Patterns of gastric atrophy in intestinal type gastric carcinoma

Authors

  • Hala M. T. El-Zimaity M.D.,

    Corresponding author
    1. Department of Medicine, Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas
    2. Department of Pathology, Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas
    • Director, Gastrointestinal Mucosa Pathology Laboratory, V.A. Medical Center (111-D), Rm. 3A-320, 2002 Holcombe Blvd., Houston, TX 77030E mail:
    Search for more papers by this author
    • Fax: (713) 797-0969

  • Hiroyoshi Ota M.D., Ph.D.,

    1. Department of Pathology, Shinshu University Hospital, Matsumoto, Japan
    Search for more papers by this author
  • David Y. Graham M.D.,

    1. Department of Medicine, Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas
    Search for more papers by this author
  • Taiji Akamatsu M.D., Ph.D.,

    1. Department of Medicine, Shinshu University Hospital, Matsumoto, Japan
    Search for more papers by this author
  • Tsutomu Katsuyama M.D., Ph.D.

    1. Department of Pathology, Shinshu University Hospital, Matsumoto, Japan
    Search for more papers by this author

Abstract

BACKGROUND

Multifocal atrophic gastritis (MAG) is currently considered a precancerous lesion leading to intestinal type gastric carcinoma. The current study aimed to describe the topography of atrophy in stomachs with early gastric carcinoma.

METHODS

Resected stomachs from patients with intestinal type gastric carcinoma were routinely processed, sectioned (an average of 108 sections/stomach), and stained with a triple stain. Sections were scored on a visual analog scale for Heliobacter pylori and intestinal metaplasia. The type of epithelium (antral, oxyntic, transitional) was recorded. Atrophy was defined as the loss of normal glandular components and included intestinal metaplasia and/or pseudo-pyloric metaplasia of the corpus. Pseudo-pyloric metaplasia was identified by the presence of pepsinogen I in mucosa that was topographically corpus but phenotypically antrum.

RESULTS

Sixteen stomachs with intestinal type gastric carcinoma were examined. In none of the specimens examined was MAG (independent foci of atrophy) identified. In the majority (88%), atrophy was present as a continuous sheet. Islands of intestinal metaplasia (multifocal intestinal metaplasia) were present within a sheet of pseudo-pyloric metaplasia. A few specimens (12%) had a non-atrophic corpus with almost total replacement of antral epithelium with intestinal metaplasia. Multifocal dysplasia distant from the original tumor was found both in areas with and without intestinal metaplasia.

CONCLUSIONS

Contrary to popular belief, atrophy in intestinal type gastric carcinoma is not present as independent foci, but rather as a continuous sheet. Previous studies failed to identify pseudo-pylori metaplasia as a marker for atrophy. Cancer 2002;94:1428–36. © 2002 American Cancer Society.

DOI 10.1002/cncr.10375

Before the re-discovery of Heliobacter pylori, it was known that different gastroduodenal diseases were associated with different patterns of gastritis.1–3 For example, duodenal ulcers were associated with minimal corpus gastritis,3–6 and gastric ulcers and intestinal type gastric carcinoma were associated with extensive gastritis, widespread intestinal metaplasia, and hypo- or achlorhydria.1–3, 7 Correa proposed that the intestinal form of gastric adenocarcinoma developed through a cascade of events that included multifocal atrophic gastritis (MAG).8–10 Multifocal atrophic gastritis has been defined as the presence of independent foci of atrophy and is thought to begin typically at the incisura angularis with progression up and down the lesser curve and around the anterior and posterior walls of the stomach.10

Although there is a large body of literature that deals with the relationship of chronic gastritis, intestinal metaplasia, and gastric carcinoma, current concepts largely evolved from studies that examined selected sections in cases of advanced gastric carcinoma or primarily focused on the presence of intestinal metaplasia.9, 11–16 Specimens from patients with advanced cancer are not ideal for the study of pathogenesis as the tumor growth may obliterate areas that could have been used to predict increased risk (e.g., dysplasia, intestinal metaplasia, or other possible characteristic precursor features). To obtain information on the histogenesis of gastric carcinoma, we studied the topography and features of gastritis in early cases of intestinal type gastric adenocarcinoma.

