This article is mentioned in:

Introduction

Gastric carcinoma (GC) is one of the most prevalent cancers worldwide, ranking fifth in incidence and fourth in cancer-related mortality globally (1). Risk factors for GC include Helicobacter pylori infection, dietary habits, smoking and genetic predispositions (2). Despite advancements in surgical techniques, chemotherapy and targeted therapies, early diagnosis of GC remains challenging due to non-specific symptoms, often resulting in a poor prognosis at the advanced stages (3). Current challenges in GC management include the development of treatment resistance and the need for personalized therapeutic approaches (4). Phosphoinositide-3-kinase regulatory subunit 1 (PIK3R1) is a key regulatory subunit of phosphoinositide 3-kinases, acting as a tumor suppressor by inhibiting the activity of p110α of class I PI3K (PIK3CA) (5). Mutations in PIK3R1 have been implicated in tumorigenesis across multiple organs (6), and elevated levels of PIK3R1 are associated with the progression of GC (7,8).

ATRX, a regulator of the chromatin state, gene expression, cellular senescence and DNA damage repair, belongs to the SWI/SNF protein family. Loss of ATRX expression is associated with activation of the alternative lengthening of telomeres pathway, contributing to the replicative immortality of cancer cells (9,10). Loss-of-function (LOF) mutations in ATRX are linked to aggressive traits such as tumour growth, migration, invasion and metastasis in osteosarcoma (9). Although the role of ATRX in GC is not well understood, female patients with GC harbouring ATRX mutations exhibit microsatellite instability-high subtypes, elevated tumour mutational burden and increased programmed cell death ligand-1 (PD-L1) expression. Moreover, those patients exhibit prolonged survival when treated with immune checkpoint inhibitors (11).

RNA binding motif protein 10 (RBM10), an alternative splicing factor protein, regulates RNA transcription and expression, including pre-mRNA splicing, mRNA stabilisation and mRNA transcription. RBM10 is recognised as a tumour suppressor gene that inhibits proliferation, invasion and metastasis while promoting apoptosis during tumorigenesis (12,13). However, RBM10 has also been shown to promote cancer via a negative feedback mechanism, driven by the upregulation of RBM10 or its homolog (13). As with ATRX, there have been few studies on RBM10 in GC, but upregulation of RBM10 has been confirmed in The Cancer Genome Atlas (TCGA) dataset (14).

Epstein-Barr virus (EBV) is an oncogenic virus that can contribute to the malignant transformation of normal gastric cells (15). The primary route by which EBV infects epithelial cells is through direct cell-to-cell contact mediated by B lymphocytes. Following persistent infection, EBV enters latency, during which it does not integrate into the host genome but replicates concurrently with the host cells (16). EBV exhibits a distinct gene expression profile depending on the latency type. EBV-positive GC (EBVpGC) is closely associated with latency type 1 genes, including EBV-encoded small RNA 1 (EBER1), EBER2 and EBV nuclear antigens 1 (16). EBVpGC is distinguished from other types of GC by its characteristic genomic aberrations, distinct clinicopathological features and generally favourable prognosis (17). In GC, PIK3R1 mutations are associated with EBVpGC cases, whereas TP53 inactivation is typically identified in EBV-negative GC (EBVnGC) cases (18).

In the present study, a GC case with a poor prognosis, harbouring LOF mutations in the PIK3R1, ATRX and RBM10 genes, with diverse histological features associated with EBV infection and TP53 inactivation is reported.

