Gene signature associated with chemotherapy response in serous OvCa.

<p>(A) Identification of gene signature differentially expressed in chemoresistant and chemosensitive patients. (B) The selective genes (n = 227) distinguish the chemoresistant patients from the chemosensitive patients. (C) A predictive model on the basis of the gene signature reveals an accuracy of approximately 87.9% in correctly classifying chemoresistant and chemosensitive tumors (n = 232; green square = chemosensitive, blue triangle = chemoresistant). (D) An receiver operating characteristic (ROC) curve illustrates the predictive performance, with a sensitivity of approximately 95.2% and specificity of approximately 70% at the predictive score cutoff of approximately −0.16 that serves as a threshold for patient stratification in the TCGA data set. AUC: area under curve. (E) Pathway analysis shows that the gene signature is enriched in the morphologic function at cellular, tissue, and tumor levels. The dotted line denotes the cutoff for significance (<i>P</i> = 0.05). The shaded bars show the ratio of genes enriched in each function to the 227 genes.