Reliability of PD-L1 as a Predictive Biomarker in NSCLC

Courtney C. Cavalieri, PharmD BCOP
Clinical Hematology/Oncology Pharmacist
Huntsman Cancer Institute at the University of Utah Salt Lake City, UT

Even through numerous developments in treatment, lung cancer re- mains the leading cause of cancer death in the United States.1 Many avenues have been attempted to prolong overall survival of this dis- ease. The most recent has been immunotherapy, specifically immune-checkpoint inhibitors targeted at the programmed death 1 (PD-1) receptor. Nivolumab and pembrolizumab are the two immunotherapies currently approved by the U.S. Food and Drug Administration (FDA) for the treatment of progressive non-small cell lung cancer (NSCLC). Although these agents may be similar in target, their respective journeys to approval revealed we have much to learn about the impact of these therapies.

Nivolumab was approved first for squamous NSCLC in March 2015 based on the CheckMate-017 phase 3 trial. This compared nivolumab to docetaxel in patients with advanced squamous NSCLC who had progressed on or after receiving a platinum-based chemotherapy regimen. Nivolumab provided an improved median overall survival (9.2 versus 6.0 months), response rate (20% versus 9%), and progression- free survival (3.5 versus 2.8 months) over docetaxel. Nivolumab also produced fewer grade 3 or 4 adverse events at 7%, compared with the 55% in the docetaxel group.2 In October 2015, the FDA extended the approval to non-squamous NSCLC based on the results of the CheckMate-057 phase 3 trial. CheckMate-057 compared nivolumab to docetaxel in patients with non-squamous NSCLC after progression during or after platinum-based chemotherapy. Again, nivolumab proved superior to docetaxel, with improved overall survival (12.2 versus 9.4 months) and response rates (19% versus 12%). The data for median progression-free survival did not actually favor nivolumab (2.3 versus 4.2 months). However, at 1 year, a higher percentage of patients treated with nivolumab were alive than those treated with docetaxel (19% versus 8%), representing a delay in response to therapy known to occur with immunotherapy. Nivolumab still proved to be better tolerated, with only 10% reporting grade 3 or 4 adverse events compared to the 54% in the docetaxel group.3

Pembrolizumab was approved for NSCLC in October 2015 based on the data from the phase I KEYNOTE-001 trial. The objectives of KEYNOTE-001 were to evaluate the side effects, safety, and anti- tumor activity of pembrolizumab in NSCLC patients, as well as de- fine and validate a tumor PD-L1 expression level associated with an increased likelihood of benefit from pembrolizumab. Pembrolizum- ab was given at a dose of 2 mg/ kg every 3 weeks, 10 mg/ kg every 3 weeks, or 10 mg/ kg every 2 weeks. PD-L1 positivity was determined by an immunohistochemical (IHC) assay and defined as at least 1% of cells showing membranous staining (proportion score). The investigators determined ≥50% proportion score as the cutoff for validation of PD-L1 as a predictive biomarker. The most common adverse events reported were fatigue, pruritus, and decreased appetite. The overall response rate was 19.4% across untreated and previously treated patients. Median overall survival was 12 months and progression-free survival was 3.7 months. No differences in response rates or toxicities were seen between the varying doses and schedules; therefore, the investigators recommend the 2 mg/ kg every 3 week schedule. Patients with ≥50% proportion score per the validation assay had longer progression-free survival than patients with 1% to 49% or <1% proportion scores (6.3 versus 4.1 versus 4.0 months, respectively) and longer over- all survival (not reached versus 10.6 versus 10.4 months, respectively). The FDA approval for pembrolizumab in NSCLC dictates patients must test positive for PD-L1 with the companion diagnostic assay.4

KEYNOTE-001 suggests that PD-L1 expression (at ≥50% proportion score) may represent a predictive biomarker for the treatment of NSCLC with pembrolizumab.4 However, the CheckMate trials provide conflicting reports on whether PD-L1 expression is a valid biomarker. CheckMate-017 defined samples as PD-L1 positive at pre- defined levels of 1%, 5%, and 10% and found no difference in predictive benefit of nivolumab activity.2 CheckMate-057 used the same IHC assay as CheckMate-017 and the same predefined positivity levels. At the interim analysis, an association between PD-L1 expression and clinical outcome was described. Expression above the predefined levels all correlated with better survival outcomes than docetaxel; however, in patients whose tumors had negative expression, the survival outcomes were similar to docetaxel. The authors concluded that because the safety profile of nivolumab out-performed docetaxel, nivolumab should still be considered an option regardless of PD-L1 expression.3

The question these three trials raise is whether PD-L1 is a reliable biomarker for predictive response to immunotherapy, particularly in NSCLC.

