The Cerebral Signature for Pain Perception and Its Modulation - 꼭 한번은 읽어야 할 논문

[문형철]

통증은 4단계를 거쳐 느낀다.

pain transduction

pain transmission

pain modulation

pain perception

대뇌 통증인지와 그 조절에 관한 이야기들..

pain related catastrophizing에 관한 탐구가 계속되는 중

panic bird....

The Cerebral Signature for Pain Perception and its modulation.pdf

Our understanding of the neural correlates of pain perception in humans has increased significantly since the advent of neuroimaging. Relating neural activity changes to the varied pain experiences has led to an increased awareness of how factors (e.g., cognition, emotion, context, injury) can separately influence pain perception. Tying this body of knowledge in humans to work in animal models of pain provides an opportunity to determine common features that reliably contribute to pain perception and its modulation. 

- 통증지각에 대한 신경과학은 뇌지도가 완성된 이후 꾸준히 발달함. 통증경험을 바꾸는 요소들 "인지, 감정, context, 손상" 등이 통증지각에 미치는 영향에 대한 탐구

- 통증지각과 통증 조절에 공헌하는 요소들에 대한 탐구

One key system that underpins the ability to change pain intensity is the brain stem’s descending modulatory network with its pro- and antinociceptive components. We discuss not only the latest data describing the cerebral signature of pain and its modulation in humans, but also suggest that the brainstem plays a pivotal role in gating the degree of nociceptive transmission so that the resultant pain experienced is appropriate for the particular situation of the individual.

- 통증강도를 변화시키는 능력은 뇌간의 통증조절 하행로 요소에 달려있음. 

- 우리는 통증의 대뇌신호와 조절에 대해 토의할 뿐 아니라, 뇌간에서 통증전달의 정도를 조절하는 것을 제안함. 

Pain as a Major Medical Health Problem 

Pain that persists for more than three months is defined as chronic and as such is one of largest medical health problems in the developed world. It affects approximately 20% of the adult population, particularly women and the elderly (Breivik et al., 2006). While the management and treatment of acute pain is reasonably good, the needs of chronic pain sufferers are largely unmet, creating an enormous emotional and financial burden to sufferers, carers, and society. Per annum, it is estimated that the cost of chronic pain to Europe is E200 billion and to the USA over $150 billion. 

- 통증이 3개월 이상 지속될때 만성통증이라고 정의하고, 이러한 만성통증은 전세계 의학계에 가장 큰 의학적 건강문제임. 

- 성인의 20%, 특히 여성과 노인에게서 흔함. 

- 급성통증의 치료와 관리는 좋은 효과를 내는 반면 만성통증으로 고통을 받는 사람은 적합한 치료, 관리가 안됨. 매년 유럽에서 2000억, 미국에서 1500억이상의 비용이 지출되고 있음. 

Improvements in our ability to diagnose chronic pain and develop new treatments are desperately needed but to achieve this we need robust and less subjective ‘‘readouts’’ of the pain experience. Innovative methods, like molecular and systems neuroimaging, that can assess changes within the central nervous system (CNS) of patients and relate these findings to the wealth of information from animal studies, have great potential and promise. Indeed, improvements in our ability to identify the extent of changes within the CNS, due to chronic pain, in animals and humans have strengthened the case for considering chronic pain as a disease in its own right. 

- 만성통증의 진단과 치료능력의 증가는 반드시 필요하지만 쉽지 않음. 

- 분자생물학과 뇌지도와 같은 방법의 혁신은 중추신경계내에서 변화에 도달하고....

The mechanisms that contribute to the generation and maintenance of a chronic pain state are increasingly investigated and better understood. A consequent shift in mindset that treats chronic pain as a disease rather than a symptom is accelerating advances in this field considerably. Tying this new body of knowledge from patients and normals with the extensive animal data on pain processing in the CNS is timely. Common aspects regarding how pain perception is mediated and modulated are being identified; this is the focus of our review.

Pain as a Perception 지각으로서 통증

Pain is a conscious experience, an interpretation of the nociceptive input influenced by memories, emotional, pathological, genetic, and cognitive factors. Resultant pain is not necessarily related linearly to the nociceptive drive or input; neither is it solely for vital protective functions.

- 통증은 의식적 경험으로 기억, 감정, 병리, 선천성 그리고 인지적 요소에 의해 영향을 받는 유해자극의 해석임. 

- 이어지는 통증은 직선적 입력과 연관되지 않고, 중요한 보호반응도 아님. 

This is especially true in the chronic pain state. Furthermore, the behavioral response by a subject to a painful event is modified according to what is appropriate or possible in any particular situation. Pain is, therefore, a highly subjective experience, as illustrated by the definition given from the International Association for the Study of Pain (Merksey and Bogduk, 1994): ‘‘an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage.’’

- 그래서 통증은 주관적이고..불유쾌한 감각 그리고 감정적 경험과 연관됨..

By its very nature, pain is therefore difficult to assess, investigate, manage, and treat. Figure 1 illustrates the mixture of factors that we know influence nociceptive inputs to amplify, attenuate, and color the pain experience. We know also from more recent data how a painful experience can occur without a primary nociceptive input (Derbyshire et al., 2004; Eisenberger et al., 2003; Raij et al., 2005; Singer et al., 2004), further complicating the story but perhaps providing an alternative explanation for how pain might arise in difficult clinical cases where the organic cause is not obvious. 

- 통증은 측정, 조사, 관리, 치료가 어려움..

What is clear is that many factors influencing pain percepts are centrally mediated, and our ability to unravel and neuroanatomically dissect their contribution has only been feasible since neuroimaging tools allowed us noninvasive access to the human CNS.

Determining the balance between peripheral versus central influences and ascertaining which are due to pathological versus emotional or cognitive influences will clearly aid decisions regarding the targeting of treatments (i.e., pharmacological, surgical, cognitive behavioral or physical rehabilitation). 

