Going Beyond PD-L1 for Predictive Biomarkers in Lung Cancer

Gina Battaglia
Published: Monday, Aug 15, 2016

Roy Herbst, MD, PhD

Roy Herbst, MD, PhD

Tumor expression of PD-L1 has been the conventional biomarker to predict response to immune checkpoint blockade. However, additional biomarkers, such as immune cell PD-L1 expression, mutational burden, and immune system activation should be investigated further to clarify the mechanisms behind response and nonresponse to immunotherapy in patients with lung cancer, according to Roy Herbst, MD, PhD.

Herbst, chief of Medical Oncology and associate director of Translational Research at the Yale Cancer Center, Yale School of Medicine, summarized recent data on potential immunotherapy biomarkers at the 2016 International Lung Cancer Congress.

According to Herbst, tumor PD-L1 expression is currently the best biomarker to predict response to anti–PD-1 therapy. However, multiple issues, including heterogeneity of PD-L1 expression within a tumor and the discrepancies in criteria for defining PD-L1 positivity among studies and laboratories, indicate that PD-L1 status is likely not useful as the sole biomarker. Furthermore, clinical trials vary in the cutoffs for defining PD-L1–positive tumors, which, according to Herbst, may contribute to the conflicting results observed.

The KEYNOTE-024 trial, which showed better progression-free survival (PFS) and overall survival (OS) with pembrolizumab (Keytruda) over chemotherapy in patients with non–small cell lung cancer (NSCLC), had a tumor PD-L1 cutoff of ≥50%, whereas the CheckMate-026 trial, which did not meet the primary endpoint for PFS with nivolumab (Opdivo) in a similar setting, had a cutoff of ≥5%. However, the KEYNOTE 010 phase 2/3 trial showed better OS with pembrolizumab than with docetaxel in patients with NSCLC and PD-L1 tumor proportion score ≥1%, which, according to Herbst, supports this lower cutoff when identifying PD-L1 positivity in patients’ tumors.

Although initial focus has been on PD-L1 expression in the tumor, PD-L1 is also localized with macrophages, dendritic cells, and T cells. Herbst presented data from a phase I study showing that the level of immune cell PD-L1 expression was positively associated with response to atezolizumab (Tecentriq), suggesting that immune cell PD-L1 expression also plays a role in response to anti–PD-1 therapy and should be included when determining PD-L1 status.

Herbst noted that factors outside of the tumor, such as the inflammatory state in the tumor microenvironment, likely play a role in the response to immunotherapy. A retrospective analysis of NSCLC tumor tissue samples showed that 26% tested negative for PD-L1 expression but stained positive for tumor-infiltrating lymphocytes (TILs). Furthermore, 57% of the samples tested negative for TILs, which, according to Herbst, emphasizes the importance of investigating the tumor microenvironment and activation of TILs when predicting response to immunotherapy and identifying additional immune checkpoints.

Herbst also noted that characterizing patterns of immune cell response to checkpoint inhibitors in responders and nonresponders is also instrumental to describing the mechanisms of immune response and aiding in development of future clinical trials. He and his colleagues showed that responders to pembrolizumab demonstrated increases in gene expression of several enzymes involved in T cell activity, including granzymes A and B, perforin, interferon-gamma, and tumor necrosis factor, after 4 weeks of therapy.

His study also observed 3 distinct immunologic patterns in nonresponders: immunologic ignorance (no T cells before or after therapy), nonfunctional immune response (presence of PD-L1 and CD8+ T cells, but no activation of T cells with immunotherapy), and excluded infiltrate (T cells that cluster on the outer edge of the tumor cell mass). Herbst indicated that upcoming trials, 1 as one investigating pembrolizumab with ramucirumab (Cyramza), may provide insight on whether combination therapies may influence the immunologic response in the tumor microenvironment.

The large number of neoantigens associated with a high mutational burden, as seen in many lung cancers, upregulates the immune system and increases inflammation within the tumor. According to Herbst, this may contribute to an improved response to immune checkpoint blockade. Herbst presented retrospective data showing that high mutational burden was associated with improved response to pembrolizumab. However, he cautioned that prospective studies are needed to determine the usefulness of tumor mutational burden as a predictive biomarker for immunotherapy.

View Conference Coverage
Online CME Activities
TitleExpiration DateCME Credits
Community Practice Connections™: 18th Annual International Lung Cancer Congress®Oct 31, 20181.5
Clinical Interchange™: Translating Research to Inform Changing Paradigms: Assessment of Emerging Immuno-Oncology Strategies and Combinations across Lung, Head and Neck, and Bladder CancersOct 31, 20182.0
Publication Bottom Border
Border Publication