University of Groningen
The Dysregulated Brain Haarman, Bartholomeus Cornelius Maria
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record
Publication date: 2017
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA): Haarman, B. C. M. (2017). The Dysregulated Brain: A psychoimmunological approach to bipolar disorder. University of Groningen.
Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.
Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Download date: 04-10-2021 The Dysregulated Brain A psychoimmunological approach to bipolar disorder B.C.M. Haarman The Dysregulated Brain A psychoimmunological approach to bipolar disorder
Illustration cover Alan Flemming Layout Marieke van der Vliet / dit-s.nl Printed by Pumbo.nl
The studies described in this thesis were performed at the Department of Psychiatry, University Medical Center Groningen, Groningen, The Netherlands; Department of Immunology, Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands; and Radiology Morphological Solutions, Berkel en Rodenrijs, The Netherlands.
The studies were financially supported by the European Commission EU-FP7-HEALTH-222963 ‘MOODINFLAME’ and EU-FP7-PEOPLE- 286334 ‘PSYCHAID’.
ISBN 978-90-824638-1-1 (printed version) 978-90-824638-2-8 (digital version)
© 2017 B.C.M. Haarman. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without written permission from the author. The Dysregulated Brain
A psychoimmunological approach to bipolar disorder
PhD thesis
to obtain the degree of PhD at the University of Groningen on the authority of the Rector Magnificus Prof. E. Sterken and in accordance with the decision by the College of Deans.
This thesis will be defended in public on
1 March 2017 at 16:15 hours
by
Bartholomeus Cornelius Maria Haarman
born on 7 August 1978 in Raalte Supervisor(s) Prof. W.A. Nolen Prof. H.A. Drexhage
Co-supervisor(s) Dr. R.F. Riemersma – Van der Lek Dr. H. Burger
Assessment committee Prof. A. Aleman Prof. V. Arolt Prof. J.D. Laman Table of Contents
CHAPTER 1 General background 7
PART 1 Peripheral immune system 25
CHAPTER 2 Relationship between clinical features and inflammation related monocyte gene expression in bipolar disorder 27
CHAPTER 3 Feature-expression heat maps 51
CHAPTER 4 Inflammatory monocyte gene expression 67
CHAPTER 5 Does CRP predict outcome in bipolar disorder in regular outpatient care? 81
PART 2 Neuroimmune system 99
CHAPTER 6 PET and SPECT in bipolar disorders 101
CHAPTER 7 Neuroinflammation in bipolar disorder 121
CHAPTER 8 Volume, metabolites and neuroinflammation of the hippocampus in bipolar disorder 141
CHAPTER 9 Diffusion tensor imaging in euthymic bipolar disorder 171
CHAPTER 10 Summary and general discussion 195
Dutch summary 223 List of publications 243 Acknowledgements 247 Curriculum vitae 253
CHAPTER 1 General background
Bartholomeus C.M. Haarman
Adapted from Haarman BCM, Riemersma - Van der Lek RF, Ruhé HG, Groot JC, Nolen WA, Doorduin J. Bipolar Disorders. In: Dierckx RAJO, Otte A, de Vries EFJ, Waarde A, den Boer JA, editors. PET and SPECT in Psychiatry. Heidelberg: Springer Berlin Heidelberg; 2014. p. 223–51. Introduction
Bipolar disorder: diagnostic description Bipolar disorder (BD)1 is a mood disorder characterized by episodic pathologic distur- bances in mood: (hypo)manic episodes and depressive episodes which alternate with euthymic periods, i.e. with normal mood. BD has to be distinguished from (unipolar) major depressive disorder (MDD), which is characterized by only depressive episodes. According to DSM-IV, the core criterion of a (hypo)manic episode is the occurrence of a pathologically elated (euphoria), expansive or irritable mood. DSM-5 added increased energy or activity to this list. In addition to these core criteria, there are other symptoms such as inflated self-esteem or grandiosity, decreased need for sleep, being more talkative than usual, flight of ideas, distractibility, increase in goal-directed activity or psychomotor agitation and excessive involvement in plea- surable activities that have a high potential for painful consequences. A depressive episode consists of at least one of the core symptoms: depressed mood and loss of interest or pleasure, completed with symptoms such as sleep problems, psychomotor changes, fatigue or loss of energy, feelings of worthlessness or excessive feelings of guilt, difficulty concentrating or making decisions and recurrent thoughts of death1. Two types of BD are recognized: bipolar I disorder (BD-I) and bipolar II disorder (BD-II), characterized by the occurrence of manic episode(s) or by only hypomanic episode(s), respectively. The difference between manic and hypomanic episodes (and thus between BD-I and BD-II) is that manic episodes are associated with marked impairment in occupational, relational or social functioning, which can lead to hospitalization, while hypomanic episodes do not have this marked impairment and do not lead to hospitalization. When manic and depressive symptoms co-occur (or alternate very quickly) in the same episode, in DSM-IV it is labeled as a mixed episode and in DSM-5 as a bipolar disorder, manic or depressive episode with mixed features. Manic, depressive and mixed episodes can also be complicated by the presence of concurrent psychotic symptoms. Besides the mood symptoms, many patients with BD also show cognitive dysfunctions which may persist during euthymic periods, and which involve disturbances in various domains such as attention, verbal memory and executive functioning2,3.
8 Epidemiology and burden of disease The lifetime prevalence of BD is about 2% across different countries, women being affected as frequently as men4,5. Across the world, the disorder is sixth among all 1 health conditions in terms of causing disability6 with poor clinical and functional 7 8 9 outcome , increased risk for suicidality and significant societal costs . It has been background General calculated that in the United States the average cost per case ranged from $11,720 to $624,785, based on the severity of the illness9. In the European countries societal costs for managing BD are considered to be high as well10–12.
Risk factors Established risk factors for BD include history of BD, and interestingly also other psychiatric disorders, in parents or siblings; severe childhood adversities; and excessive alcohol use, although the last two are viewed as triggers rather than causes in the presence of biological vulnerability13,14. A recent systematic review concluded that it is still unclear whether perinatal infection has a role in the etiology of BD15. This is particularly relevant in view of the present thesis.
Diagnostic and treatment complications Although the clinical picture seems clear at first glance, making the diagnosis is more complicated in practice. On average, there is a time lag of about 6 years between the first episode and the making of the right diagnosis, and another six years before the start of adequate treatment. This is in most cases partly impeded by the precedence of depressive episodes without obvious (hypo)manic symptoms at the beginning of the disease16. Because antidepressants appear less effective for the treatment of bipolar depressive episodes than for unipolar MDD17, delayed diagnosis often leads to prolonged illness and dysfunction.
Current pathophysiological models
It is generally accepted that the cause of BD is multifactorial, with multiple genes making someone vulnerable, and with psychological and social factors causing the genes to be expressed. Moreover, somatic factors are assumed to play a role. To unravel the complex interplay between genotype and phenotype researchers have tried to find intermediary processes that are related to both the underlying geno- type and the ultimate phenotype. Over the last 50 years several pathophysiological theories have been proposed for BD. Of these, we will shortly address the monoamine theory, the neuroinflammation theory, the white matter tract integrity disruption theory and the mitochondrial dysfunction theory.
9 Monoamine theory Since the 1960s, after the discovery of the first antipsychotic and antidepressant drugs, the monoamine theory has been the leading pathophysiological theory for various psychiatric disorders, including MDD and BD18–20. Based on the working mechanism of these drugs, disruption of serotonergic and noradrenergic neurotrans- mission in mood disorders and dopaminergic neurotransmission in schizophrenia underlies these disorders. In this regard noradrenaline is related to alertness and energy as well as to anxiety and attention; lack of serotonin to anxiety, obsessions, and compulsions; and dopamine to attention, motivation, pleasure, and reward19. Although the neurotransmitter theory originated in the discovery of psychotropic medication and has also given rise to the development of new pharmacological treatments, this model has also been criticized. The model is often regarded to be (too) simple, not explaining all patients’ symptoms, and having an effect that likely depends more on indirect effects, e.g. receptor changes21.
Corticolimbic dysregulation theory Based on extensive molecular imaging results (see chapter 6), complemented with functional MRI (fMRI) research, overall hyperactivation of limbic brain regions in BD patients relative to controls were displayed in a meta-analysis, along with an over- all hypoactivation of frontal regions22 (see figure 1). This corresponds to findings in other mood disorders, especially MDD, which is known as the corticolimbic theory of depression23. Hypo- and hyperactivity in frontal and limbic regions, respectively, was most pronounced in manic patients, although also present in depressed and euthymic ones. Depressed patients exhibit more pronounced hypoactivation of frontal regions than euthymic patients, whereas euthymic patients display, surprisingly, more hyper- activity in limbic regions than their depressed counterparts. The corticolimbic theory has some overlap with several neurological networks that have been described and are thought to lie at the basis of physiological emotional processing. These networks can be divided into circuits that lie within the cerebral cortex and those that extend to other parts of the brain24. The limbic-cortical-striatal-pallidal-thalamic (LCSPT) circuit connects the PFC to the limbic and subcortical areas of the brain25. This LCSPT circuit is thought to be particularly important in the mediation of emotional expression, because of its relation to visceral control structures26. The mood related cortico-cortical networks interact with and extend to the LCSPT27 via top-down inhibitory control28. The orbital prefrontal network consists of the central and caudal part of the orbital cortex and the ventrolateral PFC; it includes sensory association areas such as the visual associated areas in the inferior temporal cortex and somatic-sensory associated areas in the insula and frontal operculum, as well as the olfactory and taste areas. In addition to sensory integration, this system
10 FIGURE 1 Neuroanatomical regions important in mood disorders 1 anterior cingulate cortex
prefrontal cortex background General
subgenual cingulate
orbitofrontal cortex ventral striatum
hypothalamus pituitary raphe nuclei amygdala
hippocampus locus coeruleus
nervus vagus
Neuroanatomical regions important in mood disorders. (Adapted from Patrick J. Lynch, medical illustrator, and C. Carl Jaffe, MD, cardiologist76, under the Creative Commons Attribution 2.5 Generic license (CC BY 2.5)).
codes for affective characteristics of stimuli such as reward, aversion and relative value (salience)26. The medial prefrontal network of cortical areas includes the ventromedial PFC, the dorsolateral PFC, the anterior and posterior cingulate cortex, anterior temporal cortex and the enthorhinal and posterior parahippocampal corteces. This system does not have substantial sensory connections, but is a visceromotor system that is particu- larly involved in introspective functions such as mood and emotion, and in visceral reactions to emotional stimuli24. It is widely known as the “default system”, because in fMRI imaging it appeared activated as a network of areas that become inactive in most tasks that involve external attention29. It has been proposed that the “ventral” orbital prefrontal network and the “dorsal” medial prefrontal network are reciprocally connected and that the orbital PFC may mediate connections between higher-order dorsolateral prefrontal regions and subcortical limbic regions such as the amygdala during emotion regulation30.
11 Neuroinflammation theory The “macrophage theory of depression” postulates an aberrant pro-inflammatory state of monocytes/macrophages in patients with mood disorder, and considers this aberrant state of the cells as a driving force behind the illness31. The theory is founded on a higher frequency of autoimmune diseases in mood disorders, aberrant pro-inflammatory cytokines and elevated pro-inflammatory gene expression in monocytes. Autoimmune thyroiditis is considered to be an endophenotype of BD32. Patients with BD and MDD have a raised prevalence of autoimmune thyroiditis33–35. Not only BD patients, but also their offspring (affected as well as non-affected) and their monozygotic (affected and non-affected) and dizygotic (affected, but not as much unaffected) co-twins have a raised prevalence of autoimmune thyroiditis32,36. It was hypothesized that an activated inflammatory response system in monocytes consti- tutes the shared genetic susceptibility factor for both BD and thyroid autoimmunity; this has led to the extensive investigations of neopterine, IL-1β, IL-6 and TNF-α in mood disorders, and in particular in MDD. With regard to the serum concentration of these compounds, increased levels were also described in BD when compared to controls, although not in all studies37,38. To investigate the pro-inflammatory state of monocytes in a more precise and robust manner, a Q-PCR analysis of CD14+ purified monocytes was performed in which 22 mRNAs for inflammatory, chemokinesis/mo- tility, cell survival/apoptosis and MAP kinases pathway molecules were found to have an increased expression in BD patients compared to controls39. Interactions between the immune system and the HPA-axis, as well as interactions between the immune system and the neuronal system via indoleamine 2,3 dioxygen- ase (IDO) pathways have been suggested to result in mood disorder symptomatology. The HPA-axis is a complex set of direct influences and feedback interactions among the hypothalamus, the pituitary gland, and the adrenal glands that controls reactions to stress and regulates many body processes. The adrenal glands produce cortisol, which is a major stress hormone and has effects on many tissues in the body, includ- ing the brain, where it binds to glucocorticoid receptors in the PFC, the amygdala and the hippocampus40. Moreover, glucocorticoid insensitivity has been associated with a higher risk of developing a depressive episode40. In various in vivo and ex vivo studies a strong association between the activation of the inflammatory response system and glucocorticoid insensitivity has been demonstrated, linking at least in part the overproduction of pro-inflammatory cytokines to the HPA-axis disturbances in major mood disorders41–43. Tryptophan, the precursor amino acid of serotonin, can also be metabolized to downstream metabolites, known as kynurenines, via an alternative pathway. IDO is an oxygenase that catabolizes the first and rate-limiting step in this oxidative degradation. The IDO activity in monocytes/macrophages is enhanced by proin-
12 flammatory cytokines, e.g. during infections and when there is physical or mental stress44. Under such circumstances tryptophan breakdown is increased, making it less available for serotonin synthesis. When tryptophan is degraded, the next in vivo 1 product is kynurenine, which is the first metabolite of tryptophan45. This kynurenine
is again broken down into two pathways: (1) a neuroprotective, kynurenic acid, background General NMDA receptor antagonist pathway and (2) a neurotoxic 3-hydroxy kynurenine and quinolinic acid, NMDA receptor agonist pathway46. In the brain, this latter part of tryptophan catabolism, the kynurenine pathway, occurs in the astrocytes and microglia where astrocytes produce mainly neuroprotective kynurenic acid while macrophages produce mainly neurotoxic metabolites like quinolinic acid. Normally, formation of quinolinic acid is faster, while kynurenic acid has a counteractive pro- tective role against quinolinic acid47. Based on the above, a hypothesis was proposed, that an imbalance between the neurodegenerative and neuroprotective pathways leads to neurodegeneration and brings a person to a chronically depressive episode. This imbalance might be due either to a highly increased neurodegenerative pathway activity or to a lack of sufficient neuroprotective factor activity48. Tryptophan levels and the neuroprotective kynurenic acid were significantly decreased in MDD patients when compared to controls49. Also, in IFN-α treatment of hepatitis C patients, associated with depression and fatigue, IFN-α was found to up-regulate the expression of IDO50. Furthermore, the decrease of plasma tryptophan and the increase of kynurenine and neopterine during IFN-α treatment were found to correlate with the development of depression51,52. Molecular imaging can be of added importance in investigating the neuroinflam mation theory. Microglia are the central cells involved in immune regulation in the brain. When activated, these cells present the translocator protein (TSPO) on their mitochondrial membrane53. Using the positron emission tomography (PET) ligand 11C-PK11195, areas of microglia activation in the brain can be visualized. Besides in various neurological disorders, microglia activation has been found in schizophrenia, where a clear focus of neuroinflammation was found in the hippocampus54,55.
White matter microstructure integrity disturbances Interest in the white matter tracts in BD started with the observation of diffuse cor- tical and callosal white matter pathology in structural MRI studies in BD patients56,57. With the development of diffusion tensor imaging (DTI), a MRI technique allowing for the investigation of the preferred direction and rate of water diffusion, the integ- rity of the white matter microstructure can be investigated in more detail, because in the physiological situation water diffusion is restricted by the axonal structures58. The main parameters derived from DTI are the fractional anisotropy (FA) and mean diffusivity (MD). MD measures the magnitude of water molecule diffusion and FA is an index of the degree of directionality of water diffusivity. FA is reduced in diseased
13 states known to be associated with axonal loss and destruction of myelin sheaths in several diseases, e.g. multiple sclerosis, leukoencephalopathies and Alzheimer’s disease59. In BD most studies reported reduced FA and/or elevated MD, compared to controls, involving the prefrontal lobe, corpus callosum, internal capsule, uncinate fasciculus and superior and inferior longitudinal fasciculi, and suggesting a role for white matter microstructure disturbances in BD pathophysiology60. The studies focusing on the specific mood states of BD patients revealed FA to be altered in the different mood states61. In the euthymic state FA was usually found to be increased in the genu of corpus callosum, internal capsule, anterior thalamic radiation and uncinate fasciculus compared to controls, whereas during depressive episodes a lower FA has been shown in the genu of the corpus callosum and in corona radiate compared controls. In mixed samples, higher and lower FA values were found in different brain regions62. The place of white matter microstructure disturbances in the pathophysiology, with regard to other disease mechanisms, is still controversial. It has been suggested that FA changes could be related to inflammation related processes in BD, analogous to multiple sclerosis61.
Mitochondrial dysfunction Using various techniques, scientific evidence for a cellular energy metabolism disturbance has been presented. When observed in cell biological research, abnormal mitochondrial morphology is often linked to altered energy metabolism. In BD patients, compared to controls, mitochondria in neurons and fibroblasts have been reported to be smaller and concentrated proportionately more within the perinuclear region than in distal processes of the cells63. Conversely, patients with mitochondrial diseases have a higher lifetime prevalence of MDD (54%) or BD (17%) than the aver- age population64. Magnetic resonance spectroscopy (MRS) is a neuroimaging technique that allows the investigation of the metabolism on a cellular level. This MRI technique provides additional biochemical information of a selected voxel compared to a regular T1 or T2 image. The cellular metabolites are presumed to represent different cell func- tions: N-acetyl-aspertate (NAA) relates to cell viability and choline to cell membrane phospholipid integrity, and creatine is a measure of cellular metabolism65. Creatine plays an important role as a cell energy buffer, especially in high energy consum- ing cells such as muscular and brain cells. Using the creatine energy buffer reaction (figure 2), cells with an abundance of ATP can store energy by converting creatine to phosphocreatine. When in energy demanding circumstances the ATP stock becomes depleted, ATP can temporarily be supplied by reconverting phosphocreatine to creatine until the phosphocreatine stock is also depleted or energy is resupplied
14 FIGURE 2 Creatine energy buffer reaction 1 General background General
via other routes such as the oxidative phosphorylation. With 31P-MRS, creatine and phosphocreatine concentrations can be measured separately, as can the total concentration of both metabolites. The total concen- tration can also be measured with 1H-MRS, but the separate concentrations to a lesser degree when advanced quantification tools are being used. In BD patients a decreased phosphocreatine66 and reduced total creatine67,68 were described, when compared to controls, supporting the mitochondrial dysfunction theory. Findings in other MRS metabolites such as a reduced pH and increased lactate, exponents of cell metabolism exhaustion, add indirectly to this theory66,69. A study of the nature of the metabolic dysfunction revealed a paradoxical down regulation of mitochondria-related genes to glucose deprivation in fresh lymphocytes derived from BD patients, whereas control subjects showed an upregulation. This finding would suggest that patients with BD might have impairment in molecular adaptation to energy stress70. However, there is still debate as to whether this dys regulation is based on mitochondrial DNA disturbances, mitochondria-related nuclear DNA disturbances, or the effects of other mechanisms71.
15 Aim and outline of this thesis
Although multiple pathophysiological theories on BD exist, a comprehensive pathophysiological framework, integrating the various illness models for BD, is still lacking. Since immune cells play an important role, not only protecting neurons from pathogens but as microglia cells also helping in maintaining homeostasis in the brain72,73, they may play a key role in such a framework. This thesis aims to clarify the role of the immune system in the pathophysiology of BD via several different approaches, using bio-assay and neuroimaging techniques. The immune system consists of two parts: the peripheral immune system and the neuroimmune system. Structurally distinct from the peripheral immune system, the neuroimmune system is a system of structures and processes, involving the biochem- ical and electrophysiological interactions between the nervous system and immune system, which protect neurons from pathogens. Unlike the peripheral system, the neuroimmune system is composed primarily of glial cells, in particular microglia72–74.
Part 1: Peripheral immune system The first part of the thesis describes studies of the function of the peripheral immune system in BD, focusing on monocyte pro-inflammatory gene-expression and CRP, using bio assay techniques.
Association between monocyte gene-expression and clinical features Chapter 2 discusses the associations between monocyte pro-inflammatory gene expression and illness characteristics. We a priori hypothesized that pro-inflammatory gene expression would be found more frequently in BD patients with a lifetime history of psychotic symptoms. We go on to explore the associations with the individual manic and depressive symptoms of BD. Then we elucidate the course of the pro-inflammatory gene expression by investigating in more detail the relation between the pro-inflammatory gene expression and age at onset, and duration of illness. Finally, in chapter 2, we examine the association between the current use of psychotropic medication and monocyte pro-inflammatory gene expression. The feature-expression heat map, an in-house developed method used in chapter 2 to visualize the complex associations between symptoms and gene-expression, is explained in more detail in chapter 3.
Monocyte gene-expression: state or trait? In chapter 4 we present the results of the study among the bipolar cohort of the MOODINFLAME project75, in which we investigated whether monocyte pro-inflammatory gene-expression is more a state or more a trait marker. We compare the monocyte pro-inflammatory gene-expression in euthymic BD patients
16 with HC. Moreover, we present the results of a small additional study in which BD patients are compared during a mood episode and when euthymic. 1 Does CRP predict outcome in clinical practice?
Chapter 5 describes a historic cohort study in which we investigate whether, in background General a clinical setting, higher CRP levels at baseline may predict a worse BD outcome, defined as a shorter time to relapse or a longer time to recover, depending on the mood state at baseline.
Part 2: Neuroimmune system The second part of the thesis investigates the function of the neuroimmune sys- tem, focusing on microglia activation, and using PET, MRI, MRS, DTI neuroimaging techniques.
Previous PET/SPECT studies Chapter 6 describes the findings of previous PET / single-photon emission computed tomography (SPECT) research efforts in BD, based on the literature. First, we discuss the cerebral blood flow and cerebral metabolism findings, followed by the neurotransmitter studies. Finally, we summarize the most important conclusions, and follow with remarks about the observed molecular imaging study designs specific for BD.
Microglial activation in the hippocampus In chapter 7 we present the results of the first neuroinflammation PET study in BD. We aimed to demonstrate an increased [11C]-(R)-PK11195 binding to activated microglia in BD-I in comparison to a healthy control group. We a priori hypothesized the hippocampus to be the main focus of neuroinflammation in BD. In a second model we explore the presence of neuroinflammation in other brain regions.
Associations between volume, metabolites and microglial activation In chapter 8 we investigate the relations between volume, metabolites and microglial activation of the hippocampus in a contemporaneously executed PET/MRI study. We compare hippocampal volume and metabolites in BD-I patients with HC, using MRI and MRS. We a priori hypothesized hippocampal volume and the N-acetylaspartate (NAA) metabolite to be decreased in BD patients, compared to HC. Subsequently, within the BD-I and HC groups, we post-hoc investigate whether hippocampal volume and metabolites were associated with microglial activation. Furthermore, we explore if potential illness modifying factors such as duration of illness, medication use, body mass index (BMI), exercise, smoking, number of caffeine consumptions and alcohol use did affect these hippocampal measurements within the BD-I group, and whether
17 these hippocampal measurements were associated with experienced mood and functioning.
White matter microstructure disturbances and lithium usage Using DTI, in chapter 9 we investigate the white matter microstructure in BD and HC, and differences among patients in relation to lithium usage. We first com- pare estimates of white matter microstructure (fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), axial diffusivity (AD)) between euthymic BD-I patients and HC. We a priori hypothesized that we would find a widespread decrease in FA in several white matter tracts of BD-I patients compared to HC, associated with reciprocal alterations of other white matter microstructural parameters. Subsequently, we divide the patient group into lithium-users and non-lithium-users and analyze the estimates of white matter microstructure across these three groups (non-lithium-users, lithium-users, healthy controls). In this analysis, supposing a restoring effect of lithium on myelination, we a priori hypothesized FA to be increased - and consistently a decrease of other white matter microstructural parameters - in lithium-using patients compared to non-lithium-using patients, possibly even attaining healthy control values.
18 References
1. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 5th Edition. Arlington, VA: American Psychiatric Publishing, Inc; 2013. 1 2. Arts B, Jabben N, Krabbendam L, van Os J. Meta-analyses of cognitive functioning in euthymic
bipolar patients and their first-degree relatives. Psychol Med. 2008 Jun;38(6):771–85. background General 3. Bora E, Yucel M, Pantelis C. Cognitive endophenotypes of bipolar disorder: a meta-analysis of neuropsychological deficits in euthymic patients and their first-degree relatives. J Affect Disord. 2009 Feb;113(1-2):1–20. 4. Pini S, de Queiroz V, Pagnin D, Pezawas L, Angst J, Cassano GB, et al. Prevalence and burden of bipolar disorders in European countries. Eur Neuropsychopharmacol. 2005 Aug;15(4):425–34. 5. Merikangas KR, Akiskal HS, Angst J, Greenberg PE, Hirschfeld RMA, Petukhova M, et al. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey replication. Arch Gen Psychiatry. 2007 May;64(5):543–52. 6. World Health Organization. The world health report 2001: mental health: new understanding, new hope. World Health. Geneva, Switzerland; 2001. 7. Goodwin FK. Manic-Depressive Illness: Bipolar Disorders and Recurrent Depression. 2nd ed. New York: Oxford University Press; 2007. 8. Baldessarini RJ, Tondo L. Suicide risk and treatments for patients with bipolar disorder. JAMA. 2003 Sep 17;290(11):1517–9. 9. Begley CE, Annegers JF, Swann AC, Lewis C, Coan S, Schnapp WB, et al. The lifetime cost of bipolar disorder in the US: an estimate for new cases in 1998. Pharmacoeconomics. 2001 Jan;19(5 Pt 1):483–95. 10. Hakkaart-Van Roijen L, Hoeijenbos MB, Regeer EJ, Ten Have M, Nolen WA, Veraart CPWM, et al. The societal costs and quality of life of patients suffering from bipolar disorder in the Netherlands. Acta Psychiatr Scand. 2004;110(5):383–92. 11. Young AH, Rigney U, Shaw S, Emmas C, Thompson JM. Annual cost of managing bipolar disorder to the UK healthcare system. J Affect Disord. Elsevier B.V.; 2011;133(3):450–6. 12. Ekman M, Granström O, Omérov S, Jacob J, Landén M. The societal cost of bipolar disorder in Sweden. Soc Psychiatry Psychiatr Epidemiol. 2013;48(10):1601–10. 13. Rush AJ. Toward an understanding of bipolar disorder and its origin. J Clin Psychiatry. 2003;64 Suppl 6:4–8; discussion 28. 14. Post RM, Leverich GS, Xing G, Weiss RB. Developmental vulnerabilities to the onset and course of bipolar disorder. Dev Psychopathol. 2001;13(3):581–98. 15. Barichello T, Badawy M, Pitcher MR, Saigal P, Generoso JS, Goularte JA, et al. Exposure to Perinatal Infections and Bipolar Disorder: A Systematic Review. Curr Mol Med. 2016;16(2):106–18. 16. Suppes T, Leverich GS, Keck PE, Nolen W a, Denicoff KD, Altshuler LL, et al. The Stanley Foundation Bipolar Treatment Outcome Network. II. Demographics and illness characteristics of the first 261 patients. J Affect Disord. 2001 Dec;67(1-3):45–59. 17. Sachs GS, Nierenberg AA, Calabrese JR, Marangell LB, Wisniewski SR, Gyulai L, et al. Effectiveness of adjunctive antidepressant treatment for bipolar depression. N Engl J Med. 2007 Apr 26;356(17):1711–22. 18. Schildkraut JJ. The catecholamine hypothesis of affective disorders: a review of supporting evidence. Am J Psychiatry. 1965 Nov;122(5):509–22. 19. Nutt DJ. Relationship of neurotransmitters to the symptoms of major depressive disorder. J Clin Psychiatry. 2008;69 Suppl E:4–7. 20. Malhi GS, Tanious M, Das P, Coulston CM, Berk M. Potential mechanisms of action of lithium in bipolar disorder: Current understanding. CNS Drugs. 2013. p. 135–53. 21. Hirschfeld RM. History and evolution of the monoamine hypothesis of depression. J Clin Psychiatry. 2000 Jan;61 Suppl 6:4–6.
19 22. Kupferschmidt D a, Zakzanis KK. Toward a functional neuroanatomical signature of bipolar disorder: quantitative evidence from the neuroimaging literature. Psychiatry Res. Elsevier Ireland Ltd; 2011 Aug 30;193(2):71–9. 23. Mayberg HS. Limbic-cortical dysregulation: a proposed model of depression. J Neuropsychiatry Clin Neurosci. 1997 Jan;9(3):471–81. 24. Price JL, Drevets WC. Neurocircuitry of mood disorders. Neuropsychopharmacology. Nature Publishing Group; 2010 Jan;35(1):192–216. 25. al-Mousawi a. H, Evans N, Ebmeier KP, Roeda D, Chaloner F, Ashcroft GW. Limbic dysfunction in schizophrenia and mania. A study using 18F- labelled fluorodeoxyglucose and positron emission tomography. Br J Psychiatry. 1996 Oct 1;169(4):509–16. 26. Drevets WC, Price JL, Furey ML. Brain structural and functional abnormalities in mood disorders: implications for neurocircuitry models of depression. Brain Struct Funct. 2008 Sep;213(1-2):93–118. 27. Ongür D, Ferry AT, Price JL. Architectonic subdivision of the human orbital and medial prefrontal cortex. J Comp Neurol. 2003 Jun 2;460(3):425–49. 28. Savitz J, Drevets WC. Bipolar and major depressive disorder: neuroimaging the developmental- degenerative divide. Neurosci Biobehav Rev. 2009 May;33(5):699–771. 29. Gusnard DA, Akbudak E, Shulman GL, Raichle ME. Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. Proc Natl Acad Sci U S A. 2001 Mar 27;98(7):4259–64. 30. Phillips ML, Ladouceur CD, Drevets WC. A neural model of voluntary and automatic emotion regulation: implications for understanding the pathophysiology and neurodevelopment of bipolar disorder. Mol Psychiatry. 2008 Sep;13(9):829, 833–57. 31. Smith RS. The macrophage theory of depression. Med Hypotheses. 1991 Aug;35(4):298–306. 32. Vonk R, van der Schot AC, Kahn RS, Nolen W a, Drexhage H a. Is autoimmune thyroiditis part of the genetic vulnerability (or an endophenotype) for bipolar disorder? Biol Psychiatry. 2007 Jul 15;62(2):135–40. 33. Bunevicius R, Peceliuniene J, Mickuviene N, Bunevicius A, Pop VJ, Girdler SS. Mood and thyroid immunity assessed by ultrasonographic imaging in a primary health care. J Affect Disord. 2007 Jan;97(1-3):85–90. 34. Carta MG, Loviselli A, Hardoy MC, Massa S, Cadeddu M, Sardu C, et al. The link between thyroid autoimmunity (antithyroid peroxidase autoantibodies) with anxiety and mood disorders in the community: a field of interest for public health in the future. BMC Psychiatry. 2004 Jan;4:25. 35. Benros ME, Waltoft BL, Nordentoft M, Ostergaard SD, Eaton WW, Krogh J, et al. Autoimmune diseases and severe infections as risk factors for mood disorders: A nationwide study. JAMA Psychiatry. 2013;70(8):812–20. 36. Hillegers MH, Reichart CG, Wals M, Verhulst FC, Ormel J, Nolen W a. Five-year prospective outcome of psychopathology in the adolescent offspring of bipolar parents. Bipolar Disord. 2005 Aug;7(4):344–50. 37. O’Brien SM, Scully P, Scott L V, Dinan TG. Cytokine profiles in bipolar affective disorder: focus on acutely ill patients. J Affect Disord. 2006 Feb;90(2-3):263–7. 38. Hoekstra R, Fekkes D, Pepplinkhuizen L, Loonen AJM, Tuinier S, Verhoeven WMA. Nitric oxide and neopterin in bipolar affective disorder. Neuropsychobiology. 2006 Jan;54(1):75–81. 39. Padmos RC, Hillegers MHJ, Knijff EM, Vonk R, Bouvy A, Staal FJT, et al. A discriminating messenger RNA signature for bipolar disorder formed by an aberrant expression of inflammatory genes in monocytes. Arch Gen Psychiatry. 2008 Apr;65(4):395–407. 40. Spijker a T, van Rossum EFC. Glucocorticoid sensitivity in mood disorders. Neuroendocrinology. 2012 Jan;95(3):179–86. 41. Pariante CM, Pearce BD, Pisell TL, Sanchez CI, Po C, Su C, et al. The proinflammatory cytokine, interleukin-1alpha, reduces glucocorticoid receptor translocation and function. Endocrinology. 1999 Sep;140(9):4359–66.
20 42. Almawi WY, Lipman ML, Stevens AC, Zanker B, Hadro ET, Strom TB. Abrogation of glucocorticoid- mediated inhibition of T cell proliferation by the synergistic action of IL-1, IL-6, and IFN-gamma. J Immunol. 1991 May 15;146(10):3523–7. 43. Ito K, Chung KF, Adcock IM. Update on glucocorticoid action and resistance. J Allergy Clin 1 Immunol. 2006 Mar;117(3):522–43. 44. Babcock TA, Carlin JM. Transcriptional activation of indoleamine dioxygenase by interleukin 1 and
tumor necrosis factor alpha in interferon-treated epithelial cells. Cytokine. background General 2000 Jun;12(6):588–94. 45. Bender DA, McCreanor GM. Kynurenine hydroxylase: a potential rate-limiting enzyme in tryptophan metabolism. Biochem Soc Trans. 1985 Apr;13(2):441–3. 46. Chiarugi A, Calvani M, Meli E, Traggiai E, Moroni F. Synthesis and release of neurotoxic kynurenine metabolites by human monocyte-derived macrophages. J Neuroimmunol. 2001 Nov 1;120(1- 2):190–8. 47. Perkins MN, Stone TW. An iontophoretic investigation of the actions of convulsant kynurenines and their interaction with the endogenous excitant quinolinic acid. Brain Res. 1982 Sep 9;247(1):184–7. 48. Myint AM, Kim YK. Cytokine-serotonin interaction through IDO: a neurodegeneration hypothesis of depression. Med Hypotheses. 2003;61(5-6):519–25. 49. Myint A-M, Kim YK, Verkerk R, Scharpé S, Steinbusch H, Leonard B. Kynurenine pathway in major depression: evidence of impaired neuroprotection. J Affect Disord. 2007 Feb;98(1-2):143–51. 50. Curreli S, Romerio F, Mirandola P, Barion P, Bemis K, Zella D. Human primary CD4 + T cells activated in the presence of IFN-alpha 2b express functional indoleamine 2,3-dioxygenase. J Interferon Cytokine Res. 2001 Jun;21(6):431–7. 51. Capuron L, Ravaud A, Neveu PJ, Miller AH, Maes M, Dantzer R. Association between decreased serum tryptophan concentrations and depressive symptoms in cancer patients undergoing cytokine therapy. Mol Psychiatry. 2002 Jan;7(5):468–73. 52. Wichers MC, Koek GH, Robaeys G, Verkerk R, Scharpé S, Maes M. IDO and interferon-alpha- induced depressive symptoms: a shift in hypothesis from tryptophan depletion to neurotoxicity. Mol Psychiatry. 2005 Jun;10(6):538–44. 53. Doorduin J, de Vries EFJ, Dierckx RA, Klein HC. PET imaging of the peripheral benzodiazepine receptor: monitoring disease progression and therapy response in neurodegenerative disorders. Curr Pharm Des. 2008 Jan;14(31):3297–315. 54. van Berckel BN, Bossong MG, Boellaard R, Kloet R, Schuitemaker A, Caspers E, et al. Microglia activation in recent-onset schizophrenia: a quantitative (R)-[11C]PK11195 positron emission tomography study. Biol Psychiatry. 2008 Nov 1;64(9):820–2. 55. Doorduin J, de Vries EFJ, Willemsen ATM, de Groot JC, Dierckx R a, Klein HC. Neuroinflammation in schizophrenia-related psychosis: a PET study. J Nucl Med. 2009 Nov;50(11):1801–7. 56. Kempton MJ, Geddes JR, Ettinger U, Williams SCR, Grasby PM. Meta-analysis, database, and meta-regression of 98 structural imaging studies in bipolar disorder. Arch Gen Psychiatry. 2008 Sep;65(9):1017–32. 57. Vita A, De Peri L, Sacchetti E. Gray matter, white matter, brain, and intracranial volumes in first-episode bipolar disorder: a meta-analysis of magnetic resonance imaging studies. Bipolar Disord. 2009 Dec;11(8):807–14. 58. Le Bihan D. Molecular diffusion, tissue microdynamics and microstructure. NMR Biomed. 1996;8(7-8):375–86. 59. Le Bihan D. Looking into the functional architecture of the brain with diffusion MRI. Nat Rev Neurosci. 2003 Jun;4(6):469–80. 60. Heng S, Song AW, Sim K. White matter abnormalities in bipolar disorder: insights from diffusion tensor imaging studies. J Neural Transm. 2010 May;117(5):639–54.