Understanding how complex behavioral influences such as anxiety, depression, belief states, and cognition change the pain experience in animals is difficult to assess due to the lack of sophisticated behavioral paradigms and overdependence on threshold or withdraw measures. However, a greater emphasis is now being placed on measures of spontaneous pain behaviors as well as on developing and utilizing animal models of pain that more clearly mirror specific chronic human pain conditions (Blackburn-Munro, 2004; Lindsay et al., 2005; Schwei et al., 1999). 

Additionally, animal pain models now routinely take into consideration the genetic background, age, gender, and stress levels of the animal as these have been shown to potentially have a significant impact on the pain phenotype observed in animals as well as humans (Boccalon et al., 2006; Craft et al., 2004; Mogil, 1999; Mogil et al., 1997, 2006). Indeed, a more integrated

approach for translating knowledge bidirectionally between human and animal studies is already proving beneficial, as recently demonstrated in the unexpected identification of the potential central role of GTP cyclohydrolase (GCH1), the rate-limiting enzyme for tetrahydrobiopterin (BH4) synthesis, as a key modulator of peripheral neuropathic and inflammatory pain in animal models and humans suffering chronic pain (Tegeder et al., 2006).

Basic Neuroanatomy of Central Pain Processing and the ‘‘Cerebral Signature’’ for Pain Perception

Beyond the peripheral nociceptor and dorsal horn, nociceptive information ascends to the thalamus in the contralateral spinothalamic tract (STT) and to the medulla and brainstem via a spinoreticular (spinoparabrachial) and spinomesencephalic tracts. These tracts serve different purposesrelated to both their lamina origin in the dorsal horn and final central destination (Dostrovsky and Craig, 2006).

Spinal projections to the brainstem are particularly important for integrating nociceptive activity with homeostatic, arousal, and autonomic processes, as well as providing a means to indirectly convey nociceptive information to forebrain regions after brainstem processing. The capacity for projections to the brainstem to directly influence both spinal and forebrain activity clearly suggest these pathways play a direct role in affecting the pain experience; data from animals, healthy subjects, and patients increasingly confirm‎ the central role that the brainstem plays in mediating changes in pain perception.

Functional and anatomical divisions of the thalamus, the main relay site for nociceptive inputs to cortical and subcortical structures, have been made on the basis of their connections to specific spinal cord laminae in various animal species and in humans (Craig, 2003b; Pralong et al., 2004). Lamina I STT neurons largely project to the ventral posterior nucleus (VP), the posterior part of the ventral medial nucleus (VMpo), the ventral posterior inferior nucleus (VPI), and the ventral caudal division of the

medial dorsal nucleus (MDvc). Recent evidence, however, questions the lamina I STT projection to VP (Craig, 2006). Lamina V STT axons terminate in VP, VPI, ventral lateral nucleus, and intralaminar nuclei.

However, the thalamus and its connections spinally and supraspinally are still debated in terms of nociceptive processing in humans. Nevertheless, higher-resolution imaging studies coupled to surgical investigations in humans have confirmed the relevance of nuclei identified to date from animal studies (Montes et al., 2005; Romanelli et al., 2004; Seghier et al., 2005). As a critical relay site, it’s perhaps not surprising that the thalamus is implicated in chronic pain. 

Decreased thalamic blood flow contralateral to the site of pain in patients with cancer has been shown (Di Piero et al., 1991), and in patients developing pain following lesions to the peripheral or central nervous system, thalamic hypoperfusion occurs. Of course, such hypo-perfusion could reflect either a decrease in neural activity or deafferentation. A recent study of a patient with a left medullary infarct (Wallenberg’s syndrome) attempted to

distinguish between these possibilities (Garcia-Larrea

et al., 2006). In this patient, extensive right-sided sensory

deficits were accompanied by left-sided facial pain, and

a PET scan revealed that the reduction of blood flow

occurred in the right thalamus, contralateral to the area

of pain. The repeat scan following pain relief afforded by

motor cortex stimulation showed restoration of thalamic

perfusion. This suggests that thalamic hypoperfusion

indeed reflects the pain state, although it may not be pathophysiological

per se. Future areas of investigation should

include targeted deep-brain stimulation in patients,

informed by white matter diffusion tractrography connectivity

maps, to better determine the role of specific

thalamic nuclei in pain perception and its modulation.

The Pain Matrix

Because pain is a complex, multifactorial subjective experience, a large distributed brain network is subsequently accessed during nociceptive processing. Melzack (1999) first described this as the pain ‘‘neuromatrix,’’ but it’s now more commonly referred to as the ‘‘pain matrix’’; simplistically it can be thought of as having lateral (sensory- discriminatory) and medial (affective-cognitiveeval‎uative) neuroanatomical components (Albe-Fessard et al., 1985). However, because different brain regions

play a more or less active role depending upon the precise interplay of the factors involved in influencing pain perception (e.g., cognition, mood, injury, and so forth), what comprises the pain matrix is not unequivocally defined, and the literature is not always consistent regarding what regions are to be included. 

In our opinion, for the pain matrix to retain its utility, it needs to be viewed not as a stand-alone entity but rather as a substrate that is significantly and actively modulated by a variety of brain regions, and it is this interaction that in large part determines the pain experience.

A recent meta-analysis of human data from positron

emission tomography (PET), functional magnetic resonance

imaging (fMRI), electroencephalography (EEG),

and magnetoencephalography (MEG) studies does provide

clarity regarding the commonest regions found active

during an acute pain experience (Apkarian et al., 2005).

These areas include: primary and secondary somatosensory,

insular, anterior cingulate, and prefrontal cortices as

well as the thalamus (Figure 2). That is not to say these

areas are the fundamental core network of human nociceptive

processing (and if ablated would cure all pain),

although recent studies investigating pharmacologically

induced analgesia do show predominant effects in these

brain regions (Casey et al., 2000; Geha et al., 2007; Rogers

et al., 2004; Wagner et al., 2007; Wise et al., 2002, 2004).

