A Neutrophil Extracellular Traps-Related Signature Predicts Clinical Outcomes and Identifies Immune Landscape in Ovarian Cancer

Abstract

Ovarian cancer (OvCa) is the most lethal gynaecology malignancies worldwide. Neutrophil extracellular traps (NETs), net-like protein structures produced by activated neutrophils and DNA-histone complexes, have a central role in tumours, though haven't been fully explored in OvCa. We obtained transcriptome data from TCGA-OvCa database (n = 376) as training, ICGC-OvCa database (n = 111) as validation and GTEx database (n = 180) as controls. Through LASSO-COX Regression analysis, we identified an eight-gene signature among 87 NETs-related genes, which was significantly related to poor prognosis in both TCGA-OvCa and ICGC-OvCa cohorts (Log-rank p-value = 0.0003 and 0.0014). Next, we constructed and validated a prognostic nomogram, consist of NETs-related signature and clinical features (C-index = 0.82). We evaluated 22 typical immune cell infiltration through CIBERSORT analysis, which implied upregulation of memory CD4 + T cells, follicular helper T cells and neutrophils in high-risk group. Additionally, we predicted therapy sensitivity through TIDE algorithm, indicating that high NETs-riskscore exhibited more sensitivity towards Sorafenib and less sensitivity towards immunotherapy. We initially reported that RAC2 upregulation was associated with NETs formation and poor prognosis (p-value < 0.05) through IHC analysis of tissue microarrays (n = 125). Conclusively, NETs-related signature was reliable for OvCa prognosis prediction and therapy assessment. Especially, RAC2 was predominantly related to NETs formation, thus providing hints towards anti-tumour mechanism of NETs in OvCa.

Keywords: RAC2; neutrophil extracellular traps; ovarian cancer; prognostic signature; tumour immune landscape.

FIGURE 1

Identification of differentially expressed NETs‐related genes (NRGs) in OvCa. (A) Heatmap diagram showed the expression profile of ovarian tissues from TCGA‐OvCa cohorts and GTEx normal controls. (B) Volcano plot of differentially expressed genes between OvCa and normal tissues. (C) Venn diagram identified 33 differentially expressed NRGs (DE‐NRGs). (D) Top 15 KEGG pathway enrichment analysis of 33 DE‐NRGs. The circle size represents gene ratio, and circle colour implies p‐value. Top 15 GO pathway enrichment analyses of 33 DE‐NRGs refer to (E) biological process, (F) cellular component and (G) molecular function. (H) The protein–protein interaction network plot of 33 DE‐ARGs (left), while the hub genes were highlighted in red and blue (right).

FIGURE 2

Establishment of NETs‐related prognostic signature in OvCa. (A) λ selection plot of LASSO parameter selection. (B) Forest diagram indicated prognostic ability of eight prognostic NETs‐related genes (NRGs), including RAC2, IL1B, MMP9, LCN2, MMP2, ELN, SELL and FBN1, which were evaluated by the LASSO‐Cox analysis. (C) Expression profile of eight prognostic NRGs in OvCa and control tissues. (D) Sankey diagram of NETs‐related signature and clinical characteristics (age, race, grade, clinical stage and survival status). (E) Kaplan–Meier curves for overall survival rate refer to the eight prognostic NRGs, including ELN, FBN1, IL1B, LCN2, MMP2, MMP9, RAC2 and SELL, respectively.

FIGURE 3

Construction and validation of NETs‐related nomogram for OvCa patients. Forest diagrams for (A) univariate and (B) multivariate Cox Regression analysis of NETs‐related signature riskscore and clinical features, such as age, clinical stage and pathological grade. (C) Nomogram prognostic model for OvCa patients' 1‐, 3‐ and 5‐year OS, according to NETs‐related signature and clinical features. (D) Calibration diagrams of NETs‐related nomogram to predict 1‐year (top), 3‐year (middle) and 5‐year OS (bottom). Kaplan–Meier survival curves for OvCa individuals divided by NETs‐related nomogram score, in the (E) TCGA‐OvCa training and (F) ICGC‐OvCa validation cohorts.

FIGURE 4

Analysis for tumour immune microenvironment landscape and immunotherapy/chemotherapy sensitivity related to the NETs‐related signature. (A) Boxplots represented composition of 22 immune cells infiltrating in OvCa tissues of TCGA‐OvCa patients, who were then stratified into two risk groups referring to the NETs‐related signature. (B) Violin diagrams showed expression profile of 21 typical immune cells infiltration in two risk groups. (C) Boxplots represented gene expression profile of eight typical immune checkpoints, LAG3, CD274, CTLA4, HAVCR2, PDCD1LG2, PDCD1, SIGLEC15 and TIGIT, between two NETs‐related risk groups. (D) Assessment of sensitivity towards immune checkpoint blockade therapies refers to the Tumour Immune Dysfunction and Exclusion (TIDE) score. (E) The violin plots indicated estimated IC50 values among OvCa individuals, as for eight chemotherapies, which were estimated based on the GDSC database. *p‐value < 0.05; **p‐value < 0.01; ***p‐value < 0.001; ****p‐value < 0.0001.

FIGURE 5

Pan‐cancer analysis of the NETs‐related signature. (A) Violin chart represented NETs‐related signature riskscore distribution of tumour tissues in pan‐cancer and controls. (B) The forest chart distinguished prognostic value of the signature via the Cox Regression algorithm in pan‐cancer. (C) The Kaplan–Meier curves of Glioma, Cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), Low‐grade glioma (LGG), Acute Myeloid Leukaemia (LAML), OvCa, Head and Neck squamous cell carcinoma (HNSC) and Pancreatic adenocarcinoma (PAAD) in the TCGA cohorts, which were classified by NETs‐related signature.

FIGURE 6

NETs‐related genes (RAC2 and SELL) could predict prognosis among OvCa individuals. Gene expression of RAC2 (A) and SELL (B) in OvCa tissues, measured through PCR analysis. (C) Sankey plot of NETs‐related signature and clinical characteristics, such as age, tumour size, tumour side, grade, clinical FIGO stage and survival status. Forest diagrams of (D) univariate and (E) multivariate Cox regression analysis for OvCa prognosis, based on clinical features and NRGs (RAC2 and SELL). The Kaplan–Meier survival curves for OvCa recurrence‐free survival (RFS, left) and overall survival (OS, right) were classified by expression of (F) RAC2 and (G) SELL.

FIGURE 7

Aberrant upregulation of RAC2 related to NETs formation and tumour metastasis in ovarian cancer (OvCa). (A) Representative Multiplex immunohistochemical (mIHC) images of NETs, among which CitH3 (red) and MPO (green) were stained. Nuclei were stained with DAPI (blue). (B) OvCa patients with NETs were more likely to suffered recurrence (left) and death (right). (C) The relationship between NETs towards RAC2 (left) and SELL (left) expression among OvCa patients. (D) The immunohistochemistry (IHC) staining images of RAC2 expression in normal controls (left), primary OvCa tissues (middle) and metastatic OvCa tissues (right) were presented. (E) Metastatic OvCa tissues had significantly upregulated RAC2, compared with controls and primary tumour tissues, analysed through IRS score. (F) Representative IHC images staining RAC2 of primary (up) and metastatic OvCa lesions (down) of 5 OvCa patients. The RAC2 expression analysed by the IHC staining was over‐expressed among individuals suffered (G) recurrence or (H) death. (I) The Kaplan–Meier survival curves showed that RAC2 over‐expression was significantly associated with OvCa prognosis.