Gastric Adenocarcinoma With Enteroblastic Differentiation: Lessons From a Rare Case

Gastric adenocarcinoma with enteroblastic differentiation (GAED) was first proposed by Matsunou et al¹ as a rare variant of α-fetoprotein (AFP)–producing gastric cancer¹; however, the clinicopathologic characteristics of GAED have not been well established. This subtype of gastric adenocarcinoma has been histologically characterized as having a primitive intestine-like structure, composed of cuboidal or columnar cells with clear cytoplasm. It is generally recognized that GAED produces AFP in the serum or in the tumor cells. Yet, some cases of GAED are negative for AFP production, and the relationship between GAED and AFP production remains unclear.² Although GAED is known to be distinct from other conventional gastric carcinomas (GCs), clinical information on GAED has been quite limited. In this study, we report an exceedingly rare case of GAED with AFP production treated twice with surgical intervention and diagnosed by retrospective analysis using immunohistochemical staining for Glypican 3 and spalt-like transcription factor 4 (SALL-4).

Case Presentation

We present the case of an 82-year-old man who presented 4 years ago with epigastralgia and was admitted to our hospital for suspected acute cholecystitis. He had no family history of malignancy. Laboratory data showed elevated hepatobiliary enzymes: aspartate aminotransferase level of 360 IU/mL (normal range [NR], 8–38 IU/L), alanine transaminase of 145 IU/mL (NR, 4–44 IU/L), γ-glutamyl transpeptidase of 731 IU/mL (NR, 18–66 IU/L), total bilirubin of 2.1 mg/mL (NR, 0.2–1.2 mg/dL), direct bilirubin of 1.5 mg/mL (NR, 0.0–0.3 mg/dL), and amylase of 35 IU/L (NR, 40–129 IU/L). The hepatobiliary enzymes were likely elevated because of a gallstone in the common bile duct and a history of nonalcoholic fatty liver disease. Although stones were not detected in the common bile duct on endoscopic retrograde cholangiopancreatography, we judged that the stones might have passed through the common bile duct. The hepatobiliary enzymes returned to nearly normal levels afterward. The patient had a mildly elevated fasting plasma glucose of 125 mg/dL (NR, 70–109 mg/dL), and an HbA1c of 5.9% (NR, 4.7%–6.2%). Carcinoembryonic antigen was 3.7 ng/mL (NR, 0–5.0 ng/mL) and serum carbohydrate antigen was 12.1 U/mL (NR, 0–37 U/mL), both within normal range. AFP was mildly elevated at 13.6 ng/mL (NR, 0–9.9 ng/mL). Computed tomography revealed remarkable thickening in the walls of the gallbladder (Fig. 1A), gastric antrum, and corpus (Fig. 1B), suggesting acute cholecystitis and GC. Upper endoscopy revealed a large elevated mass measuring 90 mm at the antrum of the stomach (Fig. 2). The lesion was biopsied and pathology revealed a moderately differentiated adenocarcinoma. We gave the patient a preoperative diagnosis of acute cholecystitis and conventional-type GC with a TNM classification of T2N0M0 stage IB. The patient underwent cholecystectomy and distal gastrectomy with lymph node dissection. Postoperative pathology revealed well-to-moderately differentiated tubular adenocarcinoma measuring 90 mm in diameter with invasion into the muscularis propria and the lymphatic system. There was no metastasis to lymph nodes. The final staging was T1N0M0 stage IB. The patient received no adjuvant chemotherapy on the grounds of the early stage of GC as per the Gastric Cancer Treatment Guidelines 2018.⁴ Two years after the initial operation, no recurrence or metastasis was detected. Abdominal ultrasound revealed a 40-mm–sized mass at the lateral segment of the liver 31 months postoperatively. The mass was negative for fluorine-18-fluorodeoxyglucose by fluorodeoxyglucose-positron emission tomography (Fig. 3). Contrast-enhanced magnetic resonance imaging done 33 months after the gastrectomy revealed a highly enhancing mosaic mass that grew to 60 mm in size on T1-weighted imaging (Fig. 4). Serum carcinoembryonic antigen levels were mildly elevated at 9.6 ng/mL (NR, 0–5.0 ng/mL), serum carbohydrate antigen levels were within NR (2.8 U/mL [NR, 0–37 U/mL]), and serum AFP levels were highly elevated at 376 ng/mL (NR, 0–9.9 ng/mL). Therefore, a hepatectomy instead of a liver biopsy was indicated, because the mass was well vascularized; a biopsy poses a higher risk of hemorrhage. Moreover, the mass had the possibility of being a malignancy, such as hepatocellular carcinoma, metastatic carcinoma, or liver angiosarcoma; or of being a benign mass, including hepatic adenoma or hematoma. Performing a biopsy in these cases increases the risk of tumor cell dissemination. We performed an extended left lobectomy of the liver for therapeutic diagnosis at about 34 months after the first operation. The hepatic specimens and the prior resected gastric specimens were retrospectively analyzed. Pathology revealed neoplastic cells with eosinophilic cytoplasm with a focal glandular component on hematoxylin-eosin staining. Periodic acid–Schiff staining revealed acidophilic cytoplasm, suggesting glycogen-rich clear neoplastic cells (Fig. 5). These characteristic findings drove us to analyze the specimens with immunohistochemical staining for markers of enteroblastic differentiation (ED). To confirm GAED, we examined immunohistochemical positivity for 3 markers: AFP (rabbit polyclonal, 1:600; Dako, Glostrup, Denmark), Glypican 3 (rabbit, polyclonal, 1G12, 1:200; BioMosaics, Burlington, Vermont), and SALL4 (mouse, clone 6E3, 1:1000, Abnova, Taipei, Taiwan). The gastric and hepatic specimens stained strongly positive for Glypican 3 and sporadically positive for SALL-4 but negative for AFP (Fig. 6). Thus, the primary gastric tumor was retrospectively rediagnosed as GAED. The resected hepatic lesion was also diagnosed as metastatic GAED in the same manner (Fig. 7). For genetic analysis of the tumor, we performed next-generation sequencing (NGS) using the Next Seq 2000 (Illumina Inc, San Diego, California). Genomic DNA was extracted from the tumor specimens using the QIAamp formalin-fixed, paraffin-embedded gastric and hepatic specimens. Next-generation sequencing identified mutations in the TP53 gene and loss of the APC gene. The ERBB2/HER2 amplification status was not identified (HER2/CEP ratio 1.49 <1.7) by fluorescence in situ hybridization.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Thereafter, the patient remained well, and radiologic examination revealed no recurrence occurred for 6 months after the hepatectomy. The serum level of AFP remained normal (7.3 ng/mL) for 10 months after hepatectomy. However, the patient experienced an acute cerebral infarction, the cause of which is unrelated to metastatic gastric adenocarcinoma, and died 44 months after the initial surgery.

