Match and Mismatch: Identifying the Neuronal Determinants of Pain

Despite the increased intensity and sophistication of research on pain mechanisms in the past three decades, serious uncertainties remain about the neuronal origin of pain, especially in painful clinical conditions. The identification of nociceptive afferent fibers in experimental animals [1, 2] has allowed comparison of the stimulus-response functions of these fibers in anesthetized animals, including primates, with psychophysical measurements of pain in humans. More recently, it has become possible to record the discharge frequency of similar fibers in humans and to directly correlate this activity with the perceived intensity of pain [3, 4]. Thus far, efforts have been concentrated on experimentally induced cutaneous pain. The problem of determining the relation between afferent activity and pain is more difficult in the case of subcutaneous, deep, and visceral structures because of technical problems, uncertainties about the critical variables for the noxious stimulus, and a lack of knowledge about neurophysiologic and psychophysical stimulus-response relations in humans and animals. Fortunately, progress is being made [5-8], but much remains to be learned. Three articles in this issue [9-11] address important clinical problems in these areas.

Overall, the results have shown that under controlled experimental conditions, a positive correlation can be seen between nociceptive afferent activity and the perception of pain in normal humans. However, the exact form of the stimulus-response function is not known. Important mismatches between the activity of single cutaneous afferents and pain perception point to the critical importance of central nervous system processes, such as temporal and spatial summation of afferent activity, as determinants of pain [12, 13]. In addition, concurrent activity in other nociceptive and non-nociceptive fibers can strongly modulate the perceived intensity of pain in normal persons, indicating the importance of the interacting convergence of afferent activity onto neurons in the spinal cord and at supraspinal sites [14-16]. Finally, the modulating influences on spinal nociceptive transmission of impulses descending from multiple supraspinal sites, from the brainstem through the cerebral cortex, has been well established [17-21]. Similar processes modulate the transmission of nociceptive and other somatosensory information at the brainstem, thalamic, and cortical levels [21]. Thus, multiple peripheral, segmental, and supraspinal neuronal activities control nociceptive processing at all levels of the neuraxis. It is not surprising that disease affecting the nervous system at any of these levels can alter the perception of pain in either direction, producing analgesia, hypoalgesia, hyperalgesia, or hyperpathia.

The modulation of pain is not limited to changes in perceived stimulus intensity. Almost all pain researchers now recognize that the pain experience is multidimensional, composed of a sensory-discriminative component (represented by the ability to identify the stimulus within spatiotemporal and intensive domains) and a hedonic component (through which the aversive and unpleasant qualities of pain are experienced). In addition, a cognitive component reflects the ability to evaluate the significance of the pain in terms of its meaning for survival and threat to well-being [22]. It is generally recognized that each of these three components of pain is mediated by anatomically and functionally distinct, but interacting, neuronal systems that are distributed among several forebrain structures.

Previous lesion and stimulation studies in humans and animals gave some clues about which groups of cerebral structures mediate the different components of pain [23]. These observations have since been supplemented by recordings of the activity of single neurons in animals and occasionally in humans [24]. Recently, positron emission tomography (PET) with ¹⁵O as a radioactive tracer for regional cerebral blood flow has been used to identify cerebral regions with increased neuronal (synaptic) activity during experimentally induced or neuropathic pain in humans [25-29]. The combined evidence from these observations supports the concept that pain is mediated by a constellation of cerebral structures and that certain groups of structures within that constellation mediate the different components of pain. Specifically, the sensorimotor cortex and the contralateral ventral posterior thalamus appear to be critical for mediating the discriminative component of pain, whereas premotor areas and regions with stronger functional connections with the limbic system may provide the basis for the hedonic features of pain, such as motivation, affect, and the autonomic and endocrine responses that are associated with emotional states. The activation of prefrontal cortical areas, as seen in PET activation studies, may adumbrate the development of longer-term, pain-related cognitive functions. Currently, much of this conceptualization is speculative. Much remains to be learned before we can more specifically assign functions to most of the regions that are just now beginning to be studied in conscious, behaving animals and humans. Nonetheless, sufficient information already suggests that anatomically selective differences in the activity of some cerebral structures, perhaps caused by acquired or developmental conditions, may have selective effects on one or more components of the pain experience. Our knowledge is too limited to allow us to be much more specific than this most of the time.

The three studies in this issue [9-11] address contrasting aspects of the problem of identifying the neuronal determinants of pain. Rao and colleagues [9] and Frobert and coworkers [10] each address the problem of determining the cause of angina-like chest pain when no evidence suggests cardiac ischemia. In this case, the mismatch is the presence of pain with no obvious neuronal source. Rao and colleagues [9] suggest that the solution to the mismatch may be at the afferent end, where an abnormal esophagus generates nociceptive afferent discharges that undergo central processing similar to that of myocardial afferents. Frobert and colleagues [10], however, found no relation between measurements indicating esophageal dysfunction and the presence of angina-like chest pain in patients with normal coronary angiograms. Lacking evidence for a cardiac or esophageal source of nociceptor activation, the authors suggest that the problem may lie centrally; they refer to an earlier study that suggests abnormal central nociceptive processing in patients with similar unexplained chest pains.

