Abnormal expression of M1/MUC5AC mucin in distal colon of patients with diverticulitis, ulcerative colitis and cancer



The abnormal expression of gastric M1/MUC5AC mucin in precancerous lesions and colon cancer evidenced by immunohistochemistry led us to check for its presence in the mucus obtained directly from patients undergoing surgery for cancerous (adenocarcinoma) or inflammatory (diverticulitis or ulcerative colitis) diseases. In parallel, the authors quantified aberrant crypt foci (ACF) and their immunolabelling by M1/MUC5AC in mucosae of cancer and diverticulitis patients. Immuno-Radio-Metric Assay of M1/MUC5AC mucin developed by the authors was used to detect M1/MUC5AC mucin in the colonic mucus scraped from surgical specimens. M1/MUC5AC mucin was detected in the mucus of 51/69 (74%) patients with colon adenocarcinoma, versus 7/27 (26%) patients with diverticulitis (threshold: 30 units of M1 mucin per mg protein, area under ROC curve: 0.80). M1/MUC5AC was present in significantly (p < 0.001) larger amounts in the mucus of cancer versus diverticulitis patients. All (10/10) patients with ulcerative colitis tested showed levels above the threshold and their mucosae were strongly labelled with the anti-M1/MUC5AC antibody by immunohistochemistry. Patients with cancer exhibited 3 fold more ACF than those with diverticulitis, but no significant difference was observed in the mean size and M1/MUC5AC expression pattern of ACF between these two groups. The expression of M1/MUC5AC was in correlation with their size. In macroscopically normal mucosa, ACF were the most important source of M1/MUC5AC mucin. Testing of M1/MUC5AC can enhance the detection of precancerous lesions and colon cancer. © 2007 Wiley-Liss, Inc.

The search for specific marker of early human colon carcinogenesis focused very early on sialomucin.1 This pioneering work led us to a better characterisation of the modification of mucin secretion evidenced by classical histologic staining. For that purpose, we developed antisera raised against gastrointestinal and ovarian mucins.2, 3 One of these antibodies against gastric mucin reacted specifically on histological sections from colon adenomas and not from normal colon, characterising a mucin we called M1. This M1 mucin was also expressed early in the rat during colon carcinogenesis induced by DMH.4 For a better characterisation of M1 mucin, we have produced Mabs against gastric mucin by immunohistological screening method using their specific reactivity on adenoma and their nonreactivity on the normal colon.5, 6 We obtained 12 different Mabs against the gastric M1 epitopes and later demonstrated that the M1 epitopes were coded by the M1/MUC5AC gene.7, 8, 9 Using these antibodies, we developed an IRMA for M1/MUC5AC mucin.10 This IRMA assay of M1/MUC5AC mucin in pancreatic cyst fluids has been extensively used by us for the last 15 years to improve the diagnosis of mucinous cystadenoma neoplasms.11 We also showed using immunohistochemical methods the expression of gastric M1/MUC5AC mucin in colon adenomas, hyperplastic polyps and in the colon transitional mucosa.5, 6, 12, 13 Abnormal expression of M1/MUC5AC mucin during the early stages of colorectal carcinogenesis has been confirmed by several groups using biochemical, immunohistochemical and molecular biology techniques.14, 15, 16, 17 On the other hand, colon carcinogenesis induced by intrarectal deposits of the direct carcinogen MNNG in rats enabled us to detect the abnormal expression of this M1/MUC5AC mucin 10 weeks after treatment.18 Recently, we have shown that the M1/MUC5AC mucin is secreted in the aberrant crypt foci (ACF) in humans and rats, confirming that this biomarker is expressed very early during colon carcinogenesis.18 However, this mucin has never been assayed in the mucus isolated directly from the colonic surface, which contains mainly MUC2 mucin in the normal state.19 For this purpose, we have studied samples of mucus directly scraped from fresh unwashed surgical specimens. We studied 69 cases of distal colorectal adenocarcinomas as compared with 27 cases of diverticulitis (a nonmalignant pathology currently cured by surgery). In addition, we studied specimens from 10 patients suffering from ulcerative colitis, a chronic inflammatory disease considered as a premalignant state. In parallel, for a better understanding of the change in expression of gastric M1 mucin in the macroscopically normal colon mucosa, we analysed the expression of this mucin in normal and aberrant crypts by immunohistochemistry. We have recently demonstrated that the progression of tumours arising from large ACF is associated with the MUC5AC expression during rat colon MNNG carcinogenesis.20 Therefore, we focused our attention on the ACF, their number and the number of ACs21 in the mucosae of patients with colon cancer or with diverticulitis, as nonmalignant and nonprecancerous controls. Immunohistochemical studies of the normal mucosa of 31 normal colons from our embedded collection were performed in the same conditions for comparison with mucosae of patients with cancer and inflammatory diseases.