21 61. Zanetti M V, Jackowski MP, Versace A, Almeida JRC, Hassel S, Duran FLS, et al. State-dependent microstructural white matter changes in bipolar I depression. Eur Arch Psychiatry Clin Neurosci. 2009 Sep;259(6):316–28. 62. Bellani M, Brambilla P. Diffusion imaging studies of white matter integrity in bipolar disorder. Epidemiol Psychiatr Sci. 2011 Jun;20(2):137–40. 63. Cataldo AM, McPhie DL, Lange NT, Punzell S, Elmiligy S, Ye NZ, et al. Abnormalities in mitochondrial structure in cells from patients with bipolar disorder. Am J Pathol. American Society for Investigative Pathology; 2010 Aug;177(2):575–85. 64. Fattal O, Link J, Quinn K, Cohen BH, Franco K. Psychiatric comorbidity in 36 adults with mitochondrial cytopathies. CNS Spectr. 2007 Jun;12(6):429–38. 65. Gillard JH, Waldman AD, Barker PB, editors. Clinical MR Neuroimaging. Spectroscopy. Cambridge: Cambridge University Press; 2004. 66. Kato T, Takahashi S, Shioiri T, Inubushi T. Alterations in brain phosphorous metabolism in bipolar disorder detected by in vivo 31P and 7Li magnetic resonance spectroscopy. J Affect Disord. 1993 Jan;27(1):53–9. 67. Port JD, Unal SS, Mrazek D a, Marcus SM. Metabolic alterations in medication-free patients with bipolar disorder: a 3T CSF-corrected magnetic resonance spectroscopic imaging study. Psychiatry Res. 2008 Feb 28;162(2):113–21. 68. Frey, Stanley J a, Nery FG, Monkul ES, Nicoletti M a, Chen H-H, et al. Abnormal cellular energy and phospholipid metabolism in the left dorsolateral prefrontal cortex of medication-free individuals with bipolar disorder: an in vivo 1H MRS study. Bipolar Disord. 2007 Jun;9 Suppl 1(1):119–27. 69. Dager SR, Friedman SD, Parow A, Demopulos C, Stoll AL, Lyoo IK, et al. Brain metabolic alterations in medication-free patients with bipolar disorder. Arch Gen Psychiatry. 2004 May;61(5):450–8. 70. Naydenov A V, MacDonald ML, Ongur D, Konradi C. Differences in lymphocyte electron transport gene expression levels between subjects with bipolar disorder and normal controls in response to glucose deprivation stress. Arch Gen Psychiatry. 2007 May;64(5):555–64. 71. Kato T. Molecular neurobiology of bipolar disorder: a disease of “mood-stabilizing neurons”? Trends Neurosci. 2008 Oct;31(10):495–503. 72. Beumer W, Gibney SM, Drexhage RC, Pont-Lezica L, Doorduin J, Klein HC, et al. The immune theory of psychiatric diseases: a key role for activated microglia and circulating monocytes. J Leukoc Biol. 2012 Aug 8;92(September):1–17. 73. Stertz L, Magalhães PVS, Kapczinski F. Is bipolar disorder an inflammatory condition? The relevance of microglial activation. Curr Opin Psychiatry. 2013 Jan;26(1):19–26. 74. Eyre H, Baune BT. Neuroplastic changes in depression: A role for the immune system. Psychoneuroendocrinology. Elsevier Ltd; 2012;37(9):1397–416. 75. MOODINFLAME website. http://moodinflame.eu. 2014. 76. Lynch PJ, Jaffe CC. Brain human sagittal section [Internet]. Wikimedia Commons. 2006. Available from: https://commons.wikimedia.org/wiki/File:Brain_human_sagittal_section.svg?uselang=en
22 1 General background General
23
PART 1 Peripheral immune system
CHAPTER 2 Relationship between clinical features and inflammation related monocyte gene expression in bipolar disorder Towards a better understanding of psychoimmunological interactions
Bartholomeus C.M. Haarman, Rixt F. Riemersma-Van der Lek, Huibert Burger, Mina Netkova, Roosmarijn C. Drexhage, Florian Bootsman, Esther Mesman, Manon H.J. Hillegers, Anne T. Spijker, Erik Hoencamp, Hemmo A. Drexhage, Willem A. Nolen
Bipolar Disorders. 2014 Mar 29;16(2):137–50. Abstract
Objectives Existing and previously published datasets were examined for associations between illness and treatment characteristics and monocyte pro-inflammatory gene expres- sion in bipolar disorder (BD) patients. We a priori hypothesized that increased monocyte pro-inflammatory gene expres- sion would be found more frequently in patients with a lifetime history of psychotic symptoms.
Methods Monocyte QPCR and symptom data of 64 BD patients were collected from three Dutch studies. Regression analyses were performed to analyze the various associa- tions of which feature-expression heat maps were drawn.
Results Symptoms No associations were found with lifetime psychotic symptoms, while a positive association was identified between sub-cluster 2 genes and manic symptoms.
Age at onset / duration of illness For several sub-cluster 1a genes a negative association was found with age at onset. For most sub-cluster 2 genes a positive association was found with the duration of illness.
Medication Current use of antidepressants and of anti-epileptics were associated with sub-cluster 2 gene expression, and current use of lithium and antipsychotics with sub-cluster 1a gene expression.
Conclusions Our hypothesis that lifetime psychotic features would be associated with the pro-inflammatory monocyte gene expression was not confirmed. In an explorative analysis we found: (1) a possible relation between pro-inflammatory gene expression and manic symptomatology, (2) a differential immune activation related to age at onset and duration of illness, and (3) support for the concept of an immune suppres- sive action of some of the mood regulating medications.
28 Introduction
The pathophysiology of bipolar disorder (BD) is complex. While there is no doubt that both genetic predisposition and environmental factors play a role, it remains import- ant to further investigate their neurobiological underpinnings and their interaction1. As both the stress system and the immune system interact with the brain and are influenced by the environment, they can be regarded as a linking pin. The “mono- 2 cyte-T-cell theory of mood disorders”2 considers an activated inflammatory response system (IRS) in mood disorders as a driving force behind the illness. IRS activation disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship can be regarded as a disbalance in immune regulatory processes. Pro-inflamma- tory cytokines are capable of destabilizing brain function3, which makes the brain vulnerable to stress and possibly other yet unknown factors with mood disturbances as the consequence. Padmos et al.4 described a sensitive quantitative polymerase chain reaction (Q-PCR) assay system to detect the pro-inflammatory state of circulating monocytes of naturalistically treated patients with BD and detected in the monocytes a coherent, mutually correlating set of 19 aberrantly expressed inflammatory genes (‘a pro-inflammatory signature’), supporting the concept of an activated IRS in mood disorders2. We expect the IRS to be an intermediate process between genotype and phenotype (figure 1). The pro-inflammatory signature occurred in 55% of BD patients versus 18% in healthy controls.
FIGURE 1 Schematic presentation of the monocyte-T-cell theory of mood disorders in BD
Genetic Stress vulnerability
Activated Medication - inflammatory response system
Intermediary mechanisms
Bipolar disorder Other (auto-immune) phenotype pathology
29 In a subsequent study, Drexhage et al.5 found nine inflammation related schizo- phrenia (SZ) genes using the same method, indicating that monocytes of SZ and BD patients partly overlap, but also differ in inflammatory gene expression. Combined with six specific auto-immune genes, related to BD, this resulted in a set of 34 genes (table 1). Moreover, via cluster analysis they identified the expression of sub-clusters 1a en 1b within the signature genes, relating to core inflammation and transcription, and sub-cluster 2, relating to adhesion, motility and chemotaxis (figure 2). Several of these 34 genes were also investigated in other studies in BD. Savitz et al. recently demonstrated an increased expression of TNF and eleven other genes in monocytes of a combined sample of depressed BD patients and major depressive disorder patients compared to healthy controls. On the genomic level CCL2 was found to be weakly overexpressed in BD patients in a Nordic genome wide association study. PDE4B6,7, IL68, IL19, TNF8,10,11, EGR312,13, CCL214–16 and FABP17 were investigated in gene polymorphism studies.
TABLE 1 Description of genes
Gene symbol Name of corresponding protein DUSP2 Dual specificity protein phosphatase 2 ATF3 Cyclic AMP-dependent transcription factor 3 PDE4B cAMP-specific 3',5'-cyclic phosphodiesterase 4B IL6 Interleukin 6
IL1 Interleukin 1 TNF Tumor necrosis factor TNFAIP3 Tumor necrosis factor, alpha-induced protein 3 BCL2A1 B-cell lymphoma-2-related protein A1 PTX3 Pentraxin-related protein 3 Inflammation PTGS Prostaglandin G/H synthase (cyclooxygenase) CCL20 C-C chemokine ligand 20 CXCL2 C-X-C chemokine ligand 2 EREG Epiregulin CXCL3 C-X-C chemokine ligand 3 MXD MAD protein
EGR3 Early growth response protein 3 F3 Tissue factor 3 MAFF Musculoaponeurotic fibrosarcoma oncogene homolog F THBS Thrombospondin 1
Transcription SERPINB2 Plasminogen activator inhibitor-2 RGC32 Response gene to complement 32 protein PTPN Protein tyrosine phosphatase, non-receptor type 7 NAB2 Nerve growth factor-induced protein A binding protein 2 MAPK6 Mitogen-activated protein kinase 6 EMP1 Epithelial membrane protein 1 STX1 Syntaxin-1A DHRS3 Short-chain dehydrogenase/reductase 3
chemotaxis CCL2 C-C chemokine ligand 2 CCL7 C-C chemokine ligand 7 Adhesion / motility / / motility / Adhesion CDC42 Cell division control protein 42 homolog
FABP Fatty acid-binding protein 5 CD9 Cluster of differentiation 9 antigen HSPA1 Heat shock 70 kDa protein 1 Other CCR2 C-C chemokine receptor type 2
30 FIGURE 2 The hypothetical relationship of the molecules of the pro-inflammatory gene expression in monocytes of BD patients 2 Relationship between clinical features and inflammation related monocyte gene expression in bipolar disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship
These molecules have regulatory functions in the monocyte of which some are predominantly active in inflammation (sub-cluster 1a, squared, e.g. TNF, IL1, IL6), others in regulating transcription (sub-cluster 1b, diamonds, e.g. EGR3, MXD), whereas others have their main function in cell movement functions such as adhesion, motility or chemotaxis (sub-cluster 2, encircled, e.g. CCL2, CCL7). Additional abbreviations: Gα and Gq, G protein α and q; cAMP, Cyclic adenosine monophosphate; PKA and PKC, protein kinase A and C; raf, rapidly accelerated fibrosarcom; TNF-R, tumor necrosis factor receptor; p38, protein 38; jnks, jun N-terminal kinase; elk-1, ETS domain- containing protein; egr-1, early growth response protein 1; GRE, glucocorticoid response element; RXR, retinoid X receptor; RAR, retinoid acid receptor; PPAR, peroxisome proliferator-activated receptor. (Adapted from Drexhage55, reprinted with permission)
Although in the studies by Padmos and Drexhage the BD patients as a group were characterized by a positive pro-inflammatory signature, a proportion of BD patients did not show a positive signature. This led to the subsequent question whether specific clinical characteristics in BD patients would be associated with the pro-inflammatory gene expression. In view of the aforementioned overlap of pro-inflammatory gene expression between BD and SZ we a priori hypothesized that the pro-inflammatory gene expression would be found more frequently in BD patients with a lifetime history of psychotic symptoms. Secondly, we explored the associations with the individual manic and depressive symptoms of BD, extending on previous findings relating the mood state to specific
31 pro-inflammatory gene expression changes4. Thirdly, we elucidated the course of the pro-inflammatory gene expression by inves- tigating the relation between the pro-inflammatory gene expression and age at onset and duration of illness in more detail. We presumed that patients with an earlier age at onset would have a biologically more severe form of the illness, therefore demon- strating an increased pro-inflammatory gene expression compared to patients with a later age at onset. Furthermore, we expected the pathogenic processes to cause progressing derailment of the illness, demonstrating a positive association between pro-inflammatory gene expression and duration of illness irrespective of age at onset.
Finally, we examined the association between current use of psychotropic medication and monocyte pro-inflammatory gene expression, as (part of) these medications have been reported to be immune suppressive in nature5,18–21.
Patients and Methods
Participants Study participants comprised of patients and healthy controls from three different studies4,5,22–25. All patients (n=64) had a DSM-IV BD I or II disorder, confirmed by the Structured Clinical Interview for DSM-IV axis I disorders (SCID-I)26 (table 2 and figure 3). For this extension analysis we selected all participants (64 outpatients, age range 16-61 years, 24 (38%) male) who were included in the original studies and of whom Q-PCR and questionnaire data were available. Regarding the twin study, only data from the index twin were used. There was no family relation between any of the participants. Patients needed to be free of any additional severe psychiatric disorders and of any relevant medical illness that might affect inflammation status at least 2 weeks before blood withdrawal. Healthy controls (n=63) were also recruited within the three aforementioned studies and comprised of laboratory or medical staff, medical students and high school students. All controls were free of any lifetime psychiatric disorder and of a history of these disorders in first-degree family members (self report). Controls did not use any psychotropic or other medication and were also free of any relevant medical illness that might affect inflammation status at least 2 weeks before blood withdrawal.
Ethical considerations The Medical Ethical Review Committee of the University Medical Center Utrecht approved the original studies, which were performed in accordance with the Helsinki Declaration of 1975. Written informed consent was obtained from all participants.
32 TABLE 2 Characteristics of the participants
Bipolar disorder Healthy controls Group size 64 63 DSM IV classification Bipolar I disorder 52 (81%) Bipolar II disorder 12 (19%) Age (yr) 40 (16-61) 40 (16-58) Gender 2 Male 24 (38%) 22 (34%) Female 40 (62%) 39 (66%)
Duration of illness (mean (range), yr) 15 (3-38) disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship Age at onset (mean (range), yr) 25 (6-50) Medication (n=58) Antidepressants 19/58 (33%) Lihium 43/58 (74%) Antipsychotics 9/58 (16%) Valproate 12/58 (21%) Carbamazepine 6/58 (10%) Lamotrigine 2/58 (3%) Benzodiazepines 18/58 (31%)
FIGURE 3 Patient flow diagram
58 BD patients 41 BD twin pairs 54 children of BD parents Stanley Foundation Bipolar Dutch Twin study Dutch BD Offspring study Network, Dutch site
41 index BD patients with QPCR 19 BD patients with QPCR and 4 BD patients with QPCR and and questionnaire data available questionnaire data available questionnaire data available
64 patients
62 duration of 56 item level 58 item level 58 medication 57 item level illness IDS YMRS lists SCID-I measurements
We used the data of forty-one patients who took part in an ongoing Dutch twin study on BD22, nineteen patients participated in the Dutch site of the former Stanley Foundation Bipolar Network (SFBN) 56, and four patients participated in a ongoing prospective study among the adolescent offspring of BD patients in the Netherlands24,25.
Assessments Clinical features used in the analyses were extracted from the interviews held, which included the SCID-I, the Young Mania Rating Scale (YMRS), the Inventory of Depressive Symptoms (IDS) and a demographic questionnaire. The lifetime history of psychotic symptoms was derived from the SCID-I. The YMRS is an eleven-item,
33 multiple-choice questionnaire to assess manic symptoms27. The IDS is a thirty-item, multiple-choice questionnaire to assess depressive symptoms of all symptom domains of depression28. Individual YMRS and IDS item scores were used to qualify the presence of specific mood symptoms. The scores of the individual questionnaire items were transformed to a 0-1 scale to facilitate mutual comparison. Age at onset was defined as the age when the first mood episode occurred. Current medication use was dichotomously described as being present or absent.
Laboratory methods Blood (drawn in the morning and within three days of psychiatric assessment in all cases) was collected in clotting tubes for serum preparation and stored at -80°C, and in sodium-heparin tubes for immune cell preparation. From the heparinized blood, peripheral blood mono-nuclear cell (PBMC) suspensions were prepared in the after- noon by low-density gradient centrifugation within 8 hours to avoid activation of the monocytes29. To obtain cDNA for Q-PCR, 1 mg RNA was reversed-transcribed using the cDNA high-capacity cDNA Reverse Transcription kit (Applied Biosystems, USA). Q-PCR was performed on thirty-four genes selected by whole genome profiling, previously described by R. Drexhage et al.5, using a method described by Staal et al.30.
Preprocessing The quantitative value obtained from Q-PCR is a cycle threshold (Ct) that was used to calculate household-gene (Abl) normalized Ct values (ΔCt = Ct gene – Ct house- keeping gene) via the ΔΔCt method31. By subtracting the ΔCt with the mean ΔCt for the healthy control group, the relative gene expression (ΔΔCt) was determined. The healthy controls were solely used for this purpose and were not needed in the other analyses. Z-score transformation of the relative gene expression was applied to facili- tate mutual comparison. The relative gene expression can consequently be expressed as fold change after transformation via the the ΔΔCt method (fold change = 2-ΔΔCt)31.
Statistical analyses Regression methods were used to determine size and statistical significance of the association between gene expression and clinical features. Analyses were performed using individual immunologic details and individual psychiatric details. These regarded associations between specific mRNA gene expressions e.g. PDE4B, ATF3 and specific psychiatric symptoms, e.g. auditory hallucinations, depressed mood. The associations between pro-inflammatory gene expression and lifetime history of psychotic symptoms were analyzed using ordinal regression, taking lifetime history of psychotic symptoms as dependent variable, in accordance with our pathophysio- logical model (figure 1).
34 Associations of pro-inflammatory gene expression with current manic symptoms and with current depressive symptoms were analyzed using ordinal regression, taking symptoms as dependent variable. Associations between age at onset and pro-inflammatory gene expression and between duration of illness and pro-inflammatory gene expression were analyzed using linear regression. This was done separately in univariable and in combined multivariable models without and with an interaction term for effect of age at onset 2 on the duration of illness. Theses analyses were repeated while correcting for gender. Pro-inflammatory gene expression was the dependent variable. disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship Associations between pro-inflammatory gene expression and medications were analyzed with linear regression using the pro-inflammatory gene expression as dependent variable. Continuous outcome measures (relative gene expression, duration of illness) were checked for normal distribution by graphical inspection and using the skewness and kurtosis normality test. Univariable and multivariable linear regression were used for continuous, normally distributed outcome variables (disease duration, gene expres- sion in the medication analyses). Because of the multiple individual statistical tests in the analyses, there is an increased risk of a wrongful rejection of the null hypothesis (type I error). To control for this problem, we applied correction for the false discovery rate (FDR), as described by Benjamini-Hochberg32, in most analyses and only considered clustering groups of associations to be of importance. Statistical analyses were performed using Stata Statistical Software, release 11 (StataCorp. 2009, College Station, TX).
Feature-expression heat maps To display the associations we have drawn adapted heat maps, visualizing the inde- pendent gene expression variable on the vertical y-axis and the dependent clinical symptom variables on the horizontal x-axis. In these feature-expression heat maps the associations between independent and dependent variables are represented by circles in each respective compartment. The regression coefficients indicating effect size are represented by the type (red=positive, blue=negative) and intensity of the color, whereas the statistical significance of the analyses is represented by the radius of the circles. To facilitate the visual identification of meaningful clusters of association both axes were ordered into functional categories and sequences. Monocyte gene expression was ordered on the sub-clusters 1a (inflammation), 1b (transcription) and 2 (adhesion, motility and chemotaxis) categories, based on a cluster analyses performed by R.C. Drexhage5. Clinical symptoms were ordered into symptom categories. For psychot- ic symptoms a division into delusions, hallucinations and psychomotor symptoms
35 is used. Depressive symptoms were divided into general symptoms, melancholic symptoms and atypical symptoms. Manic symptoms were not subdivided into cate- gories. Furthermore, symptoms were ordered into a phenomenological sequence for mood symptoms (core mood symptoms, thought symptoms, psychosomatic symp- toms, motor symptoms, food intake symptoms, sleep symptoms, higher functional symptoms) and psychotic symptoms. The heat maps were drawn with the corrplot package, by T. Wei33, on R, release 2.14.1 (R Development Core Team 2011, Vienna, Austria). Centered dots were added to the compartments that complied with a FDR below 0.2, thus allowing 1/5 to be false positive. See chapter 3 for a more detailed description of the feature-expression heat map method.
Results
Associations between lifetime psychotic symptoms and monocyte pro-inflammatory gene expression The associations between the psychotic symptoms and the expression of genes belonging to each subcluster were depicted in the psychosis feature-expression heat map (figure 4). This heat map demonstrated a clustering group of associations between the sub-cluster 2 genes and the psychomotor symptoms. Clustering groups of associations with delusional or hallucinatory symptoms could not be demon strated. None of the individual analyses were statistically significant after FDR correction. Although not fitting into a clustering group of associations, the negative association between thought withdrawal and sub-cluster 1 gene expression, and the positive association between olfactory hallucinations and sub-cluster 1 gene expression are conspicuous.
Associations between current mood symptoms and monocyte pro-inflammatory gene expression The manic feature-expression heat map (figure 5) was created of the associations between manic symptoms and the expression of genes belonging to each subcluster. This feature-expression heat map showed a clustering group of associations between the sub-cluster 2 genes and manic symptoms. Sixteen of these associations were significant after FDR correction, especially in the associations with the symptoms increased speech and increased motor activity. The associations between depressive symptoms and the expression of genes belong- ing to each subcluster were drawn up in the depression feature-expression heat map (figure 6). With regard to this feature-expression heat map, clustering groups of as- sociations could not be identified and none of the individual analyses were significant
36 FIGURE 4 Association between lifetime psychotic symptoms and monocyte pro-inflammatory gene expression
s t 2
elusions of being controlled DelusionsPersecutory of referenceGrandiose delusionsSomatic delusionsReligious delusionsDelusions delusionsD of guilThoughtThought insertionThought withdrawaAuditory broadcastingl Visual hallucinations hallucinationTactile GustatoryhallucinationsOlfactory hallucinationsExcessive hallucinationsEcholalia motorGrossly oractivity echopraxiaDisorganized disorganized speechbehavior disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship DUSP2 ATF3 PDE4B IL6 p IL1B 0.001 TNF 0.005 TNFAIP3 0.01 BCL2A1 0.05 PTX 0.1 PTGS inflammation 0.2 CCL20 CXCL2 EREG effect size CXCL3 3 MXD
EGR3 2 F3 MAFF 1 THBS
SERP transcription RGC32 0 PTPN NAB2 -1 MAPK EMP1 STX1 -2
SDR motility / adhesion / chemotaxis CCL2 -3 CCL7 CDC42 FABP CD9 HSPA1 CCR2 delusions hallucinations psychomotor psychotic symptoms
Heat map depicting the regression coefficient of the association between lifetime psychotic symptoms and gene expression. Psychotic symptoms were ordinally measured. gene expression was expressed as z-transformed –∆∆Ct. Statistical analysis was performed using ordered logistic regression. Contrast was set to 3. Blank compartments either represent a small regression coefficient and statistical probability or an analysis with insufficient observations. None of the analyses were significant below the 0.2 false discovery rate (FDR) threshold for multiple testing.
37 FIGURE 5 Association between manic symptoms and monocyte pro-inflammatory gene expression
r
ontent ElevatedIrritability MoodC LanguageIncreased / ThoughtIncreased Motor DisorderDecreased ActivitySpeechDisruptive orSlee EnergSexualp ory AggressiveAppearance InterestInsight Behavio DUSP2 ATF3 p PDE4B 0.001 IL6 0.005 IL1B 0.01 TNF 0.05 TNFAIP3 0.1 BCL2A1 0.2 PTX inflammation PTGS FDR<0.2 CCL20 CXCL2 EREG CXCL3 MXD
EGR3 effect F3 size 3 MAFF THBS 2 SERP transcription RGC32 PTPN 1 NAB2 MAPK 0 EMP1 STX1 -1
SDR motility / adhesion / chemotaxis CCL2 CCL7 -2 CDC42 FABP CD9 -3 HSPA1 CCR2 manic symptoms
Heat map depicting the regression coefficient of the association between actual manic symptoms (within three days of blood sampling) and gene expression. Depressive symptoms were ordinally measured. gene expression was expressed as z-transformed –∆∆Ct. Statistical analysis was performed using ordered logistic regression. Contrast was set to 3. Blank compartments either represent a small regression coefficient and statistical probability or an analysis with insufficient observations. Dotted circles represent significance below the 0.2 false discovery rate (FDR) threshold for multiple testing.
38 FIGURE 6 Association between depressive symptoms and monocyte pro-inflammatory gene expression
a
omni t 2 n d
Mood s utlook disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship O
Sad MoodIrritableAnxiou MoodSomaticSympathetic ComplaintsPanic /Arousal PhobicGastrointestinalSleep Symptoms OnsetMid-Nocturnal Symptom InsomniaSexualsConcentration Interes Ins Selft Future / Decision SuicidalOutlook. MakingInvolvemen IdeatioMoodt VariationQualityPsychomotor of MooPsychomotor SlowingDecreased AgitationDecreased AppetiteEarly Weigh MorningPleasure ReactivityInsomnia / EnjoymentLeaden of MoodIncreased ParalysisIncreased /Appetite PhysicalEnergy Weight Energy Hypersomni/ FatiguabilityInterpersonala Sensitivity DUSP2
ATF3 p PDE4B 0.001 IL6 0.005 IL1B 0.01 TNF 0.05 TNFAIP3 0.1 BCL2A1 0.2 PTX inflammation PTGS CCL20 CXCL2 EREG CXCL3 MXD EGR3 effect size F3 3 MAFF
THBS 2
SERP transcription RGC32 PTPN 1 NAB2
MAPK 0 EMP1 STX1 -1 SDR motility / adhesion / CCL2 chemotaxis CCL7 -2 CDC42 FABP -3 CD9 HSPA1 CCR2 general melancholical atypical depressive symptoms depressive symptoms depressive symptoms
Heat map depicting the regression coefficient of the association between actual depressive symptoms (within three days of blood sampling) and gene expression. Depressive symptoms were ordinally measured. Gene expression was expressed as z-transformed –∆∆Ct. Statistical analysis was performed using ordered logistic regression. Contrast was set to 3. Blank compartments either represent a small regression coefficient and statistical probability or an analysis with insufficient observations. None of the analyses were significant below the 0.2 false discovery rate (FDR) threshold for multiple testing.
39 after FDR correction. Although not fitting into clustering groups of associations, some associations e.g. with a decreased appetite, with sympathetic arousal and with sad and irritable mood are notable. No noticeable difference exists in the number of associations between the atypical, melancholic and general depressive symptoms as separate groups.
Association between age at onset, duration of illness and monocyte pro-inflammatory gene expression The associations between age at onset, duration of illness and the expression of genes belonging to each subcluster are described in table 3. Using univariable anal- yses many sub-cluster 1a genes associated significantly with the age at onset in a negative manner. Some sub-cluster 1a genes associated significantly with duration of illness in a positive manner. When entered into multivariable models, PDE4B, IL6, TNFAIP3 and PTX3 gene expression associated significantly with the age at onset in a negative manner, where no concurrent associations were demonstrated with duration of illness (table 3). The genes ATF3 and IL1 associated significantly with duration of illness in a positive manner in these analyses, where concurrent associations with age at onset could not be demonstrated. Associations between any of the sub-cluster 1b genes and age at onset or duration of illness could not be found. Many sub-cluster 2 genes associated significantly with the duration of illness in a positive manner and some sub-cluster 2 genes associated significantly with age at onset in a negative manner (table 3). When entered into multivariable models, MAPK6, EMP1, STX1, DHRS3 and CCL2 gene expression associated significantly with duration of illness in a positive manner, where age at onset was found not to be asso- ciated in any of these models for sub-cluster 2 genes (table 3). Addition of an interaction term or gender to the multivariable analyses did not alter the results markedly. Significant associations could not be demonstrated between individual gene expression and age, except for MAPK6.
40 TABLE 3 Associations between age at onset, duration of illness and individual monocyte pro-inflammatory gene expression
Univariable Multivariable mRNA Age at onset - Duration of illness - Age at onset – Duration of illness genes RC (ci) RC (ci) RC (ci) RC (ci) DUSP2 -0.0716 (-0.128 - -0.0152) ** 0.0478 (-0.0124 - 0.108) -0.0639 (-0.128 - 0.000227) * 0.0172 (-0.0490 - 0.0835) ATF3 -0.0535 (-0.0955 - -0.0115) ** 0.0678 (0.0257 - 0.110) *** -0.0294 (-0.0752 - 0.0163) 0.0537 (0.00641 - 0.101) ** PDE4B -0.0785 (-0.126 - -0.0308) *** 0.0669 (0.0161 - 0.118) ** -0.0618 (-0.115 - -0.00831) ** 0.0373 (-0.0180 - 0.0926) IL6 -0.233 (-0.403 - -0.0626) *** 0.106 (-0.126 - 0.338) -0.232 (-0.417 - -0.0478) ** 0.00281 (-0.230 - 0.236) IL1 -0.0798 (-0.163 - 0.00370) * 0.122 (0.0340 - 0.211) *** -0.038 (-0.128 - 0.0522) 0.104 (0.00546 - 0.203) ** 2 TNF -0.0653 (-0.126 - -0.00512) ** 0.0383 (-0.0316 - 0.108) -0.0623 (-0.130 - 0.00518) * 0.00781 (-0.0681 - 0.0838) TNFAIP3 -0.0948 (-0.161 - -0.0283) *** 0.0616 (-0.0283 - 0.151) -0.0886 (-0.160 - -0.0169) ** 0.022 (-0.0686 - 0.113) BCL2A1 -0.0514 (-0.107 - 0.00432) * 0.0685 (0.0121 - 0.125) ** -0.0263 (-0.0881 - 0.0354) 0.0559 (-0.00795 - 0.120) *
PTX3 -0.1000 (-0.177 - -0.0231) ** 0.0794 (-0.0227 - 0.181) -0.0889 (-0.172 - -0.00628) ** 0.0398 (-0.0647 - 0.144) disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship PTGS -0.0714 (-0.145 - 0.00273) * 0.0355 (-0.0618 - 0.133) -0.0702 (-0.150 - 0.0100) * 0.00426 (-0.0972 - 0.106) CCL20 -0.204 (-0.371 - -0.0368) ** 0.203 (-0.0139 - 0.420) * -0.168 (-0.346 - 0.00970) * 0.128 (-0.0965 - 0.353) CXCL2 -0.112 (-0.204 - -0.0206) ** 0.131 (0.0377 - 0.224) *** -0.068 (-0.169 - 0.0330) 0.0983 (-0.00613 - 0.203) * EREG -0.0931 (-0.185 - -0.00100) ** 0.0892 (-0.0118 - 0.190) * -0.0681 (-0.183 - 0.0471) 0.0458 (-0.0787 - 0.170) CXCL3 -0.086 (-0.229 - 0.0568) 0.149 (0.000770 - 0.297) ** -0.00711 (-0.180 - 0.166) 0.144 (-0.0426 - 0.331) MXD -0.0187 (-0.0571 - 0.0196) 0.0309 (-0.00966 - 0.0714) -0.00288 (-0.0502 - 0.0444) 0.029 (-0.0221 - 0.0802) EGR3 0.0152 (-0.0614 - 0.0917) 0.0177 (-0.0650 - 0.100) 0.0382 (-0.0573 - 0.134) 0.0422 (-0.0611 - 0.145) F3 -0.0885 (-0.203 - 0.0265) 0.115 (-0.00756 - 0.237) * -0.0398 (-0.182 - 0.102) 0.0891 (-0.0642 - 0.242) MAFF -0.0498 (-0.123 - 0.0234) 0.0759 (-0.000765 - 0.153) * -0.0128 (-0.102 - 0.0765) 0.0677 (-0.0289 - 0.164) THBS -0.0604 (-0.143 - 0.0221) 0.0544 (-0.0357 - 0.145) -0.0471 (-0.151 - 0.0566) 0.0243 (-0.0878 - 0.136) SERPINB2 -0.0536 (-0.173 - 0.0664) 0.0956 (-0.0309 - 0.222) -0.00196 (-0.150 - 0.146) 0.0944 (-0.0653 - 0.254) RGC32 -0.0448 (-0.140 - 0.0501) 0.107 (0.0106 - 0.203) ** 0.0209 (-0.0913 - 0.133) 0.120 (-0.00101 - 0.241) * PTPN -0.0529 (-0.104 - -0.00184) ** 0.0691 (0.0205 - 0.118) *** -0.0175 (-0.0792 - 0.0443) 0.0590 (-0.00181 - 0.120) * NAB2 -0.0246 (-0.0875 - 0.0383) 0.0284 (-0.0365 - 0.0934) -0.0151 (-0.0865 - 0.0564) 0.0212 (-0.0526 - 0.0950) MAPK6 -0.00521 (-0.0705 - 0.0601) 0.0619 (-0.00358 - 0.127) * 0.0287 (-0.0430 - 0.100) 0.0756 (0.00148 - 0.150) ** EMP1 -0.0655 (-0.124 - -0.00698) ** 0.0894 (0.0309 - 0.148) *** -0.0324 (-0.0963 - 0.0314) 0.0739 (0.00787 - 0.140) ** STX1 -0.101 (-0.213 - 0.0107) * 0.157 (0.0525 - 0.262) *** -0.0166 (-0.146 - 0.113) 0.148 (0.0164 - 0.279) ** DHRS3 -0.062 (-0.142 - 0.0183) 0.117 (0.0437 - 0.190) *** 0.00491 (-0.0842 - 0.0940) 0.120 (0.0289 - 0.211) ** CCL2 -0.0475 (-0.123 - 0.0281) 0.0969 (0.0205 - 0.173) ** -0.0059 (-0.0887 - 0.0769) 0.0940 (0.00692 - 0.181) ** CCL7 -0.130 (-0.260 - 0.000696) * 0.182 (0.0485 - 0.316) *** -0.0632 (-0.206 - 0.0800) 0.151 (-0.000107 - 0.303) * CDC42 -0.0586 (-0.100 - -0.0170) *** 0.0637 (0.0210 - 0.106) *** -0.0383 (-0.0842 - 0.00763) 0.0454 (-0.00208 - 0.0928) * FABP 0.0262 (-0.0240 - 0.0763) -0.00999 (-0.0616 - 0.0417) 0.0309 (-0.0318 - 0.0936) 0.00828 (-0.0553 - 0.0719) CD9 -0.0204 (-0.0723 - 0.0314) 0.0524 (0.00279 - 0.102) ** 0.0146 (-0.0464 - 0.0755) 0.0610 (-0.000852 - 0.123) * HSPA1 0.0530 (0.000640 - 0.105) ** -0.0357 (-0.0909 - 0.0195) 0.0492 (-0.0164 - 0.115) -0.00663 (-0.0732 - 0.0599) CCR2 0.00109 (-0.0527 - 0.0549) -0.00043 (-0.0463 - 0.0455) 0.00132 (-0.0704 - 0.0730) 0.000299 (-0.0609 - 0.0615)
Z-score transformation of the relative gene expression was applied to facilitate mutual comparison. Statistical analysis was performed using linear regression. Multivariable regression analyses consisted of age at onset and duration of illness as dependent variables. Correction for gender or the addition of an interaction term in the multivariable analyses did not alter results significantly. Results are reported as regression coefficients (RC) with 95% confidence intervals (ci). Significance is reported at the 10% (p=0.1)*, 5% (p=0.05)** and 1% (p=0.01)*** level.
