Shifting paradigm in locally advanced resectable gastric and gastroesophageal junction cancers

Abstract

Gastric cancer (GC) is the fifth most common cancer and the fifth leading cause of cancer-related mortality worldwide. The management of resectable locally advanced GC evolved with the introduction of adjuvant chemoradiotherapy in some regions, notably following the INT-0116 trial. A subsequent major advance was perioperative chemotherapy with epirubicin, cisplatin, and fluorouracil, which significantly improved 5-year overall survival compared to surgery alone. More recently, the fluorouracil, leucovorin, oxaliplatin, and docetaxel (FLOT) regimen demonstrated superior outcomes compared to epirubicin, cisplatin, and fluorouracil. Despite this advancement, nearly half of all patients (46%) experience disease recurrence within three years, underscoring a significant unmet need. In a recent real-world study by Wang et al, which assessed perioperative sintilimab plus oxaliplatin and S-1 chemotherapy vs chemotherapy alone in non-metastatic GC, the authors reported significantly improved pathological response rates and overall survival with the combination. Additionally, the safety profile showed a lower frequency of high-grade adverse events. However, this study has limitations, including its retrospective design and the use of a chemotherapy backbone (oxaliplatin and S-1) considered less effective than FLOT based on phase III evidence. Recent data from the phase III MATTERHORN trial support the addition of durvalumab to FLOT, showing significant improvements in pathological complete response and event-free survival. Based on the cumulative evidence, adding immunotherapy to perioperative chemotherapy improves outcomes for patients with resected GC and may constitute a new standard of care once confirmatory data mature and regulatory approvals are granted.

Key Words: Gastric cancer; Locally advanced; Perioperative therapy; Chemotherapy; Immunotherapy

Core Tip: The recent real-world study by Wang et al adds to the growing evidence that perioperative sintilimab combined with oxaliplatin and S-1 chemotherapy improves pathological response and overall survival in non-metastatic gastric cancer, with a manageable safety profile. However, the conclusions are tempered by the study’s retrospective design, limited power, and the use of a potentially suboptimal chemotherapy backbone. Crucially, this data aligns with the positive results of the large phase III MATTERHORN trial, solidifying the paradigm shift. The collective evidence now strongly suggests that adding immunotherapy to perioperative chemotherapy enhances cure rates for locally advanced resectable gastric and gastroesophageal junction cancers, establishing this combination as a new standard of care.

TO THE EDITOR

Gastric cancer (GC) is the fifth most common cancer and the fifth leading cause of cancer-related mortality worldwide[1]. In 2022, an estimated 968350 new cases and 659853 deaths from GC occurred globally[1]. The management of resectable locally advanced GC was transformed with the establishment of multimodal therapy. In North America, the INT-0116 trial demonstrated in 2001 that adjuvant chemoradiotherapy based on 5-fluorouracil reduced the risk of death by 35%[2,3]. In Europe, the MAGIC trial in 2006 established that perioperative chemotherapy with epirubicin, cisplatin, and fluorouracil resulted in a significant improvement in 5-year overall survival (OS) compared to surgery alone (36% vs 23%)[4]. Subsequently, the German FLOT4 trial demonstrated the superior efficacy of the fluorouracil, leucovorin, oxaliplatin, and docetaxel (FLOT) regimen over epirubicin, cisplatin, and fluorouracil, with a median OS of 50 months vs 35 months[5]. Despite these advancements, long-term outcomes from the FLOT4 trial reveal that 46% of patients experience disease recurrence within 3 years, highlighting a significant unmet need for more effective therapeutic strategies[5].

Approximately 60%-80% of patients with metastatic GC have programmed death-ligand 1 (PD-L1)-positive tumors. In the first-line setting for PD-L1 combined positive score-positive disease, the addition of immunotherapy (nivolumab, pembrolizumab, or sintilimab) to chemotherapy has significantly improved OS compared to chemotherapy alone[6-11]. However, the role of immunotherapy in early-stage, resectable GC remained uncertain.

