An idiosyncratic, zonated stroma encapsulates desmoplastic liver metastases and originates from injured liver

Abstract

AbstractColorectal cancer liver metastases (CRLM) grow in two major patterns defined by the histomorphology of the invasion front, replacement or desmoplastic. The desmoplastic pattern, in which a stromal rim separates tumor cells and liver parenchyma is a strong positive prognostic factor, implying favorable biological features. However, the origin of the perimetastatic stroma is unknown and the underlying biological mechanisms are unclear. Here, we created spatial growth pattern maps of resected CRLM at cell-level resolution using digital pathology and quantified growth pattern heterogeneity at unprecedented resolution. We manually generated > 60’ 000 individual digital annotations on 543 metastases from 263 consecutive patients. We found that, in contrast to standard growth pattern assessments, high-resolution scoring revealed the prognostic outcome to be dependent on growth pattern proportions, such that survival improved with increasing fractions of desmoplastic encapsulation. The desmoplastic pattern was coupled to decreased tumor viability and to preoperative chemotherapy, hinting at a potentially causative connection of tumor viability and fibrotic encapsulation. Analyses of the cellular constituents of the rim revealed previously unrecognized liver parenchymal remnants. Spatial quantitation of liver remnants in the rim uncovered its gradual zonation from benign-like fibrosis at the liver side to tumor-associated stroma at the metastasis edge. Together, our data suggest that the perimetatstic “desmoplastic” capsule primarily results from a reparative hepatic process in conjunction with inefficient tumor cell colonization of liver plates, rather than from actively induced desmoplasia. We posit a model in which efficient replacement-type growth that precludes a mature hepatic injury reaction determines prognosis. Our results underscore tumor-cell replacement of hepatocytes as key for liver metastatic progression and suggest that the spatial heterogeneity of tumor invasion can be leveraged to understand fundamental mechanisms of metastatic growth.