Digital immunohistochemistry: new horizons and practical solutions in breast cancer pathology

Background Digital image analysis (DA) brings new opportunities to enhance breast cancer pathology testing by providing tools to read tissue-based visual data in a more precise, accurate, and high-throughput manner compared to traditional evaluation performed by a pathologist. The applications may vary from very practical computerassisted diagnosis approaches to obtain more reproducible estimates of biomarker expression to more sophisticated efforts to retrieve and aggregate multi-modal and multi-dimensional data providing integrated metamarkers of the disease [1,2]. Immunohistochemistry (IHC) biomarkers are widely used for breast cancer categorization and provide basis for therapeutic decisions [3], therefore, standardization of IHC testing is of great importance. Applying DA tools for post-analytical IHC phase generates continuous data of broad dynamic range and can be used to analyse variance of the biomarker expression in more powerful statistical models and achieve better reproducibility. Also, analytical phase of IHC testing may also benefit from DA-based quality control systems. Another particular aspect of breast cancer diagnosis is related to HER2 gene and protein assays where better quantification systems are needed to resolve issues of tumours with equivocal and discordant HER2 status, potentially related to heterogeneity of tumour cell populations [4]. We hereby summarize our recent experiments to explore the advantages of DA for breast cancer pathology diagnosis and retrieval of new quality information based on IHC and FISH tests: DA-based quality monitor of tissue controls used for routine HER2 IHC testing, feasibility of DA to obtain automated cell-based HER2 FISH data, and the potential of factor analysis of the DA-generated IHC multi-marker expression data set to understand breast cancer immunoprofile variation.