Structural abnormalities of the nervous system in Crohn's disease and ulcerative colitis

Authors


Professor K. Geboes, Universitair Ziekenhuis Sint Rafaël, Department of Pathology, Minderbroedersstraat 12, B-3000 Leuven, Belgium.

INTRODUCTION

The broad repertoire of functions within the gastrointestinal tract demands close co-ordination between various tissues and regions of the gut. It is no longer reasonable to consider the gut as a compartmentalized organ, in which secretion, motor function, sensory perception and immune function operate independently. In health, recruitment and co-ordination of these systems is largely accomplished through the nervous system but also through the mucosal immune system. Interactions between the immune system and neuroendocrine system are evident in the innervation of immune organs, the effects of neurocrine factors on immune cells and neuroendocrine responses to cytokines derived from immunological cells.1,2 The mucosal immune system may even become more important in certain disease states as illustrated by the trophic effect of inflammation on smooth muscle cells.3

The pivotal role of the enteric nervous system (ENS) in co-ordinating and integrating various components of gut function is reflected in its complex circuitry that receives inputs from the cephalic brain and the autonomic nervous system as well as in its ability to respond to signals from the immune system of the gut.4 The mucosal immune system is responsible for discriminating between useful and harmful components of lumenal content, and for neutralizing noxious agents by humoral or cell-mediated immune responses. The immune system may also recruit other tissues in the gut to assist in containing then evicting noxious agents from the gastrointestinal tract, and this results in changes in motility and secretion. These changes in gut physiology produce symptoms of nausea, vomiting and diarrhoea.4

The pathology of IBD

Inflammatory conditions of the gastrointestinal tract may arise as a result of infection, chemical injury (e.g. acid), auto immunity, connective tissue diseases and for no apparent reason in the case of human idiopathic inflammatory bowel diseases (IBD). Clinically IBD is the most important of these entities. It includes both ulcerative colitis (UC) and Crohn's disease (CD). These terms are descriptors based largely on clinical and morphological characteristics that include differences in terms of penetration of the inflammatory infiltrate within the gut wall as well as the extent of involvement along the gastrointestinal tract. UC is restricted to the colon, is confluent in its extent and exhibits proximal extension over time in most cases. Within the gut wall tissue damage is restricted to the mucosa and lamina propria, except in the more severe case of toxic megacolon. In contrast, CD can occur in any gut region, does not spread confluently but is characterized by `skip lesions' where the bowel is ostensibly normal. The disease may penetrate the entire thickness of the gut wall with involvement of the muscularis propria and externa, resulting in deep fissuring and fistula formation or in wall thickening and stricture formation due in part to the proliferation of myocytes and their deposition of collagen. Vascular lesions are believed to occur more commonly in CD and may be important in the full-thickness involvement of the gut wall.5,6

The nervous system may be involved in IBD as a result of tissue injury, or via the effects of soluble mediators of the inflammatory process that include cytokines, arachidonic acid metabolites and oxygen-derived free radicals. In addition, specific components of nerves may be a target for immune attack, as a result of altered antigenicity following inflammation or through some aberrance of the immune response itself. While the functional involvement of nerves in intestinal inflammation has recently been reviewed,7 to our knowledge, there has been no review of the structural changes that occur in the enteric nervous system in IBD.

The normal innervation of the GI tract

The gastrointestinal tract has an extrinsic and an intrinsic innervation. The extrinsic innervation reaches the gut via the vagus nerve, the mesenteric nerves and the pelvic nerves. These nerves contain a large number of afferent (sensory) fibres, connecting the gut with the brain and other organs and modulate the gut function. The intrinsic innervation or ENS is composed of neurones and glial cells organized in ganglionated and aganglionated plexuses. The ganglia are composed of clusters of neuronal cell bodies and enteroglial cells while the neuronal extensions, also present in the non-ganglionated plexuses, form a regular network of fibres and fascicles with wide meshes. They connect the different ganglia and plexuses. Ganglionated plexuses are present in the submucosa and in the septum between the circular and longitudinal layers of the muscularis propria (Auerbach's plexus). In the submucosa a distinction can be made between the upper submucosal plexus, close to the muscularis mucosae, commonly called Meissner's plexus and the deep submucosal plexus, near the circular muscle, known as Henle's plexus. In the human colon three interconnected ganglionated networks arranged along three different planes can be identified.8 In the literature there are many data, for various mammalian species, regarding the composition of the plexuses and the numbers of ganglia and ganglion cells.9 For the human enteric nervous system however, only limited data are available. In the human, the ganglia of the submucosal plexuses are small, averaging 2–8 cell bodies per ganglion in the normal adult human ileum and 6–8 or 9–21 in the normal adult human colon depending upon the type of ganglion.10,11 Ganglia can be distinguished according to their morphology and position into larger ganglia lying at the junction of fibre tracts, and smaller ganglia including those embedded within fibre tracts, those lying alongside a fibre, and the less common types such as a drumstick type lying at the end of a fibre. Neurones in the submucosal plexuses are fewer and smaller than in the ganglia of the myenteric plexus which are larger and contain more neurones (up to 90) (Fig. 1).10–12

