The effects of peripheral inflammation on the brain – a neuroimaging perspective

Abstract

In the field of neuropsychiatry, neuroinflammation is one of the prevailing hypotheses to explain the pathophysiology of mood and psychotic disorders. Neuroinflammation encompasses an ill-defined set of pathophysiological processes in the central nervous system that cause neuronal or glial atrophy or death and disruptions in neurotransmitter signaling, resulting in cognitive and behavioral changes. Positron emission tomography (PET) for the brain-based translocator protein (TSPO) has been shown to be a useful tool to measure glial activation in neuropsychiatric disorders. Recent neuroimaging studies also indicate a potential disruption in the choroid plexus (CP) and blood brain barrier (BBB) which modulate the transfer of ions, molecules, toxins, and cells from the periphery into the brain. Simultaneously, peripheral inflammatory markers have consistently been shown to be altered in mood and psychotic disorders. However, the “crosstalk” (i.e., the communication between peripheral and central inflammatory pathways) is not well understood in these disorders and neuroimaging studies hold promise to shed light on this complex process. In the current perspective, we discuss the neuroimaging insights into neuroimmune crosstalk offered in selected work. Overall, evidence exists for peripheral immune cell infiltration into the CNS in some patients, but the reason for this is unknown. Future neuroimaging studies should aim to extend our knowledge of this system and the role it likely plays in symptom onset and recurrence.

Introduction

In recent years, neuroinflammation has been identified as one of the leading hypotheses to explain illness onset and recurrence in severe psychiatric disorders, with immunological changes in the central nervous system (CNS) observed in postmortem brain tissue, cerebrospinal fluid (CSF) and neuroimaging studies. It has been further suggested that inflammatory mechanisms may precipitate alterations in neurotransmitter systems (e.g., monoaminergic and glutamatergic), resulting in behavioral and cognitive changes¹. Increased putative neuroinflammation was observed in early-phase psychosis patients, which was stably elevated for a 12-month follow-up period², indicating that neuroinflammation has a role in the onset of illness. At the same time, there are a plethora of data that indicate increased peripheral inflammatory mediators play an important role throughout the course of psychiatric illness. Evidence for this was offered in a recent, large, registry-based Danish cohort study which showed that infections requiring hospitalization increased the risk for development of schizophrenia (SZ) spectrum disorders³. After the onset of disease, inflammation may contribute to illness recurrence, treatment resistance, and cognitive impairment in a transdiagnostic manner. A recent review article highlighted partially overlapping peripheral immune profiles (e.g., sIL2R, IL-6 and TNF-α) in SZ, major depressive disorder (MDD) and bipolar disorder (BD), implicating both innate and adaptive immune responses⁴. Moreover, peripheral levels of the cytokine tumor necrosis factor alpha (TNF-α) and its soluble receptors have recently been shown to partially mediate cognitive dysfunction in patients with BD⁵, as has also been reported in SZ⁶. Importantly, elevated immune markers may be valuable predictors of treatment response in psychiatric populations^7,8. This highlights the importance of deepening our understanding of immune dysfunction in these disorders, as a potential modifiable risk factor for poor outcome.

Our understanding of the neuroimmune crosstalk has grown massively over the last few decades. Microglia are the resident immune cells of the CNS, with essential roles in neurogenesis, response to tissue damage, remodeling of synapses in development and disease, and surveillance of the interstitial milieu⁹. Also, the meningeal space surrounding the brain and spinal cord hosts a plethora of immune cell types including adaptive immune cells (e.g., CD4+ T cells) which release cytokines that modulate neuronal activity¹⁰. The CNS is protected from the peripheral circulation by the blood brain barrier (BBB) and blood-CSF barrier, which are sites of cellular and molecular transfer under normal conditions, and in injured or diseased states¹¹. There exist various other channels of communication between the CNS and periphery. For example, the glymphatic system carries interstitial solutes from the CNS parenchyma to the CSF and bloodstream, where they contact antigen presenting cells in lymphoid tissues and leptomeningeal space¹⁰. Importantly, the CNS can mount an immune response against peripheral or central infection or disease⁹.