Other regions such as basal ganglia, cerebellum, amygdala,

hippocampus, and areas within the parietal and temporal

cortices can also be active dependent upon the

particular set of circumstances for that individual (Figure

2). Perhaps we need to move toward an individualized

neural ‘‘pain signature’’ rather than forcing this complex,

subjective experience into the constraints of a rigid neuroanatomical

pain matrix (Tracey, 2005b). This is especially

true when considering the neural representation of

chronic, ongoing, or spontaneous pain in patients, something

that has been studied only recently and appears to

not be represented necessarily by the conventional pain

matrix concept (Baliki et al., 2006). And of course data

showing activity of the near entire pain matrix without a

nociceptive input during hypnosis and empathy manipulations

support the notion it is time to reconsider how we define central pain processing with respect to the origin of

the input and resultant perception and meaning (Craig

et al., 1996; Derbyshire et al., 2004; Raij et al., 2005; Rainville

et al., 1997; Singer et al., 2004). That is not to say pain

experienced without a nociceptive input (sometimes referred

to as psychogenic pain) is any less real than ‘‘physically’’

defined pain; indeed, neuroimaging studies have

highlighted the physiological reality of such experiences

due to the extensive neural activation that occurs. Rather,

it is to say we do not yet have a central signature that unequivocally

reflects peripheral nociceptive inputs. Studies

using laser-evoked potentials (LEPs) and MEG that focus

more specifically on temporal aspects of nociceptive processing,

within spatially less well-defined brain regions,

provide signals reflecting the exogenous components

(i.e., fast direct nociceptive input represented by the operculoinsula

and/or S2 region) and endogenous components

(i.e., later integrated and convolved signal represented

by the ACC) (Bentley et al., 2004; Garcia-Larrea

et al., 2003; Hobson et al., 2005; Iannetti et al., 2005a;

Ohara et al., 2004a). Great emphasis has, therefore,

been given to either the spatial or the temporal representation

of nociceptive processing within functionally defined

brain regions, without consideration for how their

activation in concert causes a perception of pain. Pain

perception, similar to many complex experiences,

emerges from the flow and integration of information

among specific brain areas; greater emphasis on understanding

temporal integration among these spatially defined

brain regions is needed and human multimodal imaging

as well as animal studies may provide the solution.

In part, the focus on the rather simplified pain matrix is

a casualty of the intense focus and success pain researchers

have had in understanding molecular and cell

biology of primary afferent sensory neurons and their interactions

in the spinal cord (Julius and Basbaum, 2001;

Mantyh et al., 2002; Morris et al., 2004; Woolf and Salter,

2000). Over the past 20 years, this success has resulted in

a large scale ‘‘migration’’ of pain researchers studying the

involvement of higher centers of the brain (cerebral cortex,

thalamus, amygdala) to focusing on the sensory neuron

and spinal cord. However, with the advent and success

of noninvasive neuroimaging techniques in humans,

greater emphasis in animal experiments must be now

placed on how sensory neurons, the spinal cord, and

higher centers of the brain act in concert if we are to truly

begin to grasp how pain is perceived at a systems level.

Combining data from human imaging studies with neuroimaging,

cellular, molecular, and behavioral studies in

animals has the potential to make similar progress in

understanding how higher centers of the brain are

involved pain perception as has been made in understanding

the neurobiology of primary afferent nociceptors.

Interrelationship between Nociception and Pain

Perception: A Pivotal Role for the Brainstem To understand at a system and molecular level how nociceptive inputs are processed and altered to subsequently influence changes in the pain experienced, it is useful to separately examine the main factors known to alter pain perception.

Cognition and Context

Attention

Anecdotal and experimental observations provide strong

evidence that attention is effective in modulating the sensory

and affective aspects of the pain experience (Levine

et al., 1982; Miron et al., 1989; Villemure and Bushnell,

2002). FMRI and neurophysiology studies show attentionand

distraction-related modulations of nociceptive-driven

activations in many parts of the brain’s pain processing

regions, with concomittant changes in perception (Bantick

et al., 2002; Legrain et al., 2002; Ohara et al., 2004b,

2004c; Petrovic et al., 2000; Peyron et al., 1999). However,

it is not known if a specific cerebral network dedicated

to the modulation of pain by attention exists and if so, if

it is different to the network that produces analgesia in

other circumstances (i.e., during placebo, acupuncture

[Napadow et al., 2007], or pharmacological manipulation).

One candidate network that might elicit pain modulation in

a generalized fashion is the descending pain modulatory

system; another network specific perhaps to attention

could further recruit other brain regions involved in pain

perception. The Descending Pain Modulatory System

The descending pain modulatory system is a well-characterized

anatomical network that enables us to regulate

nociceptive processing (largely within the dorsal horn) in

various circumstances to produce either facilitation (pronociception)

or inhibition (antinociception) (Fields, 2005;

Hagbarth and Kerr, 1954). The pain-inhibiting circuitry, of

which the periaqueductal gray (PAG) is a part, is best

known and contributes to environmental (e.g., during the

fight-or-flight response) and opiate analgesia (Fields,

2005). There are descending pathways that facilitate

pain transmission, however, and it is thought that sustained

activation of these circuits may underlie some

states of chronic pain (see later; Gebhart, 2004; Porreca

et al., 2002; Suzuki et al., 2004). Knowledge regarding

this critically important system largely came from animal

studies. Early work repeatedly demonstrated that spinal

cord excitability was directly influenced by descending

inputs originating in higher centers of the brain and that

this descending modulation could be inhibitory and/or

facilitatory in nature (Basbaum and Fields, 1984; Porreca

et al., 2002; Ren and Dubner, 2002). The ability of higher

centers of the brain to modulate the transmission of nociceptive

information in the CNS was demonstrated in the

early 1900s by Sherrington who showed that nociceptive

reflexes were enhanced following spinal cord transection

(Sherrington, 1906). Over the last several decades, evidence

has accumulated that a variety of brain regions

are involved in this descending modulation and include the

frontal lobe, anterior cingulate cortex (ACC), insula, amygdala,

hypothalamus, PAG, nucleus cuneiformis (NCF), and

rostral ventromedial medulla (RVM). Figure 3 illustrates the 

key anatomical features of the descending pain modulatory system.

