TME is one of the main hallmarks of cancer, so it is important to identify the key druggable factors and pathways in the TME. The response to cancer immunotherapy especially immune checkpoint inhibitors were impacted by tumor immune microenvironment. Wang et al. found that an “infiltrated-excluded” or “cole” tumor immune microenvironment is predictive of poor response and low-dose metformin reprograms the TME in ESCC [17]. Research by Strizova et al. showed that FasR + NK cells, CD4+, and CD8 + T cells infiltrated lymph nodes at the lowest levels and that the FasR + DR3 + CD4 + T cells were enhanced in esophageal cancer [18]. These compartmental proportions correlated with tumor stage and tumor grade suggested new possibilities for personalized immunotherapy for patients.
Therefore, in the current study, we conducted the bioinformatics analysis of TIICs and TME-related genes in ESCC and established contacts with the clinical outcome and prognosis of ESCC patients for potential prognostic biomarker selection. ESTIMATE, as a common method for calculation of immune and stromal scores, can bring scientific evidence to further analysis. We first obtained 81 ESCC samples with clinical data from the TCGA database. Then, using the ESTIMATE algorithm, we calculated immune and stromal scores of these patients. The correlation between these scores and clinical characteristics were also analyzed. As a result, there was a significance between immune and stromal scores and tumor grades. It is suggested that the tumor immune microenvironment in ESCC had a potential influence on tumor differentiation. Besides, identifying the biomarkers related to TME may predict and even improve the prognosis of ESCC patients.
Continuously, 299 TME-related DEGs were obtained between the high and low immune/stromal score groups. GO annotation results showed that these genes were not enriched in immune-related signaling but almost be relative to muscle function and structure, which indicated a new clue for ESCC development. Skeletal muscles contain resident immune cells and there is a cross-talk between muscle and innate immune cells in physiological and pathogenic conditions, including inflammatory myopathies, endotoxemia, or different types of muscle injury/insult [19]. Paracrine/autocrine and contact interactions have been proven to be involved in these pathological events [19]. In addition, innate immune receptors such as toll-like receptors and NOD-like receptors have influences on skeletal muscle metabolism and the muscle cells have the ability to secrete factors affecting the immune system [20]. These findings showed the correlation between immune response and muscle physiological effect, but there was little research followed with interest of tumor genesis. In our study, 299 TME-related DEGs were mainly involved in muscle system process, muscle contraction, and striated muscle cell differentiation in biological process enrichment analysis. And contractile fiber part, contractile fiber, and myofibril which mainly related to muscle structure were most enriched in cellular component. In molecular function enrichment, substrate-specific channel activity, channel activity and passive transmembrane transporter activity were 3 most significant signal in selected DEGs. In the enrichment analysis of KEGG pathways, we found that only six pathways had significantly statistics including neuroactive ligand-receptor interaction, vascular smooth muscle contraction, pancreatic secretion, cGMP-PKG signaling pathway, insulin secretion and staphylococcus aureus infection. These results indicated that the TME-related genes were involved in not only tumor immune microenvironment but also other undiscovered relative signal pathways.
Based on univariate and multivariate Cox regression analyses, three prognostic genes were identified and used to establish a risk model for predicting the prognosis of ESCC patients. The AUC value of 3-year survival was infinitely close to 1, which indicated a strong capability for predicting survival in ESCC patients. Among these three genes, COL9A3 was identified as an independent prognosis factor in ESCC. And its expression was positively correlated with the clinical outcome, that is, the patients with high expression level of COL9A3 has longer survival time than low expression group (p = 0.003).
