Fractal Analysis of Breast Tumour Microscopic Images in Prognosis of Distant Metastasis Risk

摘要

Invasive breast cancer exerts an exceptional outcome heterogeneity. For this reason, highly accurate tools for prognosis of distant metastasis are necessary for effective therapeutic management. However, current markers still exert insufficient accuracy thus prompting the need for novel prognostic strategies. Based on this pressing need and the fact that primary tumour of patients with early breast cancer is the main source of information for the individual characterization of this disease, we aimed to explore the prognostic value of tumour histology image analysis. This retrospective study was done on a group of 92 patients, selected on the basis of invasive breast cancer diagnosis without systemic treatment. Median follow-up was 147 months. The standard box-counting fractal analysis was performed on digital images of primary tumour tissue sections stained with haematoxylin/eosin. Patients with low fractal dimension had significantly higher risk of distant metastasis (57%) than those with high fractal dimension (24%), p < 0.001. Log rank (Mantel Cox) regression analysis and Kaplan Meier plots revealed that the fractal dimension equalled or exceeded the prognostic performance of the most significant clinicopathological prognostic parameters such as pathological tumour size (P=0.004) and even the clinicopathological Adjuvant! composite score (P=0.004). This is the first report of association between fractal dimension of the whole H&E tumour tissue microscopic histology and disease outcome. Fractal analysis of the primary breast tumour histology may thus provide a simple, cost-effective and highly prognostically significant indicator of distant metastasis risk. High clinical relevance of the obtained results is based on the role of prognosis in early breast cancer treatment decisions and the resulting impact on quality of life and survival.