AC, aberrant crypt; ACF, aberrant crypt foci; DMH, dimethyl hydrazine; IRMA, immunoradiometric assay; MAb, monoclonal antibody; MNNG, N-methyl-N′-nitro-N-nitrosoguanidine; ROC curve, receiver operating characteristic curve; PBST, PST-Tween.

Material and methods

Human mucus and tissues

Colorectal mucosa samples were obtained from 69 patients with adenocarcinoma (mean age, 63) and from 27 patients with diverticulitis (mean age, 60) and 10 patients with ulcerative colitis (mean age, 35), taken from freshly resected surgical specimens. The study of cancer patients was performed on the distal colon and rectum of sporadic adenocarcinoma (the patients with familial adenomatous polyposis and hereditary nonpolyposis colorectal cancer genetic diseases which are systematically checked in the hospital were excluded, as well as patients treated by radiotherapy before surgery). The ulcerative colitis patients studied underwent surgery after more than 5 years of the disease (up to 30 years), 5 of them were operated for the second time.

Before washing the specimen, as much as possible mucus was scraped from the mucosa with a cell scraper (upstream the tumour for cancer specimens), diluted (2–4 times vol/vol) in PBS-Tween buffer [(0.01 mol/l phosphate buffer, pH 7.4, 0.15 mol/l NaCl and 0.1% Tween 20 (polyoxyethylene sorbitan monolaurate, Sigma-Aldrich, USA)] and frozen until assayed. Whenever possible, strips of macroscopically normal mucosa were dissected from the underlying muscularis mucosa, fixed flat on coded filter paper and immersed for 2–3 days in 95% ethanol solution. For the screening of ACF, the mucosa was stained using 0.05% methylene blue in distilled water for 1 min, and the mucosal side was directly observed at a magnification of 40× by the method of Bird.21 ACs were distinguished from the surrounding normal crypts by their slit-like opening, increased size, blue staining and pericryptal zone. The surface of each ACF was measured under a microscope with camera using “Visiolab 1000” software (Explora Nova, La Rochelle, France) and the number of ACs were counted. Small colon fragments (approximately 1 cm2) of trapezoidal shape were dissected around ACF from the entire colon mucosa for immunohistological examination. The trapezoid sample shape allowed the orientation of the ACF during the paraffin embedding, so as to obtain longitudinal sections of colon crypts (Fig. 2a). The remaining pieces of mucosa were coiled into Swiss rolls5 (Fig. 2b) and embedded into paraffin. The Swiss rolls allow the examination of the whole length of a colon sample on each section. In addition, we examined 31 normal colons of our library, stored in paraffin blocks.5 These samples were obtained from kidney donors (mean age, 26). Sections were stained with the anti-M1/MUC5AC Mabs under the same condition as used for the other samples of patients.


We used the 6 anti-M1 antibodies recognising epitopes we have previously mapped on the MUC5AC gene product [namely, 19M1, 21M1, 62M18; 1-13M1, 2-11M1, 9-13M19]. In addition, we used the 45M1, which immunoprecipitated the 6 other M1 epitopes and which is now currently used by others as an anti-MUC5AC Mab. Recently, the group of Dr. G. Hansson (University of Goteborg, Sweden) localized the 45M1 epitope in the C-terminal region of the MUC5AC apomucin (personal communication). Therefore, all these M1 epitopes can be regarded as belonging to the MUC5AC apomucin. Polyclonal antibodies were obtained by rabbit immunization with a sequence specific of MUC2 mucin peptide (EDPEEEVAPASC of the D1 domain located in the N-terminal part of the MUC2 gene22). These antibodies have been demonstrated to be specific for this peptide sequence using ELISA (not shown) and stained intestinal goblet cells of the normal human gastrointestinal tract.

