ARTICLE ABSTRACTOnly a subset of patients responds to immune checkpoint blockade (ICB) in melanoma. A preclinical model recapitulating the clinical activity of ICB would provide a valuable platform for mechanistic studies. We used melanoma tumors arising from an Hgftg;Cdk4R24C/R24C genetically engineered mouse (GEM) model to evaluate the efficacy of an anti–mouse PD-L1 antibody similar to the anti–human PD-L1 antibodies durvalumab and atezolizumab. Consistent with clinical observations for ICB in melanoma, anti–PD-L1 treatment elicited complete and durable response in a subset of melanoma-bearing mice. We also observed tumor growth delay or regression followed by recurrence. For early treatment assessment, we analyzed gene expression profiles, T-cell infiltration, and T-cell receptor (TCR) signatures in regressing tumors compared with tumors exhibiting no response to anti–PD-L1 treatment. We found that CD8+ T-cell tumor infiltration corresponded to response to treatment, and that anti–PD-L1 gene signature response indicated an increase in antigen processing and presentation, cytokine–cytokine receptor interaction, and natural killer cell–mediated cytotoxicity. TCR sequence data suggest that an anti–PD-L1–mediated melanoma regression response requires not only an expansion of the TCR repertoire that is unique to individual mice, but also tumor access to the appropriate TCRs. Thus, this melanoma model recapitulated the variable response to ICB observed in patients and exhibited biomarkers that differentiate between early response and resistance to treatment, providing a valuable platform for prediction of successful immunotherapy.
Our melanoma model recapitulates the variable response to anti–PD-L1 observed in patients and exhibits biomarkers that characterize early antibody response, including expansion of the TCR repertoire.