Discussion

Reports on individual clinical courses of GAED have been scarce. To understand GAED, we should shed light on the terminology used historically for APGC, because the terms “AFP-producing gastric carcinoma,” “hepatoid adenocarcinoma (HAC),” and “GAED” have been used interchangeably. The term “HAC” was first used for an APGC with hepatic differentiation features in 1985 by Ishikura et al.⁵ Afterward, in 1990, Ooi et al⁶ pointed out a variant type with a clear cell component that resembled fetal gut in an AFP-producing HAC. Matsunou et al¹ coined the term “enteroblastic differentiation” for an APGC with clear cytoplasm and reported its blastomatous characteristics for the first time. However, these terms have not been systematically classified, because the terms were not defined by considering the morphologically and functionally overlapping features. Subsequently, Glypican 3 and SALL4 were recognized as significant markers for HAC with ED by Ushiku et al^7,8. SALL-4 is a sensitive marker for APGCs and is especially useful to diagnose hepatoid GC or GC with retrodifferentiation to fetal gut⁸. Glypican 3 is also a sensitive marker for the hepatoid components of an APGC⁷. Therefore, GAED has been most recently defined as immunohistochemical positivity for at least 3 of these markers for enteroblastic linage: AFP, Glypican3, and SALL-4. Though we did not perform immunohistochemistry for claudin-6, it is also known as an enteroblastic lineage marker useful in distinguishing hepatic metastases from GAED from other hepatic tumors⁹. Our specimen was positive for Glypican3, and SALL-4, suggesting GAED as a diagnosis. Akazawa et al¹⁰ documented that the immunohistochemical positivity rates for AFP, Glypican 3, and SALL-4 were 31.4%, 82.4%, and 80.4% respectively, and Murakami et al² reported that Glypican 3 is the most sensitive marker for GAED (83%), followed by SALL4 (72%), and AFP (45%); conventional gastric cancer was not positive for these markers. The pathologic features of GAED include the proliferation of tubular, papillary, or solid cells with a clear cytoplasm rich in glycogen. GC with ED was coexistent with conventional adenocarcinoma.¹⁰ Our resected gastric specimens showed a well-to-moderately differentiated adenocarcinoma coexisting with a clear cell carcinoma as a major component. The pathologic characteristics were consistent with the diagnostic features for GAED.