41 Association between monocyte pro-inflammatory gene expression and current medication use The associations between the expression of genes belonging to each subcluster and current medication use were drawn up in the medication feature-expression heat map (figure 7). This feature-expression heat map showed a clustering group of nega- tive associations between current use of an antidepressant and sub-cluster 2 genes. Furthermore, a clustering group of weaker positive associations was found between current use of anti-epileptics, especially carbamazepine, and sub-cluster 2 genes. Also noticeable were relative weak, statistical non-significant associations between current use of lithium and some sub-cluster 1a genes, and between current use of antipsychotics and sub-cluster 1a genes.
FIGURE 7 Association between current medication use and monocyte pro-inflammatory gene expression
3 S F 9
DUSP2 ATF3 PDE4b Il6 Il1 TN TNFaip3 BCL2a1 PTX PTG CCL20 CXCL2 EREG CXCL3 MXD EGR3 F3 MAF THBS SERP RGC32 PTPN NAB2 MAPK EMP1 STX1 DHRS CCL2 CCL7 CDC42 FABP5 CD HSPa1a CCR2 Lithium
Valproate
Carbamazepine
Lamotrigine
Antidepressants
Antipsychotics
Benzodiazepines
p Effect size 0. 1
0. 2 -3 -2 -1 0 1 23 0.0 1 0.0 5 0.001 0.005 FDR<0. 2
Heat map depicting the regression coefficient of the association between medication and gene expression. Medication was qualitatively measured. gene expression was expressed as z-transformed –∆∆Ct. Statistical analysis was performed using linear regression. Contrast was set to 1.5. Blank compartments represent a small regression coefficient and statistical probability. The dotted circle represents significance below the 0.2 false discovery rate (FDR) threshold for multiple testing.
42 Discussion
To our knowledge this is the first study on the associations between an extensive set of clinical features and monocyte gene expression in BD and it can be regarded as a next step in the converging approach between immunology and psychopathology in unraveling the complex pathophysiological mechanisms of BD. Our a priori for mulated hypothesis that lifetime psychotic features would be associated with the 2 pro-inflammatory monocyte gene expression could not be confirmed. However, our method visualized the following interesting findings in BD patients: (1) a possible disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship relation between pro-inflammatory gene expression and manic symptomatology, (2) a differential immune activation related to an earlier age at onset, (3) an increased immune system dysregulation during the course of the disorder and (4) support for the concept of an immune suppressive action of some of the mood regulating medi- cations. The association between immune activation and manic / psychomotor symptoms is evident in both the analyses with the lifetime psychotic symptoms and in the analyses with the manic symptoms, in other words irrespective of the questionnaire used. This association between immune activation and manic symptoms is supported by a previous finding by Dickerson et al., who described the YMRS and individual manic symptoms (speech, appearance, irritability, language–thought disorder, thought content and increased motor activity/energy) to be related to an increased serum C-reactive protein(CRP)34. Findings by Brietzke et al. who described pro-inflamma- tory cytokines to be increased in manic BD patients35 and Cunha et al. who Cnha et al. ards this, there arepatients. found CRP to be increased in manic BD patients, but not in euthymic and depressed ones, are also compatible with this association. Where CRP levels are known to be induced by cytokines36, it must be noted that a dis crepancy between cytokine gene expression and protein synthesis is known to exist4. In general, the association between the pro-inflammatory gene expression and depressive symptoms is less outspoken than it is with manic symptoms. However, some associations seem to exist with individual symptoms, e.g. decreased appetite, sympathetic arousal and sad and irritable mood. These associations seem to have some analogy with what is known as sickness behavior during the course of an infec- tion37–39. We could not find indications for a specific relation between atypical or melancholic symptoms and monocyte activation. To our knowledge this is the first study investi- gating an association between the subtypes of depression and biological parameters in BD. In unipolar major depressive disorder a differential role of HPA-axis function and inflammation, including CRP and cytokine levels, has been reported in melan- cholic versus atypical depression40.
43 The association between age at onset and monocyte activation supports the notion that patients with an earlier age at onset have a biologically more severe form of the illness. Several authors have demonstrated increased morbidity in BD patients with an earlier age at onset41–43, which is in agreement with this observation. On the biological side, genetic differences between BD patients with earlier and later age at onset have been described44,45. Manenschijn, Spijker et al. demonstrated an increase in cortisol in patients with a later age at onset, suggesting BD beginning on an earlier age to be less associated with HPA-axis disturbances46. Additionally, the present study supports an association with the duration of illness and the theory that the pathogenic mechanisms cause further impairment in immune system regulatory mechanisms with the progression of the illness. This association is consistent with the findings from Drexhage et al.5, partly based on the same dataset, and from Soreca et al., who described an increased medical burden, e.g. more cardio- vascular, endocrine and metabolic disease, in BD patients with al longer duration of illness47. The observation of a differential immune activation related to an earlier age of onset and an increased immune dysregulation during the course of the disorder can also be interpreted in view of the staging hypothesis of BD. A staging model founded on neu- robiological correlates of distinct stages of BD could potentially predict clinical care needs and assist in refining treatment options48. Kauer-Sant’Anna et al. described an increased TNF and decreased IL6 cytokine levels in late-stage versus earlier-stage BD patients, all levels being elevated when compared to healthy controls49. In the pres- ent study TNF and IL6 gene expression was not found to be associated with longer duration of illness, but increased gene expression of these molecules was related to an earlier age at onset. In an attempt to explain the known discrepancy between cytokine gene expression and protein synthesis4 it could be argued that earlier-stage patients in the Kauer-Sant’Anna study potentially had a biologically more severe form of the illness. That would be consistent with the earlier age at onset of this group and may explain the differentiated cytokine expression between the early-stage and late-stage groups. Based on the present study additional research is warranted on the ATF3, IL1, MAPK6, STX1, DHRS3, CCL2 gene expression as potential neurobiological staging markers of the progression of the illness, whereas PDE4b, IL6, TNFAIP3, PTX3 gene expression could perhaps predict a more immune mediated profile in the earlier stages of the disorder. In addition to previous findings by Padmos and Drexhage4,5, who described relations between current use of lithium and antipsychotics and pro-inflammatory gene ex- pression, we found associations between current use of antidepressants (negative), and of anti-epileptics (positive) and pro-inflammatory gene expression. Our study results support the theory that the effects of lithium and antipsychotics mainly concern down regulation of some genes and this could also be the case for antide-
44 pressants. Based on the relation between monocyte activation and manic symptoms, it can be argued that the addition of anti-inflammatory medication to standard anti-mania treatment could be a beneficial addition treatment strategy for manic episodes in addition to depressive episodes50. Although treatment studies have thus far not been performed in this regard, low dose (30-80mg/day) acetylsalicylic acid was found to produce a statistically significant duration-independent reduction in the relative risk of clinical deterioration in subjects on lithium, in a large pharma- 2 co-epidemiological study51 and further research into this treatment possibility is warranted. disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship The present study has several limitations. Firstly, the present study focuses on the pro-inflammatory gene expression of monocytes, which is a select part of the com- plex immune system, and generalized statements should be considered in that regard. Secondly, all patients were naturalistically treated and none of them was ‘medication naive’. The positive association between monocyte activation and current use of anti-epileptic leaves room for argumentation that medication is a causal factor for the increased monocyte pro-inflammatory gene expression in BD. Thirdly, an effect of age or gender cannot completely be ruled out. A fourth limitation is contained within the original selection design of the analyzed genes, where only highly over- and under-expressed genes, which were clearly involved in inflammatory processes, where selected, possibly ruling out important genes. The cross-sectional design of the study being suboptimal for analyses with regard to disease progression is a specific fifth limitation concerning the age at onset and duration of illness analyses. The final and in our opinion most important limitation concerns the multiple testing in the analyses leading to an increased risk of type I errors despite the application of FDR correction, combined with relatively small sample size. Nevertheless, we consider it as important for hypothesis forming and testing in further studies. Because of the relatively small sample size of BD-II patients no separate sub-group analyzes were performed comparing patients with BD-I or BD-II disorder. The nature of the associations between psychiatric symptoms and pro-inflammatory monocyte activation is as yet unknown. With the current state of knowledge about these interactions we are unable to clarify why and how activation of sub-clusters of genes, or even specific gene activation, is related to a specific (group) of symptoms. However, we regard them as intermediary phenomenon in the neuroinflammation theory of BD, positioned between the activated inflammatory response system and the phenotypical expression (symptoms), as depicted in figure 1. A theoretical model of action originates in activation of the mononuclear phagocyte system at the level of the brain (microglia), the circulation (monocytes), and the tis- sues (macrophages) as a key element in the pathogenesis of major psychiatric disor- ders. Whether there exists a direct migration of activated monocytes to the brain in psychiatric disease needs further exploration. Also, the role of inflammatory cytokine
45 exchange among the various compartments (circulation, brain, and peripheral tissues, such as adipose tissue and lymphoid tissue) needs clarification. Several routes have been proposed for cytokines to enter and act on the brain, e.g. by altering the blood- brain barrier and by affecting neuronal afferents such as the vagus nerve52. Indeed, the vagus nerve is known to be essential for balancing anti- and pro-in- flammation during sepsis53. Supported by multiple observations of mania in vagus nerve stimulation, a treatment for refractory epilepsy thought to have its effect on the limbic system and frontal cortex54, it is tempting to question the importance of the role of the vagus nerve in the association between monocyte activation and the psychiatric symptoms in BD. In our opinion, especially the association between manic symptoms and immune activation deserves verification in further studies. The psychoimmunological model in BD does not stand on its own, but concerns other psychiatric disorders as well. This makes it interesting to study our findings also in other disorders, like SZ where motor function symptoms also play an important role.
Acknowledgements
We thank Harm de Wit and Annemarie Wijkhuijs for their excellent technical assistance and Sanne Kemner for her assistance in retrieving archived data.
46 References
1. Schulze TG. Genetic research into bipolar disorder: the need for a research framework that integrates sophisticated molecular biology and clinically informed phenotype characterization. Psychiatr Clin North Am. 2010 Mar;33(1):67–82. 2. Maes M, Smith R, Scharpe S. The monocyte-T-lymphocyte hypothesis of major depression. Psychoneuroendocrinology. 1995 Jan;20(2):111–6. 3. Calabrese JR, Skwerer RG, Barna B, Gulledge AD, Valenzuela R, Butkus A, et al. Depression, 2 immunocompetence, and prostaglandins of the E series. Psychiatry Res. 1986 Jan;17(1):41–7. 4. Padmos RC, Hillegers MHJ, Knijff EM, Vonk R, Bouvy A, Staal FJT, et al. A discriminating messenger Relationship between clinical features and inflammation related monocyte gene expression in bipolar disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship RNA signature for bipolar disorder formed by an aberrant expression of inflammatory genes in monocytes. Arch Gen Psychiatry. 2008 Apr;65(4):395–407. 5. Drexhage RC, van der Heul-Nieuwenhuijsen L, Padmos RC, van Beveren N, Cohen D, Versnel M a, et al. Inflammatory gene expression in monocytes of patients with schizophrenia: overlap and difference with bipolar disorder. A study in naturalistically treated patients. Int J Neuropsychopharmacol. 2010 Nov;13(10):1369–81. 6. Pedrosa E, Locker J, Lachman HM. Survey of schizophrenia and bipolar disorder candidate genes using chromatin immunoprecipitation and tiled microarrays (ChIP-chip). J Neurogenet. 2009 Jan;23(3):341–52. 7. Kähler AK, Otnaess MK, Wirgenes K V, Hansen T, Jönsson EG, Agartz I, et al. Association study of PDE4B gene variants in Scandinavian schizophrenia and bipolar disorder multicenter case-control samples. Am J Med Genet B Neuropsychiatr Genet. 2010 Jan 5;153B(1):86–96. 8. Clerici M, Arosio B, Mundo E, Cattaneo E, Pozzoli S, Dell’osso B, et al. Cytokine polymorphisms in the pathophysiology of mood disorders. CNS Spectr. 2009 Aug;14(8):419–25. 9. Papiol S, Molina V, Desco M, Rosa A, Reig S, Sanz J, et al. Gray matter deficits in bipolar disorder are associated with genetic variability at interleukin-1 beta gene (2q13). Genes Brain Behav. 2008 Oct;7(7):796–801. 10. Pae C-U, Lee K-U, Han H, Serretti a, Jun T-Y. Tumor necrosis factor alpha gene-G308A polymorphism associated with bipolar I disorder in the Korean population. Psychiatry Res. 2004 Jan 30;125(1):65–8. 11. Czerski PM, Rybakowski F, Kapelski P, Rybakowski JK, Dmitrzak-Weglarz M, Leszczyńska- Rodziewicz A, et al. Association of tumor necrosis factor -308G/A promoter polymorphism with schizophrenia and bipolar affective disorder in a Polish population. Neuropsychobiology. 2008 Jan;57(1-2):88–94. 12. Mansour HA, Talkowski ME, Wood J, Chowdari K V, McClain L, Prasad K, et al. Association study of 21 circadian genes with bipolar I disorder, schizoaffective disorder, and schizophrenia. Bipolar Disord. 2009 Dec;11(7):701–10. 13. Gallitano AL, Tillman R, Dinu V, Geller B. Family-based association study of early growth response gene 3 with child bipolar I disorder. J Affect Disord. 2012 May;138(3):387–96. 14. Pae C-U, Kim J-J, Yu H-S, Lee C-U, Lee S-J, Jun T-Y, et al. Monocyte chemoattractant protein-1 promoter -2518 polymorphism may have an influence on clinical heterogeneity of bipolar I disorder in the Korean population. Neuropsychobiology. 2004 Jan;49(3):111–4. 15. Roh M-S, Lee KY, Joo E-J, Lee N, Kim YS. No association of the MCP-1 promoter A-2518G polymorphism with bipolar disorder in the Korean population. Neurosci Lett. 2007 Oct 29;427(1):1–5. 16. Altamura a C, Mundo E, Cattaneo E, Pozzoli S, Dell’osso B, Gennarelli M, et al. The MCP-1 gene (SCYA2) and mood disorders: preliminary results of a case-control association study. Neuroimmunomodulation. 2010 Jan;17(2):126–31. 17. Iwayama Y, Hattori E, Maekawa M, Yamada K, Toyota T, Ohnishi T, et al. Association analyses between brain-expressed fatty-acid binding protein (FABP) genes and schizophrenia and bipolar
47 disorder. Am J Med Genet B Neuropsychiatr Genet. 2010 Mar 5;153B(2):484–93. 18. Drzyzga L, Obuchowicz E, Marcinowska A, Herman ZS. Cytokines in schizophrenia and the effects of antipsychotic drugs. Brain Behav Immun. 2006 Nov;20(6):532–45. 19. Pollmächer T, Haack M, Schuld A. Effects of antipsychotic drugs on cytokine networks. J Psychiatr …. 2000;34(6):369–82. 20. Rybakowski JK. Antiviral and immunomodulatory effect of lithium. Pharmacopsychiatry. 2000 Sep;33(5):159–64. 21. Nahman S, Belmaker RH, Azab AN. Effects of lithium on lipopolysaccharide-induced inflammation in rat primary glia cells. Innate Immun. 2012 Jun;18(3):447–58. 22. Vonk R, van der Schot AC, Kahn RS, Nolen W a, Drexhage H a. Is autoimmune thyroiditis part of the genetic vulnerability (or an endophenotype) for bipolar disorder? Biol Psychiatry. 2007 Jul 15;62(2):135–40. 23. Padmos RC, Van Baal GCM, Vonk R, Wijkhuijs AJM, Kahn RS, Nolen W a, et al. Genetic and environmental influences on pro-inflammatory monocytes in bipolar disorder: a twin study. Arch Gen Psychiatry. 2009 Sep;66(9):957–65. 24. Reichart CG, Wals M, Hillegers MHJ, Ormel J, Nolen W a., Verhulst FC. Psychopathology in the adolescent offspring of bipolar parents. J Affect Disord. 2004 Jan;78(1):67–71. 25. Hillegers MH, Reichart CG, Wals M, Verhulst FC, Ormel J, Nolen W a. Five-year prospective outcome of psychopathology in the adolescent offspring of bipolar parents. Bipolar Disord. 2005 Aug;7(4):344–50. 26. APA. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR). Arlington, VA: American Psychiatric Association; 2000. 27. Young RC, Biggs JT, Ziegler VE, Meyer D a. A rating scale for mania: reliability, validity and sensitivity. Br J Psychiatry. 1978 Nov 1;133(5):429–35. 28. Rush AJ, Weissenburger J, Giles E, Schlesser MA, Fulton CL. The Inventory Preliminary for Depressive Findings Symptomatology ( IDS ): Depression. 1986;65–87. 29. Knijff EM, Breunis MN, van Geest MC, Kupka RW, Ruwhof C, de Wit HJ, et al. A relative resistance of T cells to dexamethasone in bipolar disorder. Bipolar Disord. 2006 Dec;8(6):740–50. 30. Staal FJT, Weerkamp F, Baert MRM, van den Burg CMM, van Noort M, de Haas EFE, et al. Wnt target genes identified by DNA microarrays in immature CD34+ thymocytes regulate proliferation and cell adhesion. J Immunol. 2004 Jan 15;172(2):1099–108. 31. User Bulletin # 2 ABI P RISM 7700 Sequence Detection System SUBJECT: Relative Quantitation of Gene Expression. System. 2001;1997. 32. Hochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990 Jul;9(7):811–8. 33. Wei T. corrplot: Visualization of a correlation matrix [Internet]. R package version 0.73. 2013. Available from: http://cran.r-project.org/package=corrplot 34. Dickerson F, Stallings C, Origoni A, Boronow J, Yolken R. Elevated serum levels of C-reactive protein are associated with mania symptoms in outpatients with bipolar disorder. Prog Neuropsychopharmacol Biol Psychiatry. 2007 May 9;31(4):952–5. 35. Brietzke E, Stertz L, Fernandes BS, Kauer-Sant’anna M, Mascarenhas M, Escosteguy Vargas A, et al. Comparison of cytokine levels in depressed, manic and euthymic patients with bipolar disorder. J Affect Disord. Elsevier B.V.; 2009 Aug;116(3):214–7. 36. Ramadori G, Van Damme J, Rieder H, Meyer zum Büschenfelde KH. Interleukin 6, the third mediator of acute-phase reaction, modulates hepatic protein synthesis in human and mouse. Comparison with interleukin 1 beta and tumor necrosis factor-alpha. Eur J Immunol. 1988 Aug;18(8):1259–64. 37. Hart BL. Biological basis of the behavior of sick animals. Neurosci Biobehav Rev. 1988 Jan;12(2):123–37. 38. Charlton BG. The malaise theory of depression: major depressive disorder is sickness behavior and antidepressants are analgesic. Med Hypotheses. 2000 Jan;54(1):126–30.
48 39. Dantzer R, O’Connor JC, Freund GG, Johnson RW, Kelley KW. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci. 2008 Jan;9(1):46–56. 40. Lamers F, Vogelzangs N, Merikangas KR, de Jonge P, Beekman a TF, Penninx BWJH. Evidence for a differential role of HPA-axis function, inflammation and metabolic syndrome in melancholic versus atypical depression. Mol Psychiatry. Nature Publishing Group; 2012 Oct 23;(February):1–8. 41. Bellivier F, Golmard JL, Henry C, Leboyer M, Schürhoff F. Admixture analysis of age at onset in bipolar I affective disorder. Arch Gen Psychiatry. 2001 May 1;58(5):510–2. 42. Leboyer M, Henry C, Paillere-Martinot M-L, Bellivier F. Age at onset in bipolar affective disorders: a 2 review. Bipolar Disord. 2005 Apr;7(2):111–8. 43. Coryell W, Fiedorowicz J, Leon AC, Endicott J, Keller MB. Age of onset and the prospectively Relationship between clinical features and inflammation related monocyte gene expression in bipolar disorder bipolar in expression gene monocyte related inflammation and features clinical between Relationship observed course of illness in bipolar disorder. J Affect Disord. Elsevier; 2012 Oct 10;1–5. 44. Lin P-I, McInnis MG, Potash JB, Willour VL, Mackinnon DF, Miao K, et al. Assessment of the effect of age at onset on linkage to bipolar disorder: evidence on chromosomes 18p and 21q. Am J Hum Genet. 2005 Oct;77(4):545–55. 45. Etain B, Mathieu F, Rietschel M, Maier W, Albus M, McKeon P, et al. Genome-wide scan for genes involved in bipolar affective disorder in 70 European families ascertained through a bipolar type I early-onset proband: supportive evidence for linkage at 3p14. Mol Psychiatry. 2006 Jul;11(7):685–94. 46. Manenschijn L, Spijker AT, Koper JW, Jetten AM, Giltay EJ, Haffmans J, et al. Long-term cortisol in bipolar disorder: Associations with age of onset and psychiatric co-morbidity. Psychoneuroendocrinology. Elsevier Ltd; 2012 Dec;37(12):1960–8. 47. Soreca I, Fagiolini A, Frank E, Houck PR, Thompson WK, Kupfer DJ. Relationship of general medical burden, duration of illness and age in patients with bipolar I disorder. J Psychiatr Res. 2008 Sep;42(11):956–61. 48. Berk M, Hallam KT, McGorry PD. The potential utility of a staging model as a course specifier: a bipolar disorder perspective. J Affect Disord. 2007 Jun;100(1-3):279–81. 49. Kauer-Sant’Anna M, Kapczinski F, Andreazza AC, Bond DJ, Lam RW, Young LT, et al. Brain-derived neurotrophic factor and inflammatory markers in patients with early- vs. late-stage bipolar disorder. Int J Neuropsychopharmacol. 2009 May;12(4):447–58. 50. Müller N, Schwarz MJ, Dehning S, Douhe a, Cerovecki a, Goldstein-Müller B, et al. The cyclooxygenase-2 inhibitor celecoxib has therapeutic effects in major depression: results of a double-blind, randomized, placebo controlled, add-on pilot study to reboxetine. Mol Psychiatry. 2006 Jul;11(7):680–4. 51. Stolk P, Souverein PC, Wilting I, Leufkens HGM, Klein DF, Rapoport SI, et al. Is aspirin useful in patients on lithium? A pharmacoepidemiological study related to bipolar disorder. Prostaglandins Leukot Essent Fatty Acids. Elsevier; 2010 Jan;82(1):9–14. 52. Beumer W, Gibney SM, Drexhage RC, Pont-Lezica L, Doorduin J, Klein HC, et al. The immune theory of psychiatric diseases: a key role for activated microglia and circulating monocytes. J Leukoc Biol. 2012 Aug 8;92(September):1–17. 53. Kessler W, Diedrich S, Menges P, Ebker T, Nielson M, Partecke LI, et al. The role of the vagus nerve: modulation of the inflammatory reaction in murine polymicrobial sepsis. Mediators Inflamm. 2012 Jan;2012(2012):467620. 54. Gerson R, Murray E, Price B, Frankel M, Douglass LM, Cunningham M. Mania following vagus nerve stimulation: a case report and review of the literature. Epilepsy Behav. Elsevier Inc.; 2011 Jan;20(1):138–40. 55. Drexhage RC. Immuno-neuro-endocrine networks: A study on the inflammatory state of circulating monocytes and CD4+ T cells in psychiatric and endocrine autoimmune disease. University Medical Center Rotterdam, The Netherlands; 2011. 56. Suppes T, Leverich GS, Keck PE, Nolen W a, Denicoff KD, Altshuler LL, et al. The Stanley Foundation Bipolar Treatment Outcome Network. II. Demographics and illness characteristics of the first 261 patients. J Affect Disord. 2001 Dec;67(1-3):45–59.
49
CHAPTER 3 Feature-expression heat maps A new visual method to explore complex associations between two variable sets
Bartholomeus C.M. Haarman, Rixt F. Riemersma-Van der Lek, Willem A. Nolen, R. Mendes, Hemmo A. Drexhage, Huibert Burger
Journal of Biomedical Informatics. 2015 Oct 14;53:156–61. Abstract
Introduction Existing methods such as correlation plots and cluster heat maps are insufficient in the visual exploration of multiple associations between genetics and phenotype, which is of importance to achieve a better understanding of the pathophysiology of psychiatric and other illnesses. The implementation of a combined presentation of effect size and statistical significance in a graphical method, added to the ordering of the variables based on the effect-ordered data display principle was deemed useful by the authors to facilitate in the process of recognizing meaningful patterns in these associations.
Materials and Methods The requirements, analyses and graphical presentation of the feature-expression heat map are described. The graphs display associations of two sets of ordered variables where a one-way direction is assumed. The associations are depicted as circles representing a combination of effect size (color) and statistical significance (radius).
Results An example dataset is presented and relation to other methods, limitations, areas of application and possible future enhancements are discussed.
Conclusion The feature-expression heat map is a useful graphical instrument to explore associ- ations in complex biological systems where one-way direction is assumed, such as genotype-phenotype pathophysiological models.
52 Introduction
Tukey emphasized that exploratory data analysis relies more on graphical display, whereas confirmatory data analysis is easier to computerize1,2. Existing graphi- cal methods to explore associations in a set of multiple variables are cluster heat maps and correlation plots. Heat maps originated from two-dimensional displays of a two-by-two data matrix. Larger values were represented by darker squares and smaller values by lighter squares3. E.g., in gene expression studies, these values correspond to the amount of a particular RNA or protein expressed. The further development of the cluster heat map, which includes ordering of the columns and 3 rows to reveal structure, has been a multi-step process. Facilitating the process of detecting meaningful patterns in the visual presentation, Sneath4 displayed the Feature-expression heat maps results of a cluster analysis by permuting the rows and the columns of a matrix to place similar values adjacent to each other according to the clustering, which is based on the effect-ordered data display principle5. This principle says that in any data table or graph, unordered variables should be ordered according to what we aim to show. The ideas of similarity and grouping are derived from Gestalt psychology, but have shown to be equally useful in biology6. Ling ultimately formed the idea for joining cluster trees to the rows and columns of the heat map7. Technical advances in printing let the presentation of the graphs develop from overstruck printer characters to the use of computer programs to produce cluster heat maps with high-resolution color graphics8, as can be seen in figure 1. Correlation plots are used to visualize association matrices. These plots can be regarded as heat map style displays of multiple correlation statistics. These statistics may be drawn in several forms: as numbers, circles, ellipses, squares, bars or “pac-man” symbols. In each symbol both the sign and magnitude of the correlation coefficient is represented. This is done so by using two colors printed with varying intensity. The color indicates the sign of the coefficient and the intensity of the color increases proportionally with the magnitude of the correlation coefficient5. We entertained the idea that these visual methods could be of help in the exploration of associations between genetic data and phenotypical presentation in the investi- gation of the pathophysiology of psychiatric disorders. The pathophysiology of these disorders is still largely unknown. In an effort to unravel the genetic basis of mood dis- orders, many genome-wide association studies have been performed. However, these studies found evidence for only a few susceptibility genes, which in turn accounted for a very minor part of disease liability. This fuelled the idea that to grasp the mecha- nism of these complex illness, it is important to have a framework integrating biology and clinical phenotype9. In this model the intermediary processes that occur between the genetic information and the specific phenotypical expression of these illnesses are regarded as a black box10.
53 FIGURE 1
Cluster heat map27,28. The columns of the heat map represent genes and the rows represent samples. Each cell is colorized based on the level of expression of that gene in that sample.
To achieve a better understanding of these intermediary underlying pathophysio logical processes, we wanted to investigate patterns in the associations between specific symptoms and specific gene expression11. We hypothesized that patterns in these associations would come to light most effectively at the intersection of related genes and related symptoms, embroidering on the above mentioned principle of effect-ordered data display. Because we were exploring the physiology of these intermediary black box processes, we preferred to use an effect size measure instead of the correlation coefficient. Contrary to the correlation coefficient effect size measures describe the magni- tude of an association in measurement units. This is generally of more interest in the biomedical sciences than just the degree of linearity of an association, which is measured by the correlation coefficient. This is of special importance in explorative biology based research, which can be compared to a field biologist visiting a new
54 habitat who will begin describing the most striking features, i.e. analogous to the largest effects sizes. In addition to a measure of the magnitude of the associations of interest we wanted to implement inferential statistics to aid in drawing conclusions incorporating their certainty. Statistical significance for the given sample size was used in this regard. Summarizing, the scope of the method we had in mind was to visualize a large set of associations of variables in two sets in which one-way association was assumed, i.e. from gene expression to phenotype. This approach required a contiguously and ordered arrangement of the variables, incorporating the direction of the associations, an effect size measure and the statistical significance of individual associations. 3 Elaborating on this reasoning we developed feature-expression heat maps. In this article we will first describe the method of creating a feature-expression heat Feature-expression heat maps map. Secondly we will present an example. Finally, we will comment on this method and theorize on other areas of application.
Materials and Methods
Preprocessing The dataset for a feature-expression heat map analysis needs to consist of two sets of variables that differ in their nature. The variables in these sets are assumed to represent phenomena that occur in a certain time order according to an under- lying theoretical model and have a one-way relationship. The variables of each set are then transformed in a way that facilitates comparability within the set, such as z-transformation. Because of reasons of in-between comparability, similarity in data type (binary, ordinal, continuous) of the variables within each set is a stringent re- quirement. In order to be able to recognize meaningful patterns in the final feature-expression heat map the variables need to be arranged contiguously and ordered a priori in a way that similar variables are placed adjacent to each other in both variable sets, consistent with the effect ordered data display principle. This may be achieved by performing a cluster analysis on a correlation matrix of the variables in each variable set. Alternatively, the variables can be arranged on phenomenological similarity.
Analysis Regression methods may be used to determine the effect size and statistical signifi- cance of the associations of the individual variables of the preceding with those of the subsequent variable sets. In case of linear regression the regression coefficient β and its p-value are used as measures of effect size and statistical significance, respective- ly. In case of (ordered) logistic regression the β is used which indicates the change in
55 the logit (or log-odds of the outcome) and may be preferred over the odds ratio, the latter being asymmetric and ranging from zero to infinity. In order to control for the increased risk of wrongful rejection of the null hypothesis (type I error, false positive results) correction for the false discovery rate (FDR) can be applied, as described by Benjamini and Hochberg12. Deriving from this method, approximations of the power and sample size can be calculated13,14. The separate effect size, statistical significance and FDR results of each association are put into individual data matrices. They are ordered in such a way that for each as- sociation the statistical property is placed on the intersection between the preceding variables (columns) and the subsequent variables (rows), where the variables of the preceding and subsequent variable sets are ordered according to the above-de- scribed procedure.
Graphical presentation To display the associations of two sets of variables adapted heat maps are drawn, visualizing the preceding variable set in the columns and the subsequent variable set in the rows, each ordered facilitating the visual identification of meaningful clusters of association later on. An underlying cluster analysis tree may be added to one or both of the axes. In the feature-expression heat maps the associations between preceding and sub- sequent variables are represented by circles (figure 2). The effect size measure is represented by the type and intensity of the color, whereas the statistical significance of the analyses is represented by the radius of the circles. Shades of red are used for positive effect sizes, whereas shades of blue are used for negative effect sizes.
FIGURE 2 Overview of a feature-expression heat map
Preceding Variable set A (ordered by cluster analysis or phenomenologically)
Variable 1 Variable 2 Variable 3 ... Variable n color intensity = effect size Variable 1 circle radius = statistical significance Variable 2 blue = negative effect
Variable 3 red = positive effect
Variable 4 centered dot = association surviving false discovery rate correction
... or phenomenologically) converging associations Subsequent Variable set B Subsequent Variable (ordered by cluster analysis cluster analysis by (ordered
Variable n
56 TABLE 1 R code for corrplot Table 1 R code for corrplot
Import tables > plot_size <- as.matrix(read.table("plot_size.txt", sep = "\t", header = TRUE)) > plot_significance <- as.matrix(read.table("plot_significance.txt", sep = "\t", as Stepmatrices Codeheader = TRUE)) > fdr_template <- as.matrix(read.table("fdr_template.txt", sep = "\t", header = TRUE)) > fdr_significance <- as.matrix(read.table("fdr_significance.txt", sep = "\t", header = TRUE)) Define colors > col <- colorRampPalette(c("blue", "white", "red")) Create > corrplot(plot_significance, method =c("circle"), col=("black"), tl.cex = 0.6, tl.col = ("black"), cl.pos="n") significance plot > corrplot(plot_size, is.corr = FALSE, method =c("color"), addgrid.col = "grey", 3 Create size col = col(200), tl.cex = 0.6, tl.col = "black", cl.pos="n") measure plot > corrplot(fdr_template, method =c("circle"), col=("black"), tl.cex = 0.6, tl.col Create FDR Feature-expression heat maps = ("black"), p.mat = fdr_significance, insig = "blank", sig.level = 0.009, plot cl.pos="n") Create legends > corrplot(legend_significance, method =c("circle"), col=("grey"), tl.cex = 0.6, tl.col = ("black"), cl.pos="n") > corrplot(plot_size, is.corr = FALSE, method =c("color"), addgrid.col = "grey", col = col(200), tl.cex = 0.6, tl.col = "black", cl.pos="r", cl.lim = c(-3,3), cl.ratio=0.4, cl.length=7) R code for corrplot version 0.7315 R code for corrplot version 0.7315
Centered dots may be added to the compartments that comply with a FDR below a certain threshold, thus allowing for a selected portion to be expected false positive. Optionally, when needed in the process of drawing statistical decisions it can be preferred to only visualize the circles of associations of which the statistical signifi- cance is below a pre-defined threshold. To create these heat maps, separate plots are drawn for the effect size, statistical significance and FDR data matrices. These plots can be drawn with the corrplot package, by T. Wei15, on R (R Development Core Team 2013, Vienna, Austria)16. R pro- gramming code examples are given in table 1. In order to magnify the more significant associations, i.e. those in which p is approaching 0, applying a transformation 1-3√p to the statistical significance parameter was found to be useful empirically. Finally, the separate plots can be merged with the transparency function of a vector graphics editor.
