Optimization and validation of PD-L1 immunohistochemistry staining protocols using the antibody clone 28-8 on...

Several immunohistochemistry (IHC) assays have been developed to assess tumor programmed death-ligand 1 (PD-L1) expression levels in patients who are candidates for programmed death-1 (PD-1)/PD-L1 inhibitor therapy. The PD-L1 IHC 28-8 pharmDx kit is FDA-approved as a complementary diagnostic and CE-marked as an in vitro diagnostic device for nivolumab therapy in melanoma and specific lung cancer subtypes (and for squamous cell carcinoma of the head and neck/urothelial carcinoma in Europe only). Kit availability is limited outside the United States, and its use requires the Dako Autostainer Link 48 platform, which is unavailable in many laboratories. Validated laboratory-developed tests based on 28-8 concentrated antibody outside the kit are needed. This study compared the results from PD-L1 expression level analysis across four immunohistochemistry platforms (Dako Autostainer Link 48, Dako Omnis, Leica Bond-III, and Ventana BenchMark ULTRA) with the 28-8 pharmDx kit in lung cancer (multiple histologies), melanoma, and head and neck cancer (multiple histologies). Samples were prepared per protocol for each platform and stained using PD-L1 IHC 28-8 pharmDx kit on Dako Autostainer Link 48, and per protocol for each platform. The control samples (tonsil and placenta tissue; cell lines with prespecified PD-L1 expression levels) were tested to evaluate the specificity and the sensitivity of test assays. An agreement level of 0.90 with the pharmDx kit was set for each platform. Inter- and intra-assay reliability were assessed. Evaluable samples were lung cancer = 29; melanoma = 31; head and neck cancer = 30. Mean agreement was calculated for PD-L1 expression levels of ≥1%, ≥5%, ≥10%, and ≥50%. Mean overall agreement for all indications was 0.87-0.99. Inter- and intra-assay of scoring/classification repeatability was 100%. Analysis of PD-L1 expression levels using laboratory-developed immunohistochemistry assays with 28-8 antibody may be permissible if the platform is validated using reference samples with defined expression levels.