Mechanisms and biomarkers of immune-related adverse events in gastric cancer

In recent years, immune checkpoint inhibitors (ICIs) have achieved satisfactory results in various tumor types, greatly changing the treatment strategy of tumors and bringing more benefits to patients’ survival [1‐3]. However, the survival benefit of cancer patients depends not only on the efficacy of ICI, but also on the occurrence of related adverse events caused by ICI, namely immune-related adverse events (irAE) [4, 5]. IrAEs are very common in ICI treatment and can occur in multiple tissues and organs [6, 7]. Among them, the most commonly affected organs for grade 3 or above irAE are the digestive system, endocrine system, and lungs, while others include the nervous system, kidneys, liver, and heart [8, 9]. Kawazoe et al. [10] found in a study evaluating the safety and efficacy of pembrolizumab combined with S-1 plus oxaliplatin as first-line treatment for advanced gastric/gastroesophageal junction cancer that the incidence of grade 3 or above irAE was 57.4%. The more common irAEs were thrombocytopenia (14.8%), neutropenia (13.0%), colitis (5.6%), and adrenal insufficiency (5.6%). Other irAEs included pneumonia, type 1 diabetes, and peripheral neuropathy. For low-grade irAEs that occur during ICI treatment, corresponding routine treatments (such as steroid hormones and immunosuppressants) can be given to relieve clinical symptoms, while high-grade irAEs may consider suspending or terminating immunotherapy [11, 12]. Previous hypotheses suggested that the mechanisms behind irAEs include overactivation of the immune system, excessive release of inflammatory cytokines, elevated levels of pre-existing autoantibodies, and the presence of shared antigens between tumors and normal tissues [9, 13, 14] (Fig. 1). However, given the relatively hidden clinical features of irAE, the immature mechanism of occurrence and lack of identification of sensitive markers that may occur irAE, which makes early prediction of patients susceptible to irAE particularly difficult [9]. Thus, understanding irAE and developing predictive biomarkers for their occurrence are crucial for achieving the maximum benefit–risk ratio for ICI-treated patients. Considering the unique clinical value, convenience and accuracy of biomarkers, we list the currently known predictive biomarkers for gastric cancer irAE in this review, providing targeted insights into irAEs as a reference for future studies.

Currently, many biomarkers are used in clinical practice, including PD-L1 clinical prediction score (CPS), microsatellite instability (MSI) status, and other recognized markers such as tumor mutation burden (TMB) and gene expression score (GEP) [15‐18]. However, although the above markers have certain predictive value for ICIs, their predictive ability is still not ideal, and there are contradictions in some studies. Therefore, finding reliable biomarkers is crucial for the initial identification of patients with tumors that may benefit from ICIs treatment.

Studies have shown that tissue-resident memory T cells (TRM) and tertiary lymphoid structures (TLS) are associated with good prognosis [19, 20]. Mori et al. [19] discovered through immunohistochemical detection CD103 + T cells and evaluated the relationship between CD103 + T cells and TLS that CD103 + T cells were located around TLSs, and patients with high CD103 showed rich TLS, and patients with high CD103 cells and rich TLSs had better prognosis. In addition, CD103 + CD8 + cells in GC showed better prognosis and were associated with TLS. For patients receiving ICI treatment, high CD103 and TLSs enrichment showed excellent anti-tumor immune response. Moreover, CD103 + CD8 + T cells in GC express higher levels of PD-1, granzyme B, and interferon-γ than CD103—CD8 + T cells. In a study on the distribution and proportion of TLSs and TRMs in lung adenocarcinoma, Zhao et al. [20] showed that the proportion of TRMs within TLSs was significantly higher than that outside TLSs, and was positively correlated with patient survival. In addition, CD103 + TRMs were closely related to TLS maturity, and higher maturity TLSs showed higher proportions of CD4 + CD103 + TRMs and CD8 + CD103 + TRMs. Subsequently, Nose et al. [21] also showed that in patients with ICI treatment, those with high levels of CD103 in peripheral blood CD8 + T cells had significantly higher progression-free survival than the low-level group, and the CD103 + CD8 + T cell population was mainly composed of central memory T cells, showing high Ki-67 expression and a small amount of cytotoxic particles.

Additionally, based on the whole and single-cell RNA-seq data of tumor-infiltrating immune cells, Yang et al. [22] found that tumor-infiltrating PD-1^hiCD8 + T cells could serve as effective biomarker for ICI treatment response in multiple cancers, including GC. In clinical samples and animal models, they found that high-score PD-1^hiCD8 + T cell subsets have better therapeutic response and longer survival time. Moreover, tumor-infiltrating PD-1^hiCD8 + T cells showed better predictive performance when combined with tumor mutation burden (TMB), which can be used as an effective supplementary biomarker for TMB. Zhang et al. [23] established an EV-score based on four plasma EV proteins (ARG1/CD3/PD-L1/PD-L2) to predict immunotherapy results and dynamically monitor disease progression. They believe that high EV-score reflects stronger anti-tumor immune microenvironment characteristics, manifested as more activated CD8 + T/NK cells, higher Th1/Th2 ratio and higher IFN-γ/perforin/granzyme expression in peripheral blood. Among GC patients, those with an EV-score ≥ 1 can benefit more from ICIs, while those with EV-score < 1 could potentially benefit more from ICIs in combination with HER2 treatment.

