PREDICTION OF AXILLARY LYMPH-NODE INVOLVEMENT OF WOMEN WITH INVASIVE BREAST-CARCINOMA - A MULTIVARIATE-ANALYSIS

BACKGROUND. The increasing use of systemic therapy for women with lymp h node negative breast carcinoma and earlier stage of disease a; mammo graphic detection raises questions regarding the need for routine axil lary lymph node dissection. Predictive modeling for lymph node involve ment may be one way to reduce the need for axillary lymph node dissect ion and its morbidity. METHODS. A multivariate analysis of 12 factors predictive of axillary lymph node involvement was conducted in a popul ation-based cc,hort of 4312 women with invasive breast carcinoma diagn osed between January 1. 1993 and December 31, 1996. RESULTS. Clinical palpability, lymphatic or vascular invasion, lesion size, margin statu s, histology, and patient age were independent predictors of axillary lymph node involvement. The model correctly identified lymph node stat us in 76.6% of cases. Model accuracy and fit were equally high when ay ,plied to randomly selected halves of the study subjects. Approximatel y 32.0% of the patients in the study sample (1363/4312) were identifie d as having an extremely high (91%; n = 1102) or low (10%; n = 261) ri sk of lymph node involvement. In a second analysis, a clinically useab le, three-variable model identified a very low risk group of patients (n = 147) with a 4.8% risk of lymph node metastasis and a high risk gr oup of patients (n = 1008) with a 74.2% risk of lymph node metastasis. Greater than 90% of subjects in the high risk group received adjuvant systemic therapy even if they were lymph node negative pathologically . CONCLUSIONS. A clinically useable, three-variable model employing tu mor and lymph node palpability, size, and lymphatic or vascular invasi on can identify women with invasive breast carcinoma in whom axillary lymph node dissection is very unlikely to alter recommendations regard ing adjuvant systemic therapy. (C) 1998 American Cancer Society.