GC is a heterogenous disease. Over the past decade, an in-depth understanding of the molecular characteristics of GC has led to the advent of novel diagnostic and prognostic biomarkers as well as the development of targeted therapies. With precision medicine and an increasing ability to predict responders to various therapies, neoadjuvant and adjuvant therapies can be better utilized in GC patients.

In 2014, The Cancer Genome Atlas (TCGA) outlined a molecular classification system separating GC into four subtypes: tumors with high microsatellite instability (MSI-H), tumors that are Epstein–Barr virus positive (EBV+), genomically stable (GS) tumors, and tumors with chromosomal instability (CIN) [31]. These subtypes each have distinct genomic features, which may serve as targets for therapeutic agents. In 2015, the Asian Cancer Research Group published their own molecular classification of GC with associated prognostic differences: tumors with microsatellite instability (MSI), mesenchymal tumors (microsatellite-stable tumors with epithelial-to-mesenchymal transition phenotype (MSS/EMT)), and non-EMT MSS tumors that are p53+ and p53− [32]. Patients with MSS/EMT-type GC were younger and had the worst prognosis compared to patients with the other subtypes. In contrast, patients with MSI-H tumors had the best prognosis; the MSS/p53+ genotype was closely linked to EBV+ status, and patients with MSS/p53− tumors had a better prognosis than those with MSS/p53+ tumors.

MSI-H tumors have a high tumor mutational burden and are deficient in one or more mismatch repair proteins, i.e., mismatch repair deficient (dMMR). dMMR is a surrogate for the MSI-H phenotype. Polymerase chain reaction can be used to detect amplified microsatellite loci and is more accurate than IHC, though more expensive [33]. MSI-H tumors are associated with older age and tend to occur in the distal stomach [34]. About 10–20% of GCs are MSI-H [35,36]. MSI-H tumors tend to have fewer lymph node metastases and better OS [37,38]. In addition, MSI-H GCs frequently express immune checkpoint molecules like PD-L1, making them possibly eligible for treatment with immune checkpoint inhibitors (ICIs) [39]. MSI-H GCs can be identified using immunohistochemistry (IHC) with antibodies directed at mismatch repair proteins: MLH1, PMS2, MSH2, and MSH6 [38].

The EBV+ subtype of GC accounts for 10% of cases and frequently harbors PIK3CA and ARID1A mutations as well as amplification of JAK2, PD-L1, and PD-L2 and DNA hypermethylation [31]. Testing for EBV infection is best implemented by EBV encoding region (EBER) in situ hybridization, which would show a nuclear signal [40]. EBV-associated GC is more common in younger men and occurs primarily in the proximal stomach or postsurgical gastric stump [41]. The EBV+ subtype of GC generally has a more favorable prognosis and fewer lymph node metastases than non-EBV GC [42]. In addition, given that the EBV+ subtype of GC often is associated with modulation of the tumor microenvironment (TME) to evade the host immune response (such as by upregulation of PD-L1 expression by CD274 focal amplification and IFN-γ-mediated signaling, causing immunosuppression), immune checkpoint blockade arose as a promising avenue of treatment [43,44,45].

HER2 gene amplification occurs in 10–30% of GCs, which results in overexpression of HER2, a receptor tyrosine kinase [31]. HER2 status is mainly assessed through IHC and/or fluorescence in situ hybridization (FISH). Given its relatively low cost and ease of performance, IHC is typically used first to test for HER2 overexpression. A 3+ result is considered positive. If the results are equivocal (2+), FISH is performed [46]. HER2+ status is prognostic of poorer outcomes in GC, so identification is critical and may predict which patients will benefit from the addition of trastuzumab or other HER2-directed therapies to their chemotherapy regimen [26,47].

