A comparative study of PD-L1 IHC assays using immune cell scoring and CPS in breast cancer.

Abstract

e15262 Background: The clinical benefit of immune checkpoint inhibitors (ICIs) targeting the programmed death-1/programmed death ligand 1 (PD-L1) pathway has previously been demonstrated across a range of tumor types, including in PD-L1+ patients with metastatic triple-negative breast cancer (TNBC). Various PD-L1 immunohistochemistry (IHC) assays and scoring algorithms are being investigated to select patients with breast cancer (BC) most likely to respond to ICIs. Scoring algorithms include PD-L1 expression on tumor cells, immune cells (ICs), or both. We compared the analytical concordance of 3 PD-L1 IHC assays and evaluated PD-L1+ prevalence, using combined positive score (CPS) and % IC scoring algorithms in commercially procured TNBC and hormone receptor–positive, HER2-negative (HR+/HER2−) BC samples. Methods: PD-L1 expression was assessed by HistoGeneX (Naperville, IL) in 163 commercially procured, surgically resected, formalin-fixed, paraffin-embedded BC samples (mostly stage I–III) using the Ventana PD-L1 (SP142) and Dako PD-L1 IHC 28-8 and 22C3 pharmDx assays in conjunction with the CPS (28-8 and 22C3 assays) and % IC algorithms (SP142 and 28-8 assays). PD-L1+ prevalence with each assay and concordance between assays were calculated using CPS ≥ 1 and IC ≥ 1% cutoffs, with a single pathologist assigned to each scoring algorithm. Results: 93 HR+/HER2− BC and 70 TNBC samples were evaluable for PD-L1 expression across all assays and algorithms. Overall concordance was higher between the 28-8 and 22C3 assays (CPS cutoff of 1) than between the 28-8 and SP142 assays (IC cutoff of 1%). PD-L1+ prevalence was similar with the 28-8 and 22C3 assays (CPS ≥ 1) and higher with the 28-8 assay (IC ≥ 1%) than with the SP142 assay (IC ≥ 1%). PD-L1+ samples identified by the SP142 assay (IC ≥ 1%) were mostly included within PD-L1+ sample sets defined by the 28-8 assay (IC ≥ 1% and CPS ≥ 1). PD-L1+ prevalence was higher in TNBC vs HR+/HER2− BC for all assays (Table). No clear trend in PD-L1+ prevalence was observed across tumor grade and stage. Conclusions: High analytical concordance was observed between the 28-8 and 22C3 assays (CPS cutoff of 1) in both HR+/HER2− BC and TNBC samples. PD-L1+ prevalence varied according to IHC assay, scoring algorithm, and cutoff used. Further studies are needed to select the most appropriate PD-L1 assay and scoring algorithm for BC clinical trials. [Table: see text]