Representation of Pain in the Brain

Historical Perspective

The role of the brain in pain processing has long remained elusive. More than one hundred years ago, observed that soldiers with extensive injuries to the cerebral cortex still perceived pain and thus concluded that the cortex played only a minimal role in pain perception. reached a similar conclusion when they found that patients rarely reported pain on electrical stimulation of their exposed cerebral cortex during surgery to remove epileptic foci. Although , , and others did not agree with this view and produced clinical evidence that they thought was more consistent with the idea that the cortex is involved in pain perception, until recently the dominant viewpoint remained that the cortex has a minimal role in pain perception. Nevertheless, the complex nature of the pain experience, which encompasses both sensory features and emotional and motivational components ( , ), suggests that the conscious appreciation of pain must include the activation and interaction of multiple brain regions. It is thus not surprising that specific lesions or focal stimulation of the cortex did not produce the experience of pain.

By the late 1980s, multiple lines of evidence suggested that several regions of the cerebral cortex could participate in pain processing. Some patients with epileptic foci involving the primary or secondary somatosensory cortices (S1 and S2, respectively) had been observed to experience painful seizures ( , ). In addition, lesions involving these areas in humans had on occasion been shown to reduce pain perception ( ). A few single neurons responding to noxious skin stimulation had been identified in S1 and S2 of the monkey ( , , ) and in frontal cortices of the rat ( ). Finally, neurosurgeons had observed that lesions involving the anterior cingulate cortex (ACC) reduced the distress associated with chronic intractable pain ( , ). Nevertheless, the paucity of both animal and human evidence of involvement of the cerebral cortex in pain perception led to the continued view in medical textbooks that pain was a subcortical phenomenon.

The advent of modern human brain imaging studies in the early 1990s allowed us to begin unraveling the role of the brain in the complex experience of pain. Hemodynamic correlates of pain were first imaged in the human brain in the 1970s by with the radioisotope xenon 133. This technique provided little spatial resolution but suggested that during pain, blood flow to the frontal lobes was increased. The first three human brain imaging studies of pain using today’s technology were published in the early 1990s by and , who used positron emission tomography (PET), and by , who used single-photon emission computed tomography (SPECT). All three studies used painful cutaneous heat, and despite differences in their results, together they indicated that multiple cortical and subcortical brain areas are activated during short-duration pain induced by heat. Non-invasive human brain imaging continues to provide new insight into the human brain in health and disease at an unprecedented and unabatingly fast pace.

Defining a Pain Network in the Brain

Human brain activity can be imaged with several techniques, including PET, SPECT, functional magnetic resonance imaging (fMRI), electroencephalographic (EEG) dipole source analysis, and magnetoencephalographic (MEG) analysis. Each of these techniques has advantages and disadvantages in terms of spatial and temporal resolution, sensitivity, and cost ( Box 7-1 ). However, all provide measures that we can use as indirect indices of neural activity. Furthermore, despite the many differences among these techniques, the results derived from each are generally congruous.

Hundreds of human brain imaging studies have now examined the cortical and subcortical brain regions involved in acute pain processing in healthy subjects. Although there are many differences in activation patterns across studies, a consistent cortical and subcortical network has emerged that includes sensory, limbic, associative, and motor areas. The regions most commonly activated are the S1, S2, ACC, insular cortex (IC), prefrontal cortex (PFC), thalamus, and cerebellum ( Fig. 7-1 ). Pain-evoked activity in these areas is frequently observed with either PET or fMRI techniques, and the activation in these regions is consistent with anatomical studies that show probable nociceptive connectivity to these regions ( ).

As discussed in Chapter 2 and shown in Figure 7-2 , regions activated by pain in imaging studies receive either direct or indirect nociceptive input. In primates, S1 and S2 receive noxious and innocuous somatosensory input from the somatosensory thalamus ( , , ). The cingulate cortex receives input from medial thalamic nuclei that contain nociceptive neurons, including the nucleus parafascicularis and ventrocaudal part of the nucleus medialis dorsalis, as well as from lateral thalamic regions, including the ventral aspect of the ventroposterior nucleus and the ventroposterior inferior nucleus ( , ). Nociceptive input to the ACC is further suggested by the observations that painful stimuli evoke potentials over the human anterior cingulate gyrus and that single nociceptive neurons are present in the ACC of humans ( , ), monkeys ( ), and rabbits ( ). These data indicate a specific role for parts of ACC in pain processing that is distinct from, though probably related to, the role of the ACC in cognitive processes such as attention ( , ). The IC also receives direct thalamocortical nociceptive input in the primate ( ), and nociceptive activity has been recorded from the human IC ( ). A recent study using a viral anterograde transport technique provided the first direct evidence of the spinothalamic–thalamocortical pathway in monkeys ( ) by documenting that the pathway accesses S2, IC, and ACC more heavily than S1. Moreover, the data provide good evidence that parts of the posterior cingulate cortex receiving spinothalamic–thalamocortical input are cingulate motor areas with direct projections to the primary motor cortex. Thus the posterior cingulate regions often observed in human pain brain imaging studies potentially provide a direct route for controlling motor responses to painful stimuli. The prefrontal cortical regions are activated in a number of imaging studies of acute pain in normal subjects, and recent anatomical evidence points to massive innervation of the superficial layers of at least the dorsolateral PFC through medial thalamic neurons that respond to noxious stimuli throughout the body in rats ( ); thus this thalamocortical pathway provides the substrate for modulating higher cortical processes by nociceptive input.