Figure 1.

Normal human colon showing part of the mucosa including the basal part of the crypts, the muscularis mucosae and a ganglion of the upper submucosal plexus (Meissner) with three small neuronal bodies (arrow) (H&E obj × 25).

Aganglionated plexuses are composed of extensions of neurones covered by glial cells. They are present in the muscle coat (mainly the circular coat), in the submucosa and in the lamina propria of the mucosa. In the small intestine the mucosal innervation is more developed than in the colon. In the colon mucosal nerve fibres are scant. They form a plexus which is generally denser near the crypts in the deeper part of the lamina propria. Small S-100 positive cells with tiny processes can be identified in this plexus.13

Different types of neurones can be identified depending on their morphology, chemical content and electrophysiological characteristics. They are responsible for the processing of sensory signals, act as interneurones and as motor neurones connecting with effectors including muscle, secretory cells and blood vessels. A variety of peptides including a.o. neuropeptide Y (NPY) and vasoactive intestinal polypeptide (VIP) can be found in these neurones and several of them contain nitric oxide (NO) synthase.14–16 In the internal submucosal plexus of the human colon only few NADPH-diaphorase positive cells (indicating the presence of NO synthase) are present. In contrast numerous NADPH-diaphorase-stained nerve cell bodies are found in the plexus myentericus and the deep submucosal plexus.16 The majority of these have morphological characteristics similar to those of Dogiel type 1 neurones, i.e. possessing broad flat dendrites and one major axonal projection.16

The enteric glial cells are special cells, different from the Schwann cells found in the peripheral nervous system, and resembling more closely the astrocytes found in the cephalic brain. They express S-100 protein, and show a positive staining for antibodies directed against the glial fibrillary acidic protein (GFAP). In the normal human gut rare enteroglial cells have a membranous expression of major histocompatibility class II antigens (MHC class II) showing that they can also act as antigen-presenting cells.17 Studies of the immune system have indeed shown that MHC class II antigens are involved in antigen presentation and in the interactions between antigen-presenting cells and T lymphocytes. Enteric glial cells possess receptors for cytokines and are also capable of producing cytokines such as interleukin-6 (IL-6). In addition, they possess receptors for neurotransmitters and a recent study has shown that cytokine production by enteric glial cells may be modulated by neurotransmitters.18 These results identify glial cells as having the ability to interface between the neural and immune systems in the gut, and they may therefore have an important role in the pathophysiology of IBD.

The integrity and maintenance of the neurones depends on several factors including interactions of neurotrophins with appropriate receptors.19 One of these, a polypeptide, called nerve growth factor (NGF) can be synthesized by intestinal epithelial cells, smooth muscle cells and fibroblasts. It binds to a specific cell surface glycoprotein receptor, the nerve growth factor receptor, a 75-kD receptor protein.20,21

Functional alterations in IBD

Several disturbances of gut function have been demonstrated in IBD. These include changes in secretion by epithelial cells,22 increased sensory perception23 as well as changes in gut motility.24–28 These findings reflect both inflammation-induced changes in the responsiveness of target tissue such as epithelium or smooth muscle29 as well as the altered innervation of the gut in IBD. The latter includes not only changes in nerves intrinsic to the gut, but also changes in autonomic balance demonstrated in IBD.30,31 The impact of inflammation on intestinal neuro-motor function has been reviewed in detail elsewhere.7

Structural abnormalities of the ENS in IBD

The major structural abnormalities thus far described in IBD are summarized in Table 11. They include architectural alterations of the plexuses, nerve fibre hypertrophy and hyperplasia and alterations of the neuronal cell bodies and enteric glial cells.

