A Novel pyroptosis-related signature for predicting prognosis and evaluating tumor immune microenvironment in ovarian cancer

Abstract

Ovarian cancer (OV) is the most fatal gynecological malignant tumor worldwide, with high recurrence rates and great heterogeneity. Pyroptosis is a newly-acknowledged inflammatory form of cell death with an essential role in cancer progression, though studies focusing on prognostic patterns of pyroptosis in OV are still lacking. Our research filtered 106 potential pyroptosis-related genes (PRGs) among the 6406 differentially expressed genes (DEGs) between the 376 TCGA-OV samples and 180 normal controls. Through the LASSO-Cox analysis, the 6-gene prognostic signature, namely CITED2, EXOC6B, MIA2, NRAS, SETBP1, and TRPV46, was finally distinguished. Then, the K-M survival analysis and time-dependent ROC curves demonstrated the promising prognostic value of the 6-gene signature (p-value < 0.0001). Furthermore, based on the signature and corresponding clinical features, we constructed and validated a nomogram model for 1-year, 2-year, and 3-year OV survival, with reliable prognostic values in TCGA-OV (p-value < 0.001) and ICGC-OV cohort (p-value = 0.040). Pathway analysis enriched several critical pathways in cancer, refer to the pyroptosis-related signature, while the m6A analysis indicated greater m6A level in high-risk group. We assessed tumor immune microenvironment through the CIBERSORT algorithm, which demonstrated the upregulation of M1 Macrophages and activated DCs and high expression of key immune checkpoint molecules (CTLA4, PDCD1LG2, and HAVCR2) in high-risk group. Interestingly, the high-risk group exhibited poor sensitivity towards immunotherapy and better sensitivity towards chemotherapies, including Vinblastine, Docetaxel, and Sorafenib. Briefly, the pyroptosis-related signature was a promising tool to predict prognosis and evaluate immune responses, in order to assist decision-making for OV patients in the realm of precision medicine.

Keywords: Ovarian cancer; Prognostic signature; Pyroptosis; Tumor immune microenvironment.

Fig. 1

Identification of differentially expressed pyroptosis-related genes (DE-PRGs) in ovarian cancer (OV). (A) The flowchart of the research. (B) The heatmap of differential gene expression, among which the top 50 up-regulated and the top 50 down-regulated genes were listed. Different colors represent the different trend of gene expression between normal tissues and OV tissues. (C) The volcano plot showed the differentially expressed genes (DEGs) between normal tissues and OV tissues. The up-regulated and down-regulated DEGs were respectively highlighted in red and blue. (D) The Venn plot of the DE-PRGs. (E) The Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the 106 DE-PRGs (up). The hallmark pathway from mSigDB enrichment analysis of the 106 DE-PRGs (bottom). Here, the top 20 clusters were shown, while the size of the circles represents gene ratio and the color scale represents the p-value. (F) The protein-protein interaction (PPI) network plot of the 106 DE-PRGs (left), among which 13 hub genes with significant associations were defined (right)

Fig. 2

Construction of an ovarian cancer (OV) prognostic signature based on the pyroptosis-related genes (PRGs). (A) The λ selection plot of the 10-fold cross-validation for the LASSO tuning parameter selection. (B) The LASSO-Cox analysis for the optimal prognostic PRGs, including MIA2, XRCC2, NRAS, ALPL, TRPV4, RYR1, EXOC6B, SETBP1, CITED2, and IGF2. (C) The forest plot represented the prognostic ability of the ten optimal PRLs, which were analyzed through the Cox Regression algorithm. (D) The heatmap for the relationship among the six prognostic PRGs, namely CITED2, EXOC6B, MIA2, NRAS, SETBP1, and TRPV4. The color scale represented different correlation coefficients (red for negative relationship and blue for positive relationship). (E) The expression distribution of the six prognostic PRGs in normal tissues and OV tissues. (F) The Kaplan-Meier (K-M) survival curves of the six prognostic PRLs.

