TGFβ2 is a prognostic‐related biomarker and correlated with immune infiltrates in gastric cancer

Abstract TGFβ2 is an essential regulator of immune cell functionality, but the mechanisms whereby it drives immune infiltration in gastric cancer remain uncertain. The Oncomine and Tumor Immunoassay Resource (TIMER) databases were used for assessing the expression of TGFβ2, after which TIMER was used to explore the relationship between TGFβ2 and tumour immune infiltration. Finally, we assessed how TGFβ2 expression correlated with the expression of a set of marker genes associated with immune infiltration using TIMER and GEPIA. We determined TGFβ2 expression to be significantly correlated with outcome in multiple types of cancer in the Cancer Genome Atlas (TCGA), with the effect being particularly pronounced in gastric cancer. Furthermore, elevated TGFβ2 expression was found to be significantly correlated with gastric cancer N staging, and with the expression of a variety of immune markers associated with particular immune cell subsets. These results indicate that TGFΒ2 is associated with patient outcome and tumour immune cell infiltration in multiple cancer types. This suggests that TGFβ2 is a key factor which governs immune cell recruitment to gastric cancer tumours, potentially playing a vital role in governing immune cell infiltration and thus representing a valuable prognostic biomarker in gastric cancer patients.


| INTRODUC TI ON
Gastrics cancer (GC) remains among the deadliest forms of cancer, and it is particularly prevalent in East Asia. 1 The poor prognosis of this cancer type is in part attributable to tumour metastasis.2 Immunological mechanisms regulate the development and progression of GC, and as such, many different immunotherapies have been proposed as a means of effectively treating this cancer type. 3 In non-small cell lung cancer, immunotherapies including inhibitors of cytotoxic T lymphocyte-correlated antigen 4 (CTLA4), programmed death-1 (PD-1) and programmed death ligand-1 (PD-L1) have shown great promise. 4 In GC, however, anti-CTLA4 has shown poor efficacy in the clinic,5 and anti-PD-1 and anti-PD-L1 have shown only partial responses in advanced GC and colon cancer patients.6-8 The infiltration of immune cells into tumours is of particular relevance to patient outcome, with infiltration by tumour-associated macrophages (TAMs) and neutrophils being of particular relevance to patient prognosis and tumour chemosensitivity.9 As such, there is a clear need to better clarify the immune phenotype of GC tumours and to better understand how immune cells regulate this type of cancer in order to better identify novel immunotherapy targets in GC.
Transforming growth factor beta (TGF-β) is a cytokine particularly relevant to malignant tumour progression,10-12 with three family members-TGF-β1, TGF-β2 and TGF-β3-playing non-redundant roles in vitro. 13 TGF-β1 and TGF-β2 have been shown to influence stromal and tumour cells in order to regulate tumour progression.14,15 Most cancer cells lose the ability for TGF-β to inhibit growth, thereby overcoming its suppressive activities while simultaneously enhancing its activities which favour tumour growth. 16,17 Indeed, TGF-β1 has been shown to be independently predictive of both tumour stage and poor prognosis. whereas other studies have found that the release of TNF-α, interleukin (IL)-1β, and IFN-γ is elevated in certain cancer types, including in colon cancer upon interaction with lymphocytes.32 The mechanisms whereby TGFβ2 governs tumour progression and immune cell infiltration in GC, however, remain unclear.
Herein, we conducted a comprehensive assessment of the relationship between TGFβ2 and patient prognosis using databases including Oncomine, PrognoScan and Kaplan-Meier plotter. We further investigated the link between TGFΒ2 and immune cell infiltration of tumours using the Tumor Immunoassay Resource (TIMER).
Our results offer novel insights into the functional role of TGFβ2 in gastric cancer, thereby highlighting a potential mechanistic basis whereby TGFβ2 influences immune cell interaction with tumours.

| Oncomine database analysis
The Oncomine database compiled 86,733 samples and 715 gene expression data sets into a single comprehensive database designed to facilitate data mining efforts. 33 We therefore used this database to assess the association between TGFβ2 expression and prognostic outcome in various tumour types (https://www.oncom ine.org/resou rce/login.html).

