Identification of Features of Response and Resistance to Anti-Programmed-Death-1 Immunotherapy in Melanoma
- Author(s): Zaretsky, Jesse Meir
- Advisor(s): Ribas, Antoni
- et al.
Checkpoint blockade immunotherapy takes advantage of an endogenous anti-cancer immune response and promotes tumor killing by altering the balance of signals that control T-cell activity. Dramatic and durable responses have led to FDA approvals for anti-programmed- death-1 (anti-PD1) therapy across a broad range of cancer types, however only a fraction of patients benefit and some with an initial response will later progress. The specific mechanisms of tumor-immune evasion or resistance in this setting are not well established, but understanding them will be critical for accurate prognosis, proper therapy selection, and identification of next- generation drug targets.
To investigate mechanistic correlates of anti-PD1 response and resistance in melanoma, we used high-throughput sequencing to analyze genomic, transcriptomic, and T-cell data in whole-tumor biopsies and early passage cell lines from 1) longitudinally sampled lesions from patients with an initial response followed by late progression (acquired resistance) and 2) pre-treatment lesions in cohorts of patients with or without response (primary resistance).
Whole exome sequencing from four cases of late acquired resistance revealed evidence of in-situ clonal selection for new homozygous loss-of-function mutations in the interferon-receptor associated kinases JAK1 and JAK2 and in beta-2-microglobulin (B2M, a co-factor required for class I antigen presentation). These JAK mutations abolished sensitivity to interferon-gamma (IFNγ), including IFNγ-induced PD-L1 expression and growth arrest. In complementary work, we evaluated the frequency of interferon insensitivity (or lack of PD-L1 inducibility) and interferon pathway mutations in PD1-therapy naï¿½ve cohorts. Among the handful of JAK homozygous loss-of-function mutations we identified from patients with response data, all were found in non-responders.
Two other retrospective cohort studies revealed additional response-correlated insights. Transcriptomic data from pre-treatment biopsies showed enrichment for a coordinated set of genes associated with epithelial-to-mesenchymal transition (EMT), wound-healing, angiogenesis, and hypoxia in PD1 non-responders. By contrast, we found the rare desmoplastic melanoma subtype (DM) was unusually sensitive to anti-PD1 therapy, with a 70% response rate double that of unselected cutaneous melanoma. While often a histological diagnosis, we observed our DM cohort was characterized by a high mutational load and NF1 driver subtype.
Finally, we worked with colleagues at Caltech on enabling studies with a new technology for sensitive identification of neo-antigen specific T-cell responses that is optimized for downstream T-cell capture and receptor cloning. Here, the goal was to aid in assessment of the diversity and dynamics of defined tumor antigen-specific T-cells and to allow for their ex-vivo production.