Table 1.  Summary of the structural abnormalities of the ENS in IBD Thumbnail image of

Architectural alterations of the regular network of the ganglionated plexuses are difficult to illustrate in severely affected tissue from patients with CD because of the extensive fibrosis. They are however, observed in mildly affected areas and in areas adjacent to severe lesions. They consist of an irregular network of fibres of variable size, with occasional interruptions. These architectural abnormalities are not disease specific. They can also be observed in other chronic inflammatory conditions such as experimental colonic schistosomiasis.32

In CD irregular nerve fibre hypertrophy, an increase in size of the fibres has repeatedly been reported.33–36 Thickened nerve fibres are seen in the mucosa, the submucosa and the myenteric plexus in the ileum and colon. The abnormalities are usually most prominent in the submucosa and have been called `neuromatous lesions'35 (Fig. 2). Mucosal nerve fibre hypertrophy is only seen in areas overlying submucosal fibre hypertrophy. The mucosal nerve fibre hypertrophy cannot reliably be assessed on routinely haematoxylin- and eosin-stained slides but has been demonstrated clearly for CD of the ileum and colon using antibodies directed against neurofilaments, synaptophysin and NGF receptor.37 In UC an increase was noted using antibodies against synaptophysin but staining for NGF receptor was negative. In contrast, in samples from normal mucosa and from cases with non-specific colitis mucosal fibres were rare and usually small.37 Submucosal and myenteric nerve fibre hypertrophy is uncommon in UC. In CD its occurrence in the ileum and colon has been described using routine histology and immunohistochemistry with antibodies directed against neurofilaments12 (Fig. 3). The hypertrophy in CD is mainly found in severely affected areas, and most commonly associated with the presence of an increased inflammatory round cell infiltrate. Plasma cells, lymphocytes and mast cells can be found in the close vicinity of the fibres and granulomas in the plexus myentericus are not uncommon38,39 (Figs 4 and 5). The cellular infiltrate closely associated with the thickened nerve fibres may be very dense. In addition to hypertrophy, an increase in number or hyperplasia is also frequently noted in CD.37,40–42 This observation has been confirmed by objective measurements using point-count morphometry. Application of this method on microdissected flat-mount preparations of the plexus showed an increase of the neural tissue in both ileum and colon in CD and in UC. In CD the lesions tended to be deeper.30 The increase in fibre size and number may thus not be entirely specific for CD but is certainly more frequently seen in this condition and rare in others. In a study comparing the diagnostic value of various microscopic parameters it was shown to be highly suggestive for CD with a significant difference between UC and CD. Furthermore it was shown that these abnormalities can be detected with great reliability.43

Figure 2.

Photomicrograph of the submucosa of the terminal ileum from a case with Crohn's disease showing nerve fibre hypertrophy (arrows) and a granuloma (arrowhead) (H&E obj × 10).

Figure 3.

Photomicrograph of the terminal ileum from a case with Crohn's disease showing nerve fibre hypertrophy associated with inflammation in the submucosa and inner layer of the muscularis propria. The nerves are stained with an immunoperoxydase technique using antibodies directed against neurofilaments and appear as dark fibres (obj × 4).

Figure 4.

Transmission electron microscopy shows the presence of a plasma cell (PC) closely associated with an axon (AX) of the mucosal lamina propria plexus associated with crypt epithelial cells (EP) (× 2000).

Figure 5.

Granulomatous inflammation of the myenteric plexus. The picture shows muscle cells from the external and internal layers (MC), a granuloma with giant cells (G) and an axonal bundle with neuronal cell bodies (arrow) (H&E obj × 10).