More recently, researchers have investigated whether

alteration in people’s attention influences brainstem activity

and, therefore, nociceptive processing via these corticobrainstem

influences. In an early study using highresolution

imaging of the human brainstem, we showed

significantly increased activity within the PAG in subjects

who were distracted compared to when they paid attention

to their pain, with concomitant changes in pain ratings.

Indeed, the change in pain rating between attending

and distracting conditions correlated with the change in

PAG activity across the group, suggesting a varying

capacity to engage the descending inhibitory system in

normal individuals (Tracey et al., 2002). Further work using

a counting stroop cognitive task attempted to identify the

cortical structures involved in mediating this brainstem

influence and subsequent change in pain matrix activity

to produce behavioral analgesia (Bantick et al., 2002).

Valet and coworkers extended the work further by using

connectivity analysis, an advanced method of analyzing

functional imaging, on FMRI data collected from controls

receiving nociceptive stimulation while performing a similar

distraction/cognitive task. They showed that the cingulofrontal

cortex exerts top-down influences on the PAG

and posterior thalamus to gate pain modulation during

distraction (Valet et al., 2004). These studies provide clear

evidence for the involvement of brainstem structures in

the attentional modulation of pain perception, and recent

work using diffusion tractrography confirm‎s that anatomical

connections exist between cortical and brainstem regions

in the human brain, thereby enabling such top-down

influences (Hadjipavlou et al., 2006). Adventurous studies

examining how biofeedback aids both a normal subject’s

or a chronic pain patient’s capacity to modulate their pain

experience, using real-time FMRI data analysis procedures,

provide novel ways to help us better understand

the cortical regions involved in the attentional control of

pain, enabling novel treatment options (deCharms et al.,

2005). A clinical feature of many chronic pain patients is

‘‘hypervigilance’’ to pain and pain-related information.

This has a direct impact not only on their resultant pain

perception but also quality of life if it impacts cognitive

performance. There are a number of explanations for

this attentional effect that are often the target for interventions

such as cognitive behavioral therapies (Crombez

et al., 2005). Clearly, recognizing the central role of the

brainstem in helping to mediate the analgesia and focusing

efforts to strengthen cortical connectivities to structures

such as the PAG will be important in future work

and treatment developments.

Context

The commonest route to understand how context can influence pain perception is via a placebo manipulation. Much of our knowledge of the placebo effect has come from early animal studies based upon Pavlovian conditioning and expectancy (Benedetti et al., 2005; Haour, 2005). Recent work to translate these findings to humans has helped provide a systems framework by which the placebo effect and subsequent analgesia is mediated (Colloca and Benedetti, 2005; Price et al., 2006, 2007).

Descending influences from the diencephalon, hypothalamus, amygdala, ACC, insular, and prefrontal cortex that elicit inhibition or facilitation of nociceptive transmission via brainstem structures are now thought to occur during placebo analgesia. Using PET, Petrovic and colleagues (2002) confirmed that both opioid and placebo analgesia are associated with increased activity in the rostral ACC, but they also observed a covariation between the activity in the rostral ACC and the brainstem during both opioid

and placebo analgesia, but not during pain alone. 

Interestingly, high responders to placebo mirrored their ability to respond to real opioid injection compared to low placebo responders, possibly reflecting a genetic influence in muopioid receptors. Recently, Zubieta and colleagues (2005) confirmed that placebo analgesic effects are mediated by endogenous opioid activity on mu-opioid receptors using a molecular imaging approach in humans. Wager and colleagues (2004) extended these observations to consider

whether or not placebo treatments produce analgesia by

altering expectations. Using a conditioning design, Wager

found that placebo analgesia was related to decreased brain activity in classic pain-processing brain regions

(thalamus, insula, and ACC) but was additionally associated

with increased activity during anticipation of pain in

the prefrontal cortex (PFC); an area involved in maintaining

and updating internal representations of expectations.

Stronger PFC activation during anticipation of pain was

found to correlate with greater placebo-induced pain relief

and reductions in neural activity within pain regions.

Furthermore, placebo-increased activation of the PAG

region was found during anticipation, the activity within

which correlated significantly with dorsolateral PFC

(DLPFC) activity. These results support the concept that

prefrontal mechanisms can trigger opioid release within

the brainstem during expectancy to influence the descending

pain modulatory system and subsequently

modulate pain perception. In a very recent experiment

by Scott and colleagues, they examined the relationship

between placebo-related expectations and dopamine

release within the nucleus accumbens in humans using

molecular imaging. They found that activation of dopamine

release occurred during placebo administration

and that the extent of release was related to anticipated

effects as well as perception-anticipation mismatches

and subsequent placebo development. Furthermore,

using a reward task and fMRI, they found that expectancy

of monetary gain increased nucleus accumbens activity

proportionally to those measures obtained from the

molecular imaging study in the same subjects (Scott

et al., 2007). Studies such as these are significantly improving

our understanding of the placebo effect as well

as expectation of relief; areas of significant relevance for

assessing treatment outcomes in clinical trials.

Emotions and Mood

For both chronic and acute pain sufferers, mood and emotional state has a significant impact on the resultant pain perception and ability to cope. For example, it is a common clinical and experimental observation that anticipating and being anxious about pain can exacerbate the pain experienced. Anticipating pain is highly adaptive; we all learn in early life to avoid hot pans on stoves and not to put your finger into a candle flame. However, for the chronic pain patient it becomes maladaptive and can lead to fear of movement, avoidance, anxiety, and so forth.