MATERIALS AND METHODS

Shinshu University Hospital and Nagano Municipal Hospital in Nagano prefecture Japan were screened for complete stomachs removed for gastric carcinoma. Gross classification of tumors was done following the standard classification established in 1972 by the Japanese Research Society.17 Each stomach was processed immediately after resection. Freshly resected stomachs were opened along the greater curvature, pinned to a slab of paraffin, and placed surface down in a tray filled with 10% formalin for 24 to 48 hours at room temperature. Fixed stomachs were photographed, and the majority of specimens (88%) were sectioned in toto at multiple planes parallel to the lesser curvature at intervals of approximately 5 mm, then re-photographed (Fig. 1). The remaining stomachs had sections cut along the lesser and greater curvatures of the stomach in addition to the cancerous area. All sections were processed, embedded in paraffin wax, cut at 4 μm, and stained with a triple stain.18, 19 With this stain H. pylori are red, mucus is light blue, and the gastric tissue has the regular hematoxylin and eosin staining, with areas of intestinal metaplasia highlighted with Alcian blue.18, 19

Figure 1.

Resected stomachs were photographed, sectioned in toto at multiple planes parallel to the lesser curvature of the stomach at intervals of approximately 5 mm, and then re-photographed.

Tumors were classified in accordance with the system of Lauren.20 Intestinal type tumors were further subdivided, following the World Health Organization (WHO) classification, into three types (well-differentiated, moderately differentiated, and poorly differentiated adenocarcinoma) based on the degree of glandular formation.21 The extent of local growth and regional and distant spread was done following the TNM classification. 21

A visual analogue scale graded from 0 (absent/normal) to five (maximal intensity) was used to score the presence of H. pylori, active inflammation, chronic inflammation, and intestinal metaplasia.22 The type of epithelium (antral, oxyntic, transitional) was also recorded. All sections with intestinal metaplasia were stained with high iron diamine and sub-typed into Types I to III according to the system used by Rokkas et al. and Silva et al.23, 24 Atrophy was defined as the loss of normal glands and included intestinal metaplasia and/or pseudo-pyloric metaplasia. Pseudo-pyloric metaplasia was identified by the presence of mucosa that was phenotypically antrum, stained positive for pepsinogen I, and was anatomically in a region where corpus would be expected. The histologic findings were color-coded on the photocopy of the sectioned stomach (Fig. 2).

Figure 2.

Data were color-coded on the photocopy of sectioned stomach (white = oxyntic mucosa, yellow = antral type mucosa, orange = transitional mucosa, green = intestinal metaplasia Type I or II, brown = intestinal metaplasia Type iii, blue = dysplasia, and red = cancer). A) If a section had a score of 2 for Type II intestinal metaplasia and the section was phenotypically antrum, two fifths of the corresponding square on the photocopied image was colored green and the other half was colored yellow. B) Advancing atrophic front in intestinal type gastric carcinoma.

Separation of Antrum from Corpus Based on the Presence of Pepsinogen I

Pepsinogen I (PG I) is localized to the gastric corpus primarily within chief cells. We used PG I immunohistochemistry to identify pseudo-pyloric metaplasia, (i.e., to separate what was anatomically corpus but histologically antrum from true antrum).25 Control tissue for PG I immunostaining was chosen from sections with similar mucosa (phenotypic antrum/anatomic corpus) from duodenal ulcer patients. For immunophenotyping, 4 μm thick sections were stained using a modified streptavidin-biotin complex method. Slides were pretreated by steam for 20 minutes in a Black and Decker steamer (Black and Decker, Schaumburg, IL) in 10 mmol/L citrate buffer (pH 6.0), followed by cooling for 20 minutes. The following reagents were used in sequential steps at 36 °C: an inhibitor for endogenous peroxidase, a protein block, primary antibody (1:200) for one hour (anti-pepsinogen I from Biogenesis Kingston, NH), biotinylated secondary antibody, avidin-biotin complex with horseradish peroxidase, and 3,3'-diaminobenzidine tetrahydrochloride. Slides were counterstained with hematoxylin.

Analyses

The main objective of the current study was to determine the prevalence of specific gastritis patterns in early cases of intestinal type gastric carcinoma with particular attention to MAG.26 To answer questions concerning probability and statistical significance, we examined data using the binomial test designed for data that can be classified into two distinct categories. For example, cancer patients can be classified with regards to presence or absence of MAG. To determine sample size, we used a normal approximation to the binomial distribution in which sample size is a function of its probability.27, 28 In countries with a high-risk for gastric carcinoma, the overall prevalence of MAG in patients without gastric carcinoma is typically greater than 50%.29 We assumed a 55% prevalence rate of MAG in patients with intestinal type gastric carcinoma. A sample of sixteen is sufficient to evaluate the true prevalence of MAG with 96% confidence.

