Programmed Death Ligand-1 (PD-L1) Expression in the Programmed Death Receptor-1 (PD-1)/PD-L1 Blockade: A Key Player Against Various Cancers

The PD-1/PD-L1 pathway belongs to the immune checkpoint signaling pathways regulating T-cell–mediated local inflammatory reactions and self-tolerance.⁶  The importance of the PD-1/PD-L1 axis in cancer immunity has been demonstrated extensively in both animal models and clinical studies.⁷ 

PD-1, a cell surface protein belonging to the CD28 family, is encoded by PDCD1 gene located in chromosome 2q37.²  PD-1 is expressed on activated T cells, with particularly high expression by tumor-infiltrating T lymphocytes.²  PD-1 is also expressed on activated non-T cells, including B cells, natural killer cells, and monocytes, implying that PD-1 may also modulate immunity in a T-cell–independent manner.⁸  PD-1 has 2 major ligands: PD-L1/CD274 (encoded by PDCD1LG1 in chromosome 9) and PD-L2/CD273 (encoded by PDCD1LG2 in chromosome 9).

Binding of PD-1 to either PD-L1 or PD-L2 results in the activation of inhibitory kinases involved in T-cell proliferation, adhesion, and cytokine production/secretion via phosphatase SHP2.²  PD-1–PD-L interaction has been shown to play an important role in limiting the initial response of T cells upon antigen exposure and inducing T-cell tolerance. However, there are differences between PD-L1 and PD-L2.²  PD-L1 is broadly expressed on many immune cell types, such as T cells,⁹  B cells, macrophages, regulatory T (Treg) cells, and dendritic cells, as well as some non–immune cell types, such as vascular endothelial cells and pancreatic cells. On the contrary, PD-L2 is largely limited to antigen-presenting cells, such as macrophages and dendritic cells.

Overexpression of PD-L1 in murine tumor models results in inhibition of T-cell–mediated immune response via the PD-L1/PD-1 axis, indicating that blocking this axis has an important role in immunotherapy against cancers. PD-L1 is expressed in several solid tumors, including melanoma, glioblastoma, lung cancer, renal cancer, gastric cancer, colorectal cancer, pancreatic cancer, breast cancer, and cervical and ovarian cancers (please see Table 1 for percentage of PD-L1 expression in various tumors). PD-L1 is also found in hematologic malignancies, such as multiple myeloma, lymphoma, and various leukemia types. In contrast, PD-L2 is preferentially expressed in hematopoietic tumors, like B-cell lymphomas.¹⁰  Additionally, strong association between PD-1/PD-L1 expression and poor prognosis has been reported in several tumor types, including gastric, lung, and renal carcinomas (details in Figure 1).

Many pathways have been suggested to be involved in PD-L1 upregulation in cancers (as illustrated in Figure 2).⁶  For instance, activation of oncogenic signaling, such as through the phosphatase and tensin homolog–phosphoinositide 3-kinase–protein kinase B (PTEN-PI3K-AKT) pathway, upregulates PD-L1 on glioblastomas.¹¹  Similarly, activation of the anaplastic lymphoma kinase (ALK) signal transducer and activator of transcription 3 (STAT3) pathway is also able to induce PD-L1 expression in ALK-carrying T-cell lymphoma and chemoresistant non–small cell lung cancer (NSCLC).^12,13  In melanoma, the upregulation of PD-L1 is partially mediated by recruitment of histone deacetylase 6 (HDAC6) onto STAT3 to the PD-L1 promoter region.¹⁴  Furthermore, interferons, especially interferon γ, are the most common inducer in adaptive immune-mediated PD-L1 upregulation on tumor cells or on tumor-infiltrated leukocytes, including melanoma¹⁵  and ovarian cancer.^2,16  Other inflammatory signaling pathways, including interleukin 4, interleukin 10, vascular endothelial growth factor, granulocyte colony–stimulating factor, and bacterial lipopolysaccharide, have also been shown to induce the expression of PD-L1.¹⁷  This may have contributed to the association of chronic inflammation and cancer development.

