Abstract
Aim: To evaluate the correlation of various nuclear morphometric variables with conventional staging categories in patients with urothelial cancer of the urinary bladder. Material and Method: Nuclear morphometry using interactive image analysis system was performed on hematoxylin and eosin stained slides of 39 patients with urothelial cancer of the bladder. Thirteen patients were classified as Ta, N₀, M₀, 5 as T1, N₀, M₀, 13 patients had invasive disease (T ≥ 2, N₀, M₀), and 8 had pelvic lymph node involvement or distant metastases. Results: Univariate analysis demonstrated 19 out of 25 morphometric variables to show statistically significant difference between superficial disease (Ta, T1) and invasive disease (T ≥ 2, N+). Standard deviation of heterogeneity, standard deviation of polygonal area and standard deviation of angle were found to be highly significant stage predictors on multivariate analysis (p < 0.0001). Receiver operating characteristics analysis based on this variables revealed that the best cut point of the discriminate score was –0.5 for distinguishing between superficial (Ta, T1) and invasive disease (T ≥ 2, N+). The sensitivity of the test is 90.5% and the specificity is 88.9%. Only 2 variables demonstrate statistically significant difference between locally advanced and metastatic disease in univariate analysis, none of them were significant in the multivariate model. Conclusion: Nuclear morphometry enables objective and quantitative differentiation between superficial and invasive bladder cancer. Comparison between invasive (T ≥ 2) and nodal involvement (N+) and/or metastatic disease revealed no significant differences, suggesting that they might be biologically one entity.