Abstract
The thalamus is a key relay station for the transmission of nociceptive information to the cerebral cortex. We review the input–output connection, functional imaging, direct neuronal recording, stimulation, and lesioning studies on the involvement of thalamus in acute and chronic pain functions. Based on its specific reciprocal connection with the cerebral cortex, strong nociceptive responsiveness, and the severe chronic pain when it is damaged, the thalamus may hold the key to pain consciousness and the key to understanding spontaneous and evoked pain in chronic pain conditions. A work plan is proposed for future study.
Keywords
cerebral cortex; pain: neuropathic; pain: thalamic; thalamus;
1. Introduction
The thalamus is the gateway to the cerebral cortex. All cortical-bound somatosensory inputs relay through the thalamus. One major group of these somatosensory inputs is the nociceptive input. Nociceptive inputs from the skin, deep structures, and visceral organs converge in the thalamus en route to the cerebral cortex. A hundred years ago, Head and Holmes1 designated the thalamus the essential organ of the affective side of our sensation, especially pain. The objectives of the present review are to review the importance of the thalamus in pain function, posit main questions, and propose what needs to be done. Recent reviews on similar topics include those published by Basbaum and Jessell,2 Apkarian,3 Dostrovsky,4 Jones,5 and Lima.6
2. Anatomy
The word “thalamus” has its origin in Greek and means “the inner chamber.” The thalamus is located in the center core of the human brain. Developmentally, it comprises three parts: epithalamus, dorsal thalamus, and ventral thalamus. Among them, the dorsal thalamus is the principal part that connects intimately with the cerebral cortex. The reticular thalamic nucleus (RT) and the zona incerta (ZI) are the most important nuclei in the ventral thalamus. Both nuclei are GABAergic, and neurons in the ventral thalamus project primarily to the dorsal thalamus to modulate thalamic information flow.7, 8
The principal somatosensory thalamic nuclei are parts of the dorsal thalamus. Peripheral thinly myelinated Aδ and unmyelinated C fiber are activated by noxious stimulation. Aδ fibers are usually activated by intense mechanical stimulation and C fibers by noxious mechanical, thermal, and chemical stimulation. Aδ and C fibers terminate in the superficial laminae I, II, and deeper lamina V of the spinal cord. Nociceptive inputs can be transmitted from the spinal cord to the dorsal thalamus directly through the spinothalamic tract (STT) or indirectly through the spinoreticular, spinomesencephalic, or even mediolemniscal (ML) pathways to the thalamus.9An early study showed that spinal cord lesion, including dorsal and ventral horns, resulted in extensive atrophy in the thalamus.10 The affected thalamic nuclei included the following: ventral posterior nuclei (VP); posterior nuclei (PO), such as medial/triangular nuclei (POm/PoT); intralaminar nuclei (ILN), such as centrolateral (CL) rostrally and parafascicular nuclei (Pf) caudally; medial dorsal nucleus (MD) and many midline thalamic nuclei, such as submedius nucleus (Sm).
2.1. Medial and lateral thalamic pain pathways
Based on anatomical, electrophysiological, cognitive, and clinical evidence, the above somatosensory-related thalamic structures can be broadly divided into lateral and medial subdivisions. In the lateral thalamic pain pathway are the VP and PO, and in the medial pathway are the ILN, MD, and midline nuclei.
STT has two components: the lateral STT ascends in the lateral column of the spinal cord and the medial STT in the ventral column. However, both medial and lateral STT axons terminate into the medial and lateral thalamic nuclei. Approximately 15% of STT axons bifurcate and terminate into both parts of the thalamus. The anatomical difference is more clearly defined by the cell of origin in the spinal cord laminae. Tracer studies have shown that laminar I–II neurons in the spinal cord project to VP, POm, PoT, MD, and Sm.11, 12 By contrast, the deep lamina V projects primarily to the CL and moderately to the PoT.12 In primates, laminar 1 projects to the posterior part of the ventral medial thalamic nucleus (VMPo), the primate homologous of the Sm,13 which is specifically activated by noxious and thermal stimuli.