According to reports, the TGFβ signaling plays a key role in cancer progression by forming the tumor structures and inhibiting the anti-tumor activity of immune cells [24]. Increasing evidence suggests that TGF-β is involved in regulating the composition and behavior of immune components in the tumor microenvironment (TME), thereby inducing tumor immune escape, especially ICI [25‐27]. Related studies have confirmed that overactive TGF-β signaling is associated with ICI resistance, and the synergistic effect of TGF-β blockers and ICI can significantly reduce tumor immune tolerance [28‐30]. For example, TGF-β/PD-L1 bispecific antibodies such as YM101, BiTP, and M7824 have shown strong anti-tumor activity in preclinical and clinical models. All of them showed high binding affinity to TGF-β/PD-L1 dual targets, and had better anti-tumor activity than single anti-PD-L1 or anti-TGF-β treatment [29‐31]. Mechanistically, YM101 exerts anti-tumor effects by increasing the number of tumor-infiltrating lymphocytes and dendritic cells, increasing the proportion of M1/M2, and enhancing the production of cytokines in T cells [29]. BiTP demonstrates potent anti-tumor effects by reducing collagen deposition, enhancing CD8 + T cell penetration in the tumor microenvironment and increasing tumor T lymphocyte infiltration [30]. M7824 can activate dual anti-immunosuppressive functions through TME, induce anti-tumor activity by the innate and adaptive immune system, and block TGF-β1-induced tumor interstitialization and PD-L1-dependent immunosuppression potential [31‐33]. In addition, TGFβ2 is highly expressed and is associated with the expression of genes drive epithelial–mesenchymal transition (EMT) in GC [34]. Meanwhile, TGFβ2 was associated with high levels of multiple immune cell infiltration and cytokine expression in GC microenvironment. The expression of PD-1, PD-L1 and CTLA-4 was significantly higher in tissues with high TGFβ2 expression, and the reactivity of immune checkpoint blockade (ICB) was significantly enhanced [35].

However, there is basically no specific signaling pathway in the process of anti-tumor immunity, such as Notch signaling pathway; PI3K–AKT signaling pathway; hedgehog signaling pathway; NF-κB signaling pathway; JAK–STAT signaling pathway; Wnt/β-catenin signaling pathway, etc [17, 36‐40]. Almost all biological processes of cell proliferation and transformation involve these same signaling pathways. For example, Notch signaling is associated with anti-tumor immunity/immunotherapy [41, 42]. Co-mutations of NOTCH1-3 and homologous repair genes are associated with lasting clinical benefits [43]. Recent studies by Long et al. [44] have shown that NOTCH4 mutations have better clinical benefits in patients with gastric cancer, and NOTCH4 mutations are significantly associated with enhanced immunogenicity, including TMB, co-stimulatory molecule expression and activation of antigen processing mechanisms. In addition, Wnt/catenin pathway is also involved in tumor immunotherapy [45]. Blocking the Wnt/catenin pathway can increase the sensitivity of gastric cancer cells to PD-1 antibody.

Studies show that inflammatory biomarkers have a potential prognostic effect on ICI treatment in cancer patients [46, 47]. The GIPI nomogram established by Formica et al. [48] based on NLR, CRP and ALB showed a significant prognostic value for mGOJ/GC at the metastatic gastroesophageal junction receiving ICI. Similarly, Chen et al. [49] controlled nutritional status (CONUT) score based on total lymphocyte count (TL), total cholesterol level (T-CHOL) and serum albumin (ALB), providing a useful immunological prognostic biomarkers for cancer patients. Patients with high CONUT scores were associated with shorter PFS and OS of ICI or chemotherapy. In addition, patients with high CONUT score had lower PFS and OS in the case of PD-1/PDL1 positive expression.

With the shift towards individualized and precise therapy in the current treatment modes, circulating tumor DNA (ctDNA) has become a crucial biomarker for evaluating the therapeutic effect of ICI before or after treatment of solid tumors [50‐52]. Compared to traditional marker screening, ctDNA has the following unique advantages [53, 54]. First, ctDNA detection is less invasive and only requires a minimal peripheral blood sample instead of a biopsy. Secondly, ctDNA mainly consists of genomic DNA fragments released from cell apoptosis, necrosis or active secretion. Thus, it reflects more comprehensive tumor information compared to tissue biopsy. Besides, tissue biopsy can only be further detected after tumor progression, while ctDNA can be monitored in real time. Kim et al. [55] evaluated serum ctDNA levels in ICI-treated metastatic GC and showed that decreased ctDNA was associated with improved prognosis.