RESULTS

Gastric resections were obtained from sixteen patients with intestinal type gastric carcinoma. Patient clinical features, macroscopic findings, and results of WHO and TNM classification are shown in Table 1. Overall 1,730 sections were examined (mean/median, 108/113 section per patient; range, 32–183 sections per patient). High iron diamine staining was used to examine 1,302 sections with intestinal metaplasia (mean/median, 81/72 per patient; range, 25–152 sections per patient).

Table 1. Clinical Features and Tumor Classification:Macroscopic, WHO, and TNM Classification
PatientAge (Years)GenderJapanese grossTumor sizeWHOTNM
  1. WHO: World Health Organization; WD: well differentiated; MD: moderately differentiated; PD: poorly differentiated.

176FType 0 II c22 × 19 mmWDT1 NO MO
259MType 0 II c45 × 28 mmWDT1 NO MO
360FType 0 II a29 × 20 mmWDT1 NO MO
466MType 0 II c5 × 4 mmWDT2 N2 MO
567FType 0 II c10 × 11 mmWDT1 N2 MO
654MType 0 I42 × 25 mmWDT2 NO MO
788FType 0 II a, II c, polypoid20 × 12 mmWDT1 NO MO
868MType 0 I30 × 20 mmWDT2 N1 MO
961MType 0 II c17 × 3 mmMDT1 NO MO
1078MType 0 II b10 × 15 mmWDT1 or Tis NO MO
1172MType 0 II c22 × 20 mmPDT2 NO MO
1258MType 0 II c12 × 20 mmWDTis NO MO
1372MType 0 II c17 × 11 mmWDTis NO MO
1475FType 0 II a, pap21 × 20 mmWDT1 NO MO
1569FType 0 II c33 × 50 mmWDT1 NO MO
1677FType 0 II c40 × 30 mmWDT1 NO MO

Two atrophic patterns were recognized. The first pattern was present in 14 specimens (88%) and involved the corpus. It was designated the advancing atrophic front. The second pattern was present in two specimens and was designated the atrophic antrum.

The Advancing Atrophic Front

Atrophic changes were seen in the antrum and corpus. All cases had atrophic changes at the incisura angularis as well as at the adjacent distal half of the lesser curve and adjoining anterior and posterior walls of the antrum. In addition, atrophy was observed in variably sized adjoining regions along the lesser curve proximal to the incisura angularis as well as along the greater curve, and the adjoining anterior and posterior walls of the corpus (Figs. 2 and 3). None of the specimens showed the pattern of atrophy of the greater curve with sparing of the lesser curve.

Figure 3.

Atrophic patterns in intestinal type gastric carcinoma. A) in 88% of cases, atrophy was present as intestinal metaplasia (Type I, II, and/or III) islands amid atrophic mucosa (pseudo-pyloric metaplasia). two patients (12%) had a nonatrophic corpus with total replacement of the antrum with intestinal metaplasia (more complete than incomplete). B) The degree of replacement with intestinal metaplasia or pseudo-pyloric metaplasia varied among specimens.

Corpus

Atrophy was present as a continuous sheet of pseudo-pyloric metaplasia with or without islands of intestinal metaplasia (Figs. 2 and 3). The degree of replacement with intestinal metaplasia or pseudo-pyloric metaplasia varied among specimens. For example, five specimens had almost complete replacement with intermingling patches of Type II and Type III intestinal metaplasia. In contrast, three specimens had pseudo-pyloric metaplasia of the corpus with small, scattered foci of intestinal metaplasia (Type II being more frequent than Type III). Although a patchy distribution of intestinal metaplasia was occasionally present in normal oxyntic mucosa, it represented a small percent (1%) of the overall picture.

Antrum

Five cases (36%) had complete replacement of the antral mucosa with intermingling patches of Type II and Type III intestinal metaplasia. Nine had islands of Type II and/or Type III intestinal metaplasia surrounded with normal mucosa.

All cases were positive for H. pylori organisms and were associated with a pan-gastritis (inflammation in the antrum and corpus). On a visual analogue scale of 0-5, the median for polymorphonuclear cell infiltration was 4 in the antrum and 3 in the body, and the median for mononuclear cell infiltration was 4 in both the antrum and body.

The Atrophic Antrum pattern

The second pattern (12%, 2 cases) was a non-atrophic corpus and almost complete replacement of the antrum with intestinal metaplasia. The two stomachs with this pattern had more areas with Type II intestinal metaplasia than with Type III (Type III represented less than 10%) (Fig. 3). This pattern was associated with antral predominant gastritis. On a visual analogue scale of 0-5, the median for polymorphonuclear cell infiltration was 4 in the antrum and 2 in the body; the median for mononuclear cell infiltration was 5 in the antrum and 4 in the body.

