In this study, our aim was to evaluate the discordance of programmed cell death ligand 1 (PD-L1) expression between 2 paired paraffin blocks from surgical nonsmall cell lung cancer (NSCLC) specimens, thus providing recommendations for choosing paraffin blocks for PD-L1 immunohistochemistry detection. A total of 460 paired paraffin blocks of surgical NSCLC specimens were analyzed using the 22C3-PD-L1 assay. PD-L1 expression between 2 paired paraffin blocks was calculated using 3 classification schemes, tumor proportion score (TPS)-based 3-level classification and TPS-based binary classification using TPS ≥1% or TPS ≥50% as the cutoff. Clinicopathological characteristics were analyzed for cases with discordant PD-L1 expression. PD-L1 expression in 2 paired paraffin blocks of surgical NSCLC specimens was highly correlated from a single tumor ( R2 =0.89) in the overall trend. The overall discordance rates were 10.9%, 7.4%, and 3.5% respectively when using TPS-based 3-level classification or binary classification. No statistical differences were observed in PD-L1 expression discordance rates when patients were stratified by age, sex, smoking status, histologic types, TNM stage, or years of paraffin blocks ( P >0.05). The presence of a solid histologic pattern was associated with a higher PD-L1 expression discordance rate in adenocarcinomas ( P <0.05). Twenty-five cases with discordant PD-L1 expression were divided into 2 categories: spatial heterogeneity without recognizable morphology difference (76%) and spatial heterogeneity with significantly different morphology (24%). The discordance rate in TPS scores was much higher in cases with different morphology than those without different morphology. One representative paraffin block containing adequate neoplastic tissue may be adequate to determine PD-L1 expression in most of the surgical specimens of NSCLC. In cases that harbored different morphology in different paraffin blocks, the dual-block immunohistochemistry assessment method is recommended due to the intratumoral heterogeneity.
Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc.