Abstract
Background:
ARID1A mutations are frequent in gastric cancer (GC) and may indicate an immune-active tumor microenvironment.
Objectives:
This study aimed to assess the efficacy of the immune checkpoint inhibitor (ICI) plus chemotherapy as first-line therapy in advanced GC and explore its association with key molecular features according to the status of ARID1A-mutation.
Design:
This was a retrospective, single-center cohort study.
Methods:
We analyzed 258 patients with advanced HER2-negative GC who received ICI plus chemotherapy between 2022 and 2024. ARID1A mutation status and other molecular features were assessed using next-generation sequencing. Also, the status for Epstein–Barr virus (EBV), programmed death-ligand 1 (PD-L1), and claudin 18.2 was done. Treatment outcomes were evaluated between ARID1A-mutant and wild-type groups.
Results:
Among 258 patients, 67 (26.0%) harbored at least one ARID1A mutation. ARID1A mutant tumors (MT) showed significantly higher tumor mutational burden (median 18.0 vs 6.7 mut/Mb), EBV positivity (11.8% vs 0.6%), MSI-H status (10.6% vs 6.3%), and PD-L1 CPS ⩾50 (14.9% vs 1.0%) compared to wild-type (WT). The median progression-free survival (PFS) was numerically longer in the ARID1A MT group (9.7 vs 8.5 months; HR 0.75; 95% CI, 0.53–1.10), though not statistically significant. There was no significant difference for the ORR (55.2% vs 64.4%, p = 0.19), and the overall survival (27.4 vs 31.1 months; HR 0.83; 95% CI, 0.58–1.20) between the two groups.
Conclusion:
ARID1A mutations were associated with immune-active molecular features and a trend toward improved PFS to ICI plus chemotherapy. Further research for ARID1A as a potential biomarker for ICI is warranted in advanced GC.
Plain language summary
Gastric cancer is one of the leading causes of cancer-related death worldwide, particularly in East Asia. For patients with advanced HER2-negative gastric cancer, the current standard first-line treatment is a combination of chemotherapy and immune checkpoint inhibitors (ICIs). However, not all patients benefit equally, and identifying reliable predictive biomarkers remains an important clinical challenge. ARID1A is a tumor suppressor gene frequently mutated in gastric cancer. Previous studies have suggested that ARID1A mutations may be linked to an immune-active tumor microenvironment, characterized by features such as higher tumor mutational burden (TMB), increased PD-L1 expression, and association with Epstein–Barr virus (EBV) positivity and microsatellite instability-high (MSI-H) status — features known to correlate with better responses to immunotherapy. In this retrospective study, we analyzed 258 patients with advanced HER2-negative gastric cancer who received oxaliplatin-based chemotherapy plus ICIs as first-line treatment at a single tertiary center in Korea between 2022 and 2024. Among them, 67 patients (26.0%) had at least one ARID1A mutation. We compared molecular features and treatment outcomes between patients with and without ARID1A mutations. Tumors with ARID1A mutations showed significantly higher TMB, EBV positivity, MSI-H status, and PD-L1 CPS ⩾50 compared with wild-type tumors. The median progression-free survival was numerically longer in the ARID1A-mutant group (9.7 vs. 8.5 months), though not statistically significant. Response rate and overall survival were similar between the two groups. These findings suggest that ARID1A mutations are associated with immune-active molecular features in advanced gastric cancer and may represent a potential biomarker for selecting patients who could benefit from chemo-immunotherapy. Larger prospective studies are needed to confirm its predictive value.
Introduction
Gastric cancer (GC) remains a significant global health concern, consistently ranking among the leading causes of cancer mortality worldwide, especially in East Asian countries. 1 Despite advances in diagnosis and traditional treatment strategies, including surgery, chemotherapy, and radiotherapy, the prognosis for advanced GC patients often remains poor, characterized by high recurrence rates and limited therapeutic effects.