Encouraged by the success of preclinical studies in which PD-1 blockade via monoclonal antibodies attenuated the remote metastasis of melanoma cells or colon cancer cells, several agents to block the PD-1/PD-L1 axis have been developed and applied in different types of tumors. The journey started from a small pilot phase 1 study in which nivolumab (a humanized anti–PD-1 antibody; Table 2), manufactured by Bristol-Myers Squibb (New York, New York), was shown to be safe and to have antitumor activity in treatment-refractory patients with solid tumors (melanomas, renal cell carcinomas [RCCs], and NSCLCs) in 2010. A large trial with 296 patients then revealed that nivolumab induced an objective response in 18% to 28% of patients with refractory solid tumors (26 of 94 in melanoma, 14 of 76 in NSCLC, and 9 of 33 in RCC), and the duration of the therapeutic response was greater than 1 year for most of these patients.⁵  Nivolumab was shown to be effective in refractory classical Hodgkin lymphoma, with an objective response rate of 87% (20 of 23) reported.¹⁸  Nivolumab soon became the first Food and Drug Administration (FDA)–approved PD-1 inhibitor for metastatic melanoma (2014), NSCLC (2014), and RCC (2015).

Another PD-1 antibody, pembrolizumab (lambrolizumab), manufactured by Merck (Kenilworth, New Jersey), was demonstrated to have antitumor effects in patients with advanced melanoma in an initial phase 1 study.¹⁹  In the KEYNOTE-006 study, the response rate was 33.7% (94 of 279) in patients with advanced melanoma receiving pembrolizumab every 2 weeks.²⁰  A 19.4% (96 of 495) objective response rate and 1 year of median duration of overall response were also reported in pembrolizumab-treated patients with metastatic NSCLC.²¹  Pembrolizumab became the second FDA-approved PD-1 inhibitor for metastatic melanoma and NSCLC. A recent report expanded the application of pembrolizumab further in advanced Merkel cell carcinoma. The objective response rate was as high as 56% (14 of 25) in patients with advanced Merkel cell carcinoma, including patients testing positive and those testing negative for Merkel cell polyomavirus.²²  Another promising PD-1 inhibitor, pidilizumab, manufactured by CureTech (Yavne, Israel), was tested in patients with solid tumors and hematologic tumors, particularly lymphomas.²³  More recent data from a study that evaluated the use of a combination of pidilizumab with rituximab in patients with relapsed follicular lymphoma showed a complete response rate of 52% (15 of 29) and a partial response rate of 14% (4 of 29).²⁴ 

PD-L1–targeted therapy is also currently under investigation for use in both solid and hematologic tumors.^25,26  Atezolizumab, manufactured by Roche/Genentech (Basel, Switzerland), was listed as the Breakthrough Drug of 2014 by the FDA, given its significance in treating metastatic bladder cancer. In a recent phase 1a study in patients with RCC, atezolizumab treatment resulted in a 28.9-month median overall survival, even in some patients previously exposed to anti–vascular endothelial growth factor therapy.²⁷  Successful results have also been reported using atezolizumab in the treatment of bladder cancer.²⁸  MEDI4736 is another exciting PD-L1 antibody that is currently being tested in phase 3 trials for stage 3 NSCLC. Another exciting, newly emerged antibody is durvalumab, manufactured by AstraZeneca (London, United Kingdom), which is a more PD-L1–specific inhibitor minimizing the toxicity associated with PD-L2 inhibition.²⁹  There was a phase 1b study conducted in which a dual-antibody therapy including durvalumab and tremelimumab, manufactured by AstraZeneca, was tested in patients with advanced-stage NSCLC; the therapy demonstrated an objective response of more than 20% (6 of 26) and disease control in more than 30% (9 of 26), in both PD-L1–positive and PD-L1–negative groups.³⁰  Table 2 summarizes anti–PD-1 and anti–PD-L1 agents in clinical use or in the early phases of development.

