Tissue Biomarkers in Gastric Cancer Treatment: Present and Future

Abstract

The aggressive nature of gastric cancer often leads to late diagnosis and poor prognosis. Chemotherapy and the more recently added immunotherapy remain key treatments for this disease. Several studies have focused on identifying tissue biomarkers with prognostic and/or predictive roles and therefore the therapeutic options are rapidly growing. In this narrative review, we summarize the major tissue biomarkers routinely assessed in clinical practice. In addition, we focus on new evidence about emerging tissue biomarkers that could have a predictive role in future therapeutic approaches and also on the potential role of liquid biopsy in this neoplasm.

1. Introduction

Gastric cancer (GC) accounted for approximately 5.6% of newly diagnosed cancer in 2020 and the incidence will continue to increase due to the ageing population and rising exposure to risk factors. The incidence of GC is higher in men, and it varies by region; the highest rates are observed in Eastern Asia, Central and Eastern Europe and South America [1]. These differences in terms of incidence could be explained by the prevalence of helicobacter pylori and Epstein–Barr virus (EBV) infections, alcohol habits, high-salt diets, low fruit and vegetable intake, obesity and gastro-esophageal reflux [2]. GC aggressiveness relies on rapid tumor growth, which makes it hard to detect the malignancy in its early stages. Overall, about 60% of patients with GC are not eligible for curative treatment due to the advanced stage of the disease or comorbidities [3]. Even though the therapeutic options are rapidly growing, the prognosis of GC remains poor, with a 5-year relative survival rate of 36% [4].

Since the first evidence in 1957 [5], the fluoropyrimidine-based chemotherapy (CT) combination has remained the backbone of perioperative treatment for locally advanced disease [6,7,8] and the frontline treatment for advanced stage disease [9,10,11,12]. Recently, immunotherapy with programmed cell death-1 (PD-1) inhibitors in combination with CT has been introduced in specific subtypes in the advanced disease [13,14].

Despite initially high chemosensitivity, the rapid onset of chemoresistance dramatically impacts on patients’ prognosis; survival in clinical trials assessing the value of chemotherapy has historically been <1 year in non-Asian countries [15]. The standard second-line treatments have consisted of mono-chemotherapy paclitaxel, docetaxel or irinotecan for years, although recently the addiction of the anti-vascular endothelial growth factor receptor 2 (VEGFR2) antibody, ramucirumab, to paclitaxel, or as a single agent [16], became the new second-line standard treatment [17].

In this poor scenario, several studies have focused on identifying tissue biomarkers with prognostic and/or predictive roles. Their differentiation into intestinal and diffuse types according to the Lauren classification [18] and into papillary, tubular, mucinous (colloid) and poorly cohesive carcinomas according to the system proposed by the World Health Organization [19] showed no clinical utility. The Lauren classification [18] divides GC as follows:

(a)

Diffuse type (poor cohesion between neoplastic cells, variable component of cells showing signet ring cell morphology);

(b)

Intestinal type (composed of tubular or glandular structures similar to intestinal adenocarcinoma);

(c)

Mixed;

(d)

Undifferentiated.

The WHO classification [19] of GC comprises a more detailed classification based on the morphological and, in some cases, immune expression profile of tumors:

(a)

Adenocarcinoma NOS (tubular, parietal, papillary, micropapillary, mucoepidermoid, mucinous, signet ring, poorly cohesive, medullary, hepatoid, Paneth cell);

(b)

Squamous cell carcinoma;

(c)

Adenosquamous carcinoma;

(d)

Undifferentiated carcinoma (large cell, pleomorphic, sarcomatoid);

(e)

Gastroblastoma;

(f)

Neuroendocrine tumor.

Some subtypes can be recognized in both classifications: the Lauren intestinal subtype corresponds to the papillary and well- and moderately differentiated tubular carcinoma in the WHO classification. The Lauren diffuse subtype corresponds with the poorly cohesive categories of the WHO classification [20].

The Cancer Genome Atlas (TCGA) molecular classification of GC has identified four major genomic classes, each characterized by distinct pathway alterations. These alterations hold potential for identifying clinically relevant biomarkers and developing new therapeutic options [21]. There is some overlap in the genomic features defining each molecular subtype among the four classes.

The first group comprises Epstein–Barr Virus (EBV)-infected tumors, which exhibit a significant enrichment of a high EBV burden and extensive DNA promoter hypermethylation and are associated with a better prognosis [22]. The second group consists of microsatellite instability (MSI) tumors, which are enriched for MSI and exhibit elevated mutation rates and hypermethylation (including at the MLH1 promoter). Chromosomal instability (CIN) and genomically stable (GS) types are distinguished by the presence or absence, respectively, of extensive somatic copy-number aberrations. The GS subtype is associated with the worst prognosis [22]. Notably, receptor tyrosine kinase oncogene amplifications, such as human epidermal growth factor receptor 2 (HER2), epidermal growth factor receptor (EGFR), fibroblast growth factor receptor 2 (FGFR2) and MET, are particularly described in the CIN subtype [21].

Understanding these pathway alterations has led to the development of targeted cancer therapies. However, these therapies are more effective in specific subgroups of GC expressing specific biomarkers. The aim of this review is to provide an overview on the tissue biomarkers that are currently useful in clinical practice and other biomarkers with potential utility in the future.

