- Email: [email protected]
The network model of delirium
Medical Hypotheses 104 (2017) 80–85
Contents lists available at ScienceDirect
Medical Hypotheses journal homepage: www.elsevier.com/locate/mehy
The network model of delirium James W.S. Young ⇑ Department of Medicine, University of Toronto, Toronto, ON, Canada
a r t i c l e
i n f o
Article history: Received 5 November 2016 Accepted 25 May 2017
Keywords: Delirium Default-mode network Salience network Frontoparietal control network Dorsal attention network
a b s t r a c t The coordinated function of brain networks underlies consciousness, attention and reality testing, all of which are impaired in delirium. The default-mode network, salience network, frontoparietal control network and dorsal attention network are brain networks with integral roles in the maintenance and modulation of the aforementioned functions. Multiple lines of evidence point to their dysfunction in delirium. The convergence of neurotransmitter changes, neuroendocrine and inflammatory stressors on brain networks disrupts bottom-up and top-down attentional control. Neuroimaging and neuroanatomy correlates are potentially consistent with this hypothesis. Overall, this model appears to have significant utility in connecting the seemingly disparate precipitants of delirium while accounting for the clinical manifestations of the syndrome. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Delirium is a functional brain disorder, as there are no specific findings on conventional neuroimaging studies or pathological specimens which clearly correlate with the clinical syndrome. This has been a significant reason for our poor understanding of exactly what delirium is. Functional brain imaging studies with functional magnetic resonance (fMRI) and positron emission tomography (PET) have begun to provide insights into brain dysfunction which arises from problems in the coordinated function of the neural network rather than discrete pathological lesions. Therefore, a functional perspective has the potential to shed light on the nature of delirium. Sanders [1] and Rapazzini [2] have previously proposed the idea of changes in brain network connectivity being responsible for delirium. Functional imaging studies have reinforced the general idea that the cortex and subcortical gray matter are more usefully viewed as a collection of discrete modules, or nodes, that network together [3]. Various combinations of nodes are recruited to perform the diverse functions which the brain is capable of. Multiple functional brain networks have been identified which significantly influence attention, but several appear to be especially important: the default-mode network (DMN) [4], the dorsal attention network (DAN) [5], the salience network (SN) [6,7], and the frontoparietal control network (FPCN) [8] (Table 1). Because inattention is the
⇑ Address: Brampton Civic Hospital, William Osler Health System, Division of Geriatric Medicine, 2100 Bovaird Drive East, Brampton, ON L6R 3J7, Canada. E-mail address: [email protected] http://dx.doi.org/10.1016/j.mehy.2017.05.027 0306-9877/Ó 2017 Elsevier Ltd. All rights reserved.
core cognitive symptom of delirium, these brain networks are suspected to be dysfunctional in delirium.
Brain network function and inattention The function of individual brain networks is an evolving area, but general themes describing the function of these networks have emerged. The DMN is more active when ‘task-activated’ areas of the brain are quiet [4]. Its activity constitutes the baseline state of conscious awareness which includes broad, low-level awareness of multiple internal and external stimuli [11]. When no deliberate thinking is occurring it allows for some awareness of mostly irrelevant external sensory data (to permit detection of unexpected salient stimuli) while mostly allowing for an unstructured stream of internal thoughts to meander along. For more deliberate mental tasks such as imagination or planning the future, the DMN is required, but it is co-opted by the FPCN and directed more intentionally. Consciousness itself may arise largely from the coordinated function of the nodes of the DMN in particular, and thus any disruption of its function could lead to an altered state of consciousness. This is seen physiologically each night when we sleep [11,12], but also in altered states of consciousness due to anesthesia or brain injury [13]. Thus, the integrity of the DMN as a standalone network seems to be an important determinant of the level of consciousness and awareness of the environment. The dorsal attention network (DAN) is activated when externally focused attention is required. When attention is directed externally the DMN is suppressed while the DAN is activated. Conversely, the DAN is suppressed in the absence of external stimuli,
81
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85 Table 1 Brain networks involved in attention, their components, and symptoms of delirium which likely result from their dysfunction. Network
Components
Normal Functions
Network dysfunction leads to
Refs.
Default-Mode Network
1. Medial prefrontal cortex 2. Posterior cingulate cortex 3. Inferior parietal lobule 4. Lateral temporal cortex 5. Hippocampal formation
Active when subjects are at rest, such as in a quiet room with eyes closed, not engaged in any specific task (the ‘default-mode’). Increased activity with tasks of mental simulation and self-referential thought and ‘mind-wandering’. Suppressed when attention focused on specific stimuli or tasks in the external environment.
– Altered level consciousness – Disorganized thinking
of
[4,9]
Frontoparietal Control Network
1. Dorsolateral prefrontal cortex 2. Anterior inferior parietal lobule
Deliberate (‘top down’) control of attention. Works with elements of the salience network.
– Inattention (reduced sustained attention) – Perceptual disturbances
[6,9]
Salience Network
1. Anterior insula 2. Anterior cingulate cortex 3. Amygdala 4. Ventral striatum 5. Substantia nigra/ ventral tegmental area 6. Thalamus
Receives input about reward (ventral striatum/VTA), emotion (amygdala), pain and other senses (thalamus), autonomic activity (AI), cytokine and cortisol levels (ACC). This allows the SN to monitor and prioritize homeostatically important information (grabs attention ‘bottom up’). The SN may act as a switch between DMN and DAN modes of attention. The SN along with input from the FPCN can activate the DAN to recruit focused attention towards homeostatically significant stimuli in the external environment.
– Inattention (distractibility) – Perceptual disturbances – Agitation
[3,6,9]
Dorsal Attention Network
1. Intraparietal sulcus 2. Frontal eye fields 3. Superior parietal lobule 4. Middle temporal motion cortex
Activated when attention is focused on a specific external task or stimulus. Activity is anticorrelated with the default mode network.
