Abstract 6670: Combined signals from tumor and immune cells predict outcomes of checkpoint inhibition in melanoma

Abstract Cancer immunotherapy with checkpoint blockade has improved survival and outcomes in melanoma, but still a majority of patients do not respond. Both high tumor mutation burden (TMB) and high T cell infiltration have been associated with response, but integrative models based on DNA or RNA assays have not been comprehensively explored and validated. Focusing on melanomas from patients receiving checkpoint blockade, we generated new and aggregated existing datasets of whole exome sequencing (WES) (n = 189 total) and bulk RNA sequencing (n = 154 total) to derive genomic and transcriptomic factors that predict survival and response to immunotherapy in melanoma. We quantified T and B cell infiltrates using rearranged T cell receptor (TCR) and immunoglobulin (Ig) sequences, respectively, from DNA or RNA sequencing. High levels of rearranged TCR reads or rearranged Ig reads in RNA-seq were associated with survival (P = 0.0046, P = 0.015) and response (P = 0.0034, P = 0.047). We created RNA-based metrics of T and B cell burden (TCBRNA or BCBRNA) by normalizing the number of rearranged TCR reads by the total number of mapped reads. When we analyzed WES data in patients for whom DNA and RNA were extracted from the same region, we found that the TCBDNA correlated with TCBRNA (rho = 0.73) and BCBDNA with BCBRNA (rho = 0.41), demonstrating that the level of lymphocyte infiltration can be estimated using rearranged TCR or Ig reads from tumor WES alone. We found that TCBDNA and BCBDNA both associated with survival (P = 0.0023 and 0.0089). In a combined model, patients with high TMB and high TCB DNA survived longer (P = 2.4e-4, HR = 2.68) and had a higher response rate (Fisher P = 0.028). This combined model was superior to models with TMB or TCBDNA alone. Similarly, patients with high TMB and high BCBDNA had longer survival and higher response rates (log-rank P = 0.0029, HR = 2.64, Fisher P = 0.015). We reanalyzed stage III/IV melanomas from TCGA and found that the TMB high, TCBDNA high subgroup had increased survival (P = 0.007). Next, clustering of tumor transcriptomes identified 5 tumor subtypes based on melanocyte differentiation, immune infiltration and keratin levels. These melanoma subtypes were associated with survival outcomes after immunotherapy (P = 0.019). We found that TBX3, a tumor-expressed transcription factor enriched in poorly differentiated melanomas, was over-expressed among non-responders within the immune-infiltrated subtype and among all patients (P = 3.9e-4, P = 8.7e-5). Patients whose tumors had high immune infiltrate and low expression of TBX3 had longer survival (P = 1.6e-5, HR = 3.39), however this subgroup did not have longer survival in an independent cohort (n = 73, P = 0.10, HR = 2.63). In conclusion, we demonstrate both RNA-based (immune infiltrate and tumor subtype) and DNA-based metrics (TMB/TCB or TMB/BCB) can be used as pre-treatment predictors of survival after checkpoint blockade in melanoma. Citation Format: Samuel S. Freeman, Moshe Sade-Feldman, Jaegil Kim, Chip Stewart, Arvind Ravi, Monica Arniella, Keren Yizhak, Ignaty Leshchiner, Liudmila Elagina, Oliver Spiro, Dimitri Livitz, Daniel Rosebrock, François Aguet, Jian Carrot-Zhang, Anna Gonye, Gavin Ha, Ziao Lin, Jonathan H. Chen, Dennie T. Frederick, Michal Barzily-Rokni, Marc R. Hammond, Hans Vitzthum, Shauna M. Blackmon, Yunxin J. Jiao, Donald P. Lawrence, Lyn M. Duncan, Anat Stemmer-Rachamimov, Jennifer A. Wargo, Keith T. Flaherty, Genevieve M. Boland, Ryan J. Sullivan, Matthew Meyerson, Gad Getz, Nir Hacohen. Combined signals from tumor and immune cells predict outcomes of checkpoint inhibition in melanoma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6670.