It is nearly a century since the publication of Sherrington's influential work 'The integrative action of the nervous system'. Sherrington established a framework of ideas about the organisation and operation of the nervous system which have dominated thinking for much of this century. In relation to pain sensibility, it was Sherrington who proposed the existence of nociceptors, afferent neurones that would detect tissue damage. This idea arose from Sherrington's observations on the reflex responses triggered by the application of strong stimuli to the skin. Indeed, Sherrington (1900) wrote that 'Pain appears the psychical adjunct to protective reflexes'. Several decades after proposing the existence of nociceptors, direct experimental evidence for large numbers of such receptors in skin was obtained. However, it is difficult to apply Sherrington's concept to deep tissues such as viscera, since many forms of tissue injury (such as neoplastic destruction of the solid organs) are often not painful in man, as discussed more fully below. Additionally, some stimuli which are painful (such as distention of the hollow organs) do not damage, or even threaten to damage, the tissue. It is also not always clear what might constitute an adequate noxious stimulus for some visceral organs. For instance, a receptor found to respond to supraphysiological levels of distension of a viscus might also respond to entirely physiological pressure changes in the associated vasculature or in the mesenteric attachments of that viscus. One example can be seen in the case of renal colic. It might at first sight appear intuitively obvious that the pain arises from the mechanical stimulation offered by the passage of a sometimes large and rough stone through the restricted lumen of a ureter. However, an alternative explanation, for which there is much evidence, is that the stone merely obstructs the ureter, leading to a distension of the entire upper urinary tract. The increased hydrostatic pressure acting on receptors in the pelvis of the kidney may actually be the stimulus that is detected (Bretland, 1972).
Persistent pain of visceral origin is undoubtedly a greater clinical problem than pain from skin, but the overwhelming focus of experimental work on pain mechanisms has considered cutaneous systems, and much of what we know of mechanisms relates specifically to cutaneous sensibilities. However, until relatively recently, it was often tacitly assumed that these ideas could be transferred more or less wholesale to the visceral domain. Indeed, there has been such a reliance on the data and ideas arising from the study of somatic tissues that in some cases a visceral 'peg' has been made to fit a cutaneous 'hole'. It is arguably true that if the initial experimental studies had been undertaken on visceral tissue, then we might now have a completely different general theory of the processing of painful stimuli.
In this chapter, I will seek to directly compare the peripheral properties of somatosensory and viscerosensory systems. Such a discussion is timely since there has been a recent upturn in interest in viscerosensory processes, both from a basic science perspective and also because of the growing clinical belief that disorders such as Irritable Bowel Syndrome, previously thought to primarily motor dysfunctions, may actually be at least partly explained by altered sensory function. I will consider firstly the nature of sensations from visceral and somatic domains, and then attempt to correlate these sensibilities with the properties of primary afferent neurones innervating the two systems.
Nature of superficial and visceral pains.
(i) Effective stimuli. In normal healthy people, a variety of intense stimuli applied to skin readily produce pain. These include mechanical and thermal stimuli that might be considered noxious (i.e. tissue damaging), as well as non-damaging events such as electrical stimuli and some chemical irritants. When the intensities of cutaneous stimuli are raised above threshold, pain does not usually radiate but, conversely, frequently become more focal, as can be appreciated if a pencil point is increasingly pushed against the skin. Also, increasing the area over which a stimulus acts causes a modest increase in perceived pain, but importantly the threshold for pain is not markedly reduced (Price et al., 1989). Deep somatic tissues, such as joint and muscle, are similarly sensitive to direct tissue-threatening stimuli. For instance, strong mechanical pressure on a muscle or distortion of a joint beyond its working range cause pain. Irritant chemicals, directly injected into muscles or joints, can induce pain (see Mense, 1986).
The sensitivity of visceral tissues is markedly different. Some structures, such as the lung, liver and the parenchymatous part of the kidney, appear not to give rise to pain with any stimulus, including their gross destruction by malignant growth. In addition, surgeons working in the early part of this century, and using only local anaesthesia of the body wall, were surprised to find that a wide variety of traumatising stimuli including crushing, cutting, and burning, very rarely gave rise to any sensation when applied to healthy visceral tissue (MacKenzie, 1909; Morley, 1931; Capps, 1932). There are some exceptions e.g. the mesenteries are said by most authors to be sensitive to tension or clamping and it is recognised that the trigone region of the bladder neck can give rise to pain when probed directly or stimulated by the presence of a stone. However, viscera are sensitive to other forms of stimulation. The best recognised is distension of the hollow muscular-walled organs. Distension of the GI tract from oesophagus to rectum, the urinary tract from kidney pelvis to bladder, and of the gall bladder, all produce pain (Hertz, 1911; Pollard & Bloomfield, 1931; Denny-Brown & Robertson, 1933a; Bentley & Smithwick, 1940; Lewis, 1942; Ray & Niel, 1947; Nathan, 1952; Goligher & Hughes, 1951; Risholm, 1954; Bretland, 1972; Csendes & Sepulveda, 1980). The severity of distension-induced pain is often only modest in healthy subjects. It arises with a short latency (measured in seconds), suggesting that indirect effects (e.g. ischemia) are not the cause. Active contractions of smooth muscle, around for instance an obstruction, may exacerbate pain, and result in pain that comes in waves, as is so apparent in the case of labour pains. One can readily demonstrate similar contraction-associated pain by voluntarily checking the flow of urine in the middle of micturition. Closure of the urethral outlet causes a large reflex isovolumetric contraction of the bladder which, in most people, is distinctly painful.
One of the problems in transferring the concept of nociception from cutaneous to viscera domains is that the distending pressures associated with pain are not tissue-damaging (e.g. 30-40 mm Hg in the case of bladder distension). Estimates of the threshold pressures producing pain in a particular viscus often vary considerably. One reason is that the area of tissue stimulated may be a crucial determinant of threshold. Unlike skin, spatial summation may drastically reduce the effective threshold for pain. This viewpoint was strongly argued by Goldsheider (1920). Comparisons of different studies in man and animals suggest that spatial summation can appreciably lower the threshold for visceral pain (Lewis, 1942; Peterson & Youmans, 1945). The existence of appreciable spatial summation of visceral inputs may explain the failure of localised mechanical stimuli, even frankly damaging ones, to produce pain.
Another effective stimulus for visceral pain is ischemia. The best recognised example being that of ischemic heart disease but it is likely that ischemia of other visceral tissues produces pain (Lewis, 1942; Poole et al., 1987). With coronary occlusion there is the possibility of secondary mechanical effects (for instance the spasm of arteries, Osler, 1910) but it is frequently assumed that an important component of the stimulus is an accumulation of pain producing chemicals in the ischemic tissue (but see Malliani, 1986; Ness & Gebhart, 1990). A number of well recognised algogenic chemicals do produce pain when applied to human visceral tissues. The best studied is bradykinin, a naturally occurring agent. This substance produces pain when infused into the abdomen of healthy volunteers (Lim et al., 1967). It is less clear if it is algogenic in the heart. Euchner-Warmser et al., 1994; see also Pagani et al., 1985). Another reason for caution is that bradykinin may exert an indirect action via smooth muscle contraction (Floyd et al., 1977). Other, synthetic, algogens have been shown to induce pain in some viscera, for instance the urinary bladder (Head, 1893; Nesbit & McLellan, 1939; Maggi et al., 1989).
A final, and clinically important, circumstance where visceral pain may be triggered is in inflammatory states (Head, 1983; Wolf, 1965). In the urinary and alimentary tracts, inflammation is common and can be painful. In cystitis, for example, the sensations during bladder emptying often become unpleasant and painful (Nesbit & McLellan, 1939; Petersen & Franksson, 1955).
(ii) Hyperalgesia. In the wake of strong stimuli, the sensitivity of skin changes markedly. Previously innocuous stimuli become capable of evoking pain, and noxious stimuli produce more pain than in normal tissue. This phenomenon is called primary hyperalgesia. There is a wealth of experimental evidence that this hyperalgesia arises at least in part from a sensitization of primary sensory nociceptors. Surrounding the area of damage, skin also sometimes becomes more sensitive, a process called secondary hyperalgesia. The wide spread of secondary hyperalgesia that is sometimes seen argues strongly that the underlying cause must lie not in the properties of primary afferents but within the central nervous system, and indeed there is considerable experimental support for such a view (see McMahon et al, 1993).
The opportunities to observe such changes in visceral structures are much more limited, and it is not surprising that we have only meagre information on this point. Nonetheless, there are a number of anecdotal reports that visceral structures may become hyperalgesic, particularly in inflammatory states. Kinsella (1940) reported that direct mechanical stimulation of the inflamed, but not the healthy, appendix caused pain. Other reports exist for the ureter, kidney, bladder, ovary, stomach, and oesophagus (Head, 1983; Hertz, 1911; Hurst, 1911; McLellean & Goodell, 1943; Ruffin et al., 1953; Petersen and Franksson, 1955; Wolf, 1965). One example comes from the much-studied patient Tom, who had a gastric fistula. The gut mucosa of Tom was normally insensitive to pinching, but the same stimuli produced pain when the mucosa was inflamed. Quantitative studies of the increased sensitivity of inflamed viscera are few, but some data exist for patients with irritable bowel syndrome and non-cardiac chest pain. In one recent study Trimble et al., (1995) reported lower pain thresholds to distension in irritable bowel syndrome and functional dyspepsia (see also Mayer & Raybould, 1990). The altered sensibility of visceral tissue in pathological conditions such as inflammation may indicate the emergence of new neurophysiological processes, a view for which there is growing experimental evidence (see below).