In their interesting work, Wang et al[12] assessed the efficacy and safety of perioperative sintilimab plus oxaliplatin and S-1 (SOX) chemotherapy in patients with non-metastatic GC. Their study explores a new paradigm, the integration of immunotherapy into the perioperative strategy for resectable GC. The authors suggest that the pathological response rate and survival outcomes [event-free survival (EFS) and OS] associated with perioperative immunotherapy plus chemotherapy were superior to those observed with standard perioperative chemotherapy. Furthermore, according to the primary interim analysis of the MATTERHORN trial, recently presented at the 2025 American Society of Clinical Oncology annual meeting, the addition of durvalumab to FLOT (D-FLOT) significantly improved pathologic complete response (pCR) and EFS in stage II-IVa resectable gastric and gastroesophageal junction (G/GEJ) adenocarcinomas compared to FLOT alone[13].

Despite these encouraging results, several critical questions regarding the optimal strategy for these patients remain unresolved. These include clarifying which patients truly need perioperative treatment, determining whether perioperative immunotherapy can increase the cure rate in stage II/III GC, establishing the ongoing necessity of chemotherapy for patients with deficient mismatch repair (dMMR) GC in the immunotherapy era, and defining the need for surgery in dMMR patients who achieve a complete response. In this letter, we address these uncertainties and discuss the current evidence for perioperative immunotherapy.

The potential of perioperative immunotherapy to increase cure rates in resectable G/GEJ cancers

The use of neoadjuvant immunotherapy has a strong biological and clinical rationale in the non-metastatic setting, as extensively investigated in non-small cell lung cancer and melanoma[14,15]. Neoadjuvant immunotherapy can induce a more robust T-cell immune response in the tumor and lymph nodes compared to the adjuvant setting. For instance, the NADINA study in melanoma showed that two cycles of neoadjuvant nivolumab-ipilimumab were superior to one year of adjuvant therapy[15]. Similarly, in colorectal cancer, the NICHE-2 study reported a major pathologic response in 95% of dMMR cases (68% complete response, 27% near-complete response), with an estimated 3-year disease-free survival of 100%, despite 65% of tumors being T4-stage[16].

Wang et al[12] conducted a retrospective, single-center comparative analysis in Qinghai, China, including 299 patients with resectable stage II/IIIC GC[12]. Patients were divided into three cohorts: P-SOX (nab-paclitaxel, SOX) (n = 100), SOX (n = 118), and SOX plus sintilimab (n = 81), receiving 2-4 cycles of systemic therapy before and after surgery (6-8 cycles total) followed by D2 R0 gastrectomy. Notably, the chemotherapy-only cohorts had more poor prognostic factors (e.g., node-positive disease, stage IIIC). The SOX plus sintilimab group showed a higher objective response rate of 70.4% compared to 52.9% (SOX) and 59.3% (P-SOX), although this difference was not statistically significant. Median OS was significantly higher in the sintilimab cohort (32 months vs 29 months for SOX and 27 months for P-SOX; P = 0.007). Median EFS was also numerically higher (30 months vs 22 months and 25 months, respectively) but did not reach statistical significance (P = 0.054).

However, this study has several limitations, including its retrospective design, which introduces potential selection bias. Furthermore, the sample size (n = 299) is limited compared to recent phase III trials, and the chemotherapy backbone (SOX) is considered suboptimal compared to FLOT, which has demonstrated superior efficacy in a phase III setting and is a globally established standard[4]. Finally, the median follow-up duration was not reported.