Ultrastructural studies of CD, UC and control samples have shown that the nerve fibres or axons appear as swollen, empty, lucent structures, sometimes with large membrane-bound vacuoles, swollen mitochondria and concentrated neurofibrils.44–46 These changes are observed in the ileum and colon and in all plexuses. They can be focal or diffuse. All axons in one nerve bundle, or just a few can be affected (Fig. 6). These ultrastructural alterations of the axons have been observed in CD in affected and non-affected areas, in samples from section margins and in pouchitis. They are considered to be a sign of axonal damage and necrosis. The axonal changes are characteristic for CD and not limited to areas with concurrent inflammatory reaction. They are widespread and extensive. Their specificity for CD has, however, not been confirmed. A mean of 77.8% abnormal axons (range 52–98%) was found in one study of eight cases with CD.47 Yet in another study a mean of 29.9% (range 12.2–60%) abnormal axons was found in cases with CD while in UC a mean of 21–25% abnormal axons were counted. In the same study a mean of 12–63% of abnormal axons was found in controls (one case with diverticulitis, one case of faecal incontinence).48 A chaotic distribution of nerve fibres immunoreactive for tyrosine hydrolase, 5-hydroxytryptamine and NPY and the high frequency of enlarged varicosities in these fibres were further found in the myenteric plexus of Crohn's ileum.12 These findings confirm the ultrastructural observations and might indicate injury of the sympathetic nervous system.

Figure 6.

Nerve bundle filled with greatly swollen empty axon chambers in a biopsy from the ileum of a patient with Crohn's disease (× 8000).

The meaning of the nerve fibre hypertrophy and hyperplasia is unclear. It has been suggested that this is a non-specific event secondary to stenosis.49 While stenosis may be a triggering event, in experimental stenosis the changes occur only late. Furthermore in CD the abnormalities are usually associated with inflammation and not only limited to prestenotic or stenotic segments. The enhanced NGF receptor expression in CD and the hyperplasia support the concept of neural growth and proliferation. The enhanced NGF receptor expression in CD has been confirmed and its neuronal and enteroglial expression was found to be associated with enhanced expression of CD27, another member of the NGF receptor family, in many neurones and glial cells, in follicular and extra-follicular T cells and mucosal plasma cells, indicating a relation with the inflammatory reaction.50 The main distribution of CD27 is on T cells and plasma cells. It is present on the cell surface as a homodimer and binds to CD70 (K1–24) which is present on activated lymphocytes. In addition a positive staining for the Leu-19 antigen, which is related to the neural cell adhesion molecule (NCAM) is also noted on the nerve fibres in samples from patients with CD. The Leu-19 antigen is also expressed on different types of lymphocytes (helper, suppressor and cytotoxic T cells).41

These data support the concept of a relation between the neural hyperplasia and the inflammatory reaction. The hyperplasia may well, however, be a non-specific reaction to tissue and axonal damage of any aetiology, but this hypothesis would not account for the marked differences between UC, non-specific colitis and CD. It is also not clear whether the nerve fibre hypertrophy and hyperplasia have functional consequences or which functional changes they might induce. Some of the fibres with a prominent appearance on routinely stained slides may indeed not be functional. The swollen aspect of the axons observed with ultra structural studies might correspond with the thickened appearance of the fibres on H&E and in some immunohistochemical studies while in fact these fibres are damaged and hence not functional. Yet most of the hypertrophied fibres still show a positive staining with antibodies directed against neurofilaments indicating a more or less well preserved structural integrity.41 Summarizing the nerve fibre abnormalities it appears thus that CD, more than UC, is characterized by two types of abnormalities: damage and hyperplasia in all parts of the ENS, related to inflammation.

Abnormalities of the cellular components of the ENS reported in CD and UC include hyperplasia or an increase in the number of neuronal cell bodies, mainly in the ganglia of the submucosal plexuses, neuronal cell damage and neuronal hypertrophy (Figs 7 and 8). Hypertrophy of neurones with the appearance of prominent organelles, including numerous strands of rough endoplasmic reticulum has been identified in samples from patients with CD but is not consistently confirmed.45,51 Neuronal hypertrophy in Meissner's plexus seems more common in UC than in CD. Hyperplasia of neuronal cell bodies is a more common and constant finding than neuronal hypertrophy.36,52 A three-fold increase in the number of ganglion cells of the ileal myenteric plexus was recorded in a series of 24 cases with CD. The increase was also present in areas not actively involved in the inflammatory process.53 Increased numbers of ganglion cells were reported for the submucosal plexuses in CD when compared with control samples.54 Neuronal hyperplasia of the myenteric plexus is also reported for UC but the data available in the literature are rare and may have been confused by difficulties in the differential diagnosis between granulomatous colitis and genuine UC.55,56 Neuronal cell hyperplasia is certainly more frequently reported for CD and seems more common with a significant, statistical difference when compared with UC. This hyperplasia can be detected with high reliability.43

Figure 7.