(iii) Accuracy of localisation. Pain of cutaneous origin has distinct features. It is usually focal (that is, with well defined boundaries) and often has a burning quality. It is well localised, and even if any tactile cues are removed by block of large diameter afferent fibres, people can localise a noxious stimulus to the skin of the hand within 10-20 millimetres (Lewis, 1942). For visceral pain, two distinct types of localisation have been noted. In some cases, visceral pain may be referred to distant structure, as described in more detail below. In other cases, however, the pain is perceived as being deep within the body. This type of visceral pain, so-called 'true' visceral pain (or by early authors 'splanchnic pain', Ross, 1888), is usually perceived as arising in the midline, The pain may be perceived as anterior or posterior, and occasionally radiates over considerable distances. One example is the initial sensation perceived after myocardial infarction (Proccaci et al., 1986). Another is the early pain of appendicitis, which is initially felt in the midline. 'True' visceral pain is usually extensive rather than focal, perceived over an area much larger than that of the stimulus. Deep pain has diffuse boundaries. It is frequently associated with a sense of nausea and ill-being. Autonomic and motor reflexes associated with deep pains are often extreme and prolonged. Muscle rigidity may itself form a new source of pain, although this is contested by some. Only in exceptional circumstances is deep pain well localised.
(iv) Referred pain. In contrast to pain deriving from stimulation of skin, much visceral pain is localised to distant structures, a phenomenon known as referred pain. The area of referral is generally segmental and superficial. That is, to muscle and/or skin innervated by the same spinal nerves as the viscus giving rise to the referred sensation. A classic example is the pain that develops shortly after myocardial infarction. While the initial pain in these cases may be felt as deep, within the chest, with time (usually measured in minutes) it is often felt in parietal structures. Here it is still not well localised, but most often perceived as diffuse within the anterior chest and left arm. In some patients, the referred pain becomes yet more superficial to involve cutaneous structures, as time progresses (Procacci et al., 1986). Another example is the pain of renal colic which is felt in the iliac fossa and scrotum. The general pattern of referral is consistent enough to be of diagnostic use, although confusion can arise from viscera which share a common segmental innervation (ie. those within a viscerotome), for instance the heart and oesophagus. One notable feature of referred pain is that it masks the original 'true' visceral pain.
Descriptive studies on the nature of referred sensations in patients are sometimes confounded by the possibility that the effective stimulus moves from a visceral site to a parietal one. For instance, the rupture of an inflamed appendix is associated with the sudden appearance of a pain localised in the lower right quadrant of the abdomen (Silen, 1987). Similarly, the growth of a tumour may newly involve non-visceral tissue. Stimulation of the body wall and especially its membranous linings is well recognised to give rise to poorly localised deep pain in some cases, or a more superficial referred pain in others. It is clear, however, that some examples of referred pain cannot be explained on this basis, such as the pain of renal colic, angina, and in the extreme, the referred pain felt when the splanchnic nerve of conscious humans has been stimulated electrically (Foerster, 1933; Lerishe, 1937).
Another important feature of referred pain is that the site of referral may additionally show hyperalgesia (e.g. Head, 1983; Procacci et al., 1986). This is true for both the pain referred to muscle and to skin. Such tenderness develops slowly, taking many minutes or even hours to become manifest and, equally, persisting for prolonged periods, measured in hours.
Mechanisms of referral. Figure 1 illustrates a number of possible explanations that have been offered for referred sensations. The first case shown originated from Sinclair et al. (1948). The hypothesis is that some primary sensory neurones have widely bifurcating axons and innervate both somatic and visceral targets, thus obscuring the source of afferent activity, and explaining the segmental nature of referred sensations. In support, Bahr et al. (1981) found that 18% of a relatively small sample of unmyelinated fibres in the lumbar splanchnic nerves could be driven by electrical stimuli applied to segmentally appropriate somatic nerves. Some of these may have been sensory neurones, but no attempts were made to identify receptive fields in peripheral tissues. There have been other positive results reported for pairs of somatic nerves (Pierau et al., 1982; Taylor & Pierau, 1982), but these findings have been challenged on technical grounds (Devor et al., 1984). The only positive data for sensory neurones with receptive fields in two tissues comes a study by Mense et al. (1981) who reported single sensory neurones with both skin and muscle fields innervating the tail of the cat. Recently, Takahashi et al. (1993) provided evidence for bifurcating nociceptors with terminals in intervertebral discs and skin.
This first hypothesis, however,does not explain the time delay in the evolution of referred pain. Nor does it explain the referred hyperalgesia that frequently develops, since antidromic activity (that might invade the distant branch) does not appear capable of inducing a sensitization of peripheral terminals (Reeh et al., 1986).
Another putative mechanism of referred pain is that visceral and somatic primary sensory neurones converge onto common spinal neurones. This is the projection-convergence theory, suggested as such by Ruch (1946), but derived from earlier ideas of Sturge (1883) and Ross (1888). This proposes that the activity in ascending spinal pathways is misconstrued as originating from somatic structures. This theory can explain the segmental nature of referred pain. There is now considerable experimental evidence that somato-visceral convergence is common in spinal neurones (as reviewed in Ness and Gebhart, 1990; McMahon, 1994), but it should be remembered that many instances of such convergence may relate to the integration of somatic and visceral reflexes, rather than to viscerosensory processing. The theory does not explicitly address the issue of referred hyperalgesia. It is possible that summation of inputs from visceral and cutaneous structures could underlie cutaneous hyperalgesia, but the theory offers no explanation of the slow evolution of referred tenderness.
A variation on the theme of convergence-projection (figure 1c) derives from the ideas of MacKenzie (1909), and is the convergence-facilitation theory. Because MacKenzie was convinced that the viscera were wholly insensitive and therefore that visceral afferent activity never itself gave rise to pain, he proposed instead that this activity was capable of creating an 'irritable focus' within the spinal cord, so that other, segmentally appropriate, somatic inputs could now produce abnormal and, of course, referred pain sensations (MacKenzie, 1909). His theory did not find general acceptance, in part because it implicitly denied the existence of 'true' visceral pain. However, the theory offers an explanation for referred hyperalgesia and, perhaps, the delay in the referral of sensations (allowing for the generation of an 'irritable focus'). The concept of an irritable focus has been resurrected with another label - central sensitization, which appears to be of cardinal importance in hyperalgesia from somatic and visceral structures (see McMahon, 1994; Mayer and Gebhart, 1995).
A final view of referred pain is illustrated in figure 1d, and suggests that interactions at supraspinal levels lead to the phenomenon (Theobold, 1941). Most of electrophysiological data we have relating to projections from the spinal cord to brainstem, suggests that viscero-somatic convergence is extremely common, and such convergence is of course contrary to this theory (see McMahon, 1994). However, there is some evidence that a subset of ascending spinal neurones convey exclusively visceral information (e.g. Akeyson and Schramm, 1994) and there is additional evidence that, whatever the mechanism, some supraspinal structures may functionally respond to only visceral information (see Cechetto, 1987).
In summary, it is clear that the nature of pain from cutaneous and visceral tissues differs in a number of important respects, as summarised in Table I. There are a number of general reasons why this might arise. Firstly, quantitative differences in the density of innervation of somatic and visceral tissues may have a major impact in the precision of perceived sensations. It is also true that the tissues exhibit very different physical properties, such as their degree of visco-elasticity. Such differences may greatly alter the encoding properties of afferent terminals. Alternatively, the observed differences in sensory capacities may have arisen secondarily to the evolution of other systems. For instance, the skin is endowed with a numerically large and highly specialised system of large afferent fibres which, with great precision, relay information about tactile stimuli. This system is associated with a large and precise cortical representation of the body surface. It is therefore possible that the high degree of localisation seen for cutaneous pain derives incidently from the existence of the accurate cortical map of the body surface. Finally, visceral and somatic systems may be fundamentally different for good teleological reasons, for instance that the tissues are normally exposed to very different types of stimuli, and participate in different behavioural repertoires.
Comparison of cutaneous and visceral afferent primary sensory neurone properties.
