Development of a prognostic signature for esophageal cancer based on nine immune related genes

Esophageal cancer (EC) is the eighth commonest cancer worldwide. The National Cancer Institute estimated 16,910 new cases and 15,910 deaths from esophageal cancer in the United States in 2016 [1]. Its incidence has risen by more than six times (1999–2008) [2]. The overall five-year survival of EC and that after esophagectomy are still poor, although great improvements have been made in treatment [3]. Squamous cell carcinoma is the most common histological type of EC [4]. Tobacco, alcohol, and malnutrition are the most associated risk factors in the development of EC [5]. Once diagnosed, EC must be accurately staged prior to the initiation of treatment. TNM (tumor, lymph node, metastasis) is a staging system based on the status of tumor invasion, lymph node, and metastasis [6]. Early-stage EC is usually treated with endoscopic surgery, advanced EC with surgery with or without chemoradiation [7].

Certain specific genes and biomarkers are needed to predict the patient’s therapeutic response and increase their survival [3]. Immune responses is critical in the tumor microenvironment. Tumor cells with genomic alterations can produce new antigens that can be recognized by the immune cells [8]. Expression of IRGs can serve as efficient biomarkers. Previous research have explored the IRGs-based prognostic features in patients with non-squamous non-small cell lung cancer [9] and papillary thyroid carcinoma [10]. However, prognostic models based on IRGs for EC remain to be elucidated.

This study investigated the clinical significance of a prognostic model based on immunogenomics.

Esophageal cancer has a large number of new cases every year, and it has historically been regarded as an uncontrollable disease process. The etiology of esophageal cancer may be multifactorial, but part of it is due to the unique manifestation of this cancer [21]. At present, for the treatment of esophageal cancer, attention has shifted to the development of immunotherapy with novel immune biomarkers [22]. Somatic cells acquire malignancy through genetic alterations. Cancer cells usually evade the recognition of the immune system and develop into clinically meaningful masses [23]. Compared with conventional therapies, cancer immunotherapy shows long-lasting response with fewer adverse reactions [24]. This provides a new option for the treatment of EC.

The prognostic model for EC has been continuously updated [25‐27]. In this study, we identified 247 up-regulated and 56 down-regulated IRGs in EC and screened out survival-related IRGs. Based on these data, we established a prognostic model that divided EC patients into high-risk and low-risk groups. This model showed a good predictive performance (AUC 0.826). The model was also an independent prognostic indicator by multivariate analysis incorporating other clinical factors. KEGG analysis indicated that the main pathway was enriched in cytokine-cytokine receptor interaction. Many biological processes are regulated by cytokines, including cell growth, differentiation, immunity, inflammation, and metabolism [28]. Tumor progression can be promoted by cytokines that affect the tumor microenvironment and directly act on cancer cells [29]. Moreover, cytokines participate in the immune response of cytotoxic T lymphocytes (CTLs) by modulating the differentiation of Th1 and Th2 cells [30]. Kita Y et al. found that STC2 may be involved in lymph node metastasis, making it a potential prognostic marker for patients with EC [31]. Studies also demonstrated that STC2 may play an important role in ESCC tumorigenesis [32]. Abnormal expression of DKK1, which is regulated by DKK1-CKAP4 pathway, predicts the poor prognosis of esophageal squamous cell carcinoma (ESCC) [33]. These results are consistent with our findings. CacyBP regulates cell proliferation, tumorigenesis, differentiation or gene expression [34]. In colon cancer, CacyBP can promote the growth of cancer cells by enhancing the ubiquitin-mediated degradation of p27kip1 [35]. In addition, studies have confirmed that CacyBP level increased in gastric, nasopharyngeal carcinoma, osteogenic sarcoma and melanoma [36, 37].