Photomicrograph of a ganglion of the submucosal plexus from the colon of a patient with Crohn's disease showing hyperplasia of neuronal cell bodies – arrow in (a). The size of the neurones is normal as shown by a large magnification from another case in (b). (a: H&E obj × 4; b: H&E obj × 40).

Figure 8.

Photomicrographs of a ganglion of the submucosal plexus from the colon of a patient with ulcerative colitis showing hypertrophy of the neuronal cell bodies (a) and damage (b). The hypertrophy (arrow) is recognizable when the size of neuronal cell bodies is compared with the normal appearing cells in Figs 6(b) and 7(a) (arrow). In (b) neuronal damage is apparent by the accumulation of cytoplasmic material in the centre of the cell (arrow) (H&E obj × 40).

In addition to hypertrophy and hyperplasia of neuronal cell bodies, signs of degeneration have been described occasionally in areas of inflammation in CD and UC.34,40,57,58

Summarizing the data, it appears thus that for neuronal cell bodies the same two types of abnormalities (damage and hyperplasia) as for the nerve fibres are found.

Immunohistochemical studies have added important, yet thus far incomplete data to the above findings. In CD a relative increase of nitric oxide positive neurones has been demonstrated in the submucosal and myenteric plexuses.12,59 The increase in myenteric neurones containing NOS and also VIP and PACAP (pituitary adenylate cyclase activating peptide) might indicate an overall increase in the production of the respective substances in CD. This might cause persistent relaxation of smooth muscle of the affected segment.12 An increase in number, size and immunostain and an abnormal pattern of VIP-containing fibres have been reported in CD.54,60,61 The increase was found in histologically normal bowel, in areas with pathology both in specimens of the ileum and colon. In addition an increase in the number of VIP immunostained ganglion cells in the submucous plexus with 6–8 immunoreactive cell bodies in the ganglia in the CD samples and 0–2 in controls was noted. These findings were later confirmed and expanded. Increased VIP, nitric oxide synthase and PACAP immunoreactivity was seen in the myenteric plexus and nerve fibres of the circular muscle layer. Numbers of immunoreactive neurones in the myenteric ganglia were 7–89 for VIP in CD compared to 2–45 in controls and 7–65 compared to 4–73 for NOS.12 No abnormalities were described yet for substance P and metenkephalin containing nerves nor for endocrine cells (neurotensin, somatostatin, enteroglucagon).54 However, with digitized morphometric analysis of colonic surgical specimens a marked decrease in VIP nerve fibres was found in the lamina propria and submucosa in both UC and CD. In the lamina propria but not in the submucosa, the variation in decrease was significantly associated with the severity of the disease.62,63 The abnormal VIP staining pattern described in the rectum in CD tissue samples from involved and uninvolved areas was found to be associated with an increase in VIP content in both inflamed and histologically normal tissue.60 This finding was, however, not confirmed.61

These results are conflicting but they show at least that the VIP innervation pattern is abnormal. The differences between the studies reporting an increased VIP innervation and those showing a decrease could be related to the duration of the disease. Reactive hyperplasia or increase might be more a feature of long-standing, recurrent, disease while decrease might be an earlier phenomenon related to the damage. Furthermore the normal mucosal innervation is more intense in the small intestine than in the colon. The abnormal VIP innervation, which is repeatedly reported might adversely affect the migration of lymphocytes from the blood, especially in the ileum and impair motility and therefore contribute to the pathogenesis and symptoms.64

Data on enteric glial cells are rare, increase in number and the appearance of lipofuscin being the main findings reported in routine studies.45 Immunohistochemistry revealed an increase in MHC class II (HLA-DR) membranous expression on enteric glial cells and glial sheaths of nerve fibres in the mucosa and submucosa (Figs 9 and 10). This increased and aberrant expression is present in macroscopically involved and uninvolved areas, in the colon and ileum and correlates with aberrant or increased epithelial MHC class II expression. The enhanced MHC class II expression on the enteric glial cells is positively correlated with the local intensity of the cellular inflammatory infiltrate, especially with the presence of CD8 + T lymphocytes. The induction of HLA-DP and DQ antigens on glial cells is restricted to areas of moderate and high inflammatory activity.65,66 In addition an enhanced expression for NGF receptor is noted on enteroglial cells in samples from CD. This expression is associated with an enhanced expression of CD27.50

Figure 9.

