Gastric Composite Tumor of Alpha Fetoprotein-Producing Carcinoma/Hepatoid Adenocarcinoma and Endocrine Carcinoma with Reference to Cellular Phenotypes
Alpha-fetoprotein-producing carcinoma (AFPC)/hepatoid adenocarcinoma (HAC) and neuroendocrine carcinoma (NEC) are uncommon in the stomach. Composite tumors consisting of these carcinomas and their histologic phenotypes are not well known. Between 2002 and 2007, to estimate the prevalence of composite tumors consisting of tubular adenocarcinoma, AFPC/HAC and NEC, we reviewed specimens obtained from 294 consecutive patients treated surgically for gastric cancer. We examined histological phenotype of tumors of AFPC or NEC containing the composite tumor by evaluating immunohistochemical expressions of MUC2, MUC5AC, MUC6, CDX2, and SOX2. Immunohistochemically, AFPC/HAC dominantly showed the intestinal or mixed phenotype, and NEC frequently showed the gastric phenotype. In the composite tumor, the tubular and hepatoid components showed the gastric phenotype, and the neuroendocrine component showed the mixed type. The unique composite tumor predominantly showed the gastric phenotype, and the hepatoid and neuroendocrine components were considered to be differentiated from the tubular component.
Alpha-fetoprotein (AFP) is a fetal protein produced by yolk sac cells, fetal hepatic cells, and some fetal gastrointestinal cells . AFP-producing carcinoma (AFPC) is rare in the stomach , although gastric adenocarcinoma is one of the most common manifestations of AFPC . Ishikura et al.  reported that hepatoid adenocarcinoma (HAC) is characterized by both hepatoid differentiation and AFP production, while the histological features of hepatoid differentiation in gastric AFPC have been determined. Gastric AFPC or HAC frequently displays aggressive behavior [2, 5]; however, these tumors exhibit many unresolved clinical and histopathological features. On the other hand, neuroendocrine carcinoma (NEC), including small cell carcinoma, is rare in the stomach , and its clinicopathological features and clinical outcome have been characterized recently [7, 8]. Both AFPC and NEC were classified as special types of gastric carcinomas by the Japanese Classification of Gastric Carcinoma . Composite gastric tumors, consisting of special types of carcinoma, have been insufficiently investigated. A composite gastric tumor usually consists of both a common and a special type of carcinoma, that is, well- or moderately differentiated adenocarcinoma and HAC .
Gastric adenocarcinomas have been histopathologically classified into four categories based on their cellular phenotype . Several cellular markers, including MUC2, MUC5AC, and MUC6, have been used to histochemically investigate these phenotypes. It has been identified that intestinal goblet cells, gastric foveolar epithelial cells, and pyloric gland cells cause the expression of MUC2 , MU5AC , and MUC6 , respectively. Measurement of CDX2 and SOX2 expression to investigate cellular phenotype has been adopted, recently. CDX2, homeobox gene has been known to play a role in the development of small and large intestines , while transcription factor SOX2 is expressed in normal stomach but not in the colon .
We examined the histological phenotypes of a composite gastric tumor consisting of AFPC/HAC, NEC, and tubular adenocarcinoma and investigated the cellular phenotypes of AFPC and NEC by evaluating the immunohistochemical expressions of MUC2, MUC5AC, MUC6, CDX2, and SOX2.
2. Material and Methods
2.1. Patients and Materials
To estimate the prevalence of composite gastric tumors consisting of tubular adenocarcinoma, AFPC/HAC, and NEC, we reviewed gastrectomy specimens obtained from 294 consecutive patients treated surgically for gastric cancer during a 5-year period (2002–2007) at Shinshu University Hospital. In addition to these consecutive case series, one case of composite gastric tumor of tubular adenocarcinoma, AFPC/HAC, and NEC was included. These specimens were examined after obtaining informed consent.
Three-micrometer-thick serial paraffin sections were obtained from representative formalin-fixed, paraffin-embedded blocks and stained with hematoxylin and eosin (HE) for histological examination or subsequent immunohistochemical staining to investigate their cellular phenotype.
