Prediction and evaluation of response to breast cancer chemotherapy by use of multifractal analysis

Neoadjuvant chemotherapy increases survival of patients with locally advanced breast cancer. Very accurate predictors of chemotherapy response are essential for effective chemotherapeutic management due to the pronounced individual heterogeneity in breast cancer. Predictive molecular determinants for conventional chemotherapy are only emerging and still incorporate a high degree of predictive variability. Taken together, there is a pressing need for improvements of predictive performance. We addressed this issue by exploring the value of tumour histology image analysis as a novel tool for prediction and evaluation of chemotherapy response. Fractal analysis was applied to hematoxylin/eosin stained archival diagnostic breast tumour biopsies derived from 106 patients diagnosed with invasive breast cancer and treated with anthracycline. Based on the results it is concluded that multifractal analysis of breast tumour tissue prior to chemotherapy can predict the pathological complete response, partial pathological response and progressive/stable disease with accuracies ranging from 82% 91%. Multifractal comparison of tumour sections before and after chemotherapy, also based on 350 representative histology images for each group, has achieved a discrimination accuracy between the groups of 73%. This study indicates for the first time the potential value of multifractal analysis as a simple and cost-effective quantitative clinical tool for prediction and evaluation of chemotherapy response. Keywords—breast cancer, chemotherapy, histology,