Development of a relapse-related RiskScore model to predict the drug sensitivity and prognosis for patients with ovarian cancer.

BACKGROUND

Ovarian cancer (OC) is a highly aggressive malignancy in the reproductive system of women, with a high recurrence rate. The present research was designed to establish a relapse-based RiskScore model to assess the drug sensitivity and prognosis for patients with OC.

METHODS

Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases were accessed to obtain relevant sample data. The single-cell atlas of primary and relapse OC was characterized using the "Seurat" package. Differentially expressed genes (DEGs) between primary and relapse samples were identified by FindMarkers function. Subsequently, univariate Cox, least absolute shrinkage and selection operator (LASSO) and stepwise regression analysis were employed to determine independent prognostic genes related to relapse in OC to establish a RiskScore model. Applying "timeROC" package, the predictive performance of RiskScore model was assessed. Drug sensitivity of different risk groups was evaluated using "pRRophetic" package. The effects of relapse-related prognostic genes on OC cells were detected with assays.

RESULTS

The single-cell atlas revealed that compared to primary OC, fibroblasts were reduced but epithelial cells were increased in relapse OC. Five prognostic genes (, , , , and ) independently linked to relapse in OC were identified to construct a RiskScore model, which showed high robustness in the prognostic prediction for OC patients. High-risk group tended to have worse outcomes in terms of different clinical features than the low-risk group. Further, six drugs (Vinorelbine, GW-2580, S-Trityl-L-cysteine, BI-2536, CP466722, NSC-87877) were found to be correlated with the RiskScore. While the high-risk group had higher IC values to these drugs, the low-risk group was more sensitive to the six drugs. In addition, silencing markedly inhibited the invasion and migration of OC cells.

CONCLUSION

This study established a relapse-related RiskScore model based on five prognostic genes (, , , , and ), offering novel insights into the recurrence mechanisms in OC and contributing to the development of individualized treatment strategies.