| PrognoScan database analysis
The PrognoScan database is designed to facilitate meta-analyses of gene prognostic value by comparing the relationship between gene expression and relevant outcome including overall survival (OS) in a wide range of published cancer microarray data sets. 34 We therefore used this database to assess the relationship between TGFβ2 expression and patient outcome (http://www.abren.net/Progn oScan /).

| Kaplan-Meier plotter analysis
The Kaplan-Meier plotter offers a means of readily exploring the impact of a wide array of genes on patient survival in 21 different types of cancer, with large sample sizes for the breast (n = 6,234), ovarian (n = 2,190), lung (n = 3,452) and gastric (n = 1,440) cancer cohorts. 35 We therefore used this database to explore the association between TGFβ2 expression and outcome in patients with gastric, breast, ovarian and lung cancer, analysing the impact of both clinicopathological factors and TGFβ2 on patient outcome in gastric cancer patients (http://kmplot.com/analy sis/).

| TIMER database analysis
TIMER (https://cistr ome.shiny apps.io/timer /) is a database designed for analysing immune cell infiltrates in multiple cancers. This database employs pathological examination-validated statistical methodology in order to estimate tumour immune infiltration by neutrophils, macrophages, dendritic cells, B cells and CD4/CD8 T cells. 36 We initially employed this database to assess differences in TGFβ2 expression levels in particular tumour types using the TIMER database, and we then explored the association between this TGFβ2 expression and the degree of infiltration by particular immune cell subsets. We further conducted Kaplan-Meier curve analyses to explore differences in patient survival as a function of gene expression or immune cell infiltration. Lastly, we assessed how TGFβ2 expression correlated with the expression of particular immune infiltrating cell subset markers.

| GEPIA database analysis
GEPIA is an online database which facilitates the standardized analysis of RNA-seq data from 9,736 tumour samples and 8,587 normal control samples in the TCGA and GTEx data sets (http://gepia.cance r-pku.cn/index.html). 37 We therefore employed this database to assess the link between TGFβ2 expression and patient prognosis in multiple tumour types, and we further assessed the link between TGFβ2 expression and the expression of particular markers associated with immune cell infiltration of tumours.

| Statistical analysis
The PrognoScan, Kaplan-Meier plotter, TIMER and GEPIA databases were used for generating survival plots in respective analyse, with data including either HR and P-values or P-values derived from a log-rank test. Data from the Oncomine database are presented with information regarding ranking, fold-change and P-values.
Spearman's correlation analyses were used to gauge the degree of correlation between particular variables, with the following r values

| Assessment of TGFβ2 expression in different cancer and normal tissues
We first assessed the expression of TGFβ2 in multiple tumour and normal tissue types using the Oncomine database, revealing that expression of this gene was elevated relative to normal tissue controls for brain, breast, colorectal, oesophageal, rectal, gastric, head and neck, liver, renal and pancreatic cancers. We also found that relative to normal tissue controls, TGFβ2 expression was lower in brain, breast, renal, lung and prostate cancer tissues ( Figure 1A).
Detailed findings in particular tumour types are compiled in Table   S1. We further used the TCGA and TIMER databases to assess how TGFβ2 expression differs in particular tumour types. We found that the expression of TGFβ2 was significantly elevated relative to nor-

| The association between TGFβ2 expression and cancer patient prognosis
We next explored the link between the expression of TGFβ2 and cancer patient outcome using the PrognoScan database (Tables S2-S5).
We found that multiple cancer types exhibited a significant association between patient prognosis and TGFβ2 expression including breast, lung, blood, ovarian, prostate, brain and colon cancer ( We further used the GEPIA database to assess how TGFΒ2 expression relates to patient prognosis, analysing 33 TCGA cancer types and revealing that TGFβ2 expression correlated both with OS and DFS in ACC, LGG, STAD ( Figure S1). These results thus clearly demonstrate that TGFβ2 expression significantly correlated with poorer outcome in multiple tumour types.

