BACKGROUND: Currently, there are no accurate methods for predicting metastases or time to disease progression for patients with clinically localized prostate cancer after surgery. METHODS: In this report, histologic sections were studied from prostate cancer specimens from 100 men with clinically localized prostate cancer (clinical Stages A1 [9 cases], A2 [24 cases], B1 [27 cases], and B2 [40 cases]; pathologic Stages A1 [9 cases], A2 [22 cases], B [23 cases], C1 [8 cases], and D1 [38 cases]) to determine whether nuclear morphometry--when analyzed with clinical stage, pathologic parameters, and age in a multivariate fashion--would predict time to disease progression. RESULTS: These patients were treated with surgery alone for their clinically localized disease and were observed after surgery until disease progression or death. For each of the 100 specimens, 16 different mathematical descriptors described the shape of 150 nuclei. A series of 17 different statistical measurements were calculated to accurately describe the distribution, extremes, and variability within each descriptor. As univariate predictors, the variance of nuclear roundness, the mean of ellipticity, the Gleason score, age, and clinical stage were statistically significant predictors of disease progression when analyzed with Kaplan-Meier survival curves. A prognostic factor score calculated with multivariate analysis of clinical stage, Gleason score, age, and variance of nuclear roundness separated the patients into three statistically distinct groups and predicted time to progression by the Kaplan-Meier life table and Cox proportional hazards analysis. CONCLUSIONS: This prognostic factor score may aid in stratifying patients into high-risk and low-risk groups for testing adjuvant therapies for prostate cancer.
BACKGROUND: Currently, there are no accurate methods for predicting metastases or time to disease progression for patients with clinically localized prostate cancer after surgery. METHODS: In this report, histologic sections were studied from prostate cancer specimens from 100 men with clinically localized prostate cancer (clinical Stages A1 [9 cases], A2 [24 cases], B1 [27 cases], and B2 [40 cases]; pathologic Stages A1 [9 cases], A2 [22 cases], B [23 cases], C1 [8 cases], and D1 [38 cases]) to determine whether nuclear morphometry--when analyzed with clinical stage, pathologic parameters, and age in a multivariate fashion--would predict time to disease progression. RESULTS: These patients were treated with surgery alone for their clinically localized disease and were observed after surgery until disease progression or death. For each of the 100 specimens, 16 different mathematical descriptors described the shape of 150 nuclei. A series of 17 different statistical measurements were calculated to accurately describe the distribution, extremes, and variability within each descriptor. As univariate predictors, the variance of nuclear roundness, the mean of ellipticity, the Gleason score, age, and clinical stage were statistically significant predictors of disease progression when analyzed with Kaplan-Meier survival curves. A prognostic factor score calculated with multivariate analysis of clinical stage, Gleason score, age, and variance of nuclear roundness separated the patients into three statistically distinct groups and predicted time to progression by the Kaplan-Meier life table and Cox proportional hazards analysis. CONCLUSIONS: This prognostic factor score may aid in stratifying patients into high-risk and low-risk groups for testing adjuvant therapies for prostate cancer.
Authors: Daniel T Keefe; Nicola Schieda; Soufiane El Hallani; Rodney H Breau; Chris Morash; Susan J Robertson; Kien T Mai; Eric C Belanger; Trevor A Flood Journal: Virchows Arch Date: 2015-07-31 Impact factor: 4.064