The VP is unique in that the somatotopically organized fine-grain tactile inputs from the ML terminate specifically here. An important question left unsolved is how and why somatotopically organized nociceptive and tactile inputs converge in the VP. One possibility is that the VP may be subdivided into the core, the caudal part, and the border part. These subdivisions have differential connections and can thus serve different functions.14 Another possibility is that STT terminals cluster in specific parts of the dense ML terminal sheet in the core of the VP.15 These STT terminal clusters give precise information about the location of the painful stimulus.
Less fine-grained, yet still roughly somatotopically organized, extralemniscal, ascending, tactile inputs go to the PO. Medially located thalamic nuclei in the medial pain pathway receive mostly multisynaptic tactile inputs. Topographic information is not clear in the medial thalamic pathway. In addition, medial and lateral thalamic nuclei are distinct even in their input pattern from other subcortical structures.16
Rodents have three subdivisions in the MD, named according to their location: the medial (MDm), the central (MDc), and the lateral part of MD (MDl). The MDm receives afferents from basolateral and central amygdala,17, 18 and from the ventral pallidum of the basal forebrain.18 The MDc receives afferent inputs from basolateral and central amygdala, magnocellular preoptic, lateral preoptic area, and the diagonal band of the basal forebrain.18The MDl receives inputs from the substantia nigra reticulata and the lateral preoptic area.18 The entire MD receives inputs from the lateral hypothalamus, the ventral tegmental area, the dorsal tegmental gray, and the substantia innominata.18 In rodents and cats, ILN receive afferents from midbrain periaqueductal gray (PAG) and adjacent reticular formation, substantia nigra, superior colliculus, and anterior pretectal nucleus. ILN receive sparser afferents from the dorsal lateral parabrachial nucleus and pontomedullary reticular formation.16, 19 The VP also receive inputs from the midbrain PAG and adjacent reticular formation but far fewer compared to the medial thalamus.16
2.2. Thalamocortical interconnections
The most distinctive feature of the thalamus is its interconnection with the cerebral cortex. Not only do the three major dorsal thalamic divisions, namely ventral, anterior, and medial thalamus, have specific reciprocal thalamocortical interconnections, but it is now known that ILN and midline thalamic nuclei also have target-specific connections with the cortex.20, 21The major cell type for all dorsal thalamic nuclei is the thalamocortical neuron, a multipolar neuron that sends axon projections to the cerebral cortex. The thalamocortical axon does not have local collateral in the dorsal thalamus. Because no intrathalamic cross-talk occurs between the nuclei of the dorsal thalamus, each thalamocortical pair can be considered a relatively independent operation unit.
In the somatosensory thalamus, VP connects with S1 and S2 reciprocally and topographically. POm also connects reciprocally with S1 and S2, but the topography is not well organized.22 Using juxtacellular recording, Gauriau and Bernard23 found that nociceptive-specific PoT neurons projected to S2, and nociceptive nonspecific and tactile PoT neurons projected to the posterior part of the insular cortex (PIC).
MD and ILN have distinct cortical targets. These cortical structures are all closely related to pain function. MD topographically projects to the prefrontal cortex.24 In primates, the dorsomedial part of the MD projects to the orbitofrontal cortex. The ventromedial part projects to the insular cortex. The dorsocentral part projects to the medial prefrontal cortex (cingulate, prelimbic, and infralimbic cortices). And the lateral part of the MD projects to the lateral prefrontal cortex. The cortical projections of each MD subnucleus in rats have been investigated using anterograde tracers.25 MD thalamic nuclei directly project to the medial prefrontal cortex26, 27 and the insular cortex.28The MDm projects to the prelimbic cortex and the dorsal agranular insular cortex. The MDc projects to the ventral agranular insular cortex and the lateral orbital cortex. The MDl projects to the anterior cingulate cortex (ACC).
