PerspectiveCancer Immunotherapy

Clinical implications of tumor-intrinsic mechanisms regulating PD-L1

See allHide authors and affiliations

Science Translational Medicine  06 Feb 2019:
Vol. 11, Issue 478, eaav4810
DOI: 10.1126/scitranslmed.aav4810


Treatment with immune checkpoint inhibitors targeting programmed death receptor-1 (PD-1) or programmed death ligand-1 (PD-L1) is effective in many cancer types. Tumors harboring specific mutations modulate antitumor immune responses through the PD-1/PD-L1 axis, and this should be taken into account when designing rational combinatory treatments.


Tumor cells are frequently characterized by genetic instability, which facilitates the acquisition of new mutations. Such variability can create new potential antigens, against which effective immune responses can be generated. Consequently, tumor cells implement mechanisms, which prevent immune-mediated eradication, to cope with their newly acquired visibility to the immune system. A major mechanism of immune resistance is the up-regulation of programmed death-ligand 1 (PD-L1), which inhibits T cell effector functions (1). This detrimental effect on the immune compartment can be exploited clinically by interfering with the PD-1/PD-L1 axis. Owing to the fact that proinflammatory signaling is a major physiological stimulus that induces PD-L1, oncogenic signaling may cooperate with physiological inflammatory signaling pathways, further enhancing PD-L1 expression. An example of this concept is inflammatory Janus kinase (JAK) 2 signaling in Hodgkin’s lymphoma (HL). HL carries the 9p24.1 amplification, which results in increased expression of both the JAK2 and the PDL1 genes, thereby contributing to high PD-L1 transcription (2), which explains the high clinical response rates of this disease to PD-1/PD-L1 inhibition (3). Oncogenic JAK2 activation through the V617F point mutation also has a direct effect on PD-L1 expression in myeloproliferative neoplasms (MPN), which renders this disease susceptible to immune checkpoint inhibitors (ICI) in preclinical models (4). Whereas the susceptibility to PD-1/PD-L1 blockade has been shown for HL, clinical trials that test PD-1 inhibitors in MPN are ongoing (NCT02421354).

In agreement with a central role of JAK2 signaling for PD-L1 expression, loss-of-function mutations in JAK1/2 genes detected in melanoma and other cancer types cause resistance to PD-1/PD-L1 blockade (57) (Fig. 1). It is conceivable that, during anti-PD1 immunotherapy, cancer cells with an inactivating JAK2 mutation experience a survival advantage because they have acquired mechanisms to escape the immune response independent of the JAK2/PD-L1 axis. This observation has important clinical implications because patients with tumors carrying loss-of-function mutations in JAK1/2 genes will not benefit from PD-1/PD-L1 inhibition but still have the risk for immune-mediated side effects. In addition, it appears that cancer cells can efficiently combine intrinsic programs of cell proliferation and immune escape by activating oncogenic JAK/STAT signaling (Fig. 1). Clinically, the activation status of JAK2 signaling in different tumors may predict the sensitivity to PD-1/PD-L1 blockade, a concept that needs validation in future studies.

Fig. 1 PD-L1 expression in response to JAK2/STAT signaling might predict vulnerability to PD-1/PD-L1 blockade.

In solid tumors, IFN-γ signaling promoted by adaptive immune responses causes JAK/STAT activation that stimulates PD-L1 expression. Tumors harboring JAK/STAT inactivating mutations have reduced PD-L1, which can cause primary or acquired resistance to immune checkpoint inhibitors (left). In Hodgkin’s lymphoma and mediastinal large B cell lymphoma (middle), JAK2 amplification sustains high expression of PD-L1, which causes immune escape and supports the usage of immune checkpoint inhibitors. In myeloproliferative neoplasms (right), oncogenic JAK2V617F promotes PD-L1 expression, which induces immune escape but might also confer sensitivity to PD-1/PD-L1 blockade, which, as in the middle panel, can cause both immune escape and sensitivity to immune checkpoint inhibitors. EpoR, erythropoietin receptor; TpoR, thrombopoietin receptor.