Numerous subcortical areas have shown activation in human imaging studies. The most common subcortical pain-related activation takes place in the thalamus and cerebellum. Several nuclei in the thalamus receive nociceptive input from the dorsal horn, and the cerebellum also has reciprocal spinal connectivity ( ). The striatum, mainly the caudate putamen, is also often reported to be active in human pain imaging studies, nociceptive neurons have been described in rats and monkeys within this region ( ), and a recent viral anterograde labeling study showed that a specific population of spinal cord lamina 5 neurons project directly to the basal ganglia ( ). Some human pain imaging studies also indicate activity in the nucleus accumbens and amygdala ( , ), and this activity is probably a reflection of nociceptive transmission through spino–parabrachial–amygdala projections ( ). The periaqueductal gray (PAG) has also been observed to be active in human imaging studies of somatic and visceral pain, especially when the brain stem is specifically studied ( ). Thus there is evidence that multiple ascending nociceptive pathways are engaged in signaling the information integrated at the cortical level, in addition to the commonly cited spinothalamic pathways (see Chapter 12 for additional discussion).

Brain Processing of the Multidimensionality of Pain

There is now evidence from brain imaging, lesion, and electrophysiological studies that different cortical regions may be preferentially involved in different aspects of the complex experience of pain. Most evidence suggests that the somatosensory cortices are most important for the perception of sensory features, such as the location and duration of pain, whereas the limbic and paralimbic regions, such as the ACC and IC, are more important for the emotional and motivational aspects of pain. S1 and S2 contain neurons that code the spatial, temporal, and intensive aspects of innocuous and noxious somatosensory stimuli ( , , ), characteristics that could subserve the sensory-discriminative dimension of pain processing. Furthermore, clinical studies of lesions involving S1 and/or S2 show deficits in pain sensations ( , ). Strikingly, Ploner and colleagues observed that a patient who suffered a stroke involving S1 and S2 could not localize or describe the nature of a painful stimulus but instead reported a poorly localized and ill-defined unpleasant feeling when presented with a noxious stimulus. Thus this individual appears to have pain affect without a clear sensory-discriminative component of pain sensation. Recently, reported on two patients with unilateral IC lesions regarding pain perception and brain activity. Surprisingly, both patients had higher ratings of acute thermal pain than did controls and exhibited increased S1 activity in the absence of IC activity, thus implying that disruption of nociceptive input to the IC may be compensated by increased transmission of nociceptive information to S1 and that increased acute thermal pain sensitivity results.

The ACC and IC have long been considered to be components of the limbic (emotional) part of the brain ( , ) and therefore potential candidates for processing the affective–motivational dimension of pain. This idea was substantiated by clinical reports that patients who had undergone cingulotomy showed attenuated emotional responses to pain ( , , ). The results of imaging studies confirm these clinical reports. demonstrated selective modulation of ACC pain-evoked activity after hypnotic suggestion of changes in pain unpleasantness and also showed a significant correlation between ACC activity and subjects’ ratings of pain unpleasantness, thus strongly suggesting involvement of the ACC in the affective dimension of the pain experience. Similarly, used regression analysis to show that pain-evoked activation in the ACC is more related to the affective component of the pain experience, and showed that the affective component of the McGill Pain Questionnaire correlated with μ-opioid receptor activation in the ACC during sustained pain. The most compelling evidence for the role of the ACC in pain behavior comes from manipulating its activity in rats during associative learning. Concomitant activation of the ACC during conditioning produces aversive learning, whereas inhibiting ACC activity seems to block aversive learning ( ). Therefore, at least in rodents, involvement of the ACC seems to be necessary and sufficient for noxious stimuli to produce an aversive learned behavior.

