Anatomic Pathology / COLLAGENOUS GASTRITIS Collagenous Gastritis A Long-Term Follow-up With the Development of Endocrine Cell Hyperplasia, Intestinal Metaplasia, and Epithelial Changes Indeterminate for Dysplasia Jeffrey L. Winslow, MD, 1 Thomas D. Trainer, MD, 1 and Richard B. Colletti, MD 2 Key Words: Gastritis; Collagen; Chronic; Neuroendocrine; Hyperplasia; Intestinal metaplasia; Carcinoid; Adenocarcinoma; Dysplasia Abstract This report reviews the literature pertaining to collagenous gastritis and describes the clinicopathologic evolution of this disease in a patient during a 12-year period. We examined 109 biopsy specimens of gastric mucosa from 19 different endoscopic procedures for the severity and distribution of collagenous gastritis in a single patient. Assessments were undertaken for the presence of endocrine and gastrin cell hyperplasias and dysplastic epithelial changes. Relative to biopsy specimens from age- and sexmatched control subjects, the patient s biopsy specimens showed a significantly lower number of antral gastrin cells, along with a significant corpus endocrine cell hyperplasia, suggesting an increased risk of endocrine neoplasia. Gastric corpus biopsy specimens revealed an active, chronic gastritis, subepithelial collagen deposition, smooth muscle hyperplasia, and mild to moderate glandular atrophy. Additional findings of intestinal metaplasia and reactive epithelial changes indeterminate for dysplasia raise concerns about the potential for adenocarcinoma. Collagenous gastritis is a rare form of chronic gastritis characterized by a prominent subepithelial deposition of collagen, smooth muscle hyperplasia of the lamina propria, a mixed mucosal inflammatory infiltrate, and focal glandular atrophy. Neuroendocrine cell changes in the antrum and body mucosa have not been documented previously, and the longterm outcome for patients with respect to the development of neuroendocrine tumors or adenocarcinoma is still unknown. 1,2 In 1989, Colletti and Trainer 3 described collagenous gastritis in a 15-year-old girl. This report is a 12-year clinicopathologic follow-up of that patient, including an assessment of the gastric neuroendocrine cell population and documentation of focal intestinal metaplasia and epithelial changes indefinite for dysplasia. Materials and Methods Specimens from the patient consisted of 109 gastric mucosal biopsy specimens, 83 from the corpus and 26 from the antrum, obtained during 19 separate endoscopic examinations during a 12-year period. Biopsy specimens from the esophagus, duodenum, and colon also were examined on multiple occasions during the study period. Serving as control subjects were 13 women, ages 19 to 29 years, who had histories of various gastrointestinal complaints and whose gastric mucosal biopsies (19 from the body and 14 from the antrum) were reported as normal on histologic examination. All biopsy specimens were fixed in Hollande solution, embedded in paraffin, cut into 5-µm sections, and stained with H&E for routine microscopic examination. American Society of Clinical Pathologists Am J Clin Pathol 2001;116:753-758 753
Winslow et al / COLLAGENOUS GASTRITIS Table 1 Number of Patient and Control Biopsy Specimens From Corpus and Antrum Stained With Immunohistochemical Primary Antibodies Corpus Antrum Primary Antibody Against Source Patient Control Patient Control Chromogranin A Hybritech (San Diego, CA) 77 17 0 8 MIB-1 (Ki-67) Biogenex (San Ramon, CA) 11 0 0 0 CD3 (T cell) DAKO (Carpinteria, CA) 5 0 0 0 p53 Biogenex 13 0 0 0 Gastrin DAKO 3 8 26 14 alpha-smooth muscle actin Sigma (St Louis, MO) 2 0 0 0 Immunostaining was performed by applying a 3-step immunoperoxidase technique using 3,3'-diaminobenzidinetetrahydrochloride dihydrate (ChemMate DAB, Ventana, Tucson, AZ) as the chromogen. The primary antibodies used on both patient and control specimens of corpus and antrum-type mucosa are listed in Table 1. Biotinylated link and avidin-biotin complex (SDK605, Ventana) served as secondary antibodies. Hematoxylin was used as the counterstain. Patient and control slides containing gastric corpus or antrum mucosa biopsy specimens were identified, and additional 5-µm sections were cut from the corresponding blocks for immunohistochemical staining. Primary antibodies directed against chromogranin were applied to gastric corpus biopsy specimens on 32 slides from the patient and to 11 slides from the control group. Primary antibodies directed against gastrin were applied to antrum mucosa biopsy specimens on 13 slides from the patient and to 8 slides from the control group. Both corpus