Unraveling the “chicken or the egg” phenomenon in psychiatric illnesses is of broad interest to the field. Many unanswered questions remain: How does putative neuroinflammation originate, and what cells and molecules are involved? What is the role of neuroinflammation in the onset and recurrence of symptoms in mood and psychotic disorders? Despite promising early findings, there still exists a large gap in our understanding of how peripheral and central immune cells and molecules influence outcomes in severe psychiatric disorders. Neuroimaging studies, in combination with other central and peripheral markers of inflammation, may help address unanswered questions related to the causal influence of peripheral inflammation on neuroinflammation and vice versa and help identify the instigating nodus, which precipitates a series of potentially harmful inflammatory responses in the brain. Below, we discuss selected recent studies that examine associations between peripheral and central immune markers, and the anatomical barriers which modulate their interactions, highlighting insights gleaned from this work and how ongoing research can fill in knowledge gaps.

Neuroimmune Crosstalk: Insights from TSPO PET Studies

While understanding the relationship between peripheral and central inflammation in psychiatric disorders is important, there are no ‘perfect’ tools to measure central inflammation in vivo. For example, lumbar puncture for CSF poses some risk to patients in that it is invasive and can cause headache and back pain¹². It is also not clear how well CSF inflammatory measures reflect central brain-based inflammatory processes, because blood-CSF disruption could result in peripheral cytokines leaking into the CSF which may activate central inflammatory pathways. Neuroimaging is an alternative tool that poses fewer risks to patients, however none of the available imaging methods directly measure neuroinflammation – positron emission tomography (PET) for translocator protein (TSPO) targets glial activation, diffusion neuroimaging assays increased BBB permeability through extracellular free water, and magnetic resonance spectroscopy quantifies neurometabolites involved in the inflammatory cascade. Among those imaging biomarkers, PET neuroimaging for the marker 18 kDa TSPO— a transmembrane protein located on the outer mitochondrial membrane in glial cells (microglia or astrocytes)¹³, endothelial cells, and other cell types in the CNS— has shown the most promise in tagging putative neuroinflammation in neurological disorders, such as multiple sclerosis (MS)^14,15. Specifically, studies have shown that protein expression of TSPO increases in activated microglia during neuroinflammation¹³. However, even this method has revealed heterogenous findings in psychiatric disorders. Further, as we discuss below, the relationship between TSPO binding in the brain in neuropsychiatric disorders and peripheral immune status is not well understood.

Various studies have performed case-control comparisons between psychiatric clinical populations and healthy controls using TSPO. The results of these studies are now widely understood to be heterogeneous. This discrepancy is highlighted by two recently published meta-analyses in patients with SZ-spectrum disorders, one of which found an increase in TSPO binding¹⁶, and another that found a decrease¹⁷. It is challenging to fully interpret discrepancies in these studies, as TSPO binding is influenced by numerous variables (fully reviewed elsewhere¹⁸) including the tracer, kinetic model, medications, acute clinical symptoms of patients¹⁹, as well as age, sex, BMI, and smoking status^18,20. TSPO radiotracer perfusion might also change based on the level of peripheral inflammation, where higher levels (e.g., as measured by CRP concentrations) predict reduced perfusion into brain parenchyma²¹. Further, a widely theorized idea in the field is that there exist “immunophenotypes” in severe psychiatric illnesses, and many studies using TSPO have not parsed for this level of heterogeneity.