2. Tissue Biomarkers in Standard Treatments

2.1. Programmed Cell Death Ligand 1

The programmed cell death ligand 1 (PD-L1) is a protein expressed mainly in activated T-lymphocytes and other immune system cells and it represents the primary ligand of PD-1. This interaction is among the multiple co-inhibitory signals that modulate T cell activation against self-proteins for the maintenance of self-tolerance as well as infections and tumor antigens [23]. The expression of PD-L1 can be exploited by cancer cells as an important mechanism of immune escape [24]. Thus, to guide patient selection for immune checkpoint inhibitor (ICI) therapy, PD-L1 biomarker expression is routinely assessed by immunohistochemistry (IHC). PD-L1 expression can be explicit as a tumor proportion score (TPS), which is the ratio of PD-L1-expressing cells to all viable tumor cells; however, in GC, the combined positive score (CPS), which also includes the number of PD-L1-positive lymphocytes and macrophages, is considered a more reliable predictive biomarker for immunotherapy efficacy [25]. Therefore, major guidelines recommend the assessment of the PD-L1 combined positive score (CPS) in patients with metastatic GC to tailor first-line treatment in combination with chemotherapy [26,27]. Different IHC PD-L1 antibody assays (Dako 22C3, Dako 28–8, Ventana SP-142 and Ventana SP-263) have been recognized by the Food and Drug Administration (FDA) as companion diagnostics to assess PD-L1 and candidate patients for different ICI treatments in various tumor types [28].

One of the first evidence of ICI activity in GC was reported by the KEYNOTE-059 phase II study, evaluating the efficacy of monotherapy with pembrolizumab in heavily pretreated GC. Based on these results, the Food and Drug Administration (FDA) approved pembrolizumab as a third-line treatment for GC patients with a CPS ≥ 1 [29]. However, in the subsequent KEYNOTE-061 phase III trial, pembrolizumab failed to demonstrate a significant improvement in overall survival (OS) versus paclitaxel as a second-line therapy for patients with a PD-L1 CPS ≥ 1 GC [30]. The antibody used to determine the CPS in KEYNOTE-059 and KEYNOTE-061 was the Dako 22C3 [29,30].

The CheckMate-649 trial compared the addition of nivolumab to CT versus CT alone (either capecitabine/oxaliplatin or 5-FU/leucovorin/oxaliplatin) in treatment-naive gastric, esophago-gastric junction (OGJ) or esophageal cancer: in patients with a PD-L1 CPS ≥ 5 (evaluated with Dako 28-8 antibody) GC, the addition of nivolumab to CT improved OS with a hazard ratio (HR) of 0.71 (98% CI, 0.59 to 0.86; p < 0·0001) and progression-free survival (PFS) (HR 0.68, 98% CI, 0.56 to 0.81; p < 0·0001) [13]. Based on this evidence, ICIs targeting the PD-1/PD-L1 pathway combined with fluoropyrimidine and platinum-based CT are now standard first-line treatment for patients with HER2-negative PD-L1 CPS ≥ 5 GC [14].

Table 1. Trial with PD1 inhibitors in clinical practice. Abbreviations: GC: gastric cancer; CPS: PD-L1 combined positive score, FDA: US Food and Drug Administration, CT: chemotherapy, EMA: European Medicines Agency.

Drug	Trial and Phase	Determination of PDL1 Level	Design of the Study and Results	Ref.
Pembrolizumab	KEYNOTE-059, phase II	CPS ≥ 1 with Dako 22C3	Monotherapy in heavily pretreated GC—FDA approved	[29]
Pembrolizumab	KEYNOTE-061, phase III	CPS ≥ 1 with Dako 22C3	Pembrolizumab vs. paclitaxel as II line—no advantages	[30]
Nivolumab	CheckMate-649, phase III	PD-L1 CPS ≥ 5 with Dako 28-8	Nivolumab + CT vs. CT alone as first line—FDA and EMA approved	[14]

shows clinical trials and evidence on the combination of immunotherapy and chemotherapy in GC patients.

In the primary clinical trials involving these two assays, the question of whether the 22C3 and 28-8 pharmDx assays could be used interchangeably in GC remained unanswered. A recent study demonstrated a high overall percentage agreement between the PD-L1 22C3 and 28-8 pharmDx assays at CPS cut-offs of 1, 10 and 50 [31]. However, these findings were not corroborated by another study, which suggested that in a larger cohort, tumors that tested positive for PD-L1 using the 28-8 assay were twice as prevalent compared to those tested using the 22C3 and other assays [32]. Other criticisms regarding the use of this biomarker in clinical practice stem from variations in cut-off values and the diverse range of assays employed in clinical trials. Additionally, PD-L1 expression exhibits spatial and temporal heterogeneity [33,34], potentially leading to inconsistencies in PD-L1 expression between surgical samples and their corresponding biopsy specimens. Moreover, CT exposure can alter PD-L1 expression levels in tumor samples [35]. Various thresholds for the TPS and CPS have been employed across studies, with a CPS cut-off of ≥1 indicating PD-L1-positive tumors; the prevalence of PD-L1 CPS ≥ 1 tumors ranges between 50% and 60% [25,36].

Studying PD-L1 expression differences between core biopsies and surgical resection samples in GC remains an unmet need. Ye et al. addressed this gap by conducting a study aimed at determining the number of core biopsies needed to accurately reflect the PD-L1 expression status of resection specimens [37]. The study highlighted the spatial heterogeneity of this biomarker, revealing variations in PD-L1 staining status among different cores from the same tumor. Moreover, the sensitivity and concordance rates of a single tissue microarray core in predicting the PD-L1 status of the entire tumor were found to be very low (0.49 and 60.4%, respectively). Consequently, the study suggests that a minimum of five core biopsies is necessary to achieve high concordance with the overall tumor PD-L1 status [37].

