Central Neural Mechanisms that Interrelate Sensory and Affective Dimensions of Pain

The “lateral” spinothalamic tract

One principal ascending spinal pathway for pain is the “lateral” spinothalamic tract (Figure 2⇓). It receives synaptic input from primary nociceptive afferent neurons and its neurons of origin are from both lamina I and deep layers V-VI of the dorsal horn (10, 11) . Its main projection is to the ventroposterior lateral (VPL) nucleus of the thalamus and, from there, to the primary and secondary somatosensory cerebral cortical areas (S₁ and S₂ ), that are critical for sensory processing of pain. Thus, this entire pathway can be termed STT-VPL-S_1-2 (spinothalamic tract to ventroposteriolateral nucleus of the thalamus to primary and secondary somatosensory cerebral cortical areas). Two general points can be made about this pathway: 1) its neurons of origin within the dorsal horn contain predominantly wide dynamic range (WDR) neurons and some nociceptive-specific (NS) neurons and these two types of neurons are functionally significant for processing the different dimensions and stages of pain and, 2) this “lateral” STT–VPL–S_1-2 pathway is critical for sensory processing of pain as well as for pain affect, a point that will be supported by neurological and physiological evidence discussed below.

Two types of neurons in STT–VPL–S1-2

It is incontrovertible that nociceptive neurons within this pathway are comprised of wide dynamic range (WDR) neurons that respond differentially over a broad range of stimulus intensity, extending from very gentle to distinctly painful levels of stimulation as well as nociceptive-specific (NS) neurons that respond nearly exclusively to nociceptive stimuli. Both types of neurons receive synaptic input from primary nociceptive afferent neurons innervating cutaneous, visceral, muscular, or other tissues, and originate within the dorsal horn of the spinal cord and medulla (in the case of the trigeminal system) (Figure 2⇑). There is considerable evidence that the large majority of neurons of STT–VPL–S_1-2 are WDR neurons. The large majority (>75%) of spinothalamic and trigeminothalamic tract neurons originating from the dorsal horn and projecting to VPL and VPM are WDR (11–15) . Studies of VPL–VPM nociceptive neurons that are likely to directly receive input from this ascending pathway similarly reflect a strong preponderance of WDR neurons (16–20) . To take just two examples, Kenshalo et al. found that 89% of VPL nociceptive neurons in monkeys were WDR and that the remaining neurons were NS (19) . More recently, Lee et al. found that 87% of human VP neurons that were clearly nociceptive were WDR and the remaining ones were likely NS neurons (20) . Finally, nociceptive neurons from the somatosensory cortex (S₁ ) that are likely to receive input from nociceptive neurons in VPL–VPM are predominantly WDR (21, 22) . Thus, a conservative interpretation of both anatomical and neurophysiological studies of neurons within the STT-VPL-S_1-2 pathway is that they are predominantly (>75%) WDR. The functional meaning of the fact that most nociceptive neurons within STT–VPL–S_1-2 are WDR is based on the capacity of these neurons to precisely encode the intensity of a nociceptive stimulus. Thus, similar to the psychophysical performance of both human and monkeys, WDR neurons within STT–VPL–S_1-2 can respond differentially to a 0.2 to 0.3 °C change in stimulus intensity within a range of painful 45–51 °C skin temperatures (16, 21, 22) . NS neurons have much less of a capacity to encode these small differences (16, 21, 22) . Moreover, the stimulus-response functions of WDR neurons to this range of skin temperatures are precise positively accelerating power functions, ones that strongly resemble power functions that are obtained from human subjects who rated these same temperature stimuli (2, 12–14) . Based on the capacity of WDR to precisely encode stimulus features of nociceptive stimuli, the parallels between WDR neuron responses and human ratings of nociceptive stimuli, and the strong predominance of WDR neurons in STT–VPL–S_1-2 , it is clear that both WDR neurons and STT–VPL–S_1-2 are critical for the sensory dimension of pain (2, 11–22) .