– Inattention
[9,10]
while the DMN remains active. This reciprocal activity, termed anticorrelation, is essential for efficient deployment of attention towards different tasks or spheres of interest, and avoids interference from irrelevant brain activity [14]. Christoff et al. view attention management in terms of constraints on attention [9]. Unconstrained attention allows DMN activity to proceed spontaneously (day-dreaming in a quiet room). External stimuli can constrain attention further (‘‘automatic constraints”, e.g. your eye is automatically drawn to the bright pink car that drives by) and this occurs by SN activation of the DAN and suppression of the DMN (this is also referred to as ‘bottomup’ attentional control). Conscious constraints can similarly focus attention (‘‘deliberate constraints”, e.g. ‘‘I need to stop thinking about that movie and focus on this boring lecture”) and this occurs by FPCN-mediated suppression of the DMN and reinforcement of the SN/DAN (this is also referred to as ‘top-down’ attentional control). The FPCN and SN, therefore, can work together (either cooperatively or antagonistically) to guide attention, bottom-up and top-down. Importantly, the SN appears to be a hub which mediates switching between different modes of attention [3,7,15,16]. The SN is in the unique position of gathering information from sensory systems regarding internal and external stimuli which may have motivational significance (e.g. reward, pain, fear, anxiety, arousal, stress, etc.). The SN is a sentinel which monitors these streams, developing context for incoming sensory data, and then prioritizing the most homeostatically relevant and allocating finite attentional resources accordingly. The SN exists at a convergence point for multiple stress signals [16]. This is especially interesting as it relates to delirium, because the brain network which receives such a wide variety of stress signals also has a significant role in switching between unfocused attention (uncoupled DMN activity), automatically focused attention (SN/DAN interactions) and
deliberately-focused attention (SN/FPCN/DAN interactions). Therefore, this leads logically to the possibility that perturbation of SN function is a critical aspect of inattention in delirium (Fig. 1). Dysfunction of the SN has also been postulated to contribute to other neuropsychiatric problems [16,17]. It is well recognized that stress signals from illness can fundamentally influence brain function. ‘Sickness behaviours’ are behaviour changes which classically occur in an organism with acute infection (e.g. fatigue, poor concentration, sleep disturbances, etc.), and provide clear evidence for this [18,19]. They are mediated by a variety of neuroendocrine changes, especially cytokine release and HPA axis activation. Sickness behaviours are typically viewed as adaptive, promoting rest, energy conservation, healing, etc. This is probably true in the context of a healthy organism with an acute and reversible illness (e.g. acute infection). However, they may also provide the substrate for neurobehavioural changes which, when poorly regulated or taken to extremes, resemble many of the symptoms of delirium, which are decidedly maladaptive. Interestingly, inflammatory cytokines have been shown to act on brain areas which are nodes in the salience network [20,21]. Delirium is frequently precipitated by inflammation or drugs. Infection resulting in systemic cytokine changes may result in behaviour changes via the SN as noted above. Dementia is a known significant risk for delirium, likely due to a propensity towards exaggerated neuroinflammation [22,23] and baseline changes in brain networks [24–26]. Sanders suggest that some drugs may influence inhibitory tone within brain networks, which may reduce their connectivity and function [1]. Opioids, benzodiazepines and anticholinergics, among others, have the potential to alter thalamic and striatal function [27], which may disrupt SN function and DMN/DAN switching/anticorrelation. One thing which appears to be common among many drugs with a high potential to precipitate delirium is their ability to significantly impact on one or another
82
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
Fig. 1. A basic network model of delirium. Multiple stressors converge on the salience network (SN), potentially disturbing its function in regulating bottom-up attention. Problems with the frontoparietal control network (FPCN) result in impaired top-down control of attention. This impairs switching between the dorsal attention network (externally directed attention) and the default-mode network (internally directed attention). Problems switching between brain networks/attentional modes, as well as problems arising from individual network dysfunction, result in the cardinal manifestations of delirium (right side of figure). ACC, anterior cingulate cortex; ACh, acetylcholine; AI, anterior insula; DA, dopamine; DLPFC, dorsolateral prefrontal cortex; FEF, frontal eye fields; HPA, hypothalamic-pituitaryadrenal; IPS, intraparietal sulcus; LOC, level of consciousness; MPFC, medial prefrontal cortex; PCC, posterior cingulate cortex; PPC, posterior parietal cortex; VTA, ventral tegmental area.
component of the SN. As noted, the components of the SN are also in receipt of other internal signals which are likely to occur during times of infection, inflammation, and stress. Therefore, in the acutely ill hospitalized patient receiving psychoactive medications, the SN network appears to be bombarded by multiple insults. Anxiety, stress and fear may also be contributors to delirium. These may be present for numerous reasons, as might be expected in a person sick enough to be hospitalized, perhaps even critically ill in the intensive-care unit (ICU) or requiring major surgery. Pain, catheter insertions and the use of physical restraints, among others, may all contribute to stress [28]. Delirium symptoms themselves, such as confusion, disorientation, hallucinations or delusions may compound this effect. Psychological and immune stressors have been shown to act synergistically to promote inflammation and sickness behaviour [29]. Sleep deprivation has adverse effects on the efficient regulation of attentional networks [30]. Activation of the HPA axis, resulting in glucocorticoid and catecholamine secretion, appears to bias the brain away from ‘topdown’ network control in favour of more automatic responses [31]. This may be due to differential adrenergic receptor expression patterns in the various brain structures of the attentional networks [32]. Dopamine and acetylcholine are key neurotransmitters involved in the control of the SN and FPCN, and, coincidentally, the neurotransmitters most widely held to be dysregulated in delirium. The SN integrates contextual information to guide bottom-up allocation of attention. For example, the VTA and related dopaminergic reward areas in the midbrain can flag incoming sensory information as rewarding, or the amygdala can impart emotional context such as fear. Conscious efforts (via the FPCN) direct attention in a top-down manner. Both bottom-up and top-
down mechanisms rely on cholinergic and dopaminergic neurotransmission [33–35]. Cholinergic tone appears to be particularly important in top-down attentional control [35]. Constant updating of attentional goals and priorities appears to function via feedback loops monitoring attentional performance and reward via these two neurotransmitters. An important link between the subcortical components of the SN and the cortical components of the FPCN is the cholinergic nucleus basalis. Dopamine appears to control gating of information flowing into working memory [36]. Too little dopamine leaves working memory inaccessible for updating with relevant information. Too much may lead to inappropriately ascribing salience to irrelevant information and allowing its intrusion into working memory. In both cases, attention is affected. Thus, dopaminergic tone, as it influences attention, may function in a U-shaped curve [37]. Delirium can occur in the context of low dopamine (possibly reflecting the hypoactive end of the delirium spectrum, reduced sustained attention) or excess dopamine (hyperactive delirium, distractibility). Dopamine also appears to enhance coupling between the FPCN and DMN, and reduce coupling between the FPCN and DAN [38]. Overall, inattention in delirium may be due to a combination of problems with both bottom-up and top-down attentional control. What Christoff refers to as ‘automatic constraints’ still seem to partly function in delirium, that is, a novel stimulus can often grab the attention of the delirious person (bottom-up). In fact, this is often true to the point that it is pathological as delirious individuals are often quite distractible. In this sense, convergence of multiple stressors on the SN may tend to disrupt ‘bottom-up’ attentional control. However, dysfunction is also evident in the ‘deliberate constraints’, where there is difficulty intentionally focusing attention (e.g. when an individual with delirium struggles to recite the
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