Despite these caveats, some studies using TSPO have also examined peripheral markers of inflammation, which may shed light on neuroimmune crosstalk. Here literature is again mixed, with many studies showing that TSPO binding is generally not strongly correlated with peripheral cytokines and acute phase reactants^22–24, even when TSPO binding is increased in patients²², whereas others show significant positive or negative associations¹⁹. Coughlin et al. (2016) examined patients in early-phase SZ using the TSPO radiotracer ¹¹C-DPA713 and failed to find a significant relationship between TSPO and peripheral cytokines (IL6, IFN-γ, TNF-α, IL10), however, they observed a significant positive association between IL6 in the CSF and blood²⁴. This finding may indicate that microglia were not activated, and not the primary source of central IL-6 secretion. Many immune and non-immune cell types in the CNS can produce IL-6 in response to injury, including neurons, astrocytes, and endothelial cells²⁵. In fact, activation of cerebral endothelial cells was found to be the earliest event in the start of acute neuroinflammation in a rodent model of sepsis associated encephalopathy²⁶. Alternatively, IL-6 may have entered the CNS from the periphery through a saturable transport system, by trans-signaling, or through extracellular mechanisms^25,27. Finally, the temporal window of glial activation may not be adequately captured. For example, one study found increased TSPO binding three hours after peripheral infusion with lipopolysaccharide (LPS) in healthy volunteers²⁸, while another study failed to find increased TSPO binding 24 hours after peripheral infusion with the cytokine IFN-α in healthy individuals²⁹. One small study (n=6) in non-human primates supported this – they showed an increase in TSPO at 1 and 4 hours post intravenous LPS administration, but failed to show evidence of glial activation at 22 hours, despite an elevation in peripheral IL-6 at this later timepoint³⁰. Glial activation might therefore be associated with more acute symptomatology; in fact, De Picker (2018) found a positive association between peripheral quinolinic acid (QA) - kynurenic acid (KA) ratios and TSPO binding in acutely ill psychosis patients – an association that disappeared at the follow-up timepoint when patients were stable¹⁹. In sum, there is much still to learn about microglial behavior in general, and novel research has begun to shed light on disease-stage specific microglial activation³¹. More longitudinal follow-up studies using TSPO PET (with adequate sample sizes to control for relevant covariates) are needed to delineate causal relationships relevant to neuroimmune crosstalk.

The Barriers Modulating Immune Crosstalk: The Role of BBB and Choroid Plexus

To better understand the crosstalk between peripheral and central environments in the context of inflammation, recent neuroimaging studies have targeted the barriers that protect the brain from the outside world. The blood-brain-barrier (BBB) is a meshwork of microvasculature that regulates the entry of nutrients, ions and cells into the brain parenchyma from the circulation³². In diseased or injured states, activated T cells, neutrophils and macrophages interact with the BBB and regulate its properties – in particular, they release reactive oxygen species and cytokines such as TNF-α which increase BBB permeability³³. Loss of BBB function can have many negative consequences such immune cell infiltration³³. There is limited evidence in BD and SZ to directly support this mechanism, however, recent work has shed some light on this process. A study using dynamic contrast-enhanced MRI (DCE-MRI) scanning in patients with BD found a subset of ten patients (out of 50 subjects total including healthy controls) that had ‘extensive BBB leakage’, which was associated with more severe clinical symptoms and insulin resistance³⁴. A recent in vitro study seems to support these findings: BBB endothelial cells derived from the human induced pluripotent stem cells (HiPSCs) of patients with 22q11.2 deletion syndrome and schizophrenia had compromised barrier function and disorganized claudin-5 (a tight junction protein) expression³⁵.

Others have directly examined immune cell infiltration in the brain parenchyma. One study by Schlaaff et al (2020) examined the density of T and B lymphocytes in whole brain sections of patients with SZ and major depressive disorder (MDD) and healthy controls, using tissue from the Magdeburg Brain Collection³⁶. They found elevated lymphocyte levels in 7 of 20 mood disorder patients and 9 of 22 patients with SZ, compared to 1 of 20 healthy controls. In healthy conditions, lymphocytes rarely cross the BBB, therefore higher densities suggest neuroinflammation and a weakened BBB. Also, there is apparent heterogeneity in immune cell infiltration, which is consistent with prevailing hypotheses that there is a specific subset of BD and SZ patients who have an inflammatory subtype of illness. In line with this idea, a recent study by Cai et al (2020), found that those patients with SZ with high levels of inflammation (based on elevated cortical inflammatory-related transcripts), showed evidence of increased immune cell transmigration into the brain³⁷. This study implicated the intercellular adhesion molecule 1 (ICAM1), which allows for the adhesion and transmigration of leukocytes across the BBB³⁷. This study reported increased ICAM1 expression in the dorsolateral prefrontal cortex (DLPFC) in the ‘high inflammation’ SZ group, relative to the ‘low inflammation’ SZ and healthy control groups. In sum, existing evidence suggests that for some individuals with a major mood or psychotic disorder there may be increased permeability of the BBB and lymphocyte infiltration into the CNS. Whether this is a cause or consequence of disease is unknown.

