This week Immunity published new study showing that CD8 T cell dysfunctions frequently observed in tumors is induced (a) early on (between 1-3 weeks), (b) it is antigen-specific (driven by chronic antigenic stimulation) and (c) it is irreversible even when checkpoint inhibitors are applied.
For this paper the authors have used tamoxifen-inducible, autochthonous [indigenous] liver cancer model (ASTxCre-ERT2; AST = Albumin-floxStop-SV40 large T antigen, Tag). In this model tamoxifen application induces Tag expression that inactivates two central tumor suppressor proteins, retinoblastoma protein and the p53. Analysis of Tag-specific CD8 T cells at weeks 1 or 4 showed that already by week 1 tumor antigen-specific CD8 T effector cells display signs of "exhaustion" phenotype, that was further exaggerated by week 4 (up-regulation of PD-1, 2B4, LAG3, TIM-3 and failure to express cytokines).
Furthermore, when such tumor "educated" CD8 T cells were transferred into tumor-free hosts and challenged with live Tag+ infection, only week 1 CD8 T cells could respond. Week 4 T cells were irreversibly dysfunctional. Week 4 tumor antigen-specific CD8 T cells could not be revived even with checkpoint anti-PD-1 antibody.
Finally, the authors showed that tumor induced T cell dysfunction was antigen-specific because presence of tumor did not affect non-specific CD8 T cell functionality (Tag-specific vs. OVA-specific T cells).
In summary, this study indicates that tumors could imprint irreversible "epigenetic" effector dysfunction on tumor antigen-specific T cells early on. The results in this study have important implications for clinical immunotherapy. We need to recognize that even with recent progress in application of checkpoint inhibitors for tumor management success is not guaranteed.
One drawback of this study is that the authors did not determine relationship between "antigen dosage" and CD8 T cell dysfunction. Albumin could drive huge amount of tumor antigen expression and the speed of tumor formation in these mice is clear example for such rapid "non-physiological" model. Next step would be to design tumor models where tumor specific antigen expression [dosage] could be regulated as well.