The paracentral (PC) and CL nuclei of ILN project to the lateral cortex, for example, the motor cortices.20, 27 ILN have minor projections to the cingulate cortex in the monkey.29 All cortical targets of MD and ILN of rodents, such as cingulate, limbic, orbital, and insular cortex, are elements of the pain matrix of both humans30, 31, 32 and rodents.33 The Sm has been studied in the rat and cat, and was shown to have specific reciprocal connection with the ventrolateral orbital cortex (VLOC).34, 35
In summary, based on thalamocortical connections, several thalamic and cerebral cortical pairs can be identified in rodents. These thalamocortical pairs can be used in the future as potential research targets for their pain functions. These pairs include at least the following: VP-S1, MDl-ACC, PoT-PIC, and Sm-VLOC.
3. Acute pain
Pain is defined by the International Association for the Study of Pain as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage”. Such subjective sensory and emotional experiences are difficult to identify in animal and unconscious human patients. Nociception, the encoding of noxious stimulus in the nociceptive pathways, is usually used in the case of anesthetized individuals.
3.1. Functional imaging studies
The foundation for a medial versus lateral differentiation of pain function has been established with human cognitive functional imaging studies. In one series,30, 36 test participants were hypnotized. Suggestions were made to these participants that either the stimulus was becoming more unpleasant or that the temperature of the water used for stimulation was hotter. The ACC in the medial pain pathway showed a significant correlation to “unpleasantness,” whereas S1 in the lateral pathway correlated specifically to the sensory dimension of pain. In another series of studies, Craig et al37, 38 use a thermal grill illusion to evoke imagery pain. They demonstrated using positron emission tomography (PET) that illusory pain—and therefore, the affect aspect of pain—is represented in the ACC and insular cortex of the test participants.
The thalamus is one of the areas most consistently activated by painful stimuli in human imaging studies. One meta-analysis study reviewed the brain areas activated by experimental pain and chronic pain and found that the thalamus is better activated by noxious thermal stimuli than by mechanical or chemical stimuli.39 The most frequent responses are bilateral thalamic responses. Three blood oxygenation level-dependent magnetic resonance imaging (BOLD-MRI) studies revealed an exclusively contralateral thalamic response that compared painful heat (45°C or 46°C) responses to warm (32°C, 35°C, or 42°C) responses.40, 41, 42 One study using perfusion MRI revealed exclusively ipsilateral thalamus when comparing 48–53°C and 35°C responses.43 Because of spatial resolution limits, few studies could precisely pinpoint the specific thalamic nuclei involved. More recent studies with better resolution power also place the classical view of a lateral thalamus responsible for sensory-discriminative and a medial thalamus responsible for the affective, motivational component of pain in doubt. In one study, the bilateral ventrolateral thalamic nucleus, instead of medial thalamic nuclei, showed increased BOLD signals in sad mood-related pain threshold reducing.42 In addition, studies indicate that the “pain matrix” can also be activated by innocuous but salient visual or auditory stimuli.44
BOLD-MRI and other functional MRI (fMRI) methods cannot be adopted easily in animal studies because the tested individuals must not move in the scanning chamber. When anesthetized, pain function is lost, and only the most robust nociceptive processing remains. Manganese-enhanced MRI (MEMRI) is an activity-dependent, trans-synaptic tracing technique. MEMRI has been used in the medial thalamocortical pathway to evaluate its nociceptive function.45 In this study, we show that the MD–ACC pathway is activated by painful peripheral stimulation. The biggest challenge of animal fMRI is the confounding effects of anesthesia.46 Although ways are found to perform fMRI on the conscious animal,47 this is still ethically and practically difficult for pain-related studies. In this regard, F–18 fluorodeoxyglucose PET imaging reveals the cumulative metabolic activity of brain cells. During the tracer uptake period, the animal can remain conscious and, thus, researchers can avoid the interference of anesthesia.48 MEMRI can also be performed on conscious behaving animals if manganese ions are delivered through a chronically implanted cannula.