Strictly speaking, hepatoid adenocarcinoma (HAC) differs from adenocarcinoma with ED, but they both share the ability to produce AFP. Previously, Akazawa et al¹⁰ retrospectively demonstrated that all gastric cases diagnosed as HAC were classified as solid-type adenocarcinoma with ED by histologic reevaluation with immunohistochemical staining using enteroblastic lineage marker of AFP, Glypican 3 and SALL-4. Furthermore, genetic analysis revealed a high TP53 mutation frequency. Therefore, they stated that HAC can be subcategorized as solid-type adenocarcinoma with ED.¹⁰ As for the tumorigenesis of GAED, histologic analysis supports the speculation that GAED develops from conventional GC, acquiring hepatoid or primitive gut features during its course.² Adenocarcinoma with ED is often encountered in the stomach, sporadically in the colorectum,^11,12 and in the cervical esophagus, the ampulla of Vater, and the gallbladder.^13–15 We believe that this is because immunohistochemical staining using Glypican 3 and SALL-4 is being applied commonly, and adenocarcinoma with ED may be recognized if reevaluation using immunohistochemical analysis is performed on previously published data that had diagnosed conventional adenocarcinoma, which could suggest that adenocarcinoma with ED can possibly occur in a multitude of organs. From a different perspective, Lu et al¹⁶ described recent data on somatically derived yolk sac tumors (YSTs) that significantly overlapped with APGCs. The YST cells showed clear cytoplasm and subnuclear vacuolization, resembling endometrial and fetal-gut epithelium. In addition, the somatically derived YSTs showed immunophenotypes of AFP, Glypican 3, and SALL-4 similar to those of HAC and GAED. Thus, they hypothesized that GAED and HAC are subtypes of APGC, and they may arise from somatically derived YSTs.^16–18

Moreover, we studied the genetic characteristics of GAED using next-generation sequencing and fluorescence in situ hybridization. Our results identified a TP53 gene mutation and a loss of APC. Akazawa et al¹⁰ reported that the most frequently mutated gene was TP53, and almost all cases with missense mutations showed p53 overexpression.

In general, HAC exhibits elevated serum AFP levels and usually presents with extensive vascular invasion, frequent liver metastasis, and advanced pTNM staging, all of which contribute to an extremely poor prognosis.¹⁹ Similarly, GAED is rare and highly malignant. GAED has higher vascular invasion and metastasis rates compared with conventional GC.² Even if early diagnosis as GAED could be made at the initial pathologic examination, treatment would be the same, because a specific treatment for GAED has never been established, and we have no choice but to indicate the current treatment guidelines in clinical practice. Surely, there is a report of GAED cases treated by endoscopic submucosal dissection²⁰; however, treatment results by long-term follow-up have not been verified, considering the highly malignant behaviors of GAED. Yet, we should make the observation interval for GAED shorter than for conventional GCs in consideration of aggressive behaviors. However, prognosis will not improve unless gene-targeted therapy is developed. Targeted HER2 therapy with trastuzumab may play an important therapeutic role in HER2⁺ GC. There is growing evidence that HER2 is an important biomarker and key driver of tumorigenesis in GC, with studies demonstrating amplification or overexpression in 7% to 34% of all GCs; the ToGA trial revealed 22.1% in GC without geographic difference.²¹ Although our case was negative for HER2, Fujimoto et al²² reported that HER2 is frequently overexpressed in HAC (31.1%) and GAED (42.1%), compared with other GCs (13.8%).²² We think that these data imply that HER2 expression may allow the proliferation of primitive neoplastic cells in a manner that would predispose the neoplastic growth to eventually growing into HAC or GAED.

Concerning the prognosis of stage I gastric cancers, overall survival rates between the laparoscopic group (94.2% [95% CI, 92.4%–96.0%]) and the open surgery group (93.3% [95% CI, 91.4%–95.2%]) were similar and satisfactory (log-rank P = 0.64; difference, 0.9 percentage points; 1-sided 97.5% CI, −1.6 to infinity).²³ Our case was stage IB according to TNM classification. Unexpectedly, metastasis to the liver occurred in the early postoperative period. The solitary metastatic hepatic lesion was resected; however, the patient died 44 months postoperatively, not due to GAED but due to a cerebrovascular event. Compared with conventional GCs, GAED has a poor prognosis of 2 to 39 months (median, 18 months) in any of its stages.² Therefore, tumor-specific treatment, including HER2 targeted therapy when indicated, should be considered.

Finally, we want to emphasize the most important lessons from our case. (1) If the serum levels of AFP are elevated or the clinical course is evidently different from that of conventional GC, pathology should be reconsidered to include GAED in the differential diagnosis. (2) Immunohistochemistry for Glypican 3 and SALL-4 should be performed to confirm GAED. (3) If the serum level of AFP is increasing, tumors with ED should be a candidate wherever tumors are located. (4) Because adenocarcinoma with ED has been sporadically reported in organs other than the stomach, we speculate that the neoplastic transformation acquiring ED can occur across organs.

Conclusions

Tumors with ED may produce AFP, and thus have immunopositivity for Glypican 3 or SALL-4 across organs. Our results emphasize that clinicians should keep this highly malignant tumor variant in mind when considering differential diagnoses for GC histologic subtypes. Research exploring modalities for reliable GAED diagnosis and effects of GAED is needed for earlier detection and better management.