Premalignant conditions of gastric cancer



Professor Kentaro Sugano, 3311-1 Yakushiji, Shimotsuke, Tochigi 329-0498, Japan. Email:


Premalignant lesions of gastric cancer encompass a variety of conditions such as chronic gastritis, intestinal metaplasia and dysplasia, in which elevated risk of developing gastric cancer have been documented. Among them, intestinal metaplasia is frequently encountered in our daily endoscopic examination, yet its clinical significance is often underestimated despite of a number of reports demonstrating genetic and epigenetic alterations in the intestinal metaplastic mucosa. In this review, I will describe the molecular mechanisms of phenotypic changes from gastric mucosa to intestinal metaplasia based on our analysis of mouse model of intestinal metaplasia generated by ectopic expression of CDX2 in conjunction with the studies with human intestinal metaplasia.


Premalignant condition of gastric cancer (GC) is an umbrella term for high-risk states developing gastric cancer such as atrophic corpus gastritis and intestinal metaplasia (IM). IM is not considered to be a neoplastic lesion, but there is ample evidence that intestinal metaplasia harbors a number of genetic and epigenetic changes leading to gastric cancer.[1] Moreover, most of the studies demonstrate that eradication therapy is generally ineffective in reversing the condition.[2] Therefore, IM may be considered to be a pre-neoplastic lesion that has crossed “a point of no-return.” In this review, I will focus on the molecular mechanisms that lead to the formation of IM; in particular, a critical role of caudal-related homeobox transcriptional factor, CDX2, will be discussed based on our experimental data obtained with transgenic CDX2 mice, a mouse model of IM.[3]

In the stomach with IM, hypochlorhydric gastric milieu allows overgrowth of non-Helicobacter bacteria which is responsible for high nitrite contents. Our recent study indicates that oral microbacteria colonized in the hypochlorhydric stomach may be responsible for production of nitrosamines from nitrites.

I also discuss about diagnosis and management of premalignant conditions including “dysplastic lesion” for which there is a considerable disagreement between the West and Japan.