3.2. Electrophysiological recording of thalamic activity
The notion that the thalamus is responsible for the sense of pain can be dated back to more than 100 years ago.1 An important line of evidence is that the thalamus is the major source of “nociceptive neurons” at the highest level of the nervous system in anesthetized animals, including primates, cats, and rats, in a very long series of studies. “Nociceptive neurons”, that is, noxious-stimulation-responsive neurons recorded under anesthetized conditions, have been found in the VP, MD, CL, Pf, and RT by our studies49, 50, 51, 52 and those of other groups (for a review, see Willis53); and in PoT,23 Sm,54 and ZI51by many other research groups.
The percentage of heat-responsive neurons under anesthesia is 20% in the raccoon VP55 and 32% in cat PO.56 Approximately 70–75% of mechanical nociceptive neurons are heat-responsive in rats57 and monkeys.58 Most lateral thalamic neurons can code temperatures from 43°C to 53°C,55, 56, 57 but a smaller part of these VP neurons have shown plateau responses from 50°C to 53°C.57, 58 The medial thalamic response to noxious heat under anesthesia is rarely reported because of a low incidence of responsive neurons. Only 14% of mechanical nociceptive neurons are heat-responsive in the posterior ILN region of the anesthetized rat.59 The report stated that “none (of them) exhibited a clear relationship between discharge and stimulus temperature”. Hence, the stimulus–response (S–R) function of anesthetized animals clearly indicates a specific discriminative role of the lateral thalamic neuron.
The S-R function of VP neurons has been reexamined in wakeful monkeys. The incidence of heat-responsive neurons increased to 77%. The S-R function of VP neurons linearly coded the surface temperature of the face from 47°C to 49°C.60 The same research group also evaluated the S-R function of the medial thalamus. Although the incidence of mechanical nociceptive neurons was still low (7 of 52, 13.4%), 100% of noxious mechanical-responsive neurons were heat-responsive. Strikingly, medial thalamic neurons can encode temperatures from 46°C to 49°C.61 However, the response sensitivity (firing rate per °C) of medial thalamic neurons (100% increment of 49°C compared to 46°C) was still lower than that of lateral thalamic neurons (300% increment of 49°C compared to 46°C). One important difference between medial and lateral thalamus highlighted by these experiments is that the S-R function of the VP was not affected by the attention shift of the monkey from thermal changes to visual cues. By contrast, the medial thalamic response was significantly reduced by visual cues and ketamine anesthesia.
The magnitude of nociceptive coding ability between medial and lateral pain pathways have been simultaneously recorded from S1 and ACC of free-moving rats.62 The data showed that both cortices had neurons whose activity was changed by laser intensity, but the percentage is lower in the ACC (26%) compared to that in the S1 (69%). The coding ability of S1, especially by the short-latency response, is stronger than that of the ACC.63 Other studies simultaneously recorded ACC, S1, MD, and VP neurons in free-moving rats.64Surprisingly, the S-R function presented as normalized firing rate is only slightly better for neurons in the lateral pain pathway than those in the medial pain pathway in conscious behaving animals.
Nociceptive neurons can be recorded in human patients. Furthermore, the responsiveness of the human thalamic neurons can be measured while the patient is conscious and communicating. Human nociceptive neurons have been recorded in VP, VMPo,65, 66, 67 and ILN, including the central medial nucleus (CM).68 No quantitative assessment has been applied to human thalamic neurons because of ethical limitations.