Other ascending pathways

Several other ascending spinal pathways are important for pain (Figure 2⇑). An interesting difference between some of these pathways and STT–VPL–S_1-2 is that unlike the latter, the former often reflect a preponderance of NS neurons. For example, Bernard and colleagues have characterized two novel pathways for pain, termed the spino-parabrachio-amygdaloid and the spino-parabrachio-hypothalamic pathway (23, 24) . The most remarkable and consistent feature of nociceptive neurons within these two pathways is that they consist exclusively of neurons responsive to only nociceptive stimulation—that is, NS neurons, and that this pattern is nearly the reverse of STT–VPL–S_1-2 . Consistent with this unique feature is the finding that spinal cord neurons of origin of these pathways are exclusively within layer I of the dorsal horn—a region dominated by NS neurons (15, 23, 24) . Neurons within these pathways appear to encode nociceptive stimulus intensity with some degree of precision, though somewhat less than that of WDR neurons. These two pathways offer a striking comparison with STT–VPL–S_1-2 because of their very different proportions of WDR and NS neurons and because their central targets appear functionally very different. Whereas STT–VPL–S₁ is likely to be heavily involved in sensory-perceptual aspects of pain, the two pathways to the amygdala and hypothalamus are more likely to be heavily involved in autonomic processes and behaviors related to fear and defense. The central nucleus of the amygdala has been strongly implicated in fear, emotional memory and behavior, and autonomic and somatomotor responses to threatening stimuli (23, 24) . Various hypothalamic nuclei have also been implicated in these functions. Based largely on their central targets, these pathways are very likely to strongly participate in the affective dimension of pain, particularly the initial affect associated with acute pain. There also is a spinohypothalamic pathway that contains neurons like that of STT–VPL–S_1-2 and in fact many STT neurons project to both thalamus and hypothalamus (25) .

To a similar extent, spinothalamic tract pathways to more ventral and medial thalamic nuclei, such as VMpo (ventromedial portion of posterior nuclear complex) and MDvc (ventrocaudal portion of the mediodorsal nucleus) appear to have much greater proportions of NS neurons (26) . A study of nociceptive neurons within these thalamic nuclei found that, unlike neurons in VPM, they are highly modifiable by attentional and motivational state (16) . Based on these differences in response characteristics, VMpo and MDvc neurons may be somewhat more directly involved in affective motivational components of pain, whereas neurons within STT–VPL–S_{1 -2} participate in pain affect as a result of their role in sensory features of pain.

Brain pathways and regions interrelating sensory and affective dimensions of pain

The foregoing descriptions suggest the existence of several functionally different ascending pathways for pain: 1) a STT–VPL–S_1-2 pathway containing mostly WDR neurons important for encoding the capacity to recognize the sensory intensity and sensory qualitative features of pain and, 2) other pathways containing mostly NS neurons important for several aspects of pain, including immediate affective-motivational responses, autonomic and somatomotor activation, and possibly pain sensation. These different pathways are functionally diverse and they are interpreted as contributing to both pain sensory features and pain-related emotions in different ways. These pathways include both series and parallel circuitry (Figure 2⇑).