In certain malignancies, oncogenic pathways cooperate with inflammatory signaling, particularly with signals caused by interferon-γ (IFN-γ), which can be released in the tumor microenvironment to promote PD-L1 expression. For example, in multiple myeloma, the mitogen-associated protein kinase (MAPK) pathway is necessary for signal transducer and activator of transcription 1 (STAT1)–mediated PD-L1 induction downstream of IFN-γ signaling (8). In colorectal cancer and non–small cell lung cancer (NSCLC), phosphatidilinositol 3-kinase (PI3K) signaling inhibition abrogates IFN-γ–dependent PD-L1 expression (9, 10). In medulloblastoma (MB), cyclin-dependent kinase 5 (CDK5) inhibition suppressed IFN-γ–dependent PD-L1 expression through enhanced stabilization of interferon-responsive factor 2 (IRF2), which acts as a repressor of PDL1 gene transcription (11). These data from different tumor entities support the concept that cross-talk between inflammatory and oncogenic signaling pathways coordinates and sustains immune evasion.


Apart from cooperating with inflammatory signaling, certain oncogenes directly control PD-L1 expression. Oncogenic mutations in receptor tyrosine kinases (RTKs), including epithelial growth factor receptor (EGFR) (12) and anaplastic lymphoma kinase (ALK) (13), can cause PD-L1 expression through multiple mechanisms. Besides mutant RTKs, the activation of signaling molecules and transcription factors can promote PD-L1 expression. The activity of these signaling molecules depends on the tumor entity. For example, in NSCLC, oncogenic RAS signaling stabilizes PD-L1 mRNA (14), whereas in lymphomas, MYC enhanced PD-L1 transcription (15). These observations suggest that RTK inhibition and immunotherapy have redundant functions in some cases. Moreover, a synergy between RTK inhibitors and immunotherapy has been previously reported in AML, combining sorafenib and T cell transfer (16), and in melanoma, combining BRAF inhibition and anti-PD1 immunotherapy (17). The synergistic effect of kinase inhibition and T cell transfer in patients with AML was dependent on the chromatin status of the IRF7 gene and metabolic activity of T cells (16), indicating that it can be essential to identify biomarkers for response when combining targeted therapy and immunotherapy.

For a number of transcription factors involved in cell proliferation, including MYC, STATs, nuclear factor κB (NF-κB), or c-Jun, the presence of responsive elements in the PDL1 gene promoter has been demonstrated (18) (Fig. 2). Furthermore, typical cancer-associated events such as hypoxia-induced factor (HIF) transcription factor activation, which allow the cells to cope with the diminished oxygen availability within the tumor microenvironment, as well as p53 loss (19, 20) and PTEN loss (21), which contribute to reduced genomic stability, are also responsible for immune escape via PD-L1. The control over PD-L1 abundance is achieved not only through interference with gene transcription but also through mRNA and protein stabilization and degradation (18). Tumors with defective p53 have decreased miRNA-34a and increased PD-L1 expression, when compared with p53-competent tumors (20). In a syngeneic NSCLC mouse model, peritumoral administration of a miRNA-34 mimic reduced PD-L1 expression and increased CD8+ T cell infiltration (20). A functional role for the p53-miRNA-34a-PD-L1 axis has also been described for AML (22). In certain cancers, PD-L1 expression can occur when cancer-promoting kinases are inhibited. An important example is the inhibition of CDK4, which increases PD-L1 expression by reducing ubiquitination and proteasomal-mediated degradation (23). This indicates that PD-L1 induction could be a mechanism of resistance to CDK4 inhibitors and supports the combination of CDK4 inhibitors and PD-1/PD-L1 inhibition (23). Besides the PD-L1 expression per se, recent studies show that PD-L1 is highly glycosylated and that N-linked glycosylation of PD-L1 critically maintains its protein stability and is required for its interaction with PD-1 (2426). Currently, it is unclear whether oncogenic signals affect N-linked glycosylation of PD-L1. The PD-1/DL-L1 axis is only one of many tumor-intrinsic mechanisms of escaping the immune response. Other important oncogenic pathways in the context of cancer immunity include polycomb repressive complex 2 (PRC2) (27) and β-catenin (28), for example.

Fig. 2 Oncogenic mutations and pathways control PDL1 gene transcription by activating transcriptional regulators.

Activation of oncogenes and loss of tumor suppressor genes activate signaling pathways that cause PD-L1 induction through activation of a defined subset of transcription factors. These transcription factors such as STAT3, HIF-1, MYC, and NF-κB bind to specific sequences in the CD274 gene promoter, which encodes PD-L1, and facilitate gene transcription. Most of the displayed transcription factors are activated downstream of multiple oncogenes. The cancer entities appear in black letters, the starburst symbols represent activating oncogenic mutations, and the black “cross” represents a deletion of a tumor suppressor gene. Oncogenes appear in red, tumor suppressors appear in blue, and transcription factors appear in gray.