Anatomy. The skin of all mammalian species is richly innervated by sensory neurones. In man, maybe a million or so fibres project from the skin of the body and head to the spinal cord (Holmes and Davenport, 1940). In contrast, the abdominal and thoracic viscera receive but a sparse innervation, amounting probably to only 5% or so of the numbers of somatic afferents. This paucity is all the more impressive remembering that many visceral structures are thrown into folds and offer an immense surface area. The innervation density in viscera is therefore only a small fraction of that seen in skin. Afferents to both structures are bundled in peripheral nerves. In the case of visceral afferents, these fibres run with the sympathetic or parasympathetic nerves, with the exception of the afferent innervation of much of the peripheral vasculature (which can be considered a visceral target since it is innervated by the sympathetic system) which is innervated by afferents running in appropriate somatic nerves. This anatomy has led to some confusion over terminology: the sensory neurones in the vagal or pelvic parasympathetic nerves are often called parasympathetic afferents, and similarly, those in the hypogastric nerves, for example, are frequently referred to as sympathetic afferents. Of course, the terms sympathetic and parasympathetic strictly refer to efferent systems but there is an obvious economy in applying them adjectivally to afferent neurones. On the other hand, the use of these terms appears to bestow, a priori, some special property on visceral afferents which may be unwarranted. Some texts claim that only 'sympathetic' afferents are responsible for signalling visceral nociceptive events. This generalisation is certainly at least an oversimplification. For instance, for the urinary bladder (which receives afferents via both sympathetic and parasympathetic nerves), clinico-pathological investigations, and studies after surgical interruption of individual nerves, have determined that the pain of acute over-distension or cystitis can be signalled by primary afferents in the parasympathetic pelvic nerve (Head, 1893; Head and Riddoch, 1917; Riddoch, 1921; Learmonth, 1931; Denny-Brown and Robertson, 1933b; Ray and Neill, 1947; White et al., 1952; Petersén and Franksson, 1955; Bors and Comarr, 1971; Gunterberg et al., 1975). Indeed, for this organ, little information exists about the function of the sympathetic afferents, projecting to the thoraco-lumbar spinal cord. Interruption of these pathways (as today is often the case in radical retroperitoneal lymphadectomy for testicular cancer) does not appreciably interfere with bladder sensation. Similarly, the ability of visceral afferents to induce c-fos in central neurones, often tacitly assumed to represent activation of pain-signalling pathways, actually appears much more pronounced for 'parasympathetic' than 'sympathetic' afferents (e.g. Traub et al., 1994).
Visceral afferents have the same general anatomy as their somatic counterparts, with terminals in both peripheral targets and spinal cord/brainstem, and cell bodies in dorsal root ganglia or ganglia of cranial nerves. While afferents from skin project to the brainstem and along the entire length of the spinal cord, visceral afferent projections are more restricted. In the spinal cord, the afferents running with sympathetic nerves project to thoracic and upper lumbar segments, while those running with parasympathetic nerves project to the 2nd - 4th sacral segments in man.
A major difference in cutaneous and visceral afferents is seen in the size distribution of fibres present. The classification of afferent fibres that is applied to skin derives from the work of Gasser (1939). Three groups are recognised: large myelinated (Aß) fibres, small myelinated (A ) fibres, and unmyelinated (C) fibres. In a typical cutaneous nerve the Aß fibres amount to some 20-25% of the total, the A 's about 10-15% and the C's 60-70%. There is a great deal of functional specialisation amongst the cutaneous afferent fibres. The large myelinated afferents respond to innocuous events such as light touch or limb movement. Many of the smaller diameter somatic afferents, conducting in the A and C velocity range, are nociceptors (see below). In visceral nerves, very few large myelinated fibres are present, numbering only a few percent of the total population. These appear to innervate Pacinian or Pacinian-like corpuscles located mostly in the mesenteries (see Janig and Morrison, 1986). The vast majority of afferents are A and C fibres, and these have to encode both innocuous and noxious stimuli. The ratio of unmyelinated:myelinated fibres, unlike skin, is about 10:1 (see Janig and Morrison, 1986; Willis and Coggeshall, 1991).
One final anatomical difference is seen in the preponderance of ventral root afferents. In a variety of species including rat, cat and man, it has been reported that some afferent fibres project not in the dorsal root but in the ventral root (see Willis and Coggeshall, 1991). These fibres have their cell bodies, as normal, in the dorsal root ganglion, and while some of the ventral root fibres may be loops or branches or dorsal root projections, in some cases the ventral root projection appears to be the only one (Habler et al., 1990c). These ventral root projections are much more common in spinal segments receiving a visceral projection and a relatively high proportion of visceral afferents have these ventral root projections (Habler et al., 1990c). However, since the functional role of this group of fibres remains unknown (indeed it is not even clear if these fibres project into the spinal cord or end blindly in the ventral root - see Willis and Coggeshall, 1991) the significance of this difference in somatic and visceral afferents is entirely speculative. It is possible that, during development, selective guidance cues operate for afferents innervating different targets, and that the nature of these cues for visceral afferents simply leads to more developmental errors in trajectory.
Physiology. The area of greatest controversy in viscerosensory processing relates to the encoding properties of visceral afferents. As described above, afferent Aß fibres are relatively abundant in somatic tissues. These fibres obviously have the highest conduction velocities and they are well suited to the rapid transmission of precise somatosensory information about the impact of the outside world on the body. The smaller cutaneous afferents, conducting in the A and C velocity range, are mostly nociceptors which respond when stimulus intensities are raised so as to threaten the integrity of the tissue. This distinction between low and high threshold afferents provides the main prop for the so-called specificity theory of sensory processing, figure 2a. The concept is straightforward. As stimulus intensity rises from liminal levels, a specialised group of low threshold, tactile, afferents is recruited. As stimulus intensities increase through the normal, innocuous, range, these afferents increase their discharge rates and then begin to saturate. With yet further increases in stimulus intensity into the noxious range, an entirely new group of afferents, small diameter nociceptors, are recruited.
In visceral nerves, as we have said, there are practically no Aß afferents. The few present are probably incapable of encoding information related to individual viscera (Janig & Morrison, 1986). This strongly suggests that both painful and non-painful sensations, and the afferent information used to regulate visceral reflexes, must be carried by the small afferent fibres. A crucial question, therefore, is whether the visceral afferents, like somatic afferents, can be divided into separate groups responding to innocuous and noxious events, respectively, conforming again to the specificity theory of pain, figure 2a. The problem of transferring the concept of 'noxious' from somatic to visceral tissues has already been discussed, and it is not always clear what properties one would expect from a specific visceral nociceptor. The problem is compounded by the fact that the response properties of visceral afferents have of course been determined in animals. Even where the nature of the effective stimulus is clear, for instance in the case of distension of the hollow viscera, one must extrapolate across species as to the levels at which the stimulus becomes painful. The use of pseudoaffective responses (such as increases in blood pressure) as a determinant of nociceptive threshold is not without its problems (see McMahon, 1994).
An alternative to the specificity theory is illustrated schematically in figure 2b. This is the intensity theory, which requires that individual fibres encode physiological, innocuous, events and, with higher discharge frequencies, supraphysiological, presumed noxious, ones. Clearly these two theories are mutually exclusive since the specificity theory denies any contribution from other than specific nociceptors and the intensity theory requires this contribution. Some workers have reported on visceral afferents that appear to conform to the specificity theory, whilst others find afferents that are clearly signalling events in both the physiological and supraphysiological ranges. Interestingly, for the most part, the electrophysiological data put forward by proponents of the two theories do not relate to the same viscus.
By way of example, the electrophysiological evidence deriving from two well studied tissues, and representing some of the conflicting interpretations that have been made, will be considered:
(i) The urinary bladder. Distension of the urinary bladder in healthy humans initially gives rise to a sensation of fullness and eventually pain as volume increases and intravesical pressure exceeds about 25-35 mm Hg (Denny-Brown and Robertson, 1933a; Nathan, 1956; Bors et al., 1956; Bors and Comarr, 1971; Morrison, 1987). Primary sensory neurones innervating the normal urinary bladder have been repeatedly and carefully studied (Floyd et al., 1976; Bahns et al., 1986, 1987; Häbler et al.,1988, 1990a; 1993a, Sengupta and Gebhart, 1994, Wen and Morrison, 1994, Dmitrieva and McMahon, 1996). Nearly all are small myelinated (A ) or unmyelinated (C), and travel with sympathetic (hypogastric) or parasympathetic (pelvic) nerves. Some or many exhibit a low level of ongoing discharge when the bladder is empty. There appear to be some species differences, as follows:
In the cat, bladder distension excites nearly all thin myelinated afferents, with pressure thresholds corresponding to the values where humans report the first sensation of fullness. Nearly all mechanosensitive afferents are activated by the intraluminal pressures reached during normal, non-painful micturition, and they form a homogenous population in terms of their stimulus-response functions. Mechanosensitive afferents respond in a graded fashion to increases of the intravesical pressure throughout the innocuous and into the supraphysiological, noxious, pressure range (Floyd et al., 1976; Häbler et al., 1990a; 1993a). These afferents reflect the magnitude and the temporal profile of intravesical pressure changes with high accuracy.
Also in the cat, the unmyelinated population of afferents projecting through the pelvic nerve differ in their properties. Very few fibres (<2.5%) respond to changes in intraluminal pressure in normal animals, and these differ significantly in their response properties from the population of thin myelinated fibres (Häbler et al.1990b). They have pressure thresholds of 30-50 mm Hg, outside, or at the top end of, the physiological range. Thus, there are only a few afferents that could be called specific nociceptors in the bladder, and which would signal only painful levels of distension. It is illuminating to estimate the magnitude of the afferent inflow arriving at the sacral spinal cord by these different afferent populations (McMahon & Koltzenburg, 1993). At an intravesical pressure of 50 mmHg (painful in man and beyond the normal physiological pressures), the total afferent discharge in the cat is about 4500 action potentials per second, of which only some 225 (around 5%) are contributed by the unmyelinated fibre population.
