Gastric cancer is the fifth most common cancer and the third leading cause of cancer-related deaths worldwide, resulting in more than one million new cases and nearly eighty thousand deaths in 2018 (1). Stomach adenocarcinoma accounts for the majority of all pathological types. Despite substantial advances in cancer treatment in recent decades, the overall survival rate of GC unfortunately remains unsatisfactory (1, 2). To date, the specific mechanisms that cause the onset and progression of GC remain unclear. Therefore, it is vital to investigate the underlying mechanism of GC progression as well as to find novel diagnostic and prognostic biomarkers.

Recently, immunotherapy has been regarded as a revolutionary treatment of malignant neoplasms (3, 4). For instance, personalized cancer vaccines, nanomaterials enhancing effect of T-cell, CAR-engineered T cells and immune checkpoint inhibitors (ICIs) have been greatly developed to overcome difficulties in treating cancer patients (5-7). Particularly, the emergence of ICIs has brought about a revolution in the treatment of multiple cancers. ICIs, including anti-PD1 inhibitors, pembrolizumab (8) and nivolumab (9) and the anti-CTLA4 inhibitor, ipilimumab (10), can considerably improve the survival outcomes in patients with cutaneous melanoma. However, the therapeutic efficacy of ICIs varies among cancer patients, and there is an urgent need to acquire effective biomarkers to guide the use of ICIs in cancer treatment. Fortunately, precision-targeted therapy has been widely recognized as a promising method against cancer due to ongoing discoveries of molecular mechanisms in tumorigenesis. The current view has described cancers as a kind of genomic diseases driven by an accumulation of both germline derived and somatic mutations. TMB is defined as the total of somatic mutations in every million bases (11). Present studies have shown a clear association between TMB and the outcome of immunotherapy for multiple cancer types (12, 13). On the one hand, mutations in driver genes are detrimental to drive tumorigenesis in humans, but on the other a large number of somatic mutations may generate many new antigens, which could be recognized and attacked by immune cells (14-16). In multiple cancers, high TMB had shown a close association with outcomes in patients using ICIs, such as non-small cell lung cancer (NSCLC) (17-19), breast cancer (20, 21), and melanoma (22).

TCGA database, which is benefitting from breakthroughs of high-throughput sequencing, contains plenty of precious bioinformatic sources including TMB available to the public. Bioinformatics analysis is a scientific method that has gained popularity in a wide range of fields, including oncology. Therefore, the present study aims to explore the significance of TMB in GC and construct a prognostic model by analyzing data from TCGA database.

With development of molecular biology in immunotherapy, many promising results have been achieved in the treatment of advanced cancers recently. Immune checkpoint inhibitors are applied in the treatment for chemo-refractory gastric cancer. A randomized trial showed antiPD1 monoclonal antibody nivolumab improved overall survival of patients with GC (25). Unfortunately, not all patients will respond well to such an immunotherapy. Therefore, it is of great clinical value to identify biomarkers that help to distinguish patients who could benefit from immunotherapy. TMB is a potential biomarker for predicting the outcome of immunotherapy in a variety of cancers (12, 26, 27). Somatic mutations are important causes of tumorigenesis, which lead to neoantigens that are recognized and attacked by immune cells (16). TTN, encoding a structural protein in striated muscles, is frequently detected in many tumors, which is associated with an increase in TMB and the objective response to ICIs (28, 29). MUC16, which encodes the cancer antigen CA-125, was shown to be strongly associated with higher TMB and favorable survival outcomes in GC patients (30). Our result illustrated that GC patients with high TMB had a higher survival rate than those with low TMB, which was consistent with the result of Zhao et al. (31), Yu et al. (32) and Wang et al. (33). Besides, higher TMB was associated with late T and N stages. Similarly, patients with high TMB had significantly higher response rates to ICIs and longer survival than those with low TMB (17). In NSCLC, it had been demonstrated that patients with high plasma TMB got significant benefits from the treatment of atezolizumab, particularly the benefit in PFS (34, 35). In addition, we identified 11 TMB-related genes including SCGB3A1, UPK1B, XG, CCL21, CDC6, PLA2G5, LAMP5, NLGN4Y, NPR3, CPA3, PPP1R1B and constructed a prognostic model showing poor prognosis in patients with high-risk score. Notably, the AUCs of ROC analysis for the prognostic model at 1, 3, and 5-year were 0.69, 0.71, and 0.70, respectively. SCGB3A1 (HIN1), secretoglobin 3A1 (a small secreted protein), is a member of the secretoglobin family (36). A previous report suggested that SCGB3A1 expression was positively correlated with the level of B-cell infiltration (37), which was consistent with the results of the present study. Methylation of its promoter had been reported to be associated with poor outcome in patients with ovarian clear cell adenocarcinoma, and expression of the SCGB3A1 gene can increase paclitaxel sensitivity via the Akt pathway (38). UPK1B, a member of the transmembrane four superfamily, was significantly associated with the prognosis and promoted the proliferation, migration and invasion by the Wnt/β-catenin signaling pathway in bladder cancer (39). XG, encoding the XG blood group antigen, was associated with lower OS in patients with Ewing’s sarcoma and played a role in metastasis (40). CCL21 is a chemokine that can affect lymph node metastasis in various cancer types. The expression of CCL21/CCR7 was significantly associated with colorectal liver metastasis (41). CDC6, cell division cycle 6, whose expression was promoted by zinc finger protein 143 (ZNF143) and accelerated hepatocellular carcinoma cell-cycle progression (42). PLA2G5 (phospholipase A2 group V) was reported to be associated with epithelial-mesenchymal transition and the isocitrate dehydrogenase one mutation status in gliomas (43). LAMP5, a lysosomal associated membrane protein, could directly target the oncogenic MLL-fusion protein, whose depletion lead to inhibition in leukemia cell growth in vivo and in vitro (44). NLGN4Y, a type I membrane protein that belonging to the family of neuroligins, was highly expressed in lung adenocarcinoma patients with poor survival (45). NPR3 (natriuretic peptide receptor 3), could inhibit cancer cells growth in osteosarcoma via blocking the PI3K/AKT pathway (46). CPA3 (carboxypeptidase A3), was demonstrated to be involved in the histone hyperacetylation signaling pathway activated during differentiation of prostate epithelial cancer cells (47). PPP1R1B encodes protein phosphatase one regulatory inhibitor subunit 1B and the knockdown of PPP1R1B impaired the ability of lung metastases in pancreatic cancer cells in mice (48).