Several ongoing clinical studies are testing for synergy between immunotherapy and targeted therapies (Table 1), including PARP inhibitors, ALK inhibitors, PI3K inhibitors, JAK1 inhibitors, multitargeted tyrosine kinase inhibitors (TKIs), anti–VEGF-R antibodies, HDAC inhibitors, BTK inhibitors, BRAF inhibitors, and MEK inhibitors. Such studies will need to address the impact of the targeted therapies not only on PD-L1 abundance in the tumor compartment but also on other factors that are relevant for achieving successful ICI responses. For example, early clinical trial data for the combination of the JAK1 inhibitor itacitinib with pembrolizumab showed decreased intratumoral CD8 T cell infiltration, resulting in a lower than expected response rate and study interruption (29). In addition, a potential disadvantage of the combinations could be that targeting central signaling circuits, such as the MAP kinase pathway, may also block effector T cell activation, thereby limiting antitumor responses. The combination of nivolumab with the small-molecule inhibitors sunitinib or pazopanib in a phase 1 trial in metastatic renal cell carcinoma demonstrated an objective response rate of 45 to 52% (30), which has led to multiple clinical combination trials in renal cell carcinoma (NCT03141177 and NCT02231749). In addition, this resulted in the development of the MRTX-500 study, a phase 2, open-label study in NSCLC with nivolumab in combination with either of two multitargeted TKIs, glesatinib and sitravatinib, and an HDAC inhibitor, mocetinostat (NCT02954991). Despite these promising results, caution is warranted when combining immunotherapy with targeted therapy, as evidenced by an aborted phase 1 trial combining vemurafenib and ipilimumab, which was stopped early because of several cases of grade 3 to 4 hepatitis (31).

Table 1 Combination of targeted therapy and immunotherapy in clinical trials.

CRC, colorectal cancer; NSCLC, non–small cell lung cancer; CLL, chronic lymphocytic leukemia; SCC, squamous cell cancer; RCC, renal cell cancer.

View this table:


Therapies targeting the PD-1/PD-L1 axis have produced impressive clinical responses and durable remission in various cancer types. An unsolved clinical problem, however, is primary resistance observed in many patients and secondary resistance that many responders ultimately develop. Therefore, a better understanding of the genetic, epigenetic, transcriptional, translational, and posttranslational regulation of PD-L1 by oncogenic mutations or loss of tumor suppressor genes may allow for the development of rational therapies combining TKIs with PD-1/PD-L1 inhibition. The observation that RTK signaling enhances immune escape via PD-L1 suggests that a combination of TKI with immunotherapy may be more effective than each treatment alone. Conversely, prolonged pretreatment of a tumor with TKIs before immunotherapy could reduce PD-L1 expression, allowing other immune resistance mechanisms to evolve. In agreement with these concepts, a synergism between inhibition of the oncogenic signal and immunotherapy has been reported for AML (16), melanoma (17), and RCC (NCT02954991). It could be a disadvantage to combine certain kinase inhibitors with anti–PD-1 immunotherapy because the kinase inhibition decreases PD-L1 expression, which reduces the vulnerability of the remaining cancer cells to anti-PD1. It could be that a combination of the oncogenic signal inhibitor with immunotherapy other than anti-PD1/PD-L1 may be more successful, for example with arginase inhibitors or TIM3 or LAG3 inhibitors. Another important point to consider is that not only PD-L1 expressed by tumor cells but also PD-L1 on antigen-presenting cells, particularly myeloid cells including DCs and macrophages, has therapeutic relevance (32, 33). The relative contribution of tumor PD-L1 versus host PD-L1 to immune escape may vary depending on the cancer type. Because p53-mutated tumors exhibit increased PD-L1 expression (20), the combination of p53-restoring agents with immune checkpoint inhibitors could also have synergistic effects. Increased understanding of tumor-intrinsic mechanisms that promote PD-L1 expression will be of service in the design of clinical trials with rational combinations of RTK inhibitors, p53-restoring agents, CDK inhibitors, and anti-PD1 immunotherapy that could achieve synergistic effects.


Stay Connected to Science Translational Medicine

Navigate This Article