Analysis of DNA and morphometry in breast carcinoma.

Feulgen-stained imprints and smears from 730 cases of invasive breast cancer were investigated using an image analysis system. From each tumor sample 100 cells were randomly scanned and several DNA and morphometrical parameters evaluated. Their prognostic value for a prediction of distant metastases within 5 years was investigated with the multivariate Cox regression analysis, which was performed for all consecutive cases, as well as for node-negative and node-positive patients separately. The multivariate analyses showed a strong prognostic value of the anisonucleosis (variation of nuclear radius) and the DNA histogram type in addition to the nodal status, the tumor size (pT), and the histological tumor grade. However, performing this analysis for both node-positive patients and for those without lymph node metastases demonstrated a different prognostic meaning of the variables. The combination of each of the group-specific variables led to a prognostic factor, which allowed an assignment of patients to several subgroups with significantly different risk for distant metastases. Thus, both a low-risk group of node-negative patients with a 5-year distant recurrence rate of only 5.8%, and a higher risk group of node-negative patients with a recurrence rate of 38.6% could be identified. Among the node-positive patients, a low-risk group with a distant recurrence rate of 8.6%, and also a high risk group with 69% distant recurrence, could be identified.