| Elevated TGFβ2 expression is linked to prognosis in gastric cancer patients exhibiting lymphatic metastasis
As we found TGFβ2 expression to be linked with poor gastric cancer patient prognosis, we next explored the underlying mechanisms via using the Kaplan-Meier plotter database to assess the relationship between TGFβ2 expression and patient clinicopathological findings. We found that TGFβ2 expression correlated significantly with OS, DFS and with patient gender, stage, T stage, N stage, M stage, Lauren classification and differentiation, with the exception of stage 1 (Table 1). We further found TGFβ2 expression to correlate with each N stage, which corresponds to the degree of lymph node metastasis in gastric cancer patients. Such lymph node metastasis is the most common type of metastasis in gastric cancer patients and is directly linked with patient prognosis.38 With respect to the relationship between TGFβ2 and DFS in gastric cancer, N stage exhibited the highest HR (HR = 4.22 (1.56-11.44, P = .0020), suggesting that TGFβ2 expression has the potential to influence gastric cancer patient prognosis via influencing lymph node metastasis in these individuals.

| TGFβ2 expression correlated with immune cell infiltration in gastric cancer
In cancer patients, survival and lymph node metastasis are independently predicted by the frequency of lymphocytes infiltrating into the tumour.39-41 As such, we next explored the relationship between TGFβ2 expression and the degree of immune cell infiltration into 39 tumour types using the TIMER database ( Figure S2).
We found that there was a significant correlation between TGFβ2 expression and the tumour purity in 24 cancer types, and between  Figure 3B). We further generated Kaplan-Meier plots using the TIMER database in order to explore the relationship between immune cell infiltration and TGFβ2 expression in MESO and STAD. We found macrophage infiltration (P = .004) and TGFβ2 expression (P < .001) to significantly correlate with STAD prognosis (Figure 3C), whereas no significant correlation between prognosis and immune cell infiltration (P = .004) or TGFβ2 expression (P < .001) was observed in MESO ( Figure 3D). This suggests that TGFβ2 plays a strong role in regulating immune cell infiltration in gastric cancer, with a particularly strong effect on macrophage infiltration.   Table 2). In contrast, TGFβ2 expression correlated with just 10 of these markers in MESO (Table 2). TGFβ2 expression was correlated with that of the majority of monocyte, TAM, M1 and M2 macrophage markers in STAD (Table 2). In particular, it was sig-   In this report, we assessed the expression of TGFβ2 as it related to the prognosis of 33 different types of cancers using the independent Oncomie and GEPIA databases, revealing clear differences between tumour and normal tissue expression of TGFβ2 in many cancers.

| D ISCUSS I ON
Oncomine data revealed elevated TGFβ2 levels in brain, breast, colorectal, oesophageal, gastric, head and neck, renal, liver, pancreatic and lymphoma cancers relative to normal tissue, whereas in certain data sets TGFβ2 levels were lower in brain, breast, kidney, lung and prostate cancer ( Figure 1A). TCGA data set analysis indicated that there was elevated TGFβ2 expression in CHOL, COAD, LIHC, STAD and thyroid THCA, whereas expression was decreased in BLCA, BRCA, KICH, KIRP, KIRC, LUAD, LUSC, PRAD and UCEC relative to adjacent controls ( Figure 1B). Altered TGFβ2 expression in a range of different cancers may be due to the different means of data collection in different studies, or it may relate to differences in the underlying biological mechanisms. Across these databases, we consistently observed a correlation between elevated TGFβ2 expression and a poor GC prognosis.
In the TCGA database, elevated TGFβ2 levels were correlated with a  An additional key finding in this study is that the expression of  Together, these results highlight the ability of TGFβ2 to potentially regulate immune cell recruitment and activation in STAD.
In summary, TGFβ2 may be an important regulator of immune cell infiltration and a valuable prognostic biomarker in gastric cancer patients.

CO N FLI C T O F I NTE R E S T
The authors declared that they have no competing interests.  Cor, R value of Spearman's correlation; None, correlation without adjustment. Purity, correlation adjusted by purity. *P < .01; **P < .001; ***P < .0001.

AUTH O R S ' CO NTR I B UTI O N S
LH and ZX conceived the project and wrote the manuscript. SW, LY, QZ, YG and YG participated in data analysis. QX participated in discussion and language editing. DH reviewed the manuscript.