Abstract

The cellular and molecular systems that determine drug responses in cancer are complex, highly individual, and incompletely understood. As a result, identifying effective treatments for individual patients is still often challenging, particularly in relapsed disease. To tackle this challenge, the Snijder Lab is developing Pharmacoscopy, a platform that allows to measure hundreds of ex vivo drug responses in small patient biopsies by immunofluorescence, automated confocal microscopy, single-cell image analysis, and machine learning. In this talk I will present: 1) Results from an interventional clinical trial, showing that the approach identifies treatments associated with improved clinical responses for patients suffering aggressive hematological malignancies; 2) How we can use deep learning to discover and quantify new cellular phenotypes, and morphologically detect malignant cells leading to improved clinical response predictions. And 3), how, when combined with biopsy-centric multi-OMIC measurements and matched patient data, the approach allows to identify the molecular and cellular systems that govern treatment response individuality.