Cancer Therapy: Clinical PD-L1 Expression and Immune Escape in Melanoma Resistance to MAPK Inhibitors Hojabr Kakavand 1,2 , Robert V. Rawson 1,3 , Gulietta M. Pupo 4 , Jean Y. H.Yang 5 , Alexander M. Menzies 1,2,6 , Matteo S. Carlino 1,7 , Richard F. Kefford 1,8 , Julie R. Howle 1,2,7 , Robyn P.M. Saw 1,2,9 , John F. Thompson 1,2,9 , James S. Wilmott 1,2 , Georgina V. Long 1,2,6 , Richard A. Scolyer 1,2,3 , and Helen Rizos 1,8 Abstract Purpose: To examine the relationship between immune activ- ity, PD-L1 expression, and tumor cell signaling, in metastatic melanomas prior to and during treatment with targeted MAPK inhibitors. Experimental Design: Thirty-eight tumors from 17 patients treated with BRAF inhibitor (n ¼ 12) or combination BRAF/MEK inhibitors (n ¼ 5) with known PD-L1 expression were analyzed. RNA expression arrays were performed on all pretreatment (PRE, n ¼ 17), early during treatment (EDT, n ¼ 8), and progression (PROG, n ¼ 13) biopsies. HLA-A/HLA-DPB1 expression was assessed by IHC. Results: Gene set enrichment analysis (GSEA) of PRE, EDT, and PROG melanomas revealed that transcriptome signatures indic- ative of immune cell activation were strongly positively correlated with PD-L1 staining. In contrast, MAPK signaling and canonical Wnt/-b-catenin activity was negatively associated with PD-L1 melanoma expression. The expression of PD-L1 and immune activation signatures did not simply reflect the degree or type of immune cell infiltration, and was not sufficient for tumor response to MAPK inhibition. Conclusions: PD-L1 expression correlates with immune cells and immune activity signatures in melanoma, but is not sufficient for tumor response to MAPK inhibition, as many PRE and PROG melanomas displayed both PD-L1 positivity and immune activation signatures. This confirms that immune escape is common in MAPK inhibitor–treated tumors. This has impor- tant implications for the selection of second-line immunotherapy because analysis of mechanisms of immune escape will likely be required to identify patients likely to respond to such therapies. Clin Cancer Res; 23(20); 6054–61. Ó2017 AACR. Introduction The MAPK pathway is constitutively activated in the major- ity of cutaneous melanomas (1), most commonly via muta- tions affecting BRAF kinase. Targeted inhibition of the MAPK pathway, with single-agent BRAF inhibitors or combined BRAF and MEK inhibitors, has improved the progression-free (PFS) and overall survival (OS) of patients with BRAF V600 - mutant metastatic melanoma (2). However, only 20% of patients remain progression free in the long term (3), and the majority will develop resistance within 12–24 months of commencing treatment via mechanisms that reactivate MAPK signaling and/or enhance PI3K/AKT pathway activity (4–6). The genetic mechanisms of resistance to MAPK inhibitors are varied and heterogeneous. Nevertheless, in 20%–40% of patients who progress while receiving combination BRAF and MEK inhibitor therapy, the mechanism of resistance remains unknown (7, 8). The immune system contributes to the antitumor activity of BRAF inhibitors (9). Inhibition of the MAPK pathway promotes a favorable immune microenvironment by increasing the expres- sion of melanoma antigens, downregulating immunosuppressive cytokines and increasing the infiltration of CD4 þ and CD8 þ lymphocytes early during treatment (EDT; within 3–15 days of initiating therapy; refs. 10–13). Significantly, the density of the intratumoral CD8 þ lymphocyte infiltrate correlates with reduc- tion in tumor size (11) and an improved response to BRAF inhibition (14). Immune suppressive components within the melanoma microenvironment may also play an important role in BRAF inhibitor responses. The absence of the immunosup- pressive programmed death receptor-ligand-1 (PD-L1) at baseline is associated with improved response to BRAF inhibition (14). We recently confirmed that tumor PD-L1 expression is signifi- cantly altered during MAPK inhibitor therapy. In patients with positive tumor PD-L1 staining in the sample taken prior to 1 Melanoma Institute Australia, North Sydney, New South Wales, Australia. 2 Sydney Medical School, The University of Sydney, Sydney, New South Wales, Australia. 3 Tissue Pathology and Diagnostic Oncology, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia. 4 The University of Sydney at the Westmead Institute for Medical Research, Centre for Cancer Research, Westmead New South Wales, Australia. 5 School of Mathematics and Statistics, The University of Sydney, Sydney, New South Wales, Australia. 6 Royal North Shore Hospital, New South Wales, Australia. 7 Crown Princess Mary Cancer Centre, Westmead Hospital, Westmead, New South Wales, Australia. 8 Faculty of Medicine and Health Sciences, Macquarie University, Sydney, New South Wales, Australia. 9 Department of Melanoma and Surgical Oncology, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia. Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). G.V. Long, R.A. Scolyer, and H. Rizos contributed equally to this article. Corresponding Author: Helen Rizos, Faculty of Medicine and Health Sciences, 2 Technology Place, Macquarie University, New South Wales 2109, Australia. Phone: 612-9850-2762; Fax: 612-9850-2701; E-mail: helen.rizos@mq.edu.au doi: 10.1158/1078-0432.CCR-16-1688 Ó2017 American Association for Cancer Research. Clinical Cancer Research Clin Cancer Res; 23(20) October 15, 2017 6054 on June 3, 2020. © 2017 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from Published OnlineFirst July 19, 2017; DOI: 10.1158/1078-0432.CCR-16-1688