A novel preclinical platform to study FGFR2 fusion-driven cholangiocarcinoma^☆

FGFR2 fusions represent the most frequent genetic alteration in intrahepatic cholangiocarcinoma (iCCA). However, clinical responses to FGFR inhibitors are suboptimal. A major limitation is that preclinical models do not mimic the immunobiology of human FGFR2-driven iCCA. We developed an autochthonous and immunocompetent FGFR2 fusion-driven mouse model.

Using the hydrodynamic tail-vein injection method, we generated a mouse model harboring the FGFR2-PPHLN1 fusion (with or without the V565F resistance mutation) in combination with activated YAP1, and established 2D lines and 3D organoids. We characterized tumors using bulk and single cell RNA-sequencing (n=3), spectral cytometry (n=3-7/group), and performed cross-species analysis using 411 human samples. Tumor-associated neutrophils (TAN) were depleted in vivo using an anti-murine Ly6G antibody. In vitro and in vivo drug treatment with FGFR inhibitors and anti-PD-L1 was performed.

FGFR2-driven murine tumors were molecularly and immunologically similar to human iCCA, with selective accumulation of TAN and impaired T cell infiltration. scRNA-seq identified pro-tumorigenic and immunomodulatory TAN subsets while In vivo TAN depletion significantly prolonged survival (log-rank p=0.009) and led to increased CD8⁺ T cells infiltration (p=0.007). 3D organoids reproduced clinical responses to irreversible FGFR inhibitors, while showing a sustained response to the reversible inhibitor derazantinib in presence of the V565F mutation (IC50=5.190uM). In vivo derazantinib reduced tumor number (unadj. p=0.03) and marginally prolonged survival (unadj. p=0.052) whereas no statistically significant benefit of anti-PD-L1 was detected, either alone or in combination.

Our mouse model faithfully recapitulates the immunobiology and therapeutic sensitivities of the human disease. Inter-species and cross-species analyses suggest a critical role of FGFR2 fusions in shaping a cold tumor microenvironment with selective accumulation of TAN in both humans and mice.

This study develops a clinically-relevant mouse model of FGFR2-driven iCCA for future translational studies. Accumulation of pro-tumorigenic and immunomodulatory TAN along with poor sensitivity to immune checkpoint inhibitors suggests the need of designing alternative strategies harnessing the unique immunosuppressive nature of these tumors.