Finally, PD-L1 IHC is routinely performed to help predict potential benefit from immunotherapy. In addition, the EBV+ and MSI-H subtypes are associated with the overexpression of PD-LI in tumor, stromal, and immune cells [31]. The combined positive score (CPS) quantifies PD-LI expression in tumor and tumor-associated immune cells, where CPS ≥ 1 is considered positive [48]. A higher score correlates with an increased probability of clinical benefit from PD-L1 inhibition [49]. An increasing number of studies have been conducted to investigate the efficacy of immunotherapy for LAGC patients, which will be discussed later in this article.

As mentioned above, MSI-H LAGC is associated with an improved prognosis [16,31,50]. In the MAGIC trial, which established perioperative chemotherapy as the standard of care for non-Asian patients with LAGC, patients with MSI-H tumors had an improved prognosis compared to patients with microsatellite-stable (MSS) tumors in the surgery-only group (hazard ratio (HR) 0.35, 95% confidence interval (CI) 0.11–1.11, p = 0.08). Perioperative chemotherapy and surgery in the MSI-H subset of patients was associated with a worse OS compared to surgery alone (HR = 2.22, 95% CI 1.02–4.85, p = 0.04) [16,51]. In the CLASSIC trial, which found that adjuvant capecitabine and oxaliplatin after curative D2 gastrectomy is effective in Asian LAGC patients, patients with MSI-H tumors did not have a survival benefit when adjuvant chemotherapy was added to surgery (5-year disease-free survival 83.5% versus 85.7%, p = 0.931) [50,52]. A pooled meta-analysis of four neoadjuvant/adjuvant chemotherapy trials (MAGIC, CLASSIC, ARTIST, ITACA-S) found that only the non-MSI-H LAGC patients benefited from chemotherapy and surgery compared to surgery only, with a five-year OS of 62% versus 53% (HR = 0.75, 95% CI 0.60–0.94) [53]. Thus, neoadjuvant/adjuvant chemotherapy is likely ineffective for MSI-H LAGC.

However, there may be a role for neoadjuvant or adjuvant immunotherapy, especially in patients with MSI-H tumors. The TME plays a key role in GC progression through immune metabolic reprogramming and alteration of immune cells to evade host defenses [54]. The immune TME in GC contains T and B lymphocytes, natural killer (NK) cells, macrophages, and neutrophils [55]. These cells can promote or inhibit tumor immunity. GC exerts a variety of mechanisms to evade the host immune system. For instance, tumors can release anti-inflammatory cytokines such as IL-10 and TGF-β, which can cause the recruitment of regulatory T cells (Tregs) and inhibit the antitumor effect of lymphocytes [55,56]. Tregs can cause immune suppression through multiple mechanisms, including CTLA-4 expression to inhibit antigen-presenting cells [57]. In addition, GCs themselves can express PD-L1, which interacts with PD-1 on the cell surface of T cells, causing anergy and/or apoptosis of T cells [58]. With this knowledge, ICIs have become a promising area of therapy for many tumor types including GC.

Several studies show an improved response to ICIs when combined with chemotherapy in a variety of cancers, such as non-small-cell lung cancer, mesothelioma, and renal cell carcinoma [59,60,61]. This synergistic effect is thought to be due to chemotherapy activating an endogenous antitumor immune response, causing the enhancement of co-stimulatory molecules like CD80 and CD86 and the downregulation of PD-L1, and inducing immunogenic tumor cell death [62]. The current Food and Drug Administration (FDA)-approved ICIs target two key T-cell signaling pathways: PD-L1/PD-1 and CTLA-4. Although PD-L1 expression in GC has been associated with poorer outcomes, anti-PD-L1 antibodies (e.g., atezolizumab, avelumab, durvalumab) and anti-PD-1 antibodies (e.g., nivolumab, pembrolizumab) have been shown to improve OS in patients with advanced or metastatic GC [63,64]. CTLA-4 inhibitors such as tremelimumab and ipilimumab have also been explored as options for the treatment of GC. The KEYNOTE-062 phase III trial randomized patients to pembrolizumab alone, pembrolizumab and chemotherapy (5-FU and oxaliplatin), and chemotherapy alone in patients with advanced GC and PD-L1 CPS ≥ 1. That study found that pembrolizumab was as effective as chemotherapy with fewer treatment-related adverse effects [65]. Overall, pembrolizumab plus chemotherapy was not more effective than chemotherapy alone [65]. Post hoc analysis of the KEYNOTE-062 trial, along with the KEYNOTE-059 and KEYNOTE-061 trials, found a significant benefit for pembrolizumab in the subset of patients with MSI-H tumors [66]. One meta-analysis analyzed the RCTs that studied first-line advanced GC treatment with chemotherapy plus ICIs vs. chemotherapy alone [67]. That analysis found a significant reduction in the risk of death and progression in those treated with combination chemotherapy plus ICIs, across patients with PD-L1 CPS ≥ 10 and CPS ≥ 1, but a larger reduction in risk of death and progression was seen in patients with CPS ≥ 10.