chromogranin-positive cells and antral gastrin cells were counted in a similar fashion using a linear micrometer calibrated to a 22-mm 10 eyepiece and a 40 objective on an Olympus BMAX series BX40 microscope (Olympus America, Melville, NY). On a single level from each slide, 1 linear millimeter of mucosa was delineated in 4 randomly chosen, nonoverlapping 0.25-mm segments with the micrometer oriented parallel to the muscularis mucosae or the base of the biopsy specimen. Cells within the base of the antral pits or within the parietal and chief cell region of the gastric corpus were interpreted as exhibiting positive staining if they showed dense, discrete cytoplasmic staining within the plane of section over an area at least the size of an adjacent epithelial cell. The nonantral endocrine populations, in both the patient and the control specimens, also were graded according to the criteria of Solcia et al. 4 Five biopsy specimens of gastric corpus-type mucosa demonstrating the greatest numbers of intraepithelial lymphocytes as observed on H&E stains were selected for CD3 immunoperoxidase staining. The number of CD3+ cells within a series of 100 consecutive surface epithelial cells was counted in each of the biopsy specimens in the areas corresponding to the greatest apparent CD3+ cell density. Based on a previous study by Haot et al, 5 a minimum of 30 intraepithelial lymphocytes per 100 epithelial cells was used as the limit for a diagnosis of lymphocytic gastritis. The extent and severity of the patient s gastritis were graded according to the updated Sydney system. 6 Selected patient biopsy specimens also were subjected to Masson trichrome, Congo red, Fontana-Masson, and Steiner stains. Case Report Clinical History In this 27-year-old woman, recurrent epigastric abdominal pain first developed at age 14 years, and acute hematemesis, hematochezia, anemia, and hypotension developed during the following year. Endoscopic examination initially revealed multiple diffusely scattered discrete gastric mucosal hemorrhages; 1 week later, there was granularity of the gastric corpus with resolution of the hemorrhages. Upper gastrointestinal contrast radiographic studies demonstrated normal pliability and a mosaic-like surface pattern within the gastric corpus. Four months later and on all subsequent examinations, the gastric body mucosa had a nodular appearance, whereas the antral mucosa appeared normal. Results of other laboratory and radiographic studies were negative, including antiparietal cell antibodies, antiintrinsic factor antibodies, antinuclear antibodies, and antineutrophil cytoplasmic antigen antibodies; radionuclide gastric emptying; and ultrasound and computed tomography scan of the stomach. Stool samples for ova, parasites, and bacterial pathogens were negative. Gastric biopsy specimens examined for Helicobacter pylori by the urease test were negative. During the 12-year period, serum gastrin levels ranged from 25 to 339 pg/ml (11.9-161.7 pmol/l), with a median of 114 pg/ml (51.0 pmol/l; reference range, <100 754 Am J Clin Pathol 2001;116:753-758 American Society of Clinical Pathologists
Anatomic Pathology / CASE REPORT Image 1 Collagenous gastritis involving gastric corpus mucosa (H&E, 50). Image 2 Subepithelial collagen deposition in a biopsy specimen of gastric corpus mucosa (trichrome, 150). pg/ml [47.7 pmol/l]; borderline, 101-200 pg/ml [48.2 pmol/l]; elevated, >200 pg/ml [95.4 pmol/l]); only 1 value was clearly elevated. Continuous 24-hour gastric ph monitoring demonstrated a ph less than 2 and appropriate inhibition by ranitidine (ie, ph >4, 45% of the time). Results of radiographic and endoscopic studies of the esophagus, duodenum, and distal colon and radiographic examination of the small bowel were normal. During the 12-year study period, the patient was treated with sucralfate (Carafate), ranitidine (Zantac), misoprostol (Cytotec), furazolidone (Furoxone), clarithromycin (Biaxin), metronidazole (Flagyl), omeprazole (Prilosec), and prednisone, without lasting benefit. There were no recurrences of gastrointestinal bleeding. She continues to have intermittent upper abdominal pain, but currently is not being treated with prescription medications. Pathologic Findings The gastric corpus biopsy specimens demonstrated a patchy, chronic active gastritis with gradual progression in disease severity over the 12-year follow-up period. The lamina propria was remarkable for a mixed inflammatory infiltrate consisting of lymphocytes and plasma cells admixed with variable numbers of neutrophils, eosinophils, and mast cells Image 1. Subepithelial