PROM1, CXCL8, RUNX1, NAV1 and TP73 genes as independent markers predictive of prognosis or response to treatment in two cohorts of high-grade serous ovarian cancer patients

Abstract

Considering the vast biological diversity and high mortality rate in high-grade ovarian cancers, identification of novel biomarkers, enabling precise diagnosis and effective, less aggravating treatment, is of paramount importance. Based on scientific literature data, we selected 80 cancer-related genes and evaluated their mRNA expression in 70 high-grade serous ovarian cancer (HGSOC) samples by Real-Time qPCR. The results were validated in an independent Northern American cohort of 85 HGSOC patients with publicly available NGS RNA-seq data. Detailed statistical analyses of our cohort with multivariate Cox and logistic regression models considering clinico-pathological data and different TP53 mutation statuses, revealed an altered expression of 49 genes to affect the prognosis and/or treatment response. Next, these genes were investigated in the validation cohort, to confirm the clinical significance of their expression alterations, and to identify genetic variants with an expected high or moderate impact on their products. The expression changes of five genes, PROM1, CXCL8, RUNX1, NAV1, TP73, were found to predict prognosis or response to treatment in both cohorts, depending on the TP53 mutation status. In addition, we revealed novel and confirmed known SNPs in these genes, and showed that SNPs in the PROM1 gene correlated with its elevated expression.

Fig 1. Characteristics and comparison of the Cox regression models (gene: PROM1).

The models allowed for the assessment of the risk of recurrence, depending on either a single independent variable (PROM1 mRNA expression (exp)–the univariate model or three independent variables (exp, FIGO and RT)–the multivariate model. Fig A and B show how the AUC values (and thus also discriminating abilities of each model) change in time for original models (A) and models obtained after a bootstrap-based cross-validation of the original data set (B). The bigger the AUC, the higher the performance of a model. A red dashed line marks the same time point which was used to draw the time-dependent ROC curve (C) for both models. In Fig C, an optimal cut-off point was calculated for the multivariate model based on the Youden index. Sensitivity and specificity for this cut-off point are also provided. In addition, AUC values [%] are listed alongside the 95% CI values, shown in square brackets, if calculable. Fig D depicts the Kaplan-Meier survival curves obtained for the multivariate model at the same time point as in the remaining plots. The risk was classified as either higher (high) or lower (low) than in the cut-off point. The Kaplan-Meier curves are supplemented with the result of the log-rank test, as well. Abbreviations used: DFS–Disease-Free Survival, FIGO–clinical stage; RT–Residual Tumor.

Fig 2. Characteristics and comparison of the Cox regression models (gene: CXCL8).

The models allowed for the assessment of the risk of death, depending on either a single independent variable (CXCL8 mRNA expression (exp)–the univariate model or three independent variables (exp, FIGO and RT)–the multivariate model. Fig A and B show how the AUC values (and thus also discriminating abilities of each model) change in time for original models (A) and models obtained after a bootstrap-based cross-validation of the original data set (B). The bigger the AUC, the higher the performance of a model. A red dashed line marks the same time point which was used to draw the time-dependent ROC curve (C) for both models. In Fig C, an optimal cut-off point was calculated for the multivariate model based on the Youden index. Sensitivity and specificity for this cut-off point are also provided. In addition, AUC values [%] are listed alongside the 95% CI values, shown in square brackets, if calculable. Fig D depicts the Kaplan-Meier survival curves obtained for the multivariate model at the same time point as in the remaining plots. The risk was classified as either higher (high) or lower (low) than in the cut-off point. The Kaplan-Meier curves are supplemented with the result of the log-rank test, as well. Abbreviations used: OS–Overall Survival; RT–Residual Tumor.

Fig 3. Characteristics and comparison of the Cox regression models (gene: RUNX1).

The models allowed for the assessment of the risk of tumor recurrence, depending on either a single independent variable (RUNX1 mRNA expression (exp)–the univariate model or three independent variables (exp, FIGO and RT)–the multivariate model. Fig A and B show how the AUC values (and thus also discriminating abilities of each model) change in time for original models (A) and models obtained after a bootstrap-based cross-validation of the original data set (B). The bigger the AUC, the higher the performance of a model. A red dashed line marks the same time point which was used to draw the time-dependent ROC curve (C) for both models. In Fig C, an optimal cut-off point was calculated for the multivariate model based on the Youden index. Sensitivity and specificity for this cut-off point are also provided. In addition, AUC values [%] are listed alongside the 95% CI values, shown in square brackets, if calculable. Fig D depicts the Kaplan-Meier survival curves obtained for the multivariate model at the same time point as in the remaining plots. The risk was classified as either higher (high) or lower (low) than in the cut-off point. The Kaplan-Meier curves are supplemented with the result of the log-rank test, as well. Abbreviations used: DFS–Disease-Free Survival, RT–Residual Tumor.

Fig 4. Characteristics and comparison of the logistic regression models (genes: NAV1, TP73).

The models allowed for the assessment of platinum sensitivity (PS), depending on either a single independent variable (mRNA expression (exp)–the univariate model or three independent variables (exp, FIGO and RT)–the multivariate model. In Fig A, ROC curves for both models for the NAV1 gene are presented. An optimal cut-off point was calculated for the multivariate model based on the Youden index. Sensitivity and specificity for this cut-off point are also provided. In addition, AUC values [%] are listed alongside the 95% CI values, shown in square brackets, if calculable. Fig B compares discriminating capabilities of both the univariate and multivariate models for the NAV1 gene. Fig C and D depict the results of the same analyses as Fig A and B but for the TP73 gene. Abbreviations used: PS–Platinum Sensitivity; FIGO–clinical stage; RT–Residual Tumor.

Fig 5. PROM1 gene analysis.

The PROM1 gene alterations (denoted by “1”) and overexpression were found to be positively correlated (A). In Fig B, the chr4:g.15980539T>A (ENST00000447510.7:c.2374-2A>T) genetic alteration is shown, being a novel, splice acceptor variant, leading to the formation of an abnormal mRNA transcript (marked with red arrows). Since the PROM1 gene is encoded by the minus DNA strand, the reference sequence of this strand is displayed.