days of the week backwards, a common bedside test of attention). This suggests that the top-down control via the FPCN is also disrupted in delirium. Perceptual abnormalities Higher level processing failure also appears to explain perceptual disturbances in delirium. Executive areas, especially the right dorsolateral prefrontal cortex (DLPFC) (part of the FPCN), appear to have a role in reality testing. Dysfunction in the right DLPFC [39] and abnormalities in elements of the SN [17] are associated with delusions and hallucinations. Gerrans views reality testing as hypothesis testing [40]. That is, sensory systems register data and generate a basic hypothesis which becomes an initial perception, and then higher systems provide context to help confirm the accuracy of the hypothesis. When incongruous information is registered by lower sensory systems, higher cognitive systems must provide additional analysis using background knowledge and contextual information to allow resolution of the inconsistency. Sometimes an anomaly must be brought forward to the level of conscious attention and thought through to be resolved (‘‘Is that really a man in the corner of my dark bedroom or is that just the pile of clothes on a chair?”) Gerrans proposes a two-hit model, whereby aberrant sensory information can trigger a misperception, but it only persists as a full-blown delusion or hallucination when this second step in analysis also fails. This highlights the theme of problems with top-down processing – if these brain areas are not functioning an delirious individual may not be able to voluntarily address anomalies, and in fact may not even recognize the need to do so. Gaudreau and Gagnon note that anticholinergic and GABAergic drugs, well-known to contribute to delirium, may reduce striatal inhibition of the thalamus, leading to excessive or aberrant release of sensory information to the cortex [27]. This may be the first step in triggering perceptual abnormalities. Not surprisingly, several of the sensitive elements in their model are components of the SN. In delirium, these anomalous events may be ascribed inappropriate salience, and are not subject to secondary scrutiny by executive areas, which are hypoactive, leading to delusions and/or hallucinations. A familiar example is dreaming, a highly ‘unconstrained’ state. Executive areas are conspicuously inactive during REM sleep when most dreaming occurs, while elements of the DMN remains quite active [9]. Dreams often have tangential narratives (perhaps akin to the disorganized thinking of delirium) and bizarre, anomalous content, which is typically accepted without question during the dream. This may reflect unconstrained DMN activity with no executive oversight. Delirium as a state of consciousness between full awareness and unconsciousness may have close analogy with REM-sleep, the stage of sleep most proximal to wakefulness. Delirium is described by those who recover from it as having a dreamlike quality with bizarre experiences. As in a dream, anomalous perceptions may occur entirely free of the usual FPCN/SN oversight function which might otherwise flag them as anomalous and consciously reject them, thereby allowing their perpetuation. Neuroimaging and neuroanatomy The single fMRI study of delirious individuals to date shows changes in anticorrelation between nodes of the DMN and FPCN [41]. Control subjects demonstrated anticorrelated activations of their posterior cingulate cortex (PCC) and DLPFC, as expected. During delirium, individuals lost this anticorrelation, and began to have a positive correlation between activity in these networks. Specifically, positive correlation was seen between the PCC (a DMN node) and bilateral DLPFC (FPCN nodes). Following resolution
83
of delirium these changes resolved. It may be that increasing coupling of the DMN and FPCN results in attention being directed to internal awareness via activity of the DMN, with reduced ability to engage the DAN, and therefore reduced awareness of the external environment. Perhaps DMN activity becomes intrusive and difficult to suppress during delirium. At a minimum, changes were seen in the regulation of key elements of the DMN and FPCN. As noted earlier, dopamine excess may contribute to some of these changes. Connectivity changes were also seen between six subcortical regions. Of note, connections were lost during delirium between areas which are key nodes of the SN (VTA, intralaminar thalamic nuclei), and an important connection between the SN and FPCN (the cholinergic nucleus basalis). Another small functional imaging study of 22 patients who underwent SPECT scanning during an episode of delirium revealed heterogeneous findings, but some evidence of frontal and parietal perfusion abnormalities, which is potentially consistent with localization to elements of the DMN and/or FPCN [42]. A review of multiple SPECT studies across delirium of various etiologies (postoperative, post-stroke, hepatic encephalopathy, etc.) showed great heterogeneity in terms of cerebral perfusion abnormalities [43]. Hypoperfusion was the usual abnormality, and resolution of delirium was accompanied by normalization of blood flow. The authors stated that frontal and parietal perfusion abnormalities were probably most common despite the heterogeneity. Further supporting the idea that brain network dysfunction is the basis for delirium is a review of delirium in stroke patients. Perez et al. reviewed studies that examined the righthemispheric correlates of non-agitated versus hyperactive delirium [44]. Overall, they found that, despite somewhat limited evidence, right hemispheric infarcts in multiple regions including the posterior parietal cortex, temporal, occipital, prefrontal cortices, and also subcortical lesions, have been implicated in both non-agitated confusion as well as hyperactive delirium. Nonagitated confusional states in patients with right hemispheric strokes were preferentially associated with lesions in attentional and perceptual association cortices (prefrontal, posterior parietal and interconnected subcortical regions) which would implicate elements of the FPCN and/or SN. Hyperactive delirium involved regions more associated with the limbic and paralimbic cortices, which would include elements of the DMN and/or SN. Thus, it is tempting to infer that agitation in delirium is more localizable to the limbic components of the SN and/or DMN, whereas inattention arises from problems in the SN and/or FPCN. Another interesting example is that of stroke in the dorsomedial thalamus. This area has been proposed to be essential in binding together the SN nodes into a circuit [6], and stroke here has been reported to cause delirium [45].
Summary In conclusion, delirium is proposed to be a state of dysfunctional brain networks. Changes in DMN connectivity lead to an altered level of consciousness. Problems switching between the DMN and DAN may reduce awareness of the environment. The convergence of multiple neuroendocrine and inflammatory stressors on the SN disrupts bottom-up attentional control leading to distractibility. Drugs altering cholinergic or dopaminergic pathways disrupt SN and/or FPCN function. Inflammation and the neuroendocrine stress response diverts homeostatic resources away from top-down control by the FPCN. Problems with top-down control impair deliberate control of attention. Loss of FPCN/SN function also impairs reality testing which facilitates delusions and hallucinations. Many of these predictions should be testable (Table 2).
84
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
Table 2 Brain networks predicted to be abnormal in delirium and conditions to examine with functional neuroimaging studies which may support the network model. Network
Issues in delirium
Functions to Probe
DMN
1. Altered LOC 2. Loss of anti-correlated activity with DAN
1. Connectivity among core DMN regions before/during/after delirium 2. Task-induced deactivation of DMN/ activation of the DAN by external stimuli during delirium 3. 3. Effect of antipsychotics on coupling/uncoupling of DMN/DAN/FPCN before/during/after delirium
SN
1. Inattention 2. Perceptual abnormalities
1. Connectivity amongst SN nodes before/during/after delirium 2. Activity of SN during tasks requiring switching from non-directed to externally directed attention during delirium 3. Association of SN changes with reported/observed delusions/ hallucinations 4. Association of serum/CSF cytokine levels with connectivity and/or switching functions of SN
FPCN
1. Inattention 2. Perceptual disturbances
1. Recruitment of FPCN in normal versus delirious individuals at rest and with attentional demands 2. Correlation of activity in DLPFC (especially RDLPFC) with delusions and hallucinations
One of the strengths of this model is that it attempts to provide a coherent pathway linking the precipitants of delirium to the final common clinical manifestations – something which has been lacking in the delirium literature to date. There are many nuances to this theory which warrant further exploration. The brain network paradigm helps to reconcile the diverse array of seemingly unrelated stressors which may precipitate delirium, and appears to be consistent with much of what is already known about delirium. Further fMRI and PET studies, despite potential logistical challenges, may offer greater insights into this theory of delirium. Better treatment of delirium will only come with a more thorough understanding of its neurobiological basis. Declaration of sources of funding None. Conflicts of interest The author declares no conflicts of interest. Acknowledgements Thanks to Drs. David Tang-Wai, Cheryl Grady, Mark Lachmann and Camilla Wong for their feedback on the manuscript. References [1] Sanders RD. Hypothesis for the pathophysiology of delirium: role of baseline brain network connectivity and changes in inhibitory tone. Med Hypotheses 2011;77:140–3. http://dx.doi.org/10.1016/j.mehy.2011.03.048. [2] Rapazzini P. Functional interrelationship of brain aging and delirium. Aging Clin Exp Res 2016;28:161–4. http://dx.doi.org/10.1007/s40520-015-0379-3. [3] Bressler SL, Menon V. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn Sci 2010;14:277–90. http://dx.doi.org/ 10.1016/j.tics.2010.04.004. [4] Buckner RL, Andrews-Hanna JR, Schacter DL. The Brain’s default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci 2008;1124:1–38. http://dx.doi.org/10.1196/annals.1440.011.
