DIFFERENT RISK GROUPS IN NODE-NEGATIVE BREAST-CANCER - PROGNOSTIC VALUE OF CYTOPHOTOMETRICALLY ASSESSED DNA, MORPHOMETRY AND TEXTURE

Feulgen-stained imprints and fine-needle aspirates from 528 lymph-node-negative breast cancers were investigated by means of an image analysis system. Several DNA, morphometric and textural parameters were evaluated for each patient. The prognostic value of the parameters was investigated by multivariate Cox regression analysis. As prognostic criteria, a distant recurrence-free survival of 5 years and an overall survival of 8 years were considered. In multivariate analyses the anisokaryosis (standard deviation of nuclear radius, Rad-SD) was the strongest parameter in predicting the clinical course of nodenegative patients. This was followed by a textural parameter (run-length, NR2-M) and the tumor size (pT). The DNA histogram type could also add prognostic information concerning distant recurrence-free survival, but not overall survival. In both approaches a multivariate prognostic factor was calculated for each of the node-negative patients by a linear combination of the selected variables. Using this factor, patients could be split into 5 subgroups with significantly different risks of distant metastases. Thus, a low-risk subgroup, with a 5-year distant recurrence rate of only 3%, and a subgroup with a considerably higher risk and a distant recurrence rate of 35%, could be distinguished. In survival analysis the low-risk group of node-negative patients showed an I-year death rate of only 3%, whereas in the high-risk group 30% of the patients had died at 8 years. Thus DNA, morphometric and textural parameters can provide powerful prognostic information in node-negative breast carcinomas. The multivariate combination of the relevant variables may allow a better selection of those node-negative patients with a proven good prognosis, and of those who are at risk of distant recurrence and therefore may benefit from adjuvant treatment. (C) 1995 Wiley-Liss, Inc.