Gene expression profiling reveals six molecular subtypes in ovarian cancer, highlighting distinct prognostic profiles driven by mesenchymal features and reactive stroma.
Key Points
Identify novel molecular subtypes of serous and endometrioid ovarian cancer using gene expression profiling and determine their associations with clinical and pathologic features.
Analyzed microarray gene expression data from 285 serous and endometrioid tumors of the ovary, peritoneum, and fallopian tube using k-means clustering.
Validated molecular signatures using laser capture microdissection, pathology review, immunohistochemistry, and class prediction in an independent validation dataset.
Evaluated patient survival across subtypes using Cox proportional hazards models and semisupervised survival analysis.
Identified six robust molecular subtypes: two representing low-grade or low-malignant potential tumors and four representing higher-grade, advanced-stage malignancies.
Characterized a novel mesenchymal subtype marked by high N-cadherin and P-cadherin expression with low CA125 and MUC1, alongside a poor-prognosis subtype defined by a reactive stroma signature and extensive desmoplasia.
Observed distinct immune cell infiltration patterns across subtypes and replicated subtype-specific prognostic trends in an independent dataset.