Relation of Tumor Site and the Pattern of Atrophy

The tumor was located in the antrum in eight patients (50%), in the corpus in six (38%), and at the incisura in two (12%). Six tumors (38%) were completely surrounded by intestinal metaplasia; three were surrounded by Type II, one was surrounded by Type III, and two were surrounded by Types II and III. The most common pattern was for the tumor to be surrounded by combinations of intestinal metaplasia (both Type II and Type III), and phenotypic antral mucosa (56%, 9 cases). Mucous type glands surrounded one tumor; intestinal metaplasia (Types II and III) was present in other areas of the specimen (intestinal metaplasia represented 20% of atrophic mucosa).

Type III intestinal metaplasia was more prevalent in the antrum than in the corpus in 81% (13 cases). Three cases (19%) had an equal proportion of Type III intestinal metaplasia in the antrum and corpus. None had a higher proportion of Type III intestinal metaplasia in the corpus. Of the six cases with the tumor located in the corpus, two were surrounded with Type II intestinal metaplasia, one was surrounded with Type II intestinal metaplasia and mucous type glands, and three were surrounded with mucous glands and Types II and III intestinal metaplasia with a preponderance of Type II.

All carcinomas were found within the atrophic zone. For example, patients with a non-atrophic corpus and an atrophic antrum (the atrophic antrum) had the carcinoma and dysplasia in the antrum. One was near the border zone (antrum-corpus junction), and the other was more distal. More importantly, 13 patients with gastric carcinoma also had areas of dysplasia in other parts of the stomach, including the cardia. Multifocal dysplasia was present in areas without intestinal metaplasia in 40% of sections (41 of 102) with dysplasia. Of those, 12 (12%) from 6 patients were high-grade dysplasia.

Multifocal Atrophic Gastritis

Multifocal atrophic gastritis, or islands of intestinal metaplasis within non-atrophic mucosa, was not identified in any of the specimens examined (0 of 16: 95% C.I. 0% = 21%).

Pepsinogen I Localization

The pepsinogen I antibody was used to evaluate 259 slides from patients with gastric carcinoma, and was positive in 132 phenotypic antrum/anatomic corpus sections (51%). In 75 sections (29%), areas positive for PG I were present in more than 10% of the section (Fig. 4).

Figure 4.

Pepsinogen i in phenotypic antrum. A) Duodenal ulcer disease; original magnification, × 4. B) Gastric carcinoma; original magnification, × 4.

To control for PG I immunostaining, 63 slides from duodenal ulcer cases were evaluated. In duodenal ulcer cases, five (8%) phenotypic antrum/anatomic corpus sections were positive for PG I; all were obtained from the transition zone between the anatomic antrum and corpus (Fig. 4) (P < 0.001 compared to gastric carcinoma).

DISCUSSION

The term multifocal atrophic gastritis was coined by Lambert in 1972 to describe normal mucosa alternating with areas of heavy cellular infiltration in which the glands appeared atrophic.2 Correa10 later adopted the term to describe independent foci of atrophy thought to begin at the incisura angularis and the antrum-corpus junction with subsequent spread of foci up and down the lesser curvature and the anterior and posterior wall. In recent years, atrophy has often become synonymous with intestinal metaplasia.10

The current hypothesis is that most cases of the intestinal form of gastric adenocarcinoma develop through a cascade of events that involve multifocal atrophic gastritis, complete intestinal metaplasia, incomplete intestinal metaplasia, and finally dysplasia.8–10 The current study searched for markers in early cancer based on the idea that the neighboring mucosa would presumably retain the characteristics of the mucosa from which the cancer developed. All cancer patients in this study had H. pylori infection, gastritis, and intestinal type gastric carcinoma. Atrophy was present in every case. The pattern was not one of MAG, but rather of islands of intestinal metaplasia (multifocal intestinal metaplasia [MIM]) amid a sheet of pseudo-pyloric metaplasia (Figs. 2 and 3). In the majority of cases (88%), atrophy was present at the incisura angularis, the adjacent distal half of the lesser curve, and the adjoining anterior and posterior walls of the antrum. In addition, atrophy was present in variably sized regions of the adjacent greater curve, the proximal half of the lesser curve, and neighboring anterior and posterior walls of the corpus (Figs. 2 and 3).