Two major recent phase III trials have demonstrated that perioperative immunotherapy improves pCR and EFS (Table 1). The MATTERHORN trial randomized 948 patients with stage II-IVa G/GEJ cancer to receive D-FLOT or FLOT plus placebo[13]. Tumor characteristics included 90% PD-L1 positive, 90% T3-T4b, and 5% dMMR. MATTERHORN met its primary endpoint, showing a significant improvement in 24-month EFS from 59% with FLOT to 67% with D-FLOT (P < 0.001). OS data were immature but showed a promising trend (24-month OS: 76% vs 70%; P = 0.024). The pCR rate nearly tripled from 7% to 19%. The most common toxicities during neoadjuvant therapy (diarrhea, nausea, neutropenia, anemia) were similar between arms. These results support D-FLOT as a potential new standard of care. The KEYNOTE-585 phase III trial also evaluated a perioperative immunotherapy strategy, combining pembrolizumab with a cisplatin-based chemotherapy doublet. While it significantly improved the pCR rate compared to chemotherapy alone (12.9% vs 2.0%), the improvement in EFS did not meet its pre-specified significance threshold despite a favorable hazard ratio. The safety profile was manageable and similar between arms[17].

Table 1 Key trials investigating the role of immunotherapy in the perioperative setting for gastric and gastroesophageal junction cancers.

Ref.	Trial (design), participants (n)	Intervention arms vs comparator	Efficacy	Adverse events (grade 3 and 4)
[13]	MATTERHORN (phase III), 948	Durvalumab + FLOT (2 cycles), surgery, durvalumab + FLOT (2 cycles), durvalumab (10 cycles)	pCR: 19% vs 7%; 2-years EFS: 67% vs 59% (HR = 0.73; 95%CI: 0.60-0.89; P < 0.001); 2-year OS: 76% vs 70% (P = 0.024)	Grade 3-4 AEs during neoadjuvant phase: Similar between arms. Most common: Neutropenia, diarrhea, nausea, anemia
		Comparator: Placebo + FLOT (2 cycles), surgery, placebo + FLOT (2 cycles), placebo (10 cycles)
[17]	KEYNOTE-585 (phase III), 804	Pembrolizumab + CT: Pembrolizumab + cisplatin-based doublet (3 cycles), surgery, pembrolizumab + doublet (3 cycles), pembrolizumab (11 cycles)	pCR: 12.9% vs 2.0% (P < 0.0001); median EFS: 44.4 months vs 25.3 months (HR = 0.81; 95%CI: 0.67-0.99; P = 0.0178, NS); median OS: 60.7 months vs 58.0 months (P = 0.174, NS)	Grade 3-4 AEs: 78% (pembrolizumab + CT) vs 74% (CT). Most common: Neutropenia, anemia, diarrhea, nausea, vomiting, decreased appetite
		Comparator: Placebo + cisplatin-based doublet (3 cycles), surgery, placebo + doublet (3 cycles), placebo (11 cycles)
[12]	Retrospective, 299	Sintilimab + SOX: Sintilimab + SOX (2-4 cycles), surgery, sintilimab + SOX (2-4 cycles)	ORR: 70.4% vs 52.9% vs 59.3%; median EFS: 30 months vs 22 months vs 25 months (P = 0.054); median OS: 32 months vs 29 months vs 27 months (P = 0.007)	Grade 3-4 AEs: Overall incidence: 19.8% (sintilimab + SOX) vs 16.1% (SOX) vs 20.0% (P-SOX)
		SOX: SOX (2-4 cycles), surgery, SOX (2-4 cycles)
		P-SOX: P-SOX (2-4 cycles), surgery, P-SOX (2-4 cycles)

FLOT: 5-fluorouracil, leucovorin, oxaliplatin, docetaxel; pCR: Pathologic complete response; EFS: Event-free survival; HR: Hazard ratio; CI: Confidence interval; OS: Overall survival; NS: Not statistically significant; ORR: Objective response rate; AEs: Adverse events; CT: Chemotherapy; SOX: Oxaliplatin and S-1; P-SOX: Nab-paclitaxel, oxaliplatin, S-1.