2.2. Human Epidermal Growth Factor Receptor 2 (HER2)

HER2 is a transmembrane protein member of the epidermal growth factor receptor (EGFR) family. HER2 heterodimerization initiates a signal transduction cascade that promotes cell proliferation [38]. When amplificated and/or overexpressed, HER2 acts as an oncogene in numerous types of human cancer, especially in breast cancer and GC [39].

The overexpression of HER2 is reported in approximately 20% of patients with GC, being more prevalent in GEJ tumors, in the intestinal subtype and in well-differentiated GC [40,41]. Several studies have demonstrated that HER2 overexpression represents an independent negative prognostic marker in GC as it correlates with tumor size and lympho-vascular invasion [42,43,44]. After the development and success of molecular targeted drugs, HER2 expression has become an important predictive biomarker; thus, its accurate determination is crucial to identify patients who might benefit from HER2-targeting therapy.

According to the guidelines, HER2 status determined by IHC to assess protein overexpression or by fluorescence in situ hybridization (FISH) to test for HER2 gene amplification is recommended to guide the choice of first-line therapy [45,46]. Since 2010, the combination of trastuzumab and standard CT has represented the first-line treatment for patients with advanced HER2-overexpressing GC: in the phase III ToGA trial, the addition of trastuzumab to platinum/fluoropyrimidine-based CT significantly improved both the PFS (6.7 vs. 5.5. months, HR of 0.71, p = 0.0002) and OS (13.8 vs. 11.1 months; HR 0.74, p = 0.0046) [47].

In recent years, preclinical research has investigated the crosstalk between the PDL-1/PD1 axis and HER2, leading to clinical trials aimed at assessing the potential synergistic activity of a dual PD-1/HER2 blockade [48,49]. Recently, in the phase III KEYNOTE-811 trial, the combination of pembrolizumab with trastuzumab and CT as the first-line treatment for HER2-positive advanced GC showed higher overall response rate (ORR) and significantly improved PFS compared to the standard of care (SoC), especially in patients with a PD-L1 CPS ≥ 1 (10.8 months versus 7.2 months; HR 0.70). Based on these findings, both the FDA and EMA granted accelerated approval for the pembrolizumab/trastuzumab/chemotherapy combination, which currently represents the treatment of choice for first-line therapy in HER2 advanced GC with a PD-L1 CPS ≥ 1 [50].

The treatment landscape is rapidly evolving, and new molecules have demonstrated promising activity. Regarding first-line therapy, the use of zanidatamab (ZW25), an HER2-targeted bispecific antibody, combined with chemotherapy, demonstrated durable disease control, with encouraging PFS and OS results: an ongoing global phase III study (HERIZON-GEA-01) is evaluating zanidatamab plus standard chemo with or without the PD-1 inhibitor, tislelizumab, for the first-line treatment of advanced HER2-positive advanced GC [51].

Margetuximab is an Fc-engineered, anti-HER2 monoclonal antibody designed to increase antibody-dependent cytotoxicity (ADCC) and enhance PD-1/PD-L1 axis expression [52]. In cohort A of the phase II/III MAHOGANY trial, the combination of margetuximab and retifanlimab, an anti-PD1 agent, demonstrated a favorable toxicity profile and ORR compared with the historical outcomes of CT plus trastuzumab in first-line HER2-positive/PD-L1-positive advanced GC [53].

Moving to the second-line setting, several phase III trials with other HER2-targeting agents, such as pertuzumab, lapatinib and trastuzumab emtansine (T-DM1), failed to demonstrate any significant clinical benefit for HER2-positive GC or GEJ cancer [54,55,56]. This discrepancy may be due to both intra- and inter-tumoral heterogeneity in HER2 expression and its downregulation following trastuzumab therapy [57,58]. For these reasons, the reassessment of HER2 status through tissue biopsy is generally required in clinical trials and is recommended, even though recent studies suggest the potential role of the less invasive circulating tumor cell DNA in tracking trastuzumab resistance and evaluating HER2 expression [59,60,61]. Trastuzumab deruxtecan (T-DXd) is an antibody–drug conjugate (ADC) consisting of an anti-HER2 antibody, a cleavable tetrapeptide-based linker and a cytotoxic topoisomerase I inhibitor. With its high drug-to-antibody ratio (8:1) and membrane permeability of its payload, T-DXd does not require high-level HER2 expression and can be active against cancers with low HER2 expression using the bystander effects of the payload [62]. T-DXd proved to be effective in advanced HER2-positive GC: DESTINY-Gastric01 showed an ORR of 51% of the patients treated with T-DXd versus 14% of those in the physician’s choice group (p < 0.001) and an improvement in OS in heavily treated patients, and DESTINY-Gastric02 also confirmed the results in a second-line setting [63,64]. The ongoing randomized phase III DESTINY-Gastric04 study is investigating the efficacy and safety of T-DXd compared with the SoC ramucirumab and paclitaxel and will define the optimal second-line therapy in patients progressed from a first-line treatment containing trastuzumab [65].

Concerning third- or later-line treatment options, a randomized ongoing phase III study, RC48-C007, want to compare a novel ADC, disitamab vedotin, with the physician-selected standard treatment in patients with heavily pretreated advanced GC with HER2 overexpression [66].

Table 2. Anti-HER2 treatment in GC. Abbreviations: CT: chemotherapy; SoC: standard of care, FDA: US Food and Drug Administration, EMA: European Medicines Agency.