Although evidence can be provided to show that the discriminative ability of nociceptive neurons are stronger for VP neurons than for medial thalamic neurons, it is not easy to design a decisive experiment to test the affect function of the lateral versus the medial pain system at the cellular level. We have attempted this in a previous study.63 In this study, graded laser heat stimulation applied to the tail was used to construct a quantitative S-R function of S1 and ACC neurons. The fear-potentiated startle task was used to sample the coding of the fear affect by the ACC and S1 neurons recorded in the rat brain. We found a significantly superior coding of the S-R function by the S1 neuron and, in contrast, better coding of the fear affect by the ACC neuron.
Another important functional property of the thalamic neuron is its firing pattern. Thalamic neurons have two firing modes: tonic and burst.69Thalamocortical neurons switch between the two firing modes depending on their resting excitability. Under anesthesia or sleeping conditions, thalamic neurons are depressed and fire mostly in bursts. Their firing rate increases and becomes tonically active when the person or the animal is awake and attentive. Suppression of thalamic burst by genetic or pharmacological means changes the nociceptive function of the animal.70 In a recent paper, we found that calcium channel Cav3.2 knockout mice had a deficit burst pattern in RT and VP thalamic nuclei.71 These mice show enhanced nocifensive behaviors.70
3.3. Experimental lesion of thalamus
Accepting that both medial and lateral thalamus are critical relays in pain processing, it is reasonable to hypothesize that bilateral lesion of either the medial or lateral thalamus of normal animals should change pain behaviors. Rats with bilateral N-methyl-d-aspartic acid-produced medial thalamic lesions showed a reduced duration of writhing behaviors after an intraperitoneal injection of acetic acid.72 However, another study of rats with bilateral electrolytic medial thalamic lesions showed no changes in the paw lifting and licking behavior in the formalin test.73 By contrast, unilateral thalamic lesions resulted in increased pain behaviors.74 Electrolytic lesions on either the unilateral medial or the lateral thalamus decreased withdrawal latency to paw pressure and a 53°C hot plate. Increased pain-related behaviors started 1 week after thalamic lesion and continued for 8 weeks. Both medial and lateral thalamic lesions affected the contralateral paw first and then the ipsilateral paw. It is noteworthy that increased pain behavior of the ipsilateral paw was more permanent in lateral thalamus-lesioned rats. Another difference in medial and lateral lesions is that unilateral lateral thalamic lesion increased the pain scores in the second phase of the formalin test. However, unilateral medial thalamic lesion increased the pain scores in the first phase of the formalin test.74 Kainate lesion on unilateral VP of normal rats decreased von Frey hair threshold in evoking withdrawal behavior and decreased withdrawal latency to radiant heat. The increased pain-related behavior started at 24 hours and lasted up to 48 hours after VP thalamic lesion.75
3.4. Stimulation of the thalamus
Direct deep brain stimulation (DBS) in the VP thalamus from patients without pain typically evoked nonpainful, paraesthetic sensation.76, 77 DBS at the core and posterior inferior region of the VP thalamus can evoke pain sensation without specific topographic distribution.65, 67 However, warm sensation was more frequently evoked in the posterior than in the core region.65 Intensity of sensation was dependent on current and frequency of microstimulation. By using 300 Hz, 5 μA, microstimulation at where nociceptive cells were recorded can evoke pain sensation, and microstimulation at where low threshold cells were recorded can evoke warmth. The S-R functions of microstimulation can be either binary or graded. A binary function signals the presence of painful sensation. The graded function reflects the stimulation intensity.78
4. Chronic pain
Acute pain protects us from potential and more severe injuries. Chronic pain, however, is maladaptative. This review will focus on neuropathic pain, that is, pain caused by disease or damage to the somatosensory nervous system. Particular focus will be placed on mononeuropathic peripheral neuropathic pain. As Devor79 indicated, chronic pain following nerve damage is a paradox. Where cutting a telephone line leaves the line dead, damage to a nerve produces many positive symptoms. The main positive symptoms are threefold: spontaneous pain, allodynia, and hyperalgesia. The prevailing hypothesis for these enhanced activities is a sensitization of the central nervous system.80, 81 This prompts the following questions: (1) What is triggering the sensitization, and how? (2) Which brain areas are sensitized, and how do they initiate and maintain chronic pain?