In the rat, most afferents in the pelvic nerve appear to have some spontaneous activity (Sengupta and Gebhart, 1994). The clear distinction between A and C fibre properties seen in the cat, is much less clear. Fibres which do not respond to even supraphysiological levels of distension are found by all workers (Sengupta and Gebhart, 1994; Wen and Morrison, 1994, Dmitrieva and McMahon, 1996), but they are less common than in the cat, represent only some of the unmyelinated population, and are not restricted to this population. The mechanosensitive population, as in the cat, largely form a homogenous population with pressure thresholds in the innocuous range, but responsiveness extending clearly into the supraphysiological range. However, Sengupta and Gebhart (1994) reported that 20% of mechanosensitive fibres (and therefore maybe 10-15% of all bladder afferents) had higher pressure thresholds, averaging 34 mm Hg. A similar proportion of bladder afferents (7%) from a smaller sample were found by Dmitrieva and McMahon (1996) to have high pressure thresholds.
The controversy relates to the interpretation of this data. On the one hand, some have argued that the existence of a subgroup of afferents with thresholds for activation in the range that might be associated with pain in humans, supports the specificity theory in relation to viscero-sensory processing. An alternative interpretation is that these fibres constitute the 'tail end' of a distribution of fibres that are essentially intensity encoding. The latter interpretation is supported by the relative paucity of these high-threshold fibres, and the fact that they are not distinguishable from intensity-encoding afferents in other respects (e.g. presence or absence of ongoing activity, sensitivity to bradykinin cf. Sengupta et al., 1992).
(ii) Internal reproductive organs. Kumazawa and co-workers have made extensive studies of the response properties of afferent nerve fibres in the superior spermatic nerve of the dog (see Kumazawa, 1986), and Berkley et al., and Hong et al., have studied in some detail the innervation of female reproductive organs. In the testes, other than a small population of rapidly adapting low threshold mechanoreceptors (numbering about 3%), Kumazawa found that afferent fibres form a homogenous group (of both myelinated and unmyelinated afferents) with polymodal receptors in testis and/or epididymis. They can be excited in a slowly adapting fashion to stimuli applied to one or more sensitive spots, each about a millimetre or so in diameter. The threshold for activation varies over a wide range but about 80% of afferents respond to mechanical stimuli of less than 17g/mm2. The afferents are polymodal in the sense that they respond to stimulation with algesic chemicals and heating as well as mechanical stimuli. Bradykinin and hypertonic saline solutions are effective stimuli for the afferents. Prostaglandins do not excite but sensitize the afferents to other stimuli. Heating the exposed testis excites afferents when stimuli exceeded about 45oC.
Kumazawa and colleagues correlated the response of canine testicular afferents with earlier psychophysical studies on the thresholds of testicular compression that causes pain in man (Woollard and Carmicheal, 1933). The testicular afferents could encode the level of compression up to several kilograms of force but most had thresholds below 50g. Woollard and Carmicheal (1933) reported that pain was felt in man with compressive forces over 200g or so. Kumazawa concluded that these afferents, while similar in some respects to the polymodal fibres innervating somatic tissues, could not be considered specific nociceptors since they encoded innocuous and well as noxious stimuli.
A sizeable population of hypogastric afferents supplying the uterus of the cat was studied by Hong et al., (1993). These sensory neurones had many features in common with testicular afferents. Nearly all responded in a slowly adapting fashion to mechanical stimuli. The large majority were also polymodal in that they were excited by chemical (bradykinin, capsaicin or potassium chloride) as well as mechanical stimuli. There response threshold to stimulation with von Frey hairs varied over two orders of magnitude, with about 20% of the fibres forming a high threshold group activated only by high intensity stimuli. Surprisingly, perhaps, relatively few of the fibres responded to spontaneous contractions of the uterus, and very high levels of intra-uterine distension (about 100 mm Hg) were necessary to activate two afferent neurones.
In female rats, Berkley and co-workers have demonstrated a marked dichotomy between uterine and vaginal sensory systems. They have shown that sensory fibres in the hypogastric nerve and pelvic nerve supply the uterus and vagina, respectively (Berkley et al., 1993). Moreover, they find that the pelvic afferents innervating the vagina encode low and high levels of distension, while the uterine afferents respond only to extreme levels of distension that are associated with ischemia. The same group (Berkley et al., 1995) have recently correlated the response properties of these afferents with behavioural responses to distension. For vaginal distension they found that animals exhibited detection thresholds that were lower than escape thresholds, and that both of these were in the range encoded by pelvic nerve afferents. Thus, one could propose an intensity coding function for this structure. In contrast, however, in the uterus, detection thresholds to distension were no lower than escape thresholds, and escape thresholds were very much greater than for vaginal stimulation. In fact, most animals did not respond to 100% of stimuli at any pressure, and a substantial minority failed to respond at all. One interpretation of these data is that hypogastric innervation of the uterus is specialised to respond to supraphysiological mechanical events, perhaps only those associated with ischemia. However, there are a number of factors complicating this interpretation. One is that the uterine pressures producing escape were greater than those ever likely (or unlikely) to occur in the non-pregnant rat. Secondly, the escape threshold were substantially higher than the threshold necessary to evoke hypogastric nerve afferent activity. The authors conclude "..the results also indicated that activity produced in these [hypogastric afferent] fibres, even by abnormal stimuli, does not inevitably result in behaviour." (Berkley et al, 1995). (A similar conclusion was also reached by Eucher-Wamser et al. (1994) for chemical stimulation of cardiac afferents.) A final complication for the female reproductive organs is that there appears a marked variation in the sensitivity of afferent systems through the estrous cycle (Robbins et al., 1992). Given these ambiguities, one can see how contrary interpretations of the data have been possible.
The properties of afferents innervating some visceral organs are held as demonstrating a clear case for intensity coding: Distension of the gall bladder in man, both pathologically (following obstruction of the bile ducts) and experimentally, causes pain when intraluminal pressures exceed about 35-45 mm Hg (Ray & Neill, 1947; Newman & Northrup, 1956; Csendes & Sepulveda, 1980). One study on the properties of afferents innervating the biliary system of the ferret (Cervero, 1982a) reports the existence of fibres with high pressure thresholds to distension that might therefore be considered nociceptors. These afferents, travelling via the sympathetic splanchnic nerves, have no ongoing activity and respond to direct tactile stimuli applied to restricted sites in the gall bladder and its ducts. These afferents represent perhaps the clearest case of specific nociceptors in visceral tissue. Yet even here all the data are not consistent with a simple intensity theory for pain, because studies on the spinal representation of biliary information disagree in some respects. Cervero (1982b, 1983) reported the existence of dorsal horn neurones with similarly high thresholds to gall bladder distension, but Ammons et al (1984a, b) found in their studies that pressure thresholds were generally in the range 0-10 mm Hg, well below what might be considered noxious.
It is difficult to reconcile some of the differences reported for visceral afferent encoding properties. One potentially complicating factor is that the stimulus-response functions of individual afferent fibres exhibit a continuum of mechanical thresholds, and so the situation depicted in figure 2c may actually represent physiological reality in many cases. This form of the intensity theory is of course compatible with the existence of a sub-population of afferents which appear to fulfil the criteria of specific nociceptors in a particular viscus (that is, the stimulus-response functions illustrated in both figure 2a and 2b could be present amongst the afferents innervating a single viscus).
Another consideration is that views on processing of cutaneous noxious stimuli are changing somewhat. The classic description of specific nociceptors has been largely confirmed by microneurographic studies in man in which the sensations evoked by stimuli have been correlated with the firing of individual afferent fibres. A clear conclusion can be when thermal stimuli are applied. When skin temperatures reach levels that subjects judge painful, nociceptors are recruited. Increasing stimulus temperatures are associated with increasing discharge rates of these nociceptors. However, it has been known for some time that the discharge frequencies that are associated with a just-painful mechanical stimulus are much higher (van Hees and Gybels, 1981). More recently, a careful microneurographic study on the relationship between evoked sensations and afferent nociceptive activity using graded mechanical stimuli has findings equally well suited to the intensity theory of sensations, with a significant degree of 'nociceptor' activity at stimulus levels not judged painful (Koltzenburg et al, 1994).
Sensitization and the recruitment of 'silent' afferents. It is very well established that the encoding properties of cutaneous afferent nociceptors can alter in the wake of a strong stimulus. The classic descriptions relate to a leftward shift in the stimulus-response functions of nociceptors to thermal stimuli after the skin containing the nociceptive terminal is subjected to a mild burn. That is, some of these afferents show a lowered heat threshold (and are activated by what are normally innocuous temperatures), and increases in their responses to suprathreshold heating (see Campbell et al., 1994 for review). It is now clear that a variety of strong stimuli, mechanical and chemical as well as thermal, can trigger this sensitization process. However, it has been much more difficult to demonstrate peripheral sensitization to mechanical stimuli in these polymodal nociceptors. It is not seen, for instance, with the same mild burns of skin. Very damaging stimuli can induce some mechanical sensitization of cutaneous afferents (Woolf & McMahon, 1985) but desensitization is as likely with these stimuli. In the visceral domain, there have been far fewer attempts to demonstrate peripheral sensitization. For the most part these have not considered thermal sensitization, since this is rarely a normal stimulus for visceral afferents. (In the case of testicular afferents, thermal sensitization has been reported.) However, in contrast to cutaneous afferents, mechanical sensitization has been repeatedly observed. For instance, for mechanosensitive afferents innervating the urinary bladder, the inflammation induced by intraluminal injection of chemical irritants such as turpentine and Mustard oil has been shown to excite and sensitize appreciable numbers of afferents to mechanical stimuli (Häbler et al., 1993b), with a leftward shift of the stimulus-response function to changes of intravesical pressure. We have recently observed a rapid onset mechanical sensitization of a very large proportion of vesical afferents to peripheral application of nerve growth factor (Dmitrieva and McMahon, 1996). Cervero & Sann (1989) have provided some evidence for a lowering of mechanical threshold of ureteric afferents in some conditions.