COL9A3 encodes the major collagen component of hyaline cartilage, which is one of the three alpha chains of type IX collagen. Type IX collagen, a heterotrimeric molecule, was usually found in tissues containing type II collagen, a fibrillar collagen [21]. Mutations in this gene were usually found in the patients with multiple epiphyseal dysplasia type 3. Previous study has proven that the allelic variants in the collagen IX gene-COL9A3 was a genetic risk factor for intervertebral disc disease [22], and two single nucleotide polymorphisms introducing in COL9A3 were linked to an increased risk of lumbar disc disease[23]. In addition, in X-linked adrenoleukodystrophy patients, the combination of methylation levels of SPG20, UNC45A, and COL9A3 and also the expression levels of ID4 and MYRF would be a good marker for distinguishing the discriminating childhood from adult inflammatory phenotypes [24]. As for tumor-related research, COL9A3 was identified as tumor suppressor gene in rectal cancer [25], and it was also significantly associated with the prognosis of triple-negative breast cancer as an independent prognostic signature [26]. Seldom research about the relationship between COL9A3 and ESCC was taken by now.
GFRA2 named Glial cell line-derived neurotrophic factor family receptor alpha 2. In human neuroblastoma cells and tissues, GFRA2 was upregulated. It can promote cell proliferation by interacting with the tumor suppressor PTEN in neuroblastoma [27]. Similarly, a high expression level of GFRA2 leads to PTEN inactivation and then promotes tumor cell growth and chemoresistance in pancreatic cancer [28, 29]. It is suggested that GFRA2 may have the same effects on ESCC, as another tumor type in the digestive system. But the specific relationship between GFRA2 expression and ESCC development deserves further test and verification.
VSIG4 encodes a protein that may be a negative regulator of T-cell response. It broadly expressed in placenta, lung, and 19 other tissues. Byun et al. showed that high VSIG4 expression of cancer tissue was associated with a longer disease-free interval in benign ovarian tumors [30] and hepatitis B virus-related hepatocellular carcinoma [31]. Both Xu et al. and Hu et al. found that VISG4 could be used as a prognostic factor and a potential immunotherapeutic target for glioma [32] and clear cell renal cell carcinoma[33]. Similarly, Waldera-Lupa et al. found that together with other 2 genes, VSIG4 could be a novel biomarker for supporting the diagnosis of primary central nervous system lymphomas [34]. Tumor-associated macrophage is the prominent component of lung cancer stroma and VSIG4 may play a cancer-promoting effect in lung carcinoma development [35]. In summary, specific targeting of VSIG4 may prove to be an efficacious strategy for the treatment of ESCC, but more research should be taken for further investigation.
Plasma cell was derived from small B lymphocytes after their activation and related with some important process in tumor progression. It showed that tumor-associated plasma cell signatures emerged as a significant signal of survival for diverse solid tumors, but its infiltrated levels was associated with poor prognosis of patients both in breast and lung adenocarcinomas [36]. In triple-negative breast cancers, the infiltration level of plasma cells was highly connected with the disease recurrence [37]. With context-dependent immune responses influenced by oncogenic drivers and the presence of inflammation, CD4 + T cells carried complex and important roles within tumor microenvironments[38]. Tumor-associated macrophages are heterogeneous with diverse functions. For example, M1 macrophages inhibit tumor growth, as M2 macrophages promote tumor growth. And their phenotype and functions are regulated by the surrounding micro-environment especially TME. Due to the key roles in tumor progression, cell invasion, and metastasis [39], direct targeting tumor-related macrophages may be a potential therapy strategy for patients. These results indicated that the infiltration level may have potential significance in ESCC. Using spearman rank analysis, we found that the risk score calculated by risk model was negatively correlated with the proportions of plasma cells and positively correlated with the proportions of activated CD4 memory T cells, M1 Macrophages and M2 Macrophages. It is suggested that ESCC patients with high infiltration level in activated CD4 memory T cells, M1 Macrophages and M2 Macrophages need more attention in clinical therapy. In contrast, patients with high infiltration level in plasma cells may have better prognosis and more survival time than other patients.
Finally, GSEA was performed and confirmed the close relationship between the risk scores and immune pathways. As it shown that the pathways significantly enriched in the high-risk group involved immune response and immune system process, suggesting that immunosuppression exists in high-risk ESCC patients and these high-risk patients may have a poor outcome due to un-worked immune response.