Many studies aimed at understanding how anticipation and anxiety cause a heightened pain experience have been performed over the past decade (Hsieh et al., 1999; Ploghaus et al., 1999, 2000, 2001; Porro et al., 2002, 2003; Song et al., 2006). Critical regions involved in amplifying or exacerbating the pain experience include the entorhinal complex, amygdalae, anterior insula, and prefrontal cortices. More recently, we have found that the degree of anticipation to a pain event positively correlates with the reported pain intensity across a group of healthy individuals, and this amplification is mediated in part via activity

within the ventral tegmentum area of the brainstem and entorhinal cortex, as well as the PAG (Fairhurst et al.,

2007). This data obtained in humans correlates well with

animal data in demonstrating that there is a clear interaction

between pain, anxiety, and mobility. While the body

of animal data is at times conflicting (anxiety can be proor

antinociceptive depending on the models used and

the endpoints assessed), what is clear is that the pain

response of the animal is emotion-specific, i.e., higher

centers of the brain in large part determine the behavioral

response to the same noxious stimulus. What is largely

lacking, however, is a cellular, molecular, and systems understanding

of how distinct areas of the brain interact to

cause a heightened or diminished pain experience and

how prior ‘‘memories of pain’’ are stored so as to influence

current and future experiences of pain. Incorporating our

understanding of noradrenergic, serotonergic, opioidergic,

and now dopaminergic function in acute and chronic pain

processing from animal studies with the capacity to image

some of these neurotransmitters via molecular imaging in

humans and manipulate their levels with pharmacological

agents will lead to rapid advances in our understanding

of how complex moods influence pain experience.

Depressive disorders often accompany persistent pain.

Central neuronal plasticity may underlie both conditions,

further complicating our ability to dissect the components

contributing to clinical pain disorders (Castren, 2005).

Although the exact relationship between depression and

pain is unknown, with debate regarding whether one

condition leads to the other or if an underlying diathesis

exists, studies have attempted to isolate brain regions,

such as the amygdale, that may mediate their interaction

(Neugebauer et al., 2004). In another fMRI study, Giesecke

and colleagues (2005) showed that activation in amygdala

and anterior insula differentiated patients with fibromyalgia

with and without major depression; however, more

studies that specifically address the interaction between

pain and depression are needed if we are to resolve the

neuroanatomical basis for the comorbidity.

Another negative cognitive and mood affect that

impacts pain is catastrophizing. This construct incorporates

magnification of pain-related symptoms, rumination

about pain, feelings of helplessness, and pessimism about

pain-related outcomes (Edwards et al., 2006), and it is defined

as a set of negative emotional and cognitive processes

(Sullivan et al., 2001). A study on fibromyalgia patients

found that pain catastrophizing, independent of the

influence of depression, was significantly associated with

increased activity in brain areas related to anticipation of

pain (medial frontal cortex, cerebellum), attention to pain

(dorsal ACC, dorsolateral prefrontal cortex), emotional aspects

of pain (claustrum, closely connected to amygdala),

and motor control (Gracely et al., 2004). Clearly, these

results support the notion that catastrophizing influences

pain perception through altering attention and anticipation,

as well as heightening emotional responses to pain.

It is interesting to speculate whether activity in such

‘‘emotional’’ brain regions due to chronic pain impacts

performance in tasks requiring emotional decision making.

Apkarian and colleagues (2004a) showed that during

the Iowa Gambling Task, a card game developed to study emotional decision making, chronic pain patients displayed

a specific cognitive deficit compared to controls,

suggesting such an impact might exist in everyday life.

Such experiments are hard to reproduce in animal studies;

however, if we are to understand what neural systems mediate

this potential disruption, more work is needed combining

these more complex paradigms with neuroimaging

techniques (Seymour et al., 2007).

Prefrontal, Frontal, and Insular Cortex in Chronic Pain

It is clear from these few studies described and others in the literature (Apkarian et al., 2001, 2005; Lorenz et al., 2003; Phillips et al., 2003; Witting et al., 2006), that rostral anterior insula and pronounced PFC activation are consistently found across clinical pain conditions, irrespective of underlying pathology. A recent meta-analysis by Schweinhardt and colleagues (2006) highlighted that clinical pain is located significantly more rostrally in the anterior insula

than nociceptive pain in healthy volunteers, consistent

perhaps with current theories regarding interoception

and body awareness (Craig, 2003a; Craig et al., 2000;

Critchley et al., 2004). Indeed, anterior insular activity is

found not only during subjective feelings of pain, but is

associated with anxiety, depression, irritable bowel syndrome,

chronic fatigue, fibromyalgia, somatization, and

fear. Paulus and Stein (2006) have recently proposed

a role for the anterior insula in generating an altered interoceptive

prediction signal in individuals prone to anxiety.

In their model, an increased predictive signal of a prospective

aversive body state (i.e., pain) triggers an increase in

anxiety, worried thoughts, and avoidance behaviors,

with possible pain amplification. This model certainly fits

with current data. We are only beginning to unravel the roles of specific

prefrontal and frontal cortical regions in pain perception;

from other areas of cognitive neuroscience we can postulate

roles reflecting emotional, cognitive, and interoceptive

components of pain conditions, as well as perhaps

processing of negative emotions, response conflict, decision

making, and appraisal of unfavorable personal outcomes

for more medial FC, ventrolateral, and medial

PFC (Dolan, 2002; Kalisch et al., 2006; Ridderinkhof

et al., 2004; Rushworth et al., 2004, 2005, 2007; Sakagami

and Pan, 2007). Baliki recently showed in chronic back

pain patients increased activity in mPFC, including rostral

ACC, during episodes of sustained high ongoing pain.

