Immunohistochemistry (IHC) with PD-L1 is in use to predict ICI response in NSCLC patients. Inter-observer, inter- and intra-laboratory variability is, however, a known issue. In predictive IHC and to a lesser extent in panel based prognostic IHC, quality assurance principles and methods are becoming more relevant for optimal scoring. To improve scoring of PD-L1, artificial intelligence (AI) might be the way forward. Before AI can be used for daily practice, however, laboratories need to make sure that their IHC is a consistent assay instead of a routine stain.

Nils ‘t Hart¹, Lars Gal¹, Morten Grønning Nielsen² and Mateusz Tylicki²

1. Isala, Zwolle, The Netherlands
2. Visiopharm, Hørsholm, Denmark