Successful External Validation of the Calculator for Ovarian Subtype Prediction in a Clinical Trial Case Series
Stefan Kommoss, Jacobus Pfisterer, Blake Gilks, Friedrich Kommoss, Martin Kobel, Christine Chow, David Huntsman, Steve E Kalloger. University of British Columbia, Vancouver, BC, Canada; AGO Study Group, Wiesbaden, Germany; Vancouver General Hospital, Vancouver, BC, Canada; University of Calgary, Calgary, AB, Canada; British Columbia Cancer Agency, Vancouver, BC, Canada
Background: With the growing certainty of future ovarian cancer treatment being type specific it becomes more and more desirable to reliably type ovarian carcinoma. Recently it has been shown, that the application of a set of nine immunohistochemical markers can objectively support the classification of a tumor into one of the five major types of ovarian cancer by entering immunoscores into the Calculator for Ovarian Subtype Prediction (COSP). It was the aim of this study to externally validate the nine-marker panel's prediction equation for the first time.
Design: A cohort of 408 cases which were derived from an international chemotherapy trial including all histological types was immunohistochemically stained for PR, p53, DKK1, WT1, p16, Vimentin, HNF1B, TTF3 and MDM2. Scoring was performed by one of the authors who was not previously involved in generating the original data for development of the COSP. In this validation experiment, two antibodies (HNF1B & MDM2) had to replaced for of lack of availability or quality reasons.
Results: Of the 350 tumours that were assigned to the serous subtype, the COSP predicted 315 of these to be high-grade serous and 5 to be low-grade serous (table1). The overall concordance rate between the COSP preciction and specialized gynecopathology review is 87.7%.