Quantitative assessment of bladder cancer by nuclear texture analysis using automated high resolution image cytometry

The aim of this study was to investigate the possibility of identifying urothelial neoplasia by nuclear chromatin texture feature analysis using high resolution image cytometry to improve the diagnostic accuracy of cytologic examination in the detection and monitoring of bladder cancer. Touch imprints of transurethral resection material of 56 control group (CG) cases of nonmalignant urothelium and 94 tumor group (TG) cases of bladder cancer were analyzed. The specimen collection was divided randomly into a training set and a test set. Cells were stained specifically for DNA by the Feulgen method. Only diploid cell nuclei were analyzed from both groups. A discriminator comprised of three nuclear texture features was derived from the training set of cases to separate CG from TG cases. This discriminator was then applied to the independent test set. CG cases were separated from TG cases with a sensitivity of 97% and a specificity of 95% on the independent test set of cases. When dividing TG cases into high-risk and low-risk groups, sensitivity in the low-risk group was 93%. None of the high-risk cases was misclassified (sensitivity, 100%). This retrospective investigation demonstrates that by high-resolution image cytometry it is possible to distinguish between urothelial neoplasia and normal urothelium with high reliability when examining diploid cell nuclei only. This method is superior to DNA ploidy analysis using image or flow cytometry and may become clinically relevant as a supplement to conventional cytologic examination. These promising results should be confirmed on cytologic preparations derived from bladder washings or voided urine.