Ganglion of the submucosal plexus in the ileum from a patient with Crohn's disease showing positive expression for MHC class II antigens on the glial sheath surrounding neuronal cell bodies. (Immunoperoxydase technique with antibodies directed against HLA-DR, technical information: ref. 65, obj × 40.)

Figure 10.

Positive expression for MHC class II antigens on the enteric glial sheath in a nerve bundle of the myenteric plexus from the colon from a patient with Crohn's disease. (Immuno-electron microscopy, technical information: ref. 65, × 6300.)

Overall the morphologic findings suggest that the abnormalities of the ENS in IBD and especially CD are mediated by the inflammatory reaction and factors released from the lymphohistiocytic infiltrate.

Abnormalities of the extrinsic innervation

Apart from the abnormalities of the ENS some alterations of the central nervous system have been associated with CD. The concurrence of multiple sclerosis and idiopathic inflammatory bowel disease, mainly CD, and a more common occurrence of multiple sclerosis in patients with IBD than in the general population have been reported.67,68

There are some reports of cerebrovascular lesions occurring in patients with CD and of an association of peripheral neuropathy (Miller–Fisher's syndrome) and CD.69

Using magnetic-resonance imaging, hyper intense focal white matter lesions ranging from 2 to 8 mm in diameter were found in 20/48 patients with CD (42%), in 11/24 patients with UC (46%) and in 8/50 healthy volunteers. These might represent another extra-intestinal manifestation of inflammatory bowel disease.70 These data need, however, further confirmation.71 While the functional correlate of this finding is currently not known, results from animal studies provide clear demonstrations that intestinal inflammation alters function in the CNS.72

Lesions of the nervous system in animal models

Nerve fibre depletion occurs in the synovium in an antigen-induced and a hydrogen peroxide-induced model of arthritis but not in a synovitis induced by latex spheres. The latter is a mainly macrophage foreign body type reaction, while the former two models are associated with a lymphocytic reaction.73 Experiments with capsaicin pretreatment have shown that sensory nerves exert a protective and healing-promoting function in the gut.74 Yet, although nerve fibre lesions are known to occur in inflammatory animal models32 and although various animal models of inflammatory bowel disease have been developed, the effect of inflammation on the structure of nerve fibres and the enteric nervous system has not been studied extensively neither in induced nor in spontaneous colitis models. Most models resemble more or less UC. The model with a high degree of homology to human IBD is the spontaneous colitis that develops in cotton top tamarins. The most widely used models are those induced, administering toxic chemicals such as acetic acid, formalin, trinitrobenzene sulphonic acid (TNBS)/ethanol but also polysaccharides. Nerve lesions in UC are not so common in humans and therefore their relative absence in animal models is not unlikely, especially as most models are subacute. Nerve lesions are more common in human CD. Animal models resembling CD are, however, rare. The presence of mucosal and submucosal granulomas and other lesions analogous to CD has been reported in a TNBS colitis model and in sulphydryl blocker induced small intestinal inflammation in rats.75,76 Lesions of the smooth muscle cells, mainly hyperplasia, have been reported repeatedly in the TNBS model but lesions of the ENS are not reported, neither for the models resembling UC nor for the sulphydryl, TNBS model or indomethacin enteritis, showing features reminiscent of CD. In most protocols, a proper study of the ENS is, however, not included in the evaluation of the specimen and major attention is given to inflammatory features. Furthermore, the vast majority of work carried out on the ENS in animal models is functional. In the future, a more precise analysis of possible structural alterations of the ENS in animal models of inflammation may be appropriate, especially as it is known that functional alterations are not uncommon.7