4.1. Imaging studies of neuropathic pain
Patients suffering from chronic spontaneous pain show altered regional cerebral blood (rCBF) flow in the thalamus. PET studies using O-15 showed either increased82 or decreased83, 84 thalamic rCBF during episodes of spontaneous pain. Medial or lateral thalamic nuclei were rarely distinguished in these studies. Compared to PET, fMRI has better spatial resolution and is able to distinguish subnuclei in the thalamus. Through the correlation activity pattern of individual brain areas, one recent fMRI study revealed a decreased thalamocortical connectivity of VP and MD in the resting state of diabetic pain patients.85 The other fMRI showed a reduced connectivity between the thalamus and the insular cortex of fibromyalgia patients at rest.86
Brush-induced allodynia increases rCBF (O-15 PET) in both medial and lateral thalamus, compared to no stimuli or to brushing the normal skin of the same patient.87 Topical capsaicin treatment on normal volunteers can induce thermal allodynia and more unpleasantness than equally intense painful heat stimulation as evaluated using the McGill pain questionnaire. An O-15 PET study showed that warm allodynia induced higher rCBF in the medial thalamus. By contrast, rCBF in the lateral thalamus was not changed.88This study indicated that allodynia-induced unpleasantness was encoded in the medial thalamic pathway. Another O-15 PET study further compared the rCBF of central lesion patients with and without pain and found that heat-induced pain evoked increased intralaminar activity.89
Functional imaging has the advantage of longitudinally monitoring the global brain condition during each phase of the development of neuropathic pain. Unfortunately, until recently, there has been no report of a chronic longitudinal brain imaging study of neuropathic animals. One long-term anatomical study followed spared nerve injury rats before and after surgery, where the decreased frontal cortex volume fitted well to the onset time point of anxiety behavior. Increased hyperalgesia correlated with decreased volume in contralateral S1, ACC, and insular cortex,90 which is consistent with human studies.91, 92 BOLD-MRI showed enhanced barrel cortex signals, and functional tracing by MEMRI showed enhanced VP–S1 projections during electrical stimulation of the whisker pad after unilateral infraorbital nerve resection.93
4.2. Electrophysiological studies on neuropathic pain
Among human patients, the VP thalamus showed greater spontaneous firing in neuropathic pain patients compared to movement disorder patients without pain. In addition, in these chronic pain patients, the thalamic activity showed excessive burst firing.94, 95, 96 The distribution maps of low-threshold neurons and nociceptive neurons changed after stroke or after spinal cord injury. The data implied that plasticity changes occur in the thalamus of chronic pain patients.97
For peripheral neuropathic pain models,98, 99 VP recordings also reveal spontaneous hyperexcitability, evoked hyperactivities, expansion of receptive fields, and bursting firing. However, the above-mentioned studies recorded thalamic activity in brain slices or in anesthetized animals in which activity was supported by incomplete circuits or suppressed by anesthetic drugs. In addition, most studies recorded thalamic activity only during the maintenance phase of the neuropathic pain. Recently, we developed a long-term recording method100 to continuously monitor peripheral ganglion cells and VP activities before and after inferior alveolar nerve transaction. We found that spontaneous hyperactivity first occurs in the injured trigeminal ganglion (TG) neurons. The hyperactivity spread next to adjacent uninjured TG neurons. These ectopic activities might be triggering the ensuing sensitization immediately after the nerve injury. Neuropathic-specific hyperactivity was found several days later in the VP neurons. In the maintenance phase, the receptive field expansion, the modality shift, and long-lasting afterdischarges occurred only in VP but not in TG ganglion neurons.101