The serial interconnections are established by the fact that the STT–VPL–S_1-2 pathway is anatomically interconnected with a ventrally directed cortico-limbic somatosensory pathway that integrates somatosensory input with other sensory modalities such as vision and audition and with learning and memory (27) . This pathway proceeds from S₁ –S₂ to posterior parietal cortical areas and to the insular cortex (IC), and from IC to amygdala, the perirhinal cortex, and hippocampus (27) . This cortico-limbic pathway is critical for pain affect because damage to large parts of insular cortex has been found among patients with pain asymbolia (28, 29) . Patients with this condition do not display behavior indicative of threat or intrusion in response to painful stimuli despite their capacity to still appreciate the sensory qualities of painful stimuli. A recent study by Greenspan et al. showed that patients harboring only anterior insular cortical lesions had entirely normal thermal and mechanical pain thresholds yet had longer tolerance times when tested with painful ice-water (30) . Presumably this latter test reflected the affective dimension of pain to a greater extent than sensory discrimination. Similar results have been found for non-human primates. Focal damage to S-2/7b (secondary somatosensory and adjacent infraparietal cortical areas) in the monkey results in an absence of escape responses to painful temperatures despite preservation of the ability to detect the offset of noxious thermal stimuli (31) . An area that receives input from areas S-2/7b is insular cortex (32) , the same area that is associated with pain asymbolia in humans. This serial corticolimbic somatosensory pathway ultimately converges on the same limbic and subcortical structures that may be more directly accessed by other ascending spinal pathways. The latter reflect parallel processing. This dual convergence may be related to a mechanism whereby multiple neural sources, including somatosensory cortices S₁ and S₂ , contribute to pain affect. Therefore, this dual convergence reflects both parallel and serial processing.

Electrophysiological recordings of neurons of infraparietal cortical area S-2/7b neurons in monkeys help explain how this mechanism may be involved in pain affect (33, 34) . Neurons in this region responded to nociceptive stimuli (44–47 °C) and yet were enhanced by antecedent or concurrent visual stimuli (33) . However, this enhancement only occurred if the target location or direction of motion within the visual receptive field was spatially aligned with the cutaneous receptive field. The enhancement was much greater for mild nociceptive stimuli (44–45 °C) than for stronger stimuli (47 °C) (33) . The neural organization of this region of the posterior parietal cortex appears to be that of integrating nociceptive inputs with other sensory inputs in a manner that conveys information about the overall degree of threat presented to an organism. This integration is especially critical at the low end of the nociceptive stimulus range, wherein an organism must make a relevant decision about the extent of threat presented by an object. The processing of pain requires an evaluation of sensation in relationship to its overall context—an evaluation that may help link sensation with affect. This function would require integration of somatosensory input with other sensory modalities and with memory.

The contribution of somatosensory cortices (S1 and possibly S2) to pain sensation and pain affect

This corticolimbic serial mechanism directly implies that somatosensory cortices are involved in pain affect, an implication consistent with the previously described psychophysical studies in which pain unpleasantness was shown to be in series with and the result of pain sensation intensity (2, 6) . This explanation requires that somatosensory cortices directly function in the sensory aspects of pain and it predicts that lesions within STT–VPL–S_1-2 would result in deficits in both pain sensation and affect. The neurological evidence for the involvement of STT–VPL–S_1-2 in sensory dimensions of pain is based on several lines of evidence. First, considerable evidence now exists for the involvement of S₁ in sensory features of pain. Lesions within S₁ produce deficits in ability to detect sensory features of pain in both monkeys (35) and humans (30, 36, 37) . These deficits are sometimes small, particularly when the lesions of S₁ are restricted to only a portion of S₁ (35) . Second, several brain imaging studies have now shown that neural activity in S₁ (measured by regional cerebral blood flow) closely co-varies with temporal and/or intensive aspects of pain sensations (38, 39) . Although not all brain imaging studies of pain show activation within S₁ , negative results are likely the result of technical difficulties in neuroimaging activity in this portion of parietal cortex (40) . Third, as described above, nociceptive neurons definitely exist within S₁ in several mammalian species, including primates (2, 21, 22) and their response characteristics parallel psychophysical characteristics of pain in human and non-human primates (2, 21, 41, 42) . Thus, the involvement of somatosensory cortices in sensory-discriminative aspects of pain, a requirement for the proposed corticolimbic serial mechanism, stands on firm ground.

