[Reliability of the response rate in oncology: analysis of the causes for variation].

We evaluated the impact of an evaluation committee (EC) on patients' overall response status in a large multicenter trial in oncology. We identified reasons for disagreements between investigators and the EC. The Cancer Renal Cytokine (CRECY) study was a French multicenter trial that tested cytokine therapy in 489 patients with metastatic renal cell carcinoma. Objective response (OR) evaluation included medical imaging and was studied according to international guidelines. A blinded peer review of all responders and litigious cases was performed by an EC. Major disagreements occurred in 43% and minor disagreements in 10.5% of the reviewed files. The number of significant tumor responses was reduced by 23.2% after review by the EC. Reasons for disagreements included errors in tumor measurements, errors in selection of measurable targets, intercurrent diseases, and radiologic technical problems. These reasons for disagreements are analyzed and discussed. We conclude that all therapeutic trial results should be reviewed by peer analysis of all presumed responders by an EC. International guidelines for response evaluation should be updated by including more reliable methods of measurements and definition of minimal imaging procedures.