AMD3100 Disrupts Stellate-Cell-Mediated T-Cell Exclusion in Fibrolamellar Carcinoma

Researchers applied single-nucleus RNA-seq to fresh fibrolamellar resections and matched normal liver, revealing a CXCL12-rich niche created by activated stellate cells that physically sequesters CD8+ T cells in peritumoral fibrous bands. Ex-vivo treatment with AMD3100 reduced CXCL12 signaling, increased T-cell motility toward tumor clusters, and restored interferon-gamma signatures absent in untreated samples. These spatial and transcriptional changes were absent in standard bulk RNA data, explaining why prior checkpoint-inhibitor trials reported near-zero response rates. The work directly links the pathognomonic lamellar fibrosis to immune evasion rather than treating it as a passive histologic feature.

Fibrolamellar carcinoma shares T-cell exclusion signatures with pancreatic ductal adenocarcinoma and desmoplastic melanoma, where CXCR4 blockade has already entered early-phase testing. By repurposing an existing agent whose safety profile is established in WHIM syndrome and stem-cell mobilization, the Cornell-Washington team bypasses the usual multi-year drug-development timeline. However, the study used only resected, treatment-naïve specimens; whether AMD3100 penetrates established metastases or synergizes with anti-PD-1 in vivo remains untested in patients.

A critical gap is the absence of longitudinal patient-derived xenograft or organoid models that could quantify the minimum drug exposure needed to sustain T-cell infiltration. Planned correlative analyses in an upcoming phase Ib window-of-opportunity trial will measure intratumoral T-cell density before and after 14 days of AMD3100, providing the first human pharmacodynamic benchmark.