RESUMO
Our aim was to design and validate a model of CT findings that predict suboptimal cytoreduction in primary surgery (PS) for Stage III-IV epithelial ovarian cancer (EOC). We performed a retrospective review of preoperative CT scans of patients undergoing PS for EOC in a cancer centre in London, UK, between November 1995 and October 2003 (n = 91). Radiological features predictive of suboptimal cytoreduction were identified and the model tested in a second cohort undergoing PS in Manchester, June 2005 - March 2007 (n = 35). In the London cohort, liver surface disease and infrarenal para-aortic lymph node involvement predicted suboptimal cytoreduction with 80% accuracy. Accuracy of these predictors dropped to 63% when applied to the Manchester cohort. We concluded that CT prediction of suboptimal cytoreduction is unreliable and may not be reproducible. In the absence of favourable data from larger, prospective trials, it should not be used to guide management.
