Nucleus image analysis as a possible prognostic tool in grading breast cancer.

ABSTRACT

The WHO grading system for breast cancer is based on the subjective, poorly reproducible evaluation of two cytological criteria (Mitotic Activity MA and Nuclear Pleiomorphism NP) and one histological criterion (Tubular Differentiation TD). In order to improve the reproducibility of the assessment of tumour differentiation, we have looked for nuclear cytophotometric parameters (densitometry, geometry and texture) that could be measured objectively on smears stained by the Feulgen method. Tumour cell populations from 36 breast cancers were investigated by nuclear image analysis according to NP scores, MA scores, TD scores (ie for each variable, three categories), WHO total scores (3 to 9 or 7 categories) and WHO grades (grade I 13 patients, whereas 13 and 10 for grade II and III respectively). Discrimination between each score or grade could be displayed by the average profiles of the nuclear cytophotometric parameters provided by multivariate analysis. Discrimination between TD scores was based on two parameters of nuclear texture. All these data suggest that WHO grading could be obtained in an objective manner by nuclear image analysis.