Whole Slide Image vs Microscope Glass Slide Performance for the Evaluation of Programmed Death-Ligand 1 (PD-L1) Expression in 14 Tumor Types Using PD-L1 IHC 22C3 pharmDx.

Adams, Micki; Moquin, Deanna; Tabuena-Frolli, Siena; Ruvalcaba-Rodarte, Jim; Morones, Amber; Marczak, Ryan; Simmons, Epiphani; Hund, Stephanie · Mod Pathol · 2025

Where this comes from

Abstract

Agilent Technologies Inc investigated the concordance of programmed death-ligand 1 (PD-L1) expression level results (binary: positive/negative) recorded by qualified study readers using microscope glass slide (MGS) and Aperio AT2 scanner-generated whole slide image (WSI) scoring for 14 tumor types and cutoffs. All study readers were qualified by completing Agilent training and certification testing specific to the tumor types and cutoffs they scored for the study. Formalin-fixed, paraffin-embedded specimens were stained using the qualitative immunohistochemical assay, PD-L1 immunohistochemical 22C3 pharmDx, on Autostainer Link 48 and scored using the Tumor Proportion Score or Combined Positive Score algorithm at predefined cutoffs. The objective was to demonstrate comparable performance of scoring for MGS and WSI. Three study readers determined the numerical score and the corresponding PD-L1 expression level for specimens of each tumor type using (1) MGS and (2) WSI viewed with Aperio ImageScope with a ≥14-day washout period between MGS and WSI reads. Agreement between MGS (reference condition) and WSI was evaluated for each individual tumor type and cutoff. Concordance correlation coefficient analysis was also performed on individual tumor type MGS and WSI numerical score data. MGS and WSI percent agreement results for all individual tumor types and cutoffs were ≥89%. Concordance correlation coefficient value results for all tumor types and cutoffs were ≥0.80. The high concordance and the low variability and bias between scoring modalities demonstrate comparable performance between MGS and WSI scoring for PD-L1 expression evaluation across multiple tumor types and cutoffs.

Medical subject headings