Factors influencing HER2 positivity in breast cancer according to statistical modeling and data mining techniques based on a real-world national database: HER-France.

e12086Background: HER-France is a national database collecting information on HER2 status in invasive breast cancer supplied by 124 French pathology labs (FPL). It is conducted by the French Association for Quality Control in Histopathology (AFAQAP) since October 2011. Its goal is to follow the national HER2 rate and its change over time. An analysis of this database was carried in October 2015 on 105,717 records to search for factors associated with HER2 positivity. Methods: To identify these factors, multivariate regression models were applied, as well as supervised and unsupervised Data Mining techniques such as Bayesian Networks and Decision trees were used to represent dependencies between tumor characteristics and HER2 positivity. HER2 status was determined by immunohistochemistry and in situ hybridization using ASCO/CAP 2013 guidelines. To avoid bias due to adverse pre-analytic conditions, only results on Core-Biopsies (CB) are presented hereafter. Results: Out of 105,717 records, 45,785 (43.3%) CB...