IRMA of gastric M1 mucin

The IRMA was performed as described before10 with some modifications. Briefly, a double antibody sandwich system was performed using a solid phase: acrylonitrile butadiene styrene polymers stars and special polystyrene tubes, kindly provided by Cis-Bio International, Saclay, France. For the first layer, stars were coated with 1-13M1 antibody as described10 (coated stars could be stored up to 6 months at 4°). The second layer was the antigen: 300 μl of solution to be tested or various concentrations of M1 standard diluted in PBST were loaded onto the stars, and the tubes were incubated overnight at 4°C. After washing, the third layer consisting of 300 μl of a mix of purified IgG of the 7 antibodies (1-13M1, 2-11M1, 9-13M1, 19M1, 21M1, 45M1 and 62M1) labelled with iodine 125 (Perkin Elmer, Boston, MA) was incubated overnight at 4°C. After washing, the radioactivity was counted and the quantity of M1/MUC5AC mucin was determined against a standard curve. Standard M1 antigen was prepared from the lyophilized fluid of an ovarian mucinous cyst of a pure endocervical type containing a great amount of gastric M1 mucin.10 The preparation of a standard batch was done using 20 mg of powder first resuspended in 9 mol/l urea using a Potter to grind powder aggregates, cleared by centrifugation and the volume adjusted to 100 ml in PBST. Each new batch of standard was calibrated against the former one; for comparison purposes, we arbitrarily considered that at saturation the preparation contained 10,000 M1 units/ml.

Protein assay

To solve problems linked to the presence of coloured material in the mucus (blood, feces, bile salts) we used the “Non-Interfering Protein Assay”™ kit (Genotech, St-Louis, MO) which allowed reproducible results.


Deparaffinised sections (ACF and colon mucosae coiled into Swiss rolls) were incubated in “R-Buffer A, pH 6.0” solution (Pickcell, Amsterdam, The Netherlands) in the “Retriever 2100™” device (Pickcell, Amsterdam, The Netherlands) which heats the samples at 120°C for 20 min in an automatically controlled pressure. The sections were thereafter preincubated for 3 min in PBST, and then incubated for 30 min with a mixture of undiluted hybridoma supernatants of the 7 anti-M1/MUC5AC Mabs (equal volume of each Mab). After 3 rinses in PBST, the sections were incubated for 30 min with peroxidase-coupled anti-mouse IgG (H+L), Santa Cruz, CA diluted 1/200 in PBST. After 3 rinses in PBST, peroxidase activity was revealed by incubating for 4 min in a solution freshly prepared with 60 mg of amino-ethylcarbazol (Sigma-Aldrich, MO) diluted in 5 ml of dimethylformamide, 100 ml of 0.05M acetate buffer pH 5 and 0.1 ml of 30% solution of hydrogen peroxide (Sigma Aldrich, MO). Cell nuclei were stained for 5 min with 1% Mayer's hematoxylin (Fluka, Switzerland). The specificity of the immunoreactivity was verified by inhibition of staining after absorption of 1 ml of the hybridoma supernatant with 10 mg (dry weight) of lyophilised fluid of an ovarian mucinous cyst rich in gastric M1 mucin.2

Score of labelling


One ACF per patient, taken at random from the embedded ACF collection, was included in this study, and the number of ACs in each ACF studied was noted. Immunoreactivity was scored according to the percentage of positive goblet cells in the ACF: score 1: 1–5%; score 2: 5–25%, score 3: 25–50% and score 4 more 50% of M1/MUC5AC positive goblet cells. The results obtained by one observer are shown. Another observer obtained similar results independently.

Macroscopically normal mucosae.

To estimate the expression of M1/MUC5AC in the normal mucosa coiled into Swiss rolls, we measured the length of each mucosa strip using the “Visiolab 1000” software. We counted under the microscope the total number of positive normal crypts considering that a crypt is positive for M1/MUC5AC if one goblet cell or more was strongly stained by the anti-M1/MUC5AC Mabs. The results were expressed as the number of positive normal crypts per mm of mucosa.

