Patients with recurrent and metastatic non-small cell lung cancer (NSCLC) and head and neck squamous cell carcinoma (HNSCC) have considerable unmet medical needs and require improved treatments with the potential to provide prolonged survival and reduced exposure to toxic chemotherapies [3–5].
It has long been recognized that the immune system is capable of recognizing cancer as foreign and developing a specific immune response. A number of stimulatory and opposing inhibitory proteins have been shown to regulate the quality and magnitude of the antitumor response [6–9]. For example, tumors can evade detection by the immune system by exploiting one or more of the inhibitory (checkpoint) pathways that suppress antitumor T-cell responses [9]. The therapeutic blockade of these checkpoint molecules is actively being explored across different cancers [6, 7]. One of the most promising approaches is antibody blockade of the programmed cell death-1 (PD-1)/ programmed cell death ligand-1 (PD-L1) pathway [6, 7]. The anti-PD-1 agents nivolumab and pembrolizumab have demonstrated clinical activity across several tumor types and are now approved for various indications worldwide [10–13]. While dramatic responses have been seen in a number of patients treated with anti-PD-1 and anti-PD-L1 antibodies, responses have typically represented only a fraction of treated patients. This has led to a further search for biomarkers predictive of response. Chief among these is the demonstration of PD-L1 in tumor tissues using immunohistochemistry (IHC). Two different IHC assays have so far been approved by the Food and Drug Administration (FDA) for use as diagnostic tests in advanced NSCLC; a complementary diagnostic for nivolumab [14] and a companion diagnostic for pembrolizumab [15].
PD-L1 (B7-H1, CD274) is part of a complex system of receptors and ligands that are involved in regulating T-cell activation. Its main function is to regulate the balance between T-cell activation and tolerance through interaction with its two receptors, PD-1 (CD279) and CD80 (B7-1). In normal tissue, PD-L1 has been reported to be expressed on a subset of T cells, B cells, dendritic cells, macrophages, mesenchymal stem cells, and bone marrow-derived mast cells, as well as various non-hematopoietic cells [16, 17]. PD-L1 is expressed in a limited set of normal epithelial cells, such as placental trophoblast cells and crypt epithelium of the tonsil [18]. Its expression in these locations is thought to provide protection from cell-mediated attack. Importantly, and in a similar fashion, PD-L1 is also expressed in a broad range of carcinomas and other cancers [19–22] including NSCLC and HNSCC [9, 16]. In the tumor microenvironment, PD-L1 expressed on tumor cells binds to PD-1 on activated T cells reaching the tumor. This delivers an inhibitory signal to those T cells, preventing them from killing target tumor cells, and protecting the tumor from immune elimination [9].
PD-L1 biology is further complicated in additional ways. For example, full-length PD-L1 is composed of an extracellular domain that contains the PD-1 binding domain, a shorter intracellular domain, and a short transmembrane region. A number of antibodies to the extracellular and intracellular domains have been generated and several of those appear useful as IHC reagents [22–24]. It is also the case that splice variants of PD-L1 have been identified [25, 26] and individual isoforms may localize to the cytoplasm or membrane of cells in vitro. It should be noted, however, that particular forms of the PD-L1 protein may be recognized by some but not all of the other available antibodies. Cytoplasmic or membrane immunolabeling of neoplastic cells and macrophages have also been shown in tumor tissues [19, 27]. PD-L1 localization to the cell membrane is likely required for interaction with PD-1. In addition, PD-L1 demonstrates a range of apparent expression levels in cells, whether cytoplasmic or membranous. Tumor cells or immune cells with variable intensities of PD-L1 expression could therefore suppress PD-1-expressing T lymphocytes to different degrees, although this has not yet been proven. In addition to the biological heterogeneity of PD-L1 expression (notably intra- and inter-tumoral and temporal variations), [28, 29] there is also technical variability due to a number of different assays (developed for diagnostic or research use) being used to label PD-L1 in tumor tissue. These have variable target specificity and selectivity (being raised to different epitopes on the PD-L1 molecule), and use different methods for antibody detection and approaches to determining PD-L1 expression (scoring systems and staining cut-offs) [28].