Molecular mechanisms leading to the intestinal metaplasia

In a so-called Correa's cascade, chronic gastric inflammation predisposes atrophic gastritis and IM. Now it is established that the most important cause of chronic gastritis is Helicobacter pylori (H. pylori) infection. However, the precise molecular mechanisms leading to IM had been unknown. To elucidate the mechanisms, we focused our research on the homeobox transcriptional factors, CDX1 and CDX2, which were reported to be critical in conferring intestinal phenotype[4] (Fig. 1). In humans, two homologous CDX1 and CDX2 are known, and CDX1 had been reported to be ectopically expressed in the Barrett's esophagus and IM,[5] but the role of CDX2 was not known. Thus, we examined the temporal and topological expression pattern of both CDX1 and CDX2 in patients with chronic gastritis in conjunction of intestinal marker gene expressions and found that the expression of CDX2, but not CDX1, occurred early in the mucosa without intestinal gene expression. CDX1 expression was observed later in the mucosa expressing intestinal marker genes.[6] Therefore, we suspect that ectopic expression of CDX2 in the inflammatory gastric mucosa might trigger the molecular events leading to IM. To support this hypothesis, we generated CDX2-transgenic mice by expressing a transgene construct containing H+,K+-ATPase promoter attached to CDX2 gene to guide its expression in the parietal cells.[3] About one month after birth, foci of intestinal metaplastic glands emerged in the corpus mucosa which spread to the entire corpus mucosa by 6 months after birth. In another mice model of IM using Foxa3 promoter construct,[7] however, the IM observed was limited in the antrum and dissimilar to the human IM in terms of structural organization and was not extensive as shown in a patchy absorptive cells, indicating the importance of the selected promoter that could influence the ectopic expression of CDX2 in a particular cell type. A detailed analysis of the sequence of molecular signature expressions in our CDX2-transgenic mice verified that CDX2 expression emerged before apparent expression of intestinal marker genes that shape intestinal phenotype, whereas CDX1 expression was observed concurrently with intestinal gene expressions.[8] In this mouse model of IM, pseudopyloric metaplasia (spasmolytic polypeptide-expressing metaplasia: SPEM) remained in the bottom of the metaplastic glands, indicating a hybrid nature of the gland architecture. Although CDX1-transgenic mice also showed IM, but the IM were not widespread and no cancerous lesions were observed.[9] Therefore, CDX2 seemed to be more important in inducing IM. Indeed, CDX2 can activate endogenous CDX1 gene expression.[10] Importantly, cancerous lesions developed in the IM in almost all the mice when kept for 2 years (Fig. 2). This process was shortened when CDX2-transgenic mice were crossed with p53 deficient mice or APC (adenomatous polyposis coli) mutant Min mice.[11] These experimental data indicate that IM may be a direct precursor of gastric cancer, but there are controversies that the majority of intestinal type of human gastric cancers develops from so-called gastric-intestinal mixed glands.[12] However, the gastric-intestinal mixed glands have principally been defined by mucin histochemistry. As described above, CDX2 gene expression occurs before such phenotypic changes (including mucin expression),[6, 8] and therefore CDX2 was also expressed in the so-called gastric-intestinal mixed glands.[9] Conversely, SOX2 [(Sex determining region Y)-related high-mobility-group (HMG) box transcription factor 2 more simply known as SRY/HMG box 2 transcriptional factor] gene expression whose expression is limited in the normal stomach was simultaneously observed not only in our mice model,[13] but also in human IM (Fig. 3). Thus, apparent intestinal glands showed mixed expression in terms of transcriptional factors SOX2 and CDX2 whose expressions in the normal condition are limited to the stomach and intestine, respectively. As proposed by McDonald and colleagues, multiple stem cells may exist in a single gastric unit, and it may take a long time to have an entire gastric unit replaced by progenies from a single stem cell.[14] Therefore, classification of gastric-intestinal mixed gland and intestinal gland based on expression of gastric mucins (MAC5Ac, MAC6) and intestinal mucin (MAC2), respectively seems to be too simplistic. It would be more reasonable to assume that IM is a hybrid state where multiple progenitors are changing their cell fates in a differential manner. In support of this hypothesis, microRNA (miRNA) expression profile of the IM in the CDX2-transgenic mice was more closely related to the original gastric mucosa than intestinal mucosa (unpublished observation). Thus, IM may be better defined that it is not a true trans-differentiation but a “disguised state” of gastric cells, as miRNA expression profile can be more informative in elucidating the developmental lineage.[15]

Figure 1.

Caudal-related homeobox transcriptional factor (CDX2) regulates multiple intestinal gene expressions. A number of gene signatures for differentiated enterocytes, including enteroendocrine cells, goblet cells, and absorptive cells are regulated by CDX2.

Figure 2.

Gastric cancer developed in CDX2-transgenic mice. (a) In the stomach of CDX2-transgenic mice at 100 weeks after birth, differentiated type of gastric cancers developed.[11] No tumors were seen in the control mice at the same age. (b) Histological examination showed the tumor penetrated muscularis propria.

Figure 3.

Co-expression of SOX2 and CDX2 in the human intestinal metaplasia. In human intestinal metaplasia, both SOX2 and CDX2 were co-localized in the nuclei of the metaplastic glands. Antibodies used for SOX2 and CDX2 were obtained from Sigma and Biogenex, respectively. Second antibodies used for immunofluorescence were Alexa labeled antirabbit IgG from (Invitrogen) for SOX2 and Cy3-labeled antimouse IgG (immunoglobulin G) (Jackson) for CDX2, respectively.

Genetic and epigenetic changes in the IM and dysplasia

It has been well-documented that considerable genetic and epigenetic changes occurred in IM including p53 mutations, methylations of Runx3, CDH, and p16 that have been documented as important genetic/epigenetic alterations in gastric cancers.[1, 16] Among these changes, methylation of Runx3 seems important in inducing IM, because deficiency of Runx3 function was shown to induce CDX2 expression.[17] We have confirmed that micro-dissected human IM tissue contained a number of mutations in p53 genes (Table 1). Additional genetic/epigenetic changes in the IM would lead to neoplastic, dysplastic lesions as demonstrated by an elegant study by McDonald and colleagues, who showed that dysplastic lesions contained the same genetic alterations with the surrounding IM.[18] Further genetic alterations occurring in the dysplastic lesions give rise to cancerous foci within dysplastic area, so-called “cancer in adenoma” (Fig. 4a,b). As was shown in our mice model, these human studies would support that at least in some cases, IM is directly connected to adenoma-carcinoma sequence.