Case report

In April 2023, a 62-year-old man presented to the Emergency Department of Ewha Woman's University Mokdong Hospital (Seoul, Republic of Korea) with hematemesis. The patient's initial vital signs were stable. Although the patient had a history of infarction in the left middle cerebral artery and was receiving antithrombotic therapy (aspirin and clopidogrel), there was no history of cancer. Abdominal computed tomography revealed an encircling mass in the lower gastric body (Fig. 1A), multiple metastatic lymph nodes near the celiac axis and gastroepiploic vessels as well as peritoneal seeding, suggestive of unresectable advanced GC. Following admission, numerous well-defined lesions of varying sizes, highly suspected of representing hepatic metastasis, were identified in both hepatic lobes via diffusion-weighted magnetic resonance imaging (Fig. 1B). Esophagogastroduodenoscopy confirmed active bleeding from exposed vessels with detached blood clots (Fig. 1C). Given the active bleeding and the challenges of initiating chemotherapy immediately, palliative subtotal gastrectomy and liver excisional biopsy were selected as the initial intervention. Macroscopically, a well-defined, ulceroinfiltrative mass measuring 7.0×6.0×1.2 cm was identified on the posterior wall of the gastric body and antrum (Fig. 2A). The cut surface was greyish-white, solid and necrosis-free, with direct invasion into the serosa (Fig. 2B). For histological examination, the tissues were paraffin-embedded following fixation in 10% neutral-buffered formalin at room temperature (~25°C) for 24 h. Sections were cut at a thickness of 4 µm and stained with haematoxylin for 5 min and eosin for 2 min at room temperature. The prepared slides were examined using a light microscope (Leica DM2000; Leica Microsystems, Inc.). Microscopically, the tumour consisted of three distinct histological components: Gastric adenosquamous carcinoma (GASC), GC with lymphoid stroma (GCLS) and poorly differentiated adenocarcinoma (PDAD), accounting for 30, 40 and 30% of the tumour volume, respectively. The GASC component comprised of both adenocarcinoma and squamous cell carcinoma, characterised by large round-to-oval nuclei with abundant eosinophilic cytoplasm, intercellular bridges and multifocal glandular formations (Fig. 3A). No keratin pearls were observed. This component was primarily located in the mucosal and submucosal layers. The GCLS component exhibited abundant lymphocytic infiltration around undifferentiated carcinoma clusters with lacy or tubular growth patterns, pleomorphic polygonal cells, prominent nucleoli and vesicular chromatin (Fig. 3B). The PDAD component featured relatively small, monotonous, hyperchromatic nuclei with coarse chromatin, prominent nucleoli and brisk mitotic activity. This component had a mucinous stromal background, with rare glandular formations and minimal lymphocytic infiltration (Fig. 3C). The GCLS and PDAD components were primarily located in the muscularis propria and subserosa.

Figure 1. Images and endoscopic findings. (A) Abdominal computed tomography scan of a large encircling mass in the gastric lower body (arrows). (B) Diffusion-weighted magnetic resonance imaging of multiple lesions in both hepatic lobes (arrows). (C) Esophagogastroduodenoscopy image of a ulceroinfiltrative mass with bleeding focus.

Figure 2. Macroscopic findings of the tumour. (A) A large, well-defined and ulceroinfiltrative mass (7.0×6.0×1.2 cm) as identified at the posterior wall of the lower body and antrum (arrows). (B) The tumour directly invaded the serosa (the area of the tumour is indicated by arrows).

Figure 3. Microscopic images depicting three distinct histological components of the tumour alongside results from EBV in situ hybridization and p40 staining. (A) GASC component exhibiting large round to oval nuclei with abundant eosinophilic cytoplasm, intercellular bridges and multifocal glandular formation (H&E; scale bar, 200 µm). (B) GCLS component exhibiting abundant lymphocytic infiltration around undifferentiated carcinoma cell clusters (H&E; scale bar, 200 µm). (C) PDAD component exhibiting hyperchromatic nuclei with mucinous stromal backgrounds with intermittent exhibition of glandular formation (H&E; scale bar, 200 µm). EBV was not detected in the (D) GASC component (scale bar, 200 µm), but was detected in the (E) GCLS (scale bar, 200 µm) and (F) PDAD (scale bar, 200 µm) components. Immunoreactivity for p40 was identified in the (G) GASC component (scale bar, 200 µm), but was negative in the (H) GCLS (scale bar, 200 µm) and (I) PDAD (scale bar, 200 µm) components. GASC, gastric adenosquamous carcinoma; GCLS, gastric carcinoma with lymphoid stroma; PDAD, poorly differentiated adenocarcinoma; EBV, Epstein-Barr virus.