Furthermore, the medial PFC activity was strongly related

to the intensity of chronic back pain (Baliki et al., 2006). In

other pain studies, connectivity analyses of functional

imaging data have highlighted the relevance of frontal cortical

regions in mediating or controlling the functional

interactions among key nociceptive processing brain

regions to subsequently produce changes in perceptual

correlates of pain, independent of changes in nociceptive

inputs (Eisenberger et al., 2003; Lorenz et al., 2002;

Tracey, 2005a). A specific role for the lateral PFC as

a ‘‘pain control center’’ has been put forward in a study

of experimentally induced allodynia in healthy subjects

(Lorenz et al., 2002). In this study, increased lateral PFC

activation was related to decreased pain affect, supposedly

by inhibiting the functional connectivity between

medial thalamus and midbrain, thereby driving endogenous

pain-inhibitory mechanisms. More recent studies

looking at control and pain support these concepts. Wiech

and colleagues manipulated the level of control healthy

subjects had over their pain and produced changes in

pain ratings dependent upon the control condition and

the subject’s internal locus of control. Using fMRI, they

showed that the analgesic effect of perceived control relies

on activation of right anterolateral PFC (Wiech et al.,

2006). It is perhaps important to note that the prefrontal

cortex (specifically the dorsolateral PFC) is a site of major

neurodegeneration and potential cell death in chronic pain

patients (Apkarian et al., 2004b). These unexpected findings

suggest that severe chronic pain could be considered

a neurodegenerative disorder that especially affects the

PFC. This could in turn have consequent negative effects

on the descending inhibitory system and contribute to

their chronic pain state.

There is no doubt that the extent to which a stimulus

(like pain) is identified as emotive and subsequently

produces and regulates an affective or emotive state is

dependent upon activity in many other regions such as

the amygdala, insular, ventral striatum, ACC, and hippocampus,

as well as the PFC (Phillips et al., 2003). However,

it remains to be determined whether emotional and

cognitive influences such as hypervigilance, catastrophizing,

anxiety, or depression all mediate part of their recognized

influence on pain perception in chronic pain

sufferers via the descending pain modulatory system.

Recent advances in our ability to image activity within

the human brainstem (Dunckley et al., 2005; Tracey and

Iannetti, 2006) and map white matter tracts within the

human brain noninvasively using diffusion tensor imaging

and tractography (Behrens et al., 2003; Johansen-Berg

and Behrens, 2006; Le Bihan, 2003) are already contributing

to a better understanding of the neuroanatomical connectivity

among different cortical, subcortical, and brainstem

regions and, therefore, the likelihood of finding

a functional nociceptive link for these ‘‘top-down’’ influences

(Hadjipavlou et al., 2006). It is known from animal

studies that the anterior insula is connected to brainstem

structures such as PAG, RVM, NCF, and parabrachial nucleus

(Fields, 2005); this provides a mechanism to partly

explain how emotions and mood might influence changes

in pain intensity perception. Additionally, as several of the

brainstem descending modulatory regions are either

ascending homeostatic integration sites or descending

autonomic premotor sites, it is perhaps feasible that a specific

link exists between pain, homeostasis, and interoception.

Changes in the affective and cognitive state

might influence interoception to produce a bias in behavior

and decisions that affect outcome and pain perception.

Evidence is accumulating to support such concepts linking

homeostasis and pain; a recent study has provided

the first evidence that the vanilloid receptor, TRPV1 (a cation channel that serves as a polymodal detector of pain

producing stimuli like capsaicin, protons [pH < 5.7] and

heat) is also tonically activated in vivo and as such is

involved in body temperature regulation (Gavva et al.,

2007). Another study examined whether estradiol changes

in women influence pro- and antinociceptive mechanisms

(Smith et al., 2006). They found convincing estrogenassociated

variations in the activity of mu-opioid neurotransmission

that correlated with individual ratings of the

sensory and affective perceptions of the pain, as well as

the subsequent recall of that experience. Molecular imaging

studies like these not only illustrate how systemic

biochemical changes influence behavior and perception,

but also provide novel opportunities to translate research

findings between animal models and humans. Injury

Recently, changes within the descending pain modulatory

network have been implicated in chronic pain and in functional

pain disorders (Gebhart, 2004; Porreca et al., 2002;

Suzuki et al., 2004; Tracey and Dunckley, 2004). Changes

are defined in terms of patients having either a dysfunctional

descending inhibitory system or an activated and

enhanced descending facilitatory system. There has been

convincing evidence revealed regarding the differential involvement

of the PAG, RVM, parabrachial nucleus (PB),

dorsal reticular nucleus, and NCF in the generation and

maintenance of central sensitization states and hyperalgesia

in both animal models and, for the first time in

humans, a human model of secondary hyperalgesia (Zambreanu

et al., 2005). This evidence has added to the

literature and the general notion that these structures

play an important role, in addition to the dorsal horn, in

generating and maintaining central sensitization.

Recent clinical studies are further highlighting how dysfunction

within this system can be sufficient to generate

key symptoms of chronic pain. A study by Wilder-Smith

and colleagues (2004) investigated whether patients with

irritable bowel syndrome had hypersensitivity and pain

upon distension due to abnormalities in endogenous

pain inhibitory mechanisms; they found this to be the

case for patients compared with controls. In a study of

central post-stroke pain following an ischemic brainstem

injury, patients were found to experience pain in the

body side contralateral to their lesion. Furthermore, by

studying the patients using PET and a radiolabeled opioid

receptor agonist, Willoch and colleagues (2004) found

dramatic reductions in opioid-receptor binding in several

key nociceptive processing brain regions. These findings

suggest that an imbalance of excitatory and inhibitory

mechanisms contributes to either the generation or the

modulation of a pain experience both in patients and in

controls. Mayer and colleagues (2005) examined whether

visceral hypersensitivity found in patients with IBS might

arise as a consequence of aberrant top-down descending

influences. In a PET study, they observed greater activation

of limbic and paralimbic circuits during rectal distension

in patients with IBS compared with control subjects

or patients with quiescent ulcerative colitis. Functional

connectivity analysis suggested a failure to activate the

right lateral frontal cortex permits the inhibitory effects of

limbic and paralimbic circuits on PAG activation, the

consequence of which may be visceral hypersensitivity.

The same group recently examined the longitudinal

change in perceptual and brain activation response to visceral

stimuli in IBS patients (Naliboff et al., 2006). Among

several changes, they noted a decreased brainstem activity

to both the anticipation and experience of rectal inflation

after 12 months.