2.2. Histopathological Diagnosis
Gastric AFPC was histopathologically diagnosed by a positive reaction of carcinoma cells in AFP immunostaining with or without hepatoid differentiation. NEC was suggested by the detection of undifferentiated carcinoma, including small cell carcinoma, by HE-staining, and diagnosed by a positive reaction of more than one-third of carcinoma cells for immunostaining with several neuroendocrine markers, including neural cell adhesion molecule (NCAM, CD56), chromogranin A (CGA), synaptophysin (SP), and neuron-specific enolase (NSE). Neuroendocrine markers were applied to tumors suspected of neuroendocrine phenotype based on examination of HE sections. According to the WHO 2010 classification of neuroendocrine neoplasms , NEC is diagnosed by immunostaining with Ki-67. Small cell carcinoma is defined as a tumor whose cells have scant cytoplasm, finely granular nuclear chromatin, measure less than the diameter of three small resting lymphocytes, and have faint nucleoli. Large cell carcinoma is defined as a tumor whose cells are large, with moderate to abundant cytoplasm and round to oval nuclei with frequent prominent nucleoli. All AFP and NEC cases were examined in three to six blocks. The clinicopathological features of gastric carcinomas are described in accordance with the WHO 2010 Classification of Tumours of the Stomach .
2.3. Immunohistochemistry and Classification of Cellular Phenotypes
The expression of MUC2, MUC5AC, MUC6, CDX2, and SOX2 was immunohistochemically investigated to confirm the presence cellular phenotypes of gastric AFPC or NEC. Microwave treatment was performed in 1 mM EDTA/10 mM Tris buffer (pH 8.0) for 25 min. The presence of more than one third of carcinoma cells in immunostained samples was defined as a positive reaction. The commercially available antibodies for this immunostaining are shown in Table 2. Carcinomas were classified into four categories based on their cellular phenotype in accordance with immunohistochemical findings; intestinal, mixed, gastric, or null (unclassified) phenotype . This phenotypic classification, a modified version of the definition described by Kumashiro et al. , is summarized in Table 1.
A composite gastric tumor consisting of AFPC/HAC, NEC, and tubular adenocarcinoma was observed in only one case (Case 1). In the resected specimen from this case, we observed an ulcerated tumor (75 × 110 mm) with no well-defined borders infiltrating into the surrounding tumor wall (Figure 1(a)). The NEC and AFPC components were located on the blue and red line, respectively, in cut sections of the resected specimen (Figure 1(b)). Histopathologically, this gastric carcinoma comprised of three types of components (Figure 2(a)). Moderately differentiated tubular adenocarcinoma was identified (tubular component; Figure 2(b)), while other type of carcinoma cells, with large eosinophilic and clear cytoplasm, had infiltrated with a trabecular or sheet-like pattern that displayed the morphological features of HAC (hepatoid component; Figure 2(c)). Furthermore, undifferentiated carcinoma cells had infiltrated the proximal part of the tumor. This component had polygonal carcinoma cells with prominent nucleoli in the nuclei, suggestive of large cell NEC (neuroendocrine component; Figure 2(d)). In this case, the NEC-cell component comprised of about 30% of the carcinoma, so this tumor was diagnosed not as neuroendocrine differentiation but as a gastric-mixed adenoneuroendocrine carcinoma (MANEC).
Clinicopathological features of patients with gastric AFPC/HAC or NEC, including the composite gastric tumor in Case 1, are shown in Table 3. Excluding Case 1, AFPC/HAC was observed in four cases (1.4%), while NEC was observed in six cases (2.0%). Among the six NEC cases, the large and small cell types of EC were observed in four and two cases, respectively. Among the four AFPC cases, HAC was observed in one case. There was no case of a composite gastric tumor consisting of AFPC/HAC and NEC, excluding Case 1. Synchronous expression of both AFP and neuroendocrine markers was not observed in any tumors; moreover, the cooccurrence of gastric AFPC and NEC was not observed in any tumors.