Cross-trial comparisons and limitations

While the data from Wang et al[12] (sintilimab/SOX), MATTERHORN (durvalumab/FLOT), and KEYNOTE-585 (pembrolizumab/cisplatin-doublet) are informative, direct comparisons must be made with caution. These trials differed significantly in their patient populations, chemotherapy backbones (SOX vs FLOT vs cisplatin-based regimens), primary endpoints, and statistical power. The positive results of MATTERHORN, using the highly effective FLOT backbone, provide the strongest evidence to date. The negative EFS result in KEYNOTE-585 may be attributed to the use of a less immunogenic, cisplatin-based chemotherapy backbone compared to FLOT or SOX regimens used in MATTERHORN and Wang et al’s study[12], respectively[13,17]. These cross-trial limitations preclude definitive conclusions about the superiority of one specific immunotherapy-chemotherapy combination over another.

Treating patients with dMMR GC

The high pCR rates observed with immunotherapy in dMMR colorectal cancer (68% in NICHE-2) and locally advanced rectal cancer (74% in the Cercek cohort) have prompted the exploration of non-operative management for microsatellite instability-high (MSI-H) cancers[18]. A recent small study of dostarlimab in dMMR GC (n = 16) reported a pCR rate of 57%, suggesting the potential for treating these patients with immunotherapy alone[16,18]. However, it is crucial to state that no large-scale data for this approach currently exist in GC. Outside of clinical trials, surgery remains the standard of care for resectable dMMR GC, and non-operative management should be considered highly investigational.

Biomarkers and future directions

The manuscript correctly identifies the urgent need for reliable predictive biomarkers. Beyond static markers like PD-L1 and MSI status, dynamic monitoring of minimal residual disease (MRD) using circulating tumor DNA (ctDNA) is emerging as a powerful tool to guide adjuvant therapy decisions. Data from multiple observational studies in G/GEJ cancers consistently show that the detection of ctDNA after complete resection (R0) is a highly significant prognostic factor, associated with a markedly increased risk of recurrence (hazard ratios ranging from 2.0 to 12.5 across studies)[19].

This paradigm opens the door for MRD-guided clinical trials, which aim to escalate treatment for high-risk, ctDNA-positive patients and de-escalate for low-risk, ctDNA-negative patients. Promisingly, in the MSI-H population, a proof-of-concept trial demonstrated that adjuvant pembrolizumab could clear ctDNA in a significant proportion of MRD-positive patients, translating to excellent recurrence-free survival[20]. Furthermore, several ongoing trials are now exploring the efficacy of novel agents, such as trastuzumab deruxtecan in human epidermal growth factor receptor 2-positive disease, specifically in the MRD-positive setting after surgery[21,22].

The high pCR rates observed with neoadjuvant immunotherapy-based combinations in dMMR/MSI-H G/GEJ cancers (ranging from 28% to 63% in clinical trials, and up to 60%-80% major pathological response) further underscore the potential of response-adapted strategies[13,17,23-25]. The critical question of whether chemotherapy can be omitted in this population when using potent immunotherapy doublets is being actively investigated. The future management of locally advanced G/GEJ cancers will increasingly rely on integrated biomarker platforms, combining ctDNA dynamics with comprehensive tissue-based profiling (e.g., specific T-cell infiltrate subsets, interferon-γ gene signatures) to truly personalize perioperative therapy, maximize efficacy, and minimize unnecessary toxicity.

Conclusions

Evidence from Wang et al[12] and the MATTERHORN trial indicates that perioperative immunotherapy combined with chemotherapy is associated with improved survival and pathological response rates for patients with locally advanced resectable G/GEJ cancers, providing an 8% absolute improvement in 2-year EFS. However, key questions remain about whether immunotherapy must be given to all patients in the perioperative setting, or whether the same outcome can be achieved while reducing side effects and financial toxicity with shorter neoadjuvant immunotherapy. Patients achieving a pCR may not require adjuvant chemotherapy and immunotherapy. For MSI-H patients, a non-operative management strategy should be discussed in a multidisciplinary team in the case of clinical complete response. Moving forward, the development and clinical implementation of predictive biomarkers and MRD-guided strategies will be essential to truly personalize therapeutic approaches, maximize efficacy, and minimize unnecessary treatment toxicity.