collagen bands, up to 75 µm in thickness, were present in most of the corpus biopsy specimens Image 2. The distribution and thickness of the subepithelial collagen bands was variable both within and between biopsy specimens; however, the bands did not appear to become more diffusely distributed or thicker over time. Smooth muscle hyperplasia, present in the deep lamina propria of the original biopsy specimens, became progressively more pronounced over time, with extension up to the surface epithelium in some biopsy specimens Image 3. According to the updated Sydney system criteria, 6 the biopsy specimens showed a progression from moderate to marked mononuclear cell population, from mild to marked neutrophil cell component, and from absence to a moderate degree of intestinal metaplasia Image 4. Glandular atrophy was patchy in distribution and varied from mild to moderate in severity, resulting in a shortening of Image 3 Smooth muscle hyperplasia in a biopsy of gastric corpus mucosa. (alpha-smooth muscle actin immunoperoxidase stain, 50). American Society of Clinical Pathologists Am J Clin Pathol 2001;116:753-758 755
Winslow et al / COLLAGENOUS GASTRITIS Image 4 Intestinal metaplasia in a biopsy specimen of gastric corpus mucosa (H&E, 50). Image 5 Biopsy specimen of gastric corpus mucosa demonstrating surface epithelial cells with reactive atypia interpreted as indeterminate for dysplasia ( H&E, 250). oxyntic glands; however, parietal and chief cells were present in every corpus biopsy specimen. Immunoperoxidase stains for CD3 revealed a mild increase in intraepithelial T lymphocytes, ranging from 14 to 26 intraepithelial lymphocytes per 100 surface epithelial cells, with a mean of 20 in the areas of greatest CD3+ cell density. The increased intraepithelial lymphocytes were associated with patchy epithelial cell damage and reactive epithelial cell changes, as well as focal sloughing over areas of thickened A Image 6 Biopsy specimens of gastric corpus mucosa. A, An increase in neuroendocrine cells in collagenous gastritis. B, A histologically normal control. (Chromogranin A immunoperoxidase stain, 50) B subepithelial collagen. Regenerative surface epithelial cell changes, often quite pronounced, were sufficiently abnormal in a few biopsy specimens to warrant an interpretation of indeterminate for dysplasia Image 5. The epithelial cells in these areas reacted positively to MIB-1 antibody in a focal and linear pattern, but none of these cells reacted positively to p53 antibody. Steiner stains, done on multiple occasions, and Congo red stains were negative for H pylori like organisms and amyloid, respectively. Six of the patient s 83 corpus biopsy specimens were unavailable for immunohistochemical study owing to previous sectioning into the paraffin block. In the remaining 77 corpus biopsy specimens, the average number of chromogranin-positive cells per linear millimeter of mucosa was 201 for our patient (median = 196; SD = 47). The average for the 17 control biopsy specimens was 132 (median = 116; SD = 57). A 2-tailed Student t test for independent samples revealed that the number of chromogranin-positive cells per linear millimeter was significantly greater in the patient than in the control group (P =.0003) Image 6. In addition, many of our patient s biopsy specimens met the criteria of Solcia et al 4 for linear endocrine cell hyperplasia, namely the presence in multiple specimens of a mean of at least 2 chains of 5 or more neuroendocrine cells per linear millimeter of mucosa. However, a combination of both chromogranin and MIB-1 stains failed to show proliferative activity in the chromogranin-positive cells. None of the patient s corpus cells marked positively with the Fontana-Masson stain or with antibodies to serotonin, gastrin, or somatostatin. Throughout the study period, biopsy specimens of the antrum were remarkable only for a mild to moderate 756 Am J Clin Pathol 2001;116:753-758 American Society of Clinical Pathologists
