Navigating clinical decisions through cancer cartography

Multiplexed spatial profiling has shown that tumor architecture influences therapy response, yet few spatial phenotypes are in clinical use. We opine that the immediate priority is simplification and validation of complex spatial features. We draw a parallel to cartography: maps are useful because they carry a limited set of named features whose meaning is agreed. Tumor maps could be built the same way: from the presence of a defined cell type, its proximity to another, its location within a tissue compartment, its state of activation, or combinations of these. Such features may have more clinical applicability than variable, unsupervised two-dimensional 'neighborhoods'. We propose cancer cartography as the process of assembling such validated features into tumor maps that guide treatment selection.