[5] Fox MD, Corbetta M, Snyder AZ, et al. Spontaneous neuronal activity distinguishes human dorsal and ventral attention systems. Proc Natl Acad Sci 2006;103:10046–51. http://dx.doi.org/10.1073/pnas.0604187103. [6] Seeley WW, Menon V, Schatzberg AF, et al. Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci 2007;27:2349–56. http://dx.doi.org/10.1523/JNEUROSCI.5587-06.2007. [7] Menon V, Uddin LQ. Saliency, switching, attention and control: a network model of insula function. Brain Struct Funct 2010;214:655–67. http://dx.doi. org/10.1007/s00429-010-0262-0. [8] Vincent JL, Kahn I, Snyder AZ, et al. Evidence for a frontoparietal control system revealed by intrinsic functional connectivity. J Neurophysiol 2008;100:3328–42. http://dx.doi.org/10.1152/jn.90355.2008. [9] Christoff K, Irving ZC, Fox KCR, et al. Mind-wandering as spontaneous thought: a dynamic framework. Nat Rev Neurosci 2016. http://dx.doi.org/10.1038/ nrn.2016.113. [10] Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven attention in the brain. Nat Rev Neurosci 2002;3:201–15. http://dx.doi.org/ 10.1038/nrn755. [11] Horovitz SG, Braun AR, Carr WS, et al. Decoupling of the Brain’s default mode network during deep sleep. Proc Natl Acad Sci U S A 2009;106:11376–81. http://dx.doi.org/10.1073/pnas.0901435106. [12] Samann PG, Wehrle R, Hoehn D, et al. Development of the Brain’s default mode network from wakefulness to slow wave sleep. Cereb Cortex 2011;21:2082–93. http://dx.doi.org/10.1093/cercor/bhq295. [13] Heine L, Soddu A, Gómez F, et al. Resting state networks and consciousness. Front Psychol 2012. http://dx.doi.org/10.3389/fpsyg.2012.00295. [14] Fox MD, Snyder AZ, Vincent JL, et al. The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A 2005;102:9673–8. http://dx.doi.org/10.1073/pnas.0504136102. [15] Sridharan D, Levitin DJ, Menon V. A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks. Proc Natl Acad Sci 2008;105:12569–74. http://dx.doi.org/10.1073/ pnas.0800005105. [16] Uddin LQ. Salience processing and insular cortical function and dysfunction. Nat Rev Neurosci 2014;16:55–61. http://dx.doi.org/10.1038/nrn3857. [17] Manoliu A, Riedl V, Zherdin A, et al. Aberrant dependence of default mode/central executive network interactions on anterior insular salience network activity in schizophrenia. Schizophr Bull 2014;40:428–37. http://dx. doi.org/10.1093/schbul/sbt037. [18] Dantzer R, O’Connor JC, Freund GG, et al. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci 2008;9:46–56. http://dx.doi.org/10.1038/nrn2297. [19] Miller AH. Mechanisms of cytokine-induced behavioral changes: psychoneuroimmunology at the translational interface. Brain Behav Immun 2009;23:149–58. http://dx.doi.org/10.1016/j.bbi.2008.08.006. [20] Capuron L, Pagnoni G, Demetrashvili M, et al. Anterior cingulate activation and error processing during interferon-alpha treatment. Biol Psychiatry 2005;58:190–6. http://dx.doi.org/10.1016/j.biopsych.2005.03.033. [21] Felger JC, Miller AH. Cytokine effects on the basal ganglia and dopamine function: the subcortical source of inflammatory malaise. Front Neuroendocrinol 2012;33:315–27. http://dx.doi.org/10.1016/j. yfrne.2012.09.003. [22] Cunningham C. Microglia and neurodegeneration: the role of systemic inflammation. Glia 2013;61:71–90. http://dx.doi.org/10.1002/glia.22350. [23] Perry VH, Holmes C. Microglial priming in neurodegenerative disease. Nat Rev Neurol 2014;10:217–24. http://dx.doi.org/10.1038/nrneurol.2014.38. [24] Dickerson BC, Sperling RA. Large-scale functional brain network abnormalities in Alzheimer’s disease: insights from functional neuroimaging. Behav Neurol 2009;21:63–75. http://dx.doi.org/10.3233/BEN-2009-0227. [25] Agosta F, Pievani M, Geroldi C, et al. Resting state fMRI in Alzheimer’s disease: beyond the default mode network. Neurobiol Aging 2012;33:1564–78. http:// dx.doi.org/10.1016/j.neurobiolaging.2011.06.007. [26] Seeley WW, Crawford RK, Zhou J, et al. Neurodegenerative diseases target large-scale human brain networks. Neuron 2009;62:42–52. http://dx.doi.org/ 10.1016/j.neuron.2009.03.024. [27] Gaudreau J-D, Gagnon P. Psychotogenic drugs and delirium pathogenesis: the central role of the thalamus. Med Hypotheses 2005;64:471–5. http://dx.doi. org/10.1016/j.mehy.2004.08.007. [28] Inouye SK, Charpentier PA. Precipitating factors for delirium in hospitalized elderly persons. Predictive model and interrelationship with baseline vulnerability. JAMA 1996;275:852–7. [29] Brydon L, Walker C, Wawrzyniak A, et al. Synergistic effects of psychological and immune stressors on inflammatory cytokine and sickness responses in humans. Brain Behav Immun 2009;23:217–24. http://dx.doi.org/10.1016/j. bbi.2008.09.007. [30] De Havas JA, Parimal S, Soon CS, Chee MWL. Sleep deprivation reduces default mode network connectivity and anti-correlation during rest and task performance. NeuroImage 2012;59:1745–51. http://dx.doi.org/10.1016/j. neuroimage.2011.08.026. [31] Hermans EJ, Henckens MJAG, Joëls M, Fernández G. Dynamic adaptation of large-scale brain networks in response to acute stressors. Trends Neurosci 2014;37:304–14. http://dx.doi.org/10.1016/j.tins.2014.03.006. [32] Arnsten AFT. Through the looking glass: differential noradenergic modulation of prefrontal cortical function. Neural Plast 2000;7:133–46. http://dx.doi.org/ 10.1155/NP.2000.133.
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85 [33] Sarter M, Bruno JP. Cortical cholinergic inputs mediating arousal, attentional processing and dreaming: differential afferent regulation of the basal forebrain by telencephalic and brainstem afferents. Neuroscience 2000;95:933–52. [34] Sarter M, Hasselmo ME, Bruno JP, Givens B. Unraveling the attentional functions of cortical cholinergic inputs: interactions between signal-driven and cognitive modulation of signal detection. Brain Res Brain Res Rev 2005;48:98–111. http://dx.doi.org/10.1016/j.brainresrev.2004.08.006. [35] Sarter M, Paolone G. Deficits in attentional control: cholinergic mechanisms and circuitry-based treatment approaches. Behav Neurosci 2011;125:825–35. http://dx.doi.org/10.1037/a0026227. [36] Gruber AJ, Dayan P, Gutkin BS, Solla SA. Dopamine modulation in the basal ganglia locks the gate to working memory. J Comput Neurosci 2006;20:153–66. http://dx.doi.org/10.1007/s10827-005-5705-x. [37] Vijayraghavan S, Wang M, Birnbaum SG, et al. Inverted-U dopamine D1 receptor actions on prefrontal neurons engaged in working memory. Nat Neurosci 2007;10:376–84. http://dx.doi.org/10.1038/nn1846. [38] Dang LC, O’Neil JP, Jagust WJ. Dopamine supports coupling of attention-related networks. J Neurosci 2012;32:9582–7. http://dx.doi.org/10.1523/ JNEUROSCI.0909-12.2012.
85
[39] Coltheart M. Cognitive neuropsychiatry and delusional belief. Q J Exp Psychol 2007;2006(60):1041–62. http://dx.doi.org/10.1080/17470210701338071. [40] Gerrans P. Pathologies of hyperfamiliarity in dreams, delusions and déjà vu. Front Psychol 2014. http://dx.doi.org/10.3389/fpsyg.2014.00097. [41] Choi S-H, Lee H, Chung T-S, et al. Neural network functional connectivity during and after an episode of delirium. Am J Psychiatry 2012;169:498–507. http://dx.doi.org/10.1176/appi.ajp.2012.11060976. [42] Fong TG, Bogardus ST, Daftary A, et al. Cerebral perfusion changes in older delirious patients using 99mTc HMPAO SPECT. J Gerontol A Biol Sci Med Sci 2006;61:1294–9. [43] Alsop DC, Fearing MA, Johnson K, et al. The role of neuroimaging in elucidating delirium pathophysiology. J Gerontol A Biol Sci Med Sci 2006;61:1287–93. [44] Perez DL, Catenaccio E. Confusion, hyperactive delirium, and secondary mania in right hemispheric strokes: a focused review of neuroanatomical correlates. J Neurol Neurophysiol 2011. http://dx.doi.org/10.4172/2155-9562.S1-003. [45] Bogousslavsky J, Ferrazzini M, Regli F, et al. Manic delirium and frontal-like syndrome with paramedian infarction of the right thalamus. J Neurol Neurosurg Psychiatry 1988;51:116–9.