Anatomic Pathology / CASE REPORT mononuclear inflammatory infiltrate in the lamina propria and focal mild glandular atrophy. Thickened subepithelial collagen bands were not identified. Tissue from all 26 of the patient s antrum biopsies was available for immunohistochemical study. Immunoperoxidase stains for gastrin revealed an average of 106 gastrin cells per linear millimeter of antral mucosa in our patient (median = 102; SD = 48). The 14 antrum biopsy specimens from control subjects averaged 161 gastrin cells per linear millimeter of mucosa (median = 160; SD = 46). A 2-tailed Student t test for independent samples revealed this difference to be significant (P =.0167). Steiner stains were consistently negative for H pylori like organisms. Findings from biopsy specimens from the esophagus and colon remained normal during the 12-year study period. Duodenal biopsy specimens, which were normal early in the course, were remarkable for the recent development of a mild chronic duodenitis, consisting of a mild infiltration of lymphocytes and plasma cells in the basal lamina propria and focally denuded surface epithelium without villous atrophy, collagen deposition, or increased intraepithelial lymphocytes. Discussion Collagenous gastritis is a rare entity. Of the 6 cases documented in the literature to date, 4 were females and 2 were males, ranging in age from 9 to 54 years. 3,7-11 Three of the patients had upper gastrointestinal bleeding and/or vomiting with no comorbid gastrointestinal tract disease. 3,7,9 The remaining cases had either nonspecific symptoms or symptoms attributable to a coexisting but unrelated gastrointestinal tract lesion. All patients described thus far showed endoscopic abnormalities within the stomach, including erythema or mucosal hemorrhage in 4 cases 3,7,9,11 and mucosal nodularity in 3 cases. 3,8,10 In no instance in which follow-up was provided did the disease regress, as shown by endoscopic or histologic examination, despite numerous and varied pharmacologic treatments. Collagenous gastritis has been associated with Sjögren syndrome, 7 lymphocytic colitis, 8 collagenous colitis, 10 and celiac disease with ulcerative colitis. 11 Given the known or presumed immune cause of these associated diseases, it seems reasonable to suggest a similar cause for the changes seen in collagenous gastritis. Pulimood et al 10 found ultrastructural evidence of eosinophil and mast cell degranulation in gastric biopsy specimens from their patient. They suggested that such degranulation could cause both tissue damage and recruitment of fibroblasts, leading to the deposition of the subepithelial collagen seen in collagenous gastritis. 10 This degranulation could be caused by an unidentified intraluminal antigen, whose continued presence in the diet would explain why this disease is so refractory to pharmacologic treatment. Identification of such an antigen in future studies could lead to an effective dietary treatment, analogous to the gluten-free diet used to treat celiac disease. Histopathologically, all patients showed involvement of the body and/or antrum by a gastritis characterized by a thickened band of subepithelial collagen, epithelial damage, smooth muscle hyperplasia, and a mixed lamina propria inflammatory infiltrate consisting of lymphocytes, plasma cells, eosinophils, mast cells, and variable numbers of neutrophils. Following their review of the cases in the literature, Vesoulis et al 11 proposed histopathologic diagnostic criteria consisting of a subepithelial collagen layer greater than 10 µm, lamina propria lymphoplasmacytosis, intraepithelial lymphocytes, and epithelial damage. Although their patient met the criteria of Haot et al 5 for a diagnosis of lymphocytic gastritis (>30 intraepithelial lymphocytes per 100 surface epithelial cells), our patient did not. Some forms of chronic gastritis, particularly autoimmune atrophic gastritis, are associated with gastric endocrine cell hyperplasia, usually of the enterochromaffinlike (ECL) cells. 2,4,6 ECL cells produce histamine and can be identified immunohistochemically, using fresh-frozen or freeze-dried tissue, neither of which was available for the present study. Indirect evidence for the ECL origin of the hyperplastic endocrine cells in our patient is derived from their negative reactions to the Fontana-Masson argentaffin stain, as well as to antibodies directed toward serotonin, somatostatin, and gastrin. In atrophic gastritis, the ECL cell hyperplasia in the gastric corpus frequently is associated with hypergastrinemia. Presumably, the loss of parietal cells and subsequent achlorhydria led to gastrin cell hyperplasia and an increase in circulating gastrin. The latter stimulates the histamine-producing ECL cells in the corpus to undergo hyperplasia, which may be in a linear or micronodular pattern and, in a small proportion of patients, may progress to carcinoid tumor. 2,4,12 In our case, there is no evidence of gastrin cell hyperplasia; on the contrary, our patient has a significantly lower number of antral gastrin cells than do age- and sexmatched controls. Moreover, only borderline elevations in the serum gastrin level were seen in our patient, and they were seen only during therapy with either ranitidine or omeprazole. These findings suggest that our patient s endocrine hyperplasia may be the result of a gastrin-independent mechanism. The possibility exists that endocrine cell hyperplasia is part of or secondary to collagenous gastritis itself. Eissele et al 13 identified linear and/or micronodular endocrine cell hyperplasia in the corpus of normogastrinemic patients who American Society of Clinical Pathologists Am J Clin Pathol 2001;116:753-758 757