Given the efficacy of PD-1/PD-L1 inhibition in MSI-H advanced GC, the question arose of whether similar effects would be seen in MSI-H LAGC. The GERCOR NEONIPIGA phase II trial evaluated the pathological complete response (pCR) rate after perioperative chemotherapy in GC/gastroesophageal junction adenocarcinoma (GEJC) with dMMR/MSI-H status [68]. Specifically, 32 patients with cT2-T4/NX/M0 resectable dMMR/MSI-H GC/GEJC underwent neoadjuvant nivolumab and ipilimumab for four cycles followed by surgery and adjuvant nivolumab for nine cycles [68]. Overall, 59% of tumors had complete regression with no residual tumor (TRG1a) and 14% had less than 10% residual tumor (TRG1b) [68,69]. At the most recent follow up of those patients who achieved Becker TRG 1a/1b, no patients had died or had recurrence. The authors concluded that neoadjuvant and adjuvant nivolumab and ipilimumab was safe and feasible in patients with MSI-H LAGC [68].

The phase II INFINITY trial enrolled patients with MSI-H cT2 or greater (any N, M0) GC/GEJC (n = 15) and treated these patients with tremelimumab (anti-CTLA-4) and durvalumab (anti-PD-L1) for one cycle followed by durvalumab every four weeks for two cycles followed by surgery [70]. Ultimately, 60% of patients achieved pCR, and 80% had <10% viable tumor. Notably, only one out of seven patients with T4 tumors achieved pCR [70]. In summary, neoadjuvant and adjuvant immunotherapy with anti-PD-L1 and anti-CTLA-4 agents has demonstrated promising tumor-eradicating activity in patients with dMMR/MSI-H LAGC.

Although immune checkpoint blockade has been shown to be effective in MSI-H LAGC, most LAGCs are MSS [36]. ICIs in combination with chemotherapy may be effective in non-MSI-H LAGC based on studies in advanced patients. The CheckMate-649 phase III trial found that, patients with advanced GC/GEJC treated with nivolumab, an anti-PD-1 antibody, obtained a survival benefit when combined with chemotherapy compared to chemotherapy alone in patients with tumors with PD-L1 CPS ≥ 1 and ≥5 [49]. This led to the FDA approval of nivolumab in April 2021 for use in advanced GC/GEJC regardless of PD-L1 CPS, though the National Comprehensive Cancer Network (NCCN) guidelines specify their category 1 designation for CPS ≥ 5 [71]. The CheckMate-577 trial showed that nivolumab can also prolong survival compared to placebo in the adjuvant setting in patients with resected stage II and III esophageal or GEJC who also received neoadjuvant chemoradiotherapy [72]. Recent data from key phase III RCTs performed in China, Rationale-305 and ORIENT-16, show an OS benefit in patients with PD-L1+ unresectable GC/GEJC treated with PD-L1 inhibitors in combination with chemotherapy compared to placebo and chemotherapy [73,74]. Rationale-305 found that the PD-L1+ group (defined as CPS ≥ 5) treated with tislelizumab and chemotherapy had improved progression-free survival (PFS) and median OS (17.2 vs. 12.6 months, one-sided p-value = 0.0056) [74]. ORIENT-16 demonstrated superior OS with sintilimab and chemotherapy in all patients regardless of PD-L1 status (31.9% risk reduction, p < 0.0001), with a stronger reduction in patients with CPS ≥ 5 (41.3% risk reduction, p < 0.0001) [73]. ATTRACTION-4, a phase II-III multicenter trial across Japan, Republic of Korea, and Taiwan, randomized patients with unresectable advanced or recurrent GC/GEJC to either nivolumab with chemotherapy or placebo and chemotherapy; they found that the addition of nivolumab significantly improved PFS (10.45 vs. 8.34 months, p = 0.0007) but not OS [75].