Contents lists available at ScienceDirect
Medical Hypotheses journal homepage: www.elsevier.com/locate/mehy
The network model of delirium James W.S. Young ⇑ Department of Medicine, University of Toronto, Toronto, ON, Canada
a r t i c l e
i n f o
Article history: Received 5 November 2016 Accepted 25 May 2017
Keywords: Delirium Default-mode network Salience network Frontoparietal control network Dorsal attention network
a b s t r a c t The coordinated function of brain networks underlies consciousness, attention and reality testing, all of which are impaired in delirium. The default-mode network, salience network, frontoparietal control network and dorsal attention network are brain networks with integral roles in the maintenance and modulation of the aforementioned functions. Multiple lines of evidence point to their dysfunction in delirium. The convergence of neurotransmitter changes, neuroendocrine and inflammatory stressors on brain networks disrupts bottom-up and top-down attentional control. Neuroimaging and neuroanatomy correlates are potentially consistent with this hypothesis. Overall, this model appears to have significant utility in connecting the seemingly disparate precipitants of delirium while accounting for the clinical manifestations of the syndrome. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Delirium is a functional brain disorder, as there are no specific findings on conventional neuroimaging studies or pathological specimens which clearly correlate with the clinical syndrome. This has been a significant reason for our poor understanding of exactly what delirium is. Functional brain imaging studies with functional magnetic resonance (fMRI) and positron emission tomography (PET) have begun to provide insights into brain dysfunction which arises from problems in the coordinated function of the neural network rather than discrete pathological lesions. Therefore, a functional perspective has the potential to shed light on the nature of delirium. Sanders [1] and Rapazzini [2] have previously proposed the idea of changes in brain network connectivity being responsible for delirium. Functional imaging studies have reinforced the general idea that the cortex and subcortical gray matter are more usefully viewed as a collection of discrete modules, or nodes, that network together [3]. Various combinations of nodes are recruited to perform the diverse functions which the brain is capable of. Multiple functional brain networks have been identified which significantly influence attention, but several appear to be especially important: the default-mode network (DMN) [4], the dorsal attention network (DAN) [5], the salience network (SN) [6,7], and the frontoparietal control network (FPCN) [8] (Table 1). Because inattention is the
⇑ Address: Brampton Civic Hospital, William Osler Health System, Division of Geriatric Medicine, 2100 Bovaird Drive East, Brampton, ON L6R 3J7, Canada. E-mail address: [email protected] http://dx.doi.org/10.1016/j.mehy.2017.05.027 0306-9877/Ó 2017 Elsevier Ltd. All rights reserved.
core cognitive symptom of delirium, these brain networks are suspected to be dysfunctional in delirium.
Brain network function and inattention The function of individual brain networks is an evolving area, but general themes describing the function of these networks have emerged. The DMN is more active when ‘task-activated’ areas of the brain are quiet [4]. Its activity constitutes the baseline state of conscious awareness which includes broad, low-level awareness of multiple internal and external stimuli [11]. When no deliberate thinking is occurring it allows for some awareness of mostly irrelevant external sensory data (to permit detection of unexpected salient stimuli) while mostly allowing for an unstructured stream of internal thoughts to meander along. For more deliberate mental tasks such as imagination or planning the future, the DMN is required, but it is co-opted by the FPCN and directed more intentionally. Consciousness itself may arise largely from the coordinated function of the nodes of the DMN in particular, and thus any disruption of its function could lead to an altered state of consciousness. This is seen physiologically each night when we sleep [11,12], but also in altered states of consciousness due to anesthesia or brain injury [13]. Thus, the integrity of the DMN as a standalone network seems to be an important determinant of the level of consciousness and awareness of the environment. The dorsal attention network (DAN) is activated when externally focused attention is required. When attention is directed externally the DMN is suppressed while the DAN is activated. Conversely, the DAN is suppressed in the absence of external stimuli,
81
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85 Table 1 Brain networks involved in attention, their components, and symptoms of delirium which likely result from their dysfunction. Network
Components
Normal Functions
Network dysfunction leads to
Refs.
Default-Mode Network
1. Medial prefrontal cortex 2. Posterior cingulate cortex 3. Inferior parietal lobule 4. Lateral temporal cortex 5. Hippocampal formation
Active when subjects are at rest, such as in a quiet room with eyes closed, not engaged in any specific task (the ‘default-mode’). Increased activity with tasks of mental simulation and self-referential thought and ‘mind-wandering’. Suppressed when attention focused on specific stimuli or tasks in the external environment.
– Altered level consciousness – Disorganized thinking
of
[4,9]
Frontoparietal Control Network
1. Dorsolateral prefrontal cortex 2. Anterior inferior parietal lobule
Deliberate (‘top down’) control of attention. Works with elements of the salience network.
– Inattention (reduced sustained attention) – Perceptual disturbances
[6,9]
Salience Network
1. Anterior insula 2. Anterior cingulate cortex 3. Amygdala 4. Ventral striatum 5. Substantia nigra/ ventral tegmental area 6. Thalamus
Receives input about reward (ventral striatum/VTA), emotion (amygdala), pain and other senses (thalamus), autonomic activity (AI), cytokine and cortisol levels (ACC). This allows the SN to monitor and prioritize homeostatically important information (grabs attention ‘bottom up’). The SN may act as a switch between DMN and DAN modes of attention. The SN along with input from the FPCN can activate the DAN to recruit focused attention towards homeostatically significant stimuli in the external environment.
– Inattention (distractibility) – Perceptual disturbances – Agitation
[3,6,9]
Dorsal Attention Network
1. Intraparietal sulcus 2. Frontal eye fields 3. Superior parietal lobule 4. Middle temporal motion cortex
Activated when attention is focused on a specific external task or stimulus. Activity is anticorrelated with the default mode network.
