Comments on and summary of panel on assessment of in vitro and in vivo mutagenicity data with regard to safety evaluation.

Sir Drs Schouten and Kiemeney raise the important issues of comparability of bladder cancer rates between cancer registries and the interpretation of observed trends within a registry if the definition or coding of malignancy changes over time. Thames Cancer Registry takes great care to assign the correct behaviour code to tumours: Table II of our paper showed the Registry's normal coding practice, with the five tumours registered as in situ assigned to behaviour code 2, while the tumour of uncertain malignancy and the urothelial papilloma both have a behaviour code of 1. The small proportion of non-invasive cases in the study reflected the unavailability of case-notes 8-10 years after diagnosis. Although pathologists may now classify as malignant some papillomas which would previously have been classified as benign (Muir, 1987), coding practice here has been unchanged for over 20 years, and bladder cancer incidence trends are interpretable (Coleman et al., 1993). The terms used by pathologists in our area may differ from those used in Eindhoven. We rarely receive reports of papillary transitional carcinoma in situ: only two cases in 5 years were so described. We suspect that the 'low grade papillary shaped tumours' referred to by Schouten and Kiemeney may refer to lesions usually reported to us as urothelial papillomas, which we code to M8120/1.