CIBERSORT, an emerging approach that integrates the deconvolution algorithm and genomic profiles, can calculate relative proportions of tumor-infiltrating immune cells, which performs better than immunohistochemistry-based analysis (24), and has been increasingly used to estimate the infiltration of immune cells due to its favourable performance (49-52). We applied CIBERSORT to access the infiltration level of 22 types of immune cells and found higher immune infiltration levels of memory B cells, CD4⁺ memory activated T cells, M0 and M1 macrophages in the high-TMB group. Li et al. demonstrated that CD4⁺ T cells could help activate M1 macrophages, and the infiltrate levels of CD4⁺ and CD8⁺ T cells were negatively correlated with tumor size in gastric cancer (53). Macrophages are antigen-presenting cells that have the ability to bind specifically to tumor cells and can directly phagocytose and deliver drugs to tumors (54). Many studies have implied that macrophages can exert anti-cancer effects through targeted drug delivery (55, 56). M1 macrophages can inhibit tumor growth, while M2 macrophages can promote tumor growth, so the enhancement of M1 macrophages polarization and inhibition of M2 macrophages polarization are regarded as an effective way to the treatment of tumors. Consistently, our study indicated that the immune infiltration level of M1 macrophages was increased in the high-TMB group, which showed a favorable prognosis. Meanwhile, we found lower infiltration levels of naive B cells, resting CD4⁺ memory T cells, regulatory T cells (Tregs), monocytes and resting mast cells in the high-TMB group. Gu Y et al. reported that the infiltration of B cells could increase lymph node metastases by the production of pathogenic IgG in breast cancer (57). Olkhanud et al. found that breast cancer metastasis was promoted by tumor-evoked regulatory B cells through converting resting CD4⁺ T cells to regulatory T cells (58). Mast cells may play a role in tumor progression by supporting angiogenesis (59). In melanoma, mast cell-derived hypoxia-inducible factor 1 (HIF-1) could exacerbate the growth of tumor. TP53 plays a critical role in the prevention of oncogenesis in several cancer types and has been shown to be involved in the physiological disruption of the M2 macrophages polarization process through the TP53/MDM2/c-MYC axis (60). Another research suggested that the inhibition of MDMX phosphorylation could prevent the reduced expression of P53, which hindered M2 polarization and promoted M1 polarization in vivo. Combining with our findings, the mutation of TP53 might be an important factor contributing to differential immune infiltration of macrophages. In general, we concluded the landscape of TMB in GC, analyzed the differences of TMB in different clinical subgroups and constructed a prediction model. Taken together, our study indicated TMB was closely associated with the prognosis of patients with GC. A novel prognostic model and nomogram were built to predict the prognosis of GC.