– Inattention
[9,10]
while the DMN remains active. This reciprocal activity, termed anticorrelation, is essential for efficient deployment of attention towards different tasks or spheres of interest, and avoids interference from irrelevant brain activity [14]. Christoff et al. view attention management in terms of constraints on attention [9]. Unconstrained attention allows DMN activity to proceed spontaneously (day-dreaming in a quiet room). External stimuli can constrain attention further (‘‘automatic constraints”, e.g. your eye is automatically drawn to the bright pink car that drives by) and this occurs by SN activation of the DAN and suppression of the DMN (this is also referred to as ‘bottomup’ attentional control). Conscious constraints can similarly focus attention (‘‘deliberate constraints”, e.g. ‘‘I need to stop thinking about that movie and focus on this boring lecture”) and this occurs by FPCN-mediated suppression of the DMN and reinforcement of the SN/DAN (this is also referred to as ‘top-down’ attentional control). The FPCN and SN, therefore, can work together (either cooperatively or antagonistically) to guide attention, bottom-up and top-down. Importantly, the SN appears to be a hub which mediates switching between different modes of attention [3,7,15,16]. The SN is in the unique position of gathering information from sensory systems regarding internal and external stimuli which may have motivational significance (e.g. reward, pain, fear, anxiety, arousal, stress, etc.). The SN is a sentinel which monitors these streams, developing context for incoming sensory data, and then prioritizing the most homeostatically relevant and allocating finite attentional resources accordingly. The SN exists at a convergence point for multiple stress signals [16]. This is especially interesting as it relates to delirium, because the brain network which receives such a wide variety of stress signals also has a significant role in switching between unfocused attention (uncoupled DMN activity), automatically focused attention (SN/DAN interactions) and
deliberately-focused attention (SN/FPCN/DAN interactions). Therefore, this leads logically to the possibility that perturbation of SN function is a critical aspect of inattention in delirium (Fig. 1). Dysfunction of the SN has also been postulated to contribute to other neuropsychiatric problems [16,17]. It is well recognized that stress signals from illness can fundamentally influence brain function. ‘Sickness behaviours’ are behaviour changes which classically occur in an organism with acute infection (e.g. fatigue, poor concentration, sleep disturbances, etc.), and provide clear evidence for this [18,19]. They are mediated by a variety of neuroendocrine changes, especially cytokine release and HPA axis activation. Sickness behaviours are typically viewed as adaptive, promoting rest, energy conservation, healing, etc. This is probably true in the context of a healthy organism with an acute and reversible illness (e.g. acute infection). However, they may also provide the substrate for neurobehavioural changes which, when poorly regulated or taken to extremes, resemble many of the symptoms of delirium, which are decidedly maladaptive. Interestingly, inflammatory cytokines have been shown to act on brain areas which are nodes in the salience network [20,21]. Delirium is frequently precipitated by inflammation or drugs. Infection resulting in systemic cytokine changes may result in behaviour changes via the SN as noted above. Dementia is a known significant risk for delirium, likely due to a propensity towards exaggerated neuroinflammation [22,23] and baseline changes in brain networks [24–26]. Sanders suggest that some drugs may influence inhibitory tone within brain networks, which may reduce their connectivity and function [1]. Opioids, benzodiazepines and anticholinergics, among others, have the potential to alter thalamic and striatal function [27], which may disrupt SN function and DMN/DAN switching/anticorrelation. One thing which appears to be common among many drugs with a high potential to precipitate delirium is their ability to significantly impact on one or another
82
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
Fig. 1. A basic network model of delirium. Multiple stressors converge on the salience network (SN), potentially disturbing its function in regulating bottom-up attention. Problems with the frontoparietal control network (FPCN) result in impaired top-down control of attention. This impairs switching between the dorsal attention network (externally directed attention) and the default-mode network (internally directed attention). Problems switching between brain networks/attentional modes, as well as problems arising from individual network dysfunction, result in the cardinal manifestations of delirium (right side of figure). ACC, anterior cingulate cortex; ACh, acetylcholine; AI, anterior insula; DA, dopamine; DLPFC, dorsolateral prefrontal cortex; FEF, frontal eye fields; HPA, hypothalamic-pituitaryadrenal; IPS, intraparietal sulcus; LOC, level of consciousness; MPFC, medial prefrontal cortex; PCC, posterior cingulate cortex; PPC, posterior parietal cortex; VTA, ventral tegmental area.
component of the SN. As noted, the components of the SN are also in receipt of other internal signals which are likely to occur during times of infection, inflammation, and stress. Therefore, in the acutely ill hospitalized patient receiving psychoactive medications, the SN network appears to be bombarded by multiple insults. Anxiety, stress and fear may also be contributors to delirium. These may be present for numerous reasons, as might be expected in a person sick enough to be hospitalized, perhaps even critically ill in the intensive-care unit (ICU) or requiring major surgery. Pain, catheter insertions and the use of physical restraints, among others, may all contribute to stress [28]. Delirium symptoms themselves, such as confusion, disorientation, hallucinations or delusions may compound this effect. Psychological and immune stressors have been shown to act synergistically to promote inflammation and sickness behaviour [29]. Sleep deprivation has adverse effects on the efficient regulation of attentional networks [30]. Activation of the HPA axis, resulting in glucocorticoid and catecholamine secretion, appears to bias the brain away from ‘topdown’ network control in favour of more automatic responses [31]. This may be due to differential adrenergic receptor expression patterns in the various brain structures of the attentional networks [32]. Dopamine and acetylcholine are key neurotransmitters involved in the control of the SN and FPCN, and, coincidentally, the neurotransmitters most widely held to be dysregulated in delirium. The SN integrates contextual information to guide bottom-up allocation of attention. For example, the VTA and related dopaminergic reward areas in the midbrain can flag incoming sensory information as rewarding, or the amygdala can impart emotional context such as fear. Conscious efforts (via the FPCN) direct attention in a top-down manner. Both bottom-up and top-
down mechanisms rely on cholinergic and dopaminergic neurotransmission [33–35]. Cholinergic tone appears to be particularly important in top-down attentional control [35]. Constant updating of attentional goals and priorities appears to function via feedback loops monitoring attentional performance and reward via these two neurotransmitters. An important link between the subcortical components of the SN and the cortical components of the FPCN is the cholinergic nucleus basalis. Dopamine appears to control gating of information flowing into working memory [36]. Too little dopamine leaves working memory inaccessible for updating with relevant information. Too much may lead to inappropriately ascribing salience to irrelevant information and allowing its intrusion into working memory. In both cases, attention is affected. Thus, dopaminergic tone, as it influences attention, may function in a U-shaped curve [37]. Delirium can occur in the context of low dopamine (possibly reflecting the hypoactive end of the delirium spectrum, reduced sustained attention) or excess dopamine (hyperactive delirium, distractibility). Dopamine also appears to enhance coupling between the FPCN and DMN, and reduce coupling between the FPCN and DAN [38]. Overall, inattention in delirium may be due to a combination of problems with both bottom-up and top-down attentional control. What Christoff refers to as ‘automatic constraints’ still seem to partly function in delirium, that is, a novel stimulus can often grab the attention of the delirious person (bottom-up). In fact, this is often true to the point that it is pathological as delirious individuals are often quite distractible. In this sense, convergence of multiple stressors on the SN may tend to disrupt ‘bottom-up’ attentional control. However, dysfunction is also evident in the ‘deliberate constraints’, where there is difficulty intentionally focusing attention (e.g. when an individual with delirium struggles to recite the
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