The IC has been implicated in visceral sensory and motor integration, emotional responses, and memory functions and is also consistently activated by painful stimuli. observed a systematic relationship between the intensity of painful heat stimuli and IC activation, and observed a similar correlation between IC activity and the intensity of cold stimuli, thus suggesting that the IC may be involved in coding of noxious and innocuous temperature. Other data indicate that IC activity may be important for pain affect. Insular lesions are sometimes characterized by the condition of pain asymbolia, in which pain sensations appear to be normal but behavioral and physiological responses to the noxious stimulus are inappropriate ( ). Recent studies continue to provide evidence that the IC may be critical and the most specific portion of the cortex in pain perception, with a posterior area encoding nociceptive stimulus properties and a more anterior region related to the subjective experience of pain. The posterior IC is the cortical area that receives the densest spinothalamic–thalamocortical nociceptive input in the cortex ( ). Stimulation of the human IC has been shown to provoke extremely unpleasant painful sensations ( ), and more recent studies indicate that these sensations are topographically organized and seem to be evoked by electrical stimulation of the IC to the exclusion of the rest of the cortex ( ) ( Fig. 7-3 ). Although large unilateral IC lesions may enhance pain perception, focal posterior IC lesions seem to lead to central pain and specific deficits in acute pain perception ( ). Both animal and human studies implicate the IC in autonomic control ( , ). , in integrating these findings, has stated that pain has a specific emotion that reflects the homeostatic behavioral drive and suggests that IC pain-evoked activity is central to this drive. Two separate approaches for studying the role of the IC in human pain have reached the conclusion that the anterior portion of the IC uniquely encodes the subjective magnitude of painful stimuli ( , ). This integrated subjective perception may in fact be the signal dictating, for acute pain, the related behavioral drive.

Prefrontal activity, mainly in the vicinity of area 10, has been observed in a number of imaging studies. However, noted that prefrontal pain-evoked activity does not show the same systematic relationship to perceived pain intensity seen in other regions but instead exhibits the highest activity when a stimulus just becomes painful, with lower activation being associated with higher levels of pain ( Fig. 7-4 ). Similarly, found significantly more prefrontal activity in response to a painful cutaneous stimulus than to a painful visceral stimulus despite the visceral stimulus being perceived as more unpleasant. Thus, the prefrontal pain-evoked activity may be related to the cognitive aspects of pain perception rather than directly to pain sensation or affect. shed further light on the specific role of subregions of the frontal cortex in pain perception. Using capsaicin-evoked thermal allodynia, the authors compared brain activity evoked during capsaicin-produced allodynia and normal heat-induced pain of equal intensity. The contrast showed large activity in the allodynia case that included multiple frontal regions, as well as medial thalamus, nucleus accumbens, and midbrain activity. Network analysis of this activity demonstrated that dorsal frontal and orbital frontal cortical activities were antagonistic to each other, with the dorsal region limiting the activity of the orbital region and the latter acting in concert with other regions. They thus concluded that the orbital frontal–accumbens–medial thalamus network is engaged in affective perception of pain whereas the dorsal frontal cortex acts as a “top-down” controller that modulates pain and therefore limits the extent of suffering.

Another brain region activated in the majority of pain studies is the cerebellum. The cerebellum has been implicated in the control of various functions, including motor, sensory, and cognitive, and according to recent evidence, it is involved in nociceptive activities as well ( , , , ). Several human and animal imaging studies have reported activity in the cerebellum following painful somatic and visceral stimulation ( , , , , , ). Furthermore, studies using electrical and chemical stimulation of the cerebellar cortex show that the cerebellum plays a role in the modulation of both visceral and somatic nociceptive responses ( , ). Pain-evoked cerebellar activation is present in anesthetized humans, who are not consciously aware of the pain ( ), thus further suggesting that such activity may be more important in the regulation of afferent nociceptive activity than in the perception of pain.

How do we Distinguish Location and Quality of Pain?

Neuroimaging studies have examined brain regions activated by many types of painful stimulation, including noxious heat and cold, muscle stimulation using electric shock or hypertonic saline, topical and intradermal capsaicin, colonic distention, rectal distention, gastric distention, esophageal distention, ischemia, cutaneous electric shock, ascorbic acid, and laser heat, as well as an illusion of pain evoked by combinations of innocuous temperatures. Despite the differences in sensation, emotion, and behavioral responses provoked by these different types of pain, individuals can easily identify each as being painful. Thus there appears to be a common construct of “pain” with an underlying network of brain activity in the areas described above. Nevertheless, despite the similarities in pain experiences and similarities in neural activation patterns, each pain experience is unique. Subjects can usually differentiate noxious heat from noxious cold from noxious pressure. This ability to differentiate types of pain is particularly puzzling since there is ubiquitous convergence of information from cutaneous, visceral, and muscle tissue throughout the afferent nociceptive system ( ). The convergence and the similarities in brain regions activated by different types of pain are consistent with phenomena such as referred pain but cannot explain either the ability to identify the origin of the pain or the contrasting behavioral reactions to cutaneous and visceral pain (withdrawal versus quiescence).

