Amphicrine carcinoma of the stomach and intestine: a clinicopathologic and pan-cancer transcriptome analysis of a distinct entity

This study was performed in accordance with local ethical and legal requirements after approval by the Ethics Committee of Fudan University Shanghai Cancer Center (FUSCC), and written informed consent was obtained from all participants or their appropriate surrogates. A total of ten cases of amphicrine carcinoma of the stomach or intestine were retrieved from the consultation and surgical pathology files of the Department of Pathology at FUSCC between 2009 and 2017. Available medical records, including imaging study reports, were reviewed to obtain clinical data such as age, sex, presenting symptoms, endoscopic descriptions, treatment, and outcome; for the consultation cases, contributing physicians were contacted.

Available gross images, descriptions and histologic sections were reviewed by three pathologists (authors: DH, FR and SJN) to confirm the diagnosis and further characterize the histological findings. The pathologic criteria for diagnosis of amphicrine carcinoma requires exocrine and neuroendocrine features in the same neoplastic cell, which shows a divergent immunophenotype [1]. According to histologic features, our cases were divided into low, intermediate and high grades using the grading system recommended by Yozu [8]. This methodology is used to grade appendiceal goblet cell carcinomas by assessing the proportion of the tumor with tubular or clustered growth. Tumors with > 75% tubular or clustered growth were classified as having a low-grade pattern, which is characterized by small and cohesive clusters of goblet cells with or without lumina. The cells in these clusters had low to at most moderate cytologic atypia and infrequent mitoses, sometimes with peripheral localization of nuclei. Tumors with 50% to 75% tubular growth were classified as having an intermediate-grade pattern, and tumors with < 50% tubular growth were classified as having a high-grade pattern. These growth patterns deviated from the low-grade pattern and showed several forms, including single-file or sheet-like growth of signet ring-like cells. Clusters of tumor cells, especially in mucin-poor areas, had increased cytologic atypia and increased mitotic activity.

Immunohistochemistry for pankeratin (clone AE1/3, 1:150, Dako), CgA (clone LK2H101 + PHE5, Roche), Syn (clone MRQ-40, 1:400, Roche), CD56 (clone 123C3, 1:80, Dako) and Ki67 (clone 30-9, Roche) were performed using a Ventana BenchMark XT Automated Staining System. Alcian blue staining was performed to evaluate mucin content using an Alcian blue staining kit (BASO) following the manufacturer’s instructions.

The genetic data generated for amphicrine carcinomas were compared with data from a set of neuroendocrine tumors and gastric adenocarcinomas, which were genetically analyzed by the same 90-gene real-time PCR assay [9]. In brief, manual macrodissection of tumor-rich areas from unstained slides of formalin-fixed, paraffin-embedded tissue was performed under microscopy with guidance from hematoxylin and eosin staining. Total RNA was isolated from formalin-fixed, paraffin-embedded (FFPE) tissue sections using an FFPE Total RNA Isolation Kit (Canhelp Genomics, Hangzhou, China). The concentration and purity of total RNA were determined by spectrometry according to the manufacturer’s instructions. Next, cDNA was generated from isolated total RNA using a high-capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Applied Biosystems, Foster City, CA, United States). The A260/A280 of total RNA isolated from tissue sections from amphicrine carcinoma, NET and gastric cancer patients was 1.89–2.00. The expression level of each of the 90 genes was measured in an Applied Biosystems 7500 Real-Time PCR system using TaqMan Gene Expression Assays (Applied Biosystems). Normalized gene expression intensities were shifted to set the mean to 0 and rescaled to set the STD to 1 to enhance the expression differences. The average linkage hierarchical clustering method was performed, where the metric of similarity was the Pearson correlation between every pair of samples. In addition, relative mRNA expression intensities were triple detected for each specimen of all samples, including amphicrine carcinoma (AC), neuroendocrine tumor (NET) and stomach adenocarcinoma (STAD) samples.

Biological network analysis and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis were performed using NetworkAnalyst software (version 3.0) (Cite https://​doi.​org/​10.​1093/​nar/​gkz240). Protein–protein interactions have been retrieved from IMEx Interactome Database (Cite https://​doi.​org/​10.​1093/​nar/​gks1147). A minimum network was generated by keeping seed proteins as well as minimum essential non-seed proteins to study the fundamental interactions.

The clinical features and staging parameters for our ten amphicrine cancer cases are summarized in Table 1. All patients were male, with a mean age of 63 years (ranging from 56 to 69). The presenting symptoms in seven patients included upper abdominal pain, hematochezia and hematemesis. The eight gastric neoplasms were located throughout the stomach, and the antrum was the most commonly involved region. Another two cases in the intestine arose from the rectum. According to the histologic evaluation of tumor grades, most low-grade tumors were in an early T stage (T1 or T2). No patient had nodal or distant metastasis at presentation; however, one patient with stage IIIA disease had one lymph node involved. Lymphovascular invasion was present in only that patient (25%), and perineural invasion was not observed in any low-grade sample. In contrast, the high-grade group and mixed group had a high proportion of late-stage disease. The percentage of patients with lymph node involvement was increased (75%), as did the percentage of patients with synchronous distant metastasis (33%). Perineural invasion was observed in 2 high-grade tumors, and lymphovascular invasion was present in 3 cases.

Grossly, the tumor sizes ranged from 2 to 5 cm (mean, 3.6 cm) in the maximum dimension. In the 9 cases for which endoscopic or gross information was available, the ulcerative nature of the tumor was described in seven (Fig. 1a). The remaining two tumors were documented as fungating lesions. After assessment of tubular and clustered components, 4 tumors were categorized as low-grade, 6 as high-grade, and none as intermediate-grade.