Medial thalamic neuronal activity changes have also been studied in rodent neuropathic pain models. Pharmacologically induced allodynia in anesthetized rat showed the receptive field expansion of the MD, CM, reuniens thalamic nucleus, rhomboid thalamic nucleus, Sm, and anteromedial thalamic nucleus. The unit became a low threshold-responsive unit and increased firing during brush stimulation.102 Similar to reports on chronic patients, spontaneous neuronal activity increased after spinal nerve lesion in both ILN103, 104 and MD.105 The issue of the firing pattern of thalamic-projecting neurons influencing perception has recently been tested on behavioral mice using an inflammatory model. The authors found that the nociceptive pain response was positively correlated to tonic firing and negatively correlated to burst firing. In addition, the intraburst interval reliably correlated with nociceptive responses.106
4.3. Experimental thalamic lesion and neuropathic pain
If thalamic sensitization contributes to the maintenance of neuropathic pain, a lesion made to the thalamic nuclei should help alleviate the symptoms. This has been tested by Saade et al.107, 108 Although continuous blocking of the medial thalamic nuclei by minipump lidocaine infusion did alleviate mechanical allodynia, cold allodynia was only partially reversed. In addition, permanent lesion of either the medial thalamic107 or the lateral thalamic108nuclei partially and transiently reversed mechanical allodynia, cold allodynia, and heat hyperalgesia. These results indicate a more complicated forebrain circuitry involvement for neuropathic pain.
4.4. Stimulation of the thalamus in neuropathic pain patients
Microstimulation in VP evoked a higher incidence of pain in central post-stroke pain (CPSP) patients than in nonstroke pain and nonpainful patients.76Microstimulation in VP of amputees evoked phantom sensation, including pain.109 There are no studies describing the effects of medial thalamic stimulation in neuropathic pain patients.
4.5. Thalamic pain syndrome
The CPSP of thalamic origin consists of intensive, ongoing burning pain, mechanical and thermal allodynia, and hyperalgesia. CPSP can be induced by infarction of four main arterial territories (anterior, paramedian, inferolateral, and posterior110) in the thalamus. Medial pain pathway related nuclei (MD, CM, Pf, CL) are irrigated by the paramedian territory and lateral pain pathway-related nuclei (VP, VL) are irrigated by the inferolateral territory. Typically, infarction of the paramedian territory induces consciousness deficit,110impaired learning and memory, aphasia, and altered social skills and personality.111 By contrast, inferolateral territory infarction induces sensory loss110, 111 and CPSP.112 CPSP is not only intensive and drug resistant, it is also difficult to predict for its onset. CPSP can start months or years after a thalamic stroke.113 Hence, it would be clinically valuable if the physician can predict CPSP and treat it in advance. Two studies that compared thalamic infarction patients with and without pain found that infarction containing the lateral, inferior, and posterior portion of VP and anterior pulvinar might be an indicator of CPSP.114, 115
CPSP of thalamic origin, or thalamic pain, is so impressive that Head and Holmes1 designated the thalamus the center of pain sensation. Recent studies indicate that CPSP primarily result from lesions of structures relating to the lateral thalamic pain pathway. It will be important to explore how an imbalance between the medial and lateral pain pathways leads to this severe disorder.
5. Synopsis, hypothesis, and work plan
Somatosensory-related thalamic nuclei in the lateral and the medial pathways have dense, specific, direct reciprocal connections with sensory, limbic, and motor cortices. A substantial portion of thalamic neurons remains responsive to peripheral, deep, and visceral noxious stimuli even under anesthetized condition, indicating robust nociceptiveness. Furthermore, in humans, lesions in these regions cause thalamic pain. Because the thalamus sits in the core of the forebrain, we propose that the thalamus holds the key to pain consciousness, and therefore, the key to spontaneous and evoked pain in chronic pain conditions. In addition, thalamic nuclei should be paired with their matching cortical partners and studied together.