Haupt et al. (1983) have reported that colonic afferents subjected to ischaemia can show an increase in their levels of ongoing and contraction-related activity. These changes take some considerable time to develop. The results suggest a chemosensitivity of these afferents, and this was directly demonstrated in some cases using bradykinin and potassium.
Of course, in most of these cases it is not clear what agent is finally responsible for the sensitization. However, what is clear is that the process of mechanical sensitization is relatively easily induced in visceral afferents and not in cutaneous ones. This is likely to be the case even where the same intervention/substance is compared in the two tissues. For instance, nerve growth factor appears to induce a rapid thermal sensitization of cutaneous nociceptors, and a rapid mechanical sensitization of some visceral afferents (Dmitrieva & McMahon, 1996; Andreev et al., 1994). It may be that these different propensities to sensitize depend upon different receptors expressed on the two types of afferent, as discussed below.
Silent afferents: It has recently become clear that at least some unmyelinated afferent fibres do not respond appreciably to physiological or supraphysiological forms of mechanical stimuli. Some of these afferents respond specifically to chemical stimuli and have been called 'silent' or 'sleeping' afferents (see McMahon & Koltzenburg, 1990). They appear ideally suited to signalling changes occurring in inflammatory states. At the onset of an experimental inflammation, some of these fibres become active and, moreover, develop a novel mechanosensitivity so that they now respond to events such as distension. The presence of these fibres serves to further reinforce the idea that mechanisms of pain may change dramatically when one moves from normal healthy tissue to diseased pathological states.
These silent afferents have been difficult to study experimentally, because of the obvious problems associated with their identification. Even studies specifically designed to isolate this type of fibre may result in systematic under- or overestimates of their numbers. Nonetheless, in skin there have now been several studies of this type of fibre. The emerging picture is that an appreciable minority of unmyelinated afferents (in the range 10-20%) may be of this type. In viscera, only the innervation of the urinary bladder has so far been systemically studied for this type of fibre. In the pelvic nerve of the cat, very few fibres (2.5%) respond to changes in intraluminal pressure in normal animals. The overwhelming majority of unmyelinated pelvic afferents innervating the bladder therefore appear to fall into this 'silent' category. In the rat, the incidence of these silent fibres appears lower, but now three independent groups have suggested the existence of significant numbers in the pelvic nerve (Sengupta and Gebhart, 1994; Wen and Morrison, 1994; Dmitrieva and McMahon, 1996). Most or all of these afferents are likely to have sensitivity to chemical agents, such as capsaicin (Hu-Tsai et al., 1992) or even the constituents of normal urine (Wen & Morrison, 1994).
At the onset of an acute vesical inflammation induced by intraluminal injection of chemical irritants such as turpentine and Mustard oil some of these silent afferents are excited (Häbler et al., 1990b). These neurones show an initial burst of activity which then settles to a lower level as the chemically induced inflammation progresses. Some of these initially mechanically insensitive afferents also acquire a novel mechanical sensitivity in the biologically relevant pressure range (Häbler et al., 1990b). Compared to the number of unmyelinated afferents responding in the normal animal, some four times as many can be excited by distension at the onset of an acute inflammation. The activation of a numerically significant population of initially unresponsive afferents indicates that peripheral afferent mechanisms encoding pain from pelvic viscera are highly malleable and are strongly affected by the state of the tissue. These peripheral changes are obviously likely to be important for signalling pain and discomfort in inflammatory conditions.
The major difference between skin and visceral afferents in this regard appears to be numerical, although it is not yet known whether the afferent innervation of other visceral structures will be similar to the bladder. Lombardi et al. (1981), made incidental observations suggesting that specific chemoreceptors were uncommon or absent amongst afferents innervating the heart via the sympathetic nerves. Sengupta and Gebhart (1994b) also note that some 'silent' afferents can be recruited after chemical stimulation of the colon.
Neurochemistry.
Fast neurotransmitters. The large majority of primary sensory neurones appear to use the excitatory amino acids glutamate or aspartate as their principal neurotransmitter (see McMahon et al., 1993 for review). The transmitter is localised to at least 70% of dorsal root ganglion cells. Many of the post-synaptic responses of dorsal horn neurones are blocked by amino-acid antagonists. The cutaneous receptive fields of dorsal horn neurones are abolished by the blockade of glutamate receptors. We have less, direct, information about visceral afferents, but indirect evidence is consistent with the idea that glutamate is the principal neurotransmitter: Thus, micturition contractions triggered by bladder afferent activity are depressed by glutamate antagonists (Rice and McMahon, 1994), and central sensitization induced by activity in visceral afferents is blocked by NMDA receptor antagonists (as discussed more fully below). The ability of visceral afferents to induce c-fos expression in dorsal horn neurones is also partially blocked by glutamate antagonists (Birder & de Groat, 1992). There is therefore no compelling reason to believe that visceral and cutaneous afferents differ fundamentally in the fast neurotransmitters they use.
Neuropeptides and other primary afferent markers. The cell bodies of afferent neurones innervating somatic tissues can be sub-divided into three minimally overlapping populations (Averill et al., 1994) as follows: (i) the classically described large light cells, rich in neurofilaments and stained positively with the antibody RT97. These cells have myelinated axons and mostly innervate tactile low threshold mechanoreceptors. Theyamount to 30-40% of all somatic afferents; (ii) the peptidergic afferents, marked by CGRP. These are small and medium sized cells, with some overlap with the RT97 population. Subsets of these cells contain other neuropeptides such substance P. Most cells have unmyelinated or thinly myelinated axons and innervate nociceptors. They number about 45% of the total somatic population; (iii) The non-peptidergic small dark population. These cells too mostly innervate nociceptors via unmyelinated axons. They do not contain neuropeptides but are marked in a number of other ways. They react to the antibody LA4, and express the enzyme FRAP. They also react with the lectin IB4, but this shows more overlap with the CGRP population. About 40% of cells form this group.
Visceral afferents exhibit different proportions in these classes. The paucity of large afferent neurones necessarily leads to greater proportions of small cells. It is perhaps not surprising that the numbers of peptidergic afferents is generally found to be greater than among somatic afferents. There are quantitative differences between viscera. For instance, the prevalence of substance P expressing neurones is reported to vary from about 10% in pancreatic afferents to 40% for stomach afferents in the rat (Dockray and Sharkey, 1986). In the afferents innervating the urinary bladder, some 60-70% of cells appear to express CRGP and 25% substance P (see de Groat, 1986, Hunt et al., 1992; Bennett et al., 1996). FRAP is also expressed by urinary bladder afferents (McMahon, 1986). In these respects, visceral afferents appear only quantitatively different from their cutaneous counterparts. However, there are in addition some very striking qualitative differences. Most notably, the peptide VIP is apparently expressed in very many visceral afferents (at least in the cat), but many fewer somatic ones (see de Groat, 1986). It may therefore be a relatively good marker of visceral afferents. VIP expression shows another interesting feature. Peripheral axotomy of somatic nerves leads to the induction of VIP in substantial numbers of the damaged afferents, whereas axotomy of visceral nerves leads to a reduction in VIP expression. Another marker showing high levels of target specificity may be somatostatin which is found in a small minority of cutaneous afferents but practically no visceral ones.
The enzyme nitric oxide synthase (NOS), which makes the putative neurotransmitter NO, also appears to be differentially expressed in somatic and visceral afferents. For instance, in the rat, cells in the L4 and L5 dorsal root ganglia innervate somatic targets and very few normally express NOS. In contrast, relatively large numbers of afferents express NOS in spinal ganglia innervating visceral targets (Aimi, 1992). By retrograde labelling of splanchnic nerve we have confirmed that it is visceral afferents that normally express NOS (Train, Reynolds, McMahon, Woolf, unpublished). Like VIP expression, NOS is induced in many somatic afferents after peripheral axotomy (Fiallos-Estrada, 1993) but down-regulated in visceral afferents (unpublished observations).
It is less clear if nitric oxide normally functions as a transmitter for visceral afferent neurones. It has recently been reported that NOS inhibitors do not block micturition reflexes, although they do prevent the development of bladder hyper-reflexia after experimental inflammation (Rice, 1995).
Capsaicin sensitivity. The pungent extract of chili peppers, capsaicin, has a variety of actions on sensory neurones, apparently via a specific capsaicin receptor. It can strongly activate afferents but with repeated applications can desensitize them. In developing animals capsaicin can selectively kill small diameter afferent neurones. Capsaicin activates peptidergic afferent neurons and leads to the release of, for instance, CGRP and substance P. Capsaicin acts on both cutaneous and visceral afferents. About 25% of cutaneous afferents and about 60% of visceral afferents appear to be sensitive to capsaicin (Hu-Tsai et al., 1992). Since the neuropeptides released from the peripheral terminals of afferents by capsaicin are neuroactive, it is not surprising that so-called neurogenic extravasation is well developed in skin and in at least some visceral structures. Again, the difference between visceral and cutaneous systems appears mostly one of degree.