Furthermore, the synergistic benefits of the combination of PD-1/PD-L1 inhibitors with other agents—such as vaccine, other checkpoint inhibitors, chemotherapy, and targeted inhibitors (epidermal growth factor receptor [EGFR], ALK, BRAF, and vascular endothelial growth factor inhibitors)—have been confirmed by several preclinical tests. Recently, these novel drug combinations have been tested in several tumors in various phases of clinical trials.

The most common nonspecific systemic side effects of PD-1/PD-L1–targeted therapy are fatigue and fever. Imbalance of immune tolerance/immunity induced by PD-1/PD-L1–targeted therapy sometimes leads to side effects mimicking autoimmune diseases. Toxicity associated with PD-1/PD-L1–targeted therapy occurs with less frequency than the toxicity associated with anti–cytotoxic T-lymphocyte–associated protein 4 (anti-CTL4) therapy, and grades 3 to 4 adverse events occur in only 7% to 12% of patients. The most affected organs/tissues in these immune-related adverse events are the skin, gastrointestinal system, liver, and lungs. Maculopapular rash is the most common skin pathology. Diarrhea/colitis and asymptomatic elevation of alanine aminotransferase/aspartate aminotransferase are the most common digestive system side effects. The incidence of pneumonitis is more common in patients with lung cancer or those treated with combinations of PD-1 inhibitors and conventional chemotherapeutic agents. Patients with pneumonitis may present with shortness of breath, cough, fever, and/or chest pain, as well as the associated radiologic findings mimicking acute interstitial pneumonia or acute respiratory distress syndrome. The onset of pneumonitis is usually seen in the range of 7 to 24 months following the exposure to PD-1/PD-L1 inhibitors. In general, patients with mild symptoms are normally treated by discontinuing the PD-1/PD-L1 inhibitors. Steroids and other immunosuppressants are sometimes necessary for patients with severe symptoms.

The correlation between the PD-L1 expression of cancer cells and prognosis remains to be determined. As shown in Table 1, most of the studies have indicated that expression of PD-L1 in tumor cells served as an adverse prognostic marker. However, a few studies revealed that expression of PD-L1 in tumor cells may imply better prognosis. Additionally, several studies beyond those noted in Table 1 showed that PD-L1 expression in tumor cells and tumor-infiltrating lymphocytes is correlated with poor prognosis in such cancers as NSCLC,³¹  melanoma,³²  RCC,³³  and esophageal and gastric cancers.³⁴  A meta-analysis conducted by Wang et al³¹  in 1157 patients with NSCLC showed that PD-L1 expression was significantly associated with poor differentiation of tumors (poor versus well: odds ratio, 1.91; 95% CI, 1.33–2.75; P = .001) and with worse overall survival (pooled hazard ratio, 1.75; 95% CI, 1.40–2.20; P < .001).³¹  In another meta-analysis of RCC, high level of PD-L1 expression was a negative prognostic factor that increased the risk of death by 81% (hazard ratio, 1.81; 95% CI, 1.31–2.49; P < .001).³³  In gastric carcinoma, PD-L1 expression by immunohistochemistry was associated with larger tumor size, invasion into the deep muscular layers, lymph node metastasis, and decreased survival time of patients (<2 years).³⁴  Moreover, multivariate analysis demonstrated that PD-L1 immunodetection could be used as an independent factor to evaluate the prognosis of gastric carcinoma.³⁴  Additionally, PD-L1 expression has been associated with an aggressive melanoma phenotype characterized by a fibroblast-like morphology and invasive properties leading to enhanced aggressiveness and invasiveness.³²  Also, in a meta-analysis of breast cancer patients, PD-L1 expression was associated with aggressive features, such as lymph node metastasis, poor nuclear grade, and negative estrogen receptor status. It is associated with a higher total risk of mortality risk ratio of 1.64 (95% CI, 1.14–2.34) and a higher 10-year risk of mortality risk ratio of 2.53 (95% CI, 1.78–3.59) after surgery.³⁵  In contrast, there are some studies that showed that PD-L1 upregulation served as a positive prognostic marker in breast cancer and high-grade serous ovarian carcinoma. This is likely due to an increased T-cell cytotoxic immune response in these cancers.^36–38 

The mechanisms leading to these discrepancies are uncertain. However, the current use of nonstandardized immunohistochemistry (IHC) methodologies using different monoclonal antibodies (Tables 1 and 3) for measuring PD-L1 levels in these studies may have partially contributed to these conflicting results.