DISCUSSION

Structural abnormalities of the ENS and even outside the ENS have consistently been observed in CD and, less frequently, in UC. The alterations of the ENS are characteristic for CD when compared with other diseases although not entirely specific. Axonal damage and nerve fibre hypertrophy of the myenteric plexus can be seen in diverticulitis and other conditions. It is at present not known whether the abnormalities are primary or secondary to the inflammation. The occurrence of lesions in macroscopically uninvolved areas might indicate a more primary event although most of the abnormalities are associated with inflammation. It is further known from experimental studies that inflammation can modulate the nerves.77 Hence it is likely that the abnormalities of the ENS in CD are due to inflammation. Some morphologic features suggest that the inflammatory reaction may be directed primarily against nervous elements. The perineural inflammation is often very dense. Inflammation-associated lesions of the ENS can occasionally be observed at the colonic side of the ileocaecal valve in the absence of other mucosal colonic lesions, while the ileum is inflamed. This pattern might indicate a neuronal pathway for the spreading of the inflammation. Similar lesions are found in the myenteric plexus in otherwise uninflamed section margins in specimens from patients operated for CD and the high disease recurrence rate for CD is well known.78,79

The endoscopic pattern of linear and transverse ulcerations observed in CD and the natural evolution of the aphthoid ulcers towards a linear arrangement in recurrent disease mimic the meshwork arrangement of the plexuses of the ENS and could perhaps also be explained by a spreading via nervous pathways.79 This might also partly explain the variability of the histology of the early lesions described in CD. Classically the histology of the aphthoid lesions, which are considered to be the earliest visible lesions, is described as an epi-thelial defect overlying lymphoid follicles. Yet not all early lesions occur above lymphoid follicles. They can also occur as epithelial patchy necrosis, a mountain peak ulcer, a crypt base ulcer or a vascular-associated summit lesion.80 This microscopic variability shows that the early lesion is not always related to a lymphoid follicle and may correlate with other structures like vessels in the summit lesion or perhaps nerves. The relation between the structural abnormalities of the ENS in CD and a vascular origin of the disease is not immediately clear. Vasculitis, granulomatous arteritis and lymphangitis are not uncommon in CD.81,82 Multi-focal gastro-intestinal infarction has been proposed as the pathogenic mechanism for CD. The thrombogenic injury was explained by interactions between immune cells and the vascular endothelium.5 Such a sequence of events does not exclude involvement of the innervation. Enteroglial cells surrounding the delicate nerves innervating mucosal blood vessels can show MHC class II expression during inflammation.65 An increase in overall perivascular nerve density around mesenteric arteries has been reported in IBD and has been related to a possible vasoconstrictor response.83 Experimental studies suggest that sensory nerves can exert a protective and healing-promoting effect in the gut through regulation of mesenteric and mucosal blood flow.74

There seems to be a difference between CD and UC in the type of lesions observed in the ENS. Nerve fibre damage, hypertrophy, reactive nerve fibre hyperplasia and neuronal cell hyperplasia seem more common in CD whereas neuronal hypertrophy in Meissner's plexus seems more common in UC.40,43 That nerve fibre hyperplasia is genuine is supported by morphometric analysis of microdissected flat-mount preparations of the tissue.42 The increased expression of NGF receptor and NCAM on the fibres is a further argument for this hypothesis. The NGF/NGF receptor system seems involved in CD and experimental studies have shown that lymphoid tissue can induce NGF-dependent and NGF-independent neurite outgrowth.84 The hyperplasia can be related to the enteric glial cells, which seem to be activated as indicated by increased MHC class II expression and which might synthesize NGF. The alterations of the NGF receptor and MHC Class II expression on glial cells are further positively correlated with inflammation. Even if the structural abnormalities of the ENS observed in CD are secondary to the disease and the result of inflammation, the pattern indicates the existence of important differences between CD and UC and between CD and other forms of inflammatory bowel disease.85 Furthermore there is one case report describing intestinal motility disturbances due to granulomatous visceral neuropathy of the colon in a patient with non-small cell lung carcinoma, having no signs of Crohn's disease. This case report supports the possibility of a more than secondary phenomenon.86 Viruses have been implicated in the pathogenesis of sporadic idiopathic intestinal pseudo-obstruction.87 This is another possible pathway by which the ENS might be affected in CD and viruses have been proposed as a causative factor on the basis of epidemiological, pathological and case-control studies.88,89

Thus, the inflammation-induced effect on nerves at remote non-inflamed sites in the gut is probably responsible for the widespread nature of altered physiology in IBD. Taken together, the distortion of normal neural innervation patterns, the presence of autonomic imbalance in IBD patients and its persistence after colectomy for UC, suggest that nerves may play a role not only during active inflammation, but also in the development of the disease and predisposition to postoperative recurrence.

Ancillary