The Infiltration Landscape of Immune cells in tumor microenvironment (TME) of Cervical cancer tissues
Overall, transcriptome data for infiltration of immune cells in 964 cervical cancer tissues was downloaded from TCGA (cervixuteri) and GEO (GSE30760) databases. The Fragments Per Kilobase Million (FPKM) expression spectrum of the cervical cancer tissues in the TCGA was transformed into TPMs format. The abundance of immune cells in cervical cancer tissues was quantified using CIBERSORT and ESTIMATE algorithms (Table 1). Based on the infiltration profile of immune cells, the cervical cancer patients were classified into three groups.
Long rank test revealed significant difference in survival among patients in different ICI subtypes (Figs. 1C; p = 0.028). We compared the composition of immune cell subtypes in tumor tissues to further reveal intrinsic biological differences between different ICI clinical phenotypes (Figs. 1D). ICI cluster A patients had a good prognosis and a survival time of 7.56 years. Patients in this group exhibited high infiltration of naive memory resting CD4 + T cells, resting dendritic cells, macrophages M1, monocytes, activated natural killer (NK) cells, and resting mast cells in cancer tissues. The survival time of patients in ICI Cluster B was 12.21 years, and the survival time was the longest among the three groups. Cervical cancer tissues of patients in cluster B ICI exhibited higher infiltration of CD8 + T cells, activated memory CD4 + T cells, follicular helper T cells, and resting NK cells. The survival period of patients in cluster C ICI was significantly low (median survival time was 5.57 years). Patients in this group exhibited high infiltration of activated mast cells and M0 macrophages to TME. Heat map of the correlation between ICI group and TME immune cell infiltration and the interaction diagram of immune cells are shown in Fig. 1A and Fig. 1B. The darker the color, the stronger the relationship. Negative correlation is presented by blue line whereas the red line stands for positive correlation. Kruskal-Wallis analyses revealed that PD-L1 immune checkpoint was expressed highest in ICI cluster B tissues, whereas the lowest PD-L1 expression was observed in ICI cluster C tissues. The results can be seen in Fig. 1E.
Gene Expression Among Different Subgroups
We identified 111 DEGs among the ICI clusters (Table 2). The DEGs were divided into four clusters A-D (Figs. 2A and 2B). Genes (78) that positively correlated with better prognosis were classified in to ICI gene type A. The negative correlation was named ICI gene type B (Table 3). The dimensionality of the gene expression profile was appropriated using the Boruta algorithm. The heatmap for the DEGs in different genetic clusters is shown in Fig. 2A). Enrichment analysis of ICI gene type A and ICI gene type B by Gene Ontology (GO) (Fig. 2D, Table 4; Fig. 2E, Table 5). Figure 2B shows the downregulated expression of naive B cells in cluster B genes. Contrarily, there was up-regulated expression of CD8 + T cells, activated memory CD4 + T cells and M1 macrophage in gene cluster C tissues. Kaplan-Meier analysis results indicating gene expression in cluster A and D gene correlated with better prognosis, whereas expression of cluster B and C gene clusters was associated with poor prognosis (log rank test, p = 0.031; Fig. 2C).
Also, expression of clusters B and C genes corresponded with significantly higher immune scores. The expression levels of PD-L1 also varied significantly among the different genomic clusters (Fig. 2F; p < 0.05). Particularly, ICI gene clusters B and C express PD-L at high level, while ICI gene clusters A and D express PD - L 1 at low level. These findings highlight the dual nature of the immune system, which has both positive and inhibitory effects. Overall, gene clusters were linked to specific immune profile and prognostic profile, underlining the reliability of our classification model.
Analysis Of The Ici Score
The cervical Patients were stratified into high or low ICI score groups based on the median ICI score (Table 6). The distribution of patients with the four gene clusters in ICI score is shown in Fig. 3A. Genes strongly associated with immunoactivity included SOX2, CD274, CD8A, CXCL10, TBX2, TLR9, CTLA4, CD28, GZMB, IFNG, CXCL9, FOXP3 and TNF genes. Except SOX2, TBX2 and TLR9, the rest of immune checkpoint and immune activity related genes were overexpressed in individuals in the high ICI score group (Fig. 3B). Gene set enrichment analysis (GSEA) revealed that the B cell receptor, JAK-STAT and T cell receptor signaling pathways were most dysregulated in the high ICI score group (Figs. 3C; Table 7).
Further analyses revealed that high ICI score correlated with better prognosis, while the survival time of patients with low ICI score was only 8.48 years (p = 0.041, Fig. 3D).
Relationship Between Ici Score And Tmb
TMB is one of the greatest clinical factors that influences the prognosis of cancers. In this study, TMB data of cervical cancer was obtained from the TCGA database. We found TMB was directly proportional to the ICI score (Wilcoxon test p = 0.046, Fig. 4A). Based on TMB and immune score, patients were also divided into discrete subgroups. Overall, there was a strong positive correlation between ICI score and TMB (Fig. 4B; R = 0.14; P = 0.017), consistent with the above analysis.
High TMB was also associated with better prognosis of cervical cancer (Fig. 4C; p = 0.042). Analysis shows that ICI score is an independent predictor of cervical cancer prognosis. Under the classification of ICI score, there was a significant difference in survival rate between the low and high TMB subgroups. Here, four subgroups could be recognized; TMB high and ICI high (HH), TMB high and ICI low (HL), TMB low and ICI low (LH) and finally TMB low and ICI low (LL) (Fig. 4D, p = 0.012).
Significant genes in the low ICI group and high ICI group were also investigated (Figs. 4E and 4F). Analysis of TCGA data revealed significant differences in OBSCN, DNAH9, MXRA5 and CUBN gene mutations between low and high ICI score groups (Table 8). Overall, findings of this study lay the foundation for the development and assessment of response to cervical cancer immunotherapies.