Drug	Trial and Phase	Design of the Study	Level of Development	Ref.
Trastuzumab	ToGA trial, phase III	Trastuzumab + CT vs. CT alone as first line	SoC	[47]
Trastuzumab	KEYNOTE-811, phase III	Trastuzumab + pembrolizumab + CT vs. standard of care as first line in CPS ≥ 1	FDA and EMA approved	[50]
Zanidatamab (ZW25)	HERIZON-GEA-01, phase III	Zanidatamab plus standard CT with or without the PD-1 inhibitor as first line	Ongoing trial	[51]
Margetuximab	MAHOGANY, phase II/III	Margetuximab with antiPD1 vs. standard of care	Ongoing	[53]
Lapatinib	TyTAN—A, phase III	Lapatinib + paclitaxel vs. paclitaxel as second line	Failed	[54]
Pertuzumab	JACOB, phase III	Pertuzumab + standard of care vs. standard of care as first line	Failed	[55]
Trastuzumab emtansine (T-DM1)	GATSBY, phase II/III	T-DM1 vs. paclitaxel as second line	Failed	[56]
Trastuzumab deruxtecan (T-DXd)	DESTINYGastric04, phase III	T-DXd vs. SoC as second line	Ongoing	[65]

summarizes the anti-HER2 drugs and their level of development.

2.3. Microsatellite Instability (MSI)

The mismatch repair (MMR) system comprises a set of proteins involved in repairing DNA errors generated during DNA replication [67]. The MMR proteins with clinical relevance in human cancer biology are MLH-1, MSH-2, MSH-6 and PMS-2 [68]. Deficient mismatch repair (dMMR) function leads to the accumulation of mutations in the genome, which is a common occurrence in cancer development. This can arise from either germline mutations or somatic events [69]. The most common cause of somatic MMR gene inactivation involves MLH-1, often due to the hypermethylation of its promoter sequence. It is typically linked to widespread genomic hypermethylation, known as CpG island methylator phenotype (CIMP) high status. Defective MMR results in an elevated tumor mutational burden (TMB), associated with the expression of numerous abnormal proteins acting as neoantigens. These neoantigens serve as targets for cancer immunotherapies [70].

Defective MMR can be identified through IHC to assess MMR protein expression or by detecting microsatellite instability (MSI) using the polymerase chain reaction (PCR) or next-generation sequencing (NGS) methods [71].

MSI-high (MSI-H) phenotypes represent a distinct subclass, and they are mutually exclusive with other biomarkers usually used in clinical practice [72]. Patients with MSI-H GCs who have undergone radical resection have a better prognosis compared with patients with non-MSI-H subtypes [23].

In the metastatic setting, the MSI-H phenotype has shown high responsiveness to immunotherapy, as evidenced by findings from several clinical trials such as phase II KEYNOTE-059, phase III KEYNOTE-061 and phase III KEYNOTE-062. In a post hoc analysis conducted by Joseph Chao et al., it was revealed that pembrolizumab, either as a monotherapy or in combination with chemotherapy (CT), demonstrated durable antitumor activity compared to chemotherapy alone in patients with MSI-H GC or GEJ cancer. Importantly, this effect was observed regardless of the line of therapy in which pembrolizumab was administered [29,30,73,74]. A prespecified analysis of CheckMate 649 showed a survival advantage associated with nivolumab combined with chemotherapy in the group of patients with MSI-high tumors compared to those with microsatellite-stable tumors [13].

A recent meta-analysis with retrospective data on MSI status showed that anti-PD-1 agents with or without CT significantly and consistently improved OS, PFS and ORR compared to CT alone in the subgroup of patients with MSI-H advanced GC [75]. Based on the results of the phase II KEYNOTE-158, pembrolizumab is recommended in monotherapy as a second-line treatment for MSI-H/dMMR GC [74,76]. PD1/PDL1 expression changes with MMR status: PD1/PDL1 expression and CD8+ T cell/CD68+ macrophage densities are higher in MSI GC and can enhance the efficacy of immunotherapy [77].

The combination of CT and immunotherapy or immunotherapy alone may be alternative treatment choices even in the neoadjuvant setting. Phase II IMHOTEP was designed to evaluate the safety and effectiveness of perioperative pembrolizumab in patients with localized dMMR/MSI tumors, regardless of the tumors’ anatomical site. An interim analysis revealed a limited pathological complete response (pCR) rate to short-course neoadjuvant pembrolizumab; however, prolonging the pre-operative treatment duration might enhance the pCR rate [78]. The promising findings from the phase II NEONIPIGA trial support the safety and efficacy of neoadjuvant nivolumab plus low-dose ipilimumab followed by adjuvant nivolumab in patients with resectable locally advanced dMMR/MSI-H resectable GC/GEJ adenocarcinoma [79]. The study demonstrated a high rate of pCR without an increase in expected toxicities [79]. Given the better survival association and the potential limited benefit from CT in MSI-H tumors, neoadjuvant or definitive dual CTLA-4/PD(L)-1 inhibition may allow the omission of chemotherapy or surgery. The INFINITY II trial investigated the activity and safety of tremelimumab–durvalumab as neoadjuvant or definitive treatment for MSI, dMMR and EBV-negative resectable GAC/GEJAC. Pre-operative treatment for 3 months was safe and exhibited promising efficacy in MSI dMMR patients, suggesting the possibility of exploring non-operative management in patients with a clinical, pathological and molecular complete response after treatment [80]. Although not practice-changing, these results lay the groundwork for potential paradigm shifts in the standard of care in this patient subgroup.

