A decision tool for predicting sentinel node accuracy from breast tumor size and grade.

The ability to predict axillary lymph node involvement in breast cancer patients in the preoperative setting is invaluable. This study provides a simple set of formulae to enable clinicians to make informed decisions in the management of screen-detected breast cancer. The tumor pathology reports were obtained of all 4,585 women identified between 1996 and 1999 in New South Wales (NSW) with T1 or T2 breast cancer by the statewide co-ordinated breast screening service (BreastScreen NSW). Equations predicting node positivity were calculated by linear regression analysis and, from published sentinel node false-negative rates, the probability of retrieval of a false-negative axillary lymph node by sentinel node biopsy was calculated for tumors of different size and grade. Node involvement was identified in 1,089 (23.8%) of women. A linear relationship for tumor size, grade, and nodal involvement was predicted by: frequency (%) = 1.5 x tumor size (mm) + 2 (or 6 or 10) for grade I (or II or III) tumors. Assuming a 7.5% false-negative rate, the probability of retrieving a false-negative sentinel node ranged from 0.8% for a patient with a 5 mm, grade I carcinoma to 6.0% for a 50 mm, grade III tumor. These simple formulae are easy to use in a clinical setting. The reference table enables breast surgeons to inform a patient about the absolute probability of false-negative sentinel biopsy rates for patients with screen-detected carcinomas when size can be estimated from preoperative imaging and when tumor grade is often available from preoperative core biopsy. Patients with large, T2 breast tumors may be best treated with axillary dissection rather than sentinel node biopsy alone due to the risk of under-staging the woman's disease and also the high probability of finding a positive sentinel node.