Changes within the descending pain modulatory network

in chronic pain, in terms of patients having either

a dysfunctional descending inhibitory system or an activated

and enhanced descending facilitatory system, are

clearly implicated in these and increasingly other studies

(Edwards, 2005; Goadsby, 2007; Sandrini et al., 2006). Seifert

and Maihofner recently performed an fMRI study in

healthy subjects experiencing innocuous and noxious

cold as well as menthol-induced cold allodynia. Comparing

cold allodynia with equally intense cold pain conditions,

they show increased activations in bilateral dorsolateral

prefrontal cortices and brainstem during cold allodynia

(Seifert and Maihofner, 2007); reflecting the specificity of

brainstem activity for this chronic pain symptom. These

findings are supported by another study using the capsaicin

model of hyperalgesia showing brainstem activity specific

to secondary hyperalgesia (Mainero et al., 2007), results

that fit with a clinical study showing differential

involvement of brainstem nuclei between affected and unaffected

sides in chronic neuropathic pain patients (Becerra

et al., 2006). Furthermore, recent pharmacological

studies are showing that gold-standard agents used to

treat key symptoms of neuropathic pain mediate their influence

on brainstem structures (Iannetti et al., 2005b). While activation of the descending inhibitory system is

generally viewed as desirable, it also has the potential to

mask a pain that would be useful in early diagnosis and

treatment of a disease (Mantyh, 2006). Recently, a transgenic

mouse that spontaneously develops pancreatic

cancer was used to determine if the endogenous pain inhibitory

system might be tonically active in masking earlystage

pancreatic cancer pain (Sevcik et al., 2006). These

mice, like humans with pancreatic cancer, usually only

display spontaneous morphine-reversible visceral painrelated

behaviors when the cancer is advanced, the tumor

has metastasized to vital organs, and effective treatment

or cure is no longer possible (Hawes et al., 2000). To test

whether CNS pathways might be masking early-stage

pancreatic cancer pain, mice that spontaneously develop

pancreatic cancer received subcutaneous administration

of the CNS penetrant opioid antagonists naloxone or

naltrexone. Following administration of these opioid antagonists,

mice with early pancreatic cancer, who before

demonstrated no spontaneous pain behaviors, now displayed

a robust visceral pain-related behaviors. Furthermore,

the endogenous opiates that tonically inhibit pancreatic

cancer pain appear to exert their actions in the 

CNS, as subcutaneous administration of the non-CNS

penetrant opiate antagonist naloxone-methiodide did

not induce visceral pain behaviors in early stage-mice,

whereas intracerebroventricular injection of this same

compound increased visceral pain behaviors. These

data suggest that a CNS opiate-dependent mechanism

tonically masks early-stage pancreatic cancer pain (Sevcik

et al., 2006). What is impressive about these results

is just how effectively the CNS can modulate pain. Once

pancreatic cancer pain appears, in both humans and

mice, it is frequently severe. This endogenous CNS inhibition

of pain in pancreatic cancer is reminiscent of the

impressive analgesia that was originally demonstrated

by Reynolds in 1969, where it was shown that electrical

stimulation of rat PAG in awake moving rats allowed

abdominal surgery to be conducted without the use of

general anesthesia (Reynolds, 1969).

Together, these and other studies reinforce the concept

that CNS inhibitory or facilitatory mechanisms are remarkable

in their efficacy in being able to amplify or decrease

the pain experience (Vanegas and Schaible, 2004). Therefore,

understanding which CNS areas are involved in

engaging or disengaging this descending modulatory system

has significant potential to not only further our understanding

of how pain is perceived, but also in developing

mechanism-based therapies for treating different types

of acute and chronic pain. Figure 4 summarizes our current

opinion regarding the central relevance of the brainstem

and the descending modulatory system in affecting

the pain experienced in varying circumstances.

Molecular Imaging and Metabolic Changes:

Altered Opioidergic and Dopaminergic Pathways

The availability of PET ligands for opioid and dopamine receptors

has allowed the study of these receptor systems

in several clinical pain states, providing yet further evidence

for an imbalance of excitatory and inhibitory mechanisms

contributing to the generation or modulation of

pain in patients. Early opioid ligand studies (Jones et al.,

1994) showed decreased binding in patients with chronic

pain that normalized after reduction of their pain symptoms.

Regional differences in ligand binding have also

been found in neuropathic pain studies (Jones et al.,

1999, 2004; Willoch et al., 2004) with decreased binding

in several key areas involved in pain perception. Future

studies, in particular longitudinal studies that correlate

binding potential with pain intensity, could help elucidate

whether decreased receptor availability is caused by

increased release of endogenous opioids or decreased

receptor density. A study of restless legs syndrome

showed that the opioid-binding potential is negatively correlated

with the affective dimension of the McGill Pain

Questionnaire (von Spiczak et al., 2005), suggesting a

decrease of receptor density might be responsible for

the increase in pain affect.

The dopaminergic pathways have also been implicated

in pain processing in animal (Altier and Stewart, 1999;

Schmidt et al., 2002) and patient studies (Ertas et al.,

1998; Hagelberg et al., 2004; Taub, 1973). From certain

studies, it is hypothesized that the reduced activity may

mediate increased pain behavior found in animal models

of chronic stress (da Silva Torres et al., 2003; Scheggi

et al., 2002). A recent PET study in fibromyalgia patients

by Wood and colleagues showed reduced presynaptic

dopaminergic activity in several brain regions in which

dopamine plays a critical role in modulating nociceptive

processes (Wood et al., 2007), possibly highlighting dopaminergic

dysregulation with functional pain disorders

where stress is a prominent aggravating factor (Wood,

2004). Similarly to the endogenous opioid system, the

issue of cause and effect between a ‘‘functional hypodopaminergic

state’’ and pain has yet to be resolved. The

observation that reduced pain thresholds in patients with

Parkinson’s disease normalized, with corresponding

reductions in brain activation (insula and ACC), following

administration of levodopa suggests that attenuation of

dopaminergic activity underlies some chronic pain states

(Brefel-Courbon et al., 2005). However, the current data

from animal and patient studies on the role of dopamine

mechanisms in pain, using either dopamine agonists or

antagonists, are conflicting with regard to directionality

(i.e., pro- or antinociceptive responses upon dopamine

release) and location (i.e., nigrostriatal or mesolimbic

pathways). A study by Scott and colleagues (2006)

