LAUSR

e12581Background: Breast tumors have highly variable rates of growth. In vivo breast cancer growth rate is governed by number of inter-related clinicopathological and biological parameters and is strictly regulated. This study aimed to assess the molecular and clinicopathological determinants of breast cancer (BC) growth rate in vivo and its impact on outcome. Methods: The study group comprised female BC patients who had a pair of serial mammograms wherein the tumor was missed in the first screen (n = 114). Tumor volume, at both time points, along with the span between measurements were used to develop a growth rate index. A machine learning algorithm was used to determine if various combinations of biomarkers could act as a surrogate for this growth rate and was validated in a large independent validation set of 1241 BC patients. Results: Patients with missed BC at first screening showed features associated with good prognosis, when patients were stratified into the slow (59%) and fast (41%) growing base...