Recently, research on ICI-induced irAE has also found a correlation between the occurrence of irAE and good treatment outcomes [56, 57]. Two hypotheses may explain this phenomenon [18, 58, 59]. One is that the immune response triggered in unrelated areas to the tumor may be non-specific immune or inflammatory, and the other may be the response to antigens that cross-react with tumor-associated antigens. Therefore, irAEs are considered potential clinical biomarker for predicting ICI response. Known gastric cancer ICIs related biomarkers are shown in Table 1.

As mentioned earlier, multiple studies have shown that the occurrence of irAE is associated with better clinical outcomes in ICI-treated various cancers [82, 83]. Hussaini et al. [84] conducted a study on the correlation between the incidence of irAE after using ICIs and the clinical prognosis of various solid tumors. They found that the occurrence of irAEs was positively correlated with ORR, PFS and OS, but not with tumor location and ICI type. Additionally, grade 3 or higher irAEs show a better ORR but worse OS. Similarly, Xu et al. [85] based on the correlation between irAE and ICI efficacy in patients with hepatocellular carcinoma (HCC) also showed that the occurrence of irAE was associated with clinical benefit. However, they believe that low-grade irAE is a better predictive marker for ICI treatment in HCC. Given that Hussaini et al. research was based on the evaluation of irAE and ICI treatment efficacy for multiple malignant tumors, while Xu et al. research object was only liver cancer patients, which may lead to differences in this result. Additionally, Xu et al.also found that patients with diarrhea/colitis, hyperthyroidism/hypothyroidism or rash had better prognosis [85]. Similarly, a previous meta-analysis also showed that the occurrence of endocrine, skin and gastrointestinal irAE was significantly associated with good prognosis in ICIs-treated patients, while other irAEs were not [86]. Considering the consistency of the results from multiple studies, the occurrence of irAE is significantly associated with excellent clinical prognosis. Therefore, can it be considered that the sensitive biomarkers predicting the occurrence of irAEs may also predict the therapeutic effect of ICI to some extent, and the biomarkers of the two are somewhat similar?

IrAEs are mainly caused by non-specific activation of the immune system and can manifest as specific and non-specific symptoms [9]. Previous hypotheses mainly focused on the excessive release of inflammatory cytokines, the overactivation of the immune system, the amplification of pre-existing abnormal antibodies, and the existence of shared antigens between tumors and normal tissues [13, 14]. Non-specific symptoms typically include fever, cough and fatigue, while specific symptoms usually involve adverse reactions of specific organs or tissues to ICI treatment in different tissues or organs, such as colitis, hyperthyroidism and interstitial pneumonia [87]. However, most common clinical ICB strategies are a combination of immunotherapy with chemotherapy or targeted therapy, making it difficult to determine whether adverse events are caused by immunotherapy alone [9, 87]. Therefore, it is urgent to develop easy-to-detect biomarkers to identify irAEs. Various candidate biomarkers associated with irAEs have been reported, including gene expression profiles, C-reactive protein, human leukocyte antigen (HLA) and gut microbiome, which can predict irAE [88‐93]. However, current irAE biomarkers for gastric cancer are relatively few and lack a comprehensive summary. Known irAE biomarkers are shown in Table 2.

Although the clinical application of ICIs against CTLA-4 and PD-1 has shown significant anti-tumor response and improved the treatment of various cancers, irAE often occurs [4, 5]. IrAE is considered to be impaired self-tolerance caused by loss of T cell suppression, involving all organ systems. Among them, skin, gastrointestinal tract, liver, endocrine system and lungs being the most common [8, 9]. The main treatment for irAE is corticosteroids or other immunosuppressive agents such as infliximab, and most irAE can be controlled through appropriate management, but may be fatal in some cases [132]. Additionally, the clinical features of irAE are relatively hidden, and the imaging findings are not obvious [9]. So early diagnosis and management of irAE is a challenge for doctors (Fig. 2).

IrAE is considered to be related to the toxicity of ICI antibody mechanism. Therefore, while ensuring the therapeutic effect of ICI, the occurrence of irAE should be avoided or reduced as much as possible. To the best of our knowledge, this is the first comprehensive summary of ICI efficacy and irAE biomarkers in gastric cancer patients. Unfortunately, there are currently relatively few studies on biomarkers associated with irAE in gastric cancer, so only a small number of marker studies have been shown. Given that ICI has certain prospects for current cancer treatment, it may be accompanied by more irAE. Therefore, developing better biomarkers is critical for the management of irAE.