Abstract

Background:

Hepatocellular carcinoma (HCC) is a major cause of mortality worldwide. Providing precision medicine to patients is a key unmet need. Preclinical models, when accurately reflecting human disease, offer an opportunity to understand varying therapeutic responses between different patient groups.

We aimed to develop a suite of genetically engineered mouse models (GEMMs), link them to human subtypes of HCC and integrate this with a cross-species organoid platform, to ultimately identify and test new therapies in human-relevant pre-clinical models.

Methods:

We developed a suite of 25 new GEMMs, modelled upon HCC-relevant mutations. We characterised these transcriptionally alongside classic non-genetic mouse models of HCC and used computational approaches to positionally align the mice to human HCC data. Integration of this data permitted the development of new cross-species molecular subclasses of HCC. We validated the subclasses histologically, performed organoid-based screening, and tested numerous therapeutic responses in vivo.

Results:

Our GEMMs exhibited many of the typical features of human HCC, including clonal origin, heterogeneity, histopathology, and disease progression, including metastasis. We classified four cross-species clusters (HuMo clusters) in the transcriptomic alignment using our specific algorithm. These HuMo clusters showed distinct characteristics including differentiation grade, degree of steatosis, and inflammatory profile.

As proof of concept, we showed subtype specific responses to standard of care treatments within, and differing between, murine representative models of the clusters. Furthermore, the linked in vivo – in vitro approach identified a novel therapeutic candidate capable of priming immune-checkpoint inhibition (ICI) responses in previously ICI-resistant tumors.

Conclusion:

Our newly developed GEMMs show distinct subclass-specific features of human HCC. They present a platform for informing translational research and performing more in-depth mechanistical studies of HCC.

We show that, by using multiple integrated resources, linking patient data with suitable preclinical in vivo and in vitro organoid models, we are able to propel therapeutic options forward towards precision medicine for HCC.