To assess biomarker expression and the molecular characteristics of the three histological components, mucicarmine staining, immunohistochemical (IHC) analysis, EBV in situ hybridisation and next-generation sequencing (NGS) were conducted. For mucicarmine staining, the Artisan Mucicarmine Stain Kit (Agilent Technologies; cat. no. AR168) was used. The protocol involved deparaffinizing and hydrating sections, staining with Weigert's Iron haematoxylin for 10 min at room temperature, followed by incubation in Working Mucicarmine Solution for 30–60 min at room temperature. After washing, the sections were counterstained with Tartrazine Solution for 30 sec to 1 min at room temperature, then dehydrated, cleared and mounted. This method stains mucin in pink, nuclei in black and the background in yellow. For immunostaining, the tissues were paraffin-embedded and fixed in 10% neutral-buffered formalin at room temperature for 24 h. Sections were cut at a thickness of 4 µm and dried at 55°C for 3 h. Heat-induced epitope retrieval was performed at pH 9.0 for 20 min using a Leica-Bond Autostainer and a Bond Polymer Refine Detection Kit (cat. no. DS9800; Leica Biosystems). Endogenous peroxidase activity was quenched using 3% hydrogen peroxide at room temperature for 10 min. Primary antibodies were incubated at room temperature for 1 h, with details of the antibodies and dilutions provided in Table SI. Secondary antibody incubation was performed using the aforementioned Bond Polymer Refine Detection Kit, and staining was visualized with diaminobenzidine (DAB) chromogen. The prepared slides were examined using a light microscope (Leica DM2000; Leica Microsystems, Inc.). p40 and p63 staining identified the squamous cell carcinoma component, while synaptophysin and chromogranin staining detected the neuroendocrine carcinoma component. To classify the molecular subtype according to the Asian Cancer Research Group, mutL homolog 1 (MLH1), mutS homolog 2 (MSH2), p53 and E-cadherin staining were performed (19). PD-L1 and programmed cell death protein-1 (PD-1) are known to be frequently upregulated, while human epidermal growth factor receptor 2 (HER2) typically shows negative expression in EBVpGC (20). EBV in situ hybridisation was performed to detect EBER RNA sequences in formalin-fixed, paraffin-embedded tissue sections. The tissues were fixed in 10% neutral-buffered formalin at room temperature for 24 h and embedded in paraffin. Sections were cut at a thickness of 4 µm and deparaffinized prior to hybridization, and permeabilized using proteinase K (cat. no. S3020; Dako; Agilent Technologies, Inc.) at 37°C for 15 min. Hybridization was conducted using Bond hybridization buffer (cat. no. PB0589; Leica Biosystems) with the biotin-labelled Epstein-Barr Virus Early EBER RNA Probe (Invitrogen; Thermo Fisher Scientific, Inc.; cat. no. 551019) at a concentration of 0.5 µg/ml, incubated at 45°C for 20–60 min. Post-hybridization washes were performed with saline-sodium citrate buffer at 45°C to remove excess probes, followed by blocking endogenous peroxidase activity using 3% hydrogen peroxide at room temperature for 10 min. Detection was carried out using the Bond Polymer Refine Detection Kit (Leica Biosystems, Ltd.), and visualization was achieved with DAB chromogen. Counterstaining was performed with haematoxylin for 5 sec. The slides were examined under a light microscope (Leica DM2000; Leica Microsystems, Inc.). For NGS, the 5-µm-thick formalin-fixed, paraffin-embedded tissue sections were deparaffinized using alcohol and subsequently hydrated. Sections containing tumour cell-rich regions were microdissected with an ethanol-dipped scalpel. The tissue underwent washing and digestion with phosphate-buffered saline and proteinase K (Qiagen GmbH). DNA extraction was performed using a QIAamp DSP DNA Kit (Qiagen GmbH). Nucleic acid quantification was carried out using a Qubit 4 Fluorometer and fluorescence-based quantitation assays (Thermo Fisher Scientific, Inc.). The quality and integrity of the extracted DNA were assessed using agarose gel electrophoresis and the purity was verified by measuring A260/A280 and A260/A230 ratios with a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc.). Library preparation utilized the Ion AmpliSeq Library Preparation kit with nucleic acid samples (Thermo Fisher Scientific, Inc.; cat. no. 4475345). DNA sequencing was conducted employing the IonTorrent S5 XL, alongside a control cell line mixture (Horizon Discovery; Revvity Discovery Limited) and a targeted gene panel, the Oncomine Comprehensive Assay Plus (Thermo Fisher Scientific, Inc, cat. no. A48577). The sequencing utilized 200 bp single-end reads. The final library was loaded at a concentration of 50 pM onto the Ion 550 Chip, optimized for the Ion Torrent S5 XL system. This panel facilitates the identification of single nucleotide variants and copy number variants from a pool of 500 gene mutations and 69 gene fusions. For genomic data analysis and variant calling the Ion Reporter software v5.2 was employed (Thermo Fisher Scientific, Inc.).