The combination of cytotoxic chemotherapy with immune checkpoint inhibitors (ICIs) has become a standard first-line treatment across various tumor types, including GC, as demonstrated by multiple phase III trials.2,3 However, ongoing research continues to explore predictive biomarkers that can help identify patients who are most likely to benefit from ICI-based therapies. Among these, programmed death-ligand 1 (PD-L1) expression remains the most widely used biomarker, with higher combined positive score (CPS) values generally correlating with improved efficacy across multiple malignancies. 4 However, beyond PD-L1 expression, validated predictive biomarkers for response to ICIs remain limited. Consequently, ongoing research is focused on identifying alternative biomarkers that may enable the selection of patients with PD-L1–low or –negative GC who could still derive clinical benefit from immunotherapy—such as Epstein–Barr virus (EBV) positivity, tumor mutational burden (TMB), and microsatellite instability (MSI) status.5–8
ARID1A (AT-rich interaction domain 1A), a key subunit of the SWI/SNF chromatin remodeling complex, functions as a tumor suppressor gene by regulating gene expression, DNA repair, and processes such as cell proliferation, differentiation, and apoptosis. Mutations in ARID1A, predominantly nonsense and frameshift alterations, typically lead to a loss of function or complete absence of the ARID1A protein, which can be detected via immunohistochemistry. 9 In GC, ARID1A mutations are frequently observed, with prevalence rates ranging from 8% to 27% of cases, making it the second most frequently mutated tumor suppressor gene after TP53 in this malignancy. 10 ARID1A mutation is also emerging as a significant biomarker for predicting favorable responsiveness to ICIs. This is because ARID1A-mutated GC often exhibits an elevated immune activity within the tumor microenvironment, characterized by features such as a higher TMB, an increased proportion of MSI cancers, elevated PD-L1 expression, and a greater presence of tumor-infiltrating lymphocytes (TILs). Studies consistently show that patients with ARID1A alterations experience significantly longer PFS and OS after anti-PD-1/PD-L1 immunotherapy, often independently of MSI or TMB status.11–14
This study aimed to assess the efficacy of the ICI plus chemotherapy as first-line therapy in advanced GC and explore its association with key molecular features according to the status of ARID1A-mutation.
Materials and methods
Patients
We analyzed 258 patients with advanced HER2-negative gastric or gastroesophageal junction adenocarcinoma who received oxaliplatin-based doublet chemotherapy (XELOX or FOLFOX) plus ICIs as first-line therapy at Samsung Medical Center, Seoul, Korea, between January 2022 and December 2024. These regimens are the only approved first-line ICI-chemotherapy combinations for advanced HER2-negative gastric cancer in Korea. Patients were consecutively enrolled if they had histologically confirmed disease and an available ARID1A mutation status assessed by next-generation sequencing (NGS) from tumor tissue or blood. Patients with HER2-positive disease, prior systemic chemotherapy for advanced disease, or a concurrent other primary malignancy were excluded.
We reviewed electronic medical records and extracted data on age, sex, Eastern Cooperative Oncology Group (ECOG) performance status, pathology, EBV status, PD-L1 expression, claudin 18.2 expression, MSI, TMB, and the presence and location of distant metastases at treatment initiation.
This study was approved by the Institutional Review Board (IRB) of Samsung Medical Center with the requirement for informed consent waived (IRB no. 2025-05-087) and was prepared in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies.
EBV, PD-L1, Claudin 18.2, and next-generation sequencing
EBV status was assessed using in situ hybridization. PD-L1 expression was evaluated using either the 22C3 pharmDx assay (Agilent Technologies, Santa Clara, CA, USA) or the 28-8 pharmDx assay (Agilent Technologies, Santa Clara, CA, USA) and reported as CPS. PD-L1 was considered negative if CPS was less than 5, and positive if CPS was 5 or higher. Claudin 18.2 expression was evaluated by immunohistochemistry, and tumors with ⩾75% of tumor cells showing membranous staining were classified as claudin 18.2–positive. The status of MSI, TMB, and ARID1A mutation was identified using NGS performed with the TruSight Oncology 500 assay (Illumina, San Diego, CA, USA) and the Oncomine Focus assay (ThermoFisher Scientific, Waltham, MA, USA).
Outcomes
Clinical outcomes were evaluated for objective response rate (ORR), disease control rate (DCR), progression-free survival (PFS), and overall survival (OS). Tumor response was evaluated as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD), according to the Response Evaluation Criteria in Solid Tumor, version 1.1. ORR was defined as the percentage (%) of patients with confirmed CR or PR. PFS was measured from the start of the treatment to the date of disease progression or death from any cause using RECIST 1.1. OS was calculated from the start of the treatment to the date of death from any cause.