The choroid plexus (CP) works in concert with the BBB to protect the brain from invading pathogens and toxins³⁸. The CP epithelium contains tight junctions which are rate-limiting for passive passage of solutes and migration of cells³⁹. There is some evidence that the CP is important in transmigration and stimulation of T cells in response to peripheral inflammatory signals³⁹, however there is limited evidence in BD and SZ that immune cells cross the CP. A study by Kim et al. (2016) sequenced the transcriptome in the CP in patients with SZ and healthy controls from the Stanley Array Collection, and a replication sample from the New Stanley Collection. They compared measures from the frontal cortex and serum within the same individuals and found that genes associated with inflammation were upregulated in the CP in patients and were associated with peripheral markers of inflammation⁴⁰. Similarly, a more recent study by Lizano et al. (2019) examined CP volumes in a large sample of psychosis probands (N=544) from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (BSNIP) Consortium and found that CP volume was significantly larger in probands compared to controls and their relatives⁴¹. Larger CP volume was associated with smaller total gray matter in the brain, poorer cognition, larger lateral ventricles, and lower structural connectivity in probands⁴¹. Interestingly, the authors also found an association between larger CP volume and higher concentrations of IL-6 cytokine in probands⁴². A very recent study found that CP volume was significantly positively associated with allostatic load (i.e., physiological “wear and tear” on the brain and body due to prolonged or abnormal stress) in patients is SZ, after controlling for age, sex, education, and intracranial volume⁴³. Although this work is in its infancy, there appears to be clear associations between peripheral immune measures and abnormalities in the CP (e.g., larger CP volumes). The causal relationship between these measures is unknown, and the role of the CP in the pathophysiology of mood and psychotic disease has yet to be fully elucidated.

The glymphatic system is a newly described pathway that clears wastes from the brain and is an additional target for imaging of brain barriers and crosstalk^44,45. There are a few neuroimaging methods, to map BBB water exchange⁴⁶, that have been used to study this system. Other studies have specifically targeted aquaporin-4 (AQP4) – a water channel that is highly expressed on astrocytic end feet near the BBB and blood-CSF barrier and is essential for regulating fluid exchange across these barriers^47,48. To our knowledge, these techniques have yet to be applied to neuropsychiatric disorders, which is an area requiring more research in future.

Conclusion

The processes that control neuroimmune “crosstalk” between the CNS and periphery – including the role of the BBB and CP – are highly complex and not fully understood. It is known that the CNS has an active role in communicating with the peripheral immune system and can influence its function via neural and humoral pathways⁴⁹. Also, the glymphatic system allows for antigens to drain from the CNS to cervical lymph nodes and potentially activate an immune response^50–52. Conversely, the peripheral immune system can communicate with the CNS via circumventricular organs (CVOs), saturable influx transport and retrograde axonal transport mechanisms⁴⁹. CNS homeostasis is dependent upon a well-functioning immune system, not only for protection from toxins and pathogens, but also for synaptic signaling⁵³ and brain development⁵⁴. However, chronic or inappropriate immune behavior or communication might cause damage to the CNS leading to neuronal loss and impaired cognition⁵⁵. Immune dysregulation in the CNS is seen in many neuropsychiatric illnesses, including BD, SZ, and MDD⁵⁶. Similar processes of immune dysfunction may link these disorders and cross-diagnostic overlap may exist. The process by which this occurs is likely multi-factorial and evolves over the course of illness. Existing evidence supports the intriguing idea that peripheral leukocytes infiltrate into the CNS in a subset of patients with altered blood-CSF and BBB function (via increased CP volumes and ICAM-1). Additional sources of peripheral inflammation may exacerbate a primary, brain-based dysregulation, including lifestyle factors (e.g., diet, fitness, smoking, sleep, stress), medical comorbidities, and medications, among others.

It is yet unknown whether immune dysfunction is a cause or consequence of neuropsychiatric disease; however, evidence suggests that it may contribute both to risk for illness and to poor outcomes after illness onset. Both innate and adaptive immune responses likely play an important role in the pathophysiology of mood and psychotic disorders, and their influence on the barriers which protect the brain have yet to be fully elucidated. Future work focused on longitudinal studies which combine neuroimaging and peripheral measures of inflammation will be necessary to build causal models and identify specific treatment targets for intervention and prevention of illness onset.