Not surprisingly then, given the various complexities noted, substantial discordance is seen among published studies relying on PD-L1 IHC [22–24]. Nonetheless, the ability of a PD-L1 IHC method to account for those key aspects of PD-L1 biology and tissue expression relevant to therapy with either PD-1 or PD-L1 antibodies is expected to substantially contribute the overall usefulness of the assay. In that context, for example, an assay that demonstrates distinct cell membrane staining and a dynamic range of PD-L1 expression is highly desired. In addition, the ability to optimally label PD-L1 on neoplastic cells as well as immune cells is important, as is the ability to perform across the range of tissue specimens upon which the assay is expected to be applied. These qualities are especially needed to increase the ability to accurately identify patients who may respond to immunotherapies targeting PD-L1 or PD-1. At the same time, they contribute to a more comprehensive and potentially accurate assessment of PD-L1 that reflects the complexities of its expression in a range of tumor types and that may be required to clarify aspects of its immunosuppressive role in cancer that are not yet fully understood.
Durvalumab (MEDI4736) is a selective, high-affinity human IgG1 monoclonal antibody that blocks PD-L1 binding to PD-1 and CD80 but does not bind to programmed cell death ligand-2 (PD-L2) [30]. PD-L2 plays a role in controlling inflammation in normal lung tissue (with expression in lung macrophages and antigen-presenting cells) [31] and this may help to avoid PD-L2-mediated immune-related toxicities, which have been observed in animal models [32, 33]. In an ongoing Phase 1/2 study of patients with advanced solid tumors, durvalumab monotherapy has demonstrated a manageable tolerability profile and encouraging antitumor activity across multiple tumor types, including NSCLC and HNSCC [34, 35].
Most NSCLC tumors do not express PD-L1 at high levels, although reported levels vary; [20, 36, 37] only ~20 % of NSCLC tumors were reported to show PD-L1 expression in 5 % or more of tumor cells obtained with three different IHC assays (using different PD-L1 antibody clones; E1L3N, 22C3, and 5H1) [20, 38, 39]. In HNSCC, tissue PD-L1 immunostaining increases vs. normal controls [27] with variable levels of PD-L1 expression [18, 35, 40]. The absence of detectable PD-L1 in some tumors may account for poor responses to anti-PD-1/PD-L1 therapy. A sensitivity analysis of the clinical activity of nivolumab, pembrolizumab, and the anti-PD-L1 agent atezolizumab in NSCLC [41–48] demonstrated that overall response rates were significantly lower in patients with PD-L1 low tumors (1–5 % tumor cells stained using various PD-L1 assays) than in patients with PD-L1 high tumors [42]. Evidence from some individual clinical studies of the anti-PD-1/PD-L1 agents nivolumab, durvalumab, pembrolizumab, atezolizumab, and avelumab also suggest that patients with PD-L1 high tumors can experience improved treatment benefits vs. those with PD-L1 low tumors [34, 43, 47, 49, 50].
The development and application of diagnostic tests in clinical practice to identify patients most likely to benefit from anti-PD-1/PD-L1 therapy could improve patient outcomes and decrease healthcare costs, while directing patients that are less likely to respond towards other alternative treatment options. Research into the relative performance of different PD-L1 assays and their reliability for detecting PD-L1 is ongoing [51]. These diagnostic tests, which are designed to aid treatment decisions with specific anti-PD-1 therapies [23, 29, 52], are likely to be among the key drivers for the future of personalized health care in oncology [23].
In this paper, we describe the development and validation of the VENTANA PD-L1 (SP263) Assay (PD-L1 [SP263]) (Ventana Medical Systems Inc., Tucson, AZ, USA), which has been designed for the detection of PD-L1 protein in formalin-fixed, paraffin-embedded (FFPE) NSCLC and HNSCC tumor samples. In addition, we describe the identification and validation of scoring criteria (defined with tissues from a durvalumab clinical trial) that can be used to classify samples as PD-L1 high expression or PD-L1 low/no expression in NSCLC and HNSCC tissue.