Field–Flow–Front mapping of breast cancer evolutionary dynamics in situ

Abstract

Tumour progression reflects not only which clones arise but where, when and in what context they expand--dimensions genotype-centred reconstructions leave unresolved. Here we reconstruct breast cancer evolution in situ across 34 Visium HD, 16 Xenium, 32 MIBI-TOF and 10 CODEX samples, serial-section 3D reconstruction, single-cell and bulk transcriptomes, and genome-wide CRISPR dependency profiles. We partition tumours into 296 cancer microzones--stroma-bounded units within which expansion is reconstructed--and define in each a Cancer Progression Metric(CPM) coupling a transcriptomic clock to expansion geometry. Projected onto tissue, CPM yields a Field-Flow-Front model rendering progression as a continuous physical process and resolving subclonal architecture into spatially coherent domains rather than predefined branches. Unexpectedly, the most advanced fronts were not the most proliferative but low-dependency, slow-cycling populations with directional expansion, driven by an extracellular matrix programme whose evolutionary force exceeded inflammation by nearly an order of magnitude yet whose genes were the least cell-autonomous. Under chemotherapy it persisted while its clonal carriers reshuffled, marking a transferable, stroma-coupled front process--not a fixed clone--as the unit of advance. Distilled into an evolutionary advantage load, this front process predicted recurrence and survival across independent cohorts and improved on conventional staging, establishing a tissue-embedded paradigm for mapping tumour evolutionary dynamics, from local fronts to patient outcome.