Correlation of scores with the survival rate
The correlation between the proportion of immune and stromal cells and the survival rate was evaluated by ImmuneScore and StromalScore, the higher score of which indicated a larger amount of immune or stromal components in the TME. ESTIMATEScore denotes the comprehensive proportion of immune and stromal cells. The results showed that in female samples, the StromalScore and ESTIMATEScore had evident positive correlations with the survival rate, while the ImmuneScore was the opposite (Fig. 1A-C). Meanwhile, all three scores had a notably positive relationship with the survival rate in male samples (Fig. 1D-F). The results indicated that both the immune and stromal components are important for the prognosis of GC patients.
Correlation Of Scores With Clinicopathological Staging Characteristics
To determine the influence of the immune and stromal components on the clinicopathological staging characteristics, we evaluated the correlation of the ImmuneScore, StromalScore and ESTIMATEScore with tumour stage, grade, T classification, M classification and N classification. In female samples, the ImmuneScore, StromalScore and ESTIMATEScore had significant positive correlations with tumour grade, especially between grade II and grade III (Fig. 2). In contrast, other clinicopathological staging characteristics showed no evident relationship (Fig. 2). In male samples, the ImmuneScore, StromalScore and ESTIMATEScore exhibited significant correlations with tumour grade, as well as T classification (Fig. 3). In addition, the ESTIMATEScore had an evident relationship with tumour stage. The results implied that the immune and stromal components were related to GC development.
DEGs Shared by ImmuneScore and StromalScore Were Predominantly Enriched for Immune-Related Genes
DGEs were identified by comparison analysis between high- and low-ImmuneScore/StomalScore samples and used for further GO and KEGG analysis. In female samples, a total of 811 DEGs were obtained, among which 741 were upregulated and 70 were downregulated (Fig. 4A-D). In male samples, a total of 513 DEGs were obtained, among which 439 were upregulated and 74 were downregulated (Fig. 5A-D). These DEGs were potential determining factors for TME status.
Gene ontology (GO) enrichment analysis showed that the main functions of DEGs in the female sample were cell-cell adhesion and T cell activation (Fig. 4E), and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis also showed enrichment of the cell adhesion pathway and chemokine signalling pathway (Fig. 4F). Meanwhile, GO results suggested that DEGs in male samples oversaw immune cell proliferation and migration (Fig. 5E). KEGG analysis showed the enrichment of chemokine signalling pathways and chemokine-chemokine receptor interaction pathways. Therefore, the primary functions of DEGs in females and males seemed to map to immune- and chemokine-related activities.
Intersection Analysis of the PPI Network and Univariate Cox Regression
The STRING database and Cytoscape software were applied to construct a PPI network of 811 DEGs in female samples, as well as 513 DEGs in male samples (Fig. 6A, E), and the top 30 genes ordered by the number of adjacent nodes are shown in Fig. 6B and F. Meanwhile, univariate Cox regression analysis was performed to evaluate the significant factors for the survival of GC patients in females and males (Fig. 6C and G). Subsequently, intersection analysis between the top leading genes of the PPI network and Cox regression was performed to screen the key mediators of prognosis. The results showed that the overlapping genes from the above analyses were FCGR2A in female samples and GFRA1 in male samples (Fig. 6D and H).
Correlation of FCGR2A and GFRA1 Expression with Survival and Clinicopathological Staging Characteristics in GC Patients
All female samples were grouped into FCGR2A high- and low-expression groups compared with FCGR2A median expression; similarly, all male samples were grouped into GFRA1 high- and low-expression groups. FCGR2A was evidently highly expressed in female GC samples compared to normal samples (Fig. 7A), and the same result was obtained in pairing analysis between the GC samples and normal samples derived from the same patient (Fig. 7B). GC female patients with high expression of FCGR2A had a significantly lower survival rate than those with low expression of FCGR2A (Fig. 7C). Additionally, FCGR2A was highly expressed in G2 grade compared with G3 grade (Fig. 7D). For other clinicopathological staging characteristics, there were no significant changes (Fig. 7E-H). Whereas GFRA1 was significantly downregulated in male GC samples compared to normal samples (Fig. 7I), the same result was obtained in pairing analysis between the GC samples and normal samples derived from the same patient (Fig. 7J). Male GC patients with low GFRA1 expression had a significantly lower survival rate than those with high GFRA1 expression (Fig. 7K). GFRA1 expression in stage II and stage III had significant changes compared to stage I. Additionally, GFRA1 expression in T3 and T4 had significant changes compared to T1 phase (Fig. 7M-N). For other clinicopathological staging characteristics, there were no significant changes (Fig. 7L, O-P). The above results showed that the expression of FCGR2A in the TME had a positive correlation with the prognosis of female GC patients, while GFRA1 in the TME had a negative correlation with the prognosis of male GC patients.
Gsea Of Fcgr2a And Gfra1
GSEA was performed to analyse the high- and low-expression groups compared with the median levels of FCGR2A and GFRA1 expression, respectively. The genes in the FCGR2A high expression group were mainly enriched in cell adhesion and chemokine signalling pathways (Fig. 8A), while the genes in the FCGR2A low expression group were mainly enriched in metabolic pathways (Fig. 8B). The genes in the GFRA1 high expression pathway were mainly enriched in the calcium signalling pathway and cell adhesion pathway (Fig. 8C). For the GFRA1 low-expression group, the genes were enriched in the cell cycle and DNA replication pathways (Fig. 8D). These results indicated that FCGR2A and GFRA1 might be potential indicators for the status of the TME in females and males, respectively.
Correlation of FCGR2A and GFRA1 with the proportion of TICs
To further determine the correlation between FCGR2A expression and the immune environment, the rate of tumour infiltrating immune cell subsets was analysed using the CIBERSORT algorithm. Twenty-two kinds of immune cells in female samples and 21 kinds of immune cells in male samples were obtained (Fig. 9A and B). The correlation between TICs in female and male samples is shown in Fig. 9C and D. Next, difference and correlation analyses were conducted, the results of which showed that four kinds of TICs were correlated with the expression of FGCR2A. Among these TICs, memory B cells, regulatory T cells and activated dendritic cells were negatively correlated with FCGR2A expression, whereas only M2 macrophages were positively correlated with FCGR2A expression (Fig. 10A-C). Meanwhile, 10 kinds of TICs were correlated with the expression of GFRA1. Among them, naïve B cells, memory B cells, plasma cells, resting CD4 memory T cells, regulatory T cells (Tregs), monocytes and resting mast cells were positively correlated with GFRA1 expression. Four kinds of TICs were negatively correlated with GFRA1 expression, including activated memory CD4 T cells, follicular helper T cells, M0 macrophages and M1 macrophages (Fig. 11A-C). These results further proved that FCGR2A and GFRA1 affected the immune activity of the female and male TMEs, respectively.