase in inflammatory markers and white-matter microstructural abnormalities present early in the illness process. Morphological changes indicate microglial changes, a marker of neuroinflammation, in the hippocampal, frontal and temporal brain regions of these patients post-mortem. Some PET in-vivo studies also report increased uptake of tracers for a protein of the outer mitochondrial membrane, the translocator specific protein (TSPO), that shows increased expression by activated microglia in models of neuroinflammation, in total grey matter and hippocampus of patients. However, TSPO positron emission tomography (PET) imaging presents several limitations, notably the possible expression of TSPO on astrocytes and some neuronal sub-types, the sensitivity limited by poor blood brain barrier (BBB) penetration, and variable binding affinity in humans. Other current neuroimaging methodologies characterising these alterations, such as diffusion tensor imaging (DTI), also have key limitations that can affect the interpretation of results or their clinical implications. Changes in fast diffusing extracellular water, such as free-water (FW), can bias the estimation of DTI indices, confounding the interpretation of the signal changes. Increases in extracellular water can occur due to processes such as atrophy, and due to inflammation, which, given the hypothesised role of neuroinflammation in schizophrenia, is of particular interest in the disorder. Free-water imaging quantifies extracellular water, allows elimination of partialvolume confounders from FW contamination and, thus, more specific tissue alteration

Although there have been a number of recent studies to investigate FW in schizophrenia, there have been inconsistent results, so it remains unclear if patients have significantly increased levels when compared to healthy individuals. To address this, a systematic review and meta-analysis is needed. Furthermore, it is also unclear whether inflammatory stimuli result in an increase in FW levels. To address this, a study testing the effect of an inflammatory stimulus on FW levels is needed. Finally, it is unclear how FW levels correlate with other neuroimaging markers. To address this issue, a study combining FW imaging with other methodologies targeting neuroinflammatory markers such as TSPO PET imaging is needed.

Methods and Results

The introductory chapter 1 provides an overview of schizophrenia, the neurobiological systems under examination, and the methodologies used in the context of the outstanding questions this thesis aims to address.

Chapter 2 is a systematic review and meta-analysis article that discusses recent evidence of free-water differences between patients with schizophrenia and healthy individuals. The Embase, Medline and PsycINFO databases were systematically searched for FW imaging studies examining whole-brain and specific regions of interest to measure FW differences in individuals with schizophrenia compared to controls. Of the identified 106 records, 10 articles met the eligibility criteria for inclusion in the systematic review, and 6 of these deemed suitable for inclusion in the meta-analysis. The overall sample in the latter comprised 288 patients and 326 healthy volunteers. Whole-brain FW was significantly higher in patients relative to healthy volunteers (Hedge’s g = 0.38, 95% CI 0.07–0.69, p = 0.02). The magnitude of the effect size significantly varied with sex (z = −2.54, p < 0.05), such that samples with a greater proportion of female patients were associated with smaller FW difference relative to controls.

Chapter 3 attempts to address the FW findings in first episode of psychosis patients, medicated and with ongoing symptoms. A cross-sectional study was conducted including first episode of psychosis patients within the first five years of clinical contact and healthy volunteers to look at between-group differences in extracellular FW levels. The FW bitensor model and Tract Based Spatial Statistics (TBSS) were applied and show that patients have statistically significant higher levels of whole-brain FW compared to controls. FW is negatively correlated with age, and exploring this particular correlation shows that there is a quadratic relationship between FW and age. Compared to the positive correlation seen in healthy volunteers, younger patients stand out as the subgroup driving this negative correlation, supporting the hypothesis that a first episode of psychosis in earlier stages in life might have a different impact on the brain compared to later occurrences.

Chapter 4 sets out to clarify the neuroinflammatory imaging evidence indexed by microglial activation in response to systemic administration of bacterial lipopolysaccharide (LPS) in a well-known rodent model of sickness behaviour. Freewater imaging and two other neuroimaging methods – DTI and neurite orientation dispersion and density imaging (NODDI) – were applied on the same in vivo diffusion magnetic resonance imaging (dMRI) data and their distinct characteristics were used to quantify the inflammation-induced changes on the brain tissue microstructure. Intraperitoneal administration of LPS elicits an immune response and Iba1 immunohistochemistry revealed microglial activation in the parietal association cortex, hippocampus, striatum, and prefrontal cortex of LPS-treated rat brains. All three modalities were sensitive to LPS-induced changes, together indicating an increase in water diffusivity as a result of increased extracellular and/or microglial cell volume.

Chapter 5 describes a PET-FW imaging study to test the relationship between TSPO expression and FW, in the context of schizophrenia. The PET analysis was restricted to the main regions of interest (ROI) of frontal lobe GM, temporal lobe and temporal lobe GM, hippocampus, pre-frontal cortex and striatum. A repeated-measures general linear model adjusted for the continuous covariates of age and 18 kDa translocator protein (TSPO) genotype was conducted to identify statistically significant group differences in the different ROIs, and Spearman correlations were used to analyse the relationship between white-matter FW and white-matter TSPO binding values. Patients with schizophrenia did not differ from healthy volunteers in white-matter TSPO nondisplaceable binding potential, but white-matter FW levels and TSPO BPND were significantly and positively correlated.


Chapter 6 considers the outcomes of the preceding 5 chapters. In summary, this work shows 1) FW is a sensitive marker of neuroinflammation, 2) patients with schizophrenia have a significant increase in white-matter FW levels compared to controls, 3) these levels are negatively correlated with age in younger patients, and 4) FW levels are positively and significantly correlated with TSPO non-displaceable binding potential. These findings suggest the occurrence of neuroinflammatory mechanisms such as microglial changes in patients with schizophrenia. Also, they highlight how these mechanisms may manifest differently depending on the age of onset of the first episode of psychosis. Increased extracellular free-water was shown to be caused by pathologies like neuroinflammation, atrophy, low cell density, or a breakdown in the cellular membrane. Furthermore, previous post-mortem studies have reported elevated microglial markers, including elevated TSPO binding, interpreted as increased microglial activity in schizophrenia. However, this is the first study to show an association between TSPO binding and FW in patients with schizophrenia.

Together, these findings extend the existing understanding of the extracellular and whitematter changes in schizophrenia, and they highlight the possible role of extracellular pathologies such as neuroinflammation in the pathophysiology of the disorder.

Date of Award1 Feb 2023
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorFederico Turkheimer (Supervisor) & Oliver Howes (Supervisor)

Cite this