PD-L1 testing of non-small cell lung cancer using different antibodies and platforms: a Swiss cross-validation study

With the approval of pembrolizumab for first- and second-line treatment of PD-L1+ non-small cell lung cancer (NSCLC), PD-L1 testing by immunohistochemistry (IHC) has become a necessity. However, the DAKO autostainer ASL48 for the FDA approved DAKO 22C3 pharmDx assay is not broadly available in Switzerland and other parts of Europe. The primary goal of this study was to cross-validate the 22C3 anti-PD-L1 antibody on Benchmark Ultra (VBMU) and Leica Bond (LBO) immunostainers. IHC protocols were developed for 22C3 on both platforms with the 22C3phDx using ASL48 as reference. A tissue microarray (TMA) was constructed from 23 NSCLC specimens with a range of PD-L1 staining results. Empty TMA sections and the 22C3 antibody were distributed to 16 participants for staining on VBMU (8 centers) and/or LBO (12 centers) using the centrally developed protocols. Additionally the performance of the Ventana SP263 assay was tested in five centers. IHC scoring was performed centrally. Categorical PD-L1 staining (0-49% vs. 50-100%) did not significantly differ between centers using VBMU, whereas data from LBO were highly variable (p < 0.001). The SP263 assay was well concordant with 22C3 on VBMU and with 22C3 pharmDx. PD-L1 IHC using a standardized 22C3 protocol on VBMU provides satisfactory results in most centers. The SP263 assay is confirmed as a valid alternative to 22C3 pharmDx. 22C3 PD-L1 IHC on LBO shows major staining variability between centers, highlighting the need for local validation and adjustment of protocols