Representative histological features and images of p40 staining and EBV in situ hybridisation for the three tumour components are shown in Fig. 3. The GASC component revealed diffuse immunoreactivity for p40 (Fig. 3G) and p63 as well as focal identification of mucin in the mucicarmine staining (Fig. S1) but was negative for EBV in situ hybridisation (Fig. 3D). The GCLS and PDAD components were diffusely positive for EBV in situ hybridisation (Fig. 3E and F, respectively) and negative for p40 (Fig. 3H and I, respectively) and p63 staining. All components showed negativity for chromogranin, synaptophysin, PD-1 and HER2 staining, and were positive for MLH1, MSH2 and E-cadherin staining (Fig. S2). The p53 staining showed patchy positivity in all components (Fig. S3). PD-L1 expression (SP142 and SP263) was <1% positive in immune cells and negative in tumour cells in all components. In total, 29 somatic mutations and 14 deletions were identified using the NGS panel; 2 missense mutations and 2 deletions were clinically relevant, whereas the remaining 27 mutations and 12 deletions were variants of uncertain significance. The tumour mutation burden ranged from 5.68 to 6.63, with microsatellite stability in all components. All three histologically distinct components shared common a missense mutation in PIK3R1 (c.1690A>G, p. Asn564Asp) and homozygous deletions of ATRX and RBM10 (both with copy number 0). The PDAD component exclusively harboured a missense mutation in TP53: c.730G>A (p. Gly244Ser). No PIK3CA and ARID1A mutations, which are prevalent in EBVpGC (15), were identified. The LOF or likely LOF events in PIK3R1, TP53, ATRX and RBM10, which function as tumour suppressor genes, are classified as oncogenic or likely oncogenic based on the OncoKB database (https://www.oncokb.org/). OncoKB compiles mutation effect information into a database based on various experimental data, functional evidence and in silico evidence. According to their standard operating procedure file, a predictive algorithm was utilized, incorporating two programs: SIFT and PolyPhen. The results of NGS analysis for each histologically distinct component are presented in Table I.

Table I. Clinically relevant molecular alterations identified by next-generation sequencing.

Table I.

Clinically relevant molecular alterations identified by next-generation sequencing.

Gene	Type of genetic alteration	Mutation effect based on in silico analysis	Oncogenic effect	HGVS nomenclature	Associated histology in the case
PIK3R1	Missense mutation	LOF	Oncogenic	c.1690A>G (p. Asn564Asp)	GASC, GCLSa and PDADa
ATRX	Deletion	Likely LOF	Likely oncogenic	g.76763769_77041552del	GASC, GCLSa and PDADa
RBM10	Deletion	Likely LOF	Likely oncogenic	g.47006798_47046088del	GASC, GCLSa and PDADa
TP53	Missense mutation	LOF	Oncogenic	c.730G>A (p. Gly244Ser)	PDADa

a Epstein-Barr virus-detected by in situ hybridization. HGVS, human genome variation society; LOF, loss-of-function; GASC, gastric adenosquamous carcinoma; GCLS, gastric carcinoma with lymphoid stroma; PDAD, poorly differentiated adenocarcinoma; PIK3R1, phosphoinositide-3-kinase regulatory subunit 1; RBM10, RNA binding motif protein 10.

Lymph node metastasis was identified in 14 out of 19 lymph nodes (Fig. 4A) and hepatic metastasis was identified via intraoperative liver biopsy (Fig. 4B). The histological features of the metastasis were exclusively of the GCLS component. Accordingly, the TNM stage was determined to be pT4aN3aM1 based on the 8th edition of the American Joint Committee on Cancer Staging System (21). The patient was transferred to another hospital immediately after surgery without chemotherapy or radiotherapy and passed away 6 months later.

Figure 4. Representative microscopic images of metastatic lesions. (A) lymph node and (B) hepatic parenchyma (H&E; scale bar, 200 µm).

Discussion

In the present study a rare case of GC with three distinct histological features, each displaying a different EBV infection state and molecular alteration, yet all sharing common mutations in PIK3R1, ATRX and RBM10 was reported. Specifically, the tumour comprised a GASC component without EBV infection, a GCLS component with EBV infection and active TP53, and a PDAD component with concurrent EBV infection and TP53 inactivation. Based on both the molecular and histological findings, the pathogenesis of the present case may be hypothesised as follows: A subset of the tumour, harbouring mutations in PIK3R1, ATRX and RBM10, underwent EBV-driven malignant transformation, manifesting histologically as GCLS and PDAD. Subsequently, the PDAD component of this subset acquired TP53 inactivation. Evidence from previous studies supports this hypothesis. Knockdown of ATRX can induce EBV reactivation (22). Additionally, patients with activated PI3Kδ syndrome due to heterozygous gain-of-function mutations in PIK3CD or PIK3R1 are known to be susceptible to EBV, raising the possibility of EBV susceptibility in patients with LOF mutations in PIK3R1 (23). Furthermore, RBM10 has antiviral properties, as upregulation in cells infected with dengue virus inhibits viral replication, while RBM10 knockdown promotes replication (24). Thus, RBM10 loss of function may have contributed to the development of EBVpGC.