57 Results and discussion
Example We will use some results of our study on monocyte gene expression and psychiatric symptoms of patients with bipolar disorder presented in chapter 2 to demonstrate the use of the feature-expression heat map11. In this dataset we analyzed the relation between gene expression and manic symptoms containing information from 64 patients. According to the underlying pathophysiological model manic symptoms are asso- ciated with inflammation related monocyte gene expression, albeit indirectly. The manner of how these processes interact is as yet unknown. Resulting from the patho- physiological model gene expression variables were placed in the preceding variable set, whereas manic symptoms were put in the subsequent variable symptom set. The gene variables were ordered based on a hierarchical cluster analysis of a Pearson correlation matrix, previously executed and published17. They were subdivided into three subclusters and a rest group based on the molecular function. Symptom variables were ordered into a phenomenological sequence ranging from core mood symptoms via thought symptoms, psychosomatic symptoms, motor symptoms, food intake symptoms, sleep symptoms, to higher functional symptoms. The gene expres- sion variables contained continuous, normally distributed data and Z-transformation was applied. The symptom variables contained ordered categorical data, which were all transformed on a 0-1 scale for standardization. In this example analysis associations between individual gene expression and individual manic symptoms were analyzed using ordered logistic regression. The ordered logistic regression model is a direct generalization of the commonly used two-outcome logistic model. In ordered logistic regression, an ordinal dependent variable is estimated as a linear function of independent variables and a set of cut-points18,19. For each association used to create the feature-expression heat map in this example the effect size was defined as the magnitude of the regression coefficient β indi- cating the change in the log-odds of the outcome variable per unit increase in the independent variable. Subsequently, the z-statistic was calculated for each association as the ratio of the coefficient β to its standard error. The statistical significance then was defined as the statistical probability (p) that this statistic, assumed to follow a normal distribu- tion, is as extreme as, or more so, than what would have been observed under the null hypothesis, defined by p>|z|. As stated previously, the transformation 1-3√p was applied to the statistical significance and maximization of the effect size to 3 was applied. Effect size and significance plots were exported as vector graphs, displayed in figure 3a and figure 3b.
58 . Manic . 11 11 Individual effect size plot statistical(a), significance plot and (b) FDR plot (c) constitutethat the feature- expression heat map (d) depicting the associations between manic symptoms monocyte and inflammatory expression gene symptoms were ordinally measured. Gene expression and distributed normally was z-transformed. Statistical analysis was performed using ordered logistic regression. Circles with a center dot represent significance below the 0.2 false discovery rate (FDR) threshold for multiple testing. The legends, black have annotations and lines been added manually after the creation of the heat map. Figure 3d has been reprinted permissionwith 3 2 1 0 -1 -2 -3 FDR<0.2 0.001 0.005 0.01 0.05 0.1 0.2 3 p size effect
Feature-expression heat maps chemotaxis
motility / motility
inflammation transcription / adhesion
Insight
Appearance
Sexual Interest Sexual
Disruptive or Aggressive Behavior Aggressive or Disruptive
Decreased Sleep Decreased
Increased Speech Increased
Increased Motor Activity or Energy or Activity Motor Increased Language / Thought Disorder Thought / Language
manic symptoms Content
Irritability Elevated Mood Elevated d F3 IL6 PTX TNF SDR IL1B CD9 MXD STX1 ATF3 SERP FABP CCL7 PTGS CCL2 PTPN EMP1 EREG EGR3 THBS CCR2 MAFF NAB2 MAPK HSPA1 CCL20 CXCL2 CXCL3 PDE4B DUSP2 RGC32 CDC42 BCL2A1 TNFAIP3 FDR plot c Significance plot b Effect size plot Effect a FIGURE 3
59 We set the FDR at 0.2, thus allowing 1/5 to be false positive. This resulted in a statis- tical significance threshold (q-value) of 0.009. A FDR plot was drawn with corrplot marking the corresponding circles (figure 3c). Finally, the effect size plot, statistical significance plot and FDR plot were merged with Adobe Illustrator (Adobe Systems Inc., San José, California; figure 3d). This mania feature-expression heat map allowed for the identification of a converging group of associations between the genes PTPN – CDC42, so called sub-cluster 2 genes, and manic symptoms. Sixteen of these associations were significant after FDR correction, especially in the associations with the symptoms increased speech and increased motor activity11.
Relation to other methods Marked properties of the feature-expression heat maps are the combined display of an effect size measure and the statistical significance and use of effect-ordered data display on two sets of variables. This combination aids in the recognition of associa- tion patterns in complex systems, e.g. pathophysiological models. These feature-expression heat maps are based on the original cluster heat maps and use some features that can be found in correlation plots. Both the original cluster heat maps and feature-expression heat maps facilitate the visual analysis of extensive data sets for patterns. Where original cluster heat maps allow displaying all kinds of data matrices, the feature-expression heat map limits its applicability to one-way associations between two variable sets. While limiting the area of usability, it facilitates the use of regression methods. These are almost essential to analyze the strength of the phenomena involved and are an asset in explorative research focused on deconstructing pathophysiological models. Of the various ways of displaying correlation plots the circle correlation plot, where the radius as well as the type and intensity of the color were derived from the correla- tion coefficient, drew our attention. In doing so the correlation plot utilizes two ways of visual display to show one test outcome. The visual combination of two measures of association in the feature-expression heat maps, i.e. effect size and statistical significance in one graphical display, increases the usefulness of the method. While visualizing the effect size allows observation of the strongest associations, adding the significance of the association adds the ability to observe the signal-to-noise ratio of the observations, thus aiding in the process of inference of the explorative process. In visually integrating and presenting this distinctive information this method allows for a balanced interpretation of the associations. Especially, the method allows for salient complex (patterns of) associations to become apparent. Bipartite network graphs can be regarded as an alternative to the feature-expression heat map method in visualizing two dimension (bipartite) variables20,21. For example, a bipartite network could represent genes and symptoms as nodes, and the edge
60 weight connecting the nodes could represent the significance of the respective association. The color of the edges could represent the effect size, and its style (e.g. solid versus dotted line) could represent whether or not the significance met the FDR correction. The possibility of adding more than two variable dimensions in one graph (multipartite graph) is a benefit of this method. However, in medium size datasets consisting of two variable dimensions the authors consider the surveyability of the feature-expression heat map to be favorable due to its convenient arrangement.
Limitations The scalability of the feature-expression heat map is principally limited by the 3 perception and interpretation capabilities of the interpreter. To facilitate the inter- pretability of more complex feature-expression heat maps separate panels can be Feature-expression heat maps created containing subgroups of the variables. For example, the presentation of the separate symptom dimensions (manic, depressive, psychotic symptoms) has been performed using separate panels in our recent study11. Like all methods exploring data sets with large variable lists, feature-expression heat maps may bring about an increased risk of type I errors. Although an extensive discussion about controlling for this problem is beyond the scope of this article, we have endeavored to restrict the extend of this limitation by deploying the Benjamini-Hochberg method. It is known that the Benjamini-Hochberg method offers a more powerful alternative to the traditional Bonferroni method22,23. Although a wrongful rejection of the null hypotheses cannot be fully eliminated with this method, considering only clusters of adjoining associations to be meaningful can further diminish this risk, which is a strength of the heat map method.
Future perspectives Originating in a study on the relation between monocyte gene expression and psy- chiatric symptoms we expect the feature-expression heat map method to be useful for studying many other illnesses by benefiting from a combined effect size and statistical significance plot. High throughput screening studies24,25 involving com- plex relations between genetics and biological features are obvious candidates. Even more so, the method is not limited to genotype – phenotype relations, but can easily be applied in any explorative analysis exploring multiple variables, within a group of subjects, of which a two-variable-set one-way dependency is assumed. As this method can be regarded as an evolution of the original cluster heat maps, it is tempting to reflect on possible future enhancements. At present the method relies on separate ordering of the row and column variables, based on individual cluster analyses or phenomenological similarity. Biclustering is a cluster method that allows simultaneous clustering of both rows and columns26. Biclustering of the data matrix to obtain the ordering of the variables would increase the extent of the interaction
61 effects in the feature-expression heat map. Furthermore, instead of relying on visual identification of meaningful clusters of associations, future development incorpo- rating more advanced, automated pattern recognition may aid in the discrimination between more meaningful and less meaningful clusters. By automation assigned cluster associations could be visually marked by a distinguishable background color of the compartments involved or with a bold line surrounding these compartments.
Conclusion
The feature-expression heat map is a useful graphical instrument to explore associations in complex biological systems where one-way direction is assumed, such as genotype-phenotype pathophysiological models. It utilizes the combined display of an effect size measure and the statistical significance as well as the use of effect-ordered data display of two sets of variables, both aiding in the recognition of meaningful association patterns.
62 References
1. Tukey JW. Exploratory Data Analysis. Addison-Wesley; 1977. 2. Tukey JW. We Need Both Exploratory and Confirmatory. Am Stat. 1980 Feb;34(1):23–5. 3. Loua T. Atlas statistique de la population de Paris. J. Dejey & cie; 1873. 4. Sneath PH. The application of computers to taxonomy. J Gen Microbiol. 1957 Aug;17(1):201–26. 5. Friendly M. Corrgrams: Exploratory displays for correlation matrices. Am Stat. 2002;56(4):316–24. 6. Friendly M, Kwan E. Effect ordering for data displays. Comput Stat Data Anal. 2003;43:509–39. 7. Ling RL. A computer generated aid for cluster analysis. Commun ACM. ACM; 1973 Jun 1;16(6):355–61. 8. Wilkinson L, Friendly M. The history of the cluster heat map. Am Stat. American Statistical Association; 2009;63(2):179–84. 9. Schulze TG. Genetic research into bipolar disorder: the need for a research framework that 3 integrates sophisticated molecular biology and clinically informed phenotype characterization. Psychiatr Clin North Am. 2010 Mar;33(1):67–82. Feature-expression heat maps 10. Cauer W. Theorie der linearen Wechselstromschaltungen, Vol.I. Leipzig: Akad. Verlags- Gesellschaft Becker und Erler; 1941. 11. Haarman BCM, Riemersma-Van der Lek RF, Burger H, Netkova M, Drexhage RC, Bootsman F, et al. Relationship between clinical features and inflammation-related monocyte gene expression in bipolar disorder - towards a better understanding of psychoimmunological interactions. Bipolar Disord. 2014 Mar 29;16(2):137–50. 12. Hochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990 Jul;9(7):811–8. 13. Ferreira JA, Zwinderman AH. Approximate power and sample size calculations with the Benjamini-Hochberg method. Int J Biostat. 2006;2(1). 14. Efron B. Size, power and false discovery rates. Ann Stat. 2007 Aug;35(4):1351–77. 15. Wei T. corrplot: Visualization of a correlation matrix [Internet]. R package version 0.73. 2013. Available from: http://cran.r-project.org/package=corrplot 16. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2014. 17. Drexhage RC, van der Heul-Nieuwenhuijsen L, Padmos RC, van Beveren N, Cohen D, Versnel M a, et al. Inflammatory gene expression in monocytes of patients with schizophrenia: overlap and difference with bipolar disorder. A study in naturalistically treated patients. Int J Neuropsychopharmacol. 2010 Nov;13(10):1369–81. 18. Brant R. Assessing proportionality in the proportional odds model for ordinal logistic regression. Biometrics. 1990 Dec;46(4):1171–8. 19. StataCorp. Ordered logistic regression. In: Stata 13 Base Reference Manual. College Station, TX: Stata Press; 2013. 20. Dulmage AL, Mendelsohn NS. Coverings of bipartite graphs. Can J Math. 1958 Jan 1;10:517–34. 21. Asratian AS, Denley TMJ, Häggkvist R. Bipartite Graphs and Their Applications. Cambridge University Press; 1998. 22. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B …. 1995; 23. Noble WS. How does multiple testing correction work? Nat Biotechnol. Nature Publishing Group; 2009 Dec;27(12):1135–7. 24. Giuliano KA, Haskins JR, Taylor DL. Advances in high content screening for drug discovery. Assay Drug Dev Technol. 2003 Aug;1(4):565–77. 25. Abraham VC, Taylor DL, Haskins JR. High content screening applied to large-scale cell biology. Trends Biotechnol. 2004 Jan;22(1):15–22.
63 26. Eren K, Deveci M, Küçüktunç O, Çatalyürek Ü V. A comparative analysis of biclustering algorithms for gene expression data. Brief Bioinform. 2013 May;14(3):279–92. 27. Andrade M. Heatmap [Internet]. Wikipedia. 2006. Available from: http://en.wikipedia.org/wiki/File:Heatmap.png 28. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A. 1998 Dec 8;95(25):14863–8.
64 3 Feature-expression heat maps
65
CHAPTER 4 Inflammatory monocyte gene expression Trait or state marker in bipolar disorder?
Karlijn Becking / Bartholomeus C.M. Haarman, Rixt F. Riemersma-van der Lek, Laura Grosse, Willem A. Nolen, Stephan Claes, Hemmo A. Drexhage, Robert A. Schoevers
International Journal of Bipolar Disorders. 2015 Dec 17;3(1):20. Abstract
Background This study aimed to examine whether inflammatory gene expression was a trait or a state marker in patients with bipolar disorder (BD).
Methods 69 healthy controls (HC), 82 euthymic BD patients and 8 BD patients with a mood episode (7 depressed, 1 manic) were included from the MOODINFLAME study. Six of the 8 patients who had a mood episode were also investigated when they were euthymic (6 of the 82 euthymic patients). Of these participants the expression of 35 inflammatory genes were determined in monocytes using quantitative- polymerase chain reaction of which a total gene expression score was calculated as well as a gene expression score per sub-cluster.
Results There were no significant differences in inflammatory monocyte gene expression between healthy controls, euthymic patients. Patients experiencing a mood episode however had a significantly higher total gene expression score (10.63±2.58) compared to healthy controls (p=.004) and euthymic patients (p=.009), as well as when com- pared to their own scores when they were euthymic (p=.02). This applied in particular for the sub-cluster 1 gene expression score, but not for the sub-cluster 2 gene expression score.
Conclusions Our study indicates that in BD inflammatory monocyte gene expression is especially elevated while in a mood episode compared to being euthymic.
68 Background
Disturbances in the immune system have frequently been reported in bipolar disor- der (BD) 1. Several meta-analyses found peripheral cytokines to be raised in patients compared to healthy controls (HC)2,3. However, results are heterogeneous, with also studies reporting on normal4 or even lower cytokine levels5 in BD compared to HC. This may be due to the fact that peripheral cytokines are strongly influenced by life- style and disease factors6. Focusing on the main cellular producers of these cytokines, such as circulating monocytes and macrophages, may be a better approach to find stable markers for BD. Indeed, studies from our group focusing on gene expression of circulating monocytes, found a discriminating pro-inflammatory gene expression in BD patients compared to HC7,8. It remains unclear whether these immunological disturbances are related to the mood state, or are a trait phenomenon. Most studies compared BD patients to HC, 4 without differentiating between patients in different mood states. The few available studies that examined immune disturbances across mood states found significantly Inflammatory gene monocyte expression higher levels of peripheral inflammatory markers during a mood episode compared to euthymia9–13. Regarding inflammatory gene expression, our original hypothesis prior to the study described below, was that monocyte activity might be a diagnostic biomarker for BD and thus a trait factor. However, in further analysis of our previous study we already found the expression of specific inflammatory genes to be higher in a small subsample of depressed versus euthymic patients and to a lesser extent in manic compared to euthymic patients7. Furthermore, we reported a possible relation between a sub-cluster of genes and manic symptomatology14 in BD. In this report we present the results of the MOODINFLAME study, in which we compared euthymic BD patients with HC. Moreover, we present the results in of a small additional study in which BD patients were compared both in a mood episode and when euthymic. Thus, our study aimed to examine whether inflammatory gene expression in monocytes is a trait or a state marker in BD.
69 Methods
Participants Data were derived from the EU funded MOODINFLAME study15 carried out to investi- gate possible inflammatory biomarkers in order to advance early diagnosis, treatment and prevention of mood disorders. In the MOODINFLAME study adult male and female subjects were included who were free of inflammation related symptoms including fever and current or recent infectious or inflammatory disease, uncontrolled systemic disease, uncontrolled metabolic disease or other significant uncontrolled somatic disorders known to affect mood. They did not use somatic medication, specifically any medication known to affect mood or the immune system, such as corticoste- roids, non-steroid anti-inflammatory drugs and statins. Female candidates who were pregnant or recently gave birth were excluded. The present study has been set up as a cross-sectional case-control study extended with a within-patient longitudinal design. Blood was analyzed of a sample of 159 adult participants recruited from two university psychiatry clinics in Groningen (the Netherlands) and Leuven (Belgium). The sample consisted of 69 HC, 82 euthymic BD patients (BD-Eu) and 8 BD patients with a mood episode (BD-Ep) (seven depressed, one manic). Six of the 8 BD-Ep patients were also investigated when they were euthymic (6 of the 82 BD-Eu patients). Of these patients, 1 euthymic patient was resampled after he had an episode, whereas the other five were first measured when they were euthymic and later when they were in an episode. The study was approved by the ethical committees of the participating universities, and written informed consent was obtained from all participants.
Assessments DSM-IV BD diagnoses were established using the Mini-International Neuropsychiat- ric Interview (MINI)16,17. The severity of depression was measured by the Inventory of 18 Depressive Symptoms (IDS-C30) for BD patients in a face-to-face interview, for HC with a self-report questionnaire (IDS-SR30). To determine the presence or intensity of manic symptomatology in patients the Young Mania Rating Scale (YMRS)19 was used. Mood states were defined as euthymic, manic or depressed, based on the MINI. BD-Eu patients were neither in a depressed nor (hypo-)manic episode at the time of measurement as indicated by an IDS-C30 score <22 and an YMRS score <12, respec- tively. Remaining clinical characteristics were obtained with the Patient Questionnaire from the former Stanley Foundation Bipolar Network, including separate clinician and patient chapters covering a spectrum of clinical features20. In the event of a mismatch of results from the MINI in relation to the Patient Questionnaire, diagnoses were checked with the treating physician. Age of onset was defined as the age when the first mood episode occurred, and information on psychiatric medication was dichoto- mized.
70 Laboratory methods To detect the expression of inflammatory genes of monocytes, similar methods were used as described in the original study by Padmos et al.7. In short, RNA was isolated from purified monocytes and to obtain c-DNA for quantitative-polymerase chain reaction (q-PCR), 1 μg of RNA was reverse-transcribed using the cDNA high-capac- ity cDNA Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA, USA). Then, relative to the housekeeping gene ABL1, the expression of ADM, ATF3, BCL2A1, BTG3, CCL2, CCL20, CCL7, CD9, CDC42, CXCL2, DHRS3, DUSP2, EMP1, EREG, FABP5, HSPA1A/HSPA1B, IL-1α, IL-1β, IL1R1, IL-6, IRAK2, MAFF, MAPK6, MXD1, NAB2, PDE4B, PTGS2, PTPN7, PTX3, RGCC32, SERPINB2, STX1A, THBD, TNF and TNFAIP3 was determined, using the comparative threshold cycle (CT) method21. See Table 1, for the list of genes and corresponding proteins. Data were expressed as ΔCT values (values corrected to ABL1) and to control for site (Groningen and Leuven), fold change transformation was applied. By dividing the ΔCT-scores of patients from Groningen 4 by the mean of healthy controls from Groningen and subsequently the scores of
patients from Leuven by the mean of healthy controls from Leuven, the relative gene Inflammatory gene monocyte expression expression was expressed as a fold change (FC) value21.
Gene score calculation In order to obtain a simple measure for overall monocyte activation, we calculated a gene score from the expression levels as described by Grosse et al.22. For each of the 35 genes we determined a range in HC fold change gene expression, defined by the HC mean ± 1 standard deviation (SD). Then, we used this range as a standard to compare the gene expression across the different groups. A gene was consid- ered up-regulated if the FC-value was higher than the HC mean + 1 SD, and down- regulated if the FC-value was lower than the HC mean – 1 SD. Then, we calculated a total gene expression score, by adding all up-regulated (+1), all down-regulated (-1) and all normally expressed (0) genes for each patient. This method proved to be valid, since the total gene scores showed highly significant correlations with the majority of the genes22. Additionally, we calculated two separate sub-cluster gene scores, based on previous cluster analyses performed by Drexhage et al.8. The first sub-cluster consisted pri- marily of pro-inflammatory genes (see Table 1) and the second sub-cluster consisted of chemotaxis, adhesion, differentiation and motility genes (see Table 1).
Statistical analyses All data were analyzed with SPSS version 20.0 (SPSS, Chicago, IL, USA). Sample characteristics were compared using Pearson’s chi-square and Fisher’s exact tests for dichotomous and categorical variables, and for continuous variables ANOVA and t-tests were used. To compare inflammatory gene expression scores across HC,
71 TABLE 1 List of genes with corresponding proteins
Gene symbol Name of corresponding protein ATF3 Cyclic AMP-dependent transcription factor 3 BCL2A1 B-cell lymphoma-2-related protein A1 CCL20 C-C chemokine ligand 20
CXCL2 C-X-C chemokine ligand 2 DUSP2 Dual specificity protein phosphatase 2 EREG Epiregulin IL-1β Interleukin 1β IL-6 Interleukin 6 PDE4B cAMP-specific 3',5'-cyclic phosphodiesterase 4B Inflammation PTGS2 Prostaglandin G/H synthase (cyclooxygenase) PTX3 Pentraxin-related protein 3 TNF Tumor necrosis factor TNFAIP3 Tumor necrosis factor, alpha-induced protein 3 CCL2 C-C chemokine ligand 2 CCL7 C-C chemokine ligand 7
CDC42 Cell division control protein 42 homolog DHRS3 Short-chain dehydrogenase/reductase 3 EMP1 Epithelial membrane protein 1 MAPK6 Mitogen-activated protein kinase 6 motility adhesion / / adhesion
Chemotaxis/ NAB2 Nerve growth factor-induced protein A binding protein 2 differentiation/ differentiation/ PTPN7 Protein tyrosine phosphatase, non-receptor type 7 STX1A Syntaxin-1A ADM Adrenomedullin BTG3 BTG family, member 3 CD9 Cluster of differentiation 9 antigen FABP5 Fatty acid-binding protein 5 HSPA1/HSPA1B Heat shock 70 kDa protein 1 IL-1α Interleukin 1α IL1R1 Interleukin 1 receptor, type 1 Other IRAK2 Interleukin 1 receptor associated kinase-like 2 MAFF Musculoaponeurotic fibrosarcoma oncogene homolog F MXD1 MAD protein RGCC32 Regulator of cell cylcle SERPINB2 Plasminogen activator inhibitor-2 THBD Thrombomodulin
euthymic patients, mood episode patients, and within-patient analyses, ANOVA was used. Results were reported as mean ± standard error. Because an overall inflamma- tory gene expression score was used, correction for multiple testing was not applied. As a set of sensitivity analyses, we repeated all analyses using ANCOVA controlling for sex, age and body mass index (BMI).
72 Results
Table 2 shows the sample characteristics of HC, BD-Eu patients and BD-Ep patients. Figure 1 shows he total inflammatory gene expression scores in these groups. We found no significant differences between HC (2.58±0.88) and BP-Eu (3.48±0.84), or BD-Ep patients when they were euthymic (1.17±0.94) (all p>.44). However, BD-Ep patients had a significantly higher total gene expression score (10.63±2.58) compared to HC (p=.004) and BD-Eu patients (p=.009) and compared to their own scores when they were euthymic (p=.020). For the means of sub-cluster 1 score, again no signif- icant differences were found between HC (1.13±0.41), BD-Eu patients (1.59±0.39) and BD-Ep patients when they were euthymic (0.50±1.38) (all p>.40). BP-Ep patients (5.13±1.20) again had a significantly higher sub-cluster 1 score compared to healthy controls, BD-Eu patients and compared to their own scores when they were euthymic (p=.002, p=.006 and p=.01, respectively). The mean sub-cluster 2 scores of HC 4 (0.74±0.28), BD-Eu patients (0.59±0.26), BD-Ep patients (2.0±0.81), BD-Ep patients when they were euthymic (0.17±0.94) did not differ significantly between any of the Inflammatory gene monocyte expression groups (all p>.10). When repeating the analyses adjusted for sex, age and BMI, this resulted in essentially the same results.
TABLE 2 Characteristics of patients and healthy controls (N=159)
Healthy Controls Euthymic Mood episode (n=69) (n=82) (n=8) pa Female, n (%) 39 (56.5) 41 (50.0) 5 (62.5) .63 Age, mean (sd) 44.7 (16.1) 43.1 (12.1) 41.8 (12.7) .72 BMI, mean (sd) 23.9 (3.2) 25.7 (4.2) 26.89 (4.2) .005 IDS score, mean (sd) 4.8 (3.4) 8.7 (8.0) 42.1 (14.1) <.001 Clinical characteristics YMRS score, mean (sd) - 1.3 (1.2) 4.7 (6.4) .001 Bipolar I disorder, n (%) - 53 (64.4) 5 (62.5) .90 Bipolar II disorder, n (%) - 29 (35.4) 3 (37.5) Age of onset, mean (sd) - 23.3 (9.6) 21.8 (9.1) .68 Lifetime psychotic features, n (%) 26 (31.7) 2 (25.0) .70 Psychotropic medication - Melatonin - 1 (1.2) 2 (25.0) .02 SSRI - 4 (4.9) 1 (12.5) .38 Antipsychotics - 14 (17.1) 1 (12.5) .74 Lithium - 62 (75.6) 2 (25.0) .003 Benzodiazepines - 11 (13.4) 2 (25.0) .37 Anti-epileptics - 22 (26.8) 2 (25.0) .91
Abbreviations: sd, standard deviation; BMI, body mass index; IDS, inventory of depressive symptoms; YMRS, Young mania rating scale; SSRI, selective serotonin reuptake inhibitor. a Based on χ2-tests and Fisher’s exact tests for dichotomous and categorical variables, ANOVA tests when comparing age and BMI, and t-tests when comparing continuous variables between euthymic and mood episode BD patients.
73 FIGURE 1 Total monocyte gene expression score of healthy controls, euthymic and mood episode BD patients (N=159)
Black lines represent mean and standard error of the mean per group. Lines connecting values from the BD-Eu and BD-Ep group represent the euthymic patients who were measured again when they had a mood episode. Abbreviations: HC, healthy controls; BD, bipolar disorder.
Discussion To our knowledge, the present study is the first to show an elevated inflammatory monocyte gene expression in BD patients when experiencing a mood episode, com- pared to both HC and euthymic BD patients. Furthermore, BD-Ep patients had an increased inflammatory gene expression than when they were euthymic. This indi- cates that inflammatory gene expression in BD is related to the mood state, rather than being a trait marker. Our findings are supported by several other studies examining peripheral cytokines, where the highest levels of cytokines are found in BD patients with a mood episode, although findings in these studies were not equivocal9–13. Serum levels of cytokines are known to follow a different pattern than monocyte gene expression23. Belonging to the same developmental lineage as brain microglia, monocyte activation may be more directly related to psychopathology than circulating cytokines24,25. Previous studies from our group in different samples examining inflammatory mono- cyte gene expression in relation to BD found specifically the sub-cluster 2 genes to be related to a mood episode7,8 or to severity of manic symptoms14. Although in our study the scores were also higher in BD-Ep patients, we did not find a significant difference
74 in sub-cluster 2 gene score compared to HC or BD-Eu patients. This can probably be explained by the fact that we included only one manic patient, whereas in our previ- ous studies more manic patients were included and by the fact that we used a total gene score calculation, whereas the previous studies examined the separate genes. Since sub-cluster 2 genes are associated with adhesion, cell differentiation and cell shape changes and sub-cluster 1 consists of the classic pro-inflammatory genes, it seems that in our sample having a mood episode is specifically associated with activation of the inflammatory response system. Although our finding that an increased inflammatory gene expression is more likely to be a state than a trait phenomenon, the causality and timely sequence of these associations is still difficult to interpret. Based on the present data we cannot be sure whether an increase in inflammatory gene expression preceded the mood episode, or vice versa. In an earlier study we showed that increased immune activation represent- ed by peripheral markers preceded the onset of manic symptoms in MDD patients26. 4 The only way to examine a clear causal role for immune activation in the development of a mood episode is to measure euthymic patients multiple times prior, during and Inflammatory gene monocyte expression after a mood episode. Our study has several limitations. First and most important, although the total sample consisted of 159 persons, we had only few BP-Ep patients: seven patients with a depression and only one manic patient. And since we examined only one manic patient, it is difficult to draw conclusions about inflammatory gene expres- sion in a manic state. However, because we found already significant differences in this small group, we would encourage future studies to include more patients with a mood episode and also to assess patients repeatedly during both an episode and when euthymic. Second, all our patients were treated naturalistically, which result- ed in a variety of medications that are also known to influence inflammatory gene expression, including lithium, anti-epileptics, antipsychotics and several antidepres- sants7,14,27,28. Since these effects are typically suppressive in nature, medication may have obscured a real difference in inflammatory gene expression between BD-Eu patients and HC. However, the BP-Ep patients demonstrating significantly increased gene expression compared to both HC and BD-Eu patients used approximately the same medications, suggesting a pathophysiological cause. Third, our selection of genes was based on the study of Padmos et al.7 which found these specific signature genes, possibly ruling out other important genes. Finally, our study only focused on inflammatory gene expression of monocytes, which make up around 2-8% of the total white blood cell population and is still a peripheral measurement. It would also be of interest to examine other parts of the peripheral immune system (e.g. leuko- cyte subsets), or more proximal factors such as microglial activation in the brain or cytokine concentrations in cerebrospinal fluid.
75 Conclusions
In conclusion, our study showed that in BD patients the presence of a mood episode was associated with elevated inflammatory monocyte gene expression. This may imply that immune activation found in BD may rather be a state marker than a trait marker and can be detected in monocytes. Studies in peripheral cytokines corrob- orate our findings, however our results in gene expression need to be replicated in larger samples before a firm conclusion can be drawn.
Acknowledgements
We thank Harm de Wit and Annemarie Wijkhuijs for their excellent technical assistance and Juliëtte Kalkman for accompanying the patients.
76 References
1. Leboyer M, Soreca I, Scott J, Frye M, Henry C, Tamouza R, et al. Can bipolar disorder be viewed as a multi-system inflammatory disease? J Affect Disord. Elsevier B.V.; 2012 Dec;141(1):1–10. 2. Modabbernia A, Taslimi S, Brietzke E, Ashrafi M. Cytokine alterations in bipolar disorder: a meta-analysis of 30 studies. Biol Psychiatry. Elsevier; 2013 Jul 1;74(1):15–25. 3. Munkholm K, Braüner JV, Kessing LV, Vinberg M. Cytokines in bipolar disorder vs. healthy control subjects: a systematic review and meta-analysis. J Psychiatr Res. Elsevier Ltd; 2013 Sep;47(9):1119–33. 4. Guloksuz S, Cetin EA, Cetin T, Deniz G, Oral ET, Nutt DJ. Cytokine levels in euthymic bipolar patients. J Affect Disord. Elsevier B.V.; 2010 Nov;126(3):458–62. 5. Boufidou F, Nikolaou C, Alevizos B, Liappas I a, Christodoulou GN. Cytokine production in bipolar affective disorder patients under lithium treatment. J Affect Disord. 2004 Oct;82(2):309–13. 6. O’Connor M-F, Bower JE, Cho HJ, Creswell JD, Dimitrov S, Hamby ME, et al. To assess, to control, to exclude: effects of biobehavioral factors on circulating inflammatory markers. Brain Behav Immun. Elsevier Inc.; 2009 Oct;23(7):887–97. 7. Padmos RC, Hillegers MHJ, Knijff EM, Vonk R, Bouvy A, Staal FJT, et al. A discriminating messenger 4 RNA signature for bipolar disorder formed by an aberrant expression of inflammatory genes in monocytes. Arch Gen Psychiatry. 2008 Apr;65(4):395–407. 8. Drexhage RC, van der Heul-Nieuwenhuijsen L, Padmos RC, van Beveren N, Cohen D, Versnel Inflammatory gene monocyte expression M a, et al. Inflammatory gene expression in monocytes of patients with schizophrenia: overlap and difference with bipolar disorder. A study in naturalistically treated patients. Int J Neuropsychopharmacol. 2010 Nov;13(10):1369–81. 9. Brietzke E, Stertz L, Fernandes BS, Kauer-Sant’Anna M, Mascarenhas M, Escosteguy Vargas A, et al. Comparison of cytokine levels in depressed, manic and euthymic patients with bipolar disorder. J Affect Disord. Elsevier B.V.; 2009 Aug;116(3):214–7. 10. Cunha AB, Andreazza AC, Gomes F a, Frey BN, da Silveira LE, Gonçalves C a, et al. Investigation of serum high-sensitive C-reactive protein levels across all mood states in bipolar disorder. Eur Arch Psychiatry Clin Neurosci. 2008 Aug;258(5):300–4. 11. Barbosa IG, Morato IB, de Miranda AS, Bauer ME, Soares JC, Teixeira AL. A preliminary report of increased plasma levels of IL-33 in bipolar disorder: further evidence of pro-inflammatory status. J Affect Disord. Elsevier; 2014 Mar;157:41–4. 12. Tsai S-Y, Chung K-H, Wu J-Y, Kuo C-J, Lee H-C, Huang S-H. Inflammatory markers and their relationships with leptin and insulin from acute mania to full remission in bipolar disorder. J Affect Disord. 2012;136(1-2):110–6. 13. Ortiz-Domínguez A, Hernandez M, Berlanga C, Gutierez-Mora D, Moreno J, Heinze G, et al. Immune variations in bipolar disorder: phasic differences. Bipolar Disord. 2007;(11):596–602. 14. Haarman BCM, Riemersma-Van der Lek RF, Burger H, Netkova M, Drexhage RC, Bootsman F, et al. Relationship between clinical features and inflammation-related monocyte gene expression in bipolar disorder - towards a better understanding of psychoimmunological interactions. Bipolar Disord. 2014 Mar 29;16(2):137–50. 15. MOODINFLAME website. http://moodinflame.eu. 2014. 16. Sheenan D V, Lecrubier Y, Sheehan H, Amorim P, Janavs J, Weiller E, et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): The Development and Validation of a Structured Diagnostic Psychiatric Interview for DSM-IV and ICD-10. J Clin Psychiatry. 1998;59(Suppl 20):22–33. 17. Lecrubier Y, Sheehan D, Weiller E, Amorim P, Bonora I, Harnett Sheehan K, et al. The Mini International Neuropsychiatric Interview (MINI). A short diagnostic structured interview: reliability and validity according to the CIDI. Eur Psychiatry. Éditions scientifiques et médicales Elsevier, Paris; 1997;12(5):224–31.
77 18. Rush AJ, Gullion CM, Basco MR, Jarrett RB, Trivedi MH. The Inventory of Depressive Symptomatology (IDS): psychometric properties. Psychol Med. 1996;26:477–86. 19. Young RC, Biggs JT, Ziegler VE, Meyer D a. A rating scale for mania: reliability, validity and sensitivity. Br J Psychiatry. 1978 Nov 1;133(5):429–35. 20. Leverich GS, Nolen WA, Rush AJ, McElroy SL, Keck PE, Denicoff KD, et al. The Stanley Foundation Bipolar Treatment Outcome Network. I. Longitudinal methodology. J Affect Disord. 2001 Dec;67(1-3):33–44. 21. Biosystems A. User Bulletin #2, Applied Biosystems PRISM 7700 Sequence Detection System: Relative Quantitation of Gene Expression. In: Applied Biosystems. 2001. 22. Grosse L, Carvalho L a, Wijkhuijs AJM, Bellingrath S, Ruland T, Ambrée O, et al. Clinical characteristics of inflammation-associated depression: Monocyte gene expression is age-related in major depressive disorder. Brain Behav Immun. Elsevier Inc.; 2014 Aug 20;2. 23. Mesman E, Hillegers MH, Ambree O, Arolt V, Nolen W a., Drexhage H a. Monocyte activation, brain-derived neurotrophic factor (BDNF), and S100B in bipolar offspring: A follow-up study from adolescence into adulthood. Bipolar Disord. 2014;39–49. 24. Beumer W, Gibney SM, Drexhage RC, Pont-Lezica L, Doorduin J, Klein HC, et al. The immune theory of psychiatric diseases: a key role for activated microglia and circulating monocytes. J Leukoc Biol. 2012 Aug 8;92(September):1–17. 25. Haarman BCM (Benno), Riemersma-Van der Lek RF, de Groot JC, Ruhé HG (Eric), Klein HC, Zandstra TE, et al. Neuroinflammation in bipolar disorder – A [11C]-(R)-PK11195 positron emission tomography study. Brain Behav Immun. Elsevier Inc.; 2014 Aug 3;40:219–25. 26. Becking K, Boschloo L, Vogelzangs N, Haarman BCM, Riemersma-van der Lek R, Penninx BWJH, et al. The association between immune activation and manic symptoms in patients with a depressive disorder. Transl Psychiatry. Nature Publishing Group; 2013 Jan;3(10):e314. 27. Rybakowski JK. Antiviral and immunomodulatory effect of lithium. Pharmacopsychiatry. 2000 Sep;33(5):159–64. 28. Tourjman V. In vivo immunomodulatory effects of antipsychotics on inflammatory mediators: A review. Adv Biosci Biotechnol. 2012;03(04):551–65.