Neurotrophin receptors. In development, neurons of the peripheral nervous system are dependent for survival on neurotrophic factors, which are produced in target tissues. The most important molecules for survival appear to be the neurotrophins, currently comprising in mammals nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and NT-4/5. These proteins exert their actions via another family of molecules, the trk receptors, with trkA mediating NGF responses, trkB mediating BDNF and NT-4/5 responses, and trkC mediating NT-3 responses. The neurotrophins not only regulate survival, but may also determine the phenotype of sensory neurones (Ritter et al, 1992). The neurotrophins may also continue to exert actions on sensory neurone phenotypic properties in the adult animal. For instance, the availability of NGF appears to regulate the levels of the sensory neuropeptides CGRP and substance P. It is possible that some or many of the differences between cutaneous and visceral afferents (described above) may arise because of the different types or levels of neurotrophic support that afferents receive. It is therefore of interest to ask if there are differences in neurotrophin sensitivity between the afferent groups. We have recently reported on the expression of trk mRNAs in functionally defined groups of sensory neurones (McMahon et al., 1994). We found that about 45% of somatic afferents express trkA (the receptor for NGF). These cells overlap almost completely with the CGRP population (Averill et al., 1994). Smaller numbers of afferents express trkB and trkC (about 25% and 20% respectively). The trkC expressing cells were large in diameter while the trkB expressing cells could be large or small. In marked contrast, 90% of afferents projecting through the pelvic nerve expressed trkA, and 94% expressed trkB. Only 2% were found to express trkC. A recent immunocytochemical study broadly supports these findings (Bennett at al., 1996). Clearly, nearly all pelvic visceral afferents must express both trkA and trkB. The large numbers of visceral afferents sensitive to NGF (i.e. expressing trkA) may explain why more of these cells express neuropeptides. The co-expression of trkA and trkB in these afferents also distinguishes them from cutaneous afferents.
A summary of the properties of visceral and cutaneous sensory neurones is given
in table II.
Conclusions.
Clinical observations on the nature of visceral and cutaneous sensibilities in health and disease, as discussed in the first part of this article and summaised in Table I, provided a powerful impetus for suggesting that there may be fundemental differences in the organisation and properties of the respective sensory systems. As the second part of this article discusses, there are indeed differences in the somato-sensory and viscero-sensory systems, listed in Table II. Some of these differences are qualitative, such as the fibre size spectrum of primary sensory neurones innervating the two systems, but many are quantitative, for instance the proportions of afferent neurones expressing particular neuropeptides or receptors. An importat question therefore is to what extent the observed differences in properties can account for the observed differences in sensibilities. The quantitative differences in the density of innervation of somatic and visceral tissues may alone be a sufficient explanation: The apparent insensitivity of visceral tissue to focal yet tissue-damaging stimuli, but the effectiveness of distension and ischemia, is readily explained on such a basis. Further, the diffuse nature and poor localisation of 'true' visceral pain may have the same explanation. The spatial summation of visceral stimuli, hypothesised to be a crucial determinant of the pain threshold in the hollow viscera, is at least partly explainable in terms of innervation density, although it is not clear on this basis alone why spatial summation appears not to be relevant in establsihing the pain threshold in skin, at least for thermal stimuli. Numerical diffrerences in visceral and cutaneous innervation seem at first sight less relevant to the question of referred pain. However, it is possible that one consequence of the development of a large and precise cortical representation for cutaneous but not visceral information, may have led incidently to a system whereby visceral sensations are at best poorly localised, and frequently 'default' to segmentally appropriate somatic stuctures.
An alternative view would be to suggest that the few marked qualitative differences in visceral and cutaneous innervations underpin the observed differences in sensibilities. The most significant, if controversial, difference relates to the hypothesised physiological encoding properties of the two afferent systems. If it is true that much or all visceral pain is signalled by intensity-encoding afferents, rather than a dedicated group of specific nociceptors, then it could be argued that the ineffectiveness of local trauma to produce pain in viscera arises because of insufficient activation of afferents. Another feature of an intensity-encoding mechanism is that it inherently encompasses the idea of summation, also arguably explaining the effectiveness of distension and ischemic stimuli. Encoding by an intensity mechanism does not in itself address the issue of referred pain or referred hyperalgesia, wihch would appear to require a central rather than peripheral explanation.
The observed sensitization of visceral afferents, including the sensitization of so-called silent afferents, is of course likely to contribute to primary hyperalgesia, although intensity-encoding and specific systems have equal explanatory power in this respect.
This work of the author reported in the article was supported by the Medical Research Council (UK).
| Visceral Cutaneous | Effective stimuli | |
| Direct trauma ineffective | Distension & Ischemia effective | Direct trauma effective |
| Summation | Yes? | No |
| Localisation | Local - diffuse Often referred | Local - precise Not referred |
| Primary Hyperalgesia | Yes | Yes |
| Secondary Hyperlagesia | Yes, at site of referral | Yes, around site of damage |
Table I: Features of visceral and cutaneous pain compared. See text for details.
| Visceral Afferents | Cutaneous afferents | |
| Axon calibre/velocity | Ad,C | Aß,Ad,C |
| Stimulus-response function | Frequently intensity encoding | Separate high and low threshold populations |
| Sensitization | Mechanical sensitization common | Thermal sensitization common, mechanical uncommon. |
| 'Silent' afferents | Numerous in some organs (e.g. 90% of cat pelvic C fibres) | Less common (c. 20% of C fibres?) |
| Neurotransmitters | Probably mostly glutamate | Mostly glutamate |
| Neuropeptides/ Neuromodulators | CRGP & SP common, VIP in some populations Nitric oxide constitutively expressed in many afferents | CGRP & SP common, VIP very unusual. Nitric oxide not normally expressed |
| Capsaicin sensitivity | Many afferents sensitive | Many small diameter afferents sensitive |
| Neurotrophin receptors | trkA and trkB in 90% trkC rare | trkA,B,C more evenly distributed. Many small afferents no known trk |
| Receptors | 5HT, GABAa, bradykinin, opiate, | 5HT, GABAa, bradykinin, opiate |
| Ventral root afferents | Relatively common | Uncommon |
| Neurogenic extravasation | Well developed | Well developed |
Table II: Comparison of properties of primary afferent fibres innervating visceral and cutaneous targets.
Figure 1. Diagrammatic representation of three possible encoding mechanisms for noxious events by visceral afferent nerve fibres. See text for details. (Adapted from Cervero, 1988.)
Figure 2. Summary diagram illustrating the various theoretical mechanisms of referred pain. (from Morrison, 1987)
Aimi Y, Fujimura M, Vincent S R, Kimura H 1991 Localisation of NADPH-diaphorase containing neurones in sensory ganglia of the rat. Journal of Comparative Neurology 306: 382-392
Akeyson, E.W., and Schramm L.P. (1994) Processing of splanchnic and somatic input in thoracic spinal cord of the rat. Am. J. Physiol. 266: R257-R267
Ammons W S, Foreman R D 1984a Cardiovascular and T2-T4 dorsal horn cell responses to gall bladder distension in the cat. Brain Research 321: 267-277
Ammons W S, Blair R W, Foreman R D 1984b Raphe magnus inhibition of primate T1-T4 spinothalamic cells with cardiopulmonary visceral input. Pain 20: 247-260
Andreev, N., Inuishin, M., McMahon, S.B. 1994. Nerve growth factor acutely enhances the responsiveness of dorsal horn neurones to noxious heat. Soc. for Neurosci. Abs. 20: 455.16.
Averill, S., McMahon, S.B., Clary, P.O., Reichhardt, L.F., Priestley, J.V. 1995 Immunocytohemical localisation of TrkA receptors in chemically identified subgroups fo adult sensory neurones. Eur. J. Neurosci. 7:1484-1494
Bahns E, Ernsberger U, Jänig W, Nelke A 1986 Functional characteristics of lumbar visceral afferent fibres from the urinary bladder and the urethra in the cat. Pflügers Archives 407: 510-518
Bahns E, Halsband U, Jänig W 1987 Responses of sacral visceral afferents from the lower urinary tract, colon and anus to mechanical stimulation. Pflügers Archives 410: 296-303
Bahr R, Blumberg H, Jänig W 1981 Do dichotomizing fibers exist which supply visceral organs as well as somatic structures? A contribution to the problem of referred pain. Neuroscience Letters 24: 25-28
Bennett DLH, Dmitrieva N, Priestley JV, Clary D, McMahon SB. 1996 trkA, CGRP, and IB4 expression in retrogradely labelled cutaneous and visceral primary sensory neurones. Neurosci Lett 206:1-6
Bentley F H, Smithwick R H 1940 Visceral pain produced by balloon distension of the jejunum. Lancet i-ii: 389-391
Berkley KJ, Robbins A, Sato Y. 1993 Functional differences between afferent fibres in the hypogastric and pelvic nerves innervating female reproductive organs in the rat. J Neurophysiol. 69: 533-544
Berkley KJ, Wood E, Scofield SL, Little M. 1995 Behavioral responses to uterine or vaginal distension in the rat. Pain 61: 121-131
Birder, L., and de Groat, W. 1992 The effect of glutamate antagonists on c-fos expression induced in spinal neurones by irritation of the lower urinary tract. Brain Research, 580: 115-120.