The pathogenesis of GASC, which constitutes 0.25% of all primary GC cases (20), remains unclear due to its rarity. To date, no studies have investigated the relationship between GASC and PIK3R1. EBVpGC is characterised by distinctive genomic alterations, including frequent DNA hypermethylation, PIK3CA and ARID1A mutations, and an absence of TP53 mutations (15). EBVpGC is often associated with abundant lymphocytic infiltration surrounding adenocarcinoma clusters, a feature known as GCLS (20). Patients with EBVpGC generally exhibit a longer median survival time compared with those with EBVnGC, with survival times of 8.5 vs. 5.3 years, respectively (17). Although EBVpGC accounts for <10% of all GC cases, it is regarded as a unique entity, distinct from typical gastric adenocarcinoma (15,17). Mutations in PIK3R1 and TP53 in EBVpGC were documented in a previous TCGA study. Of the 73 EBVpGC cases, 8 (11%) harboured PIK3R1 mutations and 11 (15.1%) had TP53 mutations. By contrast, among the 75 EBVnGC cases, only 1 (1.3%) exhibited a PIK3R1 mutation, while TP53 mutations were found in 47 cases (62.7%) (18). These findings suggest that although PIK3R1 mutations are rare in GC, they may be associated with EBV infection. While most TP53 mutations are linked to EBVnGC, they can also occur in EBVpGC.

In the present case, the GCLS component displayed typical histological features of EBVpGC with EBV infection. However, unlike prior studies suggesting a favourable prognosis for EBVpGC, the GCLS component resulted in hepatic and lymph node metastases, and the patient passed away within 6 months. This outcome suggests that metastasis may have been driven by the PIK3R1 mutation, consistent with reports of higher PIK3R1 expression in advanced stages (III and IV) compared with early stages (I and II) of GC (6).

To date, several cases of mixed EBVpGC with ASC have been reported with histopathological and IHC analyses, but molecular features have not been explored using NGS (25–27). Although a previous study on GC with EBV heterogeneity exists, it differs from the present case in that it did not demonstrate histological heterogeneity or include alterations in PIK3R1, ATRX and RBM10 (28). The underlying cause of EBV infection heterogeneity remains unknown. More comprehensive studies, including multicentre research, are needed to further investigate the implications of EBV infection heterogeneity across diverse histological features.

In conclusion, diverse molecular alterations and EBV infection heterogeneity can be observed in mixed GC across distinct histological components. As these alterations and heterogenous EBV status may influence patient prognosis and treatment strategies, a thorough evaluation of molecular features alongside histological characteristics is crucial.

Supplementary Material

Supporting Data

Supporting Data

Acknowledgements

Not applicable.

Funding

This work was supported by a grant from Kyung Hee University in 2024 (grant no. KHU-20241076) and a grant from the Asan Institute for Life Sciences, Asan Medical Center, Seoul, Korea (grant no. 2021IL0040).

Availability of data and materials

The NGS data generated in the present study may be found in the NCBI Sequence Read Archive under accession no. PRJNA1182944 or under the following URL: https://www.ncbi.nlm.nih.gov/sra/PRJNA1182944. All other data generated in the present study may be requested from the corresponding author.

Authors' contributions

Conceptualisation and writing of the original draft were conducted by SUJ. Case selection and investigation of the clinical data was conducted by YK. Review of the H&E slides was conducted by SUJ, EC and SK. NGS analysis and funding acquisition was conducted by SK. Molecular pathology consultation was conducted by JB. Immunohistochemical staining analysis and reviewing and editing the manuscript was conducted by SUJ and SK. SUJ and SK confirm the authenticity of all the raw data. All authors read and approved the final version of the manuscript.

Ethics approval and consent to participate

This retrospective study involves experimental analysis of human tissue specimens collected following surgical procedures. The Institutional Review Board at Ewha Woman's University Mokdong Hospital (Seoul, Republic of Korea) reviewed and approved the study protocol (approval no. 2024-05-004), including a waiver of consent due to the retrospective nature of the study and the patient's death. The study was conducted in accordance with the Declaration of Helsinki.

Patient consent for publication

The Institutional Review Board at Ewha Woman's University Mokdong Hospital (Seoul, Republic of Korea) waived the requirement of patient consent for publication due to the retrospective nature of the study and the patient's death.

Competing interests

The authors declare that they have no competing interests.

Authors' information

Se Un Jeong ORCID, 0000-0001-8399-5792; Euno Choi ORCD, 0000-0002-9284-9276; Yongil Kim ORCID, 0000-0002-4131-364X; Jaeyoung Byeon ORCID, 0009-0002-7316-6549; So-Woon Kim ORCID, 0000-0002-9840-848X.

References