78 4 Inflammatory gene monocyte expression
79
CHAPTER 5 Does CRP predict outcome in bipolar disorder in regular outpatient care?
Sonya M. Balukova, Bartholomeus C.M. Haarman, Rixt.F. Riemersma – van der Lek, Robert A. Schoevers
International Journal of Bipolar Disorders. 2016 Dec;4(1):14. Abstract
Background The association between inflammation and the course of mood disorders is receiving increased attention. This study aims to investigate whether a sub-group of patients with BD can be identified for which a higher CRP (C-reactive protein) level at baseline is associated with an unfavorable prognosis.
Methods Historic cohort study using CRP at baseline, with 15 months follow-up of mood status and medication. Cross-sectional analyses include boxplots, one-way ANOVA, Receiver Operating Characteristics (ROC) curve and chi-square test, and the longitu- dinal analysis using multivariate Cox-regression.
Results 84 bipolar disorder patients were included in the analyses. Cross-sectionally, no statistically significant difference was found in CRP distribution across mood states (p=0.372) or rapid cycling state (p=0.656). Also, no CRP cut-off level was distin- guished between euthymic and non-euthymic patients according to the ROC curve (p=0.449, AUC=0.452, 95%CI: 0.327, 0.576), and a literature-derived cut-off value (3 mg/L) again demonstrated no difference (p=0.530). Longitudinally, no associa- tion was found between CRP and prognosis of disease neither in euthymic (-2 Log Likelihood= 120.460; CRP: p=0.866, B= -0.011, OR= 0.989 (95% CI: 0.874 – 1.120) or non-euthymic patients (-2 Log Likelihood= 275.028; CRP: p=0.802, B= 0.010, OR=1.010 (95% CI: 0.937 – 1.088)). Medication use did not affect these associations.
Conclusions We found no statistically significant association between CRP and a more unfavor- able BD prognosis, suggesting that the application of CRP as a practical biomarker to predict outcome in a naturalistic outpatient care setting is not as straightforward as it may seem.
82 Background
Bipolar disorder (BD) is associated with a significant decrease in quality of life and social functioning of patients1. Despite the availability of pharmacological treatment, its efficacy is far from optimal2,3. A promising approach for optimizing the treatment is to tailor it to the specific characteristics of a patient as part of a ‘personalized medicine’ approach which encompasses not only psychological but also biological markers4. There is increasing evidence to suggest that immunological processes may con- tribute to the emergence as well as the prognosis and severity of BD5. Apart from pro-inflammatory cytokines such as IL-2 and IL-66, C-reactive protein (CRP) is an acute phase protein that is produced in response to infection and inflammation. It is also considered to be another candidate biomarker for detecting immune dysreg- ulation in BD. A number of studies suggest an association between CRP and mood disorders, especially during a manic episode7–9. Based on the idea that immunological processes play a role in the pathophysiology of BD, it can be hypothesized that an increased activity of these processes, measured with CRP, would lead to more instability in BD symptomatology and course of disease. 5 Increased inflammatory activity has been shown to be related to therapy resistance Does CRP predict outcome in bipolar disorder in regular outpatient care? outpatient regular in disorder bipolar in outcome predict CRP Does and chronicity in unipolar depression10–12. Recently, Becking et al.13 demonstrated an increased CRP to predict future development of manic symptoms in a sample of MDD (major depressive disorder) patients, also suggesting that this is a subtype with an untoward prognosis. However, to date, no studies have examined this issue in bipolar patients. The current study investigated whether, in a clinical setting, higher CRP levels at baseline may predict a worse BD outcome, defined as a shorter time to relapse (if euthymic) or a longer time to recover. As some medication may influence inflammatory processes, this was taken into account.
83 Methods
Participants and ethical considerations For the present historic cohort study we used medical files from 84 BD patients from the BD outpatient department of the psychiatry department of the University Medical Center Groningen (UMCG), the Netherlands. Patients who provided a written informed consent to participate, were included if the following criteria were fulfilled: DSM-IV-TR diagnosis of BD, age between 18 and 65, recorded CRP value, not pregnant or less than 6 months postpartum, no current seri- ous somatic illnesses (current infections or liver disease, serious un- or undertreated heart, lung or neurological disorders). All patient data were collected as part of regular outpatient care and were anony- mously used for research according to the Data Protection Act (WBP) and Medical Treatment Agreement (WGBO), as formulated in the Code of Conduct for the Use of Data in Health Research, also known as the Research Code of Conduct (see also IRB in Supporting Information.
Assessments of study parameters The study parameters included CRP, measurement date and value, BD type, presence of a rapid cycling course, and mood episodes and medication one month before and 15 months following the CRP measurement. Assessment of the psychiatric condition of patients was determined by the first author based on information from two sources: the electronic medical records of the treating psychiatrists and the LifeChart Methodology (LCM) records– (a systemat- ic collection of data on the course of illness and treatment presented in a graphical form)14. The condition was noted as one of five categories: 0. Euthymia; 1. Depressive episode; 2. Hypomania or mania; 3. Mixed episode; 4. Unstable mood. Longer-lasting clinically significant mood instability that did not fulfill the criteria for any mood episode was assessed as an “unstable” mood, while subthreshold symptoms were assessed as no change in episode15. Serum CRP (high sensitive CRP) is routinely measured at hospital admission in this Psychiatry department (and additional measurements are performed when there is a psychiatric or somatic event for a patient). CRP was assayed using a wide range turbi- dimetric CRP assay (CRPL3 assay) on a Roche Modular platform (Roche, Mannheim, Germany). Starting from the CRP measurement date at baseline (T0), the time to episode change (T1) and the corresponding psychiatric condition were gathered from the medical file. Cases with unclear information and/or diagnosis of current mental status were discussed with the psychiatrists of the BD outpatient department. Remaining decisions were taken in discussion with all authors of this study.
84 Statistical analysis Both cross-sectional and longitudinal analyses were performed. Cross-section- ally, data were visualized using boxplots and tested using histograms, P-P plots, Kolmogorov-Smirnov test, Kruskal-Wallis, a receiver operating characteristics (ROC) curve and Fisher’s exact test. The longitudinal analysis comprised a multivariate Cox-regression based on the time passed until a change of episode has occurred where the main covariate was the scale of CRP values. This analysis was done separately for euthymic patients and those who were in a mood episode at baseline (depressed, manic, mixed, unstable). This was done because the variable signifying elapsed time has a different meaning for patients in an euthymic state than for a non-euthymic one, and should thus be inter- preted differently to describe the progression of disease. Longer time before episode change signifies a better BD prognosis if a subject is euthymic (means longer time in remission), but it means a worse course if a subject is non-euthymic (means longer time in an episode). Medication was added to the Cox-regression to examine whether it affected CRP or changed the effect of CRP on BD progress. Separate analyses were done after excluding those non-psychopharmaceutical drugs that are known to have 5 anti-inflammatory and/or CRP-affecting properties16–18. Does CRP predict outcome in bipolar disorder in regular outpatient care? outpatient regular in disorder bipolar in outcome predict CRP Does As part of sensitivity analyses, all tests were repeated in each of the following sub- groups for each timepoint and then for the whole study: excluding subjects taking anti-inflammatory medication; excluding all medication that affects CRP; excluding outliers, defined as CRP values above 10 mg/L.
85 Results
The sample of this study consisted of 84 patients and Table 1 shows their character- istics. Testing the data for normality showed a non-normal distribution of the CRP data, with a positive skewness and a significant difference from a normal distribution seen by Kolmogorov-Smirnov test (D(84)=0.258, p<.001). Because of this, median values are provided for each mood state in table 1.
TABLE 1 Characteristics of patient population
Euthymic Depressed (Hypo)manic Mixed Unstable All Mean age (SD) 44.9 (12.4) 42.7 (12.4) 43.2 (10.3) 46 (7.1) 40.5 (10.2) 43.6 (11.7) Number of Subjects (%) 37 (44.0) 27 (32.1) 12 (14.3) 2 (2.4) 6 (7.1) 84 (100) Male Gender (%) 11 (37.9) 10 (34.5) 4 (13.8) 2 (6.9) 2 (6.9) 29 (34.5) Female Gender (%) 26 (47.3) 17 (30.9) 8 (14.5) 0 4 (7.3) 55 (65.5) BD type I (%)* 26 (41.9) 21 (33.9) 11 (17.7) 2 (3.2) 2 (3.2) 62 (100) BD type II (%)* 7 (41.2) 5 (29.4) 1 (5.9) 0 4 (23.5) 17 (100) Rapid Cycling (%) 4 (30.8) 3 (23.1) 2 (15.4) 0 4 (30.8) 13 (100) Metabolic Syndrome (%)** 9 (13.8) 6 (9.2) 2 (3.1) 1 (1.5) 2 (3.1) 20 (30.8)*** Median C-reactive protein (mg/L) 1.10 2.10 0.90 3.08 1.39 1.37
Abbreviations: SD =Standard Deviation; BD=bipolar disorder * Bipolar disorder type information missing for 5 patients ** Metabolic syndrome information missing for 19 patients (Criteria: fasting glucose >5.6 mmol/L and two or more of following: BMI >30 kg/m2, Hypertriglyceridemia >1.7 mmol/L, HDL-C <0.9 mmol/L in men and <1.0 mmol/L in women, Hypertension > 140/90 mmHg) *** The number in brackets in this cell represents the percentage of all subjects who have metabolic syndrome, while the rest of the percentages in this row represent the distribution of subjects with metabolic syndrome across the mood states
Cross-sectional analysis The differences between the distribution of CRP values among the mood states were first examined using Kruskal-Wallis. No statistical significance was found (p=0.372) (see table 1). Figure 1 shows the distribution of mood episodes at base- line. Using Kolmogorov-Smirnov no statistical significant difference was found between the patients that were rapid cycling and patients that were not (p=0.656, see figure 2). Repeating the cross-sectional analyses excluding all medications that are known to affect CRP, as well as excluding CRP outliers, generally yielded the same, non-significant results.
86 FIGURE 1 Boxplot of CRP distribution at baseline across mood state
5
The Y-axis depicts CRP concentration in mg/L starting from 0 mg/L and incrementing care? outpatient regular in disorder bipolar in outcome predict CRP Does with 5 mg/L. On the X-axis bars signify the interquartile range of CRP values in each mood group, and the thickened line inside them shows the median value of CRP within this mood group. The stars and circle show the boxplot outliers, while the numbers next to them signify the specific number of the patient with this outlier value. The circle is a patient with CRP value closest to the threshold value but above it, so it is also not included.
After excluding 6 CRP outliers, there were 37 (44%) who were in euthymia and 47 (56%) were non-euthymic: 27 (32.1%) were in a depressive episode, 12 (14.3%) in a manic episode, 2 (2.4%) in a mixed episode and 6 (7.1%) were unstable. Using a ROC curve, the data was tested to identify a cut-off value of CRP which could suggest whether subjects at baseline would be euthymic or in an episode. Based on the curve (Figure S1 in Supporting Information), no such cut-off value was found (AUC=0.452,p=0.449, 95%CI: 0.327, 0.576). Consequently, the cross-sectional analyses were performed using a literature-based CRP cut-off value of 3.0 mg/L and a chi-square test for the not-normal distribution of CRP value. The results showed that, cross-sectionally, higher CRPs were almost equally distributed among euthymic and non-euthymic patients (42.1% euthymic and 57.9% non-euthymic by CRP>3mg/L; 44.6% euthymic and 55.4% non-euthymic by CRP≤3mg/L). Fisher’s exact test confirmed that these results have no statistical significance (p=0.530).
87 FIGURE 2 Boxplot of CRP distribution between patients with and without rapid cyclingline
The Y-axis depicts CRP concentration in mg/L starting from 0 mg/L and incrementing with 5 mg/L. On the X-axis bars signify the interquartile range of CRP values in each patient group, and the thickened line inside them shows the median value of CRP within this mood group. The stars and circle show the boxplot outliers, while the numbers next to them signify the specific number of the patient with this outlier value.
TABLE 2 Longitudinal results from Cox regression
CRP after adjusting for: Patient Group -2 Log Sig. B Odds 95% CI of Odds Ratio Likelihood coef. Ratio Lower Upper Euthymic Patients at 120.460 .866 -.011 .989 .874 1.120 Baseline Non-Euthymic Patients at 275.028 .802 .010 1.010 .937 1.088 Baseline
Longitudinal analysis For euthymic subjects at baseline From the 37 subjects who were euthymic at Baseline, 20 had a change of episode during the trial period, while 17 patients stayed euthymic. The results are shown in table 2 and the hazard function of the covariate CRP is on figure 3. There is no cor- relation between CRP value and the event of episode change (relapsing). The odds
88 FIGURE 3 Hazard function at mean of the covariate CRP for subjects euthymic at baseline
5 This figure illustrates what the hazard ratio is for recovering of a subject with a given CRP
value compared to a subject with a CRP value of one unit lower in the course of the studied care? outpatient regular in disorder bipolar in outcome predict CRP Does period. The Y-axis represents the rate of recovering of all subjects sick at baseline. The elapsed time period in days until an euthym state has occurred is depicted on the X-axis.
ratio (=0.989) approaches equality for both groups (results are very close to the neutral line). Moreover, these findings are not statistically significant and so the null hypothesis could not be rejected.
For non-euthymic subjects at baseline There were 47 subjects who were in an episode at baseline of which 30 became euthymic, while 17 patients remained in a mood episode during the study period. The results are shown in the table 2 and in figure 4. The odds ratio is approaching equality (results are very close to the neutral line) and these findings are not statistically sig- nificant. As seen in the euthymic group, these results also do not show an association between CRP and relapsing. After adjusting for covariates and testing for interactions of CRP with medication at the two timepoints, there were no results with statistical significance in each of the two mood-state groups – euthymic and non-euthymic. Adjusting for separate medi- cation types used in any moment during the study again yielded no associations with statistical significance.
89 FIGURE 4 Hazard function at mean of the covariate CRP for subjects sick at baseline
This figure illustrates what the hazard ratio is for recovering of a subject with a given CRP value compared to a subject with a CRP value of one unit lower in the course of the studied period. The Y-axis represents the rate of recovering of all subjects sick at baseline. The elapsed time period in days until an euthym state has occurred is depicted on the X-axis.
TABLE 3 Number of subjects per medication type
Parameter T0a T1b Totalc Lithium (%) 31 (36.9) 47 (56) 48 (57.1) Valproic acid (%) 14 (16.7) 9 (10.7) 16 (19.0) Anti-Inflammatory drugs (%)* 12 (14.3) 12 (14.3) 16 (19.0) Tricyclic Antidepressants (TCA) (%)* 5 (6.0) 5 (6.0) 7 (8.3) Antidepressants (non-TCA) (%) 31 (36.9) 27 (32.1) 32 (38.1) Antipsychotics (%) 35 (41.7) 34 (40.5) 39 (46.4) Benzodiazepines (%) 31 (36.9) 31 (36.9) 39 (46.4)
a Medication used up to and including the baseline (T0); b Medication used up to and including the time of episode change is such occurred (T1); c Medication used for the total period of the study (1 month before and up to 15 months after CRP measurement, unless episode change occurred earlier); * Values are the same for both T0 and T1 timepoints;
90 Table 3 shows a summary of the used medication at the T0 and T1 timepoints as well as in total for the whole study period.
Discussion
Principal findings To our knowledge, this is the first study examining longitudinal associations between CRP level and clinical outcome in BD patients in a naturalistic treated and real-life measurement outpatient setting. In a first cross-sectional analysis, we could not distinguish a sub-group of BD patients with an elevated baseline CRP level based on affective state or rapid cycling state. In the longitudinal analysis, no statistically significant association was found between higher CRP values and relapsing in either euthymic or non-euthymic patients, as well as when comparing them.
Comparison to previous studies The results of the cross-sectional analysis can be compared to several previous studies on the association of CRP with BD. Two of the studies with a total of 202 5 patients reported that for their BD subjects higher CRP was significantly associated Does CRP predict outcome in bipolar disorder in regular outpatient care? outpatient regular in disorder bipolar in outcome predict CRP Does with manic state compared to the other mood episodes8,9. Three other studies with a total of 248 patients however did not demonstrate significant differences in CRP across the mood states, which is corresponding to the current study results19–21. The findings of the influence of medication on the reported associations between CRP values and mood states in the above studies are also conflicting. Hornig et al 19 found a significant negative interaction between lithium and CRP within BD patients, while Dickerson et al.8 found that no medication affected the CRP association with BD, as we found in the current study. In a recent meta-analysis Dargél et al. demonstrated an overall cross-sectional association between BD and CRP22. In this meta-analysis, CRP levels were elevated in manic and euthymic patients compared to HC, but not in depressed BD patients compared to HC. Recent prospective longitudinal studies demonstrated CRP al- terations across mood states23, as well as before and after various treatments24. In addition an increased CRP was found to be associated with an increased risk for developing late-onset BD in a large Danish demographic sample25. Adding to this body of knowledge, the present study shows that using CRP as a practical biomarker to predict outcome in a naturalistic outpatient care setting is not as easily applicable as it may seem.
91 Limitations The present study has several limitations. A limitation pertaining the use of histori- cal data is that CRP measurements were more or less routinely measured, and more in those patients who were psychiatrically or somatically ill. Since it was not purely measured as a routine, a selection bias cannot be ruled out. Furthermore, body fat percentage, smoking, blood pressure, physical activity and SNP polymorphisms are suggested to be associated with different baseline levels of CRP26–30 and in the present study we were not able to correct for these variations. Finally, the study would have benefitted from a larger sample size.
Proposed mechanisms There are two factors to consider which could play a role in shaping the results from this study: use of medication and the pathophysiological model of BD. Due to the naturalistic design of this study, a great variety of medications was used, some of which have been suggested to have anti-inflammatory properties. Lithium, part of the standard treatment of dipolar disorder in the Netherlands, was used by more than half of the patients in this study (57.1%). Although its mechanism of action in bipolar depression is not fully understood yet16, several studies have found evidence that lithium causes a decrease in the inflammatory marker CRP11,19 and a similar effect is also described for SSRIs (selective serotonin reuptake inhibitors) used by 38% of the subjects here18. However, there are reports that lithium and antidepressants may also exhibit pro-inflammatory properties16,31,32. These properties were postulated after observing the stimulating effect of lithium on pro-inflammatory cytokines such as Tumor Necrosis Factor-α (TNF-α), Interleukin-4 (IL-4) and Interleukin-6 (IL-6). It is not yet clear what causes these reverse effects. There could be various mechanisms via which medication is influencing CRP. However, at the same time no such effect was demonstrated in this study. Therefore other factors should be considered. Our findings therefore suggest that the inflammatory model of BD is probably much more complex than what can be shown by straightforward inflammatory alterations demonstrated by elevated markers such as CRP33 at one point in time. Compared to the inflammatory mechanism in somatic disorders where CRP, being an acute phase protein, directly correlates to acute worsening of a disease, it may be that immunological processes in BD are affected by more factors and additional mechanisms, which need to be added to the model to gain a better understand- ing. One such hypothesis is that inflammatory changes in psychiatric disorders are mediated through a shift towards the tryptophan/kynurenine pathway, which leads to the formation of further factors that may affect the functioning of the brain, such as neurotoxic accumulation of the metabolite 3-Hydroxykynurenine34. Another hypothesis is the co-occurrence of autoimmunity alongside immune dysregulation in BD. It is thus likely that BD is related to a state of immune dysregulation, rather than
92 more pure immune activation35,36. It has been suggested that the immune mediating pathogenesis of BD occurs at a much younger age, and what follows is a dysregulation of the bodily systems. The long-lasting results are then what we see at an older age: unhealthy and unexpected interactions and responses of the whole body, and of inflammatory markers as well37. Another interesting possibility is that what we see is not necessarily an inflamed but a damaged organism, given that CRP reacts not only to inflammatory stimuli, but also to damaged cells38. To increase our understanding of the position of immune-system-mediated patho- physiological processes in BD it is necessary to measure CRP and other relevant immune bio-assays prospectively in a larger sample size. In that regard it is important to investigate what happens on an intra-individual level: additional trials are necessary measuring CRP on different timepoints in a prospective longitudinal manner so that individual changes could be followed and analyzed. Perhaps, these studies will eluci- date models, adjusting for variables known to influence CRP, that eventually enable CRP to be used as a practical biomarker to predict outcome in naturalistic treatment settings. 5
Acknowledgements care? outpatient regular in disorder bipolar in outcome predict CRP Does We thank dr. J.E. Kootstra-Ros for her valuable assistance in retrieving archived data.
93 References
1. ten Have M, Vollebergh W, Bijl R, Nolen W a. Bipolar disorder in the general population in The Netherlands (prevalence, consequences and care utilisation): results from The Netherlands Mental Health Survey and Incidence Study (NEMESIS). J Affect Disord [Internet]. 2002 Apr;68(2–3):203–13. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12063148 2. Frecska E, Balla P, Falussy L, Ferencz A, Varga Z. The message of the survival curves: I. Composite analysis of long-term treatment studies in bipolar disorder [Internet]. Vol. 14, Neuropsychopharmacologia Hungarica . 2012. p. 155–64. Available from: http://www.mppt.hu/images/magazin/pdf/xiv-evfolyam-3-szam/frecska.pdf 3. Burcusa SL, Iacono WG. Risk for recurrence in depression. Vol. 27, Clinical Psychology Review. 2007. p. 959–85. 4. Hamdani N, Doukhan R, Kurtlucan O, Tamouza R, Leboyer M. Immunity, inflammation, and bipolar disorder: diagnostic and therapeutic implications. Curr Psychiatry Rep [Internet]. 2013;15(9):387. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23955004 5. Liu HC, Yang YY, Chou YM, Chen KP, Shen WW, Leu SJ. Immunologic variables in acute mania of bipolar disorder. J Neuroimmunol. 2004;150(1–2):116–22. 6. Brietzke E, Stertz L, Fernandes BS, Kauer-Sant’Anna M, Mascarenhas M, Escosteguy Vargas A, et al. Comparison of cytokine levels in depressed, manic and euthymic patients with bipolar disorder. J Affect Disord [Internet]. 2009 Aug [cited 2010 Aug 28];116(3):214–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19251324 7. Maes M, Smith R, Scharpe S. The monocyte-T-lymphocyte hypothesis of major depression. Psychoneuroendocrinology [Internet]. 1995 Jan [cited 2011 Mar 6];20(2):111–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7899532 8. Dickerson F, Stallings C, Origoni A, Boronow J, Yolken R. Elevated serum levels of C-reactive protein are associated with mania symptoms in outpatients with bipolar disorder. Prog Neuropsychopharmacol Biol Psychiatry [Internet]. 2007 May 9 [cited 2011 Jun 1];31(4):952–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17391822 9. Cunha AB, Andreazza AC, Gomes F a, Frey BN, da Silveira LE, Gonçalves C a, et al. Investigation of serum high-sensitive C-reactive protein levels across all mood states in bipolar disorder. Eur Arch Psychiatry Clin Neurosci [Internet]. 2008 Aug [cited 2011 Jun 1];258(5):300–4. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18297417 10. Raison CL, Rutherford RE, Woolwine BJ, Shuo C, Schettler P, Drake DF, et al. A randomized controlled trial of the tumor necrosis factor antagonist infliximab for treatment- resistant depression: the role of baseline inflammatory biomarkers. JAMA psychiatry [Internet]. 2013;70(1):31–41. Available from: http://archpsyc.jamanetwork.com/article. aspx?articleid=1356541 11. Sluzewska A, Sobieska M, Rybakowski JK. Changes in acute-phase proteins during lithium potentiation of antidepressants in refractory depression. Neuropsychobiology. 1997;35(3):123–7. 12. Miller AH, Maletic V, Raison CL. Inflammation and Its Discontents: The Role of Cytokines in the Pathophysiology of Major Depression. Biol Psychiatry [Internet]. 2009;65(9):732–41. Available from: http://dx.doi.org/10.1016/j.biopsych.2008.11.029 13. Becking K, Boschloo L, Vogelzangs N, Haarman BCM, Riemersma-van der Lek R, Penninx BWJH, et al. The association between immune activation and manic symptoms in patients with a depressive disorder. Transl Psychiatry [Internet]. 2013 Jan [cited 2013 Dec 2];3(10):e314. Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=3818012&tool=pmcentrez&rendertype=abstract
94 14. Denicoff KD, Leverich GS, Nolen W a, Rush AJ, McElroy SL, Keck PE, et al. Validation of the prospective NIMH-Life-Chart Method (NIMH-LCM-p) for longitudinal assessment of bipolar illness. Psychol Med [Internet]. 2000 Nov [cited 2013 Sep 13];30(6):1391–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11097079 15. Perlis RH, Ostacher MJ, Marangell LB, Hongwei Z, Wisniewski SR, Ketter, Terrence a., et al. Predictors of Recurrence in Bipolar Disorder: Primary Program for Bipolar Disorder ( STEP-BD ). Am J Psychiatry. 2006;163(4):217–24. 16. Goldstein BI, Kemp DE, Soczynska JK, McIntyre RS. Inflammation and the phenomenology, pathophysiology, comorbidity, and treatment of bipolar disorder: a systematic review of the literature. J Clin Psychiatry [Internet]. 2009 Aug;70(8):1078–90. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19497250 17. Ximenes JCM, de Oliveira Gonçalves D, Siqueira RMP, Neves KRT, Santos Cerqueira G, Correia AO, et al. Valproic acid: an anticonvulsant drug with potent antinociceptive and anti-inflammatory properties. Naunyn Schmiedebergs Arch Pharmacol [Internet]. 2013;386(7):575–87. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23584602 18. O’Brien SM, Scott L V, Dinan TG. Antidepressant therapy and C-reactive protein levels. Br J Psychiatry [Internet]. 2006;188:449–52. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16648531 19. Hornig M, Goodman D, Kamoun M, Amsterdam J. Positive and negative acute phase proteins in affective subtypes. J Affect Disord. 1998;49(1):9–18. 20. Hope S, Dieset I, Agartz I, Steen NE, Ueland T, Melle I, et al. Affective symptoms are associated with markers of inflammation and immune activation in bipolar disorders but not in schizophrenia. J Psychiatr Res [Internet]. 2011;45(12):1608–16. 5 Available from: http://dx.doi.org/10.1016/j.jpsychires.2011.08.003
21. Tsai S-Y, Chung K-H, Wu J-Y, Kuo C-J, Lee H-C, Huang S-H. Inflammatory markers and their care? outpatient regular in disorder bipolar in outcome predict CRP Does relationships with leptin and insulin from acute mania to full remission in bipolar disorder. J Affect Disord [Internet]. 2012;136(1–2):110–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21962564 22. Dargél AA, Godin O, Kapczinski F, Kupfer DJ, Leboyer M. C-Reactive Protein Alterations in Bipolar Disorder. J Clin Psychiatry [Internet]. 2015 Feb 25;76(2):142–50. Available from: http://www.psychiatrist.com/jcp/article/pages/2015/v76n02/v76n0203.aspx 23. Jacoby AS, Munkholm K, Vinberg M, Pedersen BK, Kessing LV. Cytokines, brain-derived neurotrophic factor and C-reactive protein in bipolar i disorder - Results from a prospective study. J Affect Disord [Internet]. 2016;197:167–74. Available from: http://dx.doi.org/10.1016/j. jad.2016.03.040 24. Uyanik V, Tuglu C, Gorgulu Y, Kunduracilar H, Uyanik MS. Assessment of cytokine levels and hs-CRP in bipolar I disorder before and after treatment. Psychiatry Res [Internet]. 2015;228(3):386–92. Available from: http://dx.doi.org/10.1016/j.psychres.2015.05.078 25. Wium-Andersen MK, Ørsted DD, Nordestgaard BG. Elevated C-reactive protein and late-onset bipolar disorder in 78 809 individuals from the general population. Br J Psychiatry. 2016;208(2):138–45. 26. Marnell L, Mold C, Du Clos TW. C-reactive protein: Ligands, receptors and role in inflammation. Clin Immunol. 2005;117(2):104–11. 27. Palosuo T, Husman T, Koistinen J, Aho K. C-reactive protein in population samples. Acta Med Scand [Internet]. 1986;220(2):175–9. Available from: http://www.ncbi.nlm.nih.gov/pubmed/3776692 28. Visser M, Bouter LM, McQuillan GM, Wener MH, Harris TB. Elevated C-reactive protein levels in overweight and obese adults. JAMA [Internet]. 1999 Dec 8;282(22):2131–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10591334
95 29. Davey Smith G, Lawlor DA, Harbord R, Timpson N, Rumley A, Lowe GDO, et al. Association of C-reactive protein with blood pressure and hypertension: life course confounding and mendelian randomization tests of causality. Arterioscler Thromb Vasc Biol [Internet]. 2005 May;25(5):1051– 6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15731495 30. Ford ES. Does exercise reduce inflammation? Physical activity and C-reactive protein among U.S. adults. Epidemiology [Internet]. 2002 Sep;13(5):561–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12192226 31. Hamer M, Batty GD, Marmot MG, Singh-Manoux A, Kivimäki M. Anti-depressant medication use and C-reactive protein: Results from two population-based studies. Brain Behav Immun [Internet]. 2011;25(1):168–73. Available from: http://dx.doi.org/10.1016/j.bbi.2010.09.013 32. Nassar A, Azab A. Effects of Lithium on Inflammation. ACS Chem Neurosci [Internet]. 2014;5:451–8. Available from: http://pubs.acs.org/doi/abs/10.1021/cn500038f 33. Altamura a. C, Buoli M, Pozzoli S. Role of immunological factors in the pathophysiology and diagnosis of bipolar disorder: Comparison with schizophrenia. Psychiatry Clin Neurosci. 2014;68(1):21–36. 34. Myint A-MMA, Kim Y-KKY. Network beyond IDO in psychiatric disorders: revisiting neurodegeneration hypothesis. … Neuro-Psychopharmacology Biol Psychiatry [Internet]. 2013 Jan 3 [cited 2014 Sep 17];48:304–13. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24184687 35. Rege S, Hodgkinson SJ. Immune dysregulation and autoimmunity in bipolar disorder: Synthesis of the evidence and its clinical application. Aust N Z J Psychiatry [Internet]. 2013;47(12):1136–51. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23908311 36. Haarman BCM, Riemersma-Van der Lek RF, Burger H, Netkova M, Drexhage RC, Bootsman F, et al. Relationship between clinical features and inflammation-related monocyte gene expression in bipolar disorder - towards a better understanding of psychoimmunological interactions. Bipolar Disord [Internet]. 2014 Mar 29 [cited 2014 Apr 23];16(2):137–50. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24286609 37. Beumer W, Gibney SM, Drexhage RC, Pont-Lezica L, Doorduin J, Klein HC, et al. The immune theory of psychiatric diseases: a key role for activated microglia and circulating monocytes. J Leukoc Biol [Internet]. 2012 Aug 8 [cited 2012 Nov 1];92(September):1–17. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22875882 38. Kushner I, Rzewnicki D, Samols D. What does minor elevation of C-reactive protein signify? Am J Med. 2006;119(2):17–28.
96 5 Does CRP predict outcome in bipolar disorder in regular outpatient care? outpatient regular in disorder bipolar in outcome predict CRP Does
97
PART 2 Neuroimmune system
CHAPTER 6 PET and SPECT in bipolar disorders
Bartholomeus C.M. Haarman, Rixt F. Riemersma-Van der Lek, Henricus G. Ruhé, Jan Cees de Groot, Willem A. Nolen and Janine Doorduin
Adapted from Bipolar Disorders. In: Dierckx RAJO, Otte A, de Vries EFJ, Waarde A, den Boer JA, editors. PET and SPECT in Psychiatry. Heidelberg: Springer Berlin Heidelberg; 2014. p. 223–51. Abstract
This chapter describes the findings of the PET/SPECT research efforts in bipolar disorder (BD). First, the cerebral blood flow and cerebral metabolism findings in the prefrontal cortex, limbic system, subcortical structures and other brain regions are discussed. Metabolism and blood flow oriented studies aided to study various aspects of the metabolism based disease model in which prefrontal cortex hypoactivity is accompanied by limbic hyperactivity. Second, the neurotransmitter studies are discussed. The serotonin transporter alterations are described and the variation in study results is explained, followed by an overview of the results of the various dopamine receptor and transporter molecules studies, taking into account also the relation to psychosis. Molecular imaging demon- strated the importance of serotonin transporter alterations in parts of the limbic system in BD and underscored the role of dopaminergic and cholinergic neurotrans- mission. Most molecular imaging studies in BD have unique designs, extending our knowledge of the pathophysiological mechanisms, but also complicating comparisons between studies.
102 Introduction
It is generally accepted that the cause of bipolar disorder (BD) is multifactorial, with multiple genes making someone vulnerable, and with psychological and social factors bringing the genes to expression. Moreover, somatic factors are assumed to play a role. To unravel the complex interplay between genotype and phenotype researchers have tried to find intermediary processes, so called endophenotypes. These are more related to the underlying genotype than the ultimate phenotype. Endophenotypes should be consistently associated with the illness and represent persistent “trait” rather than episodic or “state” features. By definition, they also should be found in high-risk individuals such as non-affected first-degree family members at a higher rate than in the general population1. The last two decades, many molecular neuroimaging studies have been performed in BD. Alterations of function assessed by molecular neuroimaging may be regarded as important endophenotypes. Probably the best approach in neuroimaging of BD is to study patients during their depressive and manic episodes as well as during the euthymic phase with different (functional) neuroimaging techniques. However, these are very complicated patients, both technically as well as practically (e.g. one can never be sure that the same patient will develop both manic and depressive episodes within a certain time frame). Positron emission tomography (PET) and single-photon emission computed 6 tomography (SPECT) are imaging techniques that use radiolabeled, biological active
compounds (PET or SPECT tracers) to gain information on specific functions of the disorders bipolar in SPECT and PET brain, by measuring brain metabolism or blood flow, or functions of individual cells, such as transporter mechanisms or receptors. The tracers involved are administered in such small doses that pharmacological activity or chemical toxicity is practically absent and due to the usual short half-life of the radionuclides total radiation remains within generally accepted safety levels. Where PET uses positron-emitting radionuclides, that give rise to two opposite directed 511kV gamma rays after annihilation of positrons with electrons, the radio- nuclides in SPECT directly emit gamma rays. Because the gamma rays being specifi- cally in the opposite direction, PET is able to achieve higher spatial resolutions (about 4mm) than SPECT (7-12mm). SPECT is more widely accessible due to the lower maintenance costs and generally easier to handle tracers.