Statistical analysis

All individual results (Swiss rolls or ACF) were included in data bases and the statistical significance was checked using the “Statview” software (the Kruskal–Wallis or Mann–Whitney tests were used as reported in the results and p values indicated). The alpha level of 0.05 was considered significant.

Assuming that the diverticulitis patients could be used as controls for cancer pathology, we checked the accuracy of M1/MUC5AC mucin amounts obtained with the IRMA analysis in colonic mucus to discriminate between cancer and noncancer patients. The results of mucus were compared using ROC curve (Receiver Operating Characteristic Curve, plotting the true positive rate against the false positive rate for the different possible cutoff points of diagnostic test) and the threshold giving the least false negative and false positive results was determined using the “Rockit v.0.9.1 beta” software.


IRMA of M1/MUC5AC mucin in the colonic mucus of patients

Figure 1 shows a box plot of the M1/MUC5AC contents of colonic mucus in the 69 cases of adenocarcinoma, 27 cases of diverticulitis and 10 cases of ulcerative colitis tested. The area under ROC curve (not shown) was 0.80 to discriminate adenocarcinoma from diverticulitis patients for a threshold of 30 M1 units/mg protein. The 10 cases of ulcerative colitis patients were all above the threshold. Using this threshold, 74% (51/69) of cancer patients were positive, as compared with 26% (7/27) of diverticulitis patients. The difference of M1/MUC5AC mucin concentration between diverticulitis and adenocarcinoma or ulcerative colitis groups was highly significant (p < 0.0001 for each comparison, Mann–Whitney test); in contrast, there was no difference between adenocarcinoma and ulcerative colitis groups (p = 0.06).

Figure 1.

Box plot of M1/MUC5AC assay (arbitrary M1 units per mg protein) in the mucus of patients suffering adenocarcinoma (ADK, n = 69), diverticulitis (DIV, n = 27) and ulcerative colitis (UC, n = 10).

Expression of M1/MUC5AC mucin in the macroscopically normal colonic mucosa


We analysed mucosa samples from 40 cases of adenocarcinoma and 25 cases of diverticulitis (total mucosal surface studied: 1166 and 813 cm2, respectively). Table I shows that the surface occupied by the ACF per cm2 of mucosa examined (ACF density) was more than 3 fold higher in colon cancers than in the diverticulitis. In contrast, no significant difference between these groups was found for the mean surface of ACFs. This indicates that the difference in the surface occupied by the ACF reflected a higher number of ACF in cancer patients. Indeed, the mean number of ACF/cm2 of mucosa examined was significantly higher in cancer than in diverticulitis patients (Table I).

Table I. Number and Surface of ACF in Colon from Patients with Adenocarcinoma (ADK) and Diverticulitis (DIV)
PatientsADK (n = 40)DIV (n = 25)
  • 1

    Significant difference, p = 0.0019, Mann–Whitney test.

  • 2

    No significant difference, p = 0.116, Mann–Whitney test.

  • 3

    ACF density was calculated for each patient as the ratio of total ACF surface versus total surface of sample mucosa studied. The difference between ADK and DIV, was highly significant (p = 0.0004, Mann–Whitney test).

Sample of mucosa studied/patient (cm2, mean ± SD)29 ± 1432 ± 14
ACF observed/sample (number, mean ± SD)7.7 ± 7.93.5 ± 5.9
Number of ACF/cm2 of sample (mean ± SD)10.311 ± 0.3420.106 ± 0.155
ACF surface (cm2, mean ± SD)20.012 ± 0.0120.006 ± 0.004
ACF density/sample (mean ± SD)33.37 ± 4.58 × 10−30.89 ± 2.29 × 10−3

Figure 2a shows a sample of mucosa stained with methylene blue with an ACF (arrow). Figure 2b shows an hematein–eosin staining of a Swiss roll section of mucosa (see the Methods section). Figure 2c shows that M1/MUC5AC mucin was expressed in the upper part of the crypts. The M1/MUC5AC pattern of the ACF was identical in the macroscopically normal mucosae of diverticulitis and cancer patients. In one case a microadenoma (arrow), unnoticed at the macroscopic level (all visible adenoma are removed from the surgical specimen by the pathologist for routine examination), was observed adjacent to an ACF (Fig. 2d). The ACF strongly expressed the M1/MUC5AC mucin but no immunostaining was observed in the microadenoma in this section (Fig. 2d).

