Nuclear morphology of prostatic carcinoma: comparison of computerized image analysis (CAS 200) versus video...

We compared mean nuclear size and mean nuclear shape (nuclear roundness) on Feulgen-stained smears from 113 cases of prostatic carcinoma analyzed by DynaCELL system at x 100 magnification versus CAS system at x 40 where DNA can be assessed simultaneously. Correlation coefficients were 0.67 for size and -0.07 for shape. A subgroup of cases were compared at x 40 versus x 100 on the CAS system by two observers: The correlation for size ranged 0.47 to 0.49, with correlations for shape ranging from -0.19 to 0.54. In this subgroup at x 100 magnification, the correlation between CAS and DynaCELL was 0.32 for size and 0.07 for shape. At x 100 magnification on the CAS system, intraobserver correlation was 0.58 for size and 0.37 for shape with interobserver correlations of 0.36 for size and -0.05 for shape. At x 40 on the CAS system, intraobserver correlation was 0.42 for size and 0.72 for shape, with interobserver correlations of 0.06 for size and 0.45 for shape. This study shows that in comparison to video planimetry of prostatic cancer nuclei, which has been shown to have a high inter- and intraobserver correlation, the CAS 200 system provided accurate measurements of size yet not shape. Intra- and interobserver correlations were suboptimal with CAS, showing better results at x 100 for size and x 40 for shape. These results in part reflect the relative narrow range of size and shape in prostate cancer in comparison to other tumors, and the use of Feulgen smears where nuclear RNA is not stained or measured by the CAS system, potentially leading to artifactually irregular shapes.