Statistical analysis
Patients were categorized into the ARID1A mutant (MT) group if they harbored at least one ARID1A mutation, and into the wild-type (WT) group if no ARID1A mutation was detected. Categorical and continuous variables were summarized using descriptive statistics. The association between the status of ARID1A mutation and clinicopathologic or molecular features was assessed using the Chi-square test for independence. For each comparison, the Chi-square statistic (χ2), degrees of freedom (df), and corresponding two-sided p-values were reported. A p-value <0.05 was considered statistically significant.
Survival analyses were performed using the Kaplan–Meier method, and differences between groups were compared using log-rank tests. Univariable Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). All statistical analyses were performed using IBM SPSS Statistics version 27 (IBM Corp., Armonk, NY, USA) and R version 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). The data cutoff date was February 7, 2025.
Results
Patients characteristics
Baseline characteristics of 258 patients in this study are summarized in Table 1. The median age was 58.8 years (range 47.6–70.0), and 64.3% of patients were male. Most patients (84.9%) had an ECOG performance status of 1, and 70.5% had a history of previous gastrectomy. Poorly differentiated adenocarcinoma was the most common histologic subtype (46.5%), followed by moderately differentiated (29.5%) and signet ring cell type (9.3%). EBV positivity was observed in 7 patients (2.7%), and 132 patients (51.2%) were PD-L1 positive with a CPS ⩾5. 19 patients (7.4%) were classified as MSI-H. The median TMB across the cohort was 9.6 mutations per megabase. The most frequent sites of distant metastasis were the peritoneum (64.0%), lymph nodes (33.3%), and liver (19.8%). All patients received ICI plus chemotherapy; 81.8% with nivolumab and 18.2% with pembrolizumab.
Baseline characteristics.
EBV, Epstein–Barr Virus; ECOG, Eastern Cooperative Oncology Group; MSI, microsatellite instability; TMB, tumor mutational burden.
Among the 258 patients, 67 harbored at least one ARID1A mutation (ARID1A MT), while the remaining 191 were classified as wild-type (ARID1A WT) (Table 1). EBV positivity was observed in 6 patients (9.0%) in the ARID1A MT group, compared to only 1 patient (0.5%) in the ARID1A WT group (p = 0.001). PD-L1 expression also differed significantly between the two groups: 61.2% of patients in the ARID1A MT group were PD-L1 positive (defined as CPS ⩾5), compared to 47.6% in the WT group. Notably, high PD-L1 expression (CPS ⩾50) was more frequently observed in the MT group (14.9% vs 1.0%; p < 0.001). TMB was also significantly higher in the MT group, with a median of 18.0 mutations/Mb compared to 6.7 mutations/Mb in the WT group (p < 0.001). Regarding metastatic patterns, liver metastases were more frequent and bone metastases less frequent in the ARID1A WT group than in the MT group, both with statistically significant differences. Other baseline characteristics were well balanced between the two groups.
Spectrum and characteristics of ARID1A mutations
Among the 67 patients with ARID1A mutations, the most common type was frameshift mutations (n = 44), followed by nonsense (n = 19), missense (n = 17), in-frame insertions or deletions (n = 2), and a single in-frame deletion (Table 2). Supplementary analysis showed a wide range of mutation sites, with some common alterations, such as p.Asp1850GlyfsTer4 (n = 3), p.Asp1850fs (n = 2), and p.Gln766SerfsTer67 (n = 2) (Supplemental Table 1). A few mutations were classified as known oncogenic based on OncoKB COSMIC data, including p.Asp1850GlyfsTer4, p.Gln766SerfsTer67, p.Ser682LeufsTer59, p.Gln1519ArgfsTer8, p.Gln268Ter, and p.Arg2158Ter.
Summary of ARID1A mutation types.
Association of ARID1A mutation with other biomarkers
Table 3 summarizes the co-occurrence of other molecular biomarkers with ARID1A mutations. Patients with ARID1A mutations demonstrated a significantly higher frequency of EBV positivity compared to the wild-type group (11.8% vs 0.6%, p = 0.0005), as well as a numerically higher proportion of MSI-H status (10.6% vs 6.3%, p = 0.0005). PD-L1 positivity, defined as CPS ⩾5, was observed more frequently in the ARID1A-mutant group than in the wild-type group (65.1% vs 50.3%), with a trend toward statistical significance (p = 0.0580). CLDN18.2 expression was also more common among ARID1A-mutant cases (47.1% vs 36.5%), although this difference was not statistically significant (p = 0.4432).