Cerebral blood flow and cerebral metabolism Accumulating scientific evidence supports the theory of metabolic alterations in specific parts of the brain in patients with mood disorders: the prefrontal cortex, the limbic system and subcortical regions (figure 1). With molecular imaging techniques the metabolic activity in the brain (cerebral metabolic rate (CMR)) as well as the blood
103 FIGURE 1 Neuroanatomical regions important in mood disorders
anterior cingulate cortex
prefrontal cortex
subgenual cingulate
orbitofrontal cortex ventral striatum
hypothalamus pituitary raphe nuclei amygdala
hippocampus locus coeruleus
nervus vagus
Neuroanamtomic regions important in mood disorders. (Adapted from Patrick J. Lynch, medical illustrator, and C. Carl Jaffe, MD, cardiologist62, under the Creative Commons Attribution 2.5 Generic license (CC BY 2.5)).
flow in specific regions (cerebral blood flow (CBF)) can be measured. It is generally accepted that CMR and CBF are physiologically coupled and both are indeed closely correlated in healthy controls2. This appeared also to be the case in BD. Dunn et al.3 demonstrated that CMR and CBF were coupled globally and in most regions in BD, except the left pregenual anterior cingulate cortex. CMR can be investigated with an 18F-labeled fluorodeoxyglucose (FDG) PET scan. CBF is measured in PET by 15O-labeled water. The most common SPECT tracers to measure CBF are 133Xe, 123I-labeled iodoamphetamine (IMP) and 99mTc-labeled hexamethylpropylene amine oxime (HMPAO). CMR and CBF can be measured in resting state or during various tasks. Across the whole brain level it remains unclear whether there is an overall global CMR and CBF change in BD when compared to healthy controls. When investigated across mood states some studies found reduced global CMR4–6, while in other studies no alterations where found in CMR7,8. In depressed patients CMR was found to be reduced when compared to controls and manic patients in some studies4,5 but increased in another study9 (table 1). One
104 study investigating CBF found an increased perfusion in manic patients compared tot controls10, but others did not find a difference between the different mood states11,12.
Prefrontal cortex The prefrontal cortex (PFC) is the area of the frontal lobes of the cerebral cortex that is located before the motor and premotor areas. It plays an important role in executive functioning such as planning complex behavior, personality expression, decision making and moderating social behavior13. Regions of the brain are defined as Brodmann areas (BA) based on their cytoarchitectonic structure. In general, BD patients in a depressive or manic episode have a decreased prefrontal cortex CMR and CBF, compared to euthymic patients or healthy controls. Blumberg et al. found a reduced CBF in the right orbital PFC (BA 11) and medial frontal gyrus (BA 10) in manic patients when compared to euthymic patients14 (table 1). CMR activation related to a decision making task was also decreased in manic patients in this region15. Euthymic patients demonstrated orbitofrontral CBF decrease16. The healthy siblings of BD patients demonstrated a comparable CBF decrease in the orbitofrontal PFC during induced sadness17. In manic patients, a decrease in dorsolateral PFC (BA 8, 9, 46) CBF has been demon- strated18,19. Manic patients also showed a decrease of CMR during a decision making task in the ventrolateral PFC (BA 47) when compared to controls15. Furthermore, 6 euthymic older BD patients (50-65 years) had a lower CMR in this region than
controls of the same age820. disorders bipolar in SPECT and PET
Limbic system and subcortical structures The limbic system is a combination of in origin different brain structures that are involved in visceral behavioral patterns (related to survival: eating, drinking, sexual activity), emotions, and memory. Some structures, such as the hippocampus, amyg- dala, anterior thalamic nuclei, are phylogenetically rather old structures (hence the other name paleomammalian brain), while the septum, fornix and limbic cortex are more recently developed structures. The limbic cortex consists of the parahippocampal gyrus (BA 34-36), the cingu- late gyrus (BA 23-26; 29-33) and the dentate gyrus, which are parts of the frontal, parietal and temporal cortical lobes on the medial surfaces of both hemispheres, surrounding the corpus callosum. The anterior part of the cingulate gyrus, the anterior cingulate cortex (ACC, BA 24, 25, 32, 33), plays a role in autonomic functions (regulating blood pressure, heart rate), rational cognitive functions (reward anticipa- tion, decision making, empathy), pain perception and emotion21. In BD patients with depressive or manic episodes, an increased CMR and CBF were demonstrated in various parts of the limbic system. In depressed BD patients, Drevets
105
ulate, and left head of caudate during manic manic during caudate of head left and ulate,
activation during word generation and decreased orbitofrontal orbitofrontal decreased and generation word during activation
depressives has been found after correction for grey matter volume. matter grey for correction after been found has depressives
in the middle frontal gyri bilaterally. Levothyroxin decreased relative activity in in activity relative decreased Levothyroxin bilaterally. gyri frontal middle the in
basal temporal, occipital; medial frontal; parietal regions and in the cingulate gyrus. cingulate the in and regions parietal frontal; medial occipital; temporal, basal -
s g n i
d n i f
n i a episodes. M amygdala the manic and in cortex BD left compared prefrontal patients to HC. dorsolateral Decreased left subgenual cingulate the right in activity higher significantly BD treatment, exhibited patients levothyroxin Before and cerebellar striatum, hippocampus), ventral amygdala, right right thalamus, medial temporal (right lobe left cortex, activity and hadvermis; relative lower a to episode mixed a or depression from going increased depression bipolar with CMR patients for brain whole The episode. manic or state euthymic controls. in than lower but MDD same, PFC for BD and the D were anterolateral CMRdorsal the in of results The and function between CBF neuropsychological were identified. correlations corrected Several cortex prefrontal and orbital rostral Decreased right with were mania. associated rest during activity cing anterior dorsal left in were activity an increased findings The principal and parietal right temporal, superior the left lower in perfusion significantly of interest regions Examining individual found. was group patient the in regions occipital bilateral was found. CMR groups two in the between performance in differences significant No statistically and unipolar (both ness affective with subjects in higher significantly be to wasfound metabolism cerebral Global controls. normal to compared depressed) bipolar than lower those of the of the euthymic blood flow values BD wereThe significantly cerebral mean patients regional medial the bilateral in controls both in callosum corpus the genu of the to ventral cortex prefrontal the in activity increased abnormally of An area unipolar familial and depressives bipolar familial was depressed increased in which levels, BD was plasma with stress Amygdala correlated cortisol patients. activity, BD. remitted in activity amygdala the normalize Mood stabilizers the right subgenual cingulate cortex, left thalamus, right amygdala, right hippocampus, right dorsal and ventral and hippocampus, dorsal ventral right amygdala, right right thalamus, left cortex, subgenual cingulate the right vermis. and cerebellar striatum, O
15
2
HMPAO
-
d HMPAO HMPAO PET o - - h O PET O PET t Tc Tc 15 15 e 2 2 FDGPET FDGPET FDGPET FDGPET FDG PET FDG H FDGand FDGPET CPT word generation, word generation, letter repetition, state resting state resting SPECT SPECT M 11 state resting 11 treatment with levothyroxine state resting state resting 99m state resting H H 99m state resting 11 11 99mTc SPECT state resting 11 PET state resting 11 state resting CPT electrical to stimulation the forearm 11 11
- n
i
o -
d i + + - - - + + + + + + + t e +/ a M c
2 Mi, 5 D) D
s I (15 M)
I (6 E, 5 M) E, (6 I I (6 E, 5 M) E, (6 I t
- I (9 E) (9 I - - II (1 E) c
- - e j SZ b u 10 MDD 5 HC 27 HC S BD 15 14 10 MDD 10 HC 9 BD 1 BD M, 5 BD (5 11 MDD HC 5 M) D, BD15 (10 D OCD w/o 10 OCD w/ 14 12 HC 8 3 E, D, 43 BD (12 7M) HM, 6 HC 11 BD 5 HC 11 BD D) 9 BD (9 11 MDD 21 HC D) 8 BD (8 D) BD16 (16 4 MDD 24 HC E) BD16 (16 10 HC 4 M) 8 E, D, BD21 (9 17 MDD 51 HC 9 E) D, BD15 (7 21 MDD 12 HC
9
9 1
6 5 6 2 4 6 2 1
8 5 9 2 0 2 9
3 0 9
2
9
9 0 8 9 )
5 1 4 6 2 2 0 7 1 r
7 1
0 6 9 i 5 9 0 6 2 9 a 5
8 1 2 0 m
8 8 0 e 9 9 w e 0 0 0 u g g y r 9 9 1 2 a 9
r r 0 0 r 2 a , 1 1 s
1 s s
r e e
2 a 2 b s
r r t t u
o e b b s b r y k e e e e a o h n a e h t t v v d o h m m t l c n M n u x x e e o u u u u - r r u a a a e l l o r u t l a S a B B B B B B B B B C D D (
TABLE 1 Overview of PET/SPECT studies on cerebral blood flow and cerebral metabolism in BD patients
106 and- sex
- he
posterior cortical -
anteroposterior gradient. anteroposterior anhedonia symptom cluster correlated with lower absolute metabolism in in metabolism absolute lower with correlated symptom cluster anhedonia - subjects. control healthy increased compared to HC to compared increased l cortex in the patients but an increase in this region in the siblings. the in region this in increase an but patients the in cortex l
6 PET and SPECT in bipolar disorders bipolar in SPECT and PET
s g n pregenual ACC in depressive patients versus healthy control subjects. Post hoc exploratory analysis analysis exploratory hoc Post subjects. control healthy versus patients depressive ACC in pregenual i
d n i related activation was increased in the manic patients compared with the control patients in the left dorsal ACC dorsal left the in patients control the compared with patients manic the in was increased activation related f
- n i bipolar depression a pattern of prefrontal hypometabolism was observed Additionally a cerebello a Additionally observed was hypometabolism prefrontal of pattern a depression bipolar
a but decreased in the right frontal polar region. region. polar frontal right the in decreased but M psychomotor MDD the both and BD, In CMR in normalized higher and with cortex, and temporal caudate/putamen, anteroventral claustrum, insula, right cingulate. anterior inferior in increases with perfusion, brain of redistribution an important with was associated withdrawal Lithium ACC. particularly areas, limbic in decreases and regions posterior not but manic and depressive both in was asymmetric lobes temporal the of part anterior the in CBFThe distribution t in asymmetry lobe temporal showed same patient the on sequentially taken Images state. euthymic in state. euthymic the in disappeared or diminished that mood states pathological were both systems observed in areas and limbic paralimbic cortices, CBF in the decreases prefrontal in Significant group. control healthy the with compared groups depression In subgroups. bipolar normalized hypermetabolism was all seen in Common and cingulate groups three anterior with induced sadness to were all the dorsal/rostral CBF in increases were groups the Distinguishing cortices. temporal inferior and orbitofrontal the in decreases and insula anterior fronta medial the in decreases cortex accumbens amygdala, orbitofrontal putamen,CMR and anteroventral left area, the bilateral was in increased right and anterior CBF in of reductions showed significant groups patient Both CBF. global in equivalent were groups three The and areas reduction of the normal cortical Task CBFDuring manic episode global was age with compared distribution regional and level flow blood cerebral normal showed a groups patient Both controls. normal matched subjects. control healthy and the BD CBF patients the emerged between in differences No significant additionally revealed increased metabolism in left parahippocampal, posterior cingulate, and right anterior insular insular anterior and right cingulate, parahippocampal, metabolism posterior increased left in revealed additionally versus patients depressive in cortices
-
d EMZ HMPAO o - - h O PET O PET IMP SPECT IMP t - Tc Tc 15 15 Xe SPECT Xe SPECT Xe SPECT Xe SPECT Xe I e 2 2 FDGPET FDGPET FDGPET CPT transient sadness induction probability SPECT SPECT 11 M 11 CPTauditory 99m lithium withdrawal 123 state resting 99m state resting H 11 state resting 133 state resting H decision based making task 133 133 state resting 133 resting state resting between groups and before / after medication state resting
- n i
o - -
d i
- - + + + + + + + t e +/ +/ a M c
HM, 2 M) 2 HM,
s II (22 D, 7 E) I (11 D, 4 E) II (13 D)
I (11 M) (11 I t -
- -
I (9 E) (9 I - c
-
e j b u 4 BD (14 E) 4 BD (14 S D)27 BD (27 31 MDD 1 BD13 (7 D) 6 BD (6 11 MDD 9 HC BD 14 29 BD 43 HC 9 BD 9 HS BD 13 18 HC 11 BD 11 HC M)6 BD (6 6 MDD 10 HC BD 12 16 HC 30 M) D, 40 BD (10 22 MDD 61 HC D) BD (7 7 10 MDD 9 HC 11 MDD
5 1 1
7 0 2
0
7 7
) 4 1 2
8 2
6 3 9 r
3
1 1 0 6 2 1 d 4 a 1 7 n 9 l 5 8
2 i 0 2 0 e 1 2 0 9 ö
e 7 9 9 i 0 0 3 y 0 8 t
9 n 0 k , 9 9 2 0 i 6 2 z
1 9 r
s 1 1
0 r
2 1 s
9 i r r 1
w o 1
2 y e s a n n e e 9
n d h l i i 9 h t g d u v 1 t h o n
u t b b s t 8 f ü u u a l o y u o e r u u u t i u 9 a t 1 S D G G I K K M R R R S T ( HS=healthy sibling; D=depressive episode, E=euthymic episode, M=manic episode, HM=hypomanic episode, Mi=mixed episode, episode, HM=hypomanic episode, Mi=mixed episode, M=manic episode, sibling; D=depressive E=euthymic episode, HS=healthy task discrimination ADT=auditory performance test, CPT=continuous
107 et al. found an increased CMR in the subgenual portion of the ACC (BA 25) when compared to controls, after correction for grey matter volume22 (table 1). This finding was repeated in both treated7 as well as in untreated depressed patients23. Dunn re- ported an association between this CMR increase and the presence of psychomotor and anhedonia symptoms. A similar increase in CMR was demonstrated in the pre- genual and ventral area (BA 33, 24) of the ACC24. In manic patients, an increase in CBF in the subgenual portion of the ACC (BA 25), was described compared to controls22. This increase was also found in the left dorsal ACC (BA 32) when compared to euthymic patients25. In the manic patients, CMR during a decision making task was increased in the left dorsal ACC, when compared with controls 15. In untreated manic patients a SPECT study showed that increased cingulate cortex CBF is associated with poor executive functioning 26. Goodwin et al.27 examined 14 euthymic patients on lithium with SPECT before and after acute double-blind withdrawal of lithium. As often seen clinically, rapid with- drawal was associated with an increase of manic symptoms. The increase of manic symptoms correlated with a CBF decrease in the limbic areas, particularly the ACC. Euthymic patients also demonstrated ACC CBF aberrations16. The healthy siblings of BD patients demonstrated an comparable CBF increase in the ACC during induced sadness17. The amygdala, part of the limbic system, is one of the subcortical areas that is known to be involved in BD. Others are the nucleus accumbens, globus pallidus, striatum (including nucleus caudatus), all part of the basal ganglia of the brain that play a role in higher order motor control. Individually they are involved in different functions, the nucleus accumbens in the reward circuitry, nucleus caudatus in learning and memory, particularly regarding feedback processing and the globus pallidus in visceral regula- tion such as fever induction and emotion induced tachycardia28. Initially, studies of depressed BD patients versus controls described a reduced CMR in the amygdala19 as well as the striatum4,29,30. However, thereafter, various PET studies in depressed patients showed increased activity in the striatum, together with increased activity in limbic structures including the amygdala, hippocampus and parahippocampal regions6,7,20,24,32. Additionally, amygdala and ventral striatal CMR correlated positively with depression severity and with cortisol levels6,32. The differ- ence between these initial and later studies is most probably explained by a higher signal quality and more careful patient selection in the later studies33. High CMR or CBF were also observed in the nucleus caudatus in manic patients25 and nucleus accumbens in depressed patients26.
Other cortical regions An asymmetric CBF was found in the anterior temporal cortex in manic and depressed patients but not when the patients were euthymic34 (table 1). In a more
108 recent study it was demonstrated that euthymic older BD patients (50-65 years) have a higher CMR in this region than controls of the same age20. Furthermore, CBF in the temporal cortex of BD patients was positively associated with executive functions but negatively with attention and memory26.
Neurotransmitter studies Departing from the monoamine theory of affective disorders35 PET/SPECT radioligand studies have focused on the serotonergic, dopaminergic and cholinergic systems.
Serotonin Serotonin (5-hydroxytryptamine) is a monoamine neurotransmitter that is formed out of the amino acid tryptophan. It is mainly found in the gastrointestinal tract, where its secreting cells regulate intestinal movement, in platelets, where it is released during aggregation and in the central nervous system. Serotonin has a regu- latory effect with regard to mood, sleep, sexual activity and appetite. The neurons located in the raphe nuclei, a cluster of nuclei in the brain stem, are the main source of serotonin in the brain. The axons from the raphe nuclei neurons project to nearly every part of the central nervous system. After serotonin is released in the synaptic cleft, it can bind to one of the various receptors or it can be removed by the presynaptic neuron for reuse via the serotonin transporter. 6 As the primary site of serotonergic antidepressant activity, the serotonin transporter
(SERT) is the part of the serotonin neurotransmitter system that has received the disorders bipolar in SPECT and PET most attention in molecular imaging. Among the various ligands that are available, the PET ligands trans- 1,2,3,5,6,10- -hexahydro-6-[4-(methylthio) phenyl] pyrro- lo-[2,1-a] isoquinoline (11C(+)-McNeil 5652), 3-11C-amino-4-(2-dimethylaminometh- ylphenylsulfanyl)benzonitrile (11C-DASB) and the SPECT ligand 2-([2-([dimethylami- no]methyl)phenyl]thio)-5-123I-iodophenylamine (123I-ADAM) are used in BD research. An increase of SERT density was found in the thalamus using 11C(+)-McNeil 5652 in a combined group of euthymic or mildly depressed patients36 and a reduction in the midbrain, hippocampus, thalamus, putamen and ACC in a group of untreated de- pressed patients37 (table 2). With the use of 123I-ADAM SPECT a lower SERT density was found in de midbrain of euthymic BD-I patients when compared to euthymic BD-II patients and healthy controls38. Using the more stable en selective 11C-DASB ligand, an increased SERT density was found in the thalamus, dorsal cingulate cortex, medial prefrontal cortex and insula of depressed untreated BD patients, which was comparable to MDD39,40. Although the results are inconsistent, it can be concluded that serotonin transporter alterations occur in BD, especially in parts of the limbic system. Taking the regulatory function and the observed metabolic changes into account, the SERT density alter-
109
ter
- I patients when patients I psychotic BDpsychotic
-
HT2A receptor .
- cingulate cortex cortex cingulate BD patients whenBD patients
HTT both to BP relative of mood state or striatal striatal or moodof state
- comparison subjects.
MDD and BD groups showed significantly significantly showed MDD BD groups and Is associated with genetic variation within with variation genetic associated Is - DOPA uptake rate constants in the striatum striatum the in constants rate DOPA uptake - binding was found between non - found was binding F DOPA rate constants were significantly reduced in in reduced significantly were constants DOPA rate - - 18 2 depressives had reduced 5 reduced had depressives F - II patients and healthy controls. patients II psychotic BDpsychotic patients. nuclei when compared to controls. whento compared nuclei - 18
to MDDto patients
Conversely, ventral brainstem binding was nearly nearly was binding brainstem ventral Conversely, mal controls, whereas those for striatum were striatum not for whereas those mal controls, - r betweenr the groups.
-
and SZ compared and SZ with controls for and BD psychotic
induced decrease in striatal binding. induced striatal decrease in - had significantly lower DAT availability relative to controls in in controls to relative DAT availability lower significantly had
different. HTT BP in the thalamus (24%, 14%, respectively), insula (15%) and and (15%) insula respectively), 14%, (24%, thalamus the HTT BP in s s - g g n n i i
d d binding potential was not significantly different in manic patients than in in than patients manic in different significantly not was potential binding n n
i i 2 f f
n n i i a a M was the ACC in binding Receptor be decreased found to of compared MDD to and patients controls. BD in binding receptor Decreased CHRM2. the comparison subjects in the striatum. Treatmentthe the in comparison had striatum. no with subjects valproate patients. manic in potential D2 binding on the effect significant M significantly increased 5 increased significantly a to led challenge Amphetamine baseline. at binding D2 receptor However, subjects. healthy in than BD in patients response behavioral greater between difference wasthere the two the in nogroups significant ampheta mine Treatment with valproate had no significant effect on brain 5 on brain effect no significant had valproate with Treatment patients. manic in binding with patients in increased thalamus the was in significantly potential Binding in potential whereas binding subjects, compared as mood control to disorders not the diffe midbrain did had 16%BD patients amygdala, 26% to midbrain, the lower SERTin density ACC. and putamen thalamus, hippocampus, BD euthymic of midbrain de in found was density SERT lower A compared euthymic to BD BD,In the mean SERT BP was thalamus, in dorsal increased both the group healthy to Relative HC and MDD the pontine raphe of nuclei groups the vicinity in D in difference No statistical terms in differ not did subjects and healthy BD patients in differences No significant treatment with valproate, significantly were patients the for cortex frontal the for potentials The binding nor for those than lower The D posterior right the in compared controls to binding had greater BD patients binding lower modestly had BD patients region. caudate left the in putamen and the right lower binding in examined, regions and brain a significantly all in caudate compared region higher significantly had BD patients euthymic the controls, the to Compared DAT. striatal of availability BPD subjects caudate dorsal bilateral control VMAT2 of in than Binding BD thalamus in the was patients in higher SZ patients. and subjects group. control the in than higher were and BDSZ patients and between identical (DCC), medial prefrontal cortex and insula and decreased in the brainstem at the the at brainstem the in decreased and insula and cortex prefrontal medial (DCC), raphe pontine the of level striatum (12%). The bipolar (12%). striatum the patients andthe than were patients the the in lower patients in were found between the manic patients and the comparison subjects. Af were comparison and the subjects. between found manic the patients patients and controls. Post hoc tests showed higher binding for psychotic psychotic for binding showed higher tests hoc Post controls. and patients withpatients BD patients compared to non - to compared patients
1 1
- -
SPECT
-
TZTP PET TZTP PET
McNeil 5652 McNeil McNeil 5652 McNeil d d TRODAT TRODAT N o o - - - h h ADAM IZBM SPECT IZBM 3 PET raclopide DASB PET DASB PET SCH23390 CFT PET DTBZ PET - FP - FP DOPA PET setoperone PET setoperone t t ------Tc Tc - - - - I I e e F F F F C(+) - C(+) - C C C C C C C M 18 18 PET after baseline, treatmentvalproate M SPECT 18 11 PET 11 123 11 11 11 PETmethylspiperone 123 after baseline, amphetamine induction 18 after baseline, treatment valproate 11 11 99m SPECT 99m 11 11 valproate treatmentvalproate
t t
2 e e
g g
r r
HT
2 2 2 2 1 2 a - a M M T D D D D T SERT SERT SERT SERT SERT DOPA uptake DAT DAT DAT VMAT 5
- - n n i i
o o
d d i i ------+ + + + t t e e a a M M c c
s s
II I (13 M) I (13 M) I (15 E) (15 I I t I (6 D; 5 E) D; (6 I
t
-
- - - - II (5 D) - c c
- e e j j b b u u S D) BD16 (16 17 MDD 23 HC D) BD16 (16 24 MDD 25 HC S M)7 BD (7 5 E) D, 6 BD (1 21 HC D) BD18 (18 41 HC BD 10 BD 14 28 HC D) BD18 (18 37 HC D) BD18 (18 18 MDD 34 HC 14 BD (3 D, 11 M) SZ 10 12 HC E) BD13 (13 13 HC BD 13 14 HC M) 1 6 E, D, BD10 (3 21 HC BD 13 14 HC 5 BD 10 MD 46 HC E) BD17 (17 17 HC 11 BD 13 HC BD 15 SZ 12 15 HC 7 MDD
55
41 51 39 37 60
51
49 36
40
58 61 ) 52 )
50 56 r 57 r 38 a a e e y y
, , r r
o o ra 1992ra y y h h d d ham 2005b t t t u u u u t t a a S 2006a Cannon 2011 Cannon ( S Ya 2002 Ichimiya Oquendo 2007 Chou 2010 2006b Cannon 2007 Cannon 1995 Pearlson 2000 Anand 2002b Yatham Suha 2002a Yatham 2007 Amsterdam 2010 Chang 2011 Anand 2001 Zubieta (
r r e e t t t t i i m m s s
n e n n i a a n
i r r n e t t o m n o o i t r r l a o u o u p r e e h o e N C N S D
TABLE 2 Overview of PET/SPECT studies on neurotransmitter systems in BD patients episode, HM=hypomanic episode, Mi=mixed episode, M=manic episode, sibling; D=depressive E=euthymic episode, HS=healthy task discrimination ADT=auditory performance test, CPT=continuous
110 ations may be interpreted as an exponent of a dysfunctional fronto-limbic network. It furthermore suggests that there might be (yet to be identified) modulators of gene expression or that other effects, such as serotonin transporter internalization, occur during different mood states. At the level of the post-synaptic receptors a study investigating the treatment effect 18 of valproate on the 5-HT2-receptor binding, using F-setoperone, demonstrated no difference before or after treatment in manic patients41.
Dopamine Dopamine is a catecholamine neurotransmitter that is formed out of L-DOPA, which in turn is made out of the amino acid tyrosine, while dopamine itself is the precursor of norepinephrine and epinephrine. A dopaminergic imbalance plays an important role in Parkinson’s disease and psychotic symptomatology (psychotic symptoms during mood episodes and SZ)42. Additionally it is thought to be of importance in mania because of the antimanic effect of dopamine receptor blockers (antipsy- chotics) and the mania producing effect of dopamine inducing substances, such as amphetamines43.
Five subtypes of dopamine receptors are known. The D1-like family consists of D1 and D5 receptors, which lead to the inhibition of intracellular adenylate cyclase upon activation, causing cAMP to rise. The D2-like family consists of D2, D3 and D4 recep- tors, which lead to the stimulation of intracellular adenylate cyclase upon activation, 6 causing cAMP, decrease. Overall, the D1-receptor and D2-receptor are the most abun-
dant dopamine receptor subtypes in the brain, with particularly high expression in disorders bipolar in SPECT and PET the striatum and nucleus accumbens and lower levels in the olfactory tubercle. The
D2-receptor is the prominent receptor in the substantia nigra, a region where the 44 D1-receptor is absent . After release into the synaptic cleft and having its neurotransmitting effect via the receptors, dopamine is pumped back into the cytosol of the presynaptic neuron by the dopamine transporter (DAT) from where it can be broken down by enzymes or be reused in synaptic vessels via the vesicular monoamine transporter 2 (VMAT2)45. Parts of the dopaminergic neurotransmission than can be examined with molecular imaging are the various dopamine receptors, dopamine release and the dopamine transporter. These in turn can be investigated during resting state or after an amphet- amine challenge (stimulating dopamine release).
The D2-receptor is an obvious research target because of the known effectiveness of D2-receptor blocking antipsychotic medication on manic and psychotic symp- toms46. Radioligands targeting this receptor are benzamides, such as raclopride and iodobenzamide, and butyrophenons, such as methylspiperone. The binding potential of the benzamides is known to fluctuate in with changing endogenous dopamine concentrations, e.g. after amphetamine challenge. It is proposed that benzamides
111 and butyrophenones do not bind to the same configuration of theD2-receptor. Buty- rophenones may bind primarily to the monomer form, whereas benzamides may bind to both the monomer and dimer forms of the receptor47. In untreated non-psychotic manic patients compared to controls studies with the butyrophenone methylspiperone48,49 and the benzamides iodobenzamide and 50,51 raclopride did not find striatal D2-density difference (table 2). Pearlson et al.
however, did find a higher D2-recepter density in the caudate nucleus of BD patients with psychotic features during their depressive or manic episodes when compared to BD patients during episodes without psychotic features49. Within the group with psychotic features, the severity of the psychotic symptoms correlated with the receptor density, which was not the case with severity of mood symptoms. This sug- gests that the D2-receptor density is specifically related to psychosis but not to mood symptoms. This theory is further supported by the finding that the mood stabilizing
anti-epileptic valproate sodium did not alter the D2-receptor density in non-psychotic manic patients51. 52 Concerning the D1-recepter, Suhara et al. found the binding potential of SCH23390 to be decreased in the frontal cortex of BD patients with various mood states when compared to controls. In the striatum, results were comparable among patients and controls. Dopamine synthesis can be investigated by measuring the 18F-labeled 6-fluo- ro-L-DOPA, which is a precursor to dopamine, as described above. Dopamine syn- thesis was found to be comparable among untreated non-psychotic manic patients and controls. In view of the finding that valproate did not change D2-receptor density, it is interesting that valproate was able to reduce dopamine synthesis in effectively treated manic patients51,53. Perhaps the valproate-induced reduction of dopamine synthesis might be explained by an improved function of the PFC and fronto-limbic network resulting in an enhanced regulation of dopamine in the striatum. Endogenous dopamine release can be measured with an amphetamine challenge, in which dopamine release is stimulated by blocking sequestering via DAT and VMAT2 and inhibiting the breakdown enzyme monoamine oxidase(MOA). In BD amphet- amine challenge elicited a greater behavioral response, as measured with the Brief Psychiatric Rating Scale (BPRS) and the Young Mania Rating Scale (YMRS) in BD patients compared to controls. However, a difference between D2-receptor binding potential of 123I-iodobenzamide between these groups was not found50. Because it is known that benzamide binding can fluctuate during amphetamine induced endog- enous dopamine binding, it cannot be ruled out that BD patients may have a more sensitive dopamine system to challenges with stimulants and treatment with mood stabilizers33.
112 In recent years the DAT gained scientific attention because it is hypothesized that some of the efficacy of mood stabilizing medication may be due to their action on DAT54. In SPECT studies using 99mTc TRODAT-1 DAT density was increased in the right posterior putamen and in the left caudate in depressive BD-II patients55 and in the striatum of euthymic BD-I and BD-II patients 56. However, in untreated BD-I patients, a study using [O-methyl-11C]β-CFT (11C-CFT) PET, showed decreased DAT density in the bilateral dorsal caudate. These contradictive results may be explained by differ- ences in patient groups (BD-I versus BD-II) and the difference in spatial resolution between SPECT and PET57. Using the (+)-α-11C-dihydrotetrabenazine (11C-DTBZ) ligand, a elevated VMAT2 density was found in the thalamus en ventral striatum in euthymic BD patients with a history of psychotic symptoms, which was comparable to SZ patients, but differed from controls58, This would suggest a relation with psychotic symptoms in BD, how- ever, in the absence of research describing the VMAT2 density in BD patients without psychosis, a relation with affective symptoms cannot be ruled out. Overall, it can be assumed that altered dopamine neurotransmission plays a disease modifying role, especially in BD patients that experience psychotic symptoms in addition to affective symptomatology. However, dopamine neurotransmission as a pathophysiological mechanism in non-psychotic BD patients needs further research.
Choline 6 Acetylcholine is a neurotransmitter in both the peripheral nervous system and central
nervous system. In the central nervous system, it has a variety of effects as a neuro- disorders bipolar in SPECT and PET modulator upon plasticity (specifically in learning and memory), salience of sensory stimuli, arousal and reward. Interestingly, cholinesterase inhibitors were found to increase depressive symptoms in BD and MDD patients59 (table 2). Muscarinic type 2 receptor binding was decreased in the ACC of depressed BD pa- tients when compared to MDD patients and controls, using 3-(3-(3-[18F]Flouropropyl) thio)-1,2, 5-thiadiazol-4-yl)-1,2,5,6-tetrahydro-1-methylpyridine (18F-FP-TZTP)60. This decrease in muscarinic type 2 receptor binding in BD patients was associated with a genetic variation in cholinergic muscarinic-2 receptor gene61. Furthermore, the depression and anxiety severity in BD patients were negatively correlated with the binding potentials, emphasizing a contribution of the cholinergic neurotransmitter system in BD pathophysiology.
113 Conclusion
Since the beginning of the earliest PET and SPECT studies in patients with BD-In the 1980s this field of research gave rise to many new insights in the pathophysiology of BD. The first, mainly metabolism and blood flow oriented studies aided to study various aspects of the metabolism based disease model in which PFC hypoactivi- ty is accompanied by limbic hyperactivity. This model in its comprehensive form is however probably not precise enough to account for most of the specific mood and cognitive disease features and efforts are being made to draw into detail. The role of molecular imaging as the main imaging technique in metabolism studies has been taken over by fMRI, but they are still used to answer specific questions in which fMRI falls short. Molecular imaging demonstrated the importance of serotonin transporter alterations in parts of the limbic system in BD and underscored the role of dopamine and cholinergic neurotransmission. Most molecular imaging studies in BD have unique designs, extending the knowledge on the pathophysiological mechanisms, but also complicating comparisons between studies. The earlier studies with selection of heterogeneous patient groups, including both BD-I and BD-II patients and being in different mood states (manic, depressed and euthymic) led to results that were difficult to interpret. Moreover, use of medica- tion can affect study outcomes, while studies with only medication-naïve patients, studies with washout periods and naturalistic studies all have their specific advantag- es but also disadvantages. Naturalistic study designs have the advantage that they are generally easier to perform and less burdensome for patients with this serious psychiatric disorder, but the effect of medication use can never be evaluated in a valid way. The obvious advantage of medication-naïve studies is the exclusion of these medication effects. The question arises however in how far the uniqueness of these patients in that they can function without medication, interferes with the investigat- ed mechanism (i.e. the internal validity) and limits the generalizability (i.e. the external validity). In washout studies one could argue that the withdrawal interferes with the investigated mechanism. Another complicating factor is that the molecular imaging studies are limited in patient size because of careful ethical considerations due to the ionizing nature of the technique, which complicates comparisons between subgroups. Finally, some ligands are generally expected to measure the same biological property but are later found to differ in some specific aspects of the measurement complicating com- parison between studies. Nevertheless, because of its unique selectivity emanating from a continuous extending range of possible ligands, molecular imaging remains an important tool in BD research.