Bors E, Comarr A E 1971 Neurological Urology: Physiology of Micturition, its Neurological Disorders and Sequelae. Karger, Basel
Bors E, Ma K T, Parker R B 1956 Observations on some modalities of bladder sensation. Journal of Urology 76: 566-575
Bretland P M 1972 Acute Ureteric Obstruction. Butterworths, London
Capps J A 1932 An Experimental and Clinical Study of Pain in the Pleura, Pericardium and Peritoneum. Macmillan, New York
Cervero F 1982a Afferent activity evoked by natural stimulation of the biliary system in the ferret. Pain 13: 137-151
Cervero F 1982b Noxious intensities of visceral stimulation are required to activate viscerosomatic multireceptive neurons in the thoracic spinal cord of the cat. Brain Research 240: 350-352
Cervero F 1983 Somatic and visceral inputs to the thoracic spinal cord of the cat: effects of noxious stimulation of the biliary system. Journal of Physiology (London) 337: 51-67
Cervero F, Sann H 1989 Mechanically evoked responses of afferent fibres innervating the guinea-pig's ureter: an in vitro study. Journal of Physiology 412: 245-266
Cechetto, D.F. (1987) Central representation of visceral function. Federation Proceedings, 46: 17-23.
Csendes A, Sepulveda A 1980 Intraluminal gallbladder pressure measurements in patients with chronic or acute cholecystitis. American Journal of Surgery 139: 383-384
de Groat W C 1986 Spinal cord projections and neuropeptides in visceral afferent neurons. In: Cervero F, Morrison J F B (eds) Visceral Sensation. Progress in Brain Research 67, Elsevier, Amsterdam, pp 165-187
Devor M, Wall P D, McMahon S B Dichotomizing somatic nerve fibres exist in rats but they are rare. Neuroscience Letters 49: 187-192
Denny-Brown D, Robertson E G 1933a On the physiology of micturition. Brain 56: 149-190
Denny-Brown D, Robertson E G 1933b The state of the bladder and its sphincters in complete transverse lesions of the spinal cord and cauda equina. Brain 56: 397-463
Dockray, G.J., and Sharkey, K.A. 1986 Neurochemistry of visceral afferent neurones. In 'In Progress in brain research', eds F. Cervero, J.F.B. Morrison,vol 67, 133-148, SpringerVerlag, Berlin.
Dmitrieva, N., McMahon, S.B. 1996. Sensitisation of visceral afferents by nerve growth factor in the adult rat. Pain, in press.
Euchner-Warmser, I., Meller, S.T., Gebhart, G.F. 1994 A model of cardiac nociception in chronically instrumented rats: behavioral and electrophysiological effects of pericardial adminstration of algogenic substances. Pain, 58: 117-128
Fiallos-Estrada, C.E., Kummer, W., Mayer, B., Bravo, R., Zimmerman, M., Herdegen, T. (1993). Long-lasting increase of nitric oxide synthase immumoreactivity, NADPH-diaphorase reaction and c-JUN co-expression in rat dorsal root ganglion neurons following sciatic nerve section. Neurosci. Lett. 150: 169173.
Floyd K, Hick V E, Koley J, Morrison J F B 1977 The effects of bradykinin on afferent units in intra-abdominal sympathetic nerve trunks. Quarterly Journal of Experimental Physiology 52: 19-25
Floyd K, Hick V E, Morrison J F B 1976 Mechanosensitive afferent units in the hypogastric nerve of the cat. Journal of Physiology (London) 259: 457-471
Foerster O 1933 Dermatomes in man. Brain 56: 1-39
Gasser, H.S., and Grundfast. H. 1939 Axon diamters in relation to the spike dimensions and the conduction velocity in mammalian fibres. Am J Physiol. 127: 393-417
Goldscheider A 1920 Das Schmerzproblem. Springer, Berlin
Goligher J C, Hughes E S R 1951 Sensibility of the rectum and colon: its role in the mechanism of anal continence. Lancet i-ii: 543-548
Gunterberg B, Norlén L, Stener B, Sundin T 1975 Neurologic evaluation after resection of the sacrum. Investigative Urology 13: 183-188
Häbler H-J, Jänig W, Koltzenburg M 1988 A novel type of unmyelinated chemosensitive nociceptor in the acutely inflamed urinary bladder. Agents Actions 25: 219-221
Häbler H-J, Jänig W, Koltzenburg M 1990a Properties of myelinated sacral afferents responding to filling and distension of the urinary bladder. Pflügers Archives 415: R104
Häbler H-J, Jänig W, Koltzenburg M 1993a Myelinated primary afferents of the sacral spinal cord responding to slow filling and distention of the cat urinary bladder. Journal of Physiology. 463: 449-460
Häbler H-J, Jänig W, Koltzenburg M 1993b Receptive properties of myelinated primary afferents innervating the inflamed urinary bladder of the cat. Neurophysiology 69: 395-405
Häbler H-J, Jänig W, Koltzenburg M 1990b Activation of unmyelinated afferent fibres by mechanical stimuli and inflammation of the urinary bladder in the cat. Journal of Physiology (London) 425: 545-562
Häbler H-J, Jänig W, Koltzenburg M, McMahon S B 1990c A quantitative study of the central projections of unmyelinated ventral root afferents in the cat. J. Physiol. 422: 265-288.
Haupt P, Jänig W, Köhler W 1983 Response pattern of visceral afferent fibres, supplying the colon, upon chemical and mechanical stimuli. Pflügers Archives 398: 41-47
Head H 1893 On the disturbances of sensation with especial reference to the pain of visceral disease. Brain 16: 1-133
Head H, Riddoch G 1917 The autonomic bladder, excessive sweating and some other reflex conditions, in gross injuries of the spinal cord. Brain 40: 188-263
Hertz, A.F.1911 The sensibility of the alimentary tract in health and disease. Lancet, 1: 1051-1056
Hong, S.K., Han, H.C., Yoon, Y.W., Chung, J.M. (1993) Response properties of hypogastric afferent fibres supplying the uterus in the cat. Brain Res. 622: 215-225
Hunt, S., Mantyh, P.W., and Preistley, J.V. (1992) Patterns of termination of biochemically characterized sensory neurons. In 'Sensory neurons: Diversity, Development and Plasticity' S.A. Scott (ed). Oxford University Press, New York. pp. 60-76.
Hurst J 1911 The Goulstonian Lectures on the Sensibility of the Alimentary Canal. Frowde, Hodder and Stoughton, London
Holmes, F.W. and Davenport, H.A. 1940 Cells and fibres in spinal nerves; number or neurites in dorsal and ventral roots of cats. J. Comp. Neurol, 73: 1-5
Hu-Tsai, M., Winter, J., and Woolf, C.J. 1992 Regional differences in the distribution of capsaicin-sensitive target-identified rat dorsal root ganglion neurons. Neurosci. Lett., 143: 251-254
Jänig W, Morrison J F B 1986 Functional properties of spinal afferents supplying abdominal and pelvic organs, with special emphasis on visceral nociception. In: Cervero F, Morrison J F B (eds) Visceral Sensation. Progress in Brain Research 67, Elsevier, Amsterdam, pp 87-114
Kinsella V J 1940 Sensibility in the abdomen. British Journal of Surgery 27: 449-463
Koltzenburg, M., and Handwerker, H.O. 1994 Differential ability of human cutaneous nociceptors to signal mechanical pain and to produce vasodilatation. J. Neurosci. 14: 1756-1765
Kumazawa T 1986 Sensory innervation of reproductive organs. In: Cervero F, Morrison J F B (eds) Visceral Sensation. Progress in Brain Research 67, Elsevier, Amsterdam, pp 115-132
Learmonth J R 1931 Neurosurgery in the treatment of disorders of the urinary bladder: I. Anatomic and surgical considerations. Journal of Urology 25: 531-549
Leriche R 1937 Des douleurs provoquées par l'excitation du bout central des grands splanchniques (douleurs cardiaques, douleurs pulmonaires) au cours des splanchnicotomies. Presse Médicale 45: 971-972
Lewis, T. 1942 Pain. Macmillan, New York.
Lim R K S, Miller D G, Guzman F, Rodgers D W, Rogers R W, Wang S K, Chao P Y, Shih T Y 1967 Pain and analgesia evaluated by the intraperitoneal bradykinin-evoked pain method in man. Clinical Pharmacology Therapeutics 8: 521-542
Lombardi, F., Della Bella, P., Casati, R., and Malliani A. 1981 Effects of intracoronary administration of bradykinin on the impulse activity of afferent sympathetic unmyelinated fibres with left ventricular endings in the cat. Circ. Res. 48: 69-75.
MacKenzie J 1909 Symptoms and their Interpretation. Shaw, London
Maggi C A, Barbanti G, Santicioli P, Beneforti P, Misuri D, Meli A, Turini D 1989 Cystometric evidence that capsaicin-sensitive nerves modulate the afferent branch of micturition reflex in humans. Journal of Urology 142: 150-154
Malliani A 1986 The elusive link between transient myocardial ischemia and pain. Circulation 73: 201-204
Mayer E A, Raybould H E 1990 Role of visceral afferent mechanisms in functional bowel disorders. Gastroenterology 99: 1688-1704
Mayer, E A, Gebhart, G F. 1995 Basic and clinical aspects of visceral hyperalgesia. Gastroenterology. 107: 271-293
McLellan A M, Goodell H 1943 Pain from the bladder, ureter, and kidney pelvis. Res. Publ. Assoc. Res. Nerv. Ment. Dis. 23: 252-262
McMahon S.B. 1986. The localisation of fluoride resistant acid phosphatase (FRAP) in the pelvic nerves and sacral spinal cord of rats. Neurosci. Lett., 64: 305310.