Co-occurrence of ARID1A mutation with EBV, MSI-H, PD-L1, and CLDN18.2.
EBV, Epsteine-Barr Virus; MSI, microsatellite instability.
Treatment outcomes according to the status of the ARID1A mutation
ORRs were assessed according to the status of ARID1A mutation (Table 4). The ORR was 55.2% in the ARID1A mutant (MT) group and 64.4% in the WT group (p = 0.19). The DCR, defined as the proportion of patients achieving CR, partial response, or SD, was 89.6% in the MT group and 94.2% in the WT group (p = 0.40). While both ORR and DCR tended to be lower in the MT group, these differences did not reach statistical significance in this cohort.
Objective response rates by ARID1A mutations.
The median follow-up duration for the overall cohort was 17.5 months (95% CI, 16.6–18.9). The median overall PFS was 9.7 months (95% CI, 8.0–11.1). When stratified by the status of ARID1A mutation, the median PFS was 8.5 months (95% CI, 7.2–10.5) in the ARID1A WT and 9.7 months in the MT group, with a HR of 0.75 (95% CI, 0.53–1.10; p = 0.1), suggesting a non-significant trend toward improved PFS in the MT group (Figure 1). For OS, patients in the ARID1A WT had a median OS of 31.1 months (95% CI, 22.8–NA), whereas those in the ARID1A MT group had a median OS of 27.4 months (95% CI, 24.0–NA). The corresponding HR was 0.83 (95% CI, 0.58–1.20; p = 0.19), indicating no statistically significant difference between the two groups (Figure 2).
Kaplan–Meier curve of progression-free survival by ARID1A mutation.
Kaplan–Meier curve of overall survival by ARID1A mutation.
Discussion
This analysis presented that ARID1A MT tumors had higher PD-L1 positivity (CPS ⩾5: 65.1% vs 50.3%; CPS ⩾50: 14.9% vs 1.0%; p < 0.001), EBV positivity (11.8% vs 0.6%; p = 0.0005), and MSI-H status (10.6% vs 6.3%; p = 0.0005) as compared to ARID1A WT tumors. TMB was also significantly higher in the MT group, with a median of 18.0 mutations/Mb compared to 6.7 mutations/Mb in the WT group (p < 0.001).
These findings suggested that ARID1A MT tumors were associated with immune-active features, including higher TMB, EBV positivity, MSI-H status, and PD-L1 expression, which together may contribute to a more immunoresponsive phenotype. In the Cancer Genome Atlas classification, ARID1A loss is a hallmark of both EBV-associated and MSI-H tumors, which typically exhibit elevated PD-L1 expression on tumor and immune cells. Independent studies have also shown a tight correlation between ARID1A inactivation and PD-L1 overexpression in gastric cancer. 10 These findings indicate that ARID1A-mutant tumors often share key features with other immunotherapy-responsive subtypes, such as MSI-H or EBV-positive tumors, and may already be identified through existing biomarker testing. These findings support that ARID1A status could be used to complement existing biomarkers in selecting patients for immunotherapy.
Among 258 patients, 67 patients (26.0%) harbored ARID1A mutations. The ARID1A MT group showed a numerically longer median PFS to ICI plus chemotherapy as compared to the WT group (the median PFS 9.7 vs 8.5 months, HR 0.75; 95% CI 0.53–1.10; p = 0.1). Although there was no statistical difference, patients with ARID1A-mutant gastric cancer demonstrated numerically improved PFS compared to wild-type.
This finding is consistent with previous reports linking ARID1A loss with enhanced immunotherapy sensitivity. Okamura et al. reported a median PFS of 11 months in ARID1A-altered tumors versus 4 months in wild-type across various cancers treated with PD-1/PD-L1 inhibitors (p = 0.006). 14 This trend is biologically plausible: ARID1A inactivation promotes an inflamed tumor microenvironment by impairing mismatch repair through MSH2 interaction, resulting in higher TMB and neoantigen burden. In the present analysis, ARID1A-mutant tumors also showed higher TMB, further supporting this mechanism. Preclinical data reinforce this immune-activating effect; in ovarian models, ARID1A loss was associated with increased PD-L1 expression, TIL infiltration, and improved response to PD-L1 blockade. 14 Clinically, bioinformatics studies such as those by Li et al. demonstrated ARID1A-mutant GI tumors harbor more CD8+ T cells, NK cells, and upregulated immune-cytolytic signatures, 11 along with lower tumor aneuploidy, all features favorable for checkpoint blockade.