Figure 2.

(a) Trapezoid sample shape containing an ACF (arrow); bar: 500 μm. (b) Swiss roll of macroscopically normal colon mucosa stained with hematein–eosin (bar: 500 μm). (c) ACF stained with the anti-M1/MUC5AC Mabs. Unstained normal crypts were observed adjacent to the ACF (arrows); bar: 200 μm. (d) On the right, ACF stained with the anti-M1/MUC5AC Mabs, and on the left, microadenoma (arrow); bar: 200 μm. (e) Colon mucosa of a patient with UC stained with the anti-M1/MUC5AC Mabs and with anti-MUC2 antibodies in (f); bar: 200 μm.

The expression of M1/MUC5AC mucin in ACF did not show any difference between cancer and diverticulitis patients (p = 0.28, Mann–Whitney test). Therefore, the data from adenocarcinoma and diverticulitis groups were pooled. Figure 3 shows that the expression of M1/MUC5AC mucin was dependent on the number of ACs in a focus and increased significantly with the size of ACF (p = 0.016, r = 0.43).

Figure 3.

Score of the ACF corresponding to the level of M1/MUC5AC expression according to the size of ACF: Individual values of scores of one ACF (taken at random) per patient are plotted against the number of ACs in the focus (results of 31 ACF representative of 19 patients with cancer and of 12 patients with diverticulitis were pooled). ACF containing more than 50 crypts were designated as 51 since it was not possible to count all crypts. One point may represent more than one ACF. Regression curve is indicated (p = 0.016, r = 0.43).

Normal crypts.

To estimate the contribution of the normal crypts to the production of M1/MUC5AC present in the scraped mucus, we counted the number of M1/MUC5AC positive crypts per mm of mucosa in three groups: patients with adenocarcinoma (ADK, n = 37) patients with diverticulitis (DIV, n = 25) and normal individuals (N, n = 3). (tissues from brain-dead patients of our embedded collection). Figure 4 shows a box plot of the number of labelled crypts per mm. The labelling of the normal crypts was weak in the three groups: The median values were 0.428, 0.577 and 0.012 labelled crypts per mm in ADK, DIV and N groups, respectively. Kruskall–Wallis test suggested a difference between the three groups (p = 0.049) but no significant difference was found when comparing ADK, DIV and N groups 2 by 2 with Mann–Whittney test. Since the normal individuals (N group) were younger than ADK and DIV patients (mean ages: 26, 60 and 63, respectively), this marginal difference may reflect minute changes due to ageing. Indeed, using magnifying chromoendoscopy, it was reported that some ACF were present in normal individuals and that their number increased with age.23

Figure 4.

Box plot of the expression of M1/MUC5AC in Swiss rolls of histologically normal mucosa. The numbers of labelled normal crypts per mm of mucosa in cancer patients (n = 37, median = 0.48 labelled crypts per mm mucosa) diverticulitis patients (n = 25, median = 0.57 labelled crypts per mm mucosa) and normal individuals (n = 31, median = 0.012 labelled crypts per mm mucosa) are shown.

No correlation was found at the individual level between the expression of M1/MUC5AC in sample tissue studied (ACF and normal crypts) and the amount of M1/MUC5AC detected in the mucus by IRMA. This apparent discrepancy may be due to the fact that scraped mucus is representative of a larger area of epithelium than the samples of mucosa studied. In addition, the M1/MUC5AC immunolabelling of mucosa may decrease after mucus secretion.

Ulcerative colitis.

The mucosa of ulcerative colitis patients was in a too bad state to be dissected from the muscular layer and therefore only standard immunohistochemistry could be performed. A strong expression of M1/MUC5AC mucin was found by standard immunohistochemistry throughout the mucosae when undamaged crypts were present (Fig. 2e). The intensity of the labelling of the goblet cells was not homogenous and some goblet cells did not express M1/MUC5AC mucin. On the contrary, all the goblet cells were stained with the anti-MUC2 antibodies (Fig. 2f).