In our prognostic model, the IRGs showing prognostic values included HSPA6, S100A12, CACYBP, NOS2, DKK1, OSM, STC2, ANGPTL3 and NR2F2. Among the, HSPA6 may be associated with early recurrence of HCC [38]. In ESCC, S100A12 is downregulated at the protein level [39]. In Barrett’s esophagus and related adenocarcinoma, expression of inducible nitric oxide synthase (NOS-2) is increased, and NOS-2 also plays a role in inflammation and epithelial cell growth [40]. OSM has been identified as an inhibitor of tumor cell growth in a variety of cancers, including melanoma, ovarian cancer, and glioblastoma carcinomas [41‐43]. The splice variant of oncostatin M receptor β is overexpressed in human esophageal squamous cell carcinoma [44]. Angiopoietin-like protein 3(ANGPTL3) is indicative of EC prognosis [45]. NR2F2 is involved in the progression of prostate adenocarcinoma [46], and NR2F2 expression is a prognostic factor for breast neoplasms [47]. High expression of NR2F2 in certain gastric and esophageal adenocarcinomas is associated with abnormal expression of cadherin 11, suggesting that the NR2F2-related embryonic pathways in these tumors are reactivated [48]. Proteasome dysregulation is implicated in the development of many types of cancer [49]. The proteasome is involved in cell cycle and transcription, two processes indispensable for cancer development [50]. The spliceosome catalyzes pre-mRNA splicing, a key regulatory step in gene expression [51, 52]. Mutations in genes encoding splice proteins are frequently found in cancer [53]. Small molecule inhibitors that target splice components can be used to create anti-cancer drugs [52]. RNA degradation is a key post-transcriptional regulatory checkpoint to maintain proper functions of organisms. Ribonuclease, a key enzyme responsible for RNA stability, can be used alone for RNA degradation, and can bind to other proteins in the RNA degradation complex [54].

Previous immunotherapies mainly rely on T cells in tumor immune defense [55, 56]. In the present research, the abundance of CD8 T cells and regulatory T cells in the low-risk group increased. T cells are critical in host defense against cancer [57]. The value of CD8 T cells for cancer prognosis has been assessed [58‐62]. In addition, CD8 T cells also play a role in the progression of EC [63, 64].

Tregs are divided into two major subpopulations: thymus-derived Tregs (nTregs) and inducible Tregs (iTregs) [65]. Tregs show significant versatility in their inhibitory mechanisms by releasing cytokines to directly inhibit signal transduction of effector T cells [66]. Tregs can also inhibit and kill B cells by inducing programmed cell death [67]. Indeed, Treg infiltration into the tumor has been negatively correlated to OS in a majority of human solid tumors [68, 69]. However, this correlation is highly variable, depending on the tumor type [70]. In cancers that share a common feature of prominent chronic inflammation, such as colon, breast, bladder or head and neck cancers, intra-tumor accumulations of Treg appear to associate with favorable prognosis and improved OS [71‐73] .This association has been explained by the capability of Treg to suppress “tumor promoting inflammation” (TPI). Moreover, previous study found that regulatory T cells are positively correlated with locoregional control may be through down-regulation of harmful inflammatory reaction, which could favor tumor progression in head and neck squamous cell carcinoma [71]. So it can be explained that why the abundance of regulatory T cells (Treg) in the low-risk group was higher than in the high-risk group in our finding. In high-risk group, we found that macrophages M0, M2 and activated mast cells were also significantly enriched. Tumor-associated macrophages are the most abundant cancer immune cells. Studies have found that the transcription factor forkhead box protein O1 (FOXO1) can promote the polarization of macrophages M0 to M2 and the recruitment of macrophages M2 in ESCC through transcriptional regulation [74]. Macrophage M2 can be transformed into macrophage M1, and can promote the proliferation, migration and ring-forming ability of lymphatic endothelial cells associated with EC [75]. In addition, macrophage M2 can promote the migration and invasion of ESCC cells, enhance the epithelial-mesenchymal transition process, and promote tumor progression, resulting in poor prognosis for ESCC patients [76]. Tissue kallikrein (TK1), which is highly expressed in activated mast cells, can participate in the formation of mitogenic kinin, which can stimulate the proliferation of tumor cells and enhance metastasis by increasing vascular permeability [77]. All these researches above can support the finding of our study.

It is the first time that a prognostic nomogram is developed with nine immune related genes. This nomogram can be routinely applied and is cost-effective in practice, as it does not need whole-genome sequencing for EC patients. When combined with clinical parameters like TNM stage, the nomogram can show a greater prognostic performance.

Although we constructed a novel nine-gene prognostic signature in esophageal cancer, several limitations of this study should also be acknowledged. Firstly, our prognostic signature was only based on the data from TCGA database, which is not validated in other databases or other centers across different populations. The performance of this prognostic signature might be more reliable if validation is performed with independent external data sets with long-term follow up. Secondly, this study only preliminary proposed a prognostic model and the validity of the five-gene signature model needs to be further verified by clinical trials. Our study was designed on the basis of a retrospective analysis and prospective research should be performed to verify the outcomes. Thirdly, the mechanisms underlying the nine immune-related genes in the prognosis prediction of esophageal cancer needed to be investigated through in vitro and in vivo experiments.

We identified the IRGs associated with the prognosis of EC and developed an IRGs-based prognostic signature that stratify EC patients into two subgroups with statistically different survival outcomes.