2.4. Neurotrophic Tyrosine Receptor Kinases (NTRKs)

NTRKs are a family of transmembrane receptor tyrosine kinases involved in neuronal development. The three TRK family members, TRKA, TRKB and TRKC, encoded by the NTRK1, NTRK2 and NTRK3 genes, respectively, have an extracellular ligand-binding domain, a transmembrane region and an intracellular kinase domain and their activation promotes survival and proliferation in cancer cells [81]. Rearrangements of the NTRK1, NTRK2 and NTRK3 genes are the result of intra- or inter-chromosomal translocations. The 3’ end of the NTRK gene, which contains the tyrosine kinase domain, fuses with the 5’ end of a different gene, resulting in a fusion protein that is missing the ligand-binding domain, leading to ligand-independent receptor activation and the uncontrolled initiation of downstream pathways, lastly driving oncogenesis [82]. While other molecular alterations, such as activating splice variants, somatic mutations and amplifications, have been observed, there is no evidence to suggest that they function as oncogenic drivers. Additionally, these alterations do not exhibit a noteworthy response to NTRK inhibitors [82].

Targeted RNA-sequencing methods might be considered the gold standard for NTRK fusion detection, assuming that the RNA quality is optimal. Furthermore, where tissue availability is restricted, a DNA/RNA-based approach may be preferred if DNA and/or RNA has already been extracted and is accessible; IHC could serve as screening method or to confirm the presence of the fusion [83].

NTRK-altered GCs represent a rare entity: they represent < 1% of GC. GCs represent 8.5% of NTRK-mutated tumors [84]. Usually, NTRK alterations are detected in CIN neoplasms [21]. In clinical practice, the decision to conduct NTRK fusion testing involves the consideration of various factors, such as the tumor type and probability of harboring a fusion, resource accessibility and costs, tissue availability, requirements for concomitant genomic profiling and clinical evaluation. Currently, there are no guidelines indicating the routine clinical test useful for NTRK fusion detection or its treatment in GC [84].

Larotrectinib has received FDA agnostic approval for the treatment of patients with advanced solid tumors harboring an NTRK gene fusion. Given the poor prognosis of GC and GEJ cancer patients, larotrectinib might be a potential therapeutic option [85]. Recently, the TRIDENT-1 trial showed the efficacy of repotrectinib [86], obtaining FDA accelerated approval for the treatment of NTRK-positive advanced tumors [87].

In a preliminary analysis of data from three phase I/II trials (ALKA-372-001, STARTRK-1 and STARTRK-2 [88,89]), entrectinib, a potent TRKA/B/C inhibitor, demonstrated significant and durable antitumor activity and a manageable safety profile in adults with NTRK fusion-positive (NTRK-fp) gastrointestinal cancers. An updated analysis confirmed that entrectinib maintains a high level of clinical benefit for patients with NTRK-fp solid tumors and a manageable safety profile independently of cancer primitivity, reaffirming its promising therapeutic potential [90].

3. Future Perspectives

3.1. Epstein–Barr Virus

EBV is an oncovirus involved in the genesis of different cancers (e.g., nasopharyngeal carcinoma and T cell lymphoma) but also in 9% of GCs [91,92]. In the Cancer Genome Atlas (TCGA) classification, EBV-associated GCs are characterized by high tumoral immune infiltration and show molecular signs of potential sensitivity to immunotherapy such as the amplification of the chromosome 9 locus harboring the PD-L1 and PD-L2 genes [21,93]. Derks et al. found that PD-L1 expression by tumors or tumors infiltrating immune cells is a general phenomenon in EBV-associated cancers along with MSI GCs; this provides a strong rationale for testing ICIs in this patient population and for evaluating EBV status along with MSI status as key variables in immunotherapy trials [93].

The EBV test can be performed using EBV-encoded RNA (EBER) in situ hybridization (ISH) or immunostaining, but these tests have a low sensitivity. Bai et al. developed an NGS-based EBV test for the simultaneous molecular definition of GC, which could improve the diagnostic sensitivity and specificity provided by the EBER-ISH test [94].

The gene expression profile analysis of EBV-positive GC shows an increased expression of the gene involved in the immune response with the activation of infiltrating T CD8+ T cells and macrophages and a better impact on prognosis [91,95]. However, the correlation between EBV status and treatment efficacy with nivolumab has not been evaluated in the CheckMate 649 [13] and ATTRACTION-04 trials [96]. A single-center phase II trial evaluated 61 patients with metastatic GC treated with pembrolizumab in monotherapy; the patients underwent molecular profiling including an EBV test of the primary tumor with EBV DNA sequencing. Six (9.8%) patients were confirmed as EBV-positive and all of them obtained a benefit from pembrolizumab with a partial response, a median duration of response of 8.5 months and an ORR of 100% [97]. In addition, they confirmed that EBV-associated GC and MSI GC are mutually exclusive [97].

EBV infection could be a predictive biomarker of response to ICI as effective as MSI-H, but further studies are needed to establish its potential clinical role.

3.2. Claudin 18.2

Claudins (CLDNs) play an essential role in cellular adhesion and migration. The expression of CLDNs has been reported in several types of cancer, including GC, showing its potential role as a new therapeutic target [98]. Claudin 18 isoform 2 (CLDN18.2), a member of the claudin protein family, plays a crucial role in regulating tissue permeability, paracellular transport and signal transduction through tight junctions. It is primarily expressed in the gastric mucosa and remains present even during malignant transformation. Claudin 18.2 may become more exposed in malignant tissues with the disruption of tight junctions; for this reason, it represents an attractive target for cancer treatment [99,100]. CLDN18.2 expression was assessed by immunohistochemistry (IHC) and CLDN positivity was defined as moderate-to-strong expression in ≥75% of tumor cells [101]. The prevalence of CLDN positivity is reported to be 30–33% in patients with GC/GEJC and it is associated with diffuse-type [102,103] and GS GCs [21].

