Selection of therapeutically effective T-cell receptors from the diverse tumor-bearing repertoire

Background The development of T-cell receptor (TCR)-based T-cell therapies is hampered by the difficulties in identifying therapeutically effective tumor-specific TCRs from the natural repertoire of a patient's cancer-specific T cells. Methods Here, we mimic experimentally near-patient conditions to analyze the T-cell repertoire in euthymic tumor-bearing mice responding to the H-2K b -presented neoantigen p68 S551F (mp68). We temporarily separated the time point of mp68 expression from that of cancer cell transplantation to exclude the influence of injection-induced inflammation on T-cell priming. Thus, the mp68-specific T-cell response could only develop after the acute inflammatory phase had subsided. Results We found that mp68-specific TCRs isolated from either tumor-infiltrating T cells or spleens of mice immunized with mp68-expressing cancer cells are diverse and not inherently therapeutic when introduced into peripheral T cells and used for adoptive therapy of established tumors. While measuring short-term T-cell responses in vitro was unreliable for some TCRs in predicting their therapeutic failure, assessing the persistence of cancer cell destruction by TCR-modified T cells in long-term cultures accurately predicted therapeutic outcomes. A tumor-derived TCR with optimal function was also correctly identified with this approach when analyzing human TCRs that recognize the HLA-A2-presented neoantigen CDK4 R24L. Conclusions We show that a neoantigen-directed T-cell response in tumor-bearing hosts comprises a diverse repertoire. Infiltration and expansion of certain T-cell clonotypes in the tumor do not necessarily correlate with therapeutic efficacy of their TCRs in adoptive therapy. We propose that analysis of persistent rather than immediate responses of TCR-modified T cells in vitro serves as a reliable parameter to identify TCRs that are therapeutically effective in vivo.