Response evaluation criteria of chemotherapy for bladder cancer. Intra-observer and inter-observer variability in measurement of local response with magnetic resonance imaging

Various evaluation criteria have been used in clinical trials and clinical practice in chemotherapy for bladder cancer. Although tumor, nodes and metastasis (TNM) staging system is used for bladder cancer with muscular invasion, the standardized criteria are Response Evaluation Criteria In Solid Tumors (RECIST). To date, the use of RECIST for bladder cancer has not been well assessed. We elucidated the intra- and inter-observer reproducibility of the evaluation criteria with magnetic resonance imaging (MRI) in the present retrospective study. Eligible 41 patients had two courses of the chemotherapy. Three investigators evaluated the MRI twice. Four criteria, RECIST, World Health Organization criteria (WHO), modified RECIST (the shortest diameter perpendicular to RECIST), and T-factor were compared using the kappa statistics. Mean intra-observer agreements of these four criteria were 0.81, 0.67, 0.80, and 0.44, respectively. Median inter-observer agreements were 0.82, 0.78, 0.69, and 0.57, respectively. The proportion of agreement of the response between RECIST and modified RECIST was 78% (Spearman's rho=.800) with higher response in modified RECIST. From our findings, modified RECIST is recommended as the preferred criteria for bladder cancer. However, validation should be performed in the future studies comparing the efficacy and clinical outcomes such as survival and pathological findings. (author)