Clinicopathologically, seven out of eight (7/8) GA-FGs in this study were located in the upper third and only one (1/8) tumor was observed in the middle third of the stomach. Macroscopic features indicated flatly elevated (0-IIa) (6/8) or depressed (0-IIc) lesions (2/8), and a central depression was observed in some cases with deep infiltration. This central depression was believed to be evidence of submucosal involvement11. Therefore, more samples should be collected to clarify whether the case with submucosal infiltration has special endoscopic features. The tumors were small with a maximum diameter ranging from 4–12 mm (average 6 mm). Previously published reviews showed that the average tumor diameter was 7.5 mm. Approximately 80% of all tumors were less than 10 mm in diameter at the time of diagnosis6 and the diameter of the largest reported tumor was 85 mm8. Our observations were consistent with those of previous reports (Table 1).
Careful pathological observation revealed that the superficial area of the lesions almost invariably tended to retain normal foveolar epithelium, whereas the lamina propria and submucosa tended to show irregular branching and dilatation of fundic glands. The nuclei were slightly larger than those of normal fundic glands and markedly hyperchromatic. A review showed that most GA-FGs were confined to the mucosa4. Six of the eight (75%) cases exhibited submucosal invasion despite the small size of the lesions; lymphatic or venous invasion was not observed. Singhi et al.10 suggested that “GA-FG” is an exaggeration and lesions should be considered benign owing to the lack of recurrence or progression. A review of 111 reported cases revealed that 57% GA-FG showed submucosal invasion, while 6% showed subserosal invasion due to lymphovascular spreading4. In the patients in our study, the mildly atypical glands were well-circumscribed with an abrupt transition from the normal mucosa, which is one of the signs of neoplasia. Ueyama et al.7 speculated that surface mucosal epithelial cells are maintained, as tumors barely destroy the surrounding tissue. GA-FGs may possibly grow vertically into the submucosa and develop laterally toward the surrounding tissue. The adjacent oxyntic mucosa is normal without any intestinal metaplasia or atrophy7. However, a case of GA-FG arising from gastric mucosa with atrophic changes and intestinal metaplasia was focally observed in the surrounding mucosa12. In our study, most of the adjacent tissues of the tumor did not show inflammation, intestinal metaplasia, and atrophy, except in the case of one patient, where obvious intestinal metaplasia and atrophy were observed. Thus, this lesion is uncommon and does not invariably lack atrophy and intestinal metaplasia. Most cases, including those we evaluated, were instances of solitary tumors. Cases of multiple GA-FG have been rare, and most of their clinicopathological characteristics were similar to those seen in single lesions5.
Immunohistochemistry showed that tumor cells diffusely expressed pepsinogen -I and MUC6, which suggested that GA-FG originated from the chief cell of the mucosal layer rather than the foveolar cells of the epithelium. All patients in our study were positive for CD56 and synaptophysin and most were negative for chromogranin A. These findings could have resulted in an incorrect diagnosis of neuroendocrine tumors. In previous studies, staining for synaptophysin and CD56 showed diffuse positivity in the glands, while chromogranin A staining revealed completely negative results9,11. Since foregut-derived endocrine cells are invariably positive for chromogranin A (CGA), they certainly undergo endocrine differentiation13-15. Therefore, immunohistochemical findings for pepsinogen-I and MUC6 are useful for the differential diagnosis of gastric adenocarcinoma of the fundic gland type (chief cell predominant type), especially when CGA is also positive.
The differential diagnosis of GA-FG also includes the presence of neuroendocrine tumors, pyloric gland adenoma, oxyntic gland adenoma, and well-differentiated GA. Endoscopic findings may resemble those of submucosal tumor (SMT)-like tumors, especially sporadic neuroendocrine tumors11. However, most neuroendocrine tumors are small, smooth, firm, and well-circumscribed polypoid elevations of the mucosa and submucosa. Furthermore, Fukatsu11 suggested that as neuroendocrine tumors grow, they may involve the entire thickness of the gastric wall, resulting in occasional central ulceration. The minute flat, elevated lesion did not reveal any polypoid appearance in GA-FG and can be useful in distinguishing GA-FG from gastric neuroendocrine tumors. A study showed similarities in the IHC profile and molecular phenotype, suggesting that GA-FG and pyloric gland adenoma may be closely related16 and hence, may be distinguished primarily based on morphological features. Pyloric gland adenomas (PGA) show polypoid proliferation of pyloric-type glands consisting of cuboidal/columnar cells with foamy ground-glass cytoplasm. The well-differentiated (“crawling”) gastric adenocarcinoma with foveolar and pyloric phenotypes showed mild cytology and were similar to GA-FG; however, cells lacked the admixture of chief or parietal cells and showed an invasive growth pattern17,18. Oxyntic gland adenoma is a benign epithelial neoplasm composed of columnar cells that can differentiate to chief cells, parietal cells, or both, and usually lacks the complex architecture of glands and submucosal invasion. Chan9 suggested that oxyntic gland polyp/adenoma and GA-FG showed a morphological continuum, and that the adenoma was a precursor to chief cell-predominant adenocarcinoma. Whether a subset that lacks the ability to invade or metastasize can be called an “oxyntic gland adenoma” requires further investigation.
In a previous study, nuclear β-catenin positivity (using IHC) was observed in 22 of 26 cases, and 13 cases (50%) harbored mutations in at least one of the following genes: CTNNB1, GNAS, AXIN1 or 2, and APC8. Nuclear β-catenin expression coincided with the presence of GNAS mutations in four of five cases, suggesting a role for GNAS activation in WNT signaling. Activation of the WNT-β-catenin signaling pathway is believed to be involved in tumorigenesis8. Interestingly, sporadic fundic gland polyps also show activating mutations in β-catenin19; however, GNAS mutations are either absent or infrequent in conventional gastric adenomas and adenocarcinomas20,21. Although only membrane staining for β-catenin without any nuclear staining was observed in all patients in our study, our results were consistent with those of Benedict et al.4, which may be attributed to the limited number of cases.
Among the cases with follow-up data, one patient died of carcinomatosis22 and three showed disease recurrence10,23, which may have been due to incomplete excision of the lesion. Complete surgical excision and fundectomy for some cases24,25, and ESD or EMR for most cases appear to be adequate and may lead to remission. In our study, all patients could be followed up after ESD for a period from 5–33 months. All patients have been free from recurrence or metastasis.
In conclusion, GA-FG is a well-differentiated adenocarcinoma with cytological mild atypia, which is often accompanied by submucosal infiltration, although lymphatic and vascular infiltration are uncommon. They are located mostly in the upper third of the stomach and composed predominantly of chief cells and known to characteristically change the complex structure of glands; however, cellular atypia is mild. The predominant immunohistochemical markers of GA-FG are pepsinogen-I and MUC6. Although not specific markers, tumor cells were invariably positive for SYN and CD56, while CGA expression was not common. The lesions were completely removed using ESD and there was no recurrence within this observation period. If GA-FG is suspected during endoscopy, a pathologist should perform immunohistochemical staining to confirm the diagnosis. More data should be collected to clarify whether acid inhibition is involved in disease development. Since GA-FG is different from conventional gastric adenocarcinoma, its etiology and pathogenesis deserve more attention.