A predictive marker for predicting treatment response to anti–PD-1/PD-L1 treatment is highly desired because the treatment is associated with certain toxicity as described above, although less common and less severe than conventional chemotherapy. PD-L1 expression has been investigated as a potential predictive biomarker for selecting responders to anti–PD-1/PD-L1 antibody treatment. A study conducted by Topalian et al⁵  showed an objective response to nivolumab in 36% (9 of 25) of the cancer (advanced melanoma, NSCLC, castration-resistant prostate cancer, RCC, and colorectal cancer) patients who tested positive for PD-L1 (monoclonal antibody 5H1) expression, whereas none of the patients who tested negative for PD-L1 demonstrated a response. Similarly, another recent study showed that about 22% (191 of 824) of the NSCLC tumors (both squamous and nonsquamous) tested expressed PD-L1 in at least 50% of tumor cells, and these patients had a significantly higher response rate to anti–PD-1/PD-L1 treatment (41%), compared with those with PD-L1 expressed in less than 50% of tumor cells.²¹  The study data showed a clear link between PD-L1 expression and the efficacy of pembrolizumab for patients with NSCLC, which was also reflected in the indication for use by the FDA. The companion diagnostic PD-L1 IHC 22C3 pharmDx test (Dako North America Inc, Carpinteria, California) is indicated as an aid in identifying NSCLC patients for treatment with pembrolizumab.³²  By definition, a companion diagnostic test is required and essential for the safe and effective use of the corresponding targeted therapy.

Studies have shown that atezolizumab treatment results in a median overall survival of 28.9 months in patients with metastatic RCC, with benefit more obvious in the group with strong PD-L1 IHC staining (Table 3, SP142 assay).²⁷  Similar results were also reported using atezolizumab in treating bladder cancer, with a 61% overall reduction in tumor burden, a 26% (26 of 100) overall response rate, and a median overall survival of 11.4 months in the IC2/3 subgroup based on an IHC assay, compared with a 45% reduction in tumor burden, an 11% (11 of 107) objective response rate, and a median overall survival of 6.7 months in the IC1 subgroup (Table 3). Additionally, a systemic review and meta-analysis from 20 trials including patients with metastatic melanoma, NSCLC, and RCC receiving anti–PD-1/PD-L1 antibodies (4230 metastatic melanoma, 1417 NSCLC, and 312 RCC patients) showed that PD-L1 expression is associated with lower mortality and better clinical response to anti–PD-1/PD-L1 antibodies in patients with metastatic melanoma and is associated with a better clinical response in patients with nonsquamous NSCLC.³⁹ 

However, a study by Brahmer et al⁴⁰  observed that an objective response to nivolumab could still occur in patients with squamous carcinoma of lung who test negative for PD-L1 (PD-L1 antibody [clone 28-8, Dako North America]).⁴⁰  Consequently, nivolumab has recently received FDA and European Medicines Agency approval for NSCLC, regardless of the PD-L1 expression status (for NSCLC in the United States and in European Union). Of interest, the FDA might have had some reservations about the clinical data submitted for review. The FDA indicated in the premarket approval that patients with PD-L1 expression, as detected by the PD-L1 IHC 28-8 pharmDx assay (Dako North America) in previously treated metastatic nonsquamous NSCLC, may experience enhanced survival when treated with nivolumab. Currently, this assay has been used as a complementary test based on the clinical trial results.

Additionally, in a meta-analysis of cancer patients treated with Nivolumab, pembrolizumab or atezolizumab, Carbognin et al⁴¹  reported a clinical response in 239 of 702 patients (34%) with PD-L1–positive cancers and in 154 of 773 patients (20%) with PD-L1–negative cancers. The significant difference in response based on PD-L1 IHC was observed for NSCLC and melanoma but not for RCC or bladder carcinoma. Thus, using PD-L1 IHC as a predictive marker remains undetermined. Again, the use of different nonstandardized IHC techniques for measuring PD-L1 levels in tissue may have contributed to these differences.