Figure 4.

Cancer in adenoma. (a) Cancer foci showing conspicuous cellular atypism and tissue disorganization compared with the surrounding adenoma. Arrows show the site of submucosal invasion in the cancerous foci. (b) P53 positive cells were confined to a cancerous part in the cancer in adenoma tissue. Left: HE staining, Right: P53 immunostaining.

Table 1. P53 gene mutations in human IM. Human intestinal metaplastic glands were microdissected, and p53 mutations were analyzed with the human p53 Genechip (Affymetrix) that enables screen for mutations of coding region from exon 2 to 11 including introns of human p53 gene
PatientAgeSexSite of mutationsType of mutationsAmino acid changes
 154MExon 5cgc →cacR→H
 248MIntron 10g→a
 569MIntron3, Exon 5gtc→ggcV→G
 648MIntron 10g→a
 768MExon 6g→-V→Stop

Exon3, Intron 5,

Exon 6

gtt→ttt, a→g, cga→cca

gag→ggg, ccg→cgg


V→F, R→P,

E→G, P→R,


 969FExon 8ggg→cagG→R
1049FExon 3, Intron 3

ctt→ttt, aaa→ccc,

ctg→cag, g→t, t→c

L→F, N→P,


Hypochlorhydria and bacterial overgrowth

How can we explain the mechanisms of these genetic and epigenetic changes seen in IM or dysplasia? Some of the epigenetic changes have been reported in chronic gastritis, and could be reversed by eradication. However, H. pylori may play only a limited role in the neoplastic progression from IM as it cannot colonize in IM. In support of the role of other factors in this process, continuous occurrence of gastric cancer long after successful eradication of H. pylori has been reported.[19, 20] What would be the feasible mechanisms and factors to explain this process? We propose that bacterial overgrowth and resultant increased nitrosamine production in the stomach which was once the leading theory for gastric carcinogenesis should be reevaluated to explain this process. In hypochlorhydric or achlorhydric stomach, abundant growth of bacteria, mainly from oral source, can be seen (Fig. 5). Since gastric juice is not acidic with low level of ascorbic acid in the stomach harboring the IM, nitrites from saliva can stay in the gastric juice and are converted to carcinogenic nitrosamines. It is also plausible that in other area of the stomach, H. pylori and other microbacteria may coexist, since IM distribution in the stomach is sporadic. Such coexistence of multiple organisms may be more dangerous by aggravating the inflammation and atrophy.[21] Moreover, other bacterial species can still reside in the atrophic stomach with low acid secretory state and exert harmful effects by producing potentially chemical carcinogenic substance even after eradication therapy. Thus, roles of other virulent bacterial species should be further explored to explain the reasons why more advanced stages of atrophy have a higher risk for gastric cancer and to solve the so-called Asian or African enigma.

Figure 5.

Bacterial overgrowth in the H. pylori infected stomach. In gastric juice obtained from patients with atrophic gastritis, abundant growth of non-H. pylori species was observed both in aerobic (left) and anaerobic (right) conditions.