114 References
1. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry [Internet]. 2003 Apr [cited 2012 Mar 1];160(4):636–45. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12668349 2. Drevets WC. Neuroimaging studies of mood disorders. Biol Psychiatry [Internet]. 2000 Oct 15;48(8):813–29. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11301246 3. Dunn RT, Willis MW, Benson BE, Repella JD, Kimbrell T a, Ketter T a, et al. Preliminary findings of uncoupling of flow and metabolism in unipolar compared with bipolar affective illness and normal controls. Psychiatry Res [Internet]. 2005 Nov 30 [cited 2012 Feb 27];140(2):181–98. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16257515 4. Baxter LR, Phelps ME, Mazziotta JC, Schwartz JM, Gerner RH, Selin CE, et al. Cerebral metabolic rates for glucose in mood disorders. Studies with positron emission tomography and fluorodeoxyglucose F 18. Arch Gen Psychiatry [Internet]. 1985 May [cited 2012 Feb 24];42(5): 441–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/3872649 5. Baxter LR, Schwartz JM, Phelps ME, Mazziotta JC, Guze BH, Selin CE, et al. Reduction of prefrontal cortex glucose metabolism common to three types of depression. Arch Gen Psychiatry [Internet]. 1989 Mar [cited 2011 Nov 20];46(3):243–50. Available from: http://www.ncbi.nlm.nih. gov/pubmed/2784046 6. Ketter TA, Kimbrell TA, George MS, Dunn RT, Speer AM, Benson BE, et al. Effects of mood and subtype on cerebral glucose metabolism in treatment-resistant bipolar disorder. Biol Psychiatry [Internet]. 2001;49(2):97–109. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11164756 7. Bauer M, London ED, Rasgon N, Berman SM, Frye M a, Altshuler LL, et al. Supraphysiological doses of levothyroxine alter regional cerebral metabolism and improve mood in bipolar depression. Mol Psychiatry [Internet]. 2005 May [cited 2012 Feb 27];10(5):456–69. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15724143 8. Brooks JO, Wang PW, Strong C, Sachs N, Hoblyn JC, Fenn R, et al. Preliminary evidence of 6 differential relations between prefrontal cortex metabolism and sustained attention in depressed
adults with bipolar disorder and healthy controls. Bipolar Disord [Internet]. 2006 Jun;8(3):248–54. disorders bipolar in SPECT and PET Available from: http://www.ncbi.nlm.nih.gov/pubmed/16696826 9. Buchsbaum MS, Wu J, DeLisi LE, Holcomb H, Kessler R, Johnson J, et al. Frontal cortex and basal ganglia metabolic rates assessed by positron emission tomography with [18F]2-deoxyglucose in affective illness. J Affect Disord [Internet]. 1986;10(2):137–52. Available from: http://www.ncbi.nlm.nih.gov/pubmed/2941470 10. Rush AJ, Schlesser MA, Stokeley E, Bonte FR, Altshuller KZ. Cerebral blood flow in depression and mania. Psychopharmacollogy Bull. 1982;(1):6–7. 11. Silfverskiöld P, Risberg J. Regional cerebral blood flow in depression and mania. Arch Gen Psychiatry [Internet]. 1989 Mar;46(3):253–9. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7568564 12. Tutus a, Simsek a, Sofuoglu S, Nardali M, Kugu N, Karaaslan F, et al. Changes in regional cerebral blood flow demonstrated by single photon emission computed tomography in depressive disorders: comparison of unipolar vs. bipolar subtypes. Psychiatry Res [Internet]. 1998 Sep 28;83(3):169–77. Available from: http://www.ncbi.nlm.nih.gov/pubmed/9849726 13. Miller EK, Freedman DJ, Wallis JD. The prefrontal cortex: categories, concepts and cognition. Philos Trans R Soc Lond B Biol Sci [Internet]. 2002 Aug 29 [cited 2012 Mar 5];357(1424):1123–36. Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=1693009&tool=pmcentrez&rendertype=abstract 14. Blumberg HP, Stern E, Ricketts S, Martinez D, de Asis J, White T, et al. Rostral and orbital prefrontal cortex dysfunction in the manic state of bipolar disorder. Am J Psychiatry [Internet]. 1999 Dec;156(12):1986–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10588416
115 15. Rubinsztein JS, Fletcher PC, Rogers RD, Ho LW, Aigbirhio FI, Paykel ES, et al. Decision-making in mania: a PET study. Brain [Internet]. 2001 Dec;124(Pt 12):2550–63. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11701607 16. Culha AF, Osman O, Dogangün Y, Filiz K, Suna K, Kalkan ON, et al. Changes in regional cerebral blood flow demonstrated by 99mTc-HMPAO SPECT in euthymic bipolar patients. Eur Arch Psychiatry Clin Neurosci [Internet]. 2008 Apr [cited 2011 Oct 19];258(3):144–51. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17990047 17. Krüger S, Alda M, Young LT, Goldapple K, Parikh S, Mayberg HS. Risk and resilience markers in bipolar disorder: brain responses to emotional challenge in bipolar patients and their healthy siblings. Am J Psychiatry [Internet]. 2006 Feb;163(2):257–64. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16449479 18. Rubin E, Sackeim HA, Prohovnik I, Moeller JR, Schnur DB, Mukherjee S. Regional cerebral blood flow in mood disorders: IV. Comparison of mania and depression. Psychiatry Res [Internet]. 1995 May 31 [cited 2012 Feb 27];61(1):1–10. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7568564 19. al-Mousawi a. H, Evans N, Ebmeier KP, Roeda D, Chaloner F, Ashcroft GW. Limbic dysfunction in schizophrenia and mania. A study using 18F- labelled fluorodeoxyglucose and positron emission tomography. Br J Psychiatry [Internet]. 1996 Oct 1 [cited 2012 Feb 27];169(4):509–16. Available from: http://bjp.rcpsych.org/cgi/doi/10.1192/bjp.169.4.509 20. Brooks JO, Hoblyn JC, Woodard S a, Rosen AC, Ketter T a. Corticolimbic metabolic dysregulation in euthymic older adults with bipolar disorder. J Psychiatr Res [Internet]. 2009 Feb [cited 2012 Feb 27];43(5):497–502. Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=2693401&tool=pmcentrez&rendertype=abstract 21. Luu P, Posner MI. Anterior cingulate cortex regulation of sympathetic activity. Brain [Internet]. 2003 Oct [cited 2011 Jul 23];126(Pt 10):2119–20. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12959940 22. Drevets WC, Price JL, Simpson JR, Todd RD, Reich T, Vannier M, et al. Subgenual prefrontal cortex abnormalities in mood disorders. Nature [Internet]. 1997 Apr 24 [cited 2011 Jul 27];386(6627):824–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/9126739 23. Dunn RT, Kimbrell TA, Ketter TA, Frye MA, Willis MW, Luckenbaugh DA, et al. Principal components of the Beck Depression Inventory and regional cerebral metabolism in unipolar and bipolar depression. Biol Psychiatry [Internet]. 2002 Mar 1 [cited 2012 Apr 25];51(5):387–99. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0006322301012446 24. Mah L, Zarate C a, Singh J, Duan Y-F, Luckenbaugh D a, Manji HK, et al. Regional cerebral glucose metabolic abnormalities in bipolar II depression. Biol Psychiatry [Internet]. 2007 Mar 15 [cited 2011 Oct 19];61(6):765–75. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17027930 25. Blumberg HP, Stern E, Martinez D, Ricketts S, de Asis J, White T, et al. Increased anterior cingulate and caudate activity in bipolar mania. Biol Psychiatry [Internet]. 2000 Dec 1;48(11):1045–52. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11094137 26. Benabarre A, Vieta E, Martínez-Arán A, Garcia-Garcia M, Martín F, Lomeña F, et al. Neuropsychological disturbances and cerebral blood flow in bipolar disorder. Aust N Z J Psychiatry [Internet]. 2005 Apr [cited 2012 Mar 6];39(4):227–34. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15777358 27. Goodwin GM, Cavanagh JT, Glabus MF, Kehoe RF, O’Carroll RE, Ebmeier KP. Uptake of 99mTc- exametazime shown by single photon emission computed tomography before and after lithium withdrawal in bipolar patients: associations with mania. Br J Psychiatry [Internet]. 1997 May 1 [cited 2012 Feb 27];170(5):426–30. Available from: http://bjp.rcpsych.org/cgi/doi/10.1192/ bjp.170.5.426 28. Packard MG, Knowlton BJ. Learning and memory functions of the Basal Ganglia. Annu Rev Neurosci [Internet]. 2002 Jan [cited 2011 Jun 13];25:563–93. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12052921
116 29. Bonne O, Krausz Y, Gorfine M, Karger H, Gelfin Y, Shapira B, et al. Cerebral hypoperfusion in medication resistant, depressed patients assessed by Tc99m HMPAO SPECT. J Affect Disord [Internet]. 1996 Dec 16 [cited 2012 Apr 25];41(3):163–71. Available from: http://www.sciencedirect.com/science/article/pii/s0165032796000584 30. Ito H, Kawashima R, Awata S, Ono S, Sato K, Goto R, et al. Hypoperfusion in the limbic system and prefrontal cortex in depression: SPECT with anatomic standardization technique. J Nucl Med [Internet]. 1996 Mar;37(3):410–4. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8772633 31. Cohen RM, Semple WE, Gross M, Nordahl TE, King AC, Pickar D, et al. Evidence for common alterations in cerebral glucose metabolism in major affective disorders and schizophrenia. Neuropsychopharmacology [Internet]. 1989 Dec [cited 2012 Feb 27];2(4):241–54. Available from: http://www.ncbi.nlm.nih.gov/pubmed/2610821 32. Drevets WC, Price JL, Bardgett ME, Reich T, Todd RD, Raichle ME. Glucose metabolism in the amygdala in depression: relationship to diagnostic subtype and plasma cortisol levels. Pharmacol Biochem Behav [Internet]. 2002 Mar;71(3):431–47. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11830178 33. Gonul AS, Coburn K, Kula M. Cerebral blood flow, metabolic, receptor, and transporter changes in bipolar disorder: the role of PET and SPECT studies. Int Rev Psychiatry [Internet]. 2009 Jan [cited 2011 Dec 8];21(4):323–35. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20374147 34. Gyulai L, Alavi A, Broich K, Reilley J, Ball WB, Whybrow PC. I-123 iofetamine single-photon computed emission tomography in rapid cycling bipolar disorder: a clinical study. Biol Psychiatry [Internet]. 1997;41(2):152–61. Available from: http://www.ncbi.nlm.nih.gov/pubmed/9018385 35. Schildkraut JJ. The catecholamine hypothesis of affective disorders: a review of supporting evidence. Am J Psychiatry [Internet]. 1965 Nov [cited 2012 Mar 7];122(5):509–22. Available from: http://www.ncbi.nlm.nih.gov/pubmed/5319766 36. Ichimiya T, Suhara T, Sudo Y, Okubo Y, Nakayama K, Nankai M, et al. Serotonin transporter binding in patients with mood disorders: a PET study with [11C](+)McN5652. Biol Psychiatry [Internet]. 2002 May 1 [cited 2013 Nov 28];51(9):715–22. Available from: 6 http://www.ncbi.nlm.nih.gov/pubmed/11983185 37. Oquendo M a, Hastings RS, Huang Y-Y, Simpson N, Ogden RT, Hu X-Z, et al. Brain serotonin transporter binding in depressed patients with bipolar disorder using positron emission disorders bipolar in SPECT and PET tomography. Arch Gen Psychiatry [Internet]. 2007 Feb;64(2):201–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17283287 38. Chou Y-H, Wang S-J, Lin C-L, Mao W-C, Lee S-M, Liao M-H. Decreased brain serotonin transporter binding in the euthymic state of bipolar I but not bipolar II disorder: a SPECT study. Bipolar Disord [Internet]. 2010 May [cited 2012 Oct 11];12(3):312–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20565438 39. Cannon DM, Ichise M, Fromm SJ, Nugent AC, Rollis D, Gandhi SK, et al. Serotonin transporter binding in bipolar disorder assessed using [11C]DASB and positron emission tomography. Biol Psychiatry [Internet]. 2006 Aug 1 [cited 2011 Jul 27];60(3):207–17. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16875929 40. Cannon DM, Ichise M, Rollis D, Klaver JM, Gandhi SK, Charney DS, et al. Elevated serotonin transporter binding in major depressive disorder assessed using positron emission tomography and [11C]DASB; comparison with bipolar disorder. Biol Psychiatry [Internet]. 2007 Oct 15 [cited 2011 Jun 15];62(8):870–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17678634 41. Yatham LN, Liddle PF, Lam RW, Adam MJ, Solomons K, Chinnapalli M, et al. A positron emission tomography study of the effects of treatment with valproate on brain 5-HT2A receptors in acute mania. Bipolar Disord [Internet]. 2005 Jan;7 Suppl 5:53–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16225561 42. Beaulieu J, Gainetdinov RR. The physiology, signaling, and pharmacology of dopamine receptors. Pharmacol Rev [Internet]. 2011 Mar [cited 2012 Jul 27];63(1):182–217. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21303898
117 43. Cousins D a, Butts K, Young AH. The role of dopamine in bipolar disorder. Bipolar Disord [Internet]. 2009 Dec;11(8):787–806. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19922550 44. Hartman DS, Civelli O. Molecular attributes of dopamine receptors: new potential for antipsychotic drug development. Ann Med [Internet]. 1996 Jun;28(3):211–9. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8811164 45. Little KY, Krolewski DM, Zhang L, Cassin BJ. Loss of striatal vesicular monoamine transporter protein (VMAT2) in human cocaine users. Am J Psychiatry [Internet]. 2003 Jan [cited 2012 Mar 8];160(1):47–55. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12505801 46. Yildiz A, Vieta E, Leucht S, Baldessarini RJ. Efficacy of antimanic treatments: meta-analysis of randomized, controlled trials. Neuropsychopharmacology [Internet]. 2011 Jan [cited 2012 Mar 7];36(2):375–89. Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=3055677&tool=pmcentrez&rendertype=abstract 47. Ginovart N. Imaging the dopamine system with in vivo [11C]raclopride displacement studies: understanding the true mechanism. Mol Imaging Biol [Internet]. 2005 [cited 2012 Mar 7];7(1):45– 52. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15912275 48. Wong DF, Wagner HN, Pearlson G, Dannals RF, Links JM, Ravert HT, et al. Dopamine receptor binding of C-11-3-N-methylspiperone in the caudate in schizophrenia and bipolar disorder: a preliminary report. Psychopharmacol Bull [Internet]. 1985 Jan [cited 2012 Mar 7];21(3):595–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/4034877 49. Pearlson GD, Wong DF, Tune LE, Ross CA, Chase GA, Links JM, et al. In vivo D2 dopamine receptor density in psychotic and nonpsychotic patients with bipolar disorder. Arch Gen Psychiatry [Internet]. 1995 Jun [cited 2012 Mar 7];52(6):471–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7771917 50. Anand a, Verhoeff P, Seneca N, Zoghbi SS, Seibyl JP, Charney DS, et al. Brain SPECT imaging of amphetamine-induced dopamine release in euthymic bipolar disorder patients. Am J Psychiatry [Internet]. 2000 Jul;157(7):1108–14. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10873919 51. Yatham LN, Liddle PF, Lam RW, Shiah I-S, Lane C, Stoessl AJ, et al. PET study of the effects of valproate on dopamine D(2) receptors in neuroleptic- and mood-stabilizer-naive patients with nonpsychotic mania. Am J Psychiatry [Internet]. 2002 Oct [cited 2012 Mar 7];159(10):1718–23. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12359678 52. Suhara T, Nakayama K, Inoue O, Fukuda H, Shimizu M, Mori A, et al. D1 dopamine receptor binding in mood disorders measured by positron emission tomography. Psychopharmacology (Berl) [Internet]. 1992 Jan [cited 2012 Apr 25];106(1):14–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/1531387 53. Yatham LN, Liddle PF, Shiah I-S, Lam RW, Ngan E, Scarrow G, et al. PET study of [(18)F]6-fluoro-L- dopa uptake in neuroleptic- and mood-stabilizer-naive first-episode nonpsychotic mania: effects of treatment with divalproex sodium. Am J Psychiatry [Internet]. 2002 May [cited 2012 Mar 7];159(5):768–74. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11986130 54. Yatham LN, Goldstein JM, Vieta E, Bowden CL, Grunze H, Post RM, et al. Atypical antipsychotics in bipolar depression: potential mechanisms of action. J Clin Psychiatry [Internet]. 2005 Jan [cited 2012 Mar 8];66 Suppl 5:40–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16038601 55. Amsterdam JD, Newberg AB. A preliminary study of dopamine transporter binding in bipolar and unipolar depressed patients and healthy controls. Neuropsychobiology [Internet]. 2007 Jan [cited 2012 Mar 4];55(3–4):167–70. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17657170 56. Chang TT, Yeh TL, Chiu NT, Chen PS, Huang HY, Yang YK, et al. Higher striatal dopamine transporters in euthymic patients with bipolar disorder: a SPECT study with [Tc] TRODAT-1. Bipolar Disord [Internet]. 2010 Feb;12(1):102–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20148872
118 57. Anand A, Barkay G, Dzemidzic M, Albrecht D, Karne H, Zheng Q-H, et al. Striatal dopamine transporter availability in unmedicated bipolar disorder. Bipolar Disord [Internet]. 2011 Jun [cited 2012 Feb 27];13(4):406–13. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21843280 58. Zubieta JK, Taylor SF, Huguelet P, Koeppe R a, Kilbourn MR, Frey K a. Vesicular monoamine transporter concentrations in bipolar disorder type I, schizophrenia, and healthy subjects. Biol Psychiatry [Internet]. 2001 Jan 15;49(2):110–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11164757 59. Dilsaver SC. Pathophysiology of “cholinoceptor supersensitivity” in affective disorders. Biol Psychiatry [Internet]. 1986 Jul [cited 2012 Mar 8];21(8–9):813–29. Available from: http://www.ncbi.nlm.nih.gov/pubmed/3015271 60. Cannon DM, Carson RE, Nugent AC, Eckelman WC, Kiesewetter DO, Williams J, et al. Reduced muscarinic type 2 receptor binding in subjects with bipolar disorder. Arch Gen Psychiatry [Internet]. 2006 Jul;63(7):741–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16818863 61. Cannon DM, Klaver JK, Gandhi SK, Solorio G, Peck S a, Erickson K, et al. Genetic variation in cholinergic muscarinic-2 receptor gene modulates M2 receptor binding in vivo and accounts for reduced binding in bipolar disorder. Mol Psychiatry [Internet]. 2011 Apr [cited 2012 Feb 27];16(4):407–18. Available from: http://www.pubmedcentral.nih.gov/articlerender. fcgi?artid=2895691&tool=pmcentrez&rendertype=abstract 62. Lynch PJ, Jaffe CC. Brain human sagittal section [Internet]. Wikimedia Commons. 2006. Available from: https://commons.wikimedia.org/wiki/File:Brain_human_sagittal_section.svg?uselang=en
6 PET and SPECT in bipolar disorders bipolar in SPECT and PET
119
CHAPTER 7 Neuroinflammation in bipolar disorder
A [11C]-(R)-PK11195 positron emission tomography study
Bartholomeus C.M. Haarman, Rixt F. Riemersma – Van der Lek, Jan Cees de Groot, Henricus G. Ruhé, Hans C. Klein, Tjitske E. Zandstra, Huibert Burger, Robert A. Schoevers, Erik F.J. de Vries, Hemmo A. Drexhage, Willem A. Nolen, Janine Doorduin
Brain, Behavior, and Immunity. 2014 Aug 3;40:219–25. Abstract
Background The “monocyte-T-cell theory of mood disorders” regards neuroinflammation, i.e. marked activation of microglia, as a driving force in bipolar disorder. Microglia activa- tion can be visualized in vivo using [11C]-(R)-PK11195 PET. Indirect evidence suggests the hippocampus as a potential focus of neuroinflammation in bipolar disorder. We aim to demonstrate that there is increased [11C]-(R)-PK11195 binding to activated microglia in the hippocampus of patients with bipolar I disorder when compared to healthy controls.
Material and Methods Fourteen patients with bipolar I disorder and eleven healthy controls were includ- ed in the analyses. Dynamic 60-min PET scans were acquired after the injection of [ 11C]-(R)-PK11195. All subjects underwent psychiatric interviews as well as an MRI scan, which was used for anatomic co-registration in the data analysis. The data from the PET scans was analyzed with a two-tissue-compartment model to calculate the binding potential, using the metabolite-corrected plasma and blood curve as input.
Results A significantly increased [11C]-(R)-PK11195 binding potential, which is indicative of neuroinflammation, was found in the right hippocampus of the patients when compared to the healthy controls (1.66 (CI 1.45 – 1.91) versus 1.33 (CI 1.16 – 1.53); p=0.033, respectively). Although the same trend was observed in the left hippo campus, this difference was not statistically significant.
Conclusion This study is the first to demonstrate the presence of focal neuroinflammation in the right hippocampus in bipolar I disorder.
122 Introduction
The pathophysiology of bipolar disorder (BD) is complex and its neurobiology remains largely unknown1. Both the stress system and the immune system interact with the brain and are influenced by the environment. Their interactions can be regarded as possible linking pins. The “monocyte-T-cell theory of mood disorders” 2 considers an activated inflammatory response system (IRS) in mood disorders to be the driv- ing force behind these illnesses. IRS activation can be regarded as a disbalance in immune regulatory processes. In BD this theory is supported by altered concentra- tions of immune related peripheral bio-assays, e.g. elevated serum or plasma levels of pro-inflammatory cytokines, aberrant expression of pro-inflammatory genes in circulating monocytes3, alterations in the kynurenine pathway4 and a modulating effect of several psychopharmaceuticals on the immune system3,5–8. Activation of the IRS is thought to correspond to neuroinflammation, which is reflected by an increase in activated microglia, the resident macrophages of the brain9. Indirect evidence gathered from post-mortem studies, corticosteroid treat- ment related hippocampal changes and animal models suggest the hippocampus as a potential focus of neuroinflammation in BD. Post-mortem studies in humans demonstrated an increased expression of inflammation related pro-apoptosis genes10 and oxidative damage11 to the RNA in the hippocampus of BD patients as well as a decrease in growth-associated protein12 that has been proposed to be related to neuroinflammation13. In addition, medical treatment with corticosteroids, known for their anti-inflammatory effects and associated with not only depressive symptoms but also manic symptoms, is related to both functional and morphological changes in the hippocampus14. Furthermore, multiple rodent studies demonstrate a relationship 7 between a disturbed microglia function in the hippocampus and other pathophysi- 9,15–19 ological changes which are thought to relate to neuropsychiatric disorders . In a Neuroinflammation in bipolar disorder recent rodent study stress-induction led to dynamic microglia changes in the hippo- campus alone, which were associated with depressive-like behavior18. Another study demonstrated an increased cytokine response to lipopolysaccharide challenge in the hippocampus of SERT mutant rats19. Microglia activation can be visualized in vivo with the radiopharmaceutical [ 11C]-(R)-PK11195 by means of positron emission tomography (PET). This radiophar- maceutical binds to the translocator protein (TSPO), a receptor that is upregulated in the mitochondria of activated microglia cells20. [11C]-(R)-PK11195 has been utilized successfully in models of central inflammation, such as following an injection of the endotoxin lipopolysaccharide in animal models21–23 and following infections of the central nervous system in humans24–26. In various psychiatric and neurodegenerative disorders [11C]-(R)-PK11195 PET has proven to be a useful tool for imaging neuro inflammation27–30. Using this radiopharmaceutical Doorduin et al. demonstrated the
123 hippocampus to be the primary focus of neuroinflammation in schizophrenia-related psychosis30. New radiopharmaceuticals such as [11C]-PBR28 have been developed that are potentially more sensitive for imaging of neuroinflammation. However, these revealed substantial heterogeneity in binding affinity due to polymorphisms in the TSPO, resulting in so-called high-, median- and low-affinity binders, complicating data interpretation31. Since [11C]-(R)-PK11195 binding is not affected by polymorphisms we have selected this radiopharmaceutical for our study. In the current study we aim to demonstrate an increased [11C]-(R)-PK11195 binding to activated microglia in BD patients in comparison to a healthy control group. We a priori hypothesized the hippocampus to be the main focus of neuroinflammation in BD. In a second model we explored the presence of neuroinflammation in other brain regions. In addition, we examined whether clinical characteristics would be associated with neuroinflammation.
124 Materials and Methods
Participants For the present cross-sectional case-control study we included 15 preferably euthymic patients with bipolar I disorder (BD-I) and a group of 12 controls matching demographically in age and sex that participated in the MOODINFLAME study. The patients were recruited from an outpatient clinic for BD. The healthy controls were recruited via advertisements, recruitment posters and by contacting healthy controls from previous studies that gave their consent to be asked for future studies. For the MOODINFLAME study adult male and female subjects were included who were free of inflammation related symptoms, including fever and infectious or inflammatory disease. Furthermore, they were free of uncontrolled systemic disease, uncontrolled metabolic disease or other significant uncontrolled somatic disorder known to affect mood. They did not use somatic medication known to affect mood or the immune system, such as corticosteroids, non-steroid anti-inflammatory drugs and statins. Female candidates who were pregnant or recently gave birth were excluded. Patients and controls were free of benzodiazepines at least in the last week prior to the PET- scan. They were also free of anticoagulant use or presence of coagulation disease, did not suffer from palmar arc artery insufficiency, did not participate in a prior research study involving radiation less than a year ago, and did not have any contraindication for MRI scanning. Patients were allowed to continue their regular psychopharmaceutical treatment. They were neither in a depressed nor (hypo-)manic episode at the time of scanning as indicated by an Inventory of Depressive Symptoms - Clinician Version (IDS-C30) score <22 and Young Mania Rating Scale (YMRS) score <12, respectively. Patients 7 with any other current primary major psychiatric diagnosis were excluded including: schizophrenia, schizoaffective disorder, anxiety disorder and substance use disorders. Neuroinflammation in bipolar disorder Healthy controls did not have any current or lifetime psychiatric diagnosis. Nine patients were excluded due to protocol violations (claustrophobia (3), presence of ferromagnetic objects (2), palmar arc artery insufficiency (1), coagulation disorder (1), use of benzodiazepines (1) and pregnancy (1)). Four healthy controls were ex- cluded due to protocol violations (claustrophobia (2) and presence of ferromagnetic objects (2)). After completing inclusion one participant admitted to having used a benzodiazepine on the evening prior to the PET-scan. In another participant, expe- rienced technical difficulties with the automatic blood sampling system prohibited valid determination of the input function. These two subjects were removed from the subsequent analyses.
125 Ethical Considerations The Medical Ethical Review Committee of the University Medical Center Gronin- gen approved the protocol, which was performed in accordance with the Helsinki Declaration of 1975. Written informed consent was obtained from all participants.
Assessment All subjects underwent a Mini-International Neuropsychiatric Interview 5.0.0 (MINI) to confirm the bipolar I disorder diagnosis in the patient group and the absence of psychiatric disorders in the healthy control group32. Clinical features were extracted from the interviews held according to the MOODINFLAME protocol. This included the Patient Questionnaire of the former Stanley Foundation Bipolar Network, the YMRS, the IDS-C30 and a somatic illness questionnaire. The Patient Questionnaire includes separate clinician and patient chapters covering a spectrum of clinical features including vocational, educational and economic sta- tus, onset and course of illness, family history, past treatment, cycling and seasonal patterns, medical problems, medications, ability to function and symptomatic status, precipitants of illness (e.g. substance use), treatment adherence and insight into the illness33. In the event of a mismatch of results from the MINI in relation to the Patient Questionnaire, diagnoses were checked with the treating physician. The YMRS is an 34 eleven-item, multiple-choice questionnaire to assess manic symptoms . The IDS-C30 is a thirty-item, multiple-choice questionnaire to assess depressive symptoms of all 35 symptom domains of depression . The YMRS and IDS-C30 were assessed shortly before the scans and used in the relevant analyses. The somatic illness questionnaire is a MOODINFLAME specific checklist exploring all the organ systems for current and lifetime medical symptoms.
Radiochemistry [11C]-(R)-PK11195 was labeled as described previously 30. [11C]-(R)-PK11195 was ob- tained in 32±18% radiochemical yield (n=27). The quality control was performed by
HPLC, using a Novapak C18 column (150x3.9 mm) with acetonitrile/25 mM NaH2PO4 (pH 3.5) (60/40) as the eluent at a flow of 1 ml/min. The radiochemical purity was always >95% and the specific activity was 111±130 GBq/μmol. No differences were found between healthy volunteers and patients for the injected dose (390±18 vs. 355±63 MBq, p=0.088) and injected mass (0.75±0.58 vs. 0.67±0.40 mg/L, p=0.696).
Positron Emission Tomography For arterial blood sampling a catheter was inserted in the radial artery after testing for collateral circulation with the Allen test and injection of 1% lidocaine (Fresenius Kabi Nederland BV, ‘s Hertogenbosch, The Netherlands) for local anesthesia. In the other arm, a venous catheter was inserted in the antebrachial vein for injection of
126 [11C]-(R)-PK11195. Positron emission tomography imaging was performed with the ECAT EXACT HR+ camera (Siemens, Knoxville, Tenessee). Head movement was mini- mized with a head-restraining adhesive band and a neuroshield was used to minimize the interference of radiation from the subject’s body. A 60-minute emission scan in 3D-mode was performed, starting simultaneously with the intravenous injection of [11C]-(R)-PK11195. The tracer was injected at a speed of 0.5 ml/sec (total volume of 8.3 ml). After radiotracer injection, arterial blood radioactivity was continuously monitored with an automated blood sampling system (Veenstra Instruments, Joure, The Netherlands). Five extra blood samples were collected at 10, 20, 30, 45 and 60 minutes after [11C]-(R)-PK11195 injection to determine the amount of radioactivity in blood and plasma to calibrate the sampling system. The arterial blood samples that were collected at 20, 45 and 60 minutes after [11C]-(R)-PK11195 injection were also used for metabolite analysis. The metabolite analysis was performed as described previously30.
Magnetic Resonance Imaging Axial T1 (gradient echo T1 3D, slice thickness 1.2 mm isotropic) and T2 Flair (3 mm) weighed images were acquired using a 3T MRI scanner and an eight-channel head coil (3T Intera, Philips, Best, The Netherlands). The T1 images provided the input for the normalization of the PET scans to standard brain morphology of all subjects. Both the MRI scan and the PET scan were preferably made on the same day, and no more than 1 week apart. Image Analysis 7 Attenuation correction was performed with the separate ellipse algorithm. Images were reconstructed by filtered back projection in 21 successive frames of increas- Neuroinflammation in bipolar disorder ing duration (6x 10 s, 2x 30 s, 3x 1 min, 2x 2min, 2x 3 min, 3x 5min, 3x 10 min). MRI images were co-registered to the sum of all frames of the PET scan, resulting in the most optimal co-registration, using statistical parametric mapping (SPM8; Welcome Trust Center Neuroimaging, University College London, UK). Grey matter regions of interest (ROIs) were defined by the co-registered MRI images using a probability map that was based on automatic delineation of ROIs with the PVElab software36. The ROIs were transferred to the dynamic PET images and time-activity curves were calculated. In total 15 ROIs were included: left and right hippocampus, left and right frontal cortex, left and right dorsolateral prefrontal cortex (PFC), left and right temporal cortex, left and right parietal cortex, bilateral occipital cortex, bilateral anterior cingulate, bilateral posterior cingulate, bilateral cerebellum and basal ganglia. The time-activity curves of all ROIs were used for kinetic modeling with software developed in Matlab 7.1 (Mathworks, Natick, Massachusets). The individual delay was
127 corrected for the delay in radioactivity measurements in blood, as a result of the dis- tance between the subject and the automated blood sampling system. A two-tissue compartment model was used to calculate the k1-k4 with the metabolite corrected plasma and blood curve as an input function, correcting for the individual delay and a free blood volume. The binding potential was defined as k3/k4 and was calculated for each ROI individually.
Statistical Analysis Statistical analyses were performed using Stata Statistical Software, release 11 (StataCorp. 2009, College Station, TX). The differences in demographic data between the groups were investigated with Student’s t-test (age), Pearson’s chi-squared test (gender) and Kruskal-Wallis
equality-of-populations rank test (IDS-C30 score). Student’s t-test was used to determine differences in whole-brain grey matter binding potential between the patient and healthy control group. Statistical analyses of the binding potentials in the examined brain regions were per- formed using two general linear models. The first model investigated the hypothesis that the hippocampi are the focus of neuroinflammation, incorporating the binding potentials of the left and right hippocampus as dependent variables. In the second model, exploring the other brain regions, binding potentials of all investigated brain regions were added as dependent variables. In both models the whole-brain grey matter binding potential was used as a covariate to normalize for individual global cerebral [11C]-(R)-PK11195 uptake variations. Beforehand inverse square root trans- formation was applied to the binding potentials to meet the normality assumption in the general linear models. Results of the general linear models are presented as back-transformed means of the individual ROIs with 95% confidence intervals (CI). The level of significance was defined as 0.05, two-sided, in the first (hypotheses driven) model. To correct for multiple comparisons in the second (explorative) model false discovery rate (FDR<0.1) correction was applied to the results, as described by Benjamini-Hochberg37. In order to rule out possible epilepsy associated inflammation effects38 post hoc analyses were performed excluding the patient with this comorbidity. Furthermore, to increase the homogeneity of the sample additional post hoc analyses were performed excluding the patient with mild depressive symptoms and the medication free patient.
Correlations between the binding potentials and the IDS-C30 score, YMRS score, age at onset, number of episodes and duration of illness were assessed with Spearman’s rho.
128 Results
Demographics Subject characteristics are displayed in Table 1. While all but one patient were euthymic (IDS-C30 score <12), a statistical difference in the IDS-C30 score between the patient and healthy control groups was observed (H=4.676; p=0.031). Differences between the groups in gender or age were not statistically significant.
[11C]-(R)-PK11195 PET There was no statistical difference in the whole-brain grey matter binding potential of [11C]-(R)-PK11195 in patients versus controls (1.28 (CI 1.07 – 1.50) versus 1.35 (CI 1.10 – 1.60); p=0.70). The hypothesis driven general linear model with the left and right hippocampi bind- ing potentials as the dependent variables demonstrated a significant increased [ 11C]-(R)-PK11195 binding potential in the right hippocampus of the patients when compared to the healthy controls (1.66 (CI 1.45 – 1.91) versus 1.33 (CI 1.16 – 1.53); p=0.033; figure 1). The difference between the [11C]-(R)-PK11195 binding potential of the left hippocampus of BD-I patients compared to the healthy controls was not statistically significant (1.55 (CI 1.30 – 1.90) versus 1.20 (CI 1.00 – 1.46); p=0.071). The subsequent explorative general linear model adding all the investigated brain regions revealed a lower [11C]-(R)-PK11195 binding potential in the left dorsolateral PFC of the BD-I patients when compared tot healthy controls (1.18 (CI 1.09 – 1.27) versus 1.40 (CI 1.28 – 1.53); p=0.009). However, this difference did not survive correction for the false discovery rate (table 2). Post hoc analyses that were performed excluding the patient with epilepsy demon- 7 strated an increased [11C]-(R)-PK11195 binding potential in both the right hippo- campus (1.69 (CI 1.47 – 1.96) versus 1.33 (CI 1.17 – 1.54); p=0.029) as well as the left Neuroinflammation in bipolar disorder hippocampus (1.63 (CI 1.35 – 2.00) versus 1.21 (CI 1.02 – 1.47); p=0.042) of the BD-I patients when compared to healthy controls. In the explorative model the results were comparable to the original analyses in effect size and statistical significance. In post hoc analyses excluding the patient with mild depressive symptoms and the medication free patient the results were comparable to the original analyses in effect size and statistical significance in both models.