Fig. 3

Estimation and validation of the prognostic signature based on the six pyroptosis-related genes (PRGs). The distribution of the risk score, survival time (months), and survival status of ovarian cancer (OV) patients in the TCGA training set (A) and the ICGC validation set (B). The scatter diagrams represented the risk score of different OV patients, refer to corresponding survival time and survival status (top and middle). The heatmaps (bottom) showed gene expression of the 6-gene signature between low-risk and high-risk groups. The Kaplan-Meier (K-M) curves for overall survival (OS), classified into the low-risk and high-risk groups of the TCGA training set (C) and the ICGC validation set (D). The ROC analysis of the TCGA training set (E) and the ICGC validation set (F) of OS prediction by the 6-gene signature based on PRGs.

Fig. 4

Construction and validation of the pyroptosis-related 6-gene‑based nomogram. The forest plots for univariate (A) and multivariate (B) Cox Hazard Regression analysis of overall survival (OS), based on the 6-gene signature and clinical characteristics, including age, pathological grade, and clinical FIGO stage. (C) The prognostic nomogram model for 1-year, 3-year, and 5-year OS of ovarian cancer (OV) patients, based on the 6-gene risk score and clinical indicators selected by the Cox Regression analysis. (D) The calibration diagrams of the prognostic nomogram for predicting 1-year, 3-year, and 5-year OS (top, middle, and bottom) among OV patients. (E) The Kaplan-Meier (K-M) curves (left) and Receiver Operating Characteristic (ROC) curves (right) for patients in the TCGA-OV training cohort, classified by the prognostic nomogram score. (F) The K-M curves (left) and ROC curves (right) for patients in the ICGC-OV validation cohort, related to the nomogram score

Fig. 5

Pathway enrichment analysis and immunity analysis for the pyroptosis-related 6-gene signature. (A) The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis for potential genes. (B-D) The Gene Ontology (GO) pathway enrichment analysis for potential genes in terms of the biological process (BP), the cellular component (CC), and the molecular function (MF). The size of circles indicated gene numbers, while the color scale represented -log10(P-value). (E) The violin diagrams represented the expression distribution of the 19 typical N6-methyladenosine (m6A)-associated genes, between low-risk and high-risk groups. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001; ****P-value < 0.0001

Fig. 6

The tumor immune landscape related to the 6-gene signature. (A) The Boxplots showed the composition of 22 immune cells infiltrating in OV samples, which were analyzed through the CIBERSORT algorithm. OV patients were classified into low-risk and high-risk groups by the 6-gene signature. (B) The Violin diagrams indicated the difference in the 22 immune cells infiltration between low-risk and high-risk groups. (C) The heatmaps showed the proportions and relationships of the 22 immune cells among OV patients. (D) Based on the ESTIMATE algorithm, the stromal score, immune score, and ESTIMATE score, which infers the presence of stroma, infiltration of immune cells, and tumor purity, were compared among two risk groups. *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001; ****p-value < 0.0001

Fig. 7

Estimation of the sensitivity to immunotherapy and chemotherapy among OV patients. (A) The boxplots for the distribution of 8 typical immune checkpoints gene expression (including CD274, CTLA4, HAVCR2, LAG3, PDCD1, PDCD1LG2, SIGLEC15, and TIGIT) between the two groups classified by the 6-gene signature. (B) The immunotherapy response prediction of OV patients, refer to the Tumor Immune Dysfunction and Exclusion(TIDE)score. (C) The violin diagrams for the estimated IC50 values distribution for OV patients, in terms of 8 typical chemotherapies, including Bleomycin, Cisplatin, Docetaxel, Gemcitabine, Paclitaxel, Sorafenib, Vinblastine, and Veliparib. The chemotherapy sensitivity analysis was conducted based on the Genomics of Drug Sensitivity in Cancer (GDSC) database. *P-value < 0.05; **P-value < 0.01; ****P-value < 0.0001