The reproducibility of cytomorphometrical grading of bladder tumours.

The reproducibility of cytomorphometrical nuclear area measurements on transitional cell carcinoma is studied. The inter-individual consistency is low (r = 0.55, P = 0.0005) when nuclei for measurement are chosen at random. When we use a systematic analysis based on cell situation and cytological features of malignancy to select the 50 nuclei for measurement per slide, the consistency between two technicians appears to be significantly higher (r = 0.90, P = 0.015). Not only interindividual consistency increases using the selection, but the correlation with histological tumour grade also improves significantly for both technicians. The results show that cytomorphometry is a method of grading bladder carcinoma when an accurate selection of nuclei to measure is used. Cytomorphometrical studies should contain a clear description of the way the nuclei for measurement are chosen.