Assessment of PD-L1 expression through immunohistochemical staining has been advocated as one potential biomarker, as discussed above. Immunohistochemistry has several advantages: (1) the wide availability of formalin-fixed, paraffin-embedded tissue; (2) visualization of expression in various cell populations (tumor versus immune/stromal cells) to some extent based on morphology; (3) relative rapidity of the test; and (4) the test's relatively low cost and widespread use in pathology laboratories, particularly in contrast to molecular pathology–based methods. Different clinical trials have used different IHC assays from different pharmaceutical manufacturers to measure PD-L1 expression (Table 3). These assays use different monoclonal antibody clones recognizing various epitopes of PD-L1. Various systems for amplification and detection of the signal are used for IHC, leading to different thresholds of detecting PD-L1 expression. Additionally, 3 of these assays evaluate the PD-L1 expression in the tumor cells of NSCLC only. The Ventana SP142 assay, manufactured by Spring Bioscience (Pleasanton, California), measures the PD-L1 expression in both tumor cells and tumor-infiltrating immune cells in metastatic uroepithelial cancers. Of note, studies using different PD-L1 antibodies to evaluate PD-L1 expression of tumor-infiltrating immune cells in various cancer types (head/neck squamous cell carcinoma, melanoma, and bladder cancers) are undergoing further evaluation to discern the impact of tumor-infiltrating immune cells in treatment response.

In 2015 a workshop by the Food and Drug Administration, the American Association for Cancer Research (AACR), and the American Society of Clinical Oncology led to a Blueprint Proposal developed by 4 pharmaceutical companies (Bristol-Myers Squibb Co, Merck & Co Inc, AstraZeneca PLC, and Genentech Inc), 2 diagnostic companies (Agilent Technologies Inc/Dako Corp and Roche/Ventana Medical Systems Inc), 2 professional societies (AACR–International Association for the Study of Lung Cancer), and 2 regulatory agencies (the European Medicines Agency and the FDA) to evaluate the analytic similarities of the 4 PD-L1 assays for use in NSCLC. The goal of this effort is to harmonize companion diagnostics for PD-L1 and to assess the possibility of interchangeable use of these assays. There are 2 phases in this proposal: phase 1 will evaluate analytic components by measuring PD-L1 expression on tumor or immune cells and predefine cutoffs in order to evaluate how these assays would compare using clinical samples; and phase 2 will design a statistically powered study with a large sample size based on the findings of phase 1. The preliminary data of phase 1, using 500 samples, was presented at the recent AACR annual meeting (April 16–20, 2016).⁴²  The preliminary results have indicated that 3 antibodies (22C3, 28-8, and SP263) have similar analytic performance in measuring the percentage of PD-L1–expressing tumor cells. The dynamic range reported for these 3 antibodies was between 1% and 100%. A fourth antibody, SP142, constantly labeled fewer tumor cells.⁴²  However, there is less precision in analytic performance when labeling immune cells compared with tumor cells. There is also less agreement between observers when evaluating immune cells compared with cancer cells. Additionally, the patient population defined by Ventana SP263, manufactured by Spring Bioscience, at the 25% cutoff point is similar to the group identified by the Dako 28-8 and Dako 22C3, manufactured by Dako, at the 1% cutoff. However, about 37% of the cases studied revealed discrepant results for PD-L1 expression between assays. This suggests the possibility of assignment into different diagnostic categories according to the key clinical cutoffs if assays and algorithms are mismatched. A recent study further suggested that the inherent tumor heterogeneity, or assay- or platform-specific variables may also contribute to the discordant results of these companion diagnostic tests.⁴³ 

These results suggest the challenges of extrapolating the results from one test to that of another test. This is reillustrated by Blueprint Chair Dr Fred Hirsch (professor of medicine at the University of Colorado), who said in an interview, “when pathologists used each assay in combination with its own prescribed cutoff, the assays did sometimes disagree on whether samples were PD-L1-positive.”⁴⁴  However, these results are preliminary and the planned phase 2 study is needed to further evaluate how to use these antibodies most efficiently in clinical practice. Additionally, other large-scale projects to investigate harmonization of PD-L1 antibodies are planned. For example, the National Comprehensive Cancer Network will soon begin a harmonization project in collaboration with Bristol Myers Squibb, MD Anderson Cancer Center, and Yale University. These results will be important for the efficient use of PD-L1 assay in pathology laboratories.

