RESUMO
Gene expression can be used to subtype breast cancer with improved prediction of risk of recurrence and treatment responsiveness over that obtained using routine immunohistochemistry (IHC). However, in the clinic, molecular profiling is primarily used for ER+ breast cancer, which is costly, tissue destructive, requires specialised platforms, and takes several weeks to obtain a result. Deep learning algorithms can effectively extract morphological patterns in digital histopathology images to predict molecular phenotypes quickly and cost-effectively. We propose a new, computationally efficient approach called hist2RNA inspired by bulk RNA sequencing techniques to predict the expression of 138 genes (incorporated from 6 commercially available molecular profiling tests), including luminal PAM50 subtype, from hematoxylin and eosin (H&E)-stained whole slide images (WSIs). The training phase involves the aggregation of extracted features for each patient from a pretrained model to predict gene expression at the patient level using annotated H&E images from The Cancer Genome Atlas (TCGA, n = 335). We demonstrate successful gene prediction on a held-out test set (n = 160, corr = 0.82 across patients, corr = 0.29 across genes) and perform exploratory analysis on an external tissue microarray (TMA) dataset (n = 498) with known IHC and survival information. Our model is able to predict gene expression and luminal PAM50 subtype (Luminal A versus Luminal B) on the TMA dataset with prognostic significance for overall survival in univariate analysis (c-index = 0.56, hazard ratio = 2.16 (95% CI 1.12-3.06), p < 5 × 10-3), and independent significance in multivariate analysis incorporating standard clinicopathological variables (c-index = 0.65, hazard ratio = 1.87 (95% CI 1.30-2.68), p < 5 × 10-3). The proposed strategy achieves superior performance while requiring less training time, resulting in less energy consumption and computational cost compared to patch-based models. Additionally, hist2RNA predicts gene expression that has potential to determine luminal molecular subtypes which correlates with overall survival, without the need for expensive molecular testing.
RESUMO
Technological advancements in high-throughput genomics enable the generation of complex and large data sets that can be used for classification, clustering, and bio-marker identification. Modern deep learning algorithms provide us with the opportunity of finding most significant features in such huge dataset to characterize diseases (e.g., cancer) and their sub-types. Thus, developing such deep learning method, which can successfully extract meaningful features from various breast cancer sub-types, is of current research interest. In this paper, we develop dual stage (unsupervised pre-training and supervised fine-tuning) neural network architecture termed AFExNet based on adversarial auto-encoder (AAE) to extract features from high dimensional genetic data. We evaluated the performance of our model through twelve different supervised classifiers to verify the usefulness of the new features using public RNA-Seq dataset of breast cancer. AFExNet provides consistent results in all performance metrics across twelve different classifiers which makes our model classifier independent. We also develop a method named 'TopGene' to find highly weighted genes from the latent space which could be useful for finding cancer bio-markers. Put together, AFExNet has great potential for biological data to accurately and effectively extract features. Our work is fully reproducible and source code can be downloaded from Github: https://github.com/NeuroSyd/breast-cancer-sub-types.
