Seven Genes Based Novel Signature Predicts Clinical Outcome and Platinum Sensitivity of High Grade IIIc Serous Ovarian Carcinoma

Abstract

Background: As a major subtype of ovarian cancer, high grade FIGO stage IIIc serous ovarian carcinoma (HG3cSOC), has various prognosis due to genetic heterogeneity. Methods: The transcriptome of 401 primary FIGO IIIc serous ovarian samples was screened, seven genes based prognostic model was developed. The prognostic valueof risk score in four different cohorts (TCGA-cohort, Poland-cohort, Japan-cohort and USA-cohort) was validated. The relationship between risk score and other clinical indicators were analyzed. The guide value of risk score for platinum-taxol chemotherapy was also assayed. Tissue microenvironment difference among samples with different risk scores was investigated. Results: High-risk group (N=200, median survival months: 39.6, 95% CI: 35.9-46.3 months) had a significantly worse prognosis than low-risk group (N=201, median survival months: 52.6, 95% CI: 45.2-64.9 months;). The risk score's performance was validated in Japan-cohort (N=90, Poland-cohort (N=48) and USA-cohort (N=84). The risk score is independent from age, primary tumor size, grade and treatment methods and the performance of risk score is uniform in subgroups. Furthermore, the risk score predicted the response of HG3cSOC to platinum-based regimen after surgery, and this finding was further validated in newly collected China-cohort (N=102). Gene Set Enrichment Analysis (GSEA) and tumor infiltration analysis revealed that risk score reflected the immune infiltration and cell-cell interaction status, and the migration function of candidate genes were also verified. Conclusions: The optimized seven genes-based model is a valuable and robust model in predicting the survival of HG3cSOC, and served as a valuable marker for the response to platinum-based chemotherapy.

Keywords: High grade FIGO IIIc serous ovarian carcinoma; chemotherapy; microenvironment.; model; prognosis; transcriptome.

Figure 1

The performance of risk score in TCGA dataset. The overall survival (A) and progression-free survival (B) difference between low-risk and high-risk group. Detailed survival information, risk score and gene expression pattern in TCGA dataset (C), and three-year survival ROC using risk score and other clinical indicators (D).

Figure 2

The performance of risk score in validate dataset. The survival difference between high-risk and low-risk group in Japan-cohort (A), Poland-cohort (B), USA-cohort (C). The detailed risk score, survival time and expression pattern was shown in the bottom panel.

Figure 3

Relationship between risk score and clinical indicators. The relationship between risk score and other clinical indicators were evaluated using student's t-test (A), multivariate cox regression (B), and five-year survival nomogram (C).

Figure 4

The performance of risk score in sub-categories divided by treatment. The survival difference between high-risk and low-risk group was significantly different in almost all subcategories, including residual tumor size (A, left to right, Macro-invisible, 1-10mm, 10-20mm and >20mm) and the tested drugs Avastin (B, left: untreated; right: treated).

Figure 5

Risk score predicted the platinum-based chemotherapy response of HG3cSOC. The risk score difference between high/low risk groups in TCGA (A) group. Overall survival (B) and progression-free survival (C) of taxol-platinum treated samples in low/high-risk groups were compared. The corresponding profile was similar in China-cohort (D-F).

Figure 6

KEGG pathways associated with risk score. The significantly associated KEGG pathways were shown between low-risk and high-risk groups using GSEA. The enrichment score and log 10 transformed p values (A), and cell adhesion molecules cams (B), FC gamma mediated phagocytosis (C), natural killer cell mediated cytotoxicity (D) were noticed. Correlation analyses indicated that the risk score was significantly associated immune cell infiltration (D).

Figure 7

Functional assay of the candidate genes. After knocking down of these genes, The migration ability of serous ovarian carcinoma cells was up-regulated. A showed the image and B is the statistical data.