Zolbetuximab is a novel chimeric immunoglobulin G1 antibody highly specific for CLDN18.2. It binds to CLDN18.2 on the tumor cell to activate ADCC and complement-dependent cytotoxicity [104]. The SPOTLIGHT trial, a global double-blinded phase III trial, aimed to verify the efficacy of CT with a modified schedule of 5-fluorouracile and oxaliplatin for 6 cycles (mFOLFOX6) plus Zolbetuximab compared with mFOLFOX6 plus placebo in CLDN18.2-positive advanced GC/GEJC. Both the primary endpoint of PFS and the secondary endpoint of OS were significantly improved with the addition of zolbetuximab to the standard treatment [105]. The GLOW trial showed that zolbetuximab added to first-line chemotherapy with capecitabine and oxaliplatin added a benefit in terms of PFS and OS in CLDN18.2-positive, HER2-negative GC [106].

Kubota Y et al. examined genomic alterations and immune cell markers in relation to CLDN18.2 expression. CLDN-positive tumors (cut off by 75%) could be found among various molecular subtypes, suggesting that CLDN18.2 could be targetable regardless of the molecular subtype. In addition, there were no significant differences between CLDN-positive and CLDN-negative tumors regarding the efficacy of standard first- and second-line CT, and CLDN positivity was not a prognostic factor in advanced unresectable or metastatic GC/ GEJC. Furthermore, nearly 50% of the CLDN-positive tumors showed a concomitant PD-L1 CPS ≥ 5 [101]; for this reason, an ongoing trial is evaluating the efficacy and safety of zolbetuximab in combination with immunotherapy [107].

3.3. Tumor Mutational Burden (TMB)

In the years since the immunotherapy revolution, the tumor mutational burden (TMB) has been studied as a potential predictive biomarker of response to ICIs [108]. The TMB cut-off value described for predicting a response in GC patients treated with ICIs was 13.31 mt/Mb [109]. The TMB is represented by the total number of mutations present in a tumor specimen and it has been hypothesized that highly mutated tumors express several neoantigens that could contribute to immune cell activation [110]. MSI-H tumors are generally associated with a high TMB, since defects in DNA damage repair such as dMMR lead to the accumulation of mutations, although there are microsatellite-stable (MSS) tumors with a high TMB [110].

Different trials have tried to evaluate the association between the TMB and ICI response in GC. Kim ST et al. revealed, with a small study on GC, how MSI-H, EBV positivity and high TMB were associated with high response rates to pembrolizumab [97]. A post hoc analysis of the KEYNOTE-061 second-line trial showed a prolonged OS, higher response rate and durable responses with pembrolizumab in patients with a PD-L1 CPS ≥ 10 [36]. An exploratory analysis of KEYNOTE-061 evaluated the association of the TMB with PD-L1 and MSI-H and their relationship with the clinical outcomes, revealing a significant association between the TMB and clinical outcomes in the pembrolizumab arm. However, this analysis is weakened by the small sample size [111]. Also, in the first-line setting, a favorable association between the TMB and ICI benefit emerged: the KEYNOTE-062 trial suggested a benefit in the TMB in GC treated with the addition of pembrolizumab; however, the clinical utility of the data was attenuated by the exclusion of MSI-H tumors [112].

The TMB is a significant predictive biomarker for selecting patients who may benefit from ICIs, but its clinical application is limited by the high cost and wet laboratory resources needed for its evaluation. To reduce costs and increase the reproducibility of the data, a multimodal fusion deep learning model was tested to predict TMB levels [113].

3.4. BRAF Mutations

Raf Murine Sarcoma Viral Oncogene Homolog B (BRAF) mutations are related to tumorigenesis since they can influence abnormal cell growth and proliferation through MEK and ERK pathway activation [114]. The most frequently detected BRAF mutation is a single amino acid substitution (V600E) in exon 15 and BRAF mutation is frequent in melanoma (67%) and colorectal cancer (10%) [115]. It is not yet clear if BRAF mutation is a driver mutation in GC: BRAF mutations were observed only in 2% of patients and most of them were uncommon mutations of BRAF (i.e., V599M) [116].

Lee S. et al. suggested that alteration of the rat sarcoma (RAS)–Rapidly Accelerated Fibrosarcoma (RAF) kinase pathway by BRAF mutation together with RAS mutation may play an important role in gastric carcinogenesis [117]. In colorectal cancer, KRAS and BRAF mutations appear to be mutually exclusive, while two reports demonstrated that GC could harbor KRAS and BRAF mutations simultaneously [117,118].

Because of the rarity of these alterations, no data are available about the efficacy of treatments available for BRAF-mutated tumors.

3.5. RET

REarranged during Transfection (RET) is a transmembrane receptor encoded by the RET proto-oncogene situated on chromosome 10q11.2 and plays an important role in the development of nervous and urogenital systems and cells derived from the neural crest [119,120]. Chromosomal rearrangements, point mutations and overexpression of the RET gene can result in the development and progression of various tumors [121]. There is no widely accepted standard method for detecting RET rearrangements. In general, reverse transcription polymerase chain reaction (RT-PCR) and FISH are both sensitive and effective approaches [122].

RET alterations, such as fusions or mutations, are present in multiple tumor types, in particular, in lung and thyroid cancers but also in non-canonical (e.g., gastrointestinal, breast, gynecological, genitourinary, histiocytic) cancers [123].

