Author Interviews, Biomarkers, BMJ, Ovarian Cancer / 03.11.2014

Ben Van Calster PhD Department of Development and Regeneration KU Leuven, Herestraat Leuven, Interview with: Ben Van Calster PhD Department of Development and Regeneration KU Leuven, Herestraat Leuven, Belgium   MedicalResearch: What is the background for this study? What are the main findings? Dr. Van Calster: Ovarian cancer is a very common type of cancer among women, with over 200,000 new cases per year worldwide. It is the most lethal of gynecological malignancies. Research has shown that the referral of patients with ovarian cancer to specialized gynecological oncologists in high volume centers improves survival. However, audits in Europe and the United States also show that only a minority of women with ovarian cancer are appropriately triaged to receive specialist care. In addition, different types of malignancies are not treated in the same way. Hence optimal personalized management of an ovarian tumor hinges on the detailed preoperative diagnosis of its nature. Unfortunately, current prediction models focused on the discrimination between benign and malignant tumors without further specification of the likely type of malignancy. Various prediction models and rules have been developed to help predict whether an ovarian mass is benign or malignant. A recent systematic review meta-analysis has shown that the IOTA model LR2 and simple rules perform better than any other previous test. However none of these tests give anything other than a dichotomous outcome – i.e. cancer or non-cancer. In practice the position is more nuanced. The ADNEX model estimates the likelihood that a tumor is benign, borderline malignant, stage I cancer, stage II-IV cancer, or secondary metastatic cancer. This model is the first that is able to differentiate between benign and these four subtypes of malignancy. To do so, ADNEX uses three clinical predictors (age, serum CA-125 level, and type of center), and six ultrasound characteristics of the tumor (maximum diameter of lesion, proportion of solid tissue, more than 10 cyst locules, number of papillary projections, acoustic shadows, and ascites). The model is based on data from almost 6,000 women recruited at 24 centers in 10 countries. (more…)