Recent large-scale clinical datasets supported this predictive value. Jiang et al. showed that ARID1A mutations, while associated with poor baseline prognosis, predicted improved survival in ICI-treated patients across cancer types. 13 Focusing on gastric cancer, prior studies have identified ARID1A-mutant tumors as an immunogenic subtype frequently enriched in EBV-positive and MSI-high molecular profiles. These tumors demonstrated favorable responses to both 5-FU–based chemotherapy and PD-1 blockade. A mutation-based scoring system incorporating ARID1A status was able to predict pembrolizumab responders with high sensitivity, reportedly up to 91%. 15 Mechanistically, ARID1A-mutant tumors had marked DNA repair defects and Th17 cell enrichment, which may augment chemotherapy efficacy through IL-17–mediated immune recruitment. Collectively, these findings suggest that ARID1A deficiency not only enhances ICI responsiveness but may also sensitize tumors to cytotoxic agents. Since all patients in our cohort received oxaliplatin-based chemotherapy (XELOX or FOLFOX) combined with ICIs, the relative contributions of enhanced chemotherapy sensitivity versus immunotherapy responsiveness to the observed outcomes cannot be separated in the current analysis. Our results, although limited by sample size, align with these observations and support the use of chemoimmunotherapy as a rational frontline strategy in ARID1A-mutant AGC.
ICIs are less effective in patients with liver metastases. 16 In our cohort, liver metastases were significantly more frequent in the ARID1A WT group than in the MT group (23.0% vs 10.4%, p = 0.041). In addition, ARID1A WT tumors showed lower PD-L1 expression, lower EBV positivity, and reduced TMB compared with MT tumors, indicating fewer favorable immunologic markers. Overall, these results suggest that ARID1A WT gastric cancers combine adverse clinical features with a less immunoreactive profile, which may explain their limited response to immunotherapy.
This study has several limitations inherent to its retrospective and single-center design. Selection bias may have influenced the results, as there were differences in baseline characteristics between the two groups. In addition, the limited sample size of the ARID1A-mutant group (n = 67) reduces the statistical power to detect significant differences in survival outcomes. The heterogeneity in ICI agents used (nivolumab vs pembrolizumab) and in baseline molecular features such as PD-L1, EBV, and MSI status could also confound the observed treatment effects. Moreover, some biomarker data were missing or derived from different assays, potentially affecting the consistency and interpretation of molecular correlations. Furthermore, the independent predictive value of ARID1A mutation, beyond its association with TMB and PD-L1, could not be assessed in this cohort—in particular, a subgroup analysis in PD-L1-negative patients (CPS <5) was not feasible due to the small number of ARID1A-mutant patients in this subgroup (n = 22), which warrants further investigation.
In conclusion, ARID1A mutations were associated with immune-active molecular features and a trend toward improved PFS to ICI plus chemotherapy. Further research for ARID1A as a potential biomarker for ICI is warranted in advanced GC.
Supplemental Material
sj-docx-1-tag-10.1177_17562848261446862 – Supplemental material for Efficacy of immune checkpoint inhibitor plus chemotherapy in ARID1A-mutated advanced gastric cancer
Supplemental material, sj-docx-1-tag-10.1177_17562848261446862 for Efficacy of immune checkpoint inhibitor plus chemotherapy in ARID1A-mutated advanced gastric cancer by Ji Eun Shin, Eunbyeol Lee, Sung Hee Lim, Jeeyun Lee and Seung Tae Kim in Therapeutic Advances in Gastroenterology
Supplemental Material
sj-docx-2-tag-10.1177_17562848261446862 – Supplemental material for Efficacy of immune checkpoint inhibitor plus chemotherapy in ARID1A-mutated advanced gastric cancer
Supplemental material, sj-docx-2-tag-10.1177_17562848261446862 for Efficacy of immune checkpoint inhibitor plus chemotherapy in ARID1A-mutated advanced gastric cancer by Ji Eun Shin, Eunbyeol Lee, Sung Hee Lim, Jeeyun Lee and Seung Tae Kim in Therapeutic Advances in Gastroenterology
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