Several challenges remain regarding standardization of IHC beyond the Blueprint study. Preanalytically, the time for fixation in formaldehyde can modify the level of expression of PD-L1 and needs to be controlled. Additionally, there is heterogeneous PD-L1 expression in different regions of the same tumor specimen. Therefore, the absence of PD-L1 expression on small biopsies may not reflect the systemic immunologic landscape. This may have contributed to some patients responding to anti–PD-1 or anti–PD-L1 therapy independent of PD-L1 expression. Furthermore, it remains to be clarified whether the PD-L1 expression test should be performed on the primary tumor site or metastatic sites. PD-L1 expression in tumor cells is dynamic, influenced by interferon γ, hypoxia, and previous treatments, including chemotherapy, radiation therapy, and targeted therapy.⁴⁵  Additional comments regarding PD-L1 IHC as a predictive marker have been recently published.^46,47 

As shown in Table 4, the first study with human peripheral blood cells demonstrating the feasibility of detecting PD-L1 expression with a flow cytometry method was published in 2009.⁴⁸  The authors showed that cryopreservation actually decreased the expression of both PD-1 and PD-L1 in peripheral blood cells. Using a similar approach with different antibodies, a study showed a higher expression of PD-1/PD-L1 in CD4/CD8 T lymphocytes from chronic lymphocytic leukemia patients when compared with age-matched controls.⁴⁹  In addition to peripheral blood cells or lymphocytes, Gowrishankar et al¹⁵  have shown that interferon γ induces the upregulation of PD-L1 in a nuclear factor-κB–dependent fashion in 5 different melanoma cell lines and melanoma patient–derived cells using flow cytometry.¹⁵  Furthermore, Andorsky et al⁵⁰  documented that PD-L1 is widely expressed by anaplastic large cell lymphoma, whereas it only has limited expression in diffuse large B-cell lymphoma (DLBCL).

Flow cytometry studies, compared with IHC-based methods, have the advantage of simultaneously measuring the expression of PD-1 and PD-L1 in malignant cells and various types of immune cells. This is accomplished by using combinations of phenotypic markers for immune cells (cytotoxic T cells [CD8, TIA1], natural killer cells [CD56], T-reg cells [CD4, forkhead box P3], and dendritic cells/macrophages [CD68]), tumor cells (epithelial cell adhesion molecule, epithelial membrane antigen, lymphoma/leukemia markers), and antibodies to PD-1 and PD-L1. This will likely provide a more comprehensive understanding of PD-1/PD-L1 interactions between tumor cells and the immune environment. In theory, the efficacy of anti–PD-L1 treatment relies on the blockade of the inhibitory action of anticancer cytotoxic lymphocytes induced by the interaction between PD-1, expressed on immune cells (particularly lymphocytes), and PD-L1, expressed mainly on cancer cells. Therefore, it is reasonable to hypothesize that the higher the expression of PD-1 on immune cells, the more efficient the anti–PD-L1 treatment. This method may prove to be useful for predicting PD-1/PD-L1–targeted therapy, particularly for hematologic malignancies in which flow cytometry studies have been routinely used for diagnostic purposes. Similar to flow cytometry, another method that allows us to analyze PD-1 or PD-L1 expression is immunomagnetic selection using the CellTracks analyzer (Janssen Diagnostics LLC, Raritan, New Jersey). Using this method, researchers were able to identify the expression of PD-L1 in circulating tumor cells from patients with breast cancer with a sensitivity of 68.8%, indicating the promise of this noninvasive rapid screen in the future.⁵¹ 

Both SYBR dye–based and Taqman probe–based real-time polymerase chain reaction methods have been employed to detect PD-1/PD-L1 mRNA expression in whole-blood samples, cell lines, and surgical specimens.^52,53  In a recent report from Shen et al,⁵²  it was found that there is no uniform upregulation of PD-L1 expression in cancer cell lines compared with noncancer cell lines. However, authors identified a significant correlation between the expression of PD-L1 in human osteosarcoma and prognosis (median overall survival: 89 months in PD-L1–low patients versus 28 months in PD-L1–high patients).

