Anti-GARP reverts immune suppression and extends overall survival in a MASH-HCC murine model resistant to anti-PDL1+anti-VEGFA therapy

Background and Aims: Atezolizumab (anti-PDL1) plus bevacizumab (anti-VEGFA) (atezo+bev) is one first-line standard-of-care (SOC) treatment for advanced hepatocellular carcinoma (aHCC) with objective responses in ~30% of cases. Clinical benefit is hindered in non-viral HCC, including metabolic dysfunction-associated steatohepatitis (MASH)-related cases, a raising HCC etiology. MASH-HCC tumors are enriched in TGFß signaling, which has been associated with resistance to atezo+bev. The bioavailability and the Treg-related immunosuppressive effect of the TGFß1 ligand is controlled by the TGFbeta-anchoring receptor GARP. Hence, we hypothesize that GARP blockade might revert immune suppression and overcome resistance to aPDL1+aVEGF (SOC) in MASH-HCC. Method: We used an immune-genetic MASH-HCC model that combines diet-induced steatohepatitis with MYC and CTNNB1 overexpression (via hydrodynamic gene delivery), previously shown to drive atezo+bev resistance. After tumor onset, animals were treated with SOC+anti-GARP (triple combination), SOC alone or placebo until the end of the study. Spectral cytometry was conducted at day 14 to assess changes in the tumor microenvironment (n=5/arm). Remaining animals were followed for overall survival (OS) analysis (n=7/arm). GARP expression was correlated with molecular/immune features in three independent cohorts of HCC patients (n=559). Results: SOC+anti-GARP significantly extended OS compared to placebo (median of 68 vs 59 days, p=0.045), as opposed to SOC alone (p=0.21). From the immune perspective, the triple combination induced a shift from central memory (CD62L+CD127+) to tumor-resident (CD62L-CD69+) CD8+ T cells (p=0.031) accompanied by an increase in total CD8+ (p=0.0028) and CD4+ (p=0.016) T cells compared to placebo, suggesting a more locally active T-cell response. Accordingly, tumors from SOC+anti-GARP-treated mice presented increased infiltration of both effector (CD44+PD1+TIM3-) (p=0.036) and exhausted (PD1+TIM3+) CD8+ T cells (p=0.007). Among CD4+ T cells, the triple combination also increased antigen-experienced CD44+ (p=0.045) and showed a trend of the ratio between tissue-resident memory (CD69+CD103+) and Tregs (CD25+FOXP3+) (p=0.053) when compared to the rest. None of these changes were observed with SOC alone. In HCC patients, high GARP expression significantly correlated with immune exhaustion (fold-change vs rest >1.5 in 87% of cases), with an increase of Treg infiltration compared with other patients, highlighting a subset of patients that may benefit from this combination. Conclusions: Our findings highlight GARP as a promising therapeutic target whose blockade restores antitumor immune infiltration and extends survival OS when added to SOC in an immunotherapy-resistant murine model of MASH-HCC.