Trapping Cancers as They Adapt to Survive

Molecularly targeted therapies (such as inhibitors of BRAF, MEK, EGFR, and others) and immunotherapies (such as PD-1 inhibitors) have revolutionized the treatment of patients with solid tumors. Both targeted therapies and immunotherapies can lead to tumor regression and improved overall survival, but responses are often limited by the emergence of resistance (1, 2). Combining targeted therapies and immunotherapies may increase response rates compared with either drug alone, but may also compound toxicities (3). When, and how, is it appropriate to combine targeted therapies with immunologic agents? The rational integration of these complementary approaches will require identifying resistance mechanisms, either from patient samples or from immunologically relevant preclinical models.

Song and colleagues have now systematically cataloged the evolution of tumors from patients with melanoma and immunocompetent melanoma models treated with BRAF/MEK inhibitors (collectively, MAPKi; ref. 4). Their description could lead to new strategies that combine targeted and immunologic agents. The authors characterized the expression profiles of regressing and residual tumors from patients on MAPKi treatment. Many tumors initially displayed enhanced neural differentiation signatures, whereas at later time points they exhibited increased IFN signaling and reduced melanocyte differentiation markers. Nonetheless, resistant tumors from different patients with melanoma were found to share remarkably similar alterations compared with pretreatment samples. These changes were present in resistant cell lines treated in vitro as well as in tumors from murine melanoma models. As expected, some residual tumors harbored drivers of MAPK reactivation, such as BRAF amplification, or mutation of downstream or parallel signaling pathways. However, other residual cells exhibited attenuated MAPK dependency, suggesting that they had acquired an alternative mechanism of resistance.

Among the observed changes, the authors noted a tumor-intrinsic upregulation of the gene encoding PD-L2 (PDCD1LG2) following MAPKi treatment. PD-L2, along with the more commonly described PD-L1, constitutes a family of ligands that can bind PD-1 receptors on T cells, thereby inhibiting their proliferation and activity (5). Blocking the interaction of PD-1 with its ligand(s) consequently enhances antitumor immunity in many cancer types. PD-L2 is generally expressed on antigen-presenting cells but can be expressed in some normal tissues and tumor cell lines. For example, pancreatic, lung, and liver tissues have been shown to express PD-L2, but not PD-L1. To evaluate the functional consequences of PD-L2 induction in MAPKi-resistant tumors, the authors suppressed its expression in tumor cells. They showed that PD-L2 is required for survival of melanoma cells after prolonged exposure to MAPKi. Consistent with these genetic experiments, anti–PD-L2 antibodies enhanced the antitumoricidal activity of MEKi in vivo. Finally, the authors demonstrated that CD8⁺ T cells contributed to the efficacy of combination anti–PD-L2/MAPKi inhibitor therapy.

How does PD-L2 regulate the survival of melanoma cells? Similar to previous observations for PD-L1 (6), tumor-intrinsic PD-L2 expression was required to protect cells from apoptosis. The authors suggest that PD-L2 may reduce the apoptotic propensity of melanoma cells under chronic MAPKi treatment by regulating tumor cell–intrinsic inflammation. PD-L2 expression was also associated with a hypoxia and angiogenesis gene signature. These results imply a link between PD-L2 and a key regulator of hypoxia-stimulated gene transcription, HIF1α. Although PD-L1 is a HIF1α target gene in T cells (7), the link between PD-L2 and tumor-intrinsic hypoxia signaling requires further clarification. If validated, HIF or its key targets could themselves be vulnerabilities that could be exploited therapeutically. Further mechanistic understanding of how inhibition of MAPK leads to upregulation of PD-L2 is crucial, as many targeted therapies (such as EGFR inhibitors) also dampen this pathway. Finally, it remains unclear whether PD-L2 is required for survival of tumor cells of nonmelanocyte lineages.

Beyond these mechanistic studies lies the daunting challenge of extrapolating the data to the clinic. Much therapeutic interest has focused on disrupting the interaction of PD-1 and its ligands, but currently approved antibodies target only PD-1 or PD-L1. This new study suggests that targeting PD-L2 may also have some therapeutic activity, especially in the context of MAPK inhibition. Interestingly, PD-L2 expression may predict response to the PD-1 inhibitor pembrolizumab in head and neck cancers (8), hinting that PD-L2 may play an important role in resistance to multiple types of therapies.

Should it become possible to target PD-L2 in patients, patient selection and scheduling will be critical. The authors showed that nearly all on-treatment tumors in patients exhibit upregulation of PD-L2 following MAPKi, whereas this phenomenon is shown to be relatively underestimated in cell lines. However, the upregulation of PD-L2 is transient, and later progression of tumors is associated with the loss of both PD-L2 expression and CD8⁺ T-cell infiltrate. As T cells are required for the therapeutic efficacy of PD-1 inhibitors, the window of opportunity to target this adaptive resistance mechanism will be fairly narrow. The dynamic aspect of MAPKi-induced alterations also suggests that combining MAPKi with anti–PD-L2 agents may not lead to durable clinical benefit.

The authors do not functionally characterize other potential adaptive mechanisms of resistance revealed by their analysis. However, these may also be attractive targets to overcome resistance to MAPKi. For example, the authors observed an enrichment of cancer-associated fibroblast (CAF) signatures, consistent with the possibility that CAFs might support the growth of some melanoma cells (9). They also noted increased expression of the genes encoding the receptor tyrosine kinases PDGF, AXL, and c-MET. Given that these kinases can be targeted with small-molecule inhibitors, they are worthy of further functional study. Finally, the authors showed that unlike the dynamic changes in PD-L2 expression, enrichment of remodeling and angiogenesis targets persists following MAPKi treatment. Targeting these proteins, if experimentally validated, might lead to prolonged tumor control.

Existing clinical and preclinical data have supported the concept of combining targeted therapies with immunotherapies. For example, BRAF/MEK inhibitors can result in intratumoral T-cell accumulation, MHC class I upregulation, and antigen presentation in patients with melanoma (10). However, this new study, along with other emerging data, suggests that identifying effective drug combinations requires not only careful selection of two or more therapeutic agents. The future of cancer therapeutics may additionally require approaches that also incorporate rational timing of each therapy. Such a paradigm will increasingly require addressing both how and when tumor cells adapt to each treatment.

R. Haq reports receiving a commercial research grant from Bristol-Myers Squibb. No other potential conflicts of interest were disclosed.

R. Haq gratefully acknowledges support from the Melanoma Research Alliance. R. Haq is supported by a Stand Up To Cancer Innovative Research Grant (grant number: SU2C-AACR-IRG16-17). Stand Up To Cancer (SU2C) is a program of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the Scientific Partner of SU2C.