Pain is multidimensional. Mechanisms underlying the various subfunctions of pain are certain to be diverse and complex. The following list is a strategic work plan that could guide future research in identifying the roles of thalamic or thalamocortical structures in pain function (see the following subsections).
5.1. Identifying an objective index for a clear subfunction
Taking spontaneous pain as an example, in the jerky paw-lifting movement of the injured limb of a resting spared nerve injury (SNI) rat,116 the conditioned place preference choice of a previously conditioned rodent117 or the “pain face” of a laboratory animal118, 119 or a human patient are useful indices.
5.2. Demonstrating the necessity of the thalamocortical structure in the expression of the pain index
This criterion is difficult to fulfill for a deep brain structure in a human study. Even in animal models, a localized reversible lesion is difficult to achieve. In deep structures, anatomical MRI combined with histological verification could be used for the documentation of the location and the extent of the areas of a permanent brain lesion. Likewise, an animal PET may be coupled with histological verification for a reversible lesion.
5.3. Longitudinally following the activity change of the thalamocortical structure and correlating with the expression of the pain index
Most nociceptive behaviors and expressions are flickering and rapidly changing. In addition, their occurrence rates during the course of chronic pain development are dynamic. Current functional imaging methodologies for the human brain are limited in their temporal resolution. Furthermore, longitudinal functional scanning of the same individual starting before chronic pain would be impractical or unethical. Direct electrophysiological recording combined with longitudinal repetitive whole brain imaging is the method of choice for the study of animal pain models.101 Different sets of neural structures may be involved for the initiation and for the maintenance of the pain index.120, 121, 122, 123, 124 Therefore, thalamocortical activity changes at either the first-emerging or the well-established phases of the pain index could be used as the benchmark for key structures involved in these two phases, respectively.
5.4. Manipulating the activity of the thalamocortical structure changes the expression of the pain index
Transcranial magnetic stimulation (TMS) can be coupled with functional brain mapping to study the function of surface cortical structures of human individuals.125, 126, 127, 128, 129 In addition, with an increasing use of DBS electrodes to treat various neurological and psychiatric diseases in clinical wards, it is now possible to design human patient studies to manipulate brain activities at many specific sites.130, 131, 132, 133, 134
For animal studies, this type of research has been traditionally performed using electrical135, 136, 137 and chemical138 stimulation. Recently, optogenetic methods have been used to manipulate specific neuronal populations in specific brain areas.139, 140
5.5. Treatments that ameliorate the expression of the pain index lower the thalamocortical activity in a suitable time frame
By using well-established pain treatment methods, such as opiates, it is relatively easy to suppress the expression of the targeted pain index. If enhanced thalamocortical activity is indeed involved in the expression of the pain index, it follows that a thalamocortical activity decrease should parallel the effect of the behavior suppression. Furthermore, changes in brain activity should precede the changes in behavior. When drug or treatment works through a separate pathway to suppress the behavior, a dissociation of the thalamocortical activity and the behavioral changes might occur. Another likely complication is that the thalamocortical structure under study is a part of the representation of fear, stress, alertness, etc., where pain-related responses commonly occur in a behaving human or animal under noxious stimulation. Under this situation, the thalamocortical activity change may lag behind the behavioral change.
Head and Holmes1 designate the thalamus the center of pain primarily based on stroke patients who developed thalamic pain. Their study marked the beginning of the era of the search for localization of functions of forebrain structures. A hundred years later, we are still searching. The above work plan can be viewed as a short guide map for seekers. What is at stake? And what are the prices? The seat of the consciousness of pain intensity is the best objective quantitative measure of our pain. Likewise, the representation of the pain affect in the brain would give us the measure of the suffering of the patients. These objective indices can be used to titrate the effectiveness of treatment and to assess new treatment developments.
Acknowledgments
This work was supported by grants from the National Science Council(NSC100-2311-B-002-002-MY3) and the National Health Research Institute(NHRI-EX102-10104NI), Taiwan.