days of the week backwards, a common bedside test of attention). This suggests that the top-down control via the FPCN is also disrupted in delirium. Perceptual abnormalities Higher level processing failure also appears to explain perceptual disturbances in delirium. Executive areas, especially the right dorsolateral prefrontal cortex (DLPFC) (part of the FPCN), appear to have a role in reality testing. Dysfunction in the right DLPFC [39] and abnormalities in elements of the SN [17] are associated with delusions and hallucinations. Gerrans views reality testing as hypothesis testing [40]. That is, sensory systems register data and generate a basic hypothesis which becomes an initial perception, and then higher systems provide context to help confirm the accuracy of the hypothesis. When incongruous information is registered by lower sensory systems, higher cognitive systems must provide additional analysis using background knowledge and contextual information to allow resolution of the inconsistency. Sometimes an anomaly must be brought forward to the level of conscious attention and thought through to be resolved (‘‘Is that really a man in the corner of my dark bedroom or is that just the pile of clothes on a chair?”) Gerrans proposes a two-hit model, whereby aberrant sensory information can trigger a misperception, but it only persists as a full-blown delusion or hallucination when this second step in analysis also fails. This highlights the theme of problems with top-down processing – if these brain areas are not functioning an delirious individual may not be able to voluntarily address anomalies, and in fact may not even recognize the need to do so. Gaudreau and Gagnon note that anticholinergic and GABAergic drugs, well-known to contribute to delirium, may reduce striatal inhibition of the thalamus, leading to excessive or aberrant release of sensory information to the cortex [27]. This may be the first step in triggering perceptual abnormalities. Not surprisingly, several of the sensitive elements in their model are components of the SN. In delirium, these anomalous events may be ascribed inappropriate salience, and are not subject to secondary scrutiny by executive areas, which are hypoactive, leading to delusions and/or hallucinations. A familiar example is dreaming, a highly ‘unconstrained’ state. Executive areas are conspicuously inactive during REM sleep when most dreaming occurs, while elements of the DMN remains quite active [9]. Dreams often have tangential narratives (perhaps akin to the disorganized thinking of delirium) and bizarre, anomalous content, which is typically accepted without question during the dream. This may reflect unconstrained DMN activity with no executive oversight. Delirium as a state of consciousness between full awareness and unconsciousness may have close analogy with REM-sleep, the stage of sleep most proximal to wakefulness. Delirium is described by those who recover from it as having a dreamlike quality with bizarre experiences. As in a dream, anomalous perceptions may occur entirely free of the usual FPCN/SN oversight function which might otherwise flag them as anomalous and consciously reject them, thereby allowing their perpetuation. Neuroimaging and neuroanatomy The single fMRI study of delirious individuals to date shows changes in anticorrelation between nodes of the DMN and FPCN [41]. Control subjects demonstrated anticorrelated activations of their posterior cingulate cortex (PCC) and DLPFC, as expected. During delirium, individuals lost this anticorrelation, and began to have a positive correlation between activity in these networks. Specifically, positive correlation was seen between the PCC (a DMN node) and bilateral DLPFC (FPCN nodes). Following resolution
83
of delirium these changes resolved. It may be that increasing coupling of the DMN and FPCN results in attention being directed to internal awareness via activity of the DMN, with reduced ability to engage the DAN, and therefore reduced awareness of the external environment. Perhaps DMN activity becomes intrusive and difficult to suppress during delirium. At a minimum, changes were seen in the regulation of key elements of the DMN and FPCN. As noted earlier, dopamine excess may contribute to some of these changes. Connectivity changes were also seen between six subcortical regions. Of note, connections were lost during delirium between areas which are key nodes of the SN (VTA, intralaminar thalamic nuclei), and an important connection between the SN and FPCN (the cholinergic nucleus basalis). Another small functional imaging study of 22 patients who underwent SPECT scanning during an episode of delirium revealed heterogeneous findings, but some evidence of frontal and parietal perfusion abnormalities, which is potentially consistent with localization to elements of the DMN and/or FPCN [42]. A review of multiple SPECT studies across delirium of various etiologies (postoperative, post-stroke, hepatic encephalopathy, etc.) showed great heterogeneity in terms of cerebral perfusion abnormalities [43]. Hypoperfusion was the usual abnormality, and resolution of delirium was accompanied by normalization of blood flow. The authors stated that frontal and parietal perfusion abnormalities were probably most common despite the heterogeneity. Further supporting the idea that brain network dysfunction is the basis for delirium is a review of delirium in stroke patients. Perez et al. reviewed studies that examined the righthemispheric correlates of non-agitated versus hyperactive delirium [44]. Overall, they found that, despite somewhat limited evidence, right hemispheric infarcts in multiple regions including the posterior parietal cortex, temporal, occipital, prefrontal cortices, and also subcortical lesions, have been implicated in both non-agitated confusion as well as hyperactive delirium. Nonagitated confusional states in patients with right hemispheric strokes were preferentially associated with lesions in attentional and perceptual association cortices (prefrontal, posterior parietal and interconnected subcortical regions) which would implicate elements of the FPCN and/or SN. Hyperactive delirium involved regions more associated with the limbic and paralimbic cortices, which would include elements of the DMN and/or SN. Thus, it is tempting to infer that agitation in delirium is more localizable to the limbic components of the SN and/or DMN, whereas inattention arises from problems in the SN and/or FPCN. Another interesting example is that of stroke in the dorsomedial thalamus. This area has been proposed to be essential in binding together the SN nodes into a circuit [6], and stroke here has been reported to cause delirium [45].
Summary In conclusion, delirium is proposed to be a state of dysfunctional brain networks. Changes in DMN connectivity lead to an altered level of consciousness. Problems switching between the DMN and DAN may reduce awareness of the environment. The convergence of multiple neuroendocrine and inflammatory stressors on the SN disrupts bottom-up attentional control leading to distractibility. Drugs altering cholinergic or dopaminergic pathways disrupt SN and/or FPCN function. Inflammation and the neuroendocrine stress response diverts homeostatic resources away from top-down control by the FPCN. Problems with top-down control impair deliberate control of attention. Loss of FPCN/SN function also impairs reality testing which facilitates delusions and hallucinations. Many of these predictions should be testable (Table 2).
84
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85
Table 2 Brain networks predicted to be abnormal in delirium and conditions to examine with functional neuroimaging studies which may support the network model. Network
Issues in delirium
Functions to Probe
DMN
1. Altered LOC 2. Loss of anti-correlated activity with DAN
1. Connectivity among core DMN regions before/during/after delirium 2. Task-induced deactivation of DMN/ activation of the DAN by external stimuli during delirium 3. 3. Effect of antipsychotics on coupling/uncoupling of DMN/DAN/FPCN before/during/after delirium
SN
1. Inattention 2. Perceptual abnormalities
1. Connectivity amongst SN nodes before/during/after delirium 2. Activity of SN during tasks requiring switching from non-directed to externally directed attention during delirium 3. Association of SN changes with reported/observed delusions/ hallucinations 4. Association of serum/CSF cytokine levels with connectivity and/or switching functions of SN
FPCN
1. Inattention 2. Perceptual disturbances
1. Recruitment of FPCN in normal versus delirious individuals at rest and with attentional demands 2. Correlation of activity in DLPFC (especially RDLPFC) with delusions and hallucinations
One of the strengths of this model is that it attempts to provide a coherent pathway linking the precipitants of delirium to the final common clinical manifestations – something which has been lacking in the delirium literature to date. There are many nuances to this theory which warrant further exploration. The brain network paradigm helps to reconcile the diverse array of seemingly unrelated stressors which may precipitate delirium, and appears to be consistent with much of what is already known about delirium. Further fMRI and PET studies, despite potential logistical challenges, may offer greater insights into this theory of delirium. Better treatment of delirium will only come with a more thorough understanding of its neurobiological basis. Declaration of sources of funding None. Conflicts of interest The author declares no conflicts of interest. Acknowledgements Thanks to Drs. David Tang-Wai, Cheryl Grady, Mark Lachmann and Camilla Wong for their feedback on the manuscript. References [1] Sanders RD. Hypothesis for the pathophysiology of delirium: role of baseline brain network connectivity and changes in inhibitory tone. Med Hypotheses 2011;77:140–3. http://dx.doi.org/10.1016/j.mehy.2011.03.048. [2] Rapazzini P. Functional interrelationship of brain aging and delirium. Aging Clin Exp Res 2016;28:161–4. http://dx.doi.org/10.1007/s40520-015-0379-3. [3] Bressler SL, Menon V. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn Sci 2010;14:277–90. http://dx.doi.org/ 10.1016/j.tics.2010.04.004. [4] Buckner RL, Andrews-Hanna JR, Schacter DL. The Brain’s default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci 2008;1124:1–38. http://dx.doi.org/10.1196/annals.1440.011.