There is evidence from single-neuron recordings, MEG, PET, and fMRI that neural activity in the S1 cortex could underlie identification of the locus of cutaneous pain. showed that S1 nociceptive neurons have discrete receptive fields such that different neurons respond to painful stimulation in different skin areas. Correspondingly, EEG, PET, and fMRI studies have shown a topographic organization of nociceptive responses in the S1 cortex similar to the organization of tactile responses (i.e., mediolateral organization of foot, hand, face, and intra-abdominal areas) ( , , , , ), thus suggesting that responses in the S1 cortex may be important for pain localization. Yet it is now apparent that the IC also participates in pain localization ( ). used fMRI to directly compare brain activation produced by esophageal distention and by cutaneous heat on the chest that were matched for pain intensity. They found that the two qualitatively different types of pain produced different primary loci of activation within the IC, S1, motor cortex, and PFC. Such local differences in responses within the “pain network” might subserve our ability to distinguish visceral and cutaneous pain, as well as the differential emotional, autonomic, and motor responses associated with these different sensations.

Temporal Sequence of Cortical Activity during Pain Perception

Most information about the temporal sequence of pain-evoked brain activation comes from EEG or MEG studies. The dual pain sensation elicited by a single brief painful stimulus that is due to the different conduction times in nociceptive A and C fibers (about a 1-second difference) is reflected in two sequential brain activations on EEG and MEG recordings from the S1, S2, and ACC ( , , , , , , , ). EEG mapping studies ( , ), source analysis ( , , ), and intracranial recordings ( , ) show that the earliest pain-induced brain activity originates in the vicinity of S2. In contrast, tactile stimuli activate this region only after processing in S1 ( ). The adjacent dorsal IC is activated slightly but significantly later than the operculum ( ). These observations support the suggestion derived from anatomical studies that the S2 region and adjacent IC are primary receiving areas for nociceptive input to the brain ( , , ). Another study has shown that brief painful stimuli evoke sustained cortical activity corresponding to sustained pain perception. The time courses of activation disclosed that the first pain was particularly related to activation of S1 whereas the second pain was closely related to activation of the ACC. Both sensations were associated with S2 activation. These results are interpreted in view of the different biological functions of the first and second pain. The first pain signals threat and provides precise sensory information for immediate withdrawal, whereas the second pain attracts longer-lasting attention and motivates behavioral responses to limit injury and optimize recovery ( ). An MEG study by differentiated between brain regions involved in cold perception from those involved in painful cold. Cold perception resulted in activity in the posterior IC with a mean peak latency of 190 msec for contralateral and 240 msec for ipsilateral insular activity. Noxious cold stimulation initially activated the IC in the same latency ranges as innocuous cold stimuli. Additionally, activations were found in the contralateral and ipsilateral S2 areas (peak latencies of 304 ± 22.7 and 310.1 ± 19.4 msec, respectively), and more variable activation was found in the cingulate cortex. Neither cold nor painful cold produced detectable activation of S1. The results suggest different processing of cold, painful cold, and touch in the human brain.

Temporal differences in brain activity for pain have also been studied by blood flow techniques. showed that the brain activity pattern for a repetitive thermal painful stimulus is different depending on the past history. When the brain was imaged immediately after the start of the pain, the authors observed very different brain regional responses than when the scan began after the pain had been present for 40 seconds. Perceptually, participants perceived the ongoing pain as more intense and unpleasant than the same stimulus at its onset. A number of brain regions increased in activity between early and late scans (e.g., contralateral S1, bilateral S2, parts of the IC and thalamus). Other brain regions were active only in the early stimulation scans (e.g., perigenual ACC, PFC, and anterior IC). The full significance of these differences remains unclear, although these results undoubtedly imply that the temporal dynamics of brain activity should be taken into consideration to understand the role of the brain circuitry in pain perception. More recently, fMRI has been used to examine the temporal sequence of brain activity for acute thermal pain ( ; see also ), and the authors could chart transfer of brain regional information as noxious stimuli are converted to perceived pain. The analysis showed that parts of the ACC and amygdala responded in a predictive pattern; the thalamus, basal ganglia, part of the IC, and the supplementary motor region were activated next, and these regions better related to the stimulus; and the perception-related part of the IC was activated last, together with the higher cognitive frontal and parietal regions ( Fig. 7-5 ). Thus there seems to be a well-organized temporal sequence of brain activity that transforms noxious stimulus parameters to pain perception, even to pain relief (for more details see also ).

The Brain’s Role in Modulating Pain

In the last 30 years, much attention has been directed toward nociceptive modulation in the spinal cord involving both intrinsic mechanisms and descending control from the brain stem ( Chapter 8 ). However, descending projections from the cerebral cortex feed into these modulatory systems, and additional pain modulation can also take place within the cortex ( Fig. 7-6 ).