Quantitative histopathology in the prognostic evaluation of patients with transitional cell carcinoma of the urinary bladder.

Abstract

Morphologic grading of malignancy is considered to be of prognostic value in patients with transitional cell carcinomas of the urinary bladder (TCC). This qualitative approach is, however, associated with low reproducibility. Grading of malignancy can be carried out on a reproducible, quantitative scale. A retrospective, prognostic study of 110 patients treated for TCC in clinical Stages Ta-T4 (median follow-up time, 6 years) was performed, evaluating various grading techniques. Unbiased estimates of the volume-weighted mean nuclear volume (nuclear vV), nuclear volume fraction, estimates of nuclear mean profile area (aH(nuc)), nuclear profile density index (NI), and mitotic profile density index (MI) were obtained by stereologic and morphometric techniques. The T-stage and morphologic grade of malignancy were closely cross-correlated (+0.63 < Kendall tau < +0.71, 2P < 6.7 x 10(-16)). The estimation of nuclear vV was highly efficient, with more than 85% of the associated variation attributable to differences between tumors. A positive significant correlation between estimates of nuclear vV and aH(nuc) was detected (r = +0.79), whereas an inverse correlation was documented between nuclear vV and NI (r = -0.63). Estimates of nuclear volume fraction showed no correlation with nuclear vV. Comparisons between categorical and quantitative data revealed the following: a decrease in averaged estimates of NI for tumors in advanced T-stage and malignancy grade (2P < 0.0008); and nuclear vV and aH(nuc) increased on average, in tumors of high T-stage and malignancy grade. Estimates of MI were also positively correlated with the T-stage and the malignancy grade (+0.42 < Kendall tau < +0.49). Single-factor analyses showed prognostic effect of T-stage, grade of malignancy, and, apart from nuclear volume fraction, all quantitative histopathologic variables with regard to overall survival (2P < 0.03). None of the morphometric and stereologic parameters were of prognostic values with regard to recurrence-free survival (2P > 0.26). Multiple hazards regression analysis (Cox models) revealed that clinical stage of disease was the sole independent prognostic variable. Only estimates of nuclear vV added significant independent prognostic prediction with regard to recurrence-free survival in the 48 patients with Ta tumors (2P = 0.03). The results suggested that estimates of nuclear vV are prognostically superior to morphologic grading of malignancy in noninvasive TCC, whereas both morphologically and quantitatively based malignancy grading are without prognostic value in invasive TCC.