We show here the presence of gastric M1/MUC5AC mucin in the colonic mucus scraped directly from surgical specimens obtained from colon cancer patients. In addition, we show that the ACF are the most important lesions producing M1/MUC5AC mucin in the macroscopically normal mucosa, and that their number (but not their size) is 3 fold higher in patients with adenocarcinomas than in patients with diverticulitis.

Some of the diverticulitis mucosae were found to be positive in the assay of M1 mucin in mucus (7/27, i.e., 26%). This proportion is much lower than found in patients with cancer, but it may indicate that the diverticular disease is not equivalent of normal control. However, since surgical specimens from true controls (i.e, without pathology) are not available, we chose this pathology that is not considered cancer-related. Colons with diverticulitis secreting M1/MUC5AC mucin in detectable amounts may be in a premalignant state, considering that the age of patients suffering diverticulitis or cancer tested by us (mean ages of 60 and 63, respectively) were in the same range, and that the incidence of both diseases increases with age. We have previously performed an immunohistological study in individuals without any disease of the colon (mean age, 34), showing that the M1/MUC5AC mucin was usually not expressed in the normal colon. This observation was confirmed in the present study in conditions that improved the sensitivity of immunohistological examination (see Methods). However, we could not detect the gastric M1/MUC5AC mucin in 18/69 cases of adenocarcinoma (26%) that were under the threshold. This may be due to degradation of mucins in the specimens tested or other pitfalls. Alternatively, this may be related to the fact that there are several distinct colorectal carcinogenesis pathways24, 25 including such where the MUC5AC gene is not abnormally expressed.26

Few studies comparing ACF in diverticulitis versus colorectal cancer patients were previously reported. Dolara et al.27 claimed there was no difference in the number of ACF per surface of mucosa, but did not measure the surface of ACF, and studied a smaller number of patients than in the present study. In addition, the group of cancer patients in Dolara's report was not restricted to distal colon diseases where diverticulitis is generally located (sigmoiditis). Indeed, it was reported that ACF occur more frequently in distal colon28, 29 and that the ACF-adenoma-cancer sequence is more relevant for the distal colon. In contrast, our studies are in accordance with the results of Nascimbeni et al.30 who reported in a group of 103 cases that colon of patients with cancer had higher frequency of ACF than colon of patients with diverticulitis.

The aberrant expression of gastric M1/MUC5AC mucin was previously demonstrated in colonic polyps12 (hyperplastic polyps and adenomas), in the transitional mucosae,12, 13 as well as in tumours.31 We show here that ACF are an important source of M1/MUC5AC, as shown by immunohistochemistry. This is in accordance with our recent results on rat colon MNNG carcinogenesis.20 The specific presence of M1/MUC5AC in human ACF is a further argument for the preneoplastic characteristic of this gastric mucin in colon. It is likely that most of the gastric M1/MUC5AC detected by IRMA in the mucus originates from the ACF since in the case of cancer the mucus was scraped upstream of the tumour, and since in the case of diverticulitis patients, no tumour or precancerous lesions were present in the sample studied. The differences in the amount of M1/MUC5AC in mucus of cancer as compared to diverticulitis patients parallels the differences in the number of ACF in the two groups, suggesting a causal relationship between the number of ACF and the level of M1/MUC5AC detected in the mucus. It was not possible to test ACF in ulcerative colitis patients, but histological examinations showed that the expression of M1/MUC5AC was distributed more widely throughout the mucosa (see Fig. 2e), suggesting the possibility of another mechanism for the production of this mucin. Indeed, cancer develops in chronic inflammatory bowel diseases following genetic and epigenetic alterations that may differ from those described in sporadic cancer.32, 33

