BACKGROUND. A Bayesian statistical approach was used t o exairnine the effect of local control 0 1 1 survival of patients with carly breast carcinoina. METHODS. '['tie effect o f radiatioii on ovcwll survival was examined in the tour published randoiiiized trials on conservation surgery with or without radiation: the IJppsala-Orebro, (:anadian, NSABP-BO(i. and Milan I l l trials. Classical and Ihyesian statistical approachcs were uscd to evaluate thc, eftwt of sample size and followi i p on thr results.
RESULTS.
(~oiiiliiiied 5-year survival results of patitints with negative lymph riodes and surgical margins in the llppsala-Orebro. Canadian, and NSAI3P-BOB trials indicated a 79% prol)ability of a positive benefit with the addition of radiation and a $),6% relative reduction in the aiinual mortality rate in favor of tlie irradiated patients. For the S a m ! subgroup 0 1 patients in thc NSAUI)-HOG trial, the probability of ii positive bcnefit with radiation increased from 65% at 5 years to 87Y0 at 10 years. and is reflected in the change of ttic relative reduction in aniiiiiil mortality from 8.2% at 5 years to 17.5% at 10 years. For all paticnts who accepted randnrnized treatment in ttic NSARFRO6 trial, tlie probability of ;I positive wrviviil bcnetit with radiation was 80%, 98%, and 91% at 5, 10, and 12 pcars, resptxtively, with corresponding relative reductions in the annual mortality ratcs of 12.8%. 20.5?6, and 152%. CONCLUSIONS. The addition of radiation t(1 Iunipectomy offus a small h i t important survival advantage, even in patients with negative surgical margins arid lymph nodes. Hecause locoregional control and breast preservntion are signiticantly improved with radiation, even a small survival advant;ige substantiates the irnportance of 1-adiarion in the treatment of patitints with early tireast carcinoina.