Diagnosis and management of precancerous conditions

Valid diagnosis of IM can only be done by histological examination of biopsied mucosa. Indeed, poor agreement between endoscopic diagnosis and that of histology was documented. In a Japanese pilot study, endoscopic diagnosis of IM had a high specificity (99%) but the sensitivity was surprisingly poor; only 12% of histologically confirmed IM was diagnosed by endoscopy.[22] However, modern endoscopic imaging modalities, such as narrow band imaging (NBI), flexible intelligent color enhancement (FICE), and blue-laser imaging (BLI), can facilitate identification of IM and dysplastic or cancerous lesions[23-25] (Fig. 6a,b). Once IM is identified during endoscopic examinations, high levels of vigilance should be exerted to search for dysplastic or cancerous lesions, because patients with IM have a higher risk of harboring such neoplastic lesions (Fig. 6b). Also important is to eradicate H. pylori if positive. Eradication has dual benefits: reduction of inflammation by H. pylori, a major culprit of inflammation and restoration of acid secretion which can reduce bacterial overgrowth in the majority of patients.[26] Whether eradication of H. pylori can revert IM is controversial. Most of the studies, however, could not demonstrate significant improvement of IM,[2] which may be explained by the auto-regulatory mechanism of CDX2.[27] Thus, in the majority of cases, it is most likely that IM with genetic derangements remains after eradication therapy. As mentioned before, continuous development of gastric cancer long after successful eradication therapy also support that patients with atrophy and/or IM are recommended to receive follow-up endoscopic examinations.

Figure 6.

Endoscopic diagnosis of intestinal metaplasia. (a) Compared with conventional white light image (left), use of flexible-intelligent color enhancement (FICE) image facilitates detection of intestinal metaplasia due to enhanced gain in color contrast (right, arrowheads). (b) Blue laser imaging (BLI) mode can facilitate diagnosis of intestinal metaplasia and is useful in identifying the dysplastic area within the intestinal metaplasia. Left: Conventional white light image showing whitish slightly elevated mucosa. The BLI image of the rectangular area marked in the left photo was shown in the right. Intestinal metaplasia appeared greenish due to so-called “light-blue crest sign.”[23] In dysplastic mucosa, such coloration was lost, and it becomes easier to be noted (right).

If dysplastic lesions are detected during endoscopy, endoscopic mucosal resection should be considered if feasible. Even in Japan, where pathologists are well-experienced in the diagnosis of GC, biopsy-based diagnosis has to be corrected after entire lesions being checked (Table 2). This is not because some of the lesions satisfy the “invasion criteria,” but because distinct cellular and/or structural disorganization or identification of cancerous foci in adenomas (cancer in adenoma, Fig. 4a,b) can only become evident after examination of whole resected materials.

Table 2. Pathological diagnosis on biopsy samples and on endoscopically resected tissues Jichi Medical University (2001–2004). About 20% of biopsy diagnose of adenoma were changed to adenocarcinomas based on the entire tissue examinations obtained by endoscopic resections (ER). Conversely, about 6% of biopsy diagnose of well-differentiated adenocarcinomas were down-graded to adenomas after examination of endoscopically resected specimens.Thumbnail image of

Since in many Western countries, “invasive criterion” is necessary for the diagnosis of cancer; diagnosis based on biopsy alone tend to be insufficient, because it would be difficult to take biopsy materials targeted to a locally invaded area even with modern imaging modalities as the invasive front resides in the submucosa (Fig. 4a,b). Furthermore, follow-up studies on gastric dysplasia outcome revealed a high rate of progression to advanced cancers in a short period of time quite similar to the natural course of early gastric cancer in Japan,[28, 29] urging therapeutic intervention for patients with high-grade dysplasia diagnosed by Western pathologists.

For this reason, it should be recommended to perform endoscopic resection for high-grade dysplasia (early mucosal gastric cancer according to the Japanese criteria). To reconcile these discrepant diagnostic criteria between Japan and the West, the term “noninvasive high-grade neoplasia” was adopted in the Vienna classification. Unfortunately, however, this term has not been widely used in either side. Moreover, the term, “intraepithelial neoplasia” was introduced in the recent World Health Organization classification. In the future, we definitely need a global consensus how to deal with such “neoplastic” lesions, for which recent technological advancement would be instrumental in promoting mutual understanding.


This review article is the results of intensive clinical and research efforts of colleagues in Jichi Medical University. Special thanks to Dr Hiroyuki Mutoh who contributed to the molecular mechanisms of IM and to Dr Kiichi Satoh for the histological analysis. I also thank Dr Yoshikazu Hayashi in our department and Dr Shunji Hayashi in the department of microbiology, Jichi Medical University who contributed to gastric microbiology. Endoscopic images were kindly provided by Dr Hiroyuki Osawa in our department.