An enzyme-linked immunosorbent assay–based method aimed at detecting soluble PD-L1 (sPD-L1) was developed in 2011.⁵⁴  Using this approach, researchers were able to demonstrate that sPD-L1 is a good prognostic marker in multiple myeloma and DLBCL.^55,56  In a small study with 81 multiple myeloma patients, the mean concentration of sPD-L1 was 2.851 ng/mL compared with 0.716 ng/mL sPD-L1 in matched controls.⁵⁵  Additionally, a lower sPD-L1 in patients with multiple myeloma is associated with higher progression-free survival.⁵⁵  In a multicenter, randomized phase 3 clinical trial in patients with DLBCL lymphoma, sPD-L1 was found to be higher in patients with DLBCL compared with healthy individuals, and the levels dropped after complete response. Moreover, cutoff of sPD-L1 at 1.52 ng/mL can be used to stratify patients with DLBCL into 2 groups with respect to 3-year overall survival. This results in 2 distinct populations: 1 with a 3-year overall survival of 76% and 1 with a 3-year overall survival of 89%.⁵⁶  These results support the possibility of using PD-1/PD-L1 blockade as a treatment option in these cancers.

Using a fluorescence in situ hybridization probe targeting the gene locus encoding PD-L1 located at 9p24.1, 1 study has shown that 9p24.1 amplification was associated with increased PD-L1 expression via the janus kinase 2 pathway in nodular sclerosing Hodgkin lymphoma and primary mediastinum large B-cell lymphoma.⁵⁷  Another study using fluorescence in situ hybridization demonstrated that the incidence of 9p24.1 amplification (at least ≥3 copies) was present in 24% of patients with early-stage Hodgkin lymphoma, compared with a 50% presence in advanced-stage (stages 3–4) Hodgkin lymphoma.⁵⁸  Additionally, the patients with Hodgkin lymphoma harboring 9p24.1 amplifications had a shorter progression-free survival. These results suggest that fluorescence in situ hybridization studies may be another potential method for predicting the response to PD-1/PD-L1 treatment.

The expression level of PD-L1 has been reported to be associated with other genetic alternations. The NSCLC cell lines with epidermal growth factor receptor (EGFR) mutations tend to have higher PD-L1 expression on the cell surface.⁵⁹  In a phase 2 trial studying pembrolizumab in multiple solid metastatic tumor, patients with mismatch repair–deficient (ie, microsatellite instability–high) colorectal cancer are more likely to benefit from PD-1 blockade (pembrolizomab) than those with mismatch repair–proficient tumors. Interestingly, PD-L1 expression is also elevated in mismatch repair–deficient colorectal cancer patients compared with those with mismatch repair–proficient tumors. However, the PD-L1 expression is not significantly associated with progression-free survival or overall survival.⁶⁰  In another study, higher nonsynonymous mutation burden was found to be associated with PD-L1–positive tumors and better melanoma-specific survival in patients with stage 3 melanoma. In addition, NF-1 mutation was shown to be limited to PD-L1–positive melanoma, but BRAF and NRAS mutations are distributed equally in PD-L1–positive or PD-L1–negative melanoma.⁶¹ 

PD-1/PD-L1–targeted therapy has demonstrated the impressive power of cancer immunotherapy. PD-1/PD-L1–targeted therapy either alone or in combination with other treatment modalities will benefit patients with advanced cancers. Because of the complexity of cancer immunity, we still have not yet completely understood the mechanism by which PD-1/PD-L1–targeted therapy improves survival in cancer patients. Many questions remain to be answered, particularly a better prediction system for predicting the response of PD-1/PD-L1–targeted therapy.