Advanced HCC precision modeling reveals divergent responses to combinatorial immunotherapy

Abstract Combinatorial immunotherapy (CIT) has afforded patients with advanced HCC a potential for long-term survival. However, sustained responses are seen only in a minority of patients. Thus, there is an unmet need for precision modeling to differentiate responder vs. non-responders and uncover predictive biomarkers. Here, we establish mouse models to mimic genetic alterations of human HCC by either overexpression of MYC and Tgfa (MycOE;TgfaOE) or by combining activated b-catenin with inactivated p53 (N90-CTNNB1OE;Trp53-/-). We then performed in-depth testing to analyze the response to CIT and characterize the immune profiles. The less proliferative N90-CTNNB1OE; Trp53-/- tumors were sensitive to CIT and developed tertiary lymphoid structures (TLSs). In contrast, faster growing MycOE;TgfaOE tumors circumvent CIT with T-cell exclusion. Importantly, early TLS initiation and T-cell exclusion features predict the therapeutic response prior to CIT. Together, these data establish that precision mouse models can mimic differential responses to CIT, with implications for developing predictive biomarkers and tailored therapy.