A data- and model-driven approach for cancer treatment

All people are unique and so are their diseases. Our genomes, disease histories, behavior, and lifestyles are all different; therefore it is not too surprising that people often respond differently when administered the same drugs. Cancer, in particular, is a complex and heterogeneous disease, originating in patients with different genomes, in cells with the different epigenomes, formed and evolving on the basis of random processes, with the response to therapy not only depending on the individual cancer cell but also on many features of the patient. Selection of an optimal therapy will therefore require a deep molecular analysis comprising both the patient and their tumor (e.g., comprehensive molecular tumor analysis [CMTA]), and much better personalized prediction of response to possible therapies. Currently, we are at an inflection point in which advances in technology, decreases in the costs of sequencing and other molecular analyses, and increases in computing advances are converging, forming the foundation to build a data-driven approach to personalized oncology. In this article we discuss the deep molecular characterization of individual tumors and patients as the basis of not only current precision oncology but also of computational models (‘digital twins’), the foundation for a truly personalized therapy selection of the future.