Immunohistochemical findings for diagnosis of composite gastric tumor, AFPC/HAC and NEC, are shown in Table 4. All six NEC cases manifested a positive reaction for one or more neuroendocrine markers, while no reaction was evidenced for AFP. SP was expressed in all NEC cases, while NCAM, CGA, and NSE were expressed in three NEC cases. These neuroendocrine markers were completely expressed in only one NEC case. None of the four AFPC/HAC cases evidenced a reaction with neuroendocrine markers. In Case 1, the tubular and hepatoid components evidenced a positive reaction for AFP (Figure 3(a)), while no reaction was observed for neuroendocrine markers. The neuroendocrine component of composite gastric tumor demonstrated positive reactions for CGA (Figure 3(b)) and SP, while no reaction was observed for AFP.
Immunohistochemical findings for the cellular phenotypes of composite gastric tumor, AFPC/HAC, and NEC are shown in Table 4. MUC2 was expressed in one AFPC case, while MUC5AC and MUC6 were expressed in the other three cases. CDX2 was expressed in all AFPC/HAC cases (Figure 3(c)), while SOX2 was expressed in one AFPC case. MUC2 was expressed in one NEC case, MUC5AC was expressed in the other NEC case, and MUC6 in four NEC cases. CDX2 was not expressed in any of the NEC cases, while SOX2 was expressed in five NEC cases (Figure 3(d)). In NEC cases, MUC6 was expressed in almost the same site when compared to neuroendocrine markers. MUC2 and MUC5AC were expressed in the part of the positive site associated with the neuroendocrine markers, but in different areas to each other. In Case 1, the tubular component evidenced no reaction for MUC2, MUC5AC, or MUC6, but expressed SOX2 (Figure 3(e)), and the hepatoid component expressed MUC6 and SOX2. The neuroendocrine component did not express MUC2, but expressed MUC5AC, MUC6, SOX2, and CDX2 (Figure 3(f)). In all NEC cases, more than 20% carcinoma cells are positive for Ki-67 (Figure 3(g)).
Consequently, with reference to the cellular phenotypes in EC, gastric, mixed, and null phenotypes were observed in four, one, and one case, respectively. In AFPC/HAC, the intestinal and mixed phenotypes were observed in cases one and three, respectively. In the composite gastric tumor, the tubular and hepatoid components evidenced the gastric phenotype, while the neuroendocrine component showed the mixed phenotype.
The composition of gastric AFPC/HAC and NEC is not yet understood clinicopathologically. Only a few cases [18–20] of gastric carcinoma consisting of AFPC/HAC and NEC have been reported. Okamaoto et al.  reported a composite gastric tumor consisting of AFPC and NEC, similar to the present case. Rassidakis et al.  reported gastric HAC with extensive neuroendocrine differentiation, and Ueda et al.  reported a composite tumor consisting of poorly differentiated adenocarcinoma and NEC with synchronous AFP expression. In the present study, no case of composite gastric tumor consisting of AFPC/HAC and NEC was observed, excluding Case 1. Furthermore, no case of gastric carcinoma with synchronous expression of AFP and neuroendocrine markers was observed. It was considered that this composition may be sporadic, and its occurrence may be very rare.
Gastric HAC frequently contains well- or moderately differentiated adenocarcinoma including tubular adenocarcinoma [4, 21, 22]. Kishimoto et al.  suggested that HAC developed from tubular adenocarcinoma, and Akiyama et al.  reported that gastric HAC and adenocarcinomatous components are of monoclonal origin. In the present study, although HAC concomitant with adenocarcinoma was observed in only one case, AFP was expressed in well- or moderately differentiated adenocarcinomas. On the other hand, mixed glandular exocrine tumors or composite neuroendocrine exocrine tumors have been identified in the stomach [24, 25], and NEC has frequently been associated with an adenocarcinoma component in the stomach [24–26]. These tumors have been defined as adenocarcinoma and NEC in combination in the same tumor. NEC predominantly arises from endocrine precursor cell clones occurring in preceding adenocarcinoma components . In other tumors, such as pancreatic carcinomas , HAC was found to be associated with another cellular component, either NEC or ductal carcinoma. Furthermore, several cases of hepatocellular carcinoma with neuroendocrine differentiation have been reported . In the present study, six gastric NEC cases were of the histologically pure type with no glandular component. Although, as mentioned above, cases such as the composite case presented are rare, gastric adenocarcinoma may also have the potential to differentiate into other miscellaneous types of carcinoma including either AFPC or NEC, such as in this case.