129
– Inventory of 30
Medication
COCP(5)
roate, Thyroxine, Omeprazol (32) Omeprazol Thyroxine, roate,
(time since last medication switch (months)) switch medication last since (time
(21) ithium, Valp
Valproate, Thyroxine, Trazodone, Lamotrigine, Quetiapine (11) Quetiapine Lamotrigine, Trazodone, Thyroxine, Valproate, L (63) Lamotrigine Lithium, Thyroxine, Lithium, (43) Trazodone (14) Citalopram Carbamazepine, (25) Lamotrigine Lithium, (11) Quetiapine Valproate, (13) Trazodone Lithium, (30) Valproate (4) Quetiapine Lithium, - (10) Levetiracetam Valproate, (10) Lithium ------
a
Medical Medical ypothyroidism h
comorbidities ypothyroidism
Epilepsy - DH; - H ------Current medical comorbidities requiring medical IDS-C care. requiring medical comorbidities Current
a
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 score YMRS
1
2 2 2 2 2 0 5 0 5 0 0 0 0 0 0 6 3 3 3 4 4 11
12 14 30 C - core s IDS
episodes episodes
№
/1) depressed/(hypo)manic) (>5/>11) (>11/>5) (>5/3) (>11/>11) (>5/>5) (>20/>20) (>20/>20) (>5/3) (>11/2) (4/4) (>20/>20) (0 (1/2) (1/>11)
17 17 12 21 21 18 15 18 16 14 22 25 30 43 Age at onset at
21 21 61 41 37 22 52 25 58 55 50 55 50 26 56 56 53 66 36 36 24 40 45 49 64 Age
Sex F F M F F F M M M F M M M F F F F M F F F F M M M
1 1 7 7 2 2 5 8 5 8 6 9 6 9 3 3 4 4 11 11 № 12 10 10 13 14 Patients controls Healthy
TABLE 1 characteristics Subject Overview of the characteristics of the BD-I patients and healthy controls. Depressive Symptoms Clinician version; YMRS Mania – Young Rating Scale; DH – diaphragmatic hernia; COCP – combined oral contraceptive pill
130 FIGURE 1 Results of the hypothesis driven analysis
Mean adjusted [11C]-(R)-PK11195 binding potentials in the left and right hippocampus of healthy controls and BD-I patients. Statistical analysis on the binding potentials was performed using a general linear model, with the whole-brain grey matter binding potential as a covariate to correct for global [11C]-(R)-PK11195 uptake. Beforehand inverse square root transformation was applied to the binding potentials to meet the normality assumption in the general linear model. Results are presented as bars (back-transformed means) with error stripes (standard error). A significantly increased [11C]-(R)-PK11195 binding potential was observed in the right hippocampus of the patients when compared to the healthy controls (p=0.033). * statistically significant p<0.05. 7 Neuroinflammation in bipolar disorder
131 TABLE 2 Results of the explorative analysis
Adjusted [11C]-(R)-PK11195 Binding Potential p Region of Interest Healthy Controls BD-I Patients (mean (CI)) (mean (CI)) Left Hippocampus 1.20 (1.00 - 1.46) 1.55 (1.30 - 1.90) 0.071 Right Hippocampus 1.33 (1.16 - 1.53) 1.66 (1.45 - 1.91) 0.033 Left Frontal Cortex 1.24 (1.10 - 1.41) 1.24 (1.12 - 1.39) 0.184 Left Dorsolateral PFC 1.40 (1.28 - 1.53) 1.18 (1.09 - 1.27) 0.009 Left Temporal Cortex 1.19 (1.09 - 1.29) 1.26 (1.17 - 1.36) 0.315 Left Parietal Cortex 1.37 (1.26 - 1.50) 1.33 (1.23 - 1.44) 0.638 Right Frontal Cortex 1.20 (1.12 - 1.28) 1.27 (1.20 - 1.35) 0.184 Right Dorsolateral PFC 1.25 (1.12 - 1.42) 1.25 (1.13 - 1.39) 0.970 Right Temporal Cortex 1.21 (1.13 - 1.31) 1.32 (1.24 - 1.42) 0.111 Righ Parietal Cortex 1.40 (1.27 - 1.57) 1.40 (1.28 - 1.54) 0.956 Occipital Cortex 1.28 (1.19 - 1.37) 1.40 (1.31 - 1.49) 0.081 Anterior Cingulate 1.39 (1.24 - 1.56) 1.29 (1.17 - 1.43) 0.397 Posterior Cingulate 1.26 (1.10 - 1.46) 1.36 (1.20 - 1.55) 0.456 Cerebellum 1.01 (0.94 - 1.10) 1.13 (1.05 - 1.22) 0.055 Basal Ganglia 1.24 (1.16 - 1.31) 1.24 (1.18 - 1.31) 0.891
Mean adjusted [11C]-(R)-PK11195 binding potentials in the left hippocampus, right hippocampus, left frontal cortex, left dorsolateral prefrontal cortex (PFC), left parietal cortex, left temporal cortex, right frontal cortex, right dorsolateral PFC, right parietal cortex, right, temporal cortex, occipital cortex, anterior cingulate, posterior cingulate, basal ganglia and cerebellum of healthy controls and BD-I patients. Statistical analysis on the binding potentials was performed using a general linear model, with the whole-brain grey matter binding potential as a covariate to correct for global [11C]-(R)-PK11195 uptake. Beforehand inverse square root transformation was applied to the binding potentials to meet the normality assumption in the general linear model. Results are presented as back-transformed means with confidence interval (CI). A lower [11C]-(R)-PK11195 binding potential in the left dorsolateral PFC of the patients when compared tot healthy controls (1.18 (CI 1.09 – 1.27) versus 1.40 (CI 1.28 – 1.53); p=0.009) was revealed. However this difference did not survive correction for the false discovery rate (FDR<0.1).
Association with Clinical Features
Correlations between the hippocampi binding potentials and the total IDS-C30 score or individual items were not statistically significant in both the patient and healthy control group. All patients and healthy controls scored 0 on the YMRS. Therefore, correlations could not be calculated with manic symptoms. The correlations between the hippocampal binding potentials and the illness progres- sion characteristics (number of depressive or manic episodes, total number of mood episodes) were also not statistically significant.
132 Discussion
To our knowledge this is the first study to reveal actual neuroinflammation in vivo in BD. We partly confirmed our a priori hypothesis, demonstrating a statistically significant increased binding potential of [11C]-(R)-PK11195 in the right hippo- campus of BD-I patients as compared to healthy controls. The left hippocampus [ 11C]-(R)-PK11195 binding potential showed the same trend as the right hippocampus, with a comparable increase in binding potential, but it was not statistically significant. Although the effect size had the same magnitude as the right hippocampus, a slightly larger standard error was observed in the calculations and the study may have been underpowered to demonstrate a difference in the left hippocampus between patients and controls. This supposition is supported by a post hoc analysis excluding the patient with epilepsy that demonstrated a significant difference in the left hippocampus as well. The finding of neuroinflammation in BD corroborates previous studies which used less direct indicators of immune activation: an increase in peripheral TSPO receptors in platelets of BD patients, described by Marazziti39; peripheral blood monocyte gene expression found to be related to hemodynamic changes measured by functional MRI in the hippocampus of a combined sample of unipolar and bipolar depressed patients described by Savitz et al.40 and multiple studies investigating immune system related peripheral blood derived bio-assays, described above3,4,8,9. To increase the understanding of the role of immune activation in the pathophysiology of BD further research on the relationship between the peripheral blood derived bioassays and central nervous system neuroinflammation is necessary. The same holds true on the relationship between the various functional neuroimaging 7 observations and neuroinflammation. Animal model studies can be directive in this regard. These demonstrated that microglia have an active role in the development of Neuroinflammation in bipolar disorder mature synapses during embryogenesis41, pruning synapses postnatally42, regulating neurogenesis43 and inducing apoptosis9 in the hippocampus as well as other regions. It is tempting to speculate that these cellular processes (partially) explain the meta- bolic disturbances44,45 and the decreased neuronal viability observed in neuroimaging studies46. The decreased [11C]-(R)-PK11195 binding potential in the left dorsolateral PFC of BD-I patients compared to controls in the explorative analysis is possibly a false-positive finding as it was no longer statistically significant after FDR correction for multiple testing. However, a possible differential [11C]-(R)-PK11195 binding potential between the right hippocampus and the left dorsolateral PFC could also be regarded in the view of recent resting state fMRI connectivity studies that demonstrated aberrant connectivity between the PFC regions and limbic system regions in BD47.
133 It must be noted that our patients were almost all in the euthymic state, so they were not markedly depressed. It remains uncertain whether the inflammatory response would be greater during a depressive or manic episode. Previously neuroinflammation PET studies have been performed in schizophrenia and unipolar major depressive disorder (MDD). Our finding corresponds with the result in the study by Doorduin et al.30, albeit that the effect size in the right hippocampus in BD is smaller than that was found in schizophrenia-related psychotic patients, possibly related to the more extensive symptomatology of these patients. Another study on TSPO binding in patients with mild or moderate MDD using the radiophar- maceutical [11C]-PBR28 did not demonstrate a difference between patients and controls48. Although the hippocampus was not a specific region of interest in their study and subjects with a high sensitive C-reactive protein level of more than 5mg/l were excluded, it could be argued that perhaps neuroinflammation plays a more im- portant role in BD than in MDD. The present study has several inevitable limitations. Increased [11C]-(R)-PK11195 binding to the TSPO receptor in the brain is traditionally related to microglia ac- tivation9. It is important to note that the TSPO receptor can also be expressed in astrocytes, potentially influencing the [11C]-(R)-PK11195 binding potential signal49. However, because both cells are known to contribute to neuroinflammation50, it can be argued that regardless of activated microglia cells or astrocytes being responsible for the increased TSPO expression, the increased [11C]-(R)-PK11195 binding most likely represents a neuroinflammatory process either way. The naturalistic design of the study does not take the possible confounding effect of concomitant medication use into account. It is known that most mood stabilizing medications, including lithium, anticonvulsants and antipsychotics, have an effect on the immune system3,5–8. However, their effects are generally immunosuppressive in nature. It can be argued that in the present study most medications would actually have diminished the effect of the observed neuroinflammation, so the amount of microglia activation in medication-free euthymic BD-patients could be even larger, compared to controls.
134 Conclusion In conclusion, this study demonstrates the presence of focal neuroinflammation in the right hippocampus of BD-I patients, being a point of departure for unraveling the role of in vivo immune activation in the pathophysiology of BD.
Acknowledgements We thank Juliëtte Kalkman for accompanying the patients; Johan Wiegers, Aafke Zeilstra, Remko Koning, Eelco Severs en Paul van Snick for their assistance with the acquisition of PET scans; Anita Sibeijn-Kuiper, Judith Streurman and Remco Renken for their assistance with the acquisition of MRI images; Sarah Hamel-Brown for giving suggestions regarding grammar, flow, aesthetics and language of the manuscript.
7 Neuroinflammation in bipolar disorder
135 References
1. Schulze TG. Genetic research into bipolar disorder: the need for a research framework that integrates sophisticated molecular biology and clinically informed phenotype characterization. Psychiatr Clin North Am. 2010 Mar;33(1):67–82. 2. Maes M, Smith R, Scharpe S. The monocyte-T-lymphocyte hypothesis of major depression. Psychoneuroendocrinology. 1995 Jan;20(2):111–6. 3. Padmos RC, Hillegers MHJ, Knijff EM, Vonk R, Bouvy A, Staal FJT, et al. A discriminating messenger RNA signature for bipolar disorder formed by an aberrant expression of inflammatory genes in monocytes. Arch Gen Psychiatry. 2008 Apr;65(4):395–407. 4. Myint AM, Kim Y-K, Verkerk R, Park SH, Scharpé S, Steinbusch HWM, et al. Tryptophan breakdown pathway in bipolar mania. J Affect Disord. 2007 Sep;102(1-3):65–72. 5. Drzyzga L, Obuchowicz E, Marcinowska A, Herman ZS. Cytokines in schizophrenia and the effects of antipsychotic drugs. Brain Behav Immun. 2006 Nov;20(6):532–45. 6. Pollmächer T, Haack M, Schuld A. Effects of antipsychotic drugs on cytokine networks. J Psychiatr …. 2000;34(6):369–82. 7. Rybakowski JK. Antiviral and immunomodulatory effect of lithium. Pharmacopsychiatry. 2000 Sep;33(5):159–64. 8. Haarman BCM, Riemersma-Van der Lek RF, Burger H, Netkova M, Drexhage RC, Bootsman F, et al. Relationship between clinical features and inflammation-related monocyte gene expression in bipolar disorder - towards a better understanding of psychoimmunological interactions. Bipolar Disord. 2014 Mar 29;16(2):137–50. 9. Beumer W, Gibney SM, Drexhage RC, Pont-Lezica L, Doorduin J, Klein HC, et al. The immune theory of psychiatric diseases: a key role for activated microglia and circulating monocytes. J Leukoc Biol. 2012 Aug 8;92(September):1–17. 10. Benes FM, Matzilevich D, Burke RE, Walsh J. The expression of proapoptosis genes is increased in bipolar disorder, but not in schizophrenia. Mol Psychiatry. 2006 Mar;11(3):241–51. 11. Che Y, Wang J-F, Shao L, Young T. Oxidative damage to RNA but not DNA in the hippocampus of patients with major mental illness. J Psychiatry Neurosci. 2010 Sep;35(5):296–302. 12. Tian SY, Wang J-F, Bezchlibnyk YB, Young LT. Immunoreactivity of 43 kDa growth-associated protein is decreased in post mortem hippocampus of bipolar disorder and schizophrenia. Neurosci Lett. 2007 Jan 10;411(2):123–7. 13. Kato N, Nemoto K, Arino H, Fujikawa K. Influence of peripheral inflammation on growth- associated phosphoprotein (GAP-43) expression in dorsal root ganglia and on nerve recovery after crush injury. Neurosci Res. 2003 Mar;45(3):297–303. 14. Brown ES. Effects of glucocorticoids on mood, memory, and the hippocampus. Treatment and preventive therapy. Ann N Y Acad Sci. 2009 Oct;1179:41–55. 15. Roumier A, Béchade C, Poncer J-C, Smalla K-H, Tomasello E, Vivier E, et al. Impaired synaptic function in the microglial KARAP/DAP12-deficient mouse. J Neurosci. 2004 Dec 15;24(50):11421–8. 16. Roumier A, Pascual O, Béchade C, Wakselman S, Poncer J-C, Réal E, et al. Prenatal activation of microglia induces delayed impairment of glutamatergic synaptic function. PLoS One. 2008 Jan;3(7):e2595. 17. Costello D a, Lyons A, Denieffe S, Browne TC, Cox FF, Lynch M a. Long term potentiation is impaired in membrane glycoprotein CD200-deficient mice: a role for Toll-like receptor activation. J Biol Chem. 2011 Oct 7;286(40):34722–32. 18. Kreisel T, Frank MG, Licht T, Reshef R, Ben-Menachem-Zidon O, Baratta M V, et al. Dynamic microglial alterations underlie stress-induced depressive-like behavior and suppressed neurogenesis. Mol Psychiatry. Nature Publishing Group; 2013 Dec 17;(October):1–11.
136 19. Macchi F, Homberg JR, Calabrese F, Zecchillo C, Racagni G, Riva M a, et al. Altered inflammatory responsiveness in serotonin transporter mutant rats. J Neuroinflammation. Journal of Neuroinflammation; 2013 Jan;10(1):116. 20. Doorduin J, de Vries EFJ, Dierckx RA, Klein HC. PET imaging of the peripheral benzodiazepine receptor: monitoring disease progression and therapy response in neurodegenerative disorders. Curr Pharm Des. 2008 Jan;14(31):3297–315. 21. Choi J, Ifuku M, Noda M, Guilarte TR. Translocator protein (18 kDa)/peripheral benzodiazepine receptor specific ligands induce microglia functions consistent with an activated state. Glia. 2011 Feb;59(2):219–30. 22. Dobos N, de Vries EFJ, Kema IP, Patas K, Prins M, Nijholt IM, et al. The role of indoleamine 2,3-dioxygenase in a mouse model of neuroinflammation-induced depression. J Alzheimers Dis. 2012 Jan;28(4):905–15. 23. Dickens AM, Vainio S, Marjamäki P, Johansson J, Lehtiniemi P, Rokka J, et al. Detection of Microglial Activation in an Acute Model of Neuroinflammation Using PET and Radiotracers 11C-(R)-PK11195 and 18F-GE-180. J Nucl Med. 2014 Mar 10;55(3):466–72. 24. Wiley C a, Lopresti BJ, Becker JT, Boada F, Lopez OL, Mellors J, et al. Positron emission tomography imaging of peripheral benzodiazepine receptor binding in human immunodeficiency virus-infected subjects with and without cognitive impairment. J Neurovirol. 2006 Aug;12(4):262– 71. 25. Grover VPB, Pavese N, Koh S-B, Wylezinska M, Saxby BK, Gerhard a, et al. Cerebral microglial activation in patients with hepatitis C: in vivo evidence of neuroinflammation. J Viral Hepat. 2012 Feb;19(2):e89–96. 26. Garvey LJ, Pavese N, Politis M, Ramlackhansingh A, Brooks DJ, Taylor-Robinson SD, et al. Increased microglia activation in neurologically asymptomatic HIV-infected patients receiving effective ART. AIDS. 2014 Jan 2;28(1):67–72. 27. Banati RB. Visualising microglial activation in vivo. Glia. 2002 Nov;40(2):206–17. 28. van Berckel BN, Bossong MG, Boellaard R, Kloet R, Schuitemaker A, Caspers E, et al. Microglia activation in recent-onset schizophrenia: a quantitative (R)-[11C]PK11195 positron emission tomography study. Biol Psychiatry. 2008 Nov 1;64(9):820–2. 29. Folkersma H, Boellaard R, Yaqub M, Kloet RW, Windhorst AD, Lammertsma A a, et al. Widespread and prolonged increase in (R)-(11)C-PK11195 binding after traumatic brain injury. J Nucl Med. 2011 Aug;52(8):1235–9. 7 30. Doorduin J, de Vries EFJ, Willemsen ATM, de Groot JC, Dierckx R a, Klein HC. Neuroinflammation in schizophrenia-related psychosis: a PET study. J Nucl Med. 2009 Nov;50(11):1801–7. Neuroinflammation in bipolar disorder 31. Kreisl WC, Jenko KJ, Hines CS, Lyoo CH, Corona W, Morse CL, et al. A genetic polymorphism for translocator protein 18 kDa affects both in vitro and in vivo radioligand binding in human brain to this putative biomarker of neuroinflammation. J Cereb Blood Flow Metab. Nature Publishing Group; 2013 Jan;33(1):53–8. 32. Sheehan D V, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry. 1998 Jan;59 Suppl 2:22–33;quiz 34–57. 33. Leverich GS, Nolen WA, Rush AJ, McElroy SL, Keck PE, Denicoff KD, et al. The Stanley Foundation Bipolar Treatment Outcome Network. I. Longitudinal methodology. J Affect Disord. 2001 Dec;67(1-3):33–44. 34. Young RC, Biggs JT, Ziegler VE, Meyer D a. A rating scale for mania: reliability, validity and sensitivity. Br J Psychiatry. 1978 Nov 1;133(5):429–35. 35. Rush AJ, Weissenburger J, Giles E, Schlesser MA, Fulton CL. The Inventory Preliminary for Depressive Findings Symptomatology ( IDS ): Depression. 1986;65–87.
137 36. Svarer C, Madsen K, Hasselbalch SG, Pinborg LH, Haugbøl S, Frøkjaer VG, et al. MR-based automatic delineation of volumes of interest in human brain PET images using probability maps. Neuroimage. 2005 Feb 15;24(4):969–79. 37. Hochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990 Jul;9(7):811–8. 38. Aronica E, Crino PB. Inflammation in epilepsy: clinical observations. Epilepsia. 2011 May;52 Suppl 3:26–32. 39. Marazziti D, Dell’Osso B, Baroni S, Masala I, Nasso E Di, Giannaccini G, et al. Changes in peripheral benzodiazepine receptors in patients with bipolar disorder. Neuropsychiatr Dis Treat. 2005 Jun;1(2):151–4. 40. Savitz J, Frank MB, Victor T, Bebak M, Marino JH, Bellgowan PSF, et al. Inflammation and neurological disease-related genes are differentially expressed in depressed patients with mood disorders and correlate with morphometric and functional imaging abnormalities. Brain Behav Immun. 2013 Jul;31:161–71. 41. Paolicelli RC, Bolasco G, Pagani F, Maggi L, Scianni M, Panzanelli P, et al. Synaptic pruning by microglia is necessary for normal brain development. Science. 2011 Sep 9;333(6048):1456–8. 42. Schafer DP, Lehrman EK, Kautzman AG, Koyama R, Mardinly AR, Yamasaki R, et al. Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron. Elsevier Inc.; 2012 May 24;74(4):691–705. 43. Sierra A, Encinas JM, Deudero JJP, Chancey JH, Enikolopov G, Overstreet-Wadiche LS, et al. Microglia shape adult hippocampal neurogenesis through apoptosis-coupled phagocytosis. Cell Stem Cell. Elsevier Inc.; 2010 Oct 8;7(4):483–95. 44. Deckersbach T, Dougherty DD, Savage C, McMurrich S, Fischman AJ, Nierenberg A, et al. Impaired recruitment of the dorsolateral prefrontal cortex and hippocampus during encoding in bipolar disorder. Biol Psychiatry. 2006 Jan 15;59(2):138–46. 45. Gonul AS, Coburn K, Kula M. Cerebral blood flow, metabolic, receptor, and transporter changes in bipolar disorder: the role of PET and SPECT studies. Int Rev Psychiatry. 2009 Jan;21(4):323–35. 46. Deicken RF, Pegues MP, Anzalone S, Feiwell R, Soher B. Lower concentration of hippocampal N-acetylaspartate in familial bipolar I disorder. Am J Psychiatry. 2003 May;160(5):873–82. 47. Vargas C, López-Jaramillo C, Vieta E. A systematic literature review of resting state network-- functional MRI in bipolar disorder. J Affect Disord. Elsevier; 2013 Sep 25;150(3):727–35. 48. Hannestad J, DellaGioia N, Gallezot J-D, Lim K, Nabulsi N, Esterlis I, et al. The neuroinflammation marker translocator protein is not elevated in individuals with mild-to-moderate depression: a [11C]PBR28 PET study. Brain Behav Immun. Elsevier Inc.; 2013 Oct;33:131–8. 49. Lavisse S, Guillermier M, Hérard A-S, Petit F, Delahaye M, Van Camp N, et al. Reactive astrocytes overexpress TSPO and are detected by TSPO positron emission tomography imaging. J Neurosci. 2012 Aug 8;32(32):10809–18. 50. Hostenbach S, Cambron M, D’haeseleer M, Kooijman R, De Keyser J. Astrocyte loss and astrogliosis in neuroinflammatory disorders. Neurosci Lett. Elsevier Ireland Ltd; 2013 Oct 12;10–2.
138 7 Neuroinflammation in bipolar disorder
139
CHAPTER 8 Volume, metabolites and neuroinflammation of the hippocampus in bipolar disorder A combined magnetic resonance imaging and positron emission tomography study
Bartholomeus C.M. Haarman, Huibert Burger, Janine Doorduin, Remco J. Renken, Anita J. Sibeijn-Kuiper,Jan-Bernard C. Marsman, Erik F.J. de Vries, Jan Cees de Groot, Hemmo A. Drexhage, Richard Mendes, Willem A. Nolen, Rixt F. Riemersma – Van der Lek
Brain, Behavior, and Immunity. 2015 Sep 5;60(1):1–5. Abstract
Background The hippocampus is one of the brain regions that is involved in several pathophysi- ological theories about bipolar disorder (BD), such as the neuroinflammation theory and the corticolimbic metabolic dysregulation theory. We compared hippocampal volume and hippocampal metabolites in bipolar I disorder (BD-I) patients versus healthy controls (HC) with magnetic resonance imaging (MRI) and spectroscopy (MRS). We post-hoc investigated whether hippocampal volume and hippocampal metabolites were associated with microglial activation and explored if potential illness modifying factors affected these hippocampal measurements and whether these were associated with experienced mood and functioning.
Material and Methods Twenty-two BD-I patients and twenty-four HCs were included in the analyses. All subjects underwent psychiatric interviews as well as an MRI scan, including a T1 scan and PRESS magnetic resonance spectroscopy (MRS). Volumetric analysis was performed with Freesurfer. MRS quantification was performed with LCModel. A subgroup of 14 patients and 11 HCs also underwent a successful [11C]-(R)- PK11195 neuroinflammation positron emission tomography scan.
Results In contrast to our hypothesis, hippocampal volumes were not decreased in patients compared to HC after correcting for individual whole-brain volume variations. We demonstrated a decreased N-acetylaspartate (NAA) + N-acetyl-aspartyl-glutamate (NAAG) concentration in the left hippocampus. In the explorative analyses in the left hippocampus we identified positive associations between microglial activation and the NAA+NAAG concentration, between alcohol use and NAA+NAAG concentra- tion, between microglial activation and the depression score and a negative relation between Cr+PCr concentration and experienced occupational disability. Duration of illness associated positively with volume bilaterally.
Conclusion Compared to HCs, the decreased NAA+NAAG concentration in the left hippocampus of BD-I patients suggests a decreased neuronal integrity in this region. In addition we found a positive relation between microglial activation and neuronal integrity in vivo, corresponding to a differentiated microglial function where some microglia induce apoptosis while others stimulate neurogenesis.
142 Introduction
Bipolar disorder (BD) is a severe mental illness that is characterized by episodic pathologic disturbances in mood: (hypo)manic episodes and depressive episodes, which alternate with euthymic periods, i.e. with normal mood1. Besides the mood symptoms, many patients with BD also show cognitive dysfunctions which may persist during euthymic periods2,3. The lifetime prevalence of BD is about 2%, women being affected as frequently as men4,5. Across the world, BD is sixth among all health conditions in terms of causing disability6 with poor clinical and functional outcome7, increased risk for suicidality 8 and significant societal costs. It has been calculated Volume, metabolites and neuroinflammation of the in hippocampus bipolar disorder that in the United States the average cost per case ranged from $11,720 to $624,785, based on the severity of the illness9. In the European countries societal costs for managing BD are considered to be high as well10–12. Although the pathophysiology of BD is complex and its neurobiology remains largely unknown13, several pathophysiological mechanisms have been proposed to ex- plain at least part of the illness. Among these are microglial activation resulting in neuro-inflammation14, corticolimbic metabolic alterations15, and mitochondrial dys- function16. The neuroinflammation theory proposes an aberrant state of microglia in the brain (microglial activation) and of other immune competent cells to be the driving force behind the illness17–19. This theory is supported by findings of higher frequencies of comorbid autoimmune diseases, aberrant cytokine concentrations (e.g. tumor necrosis factor alpha, interleukin 4) and elevated inflammation-related gene expression in circulating monocytes and T cell activation20–23. The corticolimbic theory is based on many studies demonstrating an overall hyperactivation of limbic brain regions in BD patients relative to controls, along with an overall hypoactivation of prefrontal regions24. Supporting the hypothesis that mitochondrial dysfunction plays an important role in BD pathophysiology, several magnetic resonance spectros- copy (MRS) studies demonstrated cellular energy metabolism disturbances25–27 and structural anomalies in mitochondria were observed in cells of patients with BD28. The hippocampus, part of the limbic system, is one of the brain regions that is in- volved in all these pathophysiological theories29. The limbic system is a combination of in origin different brain structures that are involved in visceral behavioral patterns 8 (related to survival: eating, drinking, sexual activity), emotions, and memory. Some structures, such as the hippocampus and the amygdala, are phylogenetically rather old structures (hence the other name paleomammalian brain), while other parts such as the cingulate gyrus and the anterior cingulate cortex are more recently developed structures30. More specifically, the hippocampus has been implicated in the inhibition of stress responses31 and in the regulation of affective states and emotional behavior32, leading to contextually appropriate emotional responses29. Also, the hippocampus is known to play an important role in encoding new information33.
143 Multiple rodent studies have demonstrated a relationship between a disturbed microglia function in the hippocampus and other pathophysiological changes which are thought to relate to neuropsychiatric disorders17,34–36. One study demonstrated an increased cytokine response to lipopolysaccharide challenge in the hippocampus of SERT mutant rats37. Another rodent study showed stress-induction to induce dynamic microglia changes in the hippocampus alone, which was associated with depressive-like behavior38. In a recently published study using [11C]-(R)-PK11195 positron emission tomography, we demonstrated focal microglial activation in the right hippocampus and non-significant, trend level microglial activation in the left hippocampus of BD patients compared to healthy controls (HCs)39. This radiophar- maceutical binds to the translocator protein (TSPO), a receptor that is upregulated in the mitochondria of activated microglia cells40. [11C]-(R)-PK11195 has been used successfully in models of central inflammation, such as following an injection of the endotoxin lipopolysaccharide in animal models41–43 and investigating infections of the central nervous system in humans44–46. In various psychiatric and neurodegenerative disorders [11C]-(R)-PK11195 PET has proven to be a useful tool for imaging neuroin- flammation47–50. Using another approach, N-acetylaspartate (NAA), a metabolite of which the abso- lute concentration can be measured using 1H-MRS, has repeatedly (but not always) been found to be decreased in the hippocampus of bipolar patients51. NAA is the second most abundant amino acid in the central nervous system. It is formed in the mitochondria from acetyl co-enzyme A and aspartate and a decreased NAA is thought to represent loss of neuronal integrity. NAA is usually measured together with N-acetyl-aspartyl-glutamate (NAAG) by MRS52. Other metabolites that are reliably measured by MRS in the hippocampus are (phospho)creatine and choline compounds. Phosphocreatine (PCr) functions as an important energy buffer system with creatine (Cr) in the muscles and the brain, transporting energy generated in the mitochondria to the cytoplasm to maintain a constant ATP concentration. Gradual loss of creatine in conjunction with other major metabolites indicates tissue death or major cell death resulting from disease53. Interestingly, gene expression of creatine kinase, the enzyme that facilitates the Cr-PCr conversion, was found decreased in the hippocampus and prefrontal cortex of BD patients in a post mortem analysis54. The GPC+PCh peak of the 1H-MRS detects mainly phosphocholine (PCh) and glycero- phosphocholine (GPC), products of synthesis and breakdown of the cell membrane55. Regarding structural brain alterations a recent review summarized the scientific evidence for a small decrease in hippocampal volume that seemed to be counter acted by lithium treatment56,57. This decrease should be looked upon in the light that the whole-brain volume of BD patients is known to be smaller than that of HCs58. The relation between the various functional pathophysiological mechanisms and structural alterations of the hippocampus remains, however, unclear56.
144 In the current study we compared hippocampal volume and metabolites in bipolar I disorder (BD-I) patients with HCs using magnetic resonance imaging (MRI) and spectroscopy (MRS). We a priori hypothesized hippocampal volume and the NAA metabolite to be decreased in BD patients, compared to HCs. Subsequently, within the BD-I and HC group, we post-hoc investigated whether hippocampal volume and metabolites were associated with microglial activation. Furthermore, we explored if potential illness modifying factors such as duration of illness, medication use, body mass index (BMI), exercise, smoking, number of caffeine consumptions and alcohol use did affect these hippocampal measurements within the BD-I group and whether these hippocampal measurements were associated with Volume, metabolites and neuroinflammation of the in hippocampus bipolar disorder experienced mood and functioning.
8
145 Material and Methods
Participants For the present cross-sectional case-control study we included 22 BD-I patients and 24 HCs demographically similar in age and gender that participated in the MOODIN- FLAME study (http://www.moodinflame.eu). All subjects underwent a MRI-scan and a subgroup of the subjects also received a PET-scan (15 patients, 12 controls)59. In the MOODINFLAME study adult male and female subjects were included who were free of inflammation related symptoms including fever and current or recent infec- tious or inflammatory disease, uncontrolled systemic disease, uncontrolled metabolic disease or other significant uncontrolled somatic disorders known to affect mood. They did not use somatic medication known to affect mood or the immune system, such as corticosteroids, non-steroid anti-inflammatory drugs and statins. Female candidates who were pregnant or recently gave birth were excluded. Patients and controls did not have a contraindication for MRI scanning. Patients and controls who participated in the PET-scan were free of benzodiazepines at least in the last week prior to the scan. They were also free of anticoagulant use or presence of coagulation disease and did not participate in a prior research study involving radiation less than a year ago. Patients were allowed to continue their regular psychopharmacological treatment. They were neither in a depressed nor (hypo-)manic episode at the time of scanning as indicated by an Inventory of Depressive Symptoms - Clinician Version (IDS-C30) score <22 and a Young Mania Rating Scale (YMRS) score <12, respectively. Patients with any other current primary major psychiatric diagnosis were excluded including: schizophrenia, schizoaffective disorder, anxiety disorder and substance use disorders. HCs did not have any current or lifetime psychiatric diagnosis.
Ethical Considerations The Medical Ethical Review Committee of the University Medical Center Groningen approved the protocol, which was performed in accordance with the Helsinki Decla- ration of 201360. Written informed consent was obtained from all participants.
Assessments All subjects underwent a Mini-International Neuropsychiatric Interview (MINI) to confirm the diagnosis in the patient group and the absence of psychiatric disorders in the HCs61. Clinical features were extracted from the interviews held according to the general MOODINFLAME protocol. This protocol included the Patient Question- naire of the former Stanley Foundation Bipolar Network, the YMRS, the IDS-C30 and a somatic illness questionnaire. The Patient Questionnaire includes separate clinician and patient chapters covering a
146 spectrum of clinical features including vocational, educational and economic status, onset and course of illness, family history, past treatment, cycling and seasonal pat- terns, medical problems, medications, ability to function and symptomatic status, precipitants of illness (e.g. substance use), treatment adherence and insight into the illness62. In the event of a mismatch of results from the MINI in relation to the Patient Questionnaire, diagnoses were checked with the treating physician. The YMRS is an 63 eleven-item, multiple-choice questionnaire to assess manic symptoms . The IDS-C30 is a thirty-item, multiple-choice questionnaire to assess depressive symptoms of 64 all symptom domains of depression . The YMRS and IDS-C30 were assessed shortly before the scans and used in the relevant analyses. The somatic illness questionnaire Volume, metabolites and neuroinflammation of the in hippocampus bipolar disorder is a MOODINFLAME specific checklist exploring all the organ systems for current and lifetime medical symptoms.
Magnetic Resonance Imaging Using a 3T MRI scanner and an eight-channel head coil (3T Intera, Philips, Best, The Netherlands), we acquired anatomical axial T1 weighed MRI images, as well as two single volume point-resolved spectroscopy sequence (PRESS) MRS scans. The T1 images were used in the volumetric analysis and were used to determine the place- ment of the MRS volume of interest (VOI). These images were also used in the tissue type analysis of the MRS VOI and provided the anatomical reference for the normal- ization of the PET scans to standard brain morphology of all subjects. In the subgroup that underwent both a PET scan and an MRI, these were made no more than 1 week apart and even on the same day in 11 of the 27 subjects (41%).
Volumetric analysis The high resolution 3D T1-weighted images were acquired by 3D ultrafast spoiled gradient echo sequence (SPGR) with the following acquisition parameters: repetition time (TR) = 9.758 ms, echo time (TE) = 4.59 ms, flip angle = 8°, slice thickness = 1.2 mm, single-shot, field of view (FOV) = 220 · 174 mm, matrix = 256 · 256, voxel size=0.859 · 0.859 · 1.200 mm. A total of 130 axial images were collected for each subject, encompassing the whole-brain. PAR-REC files collected from scanning were transferred into NIFTI format data using MRIcroX software (version 1.2, 8 http://www.mccauslandcenter.sc.edu/CRNL/tools/mricro). Cortical reconstruction and volumetric segmentation was performed with the Free- surfer image analysis suite (version 5.3.0, http://surfer.nmr.mgh.harvard.edu) and was used to obtain the volume of the left and right hippocampus (Figure 1), as well as the whole-brain volume (not including the ventricles). The technical details of these procedures are described in prior publications65,66. Freesurfer is known to be a robust tool for automated segmentation of the hippocampus67. The Freesurfer volumetric analyses of the participants all passed a visual quality control check.
147 FIGURE 1 Freesurfer hippocampus volumes
Three-dimensional model of the segmented left and right hippocampus (yellow) of a female patient using Freesurfer and displayed in 3D Slicer (www.slicer.org).
Magnetic resonance spectroscopy All 1H MRS data were acquired using a point-resolved spectroscopy sequence (PRESS) with echo time (TE) = 144 ms, repetition time (TR) = 2000 ms, 128 acquisitions, sample frequency = 2 kHz, 1024 complex data points. Water unsuppressed spectra were also acquired for absolute quantification of metabolites in units of mmol/kg wet weight. Two single volume 15 · 15 · 15 mm (3375 mm3) VOIs were placed on the head of the left and right hippocampus (Figure 2). Due to technical problems 1H MRS scans failed in two healthy controls. The absolute quantification of the spectral metabolites NAA+NAAG, glutamate, glutamine, myo-inositol, Cr+PCr, GPC+PCh, taurine, alanine, aspartate, gamma-amino-butyric acid and glucose, as well as lipid and macromol- ecule resonances, was performed using the Linear Combination Model (LC Model) software68, an operator-independent fitting routine (Figure 3). Only the results of the more reliable metabolites (NAA+NAAG, Cr+PCr and GPC+PCh) were used. Quantifi- cation results in which these metabolites had a Cramer-Rao Lower Bounds of more than 20% were excluded for analysis69, providing 87 reliable MRS results (left hippo- campus 22 patients, 21 controls; right hippocampus 22 patients, 22 controls). Tissue-type segmentation was performed on the T1-weighed MRI image using an in- house developed automation script. FMRIB’s Automated Segmentation Tool (FAST) part of FMRIB’s Software Library (FSL 5.0.7, http://www. fmrib.ox.ac.uk/fsl) was used for segmentation of the whole-brain into gray matter (GM), white matter (WM) and
148 FIGURE 2 Freesurfer hippocampus volumes Volume, metabolites and neuroinflammation of the in hippocampus bipolar disorder
T1-weighted sagittal, coronal and axial MRI locations for left hippocampal 1H-MRS single voxel acquisitions in a male patient.
FIGURE 3 Examples of MRS spectra