The histology was somewhat complicated but showed three components consisting of three types of neoplasms: (1) low-grade amphicrine carcinoma (may mix with other components but less than 30% of tumor cell population); (2) high-grade amphicrine carcinoma (may mix with other components but less than 30% of tumor cell population); and (3) mixed amphicrine-neuroendocrine carcinoma (amphicrine carcinoma and other carcinoma, each of which according for more than 30% of tumor cell population). After assessment of the components and grades, 4 tumors were in the low-grade group, 4 were in the high-grade group, and 2 were in the mixed group (Table 2). In the low-grade group, one patient had another minor conventional adenocarcinoma component comprising 5% of the tumor. By definition, the low-grade group included tumors with up to 25% high-grade components, but none of the included tumors showed combination with any high-grade components. The most common histologic architectures in the low-grade category were tubular growth with intracellular mucin and peripheral placement of nuclei (Fig. 2), which resembled goblet cell carcinoid/carcinoma in the appendix. None of the cases in this group had single-file infiltration by signet ring-like cells. In 2 cases with an extracellular mucin pool, the tumor clusters maintained their cohesive, uniform appearance, and resembled disrupted intestinal crypts. In the high-grade group, one case presented one area of conventional adenocarcinoma (5%) and another area of low-grade amphicrine cancer characterized by tubular structures lined by goblet-like cells (45%). Another high-grade case had a low-grade component of 5%. Some nontubular growths were common among the high-grade tumors, which deviated from the architecture in the low-grade group. The most frequent forms were fusion of goblet cell clusters and disorganized growth by signet ring-like cells. Single files of goblet cells or signet ring-like cells were also fairly common (Fig. 1). Less common patterns included mucin-poor areas composed of tumor nests with high cytologic grade that resembled conventional adenocarcinoma and tumor clusters floating in mucin formed by signet ring-like cells with nuclear atypia and mitotic frequency. There were 2 cases of mixed amphicrine-neuroendocrine carcinoma that had a high-grade amphicrine component intermixed with areas of conventional neuroendocrine carcinoma (NEC), comprising 50% and 60% of the tumor (Fig. 3). Similar to the growth patterns in the high-grade group, the amphicrine components were aggregates and fusions of goblet cells forming complex disordered structures, which differentiated these areas from typical NEC areas in histologic architecture.

The immunohistochemical staining results for all cases is summarized in Table 3. All tumor components were positive for cytokeratin (AE1/3). Amphicrine components in all cases had at least focal or patchy positivity for synaptophysin and chromogranin A. CD56 staining, if performed, also showed cytoplasmic positivity in most tumors. Those neuroendocrine markers did not differ between the low-grade and high-grade groups, only indicating the amphicrine differentiation of tumor cells. Mucin was visualized by Alcian blue staining, identifying the goblet cells and signet ring-like cells with true intracellular mucin in amphicrine components. The range of the Ki67 index was 5–40% in low-grade tumors and 20–70% in high-grade tumors, suggesting a difference in cell proliferation between the two groups.

A previous study established a 90-gene expression signature to accurately classify a broad spectrum of tumor types [10]. In our research, the genetic data generated for amphicrine carcinomas, 4 cases with available blocks (Table 2), were compared with data from a set of four neuroendocrine tumors and four gastric adenocarcinomas. The average linkage hierarchical clustering method was performed, where the metric of similarity was the Pearson correlation between the 90-gene expression profiles of the samples (Fig. 4a). The heatmap representation of mRNA expression shows that all amphicrine carcinomas and conventional adenocarcinomas were clustered together. Relatedness of clusters was identified between amphicrine carcinomas and NETs. The relative expression intensity of the 12 genes with the greatest variation was used to differentiate related groups in these samples (Fig. 4b). The expression levels of the NPTX2, PCP4, ISL1, IGFBP2, GPX3, VEGFA, ID4 and GPM6B genes were synchronously decreased in the AC and STAD groups compared with their expression levels in NET entities. High levels of CEACAM5, AGR2, CXCL14 and S100P gene expression were simultaneously observed in the AC and STAD groups, while the NET group had significantly lower expression levels of those genes. In the further analysis of the minimum protein–protein network, 12 genes were generated as the seeds, and 10 of them are centered on VEGFA node (Fig. 4c). Next, enrichment analysis of genes was performed in the network. It revealed that the most significant KEGG pathways were signaling pathways regulating pluripotency of stem cells, TGF-β signaling pathway, pathways in cancer, etc. (Additional file 1: Table S1). Upon inspection, it was evident that amphicrine carcinoma exhibited transcriptional homogeneity with conventional adenocarcinoma and genetic diversity from NET. Moreover, the gene profiles in the low-grade and high-grade groups were indistinguishable in our limited cohort, exhibiting similar mRNA expression levels in low-grade (cases 2 and 3) and high-grade (cases 6 and 7) tumors.

Regarding treatment, seven patients underwent complete resection, and three of these received chemotherapy after surgery. Two patients received adjuvant chemotherapy or chemoradiotherapy after biopsy. One patient with early-stage disease did not receive any treatment. Survival information was available for all patients (Table 1), with a follow-up period ranging from 6 to 63 months. A total of 3 patients died of disease—at 11, 12 and 42 months, and all had tumors with a high-grade pattern. Seven patients were alive after a follow-up period ranging from 6 to 63 months. Of these patients, 6 had experienced no recurrence or metastasis after treatment; 4 of these patients had low-grade tumors. The survival rate was 100% in the low-grade group, compared to 50% in the high-grade group (P < 0.05). Kaplan–Meier survival curves showed a dramatic difference in overall survival between the low-grade and high-grade groups (Fig. 5), suggesting that high-grade morphology is associated with poorer prognosis.