attempted to clarify this issue and showed that variations

in the human pain stress experience are mediated by ventral

and dorsal basal ganglia dopamine activity. Specifically,

they found that activation of nigrostriatal dopamine

D2 receptor-mediated neurotransmission was positively

associated with individual variations in subjective ratings

of sensory and affective qualities of pain; contrasting

this, mesolimbic dopamine activation was only associated

with variations in the emotional responses of the individual during the pain challenge (i.e., increases in negative affect

and fear ratings).

Such molecular imaging studies are providing highly

novel information regarding pain processing in humans.

Although the data are not conclusive regarding causality,

it clearly shows that the brains of patients suffering chronic

pain are fundamentally disturbed in ways neither considered

nor appreciated before. New avenues for exploration

and possible treatment targets are open, and this area is

becoming an active area of exploration.

Novel Areas of Investigation

As the problem of pain and the key role of the brain

becomes increasingly well recognized, more research is

being directed toward a better understanding of the underlying

mechanisms. Some of the newest and more novel

areas of investigation are briefly summarized here.

Structural Imaging

The recent finding that significant atrophy exists in the

brains of chronic pain patients (Apkarian et al., 2004b;

Grachev et al., 2000; Schmidt-Wilcke et al., 2005) highlights

the need to perform more advanced structural imaging

measures and image analyses to quantify fully these

effects. Determining what the possible causal factors are

that produce such neurodegeneration is difficult. Candidates

include the chronic pain condition itself (i.e., excitotoxic

events due to barrage of nociceptive inputs), the

pharmacological agents prescribed, or perhaps the physical

lifestyle change subsequent to becoming a chronic

pain patient. Carefully controlled longitudinal studies are

now needed as this rapidly becomes, along with diffusion

tractography studies to detect and quantify white matter

tracts, an active area of research. Such studies might

best be performed in animals.

Spinal Cord Imaging

Clearly, to determine the extent of changes present

within the CNS, we must develop methods that allow

noninvasive access to the changes within the human

spinal cord. There is an extensive literature from animal

studies regarding nociceptive processing within the dorsal

horn to draw upon, and recent technical developments

provide hope that translation to human studies will be soon feasible (Brooks et al., 2006; Maieron et al., 2007).

Imaging Microglial Activation

Recently, there has been considerable excitement over the possible role that microglia play in the development and maintenance of chronic pain states (Watkins et al., 2001). To translate these exciting animal findings to humans requires an ability to perform in vivo imaging of the recruitment of microglial and macrophages into the spinal cord and brain during the development of chronic pain states. Ultrasmall, superparamagnetic particles of iron oxide (or USPIO) are nanoparticles that might provide,

like the PET ligand PK11195, an indication of microglial and macrophage recruitment (Bonnemain, 1998; Bulte and Frank, 2000; Banati, 2002). Linking these studies to those being currently performed provides an ideal opportunity to further explore the functional role of microglial in developing chronic pain states.

Genetics

We cannot ignore the possibility that our genes influence

both how nociceptive stimuli are processed and how the

brain reacts to peripheral injury and increased nociceptive

inputs. Similarly, we cannot ignore the central role that our

life experiences have on both these processes. Coghill

and colleagues (2003) addressed the issue that some individuals

claim to be ‘‘sensitive’’ to pain, whereas others

claim they tolerate pain well. In their experiment, individuals

who rated the pain highest exhibited more robust

pain-induced activation of S1, ACC, and PFC compared

with those who rated pain lowest. The key question is

whether this increased pain report and correlated objective

readout is nature or nurture driven. The answer is

perhaps central to a better understanding of why certain

patients develop chronic pain syndromes and others do

not and perhaps explaining differences in treatment

outcomes. Similarly, if these observations are driven by

nurture, what influences in a person’s upbringing are relevant

for altering nociceptive pathways to again alter the

processing and resultant pain perception? Studies are

beginning to link genetic influences on human nociceptive

processing with physical processes within the brain.

Zubieta and colleagues (2003) examined the influence of

a common functional genetic polymorphism affecting

the metabolism of catecholamines on the modulation of

responses to sustained pain in humans using psychophysical

assessment and PET. Individuals homozygous

for the met158 allele of the catechol-O-methyltransferase

polymorphism (val158 met) showed diminished regional

mu-opioid system responses to pain (measured using

PET) and higher sensory and affective ratings of pain compared

with heterozygotes. This provides clear evidence

that our genes influence nociceptive processing within

the brain and consequently our pain experience. The link

between our genes and pain perception during acute

and chronic pain experiences is one of the most exciting areas of pain research at present and is being led primarily by animal studies but with fast translation to human studies (Tegeder et al., 2006). Novel genes are being identified that force us to reconsider pain mechanisms as they relate to disease and perception. The hope is that this will lead to novel treatments that provide better efficacy for patients. 

Conclusion

We have attempted to summarize from a largely systems neural-processing perspective the current state of knowledge regarding how pain is perceived during varying circumstances. We propose a central role for the brainstem and the descending pain modulatory system in affecting the resultant pain experienced. Anatomical-, functional-,molecular-, and tractography-based studies are further elucidating connectivities between subcortical and cortical structures to specific regions of the brainstem.

This provides a framework for integrating nociceptive inputs with top-down influences so that appropriate modulation of these inputs, prior to higher-order processing, is achieved to ensure the resultant pain experienced is appropriate for that particular circumstance. In the chronic pain state, we believe this integration is disrupted via both bottom-up and top-down influences, contributing to the generation and maintenance of a heightened pain experience.

[정혜]

감사합니다