International Cancer Immunotherapy Conference, quick take: tumor antigens

The sessions yesterday were dominated by discussions of the role of tumor mutations in driving anti-tumor immunity. Tumor mutations can be abundant or rare depending on the indication, and this has an impact on the utility of anti-immune checkpoint therapeutics, as one example. But the question of tumor immunogenicity – can the immune system “see” the tumor – touches multiple therapeutic modalities, among them cellular therapies (TIL and engineered TCR-T cells) and the tumor vaccine field.

Two themes emerged that were not readily compatible. One theme, elegantly on display in the talk by Dr Rosenberg (NCI), is how rare and unique immune activating tumor mutations actually are, when you query patient tumors (or peripheral blood cells) for T cells that can respond to identified tumor mutations. The biology is complex, involving both CD4+ and CD8+ T cells (and the corresponding antigen recognition machinery) on the one hand and variable HLA haplotypes for peptide expression on the other hand. Only when the peptide/MHC (I or II) complex can be recognized by the TCR on a CD4 or CD8 cell can the T cell productively respond. Dr Rosenberg presented analyses of diverse tumor types, making the argument that tumor mutations that can induce T cells responses (thus tumor neoantigens), are unique across patients even within the same indication. Therefore he reasoned that expanding tumor infiltrating lymphocytes (TIL) derived from tumors (using specific cell surface markers) would give one the best chance of finding the right T cell reactivity – after all, that’s why the T cells are within the tumor. Dr Rosenberg has shown very impressive clinical results obtained exploiting these TIL. Such work may also inform efforts by cellular therapeutic companies that use TIL or TCR technology (Lion Bio, Kite, Juno etc).

A different theme emerged later in the session, focusing not on rare tumor antigens but rather on more common tumor antigens. Talks by Dr Sahin (Univ Mainz and BIONTECH Inc) and Dr Rammensee (Univ Tubingen) fell firmly in this camp, with the effort focussed on methods to identify neoantigens that could serve as vaccine components. Much of this work was preclinical, but also some interesting technology validation and early clinical application. This work has broad implications for the tumor vaccine field and perhaps the cell therapeutic modalities, as mentioned above.

So, these two themes clash over the concept the tumor neoantigens are either rare, or more common. This is a puzzle. As always, the details matter. In discussion over dinner, Taylor Schreiber, Anil Goyal and George Fromm from Heat Biologics and Amit Chaudhuri from Medgenome offered possible reasons for the discrepancies:

1) Cell selection – different methods were used to identify the specific populations of T cells to study

2) Antigen analysis – different methods were used to characterize tumor mutations and putative tumor neoantigens

3) Different algorithms – some bioinformatics tools may miss some mutations based on how they distinguish signal from noise (the cancer/testis families were offered as an example here)

So, time to go back and reread the literature.

stay tuned