McMahon, S.B. (1994) Mechanisms of cutaneous, deep and visceral pain. In "Textbook of Pain." Third edition, Ed. Wall & Melzack, pp 129-151.
McMahon, S.B., Armanini, M.P., Ling, L.H., Phillips, H.S. 1994. Expression and co-expression of TRK receptors in subpopulations of adult primary sensory neurons projecting to identified peripheral targets. Neuron, 12: 1161-1171.
McMahon, S.B., Koltzenburg, M. 1990. Novel classes of nociceptors: Beyond Sherrington. T.I.N.S. 13: 199-201.
McMahon, S.B., Lewin, G.R., Wall, P.D. 1993. Central hyperexcitability triggered by noxious inputs. Curr. Opin. Neurobiol, 3: 602-610.
McMahon S B, Koltzenburg M 1993 Changes in the afferent innervation of the inflamed urinary bladder. In " Basic and Clinical aspects of chronic abdominal pain." Ed. H.Raybould & E.Mayer. Pain Research & Clinical Management, Volume 9. Chapter 13, 155-171.
Mense S, Light A R, Perl E R 1981 Spinal terminations of subcutaneous high-threshold mechanoreceptors. In: Brown A G, Rethelyi M (eds) Spinal Cord Sensation. Scottish Academic Press, Edinburgh
Mense S 1986 Slowly conducting afferent fibres from deep tissues: Neurobiological properties and central nervous actions. Prog. Sens. Physiol. 6: 139-219
Morley J 1931 Abdominal Pain. Livingstone, Edinburgh
Morrison J F B 1987 In: Torrens M, Morrison J F B (eds) The Physiology of the Lower Urinary Tract. Springer-Verlag, Berlin, pp 53-132
Nathan P W 1956 Sensations associated with micturition. Journal of Urology 28: 126-131
Nesbit RM, McLellan F C 1939 Sympathectomy for the relief of vesical spasm and pain resulting from intractable pain infection. Surg. Gynecol. Obst. 68: 540-546
Ness T J, Gebhart G F 1990 Visceral pain: a review of experimental studies. Pain 41: 167-234
Newman H F, Northrup J D 1956 Intraluminal pressure and electrical excitability of the human gall bladder. Journal of Applied Physiology 9: 121-129
Osler W 1910 The Lumleian lectures on 'Angina pectoris' II. Lancet 1: 839
Pagani M, Pizzinelli P, Furlan R, Guzzetti S, Rimoldi O, Sandrone G, Malliani A 1985 Analysis of the pressor sympathetic reflex produced by intracoronary injections of bradykinin in conscious dogs. Circulation Research 56: 175-183
Petersén I, Franksson C 1955 The sensory innervation of the urinary bladder. Urology International 2: 108-119
Peterson C G, Youlmans W B 1945 The intestino-intestinal inhibitory reflex: threshold variations, sensitization and summation. American Journal of Physiology 143: 407-412
Pierau Fr.-K, Taylor D C M, Abel W 1982 Dichotomizing peripheral fibres revealed by intracellular recording from rat sensory neurones. Neuroscience Letters 31: 123-128
Polland W S, Bloomfield A L 1931 Experimental referred pain from the gastrointestinal tract. I. The esophagus. Journal of Clinical Investigations 10: 435-452
Poole, J.W., Sammartano, R.J., Boley, S.J. 1987 Haemodynamic basis of the pain of chronic mesenteric ischaemia.. Annals Int. Med. 153: 171-180
Price D D, McHaffie J, Larson M 1989 Spatial summation of heat induced pain; Influence of stimulus area and spatial separation of stimuli on perceived sensation intensity and unpleasantness. Journal of Neurophysiology 62: 1270
Procacci P, Zoppi M, Maresca M 1986 Clinical approaches to visceral sensation. In: Cervero F, Morrison J F B (eds) Visceral Sensation. Progress in Brain Research 67: 21-28
Ray B S, Neill C L 1947 Abdominal visceral sensation in man. Annals of Surgery 126: 709-724
Reeh P W, Kocher L, Jung S 1986 Does neurogenic inflammation alter the sensitivity of unmyelinated nociceptors in the rat? Brain Research 384: 42-50
Rice, A.S.C., and McMahon, S.B. 1994 Pre-emptive intrathecal administration of antagonist (AP-5) prevents hyper-reflexia in a model of persistent visceral pain. Pain, 57: 335-340
Rice, A 1995. Topical spinal administration of a nitric oxide synthase inhibitor prevents the hyper-reflexia associated with a rat model of persistent pain. Neurosci. Lett 187: 111-114
Riddoch G 1921 Conduction of sensory impulses from the bladder by the inferior hypogastrics and the central afferent connections of these nerves. Journal of Physiology (London) 54: 134P-135P
Risholm L 1954 Studies on renal colic and its treatment by posterior splanchnic block. Acta Chir. Scand. Suppl. 184: 1-64
Ritter, A..M., Lewin, G.R., Kremer, N.E., Mendell, L.M. 1991 Requirement for nerve growth factor in the development of myelinated nociceptors in vivo. Nature, 350: 500-502.
Robbins A, Berkley KJ, Sato Y. 1992. Estrous cycle variation of afferent fibres supplying reproductive organs in the female rat. Brain Res. 591:353-356
Ross J 1888 On the segmental distribution of sensory disorders. Brain 10: 333-361
Ruch T C 1946 Visceral sensation and referred pain. In: Fulton J F (ed) Howell's Textbook of Physiology. 15th Edition. Saunders, Philadelphia, pp 385-401
Ruffin J M, Baylin G J, Legerton C W Jr, Texter E C Jr 1953 Mechanisms of pain in peptic ulcer. Gastroenterology 23: 252-264
Sengupta JN, Saha JK, Goyal RK. 1992. Differential sensitivity to bradykinin of esophageal distension-sensitive mechanoreceptors in vagal and sympathetic afferents of the opossum. J Neurophysiol. 68: 1053-1067
Sengupta JN, Gebhart GF. 1994 Mechanosensitive properties of pelvic nerve afferent fibres innervating the urinary bladder of the rat. J Neurophysiol. 72: 2420-2430
Sengupta JN, Gebhart, GF. 1994b Characterization of mechanosensitive pelvic nerve afferent fibres innervating the colon of the rat. J Neurophysiol. 71: 2046- 2060
Sherrington C S 1900 Cutaneous sensation. In: Shafer E A (ed) Textbook of Physiology. Vol 2. Pentland, London, pp 920-1001
Silen W 1987 Cope's Early Diagnosis of the Acute Abdomen. 17th edition, Oxford University Press, New York
Sinclair D C, Weddell G, Feindel W H 1948 Referred pain and associated phenomena. Brain 71: 184-211
Sturge W A 1883 The phenomena of angina pectoris and their bearing upon the theory of counter-irritation. Brain 5: 492-510
Takahashi, Y., Nagajima, Y., Sakamoto, T., Moriya, H., Takahashi, K. 1993 Capsaicin applied to rat lumbar intervertebral disc causes extravasation in the groin skin: a possible mechanism of referred pain of the intervertebral disc. Neurosci. Lett. 161: 1-3
Taylor D C M, Pierau Fr-K 1982 Double fluorescent labelling supports electrophysiological evidence for dichotomizing peripheral sensory nerve fibres in rats. Neuroscience Letters 33: 1-6
Theobold G W 1941 Referred Pain: A New Hypothesis. Times of Ceylon, Colombo
Traub RJ, Lim F, Sengupta JN, Meller ST, Gebhart GF. 1994 Noxious distension of viscera results in differential c-fos expression in second order neurones receiving 'sympathetic' or 'parasympathetic' input. Neurosci Lett 180: 71-75
Trimble, K.C., Farouk, R., Pryde, A., Douglas, S., Heading, R.C. 1995 Heightened visceral sensation in functional gastrointestinal disease is not site-specific. Evidence for a generalized disorder of gut sensitivity. Digest. Dis Sci. 40: 1607-1713
Van Hees, J. , and Gybels, J. 1981 C-nociceptor activity in human nerve during painful and non painful skin stimulation. J. Neurol. Neurosurg. Psychiat. 44: 600-607
Wen, J., and Morrison, J.F.B. (1994) The effects of changing urinary composition on 'silent' afferents in the rat bladder. J Physiol. 476: 50P
White J C, Smithwick R H, Simeone F A 1952 The Autonomic Nervous System: Anatomy, Physiology and Surgical Application. Macmillan, New York
Willis, W.D., Coggeshall, R.E. 1991 Sensory mechanisms of the spinal cord, 2nd Edition. New York, Plenum Publishing corp.
Wolf S 1965 Gastric sensibility. In: Wolf S (ed) The Stomach. Springer-Verlag, New York, pp 88-99
Woolf, C.J. & McMahon, S.B. 1985. InjuryInduced plasticity of the flexor reflex in chronic decerebrate rats. Neuroscience., 16: 395404
Woollard H H, Carmicheal E A 1933 The testis and referred pain. Brain 56: 293-303