Based on immunohistochemistry results for cellular phenotype, Kumashiro et al.  recently reported that HAC and adenocarcinomatous components mainly evidenced the intestinal phenotype, and that the gastric phenotype was not observed in HAC. In the present study, AFPC/HAC expressed both intestinal and mixed phenotypes. On the other hand, the cellular phenotype has not yet been analyzed sufficiently in gastric NEC. Iwafuchi et al.  reported that, in Japanese subjects, gastric NEC frequently showed the intestinal phenotype in immunohistochemical analyses for CD10, MUC2, MUC5AC, and MUC6, with the complete intestinal phenotype constituting 45.5% of cases and incomplete intestinal phenotype consisting of 6.1% of the cases in gastric NEC. In the present study, four of the six EC cases evidenced the gastric phenotype immunohistochemically. In the composite tumor, both the hepatoid and the tubular components demonstrated the gastric phenotype immunohistochemically. The neuroendocrine component showed the presence of the mixed phenotype immunohistochemically, but this finding was considered as a gastric dominant type because MUC5AC, MUC6, and SOX2 were expressed in the neuroendocrine component. The finding that NEC evidenced the gastric dominant phenotype could be useful as a reference. These findings for cellular phenotype suggest that the hepatoid and neuroendocrine components may differ from the tubular component.
SOX2 has been shown to be downregulated in intestinal metaplasia in the stomach , and its expression was demonstrated in gastric adenocarcinoma with both gastric and mixed phenotype . CDX2 has been shown to be expressed in intestinal development  and was demonstrated in intestinal metaplasia and adenocarcinoma with the intestinal phenotype of the stomach [32, 33]. In the present study, gastric NEC frequently expressed SOX2 but not CDX2, while gastric AFPC/HAC frequently expressed CDX2 but not SOX2. These NEC cases showed the presence of the gastric phenotype, but AFPC/HAC may be considered as being strongly associated with the intestinal phenotype. In the composite tumor, the three components expressed SOX2, but CDX2 was expressed in the neuroendocrine component. Therefore, this composite gastric tumor was considered to be unique and different from typical gastric NEC or AFPC/HAC cases.
In conclusion, this unique composite gastric tumor predominantly evidenced the gastric phenotype, and the hepatoid and neuroendocrine components were considered to be differentiated from the tubular component. Generally, gastric AFPC/HAC predominantly evidenced the intestinal phenotype or mixed type, and NEC frequently showed the gastric phenotype.
D. Gitlin, A. Perricelli, and G. M. Gitlin, “Synthesis of -fetoprotein by liver, yolk sac, and gastrointestinal tract of the human conceptus,” Cancer Research, vol. 32, no. 5, pp. 979–982, 1972.View at: Google Scholar
Y. C. Chang, N. Nagasue, S. Abe, H. Taniura, D. D. Kumar, and T. Nakamura, “Comparison between the clinicopathologic features of AFP-positive and AFP- negative gastric cancers,” American Journal of Gastroenterology, vol. 87, no. 3, pp. 321–325, 1992.View at: Google Scholar
K. R. McIntire, T. A. Waldmann, C. G. Moertel, and V. L. Go, “Serum α-fetoprotein in patients with neoplasms of the gastrointestinal tract,” Cancer Research, vol. 35, no. 4, pp. 991–996, 1975.View at: Google Scholar
H. Ishikura, K. Kirimoto, M. Shamoto et al., “Hepatoid adenocarcinomas of the stomach: an analysis of seven cases,” Cancer, vol. 58, no. 1, pp. 119–126, 1986.View at: Google Scholar