The abnormal expression of M1/MUC5AC mucin described herein occurs in two apparently different pathological processes, cancer and inflammation, probably implicating two different mechanisms. During colon carcinogenesis, the expression of M1/MUC5AC could be regarded as the re-expression of this foetal mucin. Indeed M1/MUC5AC mucin was detected from the fourth months of gestation and was maximal during the sixth month.5 Such a foetal feature of differentiation in ACF has been also described by Fujitmitsu showing a fission mechanism of glands formation common to ACF and foetal intestinal mucosa in the rat.34 On the other hand, a different mechanism could be involved in the inflammatory process. The MUC5AC promoter was shown to be activated by various mediators of inflammation,35 but almost all these studies were performed in the airway system, where the MUC5AC gene is normally expressed. It is not known whether bacterial infection and/or inflammation may cause selective activation of the MUC5AC gene in the colon where it is not expressed in normal conditions, in contrast to MUC2.19, 36 It was shown in colon cell lines that IL-1 and IL-6 modulated the expression of mucin genes and carbohydrate content of mucins, but results were different depending on the cell lines used.37, 38 In addition, some bacterial products are able to induce the secretion and synthesis of mucins in cultured colon cancer cells.39 In this respect, we show that the gastric M1/MUC5AC mucin is present in the mucus of ulcerative colitis patients undergoing surgery. It was previously reported that the MUC5AC mucin is present in the ileoanal reservoir of patients after coloproctectomy for ulcerative colitis.40 Recently, the expression of MUC5AC mucin was reported by immunohistochemistry in children with ulcerative colitis.41 In contrast, no expression of MUC5AC was found in biopsies of patients undergoing colonoscopy with a known or probable diagnosis of ulcerative colitis.42 These discrepancies may be due to the fact that ulcerative colitis patients undergoing surgery are in a more advanced stage of the disease, after failure of medical treatment. In our series, the fact that the 10 patients suffering from ulcerative colitis tested were positive for M1/MUC5AC mucin suggests that long term chronic inflammation may induce the production of this mucin in the colonic epithelium. Therefore, the abnormal expression of M1/MUC5AC mucin in ulcerative colitis may occur late in the evolution of this disease, while it is a precocious event in sporadic colon cancer. In addition, the expression of MUC5AC in regenerating areas close to ulcerations in Crohn's disease suggests its implication in tissue repair mechanisms.43 Significantly, patients suffering ulcerative colitis are regarded as being at high risk for developing colon cancer, and this pathology may be considered as premalignant, especially after long term disease.32, 44 The expression of M1/MUC5AC in the mucosa of patients with ulcerative colitis is worth further examination, for example on biopsies done for the monitoring of this disease. Indeed, biomarkers could help to detect premalignant dysplasia for the surveillance of ulcerative colitis evolution.45

In the case of diverticulitis disease, scarce studies have reported any relationship between this inflammatory/infectious disease and cancer. Two retrospective studies of the same group have reported a two fold increase of left colon cancer risk in diverticulitis patients.46, 47 However, this disease is not currently regarded as premalignant.

In the normal colon, the mucin secreted is essentially the product of the MUC2 gene.19, 36 Our results show that M1/MUC5AC mucin is secreted in the colonic mucus of cancer patients and that it is possible to detect it by IRMA. We also show that M1/MUC5AC is a specific marker of ACF in the human mucosa, in agreement with our previous report using MNNG carcinogenesis in the rat.20 Recent studies using high magnification chromoscopic endoscopy on a large number of patients emphasized the usefulness of ACF in the rectum as markers of neoplasia.48 The M1/MUC5AC mucin is secreted into the lumen of colon of patients with cancer, together with the colonic mucin MUC2. The assay of colonic MUC2 mucin is not available, and it is likely that the gastric mucin represents only a very small fraction of the mucins present in the mucus gel. However, antibodies developed by us are able to select the gastric mucin diluted in a large quantity of colonic mucin. The detection of M1/MUC5AC mucin may be the basis for future research to improve the testing for colon cancer screening in asymptomatic population, for example during rectoscopic examinations or in the stool.


The authors thank Prof. Amazia Zimber (University of Jerusalem, Israel) for helpful discussions and assistance in editing this article for style and usage of English and Prof. Jean-Pierre Gendre (Service d'Hépato-Gastro-Entérologie et de Nutrition, Hôpital Saint-Antoine) for communicating clinical data of ulcerative colitis patients followed by his service. The technical work of Mrs Lydie Germain (laboratoire d'Histologie) was greatly appreciated.