The involvement of STT–VPL–S_1-2 in the affective dimension of pain is less well established, mainly because classical models of pain processing simply assume that sensory and affective dimensions of pain are processed by completely parallel neural systems (1) . Nevertheless, several neurological studies support a role for STT–VPL–S_1-2 in the affective dimension of pain. Perhaps the most recent and noteworthy study is that of Ploner et al. (43) . Their study involved a patient with an extensive lesion of the right S₁ and S₂ somatosensory cortex. The patient had a normal laser-evoked pain threshold on his right hand that was contralateral to his left intact cerebral hemisphere. However, neither pain sensation nor pain unpleasantness could be evoked from stimulation of the left hand (contralateral to his damaged somatosensory cortex) until the stimulus level was well above that of pain threshold of his right hand. Thus, the deficit included both sensory and affective dimensions. Interestingly, the only response obtained at the higher stimulus level was a vague feeling of unpleasantness in his left shoulder and arm, a feeling that was devoid of the usual sensory qualities of pain (43) . Other studies contain results that are similar in principle to those found by Ploner et al. Head and Holmes found patients with lesions of the lateral somatosensory thalamus (one of these was verified histologically to be within VPL) to have deficits in pain appreciation, including unpleasantness, throughout the low to moderate range of nociceptive stimulus intensities (44) . Like Ploner’s patient, they developed vague feelings of unpleasantness or “pain-related feelings” when the stimulus intensity was raised to levels well above normal pain threshold. The interruption of the pathway to the VPL or the S₁ –S₂ cortex produces deficits in appreciation of sensory qualities and intensities associated with pain and pain-related unpleasantness throughout the lower to moderate levels of nociceptive stimulation. When the stimulus intensity reaches a high enough level, other pathways contribute in a parallel manner to unpleasantness, often combined with the relative absence of the ability to recognize sensory features normally present in pain experience. Using direct scaling methods, Greenspan et al. found that a patient with damage to her lateral thalamus had distinct deficits in both pain sensation and pain unpleasantness throughout the low to moderate range of nociceptive stimulus intensities (12) .

Both pain sensations and pain affect are disrupted by STT–VPL–S_1-2 lesions because this pathway makes serial interconnections to cortico-limbic structures involved in pain-related affect (27, 32) . This anatomical connection is consistent with psychological studies that show serial interactions between pain sensation intensity and pain unpleasantness (2, 6) . Thus, a major contribution to pain affect is provided from STT–VPL–S_1-2 . This pathway projects to brain areas that integrate nociceptive input with contextual information and memory to provide cognitive mediation of pain affect. The involvement of STT–VPL–S_1-2 in both sensory and affective dimensions of pain is in direct opposition to a classic parallel processing view of the spinothalamic tract.

The pivotal role of the anterior cingulate cortex (ACC) in pain affect

As illustrated in Figure 2⇑, the anterior cingulate cortex (ACC) receives anatomical projections from several sources, including the insular cortex. ACC, in turn, is part of the brain’s attentional and motivational network because it projects to prefrontal cortical areas involved in executive functions and to the supplementary motor area involved in response selection (46–48) . Parts of the ACC are consistently activated during pain and in fact the ACC is the most consistent brain region activated in brain imaging studies of pain (2, 49) . Brain imaging studies show that ACC is more directly involved in pain affect than in appreciating the sensory qualities of pain. A positron emission tomography (PET) study used hypnotic suggestions to selectively increase or decrease unpleasantness ratings of experimental pain (50) . Significant pain-related activation occurred in the somatosensory area I, (S1), ACC area 24, and IC during control conditions. Hypnotic suggestions resulted in much larger unpleasantness ratings during the “high unpleasantness” condition as compared to the “low unpleasantness” condition, with no differences in pain sensation ratings. Consistent with these changes, activity in the posterior sector of ACC area 24 was much greater in the high as compared to the low unpleasantness condition yet no differences occurred for S₁ (Figure 3⇓). A separate regression analysis, controlling for factors such as pain sensation intensity ratings, showed that pain unpleasantness ratings were significantly associated with activity in the posterior sector of ACC area 24 (R = 0.55, p < 0.001). A subsequent study that used hypnosis to modify pain sensation intensity showed corresponding changes in S₁ cortex (51) .