Many multikinase RET inhibitors (MKIs), such as cabozantinib and lenvatinib, have demonstrated activity against RET. The relatively limited efficacy of MKIs in treating RET-driven cancers can be partially attributed to their off-target activity [124,125,126,127]. In recent years, there has been significant progress in the development of selective RET inhibitors aimed at higher potency and lower toxicity. The next-generation small molecular inhibitors pralsetinib (BLU-667) and selpercatinib (LOXO-292) have been rapidly tested in clinical practice [128,129]. Beyond lung cancers and thyroid cancers, the clinical activity of selective RET inhibitors has been seen in patients with GI cancers (pancreatic cancer and intrahepatic bile duct carcinoma) [130,131]. In total, 3.3% of RET-altered tumors are GCs [132]. Overall, further research into the potential benefits of RET inhibitors in the treatment of GC is needed.

3.6. MET

Hepatocyte growth factor receptor (MET) is a tyrosine kinase receptor with pleiotropic effects. MET might be, together with its ligand hepatocyte growth factor/scatter factor (HGF/SF), a possible therapeutic target in GC. The MET pathway is often hyperactivated in GC by overexpression, which is detectable by IHC and/or in situ hybridization in 50–60% of tumors [133]. Since its implementation in clinical practice, MET aberrations are often detected by an NGS test. Similarly to other biomarkers, MET expression in GC is heterogenous and it has been demonstrated that MET-positive GCs are characterized by poor prognosis [134].

Despite numerous studies on anti-MET agents, few notable results have been achieved. The P-phase I/II and III trials that explored monoclonal antibodies directed against the HGF/MET pathway, such as rilotumumab, onartuzumab or emibetuzumab, failed to demonstrate activity in the treatment of GC [135,136,137,138,139]. In addition, multi-tyrosine kinase inhibitors (TKIs) (e.g., crizotinib) and selective MET TKIs (capmatinib, tivantinib, savolitinib and AMG337) were studied in GC in phase II trials conducted on small groups of patients, but they showed no significant effectiveness [140,141,142,143]. The VIKTORY umbrella trial showed the most successful response to a selective MET TKI, savolitinib, in patients affected by GC and compared to the standard of care, paclitaxel plus ramucirumab as a second-line therapy. MET amplification was determined by NGS or FISH and an ORR of 50% was reached, showing the possible potential of cMET inhibitors in GC in a well-defined population [144].

Further research is needed to establish the role of MET inhibitors in GC treatment.

3.7. FGFR2

The dysregulation of the fibroblast growth factor receptor (FGFR) pathway has been studied in many different types of tumors [145]. In GC, the FGFR2 is the most frequently mutated member of the FGFR family; the prevalence of the FGFR2 amplification varies by country, ranging from 4–7.4% [146,147]. FGFR2 amplification can be detected both using ISH or NGS [148]. So far, there is no established association between FGFR2 amplification and the anatomical site, histological subtype or stage [148]. Hur et al. demonstrated that FGFR2 gene amplification is a poor prognostic factor in patients with metastatic GC and that a high copy number (≥30) identified by NGS-based genomic profiling is significantly associated with a lower PFS and a shorter OS [149].

In the phase II FIGHT clinical trial, a novel FGFR2 inhibitor, bemarituzumab, a humanized anti-FGFR2b monoclonal IgG1 antibody, has shown interesting results in patients affected by advanced GC, with FGFR2 gene amplification in the circulating tumor DNA (ctDNA); the patients received mFOLFOX6 plus bemarituzumab as a first-line therapy versus mFOLFOX6 plus placebo. The primary endpoint, median PFS, was 9.5 months in the bemarituzumab group versus 7.4 months in the placebo group (HR: 0.68, 95% CI 0.44–1.04; p = 0.073), while median OS was not reached. This finding indicates that the higher the percentage of tumor cells expressing FGFR2b, the more effective the targeted therapy [150,151]. Based on these positive results, a phase III trial in gastric cancer overexpressing FGFR2b (FORTITUDE-101) and a phase Ib/III trial combining mFOLFOX6 plus nivolumab and bemarituzumab (FORTITUDE-102) are ongoing [152].

4. Role of Liquid Biopsy

Liquid biopsy aims to collect different biological sources such as circulating tumor cells (CTCs), cell-free nucleic acids, exosomes, circulating tumor microemboli (CTM) or tumor-educated platelets (TEPs). The detection of these circulating factors, in terms of presence or concentration, is an emerging tool that can be used in clinical practice, from early diagnosis to the prediction of the best therapeutic option [153]. Among solid tumors, the potential impact of liquid biopsy has been most studied in colorectal cancer, non-small cell lung cancer and breast cancer, but is less studied in GC [154,155,156].

CtDNA and circulating free DNA (cfDNA) have been shown in several studies to help the detection of metastatic disease. CtDNA is composed of DNA fragments originating from tumor cells, while cfDNA is defined as DNA fragments released by cells and it is composed of ctDNA, circulating cell-free mitochondrial DNA and cell-free fetal DNA [157]. Kim et al. demonstrated that mean plasma cfDNA levels were the lowest among healthy subjects and highest in patients with advanced GC [158,159]. The Japanese MONSTAR-SCREEN study performed a serial circulating tumor DNA assay involving 540 patients with advanced solid tumors, of whom 133 had gastrointestinal (GI) tumors (48 with GC). The trial showed that ctDNA levels in gastrointestinal tumors were significantly higher compared to other tumor types [160].