Several studies are underway to investigate the efficacy of ICIs with neoadjuvant and/or adjuvant chemotherapy for non-MSI-H LAGC. A single-arm phase II trial (NCT0291816) recruited 34 patients with LAGC/GEJC; these patients were treated with capecitabine and oxaliplatin (CAPOX) before and after surgery as well as one cycle of pembrolizumab immediately before surgery followed by twelve months of maintenance pembrolizumab [76]. It was found that 20.6% of patients achieved pCR, surpassing the target pCR of 15% [76]. DANTE is a multicenter phase II trial that is comparing perioperative atezolizumab (anti-PD-L1) and FLOT chemotherapy against FLOT alone in patients with operable LAGC regardless of PD-L1 status (≥cT2 or N+ GC/GEJC) [77]. The preliminary results show an increase in pathological regression rates in patients who received both atezolizumab and FLOT, particularly in patients with higher PD-L1 expression (overall pT0 23% vs. 15%, pN0 68% vs. 54%). The estimated accrual completion date is 2025. In addition, the KEYNOTE-585 trial is underway to investigate the efficacy and safety of neoadjuvant and adjuvant pembrolizumab and chemotherapy (either FLOT, cisplatin and 5-FU, or cisplatin and capecitabine) compared to placebo and chemotherapy in patients with stage II-IVa GC or GEJC [78]. The primary endpoints are OS, event-free survival, and pCR rate [78]. KEYNOTE-585 is estimated to complete accrual in June 2024 and will hopefully further elucidate the role of adjuvant immunotherapy in non-MSI-H LAGC. Finally, the MATTERHORN trial is currently recruiting around 900 patients with resectable stage II or higher GC/GEJC. These patients will be randomly assigned to receive either neoadjuvant durvalumab or placebo in combination with FLOT chemotherapy, followed by either adjuvant durvalumab or placebo monotherapy after surgical resection [79]. Durvalumab in combination with FLOT has been shown to have the potential to improve outcomes in two smaller clinical studies in patients with advanced GC/GEJC, hence leading to the development of MATTERHORN [80,81]. Expected accrual completion is in 2025.

Recent studies suggest that the timing of immunotherapy with chemotherapy may be critical the in treatment of GC. ATTRACTION-5 is a phase III trial of Asian patients with stage III GC/GEJC randomized to receive either nivolumab with adjuvant chemotherapy or placebo with adjuvant chemotherapy [82]. There was no improvement in OS, and the primary endpoint of PFS was not met (HR = 0.90, p = 0.4363) [82]. A phase II randomized trial (NCT04250948) evaluated the efficacy of the addition of toripalimab to perioperative chemotherapy in patients with LAGC/GEJC [83]. In the group receiving PD-L1 inhibitors, toripalimab was added preoperatively and as a maintenance adjuvant therapy. Patients in the toripalimab plus chemotherapy arm achieved a higher proportion of pCR (24.1% vs. 9.3%, p = 0.039) and TRG 0/1 (44.4% vs. 20.4%, p = 0.009) than patients in the chemotherapy arm [83]. Although these studies were conducted in different populations, together they suggest that perioperative immunotherapy may be more effective than adjuvant only when combined with chemotherapy. The upcoming results of MATTERHORN and KEYNOTE-585 will further inform our understanding of the role of immunotherapy in LAGC.