In the CheckMate trials assessing nivolumab in NSCLC, PD-L1 expression was relatively consistent at 83% in CheckMate-017 and 78% in CheckMate-057. 2,3 In KEYNOTE-001, 23.2% of patients had a pro- portion score of at least 50% and 37.6% had 1%-49%.4 Other trials exploring immunotherapies in NSCLC describe positivity for PD-L1 expression ranging from 21% to 95%. PD-L1 expression varies with disease state, with reports for melanoma patients ranging from 38% to 100% and reports for renal cell carcinoma ranging from 14% to 54%, depending on which site of disease was tested (primary versus metas- tasis).5 The wide range of positivity could be explained by each dis- ease state’s heterogeneity; however, trials have not been consistent in which IHC assay has been used to determine PD-L1 positive expression.

As of this time, there is no standard IHC assay used to calculate expression of PD-L1, nor is there a standard definition of “positive” expression. There are about two dozen anti-human PD-L1 antibodies currently being used in IHC assays, including 28-8, 5H1, MIH1, and 20C3. In addition, manufacturers of PD-1 and PD-L1 inhibitors con- currently develop their own proprietary companion test when seeking FDA approval. All these different assays complicate the ability to possibly standardize PD-L1 positive quantification. Cutoff points for positive PD-L1 expression range from >1% to >50%, which would explain the incidence of patients who are considered PD-L1 “positive” yet do not respond to immunotherapy as expected, and conversely, the patients considered PD-L1 “negative” who do respond to therapy.6 Al- though PD-L1 expression may not be the best biomarker to include or exclude patients to receive immunotherapy, levels of expression could possibly be used to guide which regimens of immunotherapy may benefit the patient the most. It is clear that patients who have higher rates of PD-L1 expression do respond better to single-agent immuno- therapy, and patients who have lower or negative rates of PD-L1 expression may be better suited to receive combination immunotherapy. Differing levels of expression could also be used to stratify patients in clinical trials exploring new combinations of immunotherapy.7

Although PD-L1 presents as a tempting predictive biomarker for immunotherapy in NSCLC, there are a few barriers before relying on PD-L1 as a definitive biomarker to choose which patients should or should not receive immunotherapy. Standardization must occur across assays as to which anti-PD-L1 antibody is used for expression description. The cut-offs for what constitute positivity must also be ad- dressed, and may be dependent on tumor type, biopsy site, and assay used to determine positivity. As immunotherapy becomes an option for more patients with oncologic diseases, these issues will hopefully be addressed through future clinical trials.


1. National Comprehensive Cancer Network Clinical Practice  Guidelines in Oncology: Non-Small Cell Lung Cancer. V1.2016. 

2. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer.  N Engl J Med. 2015;373:123–135.

3. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer. N Engl J Med. 2015;373:1627–1639. 

4. Garon EB, Rizvi NA, Hui R, et al. Pembrolizumab for the treatment of non-small cell lung cancer. N Engl J Med. 2015;372:2018–2028.

5. Patel SP, Kurzrock R. PD-L1 expression as a predictive biomarker in cancer immunotherapy. Mol Cancer Ther. 2015;14:847–856.

6. Teixido C, Karachaliou N, Gonzalez-Cao M, et al. Assays for predicting and monitoring responses to lung cancer immunotherapy. Cancer Biol Med. 2015;12:87–95.

7. Mahoney KM, Atkins MB. Prognostic and predictive markers for the new immunotherapies. Oncology. (Williston Park). 2014;28 (Suppl) 3:39–48.