A series (n = 322) of breast carcinomas was investigated from 1991 to 1993 using digital image analysis. Nuclear morphometry and DNA content, and AgNORs were evaluated on cell imprints from fresh tissue samples which were further stored frozen (-80-degrees-C). Data were correlated to morphological prognostic factors and immunocytochemical expression of cell markers assessed on frozen sections and evaluated by densitometry after image analysis processing. Nuclear morphometric parameters, DNA nuclear content and AgNORs were independent from the tumor size, histological grades, and the tumor content of immunodetectable pHER-2/neu, Cathepsin D, ER, PR, pS2, and p53. But, DNA index and hyperploidy degree correlated with the mitosis index (p < 0.01) and Ki67 immunostaining (p = 0.003, p < 0.0001) whereas the shape factor and nucleus large diameter correlated with the degree of cell anisocytosis (p < 0.01). Nuclear surface and large diameter were greater in ductal carcinomas (p = 0.028) and in comedocarcinomas (p < 0.001) than in lobular carcinomas. These results suggest that image analysis processing provides accurate data to refine histoprognostic grading and additional parameters to evaluate tumor proliferative activity.

Related Articles