N. Koide, A. Nishio, J. Igarashi, S. Kajikawa, W. Adachi, and J. Amano, “α-Fetoprotein-producing gastric cancer: histochemical analysis of cell proliferation, apoptosis, and angiogenesis,” American Journal of Gastroenterology, vol. 94, no. 6, pp. 1658–1663, 1999.View at: Publisher Site | Google Scholar
K. Matsui, M. Kitagawa, A. Miwa, Y. Kuroda, and M. Tsuji, “Small cell carcinoma of the stomach: a clinicopathologic study of 17 cases,” American Journal of Gastroenterology, vol. 86, no. 9, pp. 1167–1174, 1991.View at: Google Scholar
K. Matsui, X. M. Jin, M. Kitagawa, and A. Miwa, “Clinicopathologic features of neuroendocrine carcinomas of the stomach: appraisal of small cell and large cell variants,” Archives of Pathology and Laboratory Medicine, vol. 122, no. 11, pp. 1010–1017, 1998.View at: Google Scholar
N. Koide, A. Suzuki, H. Saito et al., “Gastric small cell carcinoma successfully treated by surgery and postoperative chemotherapy consisting of cisplatin and S-1: report of a case,” Surgery Today, vol. 37, no. 11, pp. 989–994, 2007.View at: Publisher Site | Google Scholar
A. Japanese Gastric Cancer, “Japanese classification of gastric carcinoma—2nd english edition,” Gastric Cancer, vol. 1, no. 1, pp. 10–24, 1998.View at: Google Scholar
M. Tatematsu, T. Tsukamoto, and K. Inada, “Stem cells and gastric cancer: role of gastric and intestinal mixed intestinal metaplasia,” Cancer Science, vol. 94, no. 2, pp. 135–141, 2003.View at: Publisher Site | Google Scholar
S. K. Chang, A. F. Dohrman, C. B. Basbaum et al., “Localization of mucin (MUC2 and MUC3) messenger RNA and peptide expression in human normal intestine and colon cancer,” Gastroenterology, vol. 107, no. 1, pp. 28–36, 1994.View at: Google Scholar
C. de Bolos, M. Garrido, and F. X. Real, “MUC6 apomucin shows a distinct normal tissue distribution that correlates with Lewis antigen expression in the human stomach,” Gastroenterology, vol. 109, no. 3, pp. 723–734, 1995.View at: Publisher Site | Google Scholar
S. B. Ho, A. M. Roberton, L. L. Shekels, C. T. Lyftogt, G. A. Niehans, and N. W. Toribara, “Expression cloning of gastric mucin complementary DNA and localization of mucin gene expression,” Gastroenterology, vol. 109, no. 3, pp. 735–747, 1995.View at: Publisher Site | Google Scholar
D. G. Silberg, G. P. Swain, E. Ran. Suh, and P. G. Traber, “Cdx1 and Cdx2 expression during intestinal development,” Gastroenterology, vol. 119, no. 4, pp. 961–971, 2000.View at: Google Scholar
X. L. Li, Y. Eishi, Y. Q. Bai et al., “Expression of the SRY-related HMG box protein SOX2 in human gastric carcinoma,” International Journal of Oncology, vol. 24, no. 2, pp. 257–263, 2004.View at: Google Scholar
F. T. Bosman, F. Carneiro, R. H. Hruban, and N. D. Theise, WHO Classification of Tumours of the Digestive System, 4th edition, 2010.
Y. Kumashiro, T. Yao, S. Aishima et al., “Hepatoid adenocarcinoma of the stomach: histogenesis and progression in association with intestinal phenotype,” Human Pathology, vol. 38, no. 6, pp. 857–863, 2007.View at: Publisher Site | Google Scholar
T. Okamoto, K. Ogasahara, M. Fujiki et al., “Primary coexistent neuroendocrine carcinoma, hepatoid adenocarcinoma, and tubular adenocarcinoma of the stomach with focal trophoblastic differentiation in metastatic lymph nodes,” Journal of Gastroenterology, vol. 38, no. 6, pp. 608–610, 2003.View at: Google Scholar
G. Z. Rassidakis, J. K. Delladetsima, S. P. Letsos, A. Polyzos, and A. Yannopoulos, “Hepatoid adenocarcinoma of the stomach with extensive neuroendocrine differentiation and a coexisting carcinoid tumour,” Histopathology, vol. 33, no. 2, pp. 186–188, 1998.View at: Google Scholar
