Three Assays and Several Laboratory-Developed Tests Highly Concordant Across Platforms in PD-L1 Immunohistochemistry Testing for NSCLC

Get Permission

Caution is required for validation and further use of laboratory-developed tests, but these results will provide the basis for national (French) recommendations on PD-L1 testing in NSCLC.
— Julien Adam, MD, PhD

Tweet this quote

Programmed cell death ligand 1 (PD-L1) expression assessed by immunohistochemistry in formalin-fixed, paraffin-embedded tumor tissues is currently the main predictive biomarker for the benefit of anti–programmed cell death protein 1 (PD-1) and anti–PD-L1 agents in patients with non–small cell lung cancer (NSCLC). However, different assays—marketed by Dako (22C3, 28-8) and Ventana (SP142, SP263)—have been used as diagnostic tests in various clinical trials. According to Julien Adam, MD, PhD, a pathologist at Gustave Roussy Cancer Center in Villejuif, France, the harmonization of assays and laboratory-developed tests is urgently needed.1

PD-L1 testing should be rapidly available to patients in the first-line setting, Dr. Adam maintained. Dako and/or Ventana platforms are not available in all pathology laboratories, and multiple tests with different assays do not produce reliable results in small NSCLC samples, he pointed out at the 2016 World Conference on Lung Cancer in Vienna.

Dr. Adam and his colleagues aimed to evaluate the analytic performance of the Dako 22C3, Dako 28-8, and Ventana SP263 PD-L1 assays across various centers and to determine if laboratory-developed tests can achieve an analytic performance close to PD-L1 assays in a set of NSCLC cases. The aim was to harmonize PD-L1 testing and make it widely available on most immunohistochemistry platforms and at most centers. The SP142 assay was not examined in this study.

Study Methods

In 7 centers, immunohistochemistry with 5 anti–PD-L1 clones (28-8, 22C3, E1L3N, SP142, and SP263) was performed concomitantly on 41 NSCLC surgical specimens, selected to have various expression levels of PD-L1. The immunohistochemistry platforms used were Ventana BenchMark Ultra (2 centers), Leica Bond (2 centers) or Dako Autostainer Link 48 (3 centers). For each matching platform, 22C3, 28-8, and SP263 assays were performed.

For nonmatching platforms and other antibodies, laboratory-developed tests were used in each center, and tonsil tissue and reference pictures from NSCLC cases stained with 28-8 and 22C3 assays were used for harmonization.

A total of 35 stainings (30 protocols) were performed on 41 cases (1,435 slides) across different platforms and antibodies for each case. Seven trained thoracic pathologists, blinded from the centers, participated in the study. Each pathologist analyzed six cases and scored the proportion of tumor cells and immune cells that were positive for PD-L1. Tumor cell and immune cell PD-L1 staining was scored semiquantitatively, as recommended in PD-L1 Dako and Ventana assays.

Key Results

Across several centers, the analytic performance results of Dako (28-8, 22C3) and Ventana (SP263) assays were highly concordant for tumor cell and immune cell stainings across the 5 Dako or Ventana platforms, with an overall agreement for ≥ 50% threshold of 95.1%. For each antibody, laboratory-developed tests were compared to those three reference assays, and the tests demonstrated various levels of concordance.

Lung Cancer Diagnostic Testing

  • PD-L1 expression assessed by immunohistochemistry is the main predictive biomarker for the benefit of anti–PD-1/PD-L1 agents in NSCLC patients, but different assays are used in different centers, and harmonization is needed
  • French investigators evaluated the analytic performance of 3 PD-L1 assays (Dako 22C3, Dako 28-8, and Ventana SP263) across various centers to determine if laboratory-developed tests can achieve an analytic performance close to PD-L1 assays in a set of NSCLC cases.
  • 22C3, 28-8, and SP263 assays were highly concordant across several centers, and the SP263 clone achieved the highest concordance rate across all platforms.

Among 27 laboratory-developed tests in 7 centers on Dako, Ventana, and Leica platforms, 14 (51.8%) demonstrated similar concordance as compared to reference assays for tumor cell staining. All laboratory-developed tests using the SP263 clone were highly concordant across all platforms for both immune cell and tumor cell ­stainings.

“In evaluating immune cells, we found poor concordance between assays and laboratory-developed tests,” said Dr. Adam. “Various factors may be responsible for this, such as the fact that the intensity of staining was lower and the number of categories was higher in immune cells, and we may have some intraobserver variabilities, since the assessment of PD-L1 staining is more difficult in immune cells than in tumor cells,” he noted.

“When looking at the overall concordance of all tests performed with each antibody, we found that tests performed with the SP263 clone had the highest concordance for immune cell and tumor cell staining,” he said. “Caution is required for validation and further use of laboratory-developed tests, but these results will provide the basis for national (French) recommendations on PD-L1 testing in NSCLC.” ■

Disclosure: Dr. Adam reported no potential conflicts of interest.


1. Adam J, Rouquette D, Damotte D, et al: Multicentric French Harmonization Study for PD-L1 IHC Testing in NSCLC. 2016 World Conference on Lung Cancer. Abstract PL04a.04. Presented December 7, 2016.

Related Articles

Expert Point of View: Michael Boyer, MD

Close collaboration between treating clinicians and pathologists is required in order to understand the performance of each laboratory, and also so that we understand these results can actually help inform the choice of patients.
— Michael Boyer, MD

Tweet this quote




By continuing to browse this site you permit us and our partners to place identification cookies on your browser and agree to our use of cookies to identify you for marketing. Read our Privacy Policy to learn more.