PREDICTION OF SURVIVAL IN BREAST-CANCER - EVALUATION OF DIFFERENT MULTIVARIATE MODELS

Background. We compared the performance of multivariate models based o n mitotic activity index, lymph node status, and tumor size in the pro gnostication of breast cancer. Material and methods. Cox and discrimin ant models for survival were created for two patient groups: a) 120 br east cancer patients, and b) 86 patients with ductal infiltrating carc inoma. The models were compared with the model of Baak et al (1985). R esults. The models distinguished between dying and surviving patients with an efficiency of 70.9 - 77.9 % in mutual tests. With a single cut off the model of Baak et al was less efficient (50.8 - 65.8 %). If a r egion of uncertainty was allowed between two cutpoints, the efficienci es below and above the cutpoints increased. When the uncertain region included a third of the patients, the efficiency varied between 73.8 a nd 84.7 %. Conclusion. Multivariate models seem to need a region of un certainty between two discriminating cutpoints These models resulted i n the correct prediction of prognosis in about 75% and more of patient s. With different materials the models differed in efficiency With a r egion of uncertainty the model of Baak et al performed well with compl etely independent material.