Cancers of Unknown Primary: a model for personalized medicine

Cancers of unknown primary (CUP) consist in a heterogeneous group of malignancies characterized by the existence of metastases in the absence of detectable primary tumors. CUP represent a major challenge for both diagnosis and therapeutic management.  We have developed a deep learning-based tool based on the analysis of RNAseq data to identify the tissue of origin in CUP. This tool has been implemented recently within the bioinformatic pipeline of the national sequencing platform Seqoia and is used routinely during the national multidisciplinary tumor board (MTB) for  CUP patients in France. Integration of clinical, pathological and molecular data within this MTB enables to provide therapeutic orientation to most patients and is associated with increased survival.