Relapse of melanoma after successful adoptive T‐cell therapy: escape through inflammation‐induced phenotypic melanoma cell plasticity

Abstract

Advanced stage melanomas are refractory to most available therapies including recently approved targeted therapies. This is also true for most biological (IFN-a or IL-2) treatments and therapy involving adoptive transfer of melanoma-reactive T cells (de Souza et al., 2012). A number of mechanisms have been proposed to explain the lack of T-cell functional activity in vivo. Some of these mechanisms include emergence of antigen-loss tumor variants, induction of immune inhibitory molecules such as CTLA4 or PD1 and induction of T-reg (regulatory) cells or myeloid-derived suppressor cells (MDSC) or presence of immune-suppressive cytokines such as TGF-b or IL-10 (Grivennikov et al., 2010). A number of environmental factors can further contribute to genetic instability and induce epigenetic changes of progressing heterogeneous tumors. Presence of inflammatory factors at the tumor site can cause selection of preexisting tumor cell variants resulting in preferential growth of antigen-negative tumor variants or tumor cells that have lost MHC-class I molecules (Khong and Restifo, 2002). Immune editing or alterations of tumor antigens due to immune pressure can also be one of the reasons for lack of T-cell-mediated tumor cell death in vivo (Singh and Paterson, 2007). In clinical trials, adoptive cell therapies (ACT) with melanoma-reactive tumor-infiltrating lymphocytes (TILs) or polyclonal T cells engineered to express chimeric antigen receptor (CAR) with re-directed specificity to melanoma have shown great promise with dramatic regression of bulky tumors (Robbins et al., 2011). However, in most patients, initial regression of tumor cells was followed by appearance of new lesions with therapy resistant melanoma cells (Robbins et al., 2011). Until now, the exact nature of ACT-related tumor resistance and the cause of emergence of antigen-negative melanoma cells were largely unknown. In the current edition of Nature, Landsberg et al. describe a novel mechanism of T-cell-induced loss of melanocytic antigens as a possible cause of tumor cell resistance to ACT. To demonstrate therapy resistance, the authors have used a previously published genetically engineered hepatocyte growth factor (Hgf)-Cdk4 mouse melanoma model and anti-gp100-specific CTLs obtained from a TcR transgenic pmel-1 (gp100) mouse model. TNF-a, a proinflammatory cytokine predominantly secreted by activated T cells caused down-modulation of gp100 antigen expression on the mouse melanoma cells resulting in loss of recognition by anti-gp100-specific CTLs used in ACT. TNF-a also down-modulated pigmentrelated gene expression and tyrosinaserelated protein (TRP)-2 on melanoma cells. Interestingly, the pro-inflammatory environment of TNF-a did not downmodulate the expression of non-melanocytic antigens. On the contrary, there was an increased expression of nerve growth factor receptor (NGFR) on melanoma cells. Down-modulation of gp100 expression was reversible when the activity of TNF-a was neutralized or the resistant tumor cells were re-transplanted into a normal mouse lacking an inflammatory environment. The mechanism of non-genetic reversible antigen down-modulation is novel and has not been described before. This is in contrast to other genetic heterogeneityand immunoselection-related mechanisms described earlier. Understanding inflammation-induced plasticity and the non-genetic heterogeneity of melanoma cells are critical for the development of new therapeutic strategies for melanoma. The authors then corroborated their mouse studies using human melanoma cell lines where they report similar down-modulation of gp100 and MART1/Melan-A expression in the presence of TNF-a. Similar to mouse studies, human melanoma cell lines also showed up-regulation of NGFR expression in the presence of TNF-a. This was confirmed in melanoma patients’ tumor tissue sections with abundant infiltrating immune cells further supporting the clinical relevance of the above findings. The results of Landsberg et al. have important clinical implications if the phenotypic plasticity of melanoma cells in an inflammatory microenvironment is validated in a larger cohort of patients with melanoma after ACT. Down-modulation of gp100 expression and concurrent up-regulation of NGFR expression of melanoma cells in an inflammatory environment raises the possibility of using ACT with a cocktail of CAR-engineered T cells with dual or multiple specificities to target more than one melanoma antigen to ensure complete regression of tumor lesions. For this, a personalized approach of analyzing melanoma antigen expression on tumor cells before and after ACT therapy will be essential for the selection of CARengineered T cells with defined tumor antigen specificity. The proposed approach may enhance the long term disease-free survival of patients with metastatic melanoma.