A second PET study confirmed the results of Rainville et al. (50) by using a different experimental approach to induce selective variation in pain unpleasantness (52) . Using noxious heat and four successive experimental pain trials, Tolle et al. found that lower ratios of pain unpleasantness to pain sensation intensity were evoked on the first two trials and higher ratios on the second (52) . Regression analysis showed that although several brain structures were activated during pain, only pain unpleasantness was encoded in ACC area 24 (52) . Both studies suggest that ACC, rather than the somatosensory cortical areas, may be more proximal to the production of pain affect (50, 52) ; however, the latter also contribute to pain-affect by virtue of their interconnections to IC and, thus, to ACC.

The proximal causal role of ACC in the affective component of pain is supported by an experiment in which neurons of the rostral ACC were destroyed in rodents (53) . Using the hind-paw formalin model, Johansen et al. measured a learned-behavior place-conditioning paradigm that directly reflects the affective component of pain as well as formalin-induced nociceptive behaviors (i.e., paw lifting, licking, and flinching) that reflect the intensity and localization of the nociceptive stimulus (53) . Destruction of neurons within the rostral but not caudal ACC reduced formalin-induced conditioned place avoidance without reducing acute pain-related behaviors. Their results provide evidence that neurons in the ACC are necessary for some aspects of the affective dimension of pain, particularly those related to behavioral responses associated with avoiding or escaping the nociceptive stimulus.

The functional role of ACC and related limbic structures could be further clarified by consideration of how they fit into the general circuitry associated with pain processing. The general ways in which multiple ascending pathways and brain regions could participate in the various components of pain affect are shown in Figure 2⇑. Nociceptive pathways originating from the spinal cord dorsal horn directly activate brain structures involved in rudimentary aspects of autonomic system activation, escape, motoric orientation, arousal, and fear (2, 11–22, 49) . These include medullary and midbrain reticular formation nuclei, deep layers of the superior colliculus, central grey, amygdala, hypothalamus, and specific medial thalamic nuclei. Activation of these structures likely occurs during the early phase of pain, wherein fear, defensive behavior, and autonomic responses would take place. These responses would occur somewhat automatically and involve a minimum amount of cognition.

Posterior parietal cortical areas that integrate somatosensory input with other sensory modalities and with learning and memory project to the same cortical and subcortical limbic structures (ACC, IC, and amygdala) that receive direct input from spinal pain pathways (2, 32, 47, 48) . Convergence at the level of ACC would be consistent with a mechanism in which somatic perceptual and cognitive features of pain would be integrated with attentional and rudimentary emotion mechanisms. Based on neurological evidence (46) , the ACC may have a complex pivotal role in interrelating attentional and evaluative functions with that of establishing emotional valence and response priorities. Response priorities would be closely related to pre-motor functions that are integrally related to motivation and emotions and may be associated with immediate efforts to cope with, escape, or avoid the pain and pain-evoking situation. In this view, cortical areas controlling sensory, attentional, pre-motor, and affective functions of pain are largely in series, an interpretation supported by both psychological (2, 6) and brain-imaging studies (50–52) described above.

Response priorities change over an extended period of time. Pain unpleasantness endured over time engages prefrontal cortical areas involved in reflection and rumination over the future implications of a persistent pain condition. The ACC may serve this function by coordinating somatosensory features of pain with prefrontal cerebral mechanisms involved in attaching significance and long term implications to pain, a function associated with secondary-pain affect. Thus, ACC may be a region that coordinates inputs from parietal areas involved in perception of bodily threat with frontal cortical areas involved in plans and response priorities for pain-related behavior. Both functions would help explain observations on patients with prefrontal lobotomy and patients with pain asymbolia as described above. The former have deficits in spontaneous concern or rumination about their pain but can experience the immediate threat of pain once it is brought to their attention (37, 54) . In contrast, asymbolia patients appear incapable of perceiving the threat of nociceptive stimuli under any circumstances (28, 29, 55) .