It is tempting to speculate that the concept of MAG as a precursor lesion for gastric carcinoma arose9, 10 in part due to difficulty in identifying pseudo-pyloric metaplasia of the corpus. Thus, MIM in a sheet of atrophic mucosa would be diagnosed as MAG because the presence of atrophy was based on the presence of intestinal metaplasia. 9, 10 In addition, MIM is also present in duodenal ulcers,30 but the underlying mucosa differs, being non-atrophic in duodenal ulcers and atrophic in cancer. Three patients with gastric carcinoma in the current study had pseudo-pyloric metaplasia of the corpus with very little intestinal metaplasia, suggesting that identifying atrophy as either lesion (intestinal metaplasia and/or pseudo-pyloric metaplasia) would be more effective in screening programs.

Other investigators have reported similar findings.31–33 For example, Oi et al.31 reported that benign gastric ulcers occurred in mucosa with pyloric-type glands found proximal to the normal border zone (antrum-corpus junction). In 1969, Kimura and Takemoto described endoscopically32 the presence of an advancing atrophic border, but H. pylori was unknown then and the changes were attributed to aging. Independently, Hebbel noted that atrophic changes were invariably more pronounced on the lesser than on the greater curvature.33 The natural history of H. pylori gastritis is for the normal border zone (antrum-corpus junction)30, 31, 33 to shift proximally, such that the antrum appears to expand, replacing fundic gland mucosa with advancing atrophic gastritis.31–35

Atrophy in the form of pseudo-pyloric metaplasia has been previously observed in experimental models36 as well as in gastric remnants following distal gastrectomy with gastroenteric anastomosis.37 In fact, routine screening for gastric carcinoma in asymptomatic patients with gastric remnants often reveals pseudo-pyloric metaplasia in oxyntic type mucosa. We have previously shown a positive association between the presence of mucous glands in corpus biopsies (pseudo-pyloric or mucous metaplasia) and age. This association was more prevalent in Korea, where gastric carcinoma is common.30

While the origin of gastric carcinoma from intestinal metaplasia is widely accepted, its exclusive role is controversial38 and impossible to prove. To show this association convincingly, it will be necessary to show intestinal metaplasia with early gastric carcinoma arising exclusively from internal metaplasia. As has been noted by others,38–40 40% of dysplastic mucosae are not associated with intestinal metaplasia. It should be noted, however, that the absence of goblet cells from areas with dysplasia and/or carcinoma can represent mucin depletion, different degrees of differentiation, or different cells of origin, (e.g., foveolar cells38). The current study was entirely related to intestinal type gastric carcinoma. Whether similar patterns of histology are found in the diffuse type (e.g., signet cell type) remains to be determined.

Several studies have looked at the relationship between intestinal metaplasia subtypes and gastric carcinoma, with varying results.12–15, 41–42 Two points are worth considering. First, in the current study, as in that of Kato et al.,43 the type of intestinal metaplasia was site-dependent, with a higher prevalence of incomplete intestinal metaplasia in the antrum (Fig. 3). Second, while Type III intestinal metaplasia was present in all specimens with gastric carcinoma, it was present in very small areas in several cases. Data from the current study suggest that the extent of mucosal atrophy within a region of the stomach may have a more important relation with the intestinal type of gastric carcinoma than with the presence or type of intestinal metaplasia. While intestinal metaplasia is a form of atrophy that is easy for pathologists to recognize, our results suggest that it is also important to determine whether intestinal metaplasia is present as an isolated patch within non-atrophic mucosa or within an atrophic lawn.

Two carcinoma patients in our study showed complete replacement of the antrum mucosa with intestinal metaplasia and had oxyntic mucosa of normal appearance. It is unknown if these individuals had normal or reduced acid secretion. Continued inflammation with antral atrophy could possibly lead to enough destruction of gastrin producing cells44 to result in a fall in acid secretion.45, 46 Alternatively, contiguous sheets of intestinal metaplasia may be unstable epithelium, especially upon exposure to carcinogens.

In conclusion, extensive mapping of the stomach showed that, in intestinal type gastric carcinoma, atrophy in the corpus is present as a continuous sheet that appears to progress proximally and towards the greater curve with advancing disease. Future screening programs for gastric carcinoma should address whether knowledge of the location of the advancing atrophic border/front (Fig. 5) allows one to estimate the risk for gastric carcinoma.

Figure 5.

Endoscopy pictures. A) Normal control. B) In contrast to the normal pink color of non-atrophic mucosa (greater curvature), atrophic mucosa is white with visible vascular patterns.

Ancillary