Rosen and associates [11] address a contrasting mismatch: the absence of pain in the presence of an apparent source of nociceptive afferent discharge. Building on the evidence obtained in a previous PET activation study of cardiac pain [30], these investigators present data supporting the hypothesis that a central nervous system abnormality may be the source of the mismatch. These investigators used PET activation methods to directly compare the cerebral responses of patients with coronary artery disease during painful induced myocardial ischemia with the responses of similar patients who did not have pain during ischemia. Neither group had diabetes or any other clinically obvious systemic or neurologic abnormality. The major relevant finding in Rosen and colleagues' study was that although the thalamus was activated during (and, to a lesser degree, even after) the ischemic period in both groups, the group that did not have pain showed significantly less activation (smaller increases in regional cerebral blood flow) of mesial frontal and temporal cortical regions that have strong limbic system connections than patients who did have pain. The authors argue that because ischemia activated the thalamus in both groups of patients, the mismatch was probably caused by differences in central nervous system processing of ischemia-induced afferent activity rather than by differences in nociceptive afferent activation. The presence of thalamic activity accompanied by reduced cortical activation in the pain-free group led Rosen and colleagues to suggest that certain cortical activations are necessary for pain and that the critical central function involves the gating of cortically directed activity by the thalamus.

Clinicians are faced daily with the first mismatch: pain without an obvious cause or source. A prudent and often fruitful approach is to search for the site of nociceptive activation. If examination shows evidence of peripheral neuropathy, neuropathic pain syndromes must be considered. A more common problem, however, is that we still know so little about the adequate stimulus for nociceptors other than those innervating ectodermally derived tissues. This shortcoming should yield further basic and clinical research, such as that reported in this issue. The task is difficult, but the combined information derived from clinical, neurophysiologic, and psychophysical observations could help identify with more certainty the afferent sources of some painful clinical syndromes that present difficult diagnostic problems. The clinical benefits will include more accurate, efficient, and cost-effective diagnosis and treatment.

The failure to identify an afferent nociceptive source does not, of course, exclude the existence of one. Nonetheless, at some point after a thorough and fruitless search for peripheral nociceptive activity, the central nervous system becomes the focus of attention. If evidence suggests central nervous system disease, clinicians must consider the central pain syndromes, which follow neurologic disease or injury more commonly than has been previously acknowledged [31]. If such evidence is lacking, the clinician is faced with the possibility that the problem is, for lack of a better term, psychological or psychiatric in origin. The goal should be to establish a positive, rather than a default, psychological diagnosis. Establishing such a diagnosis is often difficult, requiring the time and expertise of specialists, but the effort should be made. Failing all this, are we left with the possibility, as Frobert and colleagues suggest, that the spectrum of neurologic disorders includes selective visceral hyperalgesia of central origin? Establishing this clinical entity will certainly be daunting, but we should be ready to openly and critically review the developing evidence.

The second mismatch is less frequently encountered, but, especially in the case of silent myocardial ischemia, it presents a serious, life-threatening problem for many patients. Here again, if evidence on examination suggests neurologic disease, peripheral or central, or if a psychological or psychiatric disorder is apparent, the difficulty of diagnosis may be somewhat alleviated. Otherwise, one is left with the possibility of a selective visceral hypoalgesia or analgesia of central origin, again without other evidence for neurologic abnormality. As extraordinary as they seem, the striking observations of Rosen and colleagues [11] must be considered as possibly the first direct evidence for such a condition. There is, after all, every reason to consider that the thalamus and the adjacent thalamic reticular nucleus exert potent controls over the flow of somatosensory information to the cerebral cortex [32]. Excitatory thalamocortical and corticothalamic projection neurons send collateral branches to the thalamic reticular nucleus, which forms a shell of inhibitory neurons around the thalamus. Inhibitory neurons of the thalamic reticular nucleus project back to the thalamus, providing an inhibitory control over the cortical transmission of somatosensory impulses. Salt and coworkers [33, 34] have presented evidence for widespread inhibitory controls of somatosensory processing in the thalamus. Furthermore, although the somatosensory thalamus has long been regarded as operating almost exclusively on somatic inputs, Bruggemann and associates [35] have recently presented neurophysiologic evidence for substantial visceral inputs converging onto somatically responsive thalamic neurons. Taken together, these findings support the hypothesized thalamic gating of both visceral and somatic input to the cerebral cortex.

If selective visceral hyperalgesia and hypoalgesia are present without other evidence of neurologic abnormality, several issues become salient. Are these conditions acquired or genetically determined? Which of these conditions, if either, is abnormal? What efforts should be initiated to identify them? How will the identification proceed, given the need to reasonably control the costs of health care? These and related issues will become important topics for discussion and debate as we gain new information about pain mechanisms.

• Copyright ©2004 by the American College of Physicians