Recent clinical trials of monoclonal antibodies that target the PD-1/PDL-1 pathway in renal cancer have demonstrated promising clinical efficacy with benefit persisting off therapy in a subset of patients.1–4 In a phase I trial of the anti-PD-1 blocking antibody nivolumab (Bristol-Myers Squibb),2 10 out of 34 patients with renal cancer experienced major tumor responses, and nine others had stable disease lasting at least 24 weeks.4 In the 16 patients treated at the 10mg/kg dose level, almost 70% were progression-free at 6 months, and four patients have yet to progress despite being off therapy for more than 16 weeks (Fig. 1). Another anti-PD-1 antibody, lambrolizumab (Merck and Co., Inc.), showed similar efficacy to nivolumab in patients with melanoma and patients with lung cancer, and is now being studied in other cancers, including renal.5 In this phase I study reported by Cho et al., the PDL-1 antibody MPDL3280A (Genentech) was administered to 53 patients with metastatic renal cancer resulting in a 13% incidence of drug-related grade 3/4 adverse event, a response rate of 13%, and 50% of patients progression-free at 6 months. Although the early results seen with PD-1/PDL-1-blocking antibodies in solid tumors, including renal cancer, have been encouraging, more work is needed to optimize the use of PD-1-based immunotherapy.
In an attempt to target the application of PD-1 blocking antibody (nivolumab) therapy in phase III trials, Topalian et al. identified tumor cell surface PDL-1 expression on primary tumor specimens as a potential predictive biomarker.2 However, data from several recent studies suggest that immunohistochemical staining for PDL-1 alone fails to accurately identify all responders to PD-1/PDL-1 blocking antibodies and that narrowing the population of patients who should receive this therapy will be challenging (Table 1).3,6 Critical unanswered translational research questions that will influence the development of PD-1 pathway inhibitors include: (1) What factors, in addition to PDL-1 expression (e.g., PDL-2), can reliably predict response to treatment? (2) What are the mechanisms of innate resistance to PD-1 pathway blockade in PDL-1 positive tumors? (3) Is PDL-1 expression uniform and stable in an individual tumor and between a primary tumor and a metastasis?
Correlation of PDL-1 Expression and Response to PD-1 Pathway Blockade
The reasons why PDL-1 negative tumors respond to PD-1 blockade are poorly understood but may include: (1) heterogeneous PDL-1 expression in the primary tumor or between primary and metastases leading to false negative results when PDL-1 is evaluated in a single portion of the nephrectomy specimen; (2) upregulation of PDL-1 in metastases following VEGFR-TKI therapy; (3) diagnostic antibodies that lack adequate sensitivity or specificity; (4) PDL-1 expression in the tumor microenvironment (e.g., on macrophages and myeloid-derived suppressor cells) rather on than tumor cells; and (5) expression of PDL-2 alone. In PDL-1-negative tumors that respond to treatment, PDL-2 expression may provide PD-1 mediated pro-tumor inhibitory signals, which can be overcome with a PD-1 blocking antibody.7 Since PDL-1 and PDL-2 are regulated by different cytokines, it is likely that their expression in RCC is not completely congruent.8 Many of these hypotheses are currently being explored in prospective clinical trials and translational pathology studies.
In contrast, there are several mechanisms that might mediate “innate resistance” to PD-1 pathway blockade in PDL-1-positive tumors; each requiring different strategies to overcome. For example, resistant PDL-1-positive tumors may require the simultaneous blockade of alternative pathways that negatively regulate the immune response (e.g., galectin-9). In tumors where exhausted or inhibitory T-cells predominate, approaches that activate T-cells (e.g., interleukin-2), target T-cell inhibitory molecules (e.g., T-cell membrane protein 3 and lymphocyte-activation gene 3)8; or deplete regulatory T-cells in the tumor microenvironment (e.g., CTLA-4 blockade) may be necessary.
Alternatively, it is possible that in some tumors PDL-1 expression is driven by either genetic events (e.g., amplifications of the PDL-1 locus seen in ∼2% of ccRCC in the TCGA dataset) or activation of an oncogenic pathway (e.g., PTEN loss).9 In these tumors, PDL-1 expression is not the result of IFNγ release by tumor-infiltrating lymphocytes and thus would not predict for the presence of tumor-infiltrating lymphocytes and consequent response to PD-1 blocking antibody. These tumors might respond more effectively to molecularly targeted approaches. The impressive clinical effect of inhibiting both the CTLA-4 and PD-1 signaling pathways recently reported in patients with melanoma suggests combination therapy may have merit and has led to the exploration of these approaches in other solid tumors, including RCC.10
In the initial phase I trial of ipilimumab and nivolumab, responses were seen with equal frequency in PDL-1-positive and negative tumors, suggesting that the addition of anti-CTLA-4 may alter factors in the tumor microenvironment making PDL-1-negative tumors more susceptible to anti-PD-1 blockade.11 In future studies, comparative analysis of pretreatment samples from responding and refractory patients might enable the identification of tumor and stromal factors contributing to innate resistance to PD-1 blockade and contribute to the development of rational PD-1-based combination strategies.
Predictive tissue biomarker research is usually conducted by analyzing the primary tumor because it is easier to obtain. However, given the significant tumor heterogeneity seen in renal cancers, nephrectomy specimens may not accurately reflect the biology that is being targeted by systemic therapy.12 For this reason, it may be necessary to evaluate expression of PDL-1 and other markers of immune activation and dysfunction in different areas of the primary tumors; as well as in their corresponding metastases. These studies will determine whether predictive biomarkers analyses can be conducted on primary tumor tissue or whether tissue from metastatic sites is more informative.
At the 2014 Genitourinary Cancers Symposium, Callea et al. report on a preliminary analysis of whole tumor tissue sections from 34 primary clear cell renal cancers (ccRCC) (nephrectomy specimens), and their corresponding metastases (surgical excisions) revealed that a subset of ccRCCs express PDL-1 at the cell membrane in tumor cell and/or inflammatory cells. Membranous staining in tumor cells was observed in 10 out of 34 (29%) primary tumors. In seven of these 10 cases (70%) the metastases were also positive. In three cases in which the primary tumor was only focally positive (<5% of positive tumor cells), the metastases were negative. In one case, the primary tumor was negative but the metastasis was positive (Fig. 2). The pattern of PDL-1 staining was highly heterogeneous in the primary tumors and tended to be restricted to areas of highest Fuhrman nuclear grade. The staining was homogeneous in most metastases. In addition, the percentage of PDL-1-positive tumor cells was higher in the metastases compared to the corresponding primary tumors. The effect of tumor heterogeneity may make the development of predictive biomarkers more challenging in RCC.
At the present time, it is unclear why some patients with renal cancer respond to PD-1 pathway blockade while many do not. Significant efforts in preclinical and translational research will be necessary to ensure the optimal targeting of PD-1 pathway blockade in patients with renal cancer.