T. Ueda, H. Neduka, S. Yamamoto et al., “A case of advanced gastric cancer of poorly differentiated adenocarcinoma associated with endocrine cell carcinoma element (in Japanese),” Rinsho Geka / Journal of Clinical Surgery, vol. 58, no. 7, pp. 1009–1012, 2003.View at: Google Scholar
S. Akiyama, G. Tamura, Y. Endoh et al., “Histogenesis of hepatoid adenocarcinoma of the stomach: molecular evidence of identical origin with coexistent tubular adenocarcinoma,” International Journal of Cancer, vol. 106, no. 4, pp. 510–515, 2003.View at: Publisher Site | Google Scholar
T. Motoyama, K. Aizawa, T. Fujiwara, Y. Endoh, and H. Watanabe, “Coexistence of choriocarcinoma and hepatoid adenocarcinoma in the stomach,” Pathology International, vol. 44, no. 9, pp. 716–721, 1994.View at: Google Scholar
T. Kishimoto, Y. Nagai, K. Kato, D. Ozaki, and H. Ishikura, “Hepatoid adenocarcinoma: a new clinicopathological entity and the hypotheses on carcinogenesis,” Medical Electron Microscopy, vol. 33, no. 2, pp. 57–63, 2000.View at: Publisher Site | Google Scholar
G. C. Yang and H. Rotterdam, “Mixed (composite) glandular-endocrine cell carcinoma of the stomach. A report of a case and review of literature,” American Journal of Surgical Pathology, vol. 15, no. 6, pp. 592–598, 1991.View at: Google Scholar
N. Rayhan, T. Sano, Z. R. Qian, A. K. Obari, and M. Hirokawa, “Histological and immunohistochemical study of composite neuroendocrine-exocrine carcinomas of the stomach,” Journal of Medical Investigation, vol. 52, no. 3-4, pp. 191–202, 2005.View at: Publisher Site | Google Scholar
K. Nishikura, H. Watanabe, M. Iwafuchi, T. Fujiwara, K. Kojima, and Y. Ajioka, “Carcinogenesis of gastric endocrine cell carcinoma: analysis of histopathology and p53 gene alteration,” Gastric Cancer, vol. 6, no. 4, pp. 203–209, 2003.View at: Publisher Site | Google Scholar
O. Hameed, H. Xu, S. Saddeghi, and H. Maluf, “Hepatoid carcinoma of the pancreas: a case report and literature review of a heterogeneous group of tumors,” American Journal of Surgical Pathology, vol. 31, no. 1, pp. 146–152, 2007.View at: Publisher Site | Google Scholar
R. Yamaguchi, O. Nakashima, T. Ogata, K. Hanada, T. Kumabe, and M. Kojiro, “Hepatocellular carcinoma with an unusual neuroendocrine component,” Pathology International, vol. 54, no. 11, pp. 861–865, 2004.View at: Publisher Site | Google Scholar
M. Iwafuchi, F. Kusama, T. Watanabe et al., “Endocrine cell carcinoma of the stomach (in Japanese),” Byori to Rinsho (Pathology and Clinical Medicine), vol. 23, no. 9, pp. 966–973, 2005.View at: Google Scholar
T. Tsukamoto, K. Inada, H. Tanaka et al., “Down-regulation of a gastric transcription factor, Sox2, and ectopic expression of intestinal homeobox genes, Cdx1 and Cdx2: inverse correlation during progression from gastric/intestinal-mixed to complete intestinal metaplasia,” Journal of Cancer Research and Clinical Oncology, vol. 130, no. 3, pp. 135–145, 2004.View at: Publisher Site | Google Scholar
T. Tsukamoto, T. Mizoshita, M. Mihara et al., “Sox2 expression in human stomach adenocarcinomas with gastric and gastric-and-intestinal-mixed phenotypes,” Histopathology, vol. 46, no. 6, pp. 649–658, 2005.View at: Publisher Site | Google Scholar
H. Seno, M. Oshima, M. A. Taniguchi et al., “CDX2 expression in the stomach with intestinal metaplasia and intestinal-type cancer: prognostic implications,” International Journal of Oncology, vol. 21, no. 4, pp. 769–774, 2002.View at: Google Scholar
R. Almeida, E. Silva, F. Santos-Silva et al., “Expression of intestine-specific transcription factors, CDX1 and CDX2, in intestinal metaplasia and gastric carcinomas,” Journal of Pathology, vol. 199, no. 1, pp. 36–40, 2003.View at: Publisher Site | Google Scholar