CfDNA and/or ctDNA detection is a more sensitive method for identifying tumor-specific genetic changes than traditional biopsies. For example, the detection rate of FGFR2 amplification was higher with ctDNA analysis despite tissue biopsy because of tumor heterogeneity, improving treatment strategies and efficacy [161]. Furthermore, cfDNA and ctDNA have been shown to be helpful in different solid malignancies for detecting new resistance mechanisms to chemotherapy and targeted therapy. Li J. et al. showed that the amount of cfDNA or ctDNA after therapy predicts patients’ outcome; in particular, during the treatment, in GC patients, the progression of the disease was associated with higher concentrations of plasma cfDNA/ctDNA over time, whereas the concentration of cfDNA/ctDNA remained stable in patients with stable disease [162]. Kim et al. showed that patients with advanced PD-L1-positive GC treated with pembrolizumab achieved an ORR of 83.3%. This was associated with a higher mutational load in ctDNA prior to treatment, while patients with a low mutational load achieved an ORR of only 7.7% [97]. In addition, it was also found that patients with reduced ctDNA six weeks after therapy had extended progression-free survival [97]. A clinical trial investigated the use of ctDNA for the detection of biomarkers of resistance to trastuzumab therapy in 39 patients with advanced GC; the authors showed a consistent correlation of clonal mutations between tumor and blood samples, identifying 32 mutations potentially related to trastuzumab resistance [163]. In summary, in the non-metastatic setting, liquid biopsy can provide biomarkers for early cancer detection, estimate tumor volume and help determine the prognosis. In the metastatic setting, the levels of cfDNA show a positive correlation with the tumor burden [158]. For this reason, the monitoring of cfDNA can help to differentiate between disease progression and radiological pseudo-progression, opening a window for immediate treatment changes in clinical practice [164,165].

CTCs can help in monitoring tumor recurrence and metastatic spread [166]. Significant correlations between CTC counts and patient survival time have been found in many studies. In a prospective trial enrolling 93 patients with respectable GC, patients with CTC levels ≥ 5/7.5 mL detected in postoperative blood samples had significantly inferior disease-free survival (DFS) and OS than those with a smaller number of CTCs; the correlation between high levels of CTCs and poor OS can be explained by CTCs’ ability to aggregate, forming CTM, facilitating cancer metastasis [167]. Further studies demonstrated that the CTC expression of specific molecules can be helpful in understanding the prognosis of patients with GC. For instance, N-cadherin is a marker related to GC relapse and N-cadherin expression on CTCs can be a predictor of a higher risk of recurrence [168]. The detection of HER2-positive CTCs in patients with advanced GCs that are HER2-negative based on tissue biopsy allowed the identification of patients who can benefit from trastuzumab treatment [169] Therefore, the number of HER2-positive CTCs can be a useful tool for monitoring the effectiveness of trastuzumab therapy [170].

Exosomes are small (30–140 nm) membrane-bound extracellular vesicles that are secreted by large multivesicular bodies and are released into the extracellular environment through fusion. Many cell types can release exosomes and they can be detected in blood, urine and cerebrospinal fluid. Non-coding RNAs (ncRNAs) can be packaged into exosomes and they can be related to angiogenesis, formation of the premetastatic niche and metastasis, indicating their role in the initiation and progression of cancer [171]. TEPs are among the newest liquid biopsy components; TEPs help cancer cells grow and escape from the immune system and they can show tumoral fingerprints. Several studies are ongoing to clearly define the role of exosomal ncRNAs and TEPs in clinical practice and to determine their prognostic value [172].

5. Conclusions

Several biomarkers have been explored and are being evaluated in GC with the aim of tailoring treatment and improving the clinical outcomes of GC patients.

Table 3. Current biomarkers in GC. Abbreviations: CT, chemotherapy; FISH, fluorescence in situ hybridization; HER2, human epidermal growth factor Receptor 2; IHC, immunohistochemistry; MSI, microsatellite instability; NGS, next-generation sequencing; NTRK, neurotrophic tyrosine receptor kinases; PD-L1, programmed cell death protein ligand 1.

Biomarker	Assessment	Drug	Setting	Clinical Trial
PD-L1	IHC	Nivolumab (plus standard CT) Pembrolizumab (plus trastuzumab and CT)	First line in PD-L1 CPS ≥ 5 First line in HER2-positive GC with PD-L1 CPS ≥ 1	CheckMate-649 KEYNOTE-811
HER2	IHC/FISH	Trastuzumab (plus standard CT) Trastuzumab deruxtecan	First line Third or later line	ToGA DESTINY-Gastric 0
MSI	IHC, PCR or NGS	Pembrolizumab	First line	KEYNOTE-158
NTRK	RNA-seq, IHC	Larotrectinib Entrectinib	Advanced disease	LOXO-TRK-14001, NAVIGATE STARTRK-1 and STARTRK-2
Claudin 18.2	IHC	Zolbetuximab (plus standard CT) Zolbetuximab plus Capecitabine and Oxaliplatin	First line First line	SPOTLIGHT GLOW

sums up the most important evidence and clinical practice.

To date, only some of these findings are currently validated as predictive factors for guiding targeted treatment in the clinical practice (e.g., CPS score for ICI in combination with standard CT, HER2 for anti-HER2 agents) [13,14,47,74,76]. However, most of them still require further extensive assessment to better define their impact on the efficacy of targeted therapy strategies. The results of ongoing clinical trials are eagerly awaited, as well as further research on other potential predictive factors, to understand the real impact of the biomarkers under study in improving survival in GC, which still has a dismal prognosis despite the standard treatments.