Winslow et al / COLLAGENOUS GASTRITIS had undergone antrectomy and were treated with the protonpump inhibitor lansoprazole. They proposed that trophic factors, rather than hypergastrinemia, were responsible for the endocrine cell hyperplasia. In our patient, a combination of trophic factors and therapy-related borderline elevations of the serum gastrin level may have contributed to the endocrine cell hyperplasia seen in the gastric corpus. Our patient s follow-up biopsy specimens, obtained during the past 12 years, showed a progression from mild to moderate glandular atrophy, which was accompanied by a significant linear neuroendocrine hyperplasia in the gastric corpus and a lower than normal number of antral gastrin cells. These neuroendocrine findings are unexpected and merit further study, but they suggest that collagenous gastritis may lead to an increased risk of gastric neuroendocrine cell tumors, possibly through a gastrin-independent mechanism. The increased risk of progression to gastric adenocarcinoma remains unknown. Although intestinal metaplasia has been considered one of the steps in the development of gastric carcinoma of the intestinal type, it also is found frequently in benign conditions such as peptic ulcers. 14 The epithelial changes interpreted as indefinite for dysplasia may represent an atypical reparative process or a true dysplasia. No typical high-grade dysplastic foci have been seen. Thus, until the magnitude of these possible risks is better defined, periodic endoscopic examinations may be appropriate for patients with collagenous gastritis. From the Departments of 1 Pathology and 2 Pediatrics, University of Vermont College of Medicine, Burlington. Supported by the Department of Pathology, University of Vermont College of Medicine. Address reprint requests to Dr Trainer: Dept of Pathology, University of Vermont College of Medicine, 111 Colchester Ave, Burlington, VT 05401. References 1. Solcia E, Fiocca R, Villani L, et al. Hyperplastic, dysplastic, and neoplastic enterochromaffin-like cell proliferations of the gastric mucosa: classification and histogenesis. Am J Surg Pathol. 1995;19(suppl):S1-S7. 2. Müller J, Kirchner T, Müller-Hermelink HK. Gastric endocrine cell hyperplasia and carcinoid tumors in atrophic gastritis type A. Am J Surg Pathol. 1987;11:909-917. 3. Colletti RB, Trainer TD. Collagenous gastritis. Gastroenterology. 1989;97:1552-1555. 4. Solcia E, Bordi C, Creutzfeldt W, et al. Histopathological classification of nonantral gastric endocrine growths in man. Digestion. 1988;41:185-200. 5. Haot J, Hamichi L, Wallez I, et al. Lymphocytic gastritis: a newly described entity: a retrospective endoscopic and histologic study. Gut. 1988;29:1258-1264. 6. Dixon MF, Genta RM, Yardley JH, et al. Classification and grading of gastritis: the updated Sydney system. Am J Surg Pathol. 1996;20:1161-1181. 7. Freeman HJ, Piercey JRA, Raine RJ. Collagenous gastritis. Can J Gastroenterol. 1989;3:171-174. 8. Groisman GM, Meyers S, Harpaz N. Collagenous gastritis associated with lymphocytic colitis. J Clin Gastroenterol. 1996;22:134-137. 9. Côté JF, Hankard GF, Faure C, et al. Collagenous gastritis revealed by severe anemia in a child. Hum Pathol. 1998;29:883-886. 10. Pulimood AB, Ramakrishna BS, Mathan MM. Collagenous gastritis and collagenous colitis: a report with sequential histological and ultrastructural findings. Gut. 1999;44:881-885. 11. Vesoulis Z, Lozanski G, Ravichandran P, et al. Collagenous gastritis: a case report, morphologic evaluation, and review. Mod Pathol. 2000;13:591-596. 12. Rindi G. Clinicopathologic aspects of gastric neuroendocrine tumors. Am J Surg Pathol. 1995;19(suppl):S20-S29. 13. Eissele R, Brunner G, Simon B, et al. Gastric mucosa during treatment with lansoprazole: Helicobacter pylori is a risk factor for argyrophil cell hyperplasia. Gastroenterology. 1997;112:707-717. 14. Antonioli DA. Precursors of gastric carcinoma: a critical review with a brief description of early (curable) gastric cancer. Hum Pathol. 1994;25:994-1005. 758 Am J Clin Pathol 2001;116:753-758 American Society of Clinical Pathologists