Gastrointestinal involvement in von Recklinghausen's disease occurs in three principal forms: hyperplasia of the submucosal and myenteric nerve plexuses and mucosal ganglioneuromatosis which leads to disordered gut motility; gastrointestinal stromal tumours showing varying degrees of neural or smooth muscle differentiation; and a distinctive glandular, somatostatin-rich carcinoid of the periampullary region of the duodenum that contains psammoma bodies and which may be associated with phaeochromocytoma. This review describes the histopathological features of these lesions and discusses potential pitfalls in their differential diagnosis. Their accurate identification has significant implications for clinical management and may even provide the first pointer to the diagnosis of neurofibromatosis.

Current evidence from experimental animals suggests that the colonic crypt is a clonal unit derived from, and maintained by, a single cell (Griffiths et al., 1988). Treatment of normal female TO strain mice with the colon-specific carcinogen dimethylhydrazine (DMH) results in a histochemically demonstrable loss of activity of the X-linked enzyme glucose-6-phosphate dehydrogenase (G6PD) in single, randomly distributed colonic crypts. This phenotypic change is uniform within the affected crypts, and its frequency is related to the dose of DMH treatment, strongly suggesting that it is the result of a carcinogen induced mutation at the G6PD locus on the active X-chromosome of a single primary crypt stem cell (Griffiths et al., 1988).A similar crypt-restricted phenotypic change has been demonstrated in normal human large bowel mucosa using the mild periodic acid-Schiff (mPAS) technique (Sugihara & Jass, 1986). This histochemical method distinguishes between 0-acetylated sialomucins, which are mPAS-negative, and non-0-acetylated sialomucins which stain a magenta colour (Veh et al., 1982). Sugihara and Jass (1986) found that in most human colons the goblet cell mucus is O-acetylated and mPAS-negative but that in a small proportion of individuals it is non-O-acetylated and diffusely mPAS-positive. However, in some cases with mPAS-negative colonic mucus glycoproteins they observed scattered individual mPAS-positive crypts. They were uncertain of the significance of this but interpreted it as a 'functional metaplasia'. A similar phenomenon has been identified by Hughes et al. (1986), who described a focal loss of immunoreactivity to a monoclonal antibody (3NM) directed against colonic goblet cells, again confined to scattered crypts in the human large intestine.Because the crypt-restricted phenotypic alterations described in these two reports are similar to the mutageninduced loss of G6PD activity in the colonic crypts of experimental mice, we considered that they may result from crypt stem cell mutation. If this were correct, the frequency of affected crypts would increase with exposure to environmental mutagens, and be age-related. The change might also occur more frequently in subjects with colorectal cancer. We have investigated this using the mPAS technique on histologically normal areas of resection specimens of the sigmoid colon and upper rectum from 30 adults (mean age 71.8 years, range 55-91) with primary large intestinal adenocarcinoma, 30 age-matched controls (mean age 71.3 years, range 52-90) with benign conditions of the sigmoid colon (diverticular disease 27, volvulus 3), and 18 infants or children (mean age 25 months, range 2 days to 9 years) with Hirschsprung's disease (16), intestinal atresia (1) or colonic duplication (1). One block of formalin-fixed, paraffin-embedded tissue from each resection margin of the colectomy specimens was taken and 5 im histological sections were cut at 50pm (Veh et al., 1982). In nine cases (three from each patient group) two adjacent levels were cut, one being stained by t...

Mild periodic acid-Schiff (mPAS) staining distinguishes O-acetylated from non-O-acetylated sialoglycoproteins. In human colonic mucosa, individuals possess one of three phenotypes: uniformly mPAS-positive (non-O-acetylated), uniformly mPAS-negative (O-acetylated), and negative with infrequent scattered positive crypts. This is due to a polymorphism in a single autosomal gene (oat). Discordant crypts have not been found in children's colons, suggesting that they result from somatic mutation in heterozygous individuals. We now present evidence to confirm this based on a study of radiation-induced changes. Comparison of mPAS staining of large intestinal mucosa from patients given radiation 4 weeks before surgery for carcinoma of the rectum with matched controls receiving surgery alone showed a similar phenotype distribution, but information irradiated patients showed an increased frequency of discordant crypts (irradiated vs. non-irradiated 14.5 +/- 8.2 x 10(-4) vs. 6.1 +/- 4.2 x 10(-4)). When these were classified as wholly or partially involved by the aberrant phenotype, the increase was most marked in partially involved crypts (7.5 +/- 4.5 x 10(-4) vs. 0.3 +/- 0.5 x 10(-4), Mann-Whitney U, P < 0.005). Two patients receiving radiotherapy many years before colectomy showed a very high total discordant crypt frequency but relatively few partially affected crypts. Studies of somatic mutation in colonic or small intestinal crypts following a single dose of mutagen in mice have shown early partial crypt involvement by the mutated phenotype and later complete crypt involvement.(ABSTRACT TRUNCATED AT 250 WORDS)