[5] Fox MD, Corbetta M, Snyder AZ, et al. Spontaneous neuronal activity distinguishes human dorsal and ventral attention systems. Proc Natl Acad Sci 2006;103:10046–51. http://dx.doi.org/10.1073/pnas.0604187103. [6] Seeley WW, Menon V, Schatzberg AF, et al. Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci 2007;27:2349–56. http://dx.doi.org/10.1523/JNEUROSCI.5587-06.2007. [7] Menon V, Uddin LQ. Saliency, switching, attention and control: a network model of insula function. Brain Struct Funct 2010;214:655–67. http://dx.doi. org/10.1007/s00429-010-0262-0. [8] Vincent JL, Kahn I, Snyder AZ, et al. Evidence for a frontoparietal control system revealed by intrinsic functional connectivity. J Neurophysiol 2008;100:3328–42. http://dx.doi.org/10.1152/jn.90355.2008. [9] Christoff K, Irving ZC, Fox KCR, et al. Mind-wandering as spontaneous thought: a dynamic framework. Nat Rev Neurosci 2016. http://dx.doi.org/10.1038/ nrn.2016.113. [10] Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven attention in the brain. Nat Rev Neurosci 2002;3:201–15. http://dx.doi.org/ 10.1038/nrn755. [11] Horovitz SG, Braun AR, Carr WS, et al. Decoupling of the Brain’s default mode network during deep sleep. Proc Natl Acad Sci U S A 2009;106:11376–81. http://dx.doi.org/10.1073/pnas.0901435106. [12] Samann PG, Wehrle R, Hoehn D, et al. Development of the Brain’s default mode network from wakefulness to slow wave sleep. Cereb Cortex 2011;21:2082–93. http://dx.doi.org/10.1093/cercor/bhq295. [13] Heine L, Soddu A, Gómez F, et al. Resting state networks and consciousness. Front Psychol 2012. http://dx.doi.org/10.3389/fpsyg.2012.00295. [14] Fox MD, Snyder AZ, Vincent JL, et al. The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A 2005;102:9673–8. http://dx.doi.org/10.1073/pnas.0504136102. [15] Sridharan D, Levitin DJ, Menon V. A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks. Proc Natl Acad Sci 2008;105:12569–74. http://dx.doi.org/10.1073/ pnas.0800005105. [16] Uddin LQ. Salience processing and insular cortical function and dysfunction. Nat Rev Neurosci 2014;16:55–61. http://dx.doi.org/10.1038/nrn3857. [17] Manoliu A, Riedl V, Zherdin A, et al. Aberrant dependence of default mode/central executive network interactions on anterior insular salience network activity in schizophrenia. Schizophr Bull 2014;40:428–37. http://dx. doi.org/10.1093/schbul/sbt037. [18] Dantzer R, O’Connor JC, Freund GG, et al. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat Rev Neurosci 2008;9:46–56. http://dx.doi.org/10.1038/nrn2297. [19] Miller AH. Mechanisms of cytokine-induced behavioral changes: psychoneuroimmunology at the translational interface. Brain Behav Immun 2009;23:149–58. http://dx.doi.org/10.1016/j.bbi.2008.08.006. [20] Capuron L, Pagnoni G, Demetrashvili M, et al. Anterior cingulate activation and error processing during interferon-alpha treatment. Biol Psychiatry 2005;58:190–6. http://dx.doi.org/10.1016/j.biopsych.2005.03.033. [21] Felger JC, Miller AH. Cytokine effects on the basal ganglia and dopamine function: the subcortical source of inflammatory malaise. Front Neuroendocrinol 2012;33:315–27. http://dx.doi.org/10.1016/j. yfrne.2012.09.003. [22] Cunningham C. Microglia and neurodegeneration: the role of systemic inflammation. Glia 2013;61:71–90. http://dx.doi.org/10.1002/glia.22350. [23] Perry VH, Holmes C. Microglial priming in neurodegenerative disease. Nat Rev Neurol 2014;10:217–24. http://dx.doi.org/10.1038/nrneurol.2014.38. [24] Dickerson BC, Sperling RA. Large-scale functional brain network abnormalities in Alzheimer’s disease: insights from functional neuroimaging. Behav Neurol 2009;21:63–75. http://dx.doi.org/10.3233/BEN-2009-0227. [25] Agosta F, Pievani M, Geroldi C, et al. Resting state fMRI in Alzheimer’s disease: beyond the default mode network. Neurobiol Aging 2012;33:1564–78. http:// dx.doi.org/10.1016/j.neurobiolaging.2011.06.007. [26] Seeley WW, Crawford RK, Zhou J, et al. Neurodegenerative diseases target large-scale human brain networks. Neuron 2009;62:42–52. http://dx.doi.org/ 10.1016/j.neuron.2009.03.024. [27] Gaudreau J-D, Gagnon P. Psychotogenic drugs and delirium pathogenesis: the central role of the thalamus. Med Hypotheses 2005;64:471–5. http://dx.doi. org/10.1016/j.mehy.2004.08.007. [28] Inouye SK, Charpentier PA. Precipitating factors for delirium in hospitalized elderly persons. Predictive model and interrelationship with baseline vulnerability. JAMA 1996;275:852–7. [29] Brydon L, Walker C, Wawrzyniak A, et al. Synergistic effects of psychological and immune stressors on inflammatory cytokine and sickness responses in humans. Brain Behav Immun 2009;23:217–24. http://dx.doi.org/10.1016/j. bbi.2008.09.007. [30] De Havas JA, Parimal S, Soon CS, Chee MWL. Sleep deprivation reduces default mode network connectivity and anti-correlation during rest and task performance. NeuroImage 2012;59:1745–51. http://dx.doi.org/10.1016/j. neuroimage.2011.08.026. [31] Hermans EJ, Henckens MJAG, Joëls M, Fernández G. Dynamic adaptation of large-scale brain networks in response to acute stressors. Trends Neurosci 2014;37:304–14. http://dx.doi.org/10.1016/j.tins.2014.03.006. [32] Arnsten AFT. Through the looking glass: differential noradenergic modulation of prefrontal cortical function. Neural Plast 2000;7:133–46. http://dx.doi.org/ 10.1155/NP.2000.133.
J.W.S. Young / Medical Hypotheses 104 (2017) 80–85 [33] Sarter M, Bruno JP. Cortical cholinergic inputs mediating arousal, attentional processing and dreaming: differential afferent regulation of the basal forebrain by telencephalic and brainstem afferents. Neuroscience 2000;95:933–52. [34] Sarter M, Hasselmo ME, Bruno JP, Givens B. Unraveling the attentional functions of cortical cholinergic inputs: interactions between signal-driven and cognitive modulation of signal detection. Brain Res Brain Res Rev 2005;48:98–111. http://dx.doi.org/10.1016/j.brainresrev.2004.08.006. [35] Sarter M, Paolone G. Deficits in attentional control: cholinergic mechanisms and circuitry-based treatment approaches. Behav Neurosci 2011;125:825–35. http://dx.doi.org/10.1037/a0026227. [36] Gruber AJ, Dayan P, Gutkin BS, Solla SA. Dopamine modulation in the basal ganglia locks the gate to working memory. J Comput Neurosci 2006;20:153–66. http://dx.doi.org/10.1007/s10827-005-5705-x. [37] Vijayraghavan S, Wang M, Birnbaum SG, et al. Inverted-U dopamine D1 receptor actions on prefrontal neurons engaged in working memory. Nat Neurosci 2007;10:376–84. http://dx.doi.org/10.1038/nn1846. [38] Dang LC, O’Neil JP, Jagust WJ. Dopamine supports coupling of attention-related networks. J Neurosci 2012;32:9582–7. http://dx.doi.org/10.1523/ JNEUROSCI.0909-12.2012.
85
[39] Coltheart M. Cognitive neuropsychiatry and delusional belief. Q J Exp Psychol 2007;2006(60):1041–62. http://dx.doi.org/10.1080/17470210701338071. [40] Gerrans P. Pathologies of hyperfamiliarity in dreams, delusions and déjà vu. Front Psychol 2014. http://dx.doi.org/10.3389/fpsyg.2014.00097. [41] Choi S-H, Lee H, Chung T-S, et al. Neural network functional connectivity during and after an episode of delirium. Am J Psychiatry 2012;169:498–507. http://dx.doi.org/10.1176/appi.ajp.2012.11060976. [42] Fong TG, Bogardus ST, Daftary A, et al. Cerebral perfusion changes in older delirious patients using 99mTc HMPAO SPECT. J Gerontol A Biol Sci Med Sci 2006;61:1294–9. [43] Alsop DC, Fearing MA, Johnson K, et al. The role of neuroimaging in elucidating delirium pathophysiology. J Gerontol A Biol Sci Med Sci 2006;61:1287–93. [44] Perez DL, Catenaccio E. Confusion, hyperactive delirium, and secondary mania in right hemispheric strokes: a focused review of neuroanatomical correlates. J Neurol Neurophysiol 2011. http://dx.doi.org/10.4172/2155-9562.S1-003. [45] Bogousslavsky J, Ferrazzini M, Regli F, et al. Manic delirium and frontal-like syndrome with paramedian infarction of the right thalamus. J Neurol Neurosurg Psychiatry 1988;51:116–9.