Deep-Learning Based Prediction of c-MET Status from Digitized H&E-Stained Non-Small Cell Lung Cancer Tissue Samples

Genetic profiling of non-small cell lung cancer (NSCLC) tissue has identified alterations in a number of genes, including c-MET. Dysregulation of c-MET, a receptor tyrosine kinase mesenchymal epithelial transition factor, is associated with worse prognosis. Therapeutics targeting c-MET such as the antibody drug conjugate (ADC) Telisotuzimab vedotin (Teliso-V) may be of benefit. Currently, immunohistochemical staining of tissue is used to determine whether c-MET protein is overexpressed. Here we report development of a machine learning (ML)-powered method that identifies features of the tumor microenvironment (TME) as predictors of c-MET overexpression status (positive vs.