Objectives: Multiple programmed cell death ligand-1 (PD-L1) immunohistochemistry assays are currently used as companion or complementary diagnostic tools for anti-programmed cell death-1 immunotherapies. We aimed to characterize two PD-L1 immunohistochemistry assays (Dako 22C3 and 28-8) for non-small cell lung cancer (NSCLC) in clinical laboratories.
Materials and methods: Surgical specimens from 420 patients with pathological stages IA to IIIA NSCLC were investigated. The archived samples were freshly cut into 5-μm-thick sections stained with antibodies 22C3 and 28-8, and tumoral PD-L1 expression was evaluated in two clinical laboratories, respectively. Overall, positive, and negative percent agreement (OPA, PPA, and NPA, respectively) at specified PD-L1 cutoffs were calculated to assess the concordance between 22C3 and 28-8 assays.
Results: Tumoral PD-L1 expression of ≥ 1% was detected by either 22C3 or 28-8 assays in 176 cases (41.9%), whereas 22C3 revealed a significantly higher tumoral PD-L1 expression compared to 28-8 (median 30% vs. 10%, p < 0.0001). OPA was 89.0, 90.2, and 91.9% at 1, 25, and 50% cutoff levels. When 22C3 was compared to a standard assay 28-8, the PPA was 85.5, 98.3, and 94.9%, whereas NPA was 91.0, 89.0, and 91.6% at 1, 25, and 50%. On the other hand, when 28-8 was compared to 22C3, PPA was 84.4% at 1%, but it decreased to 58.3 and 53.6% at 25 and 50%; whereas NPA remained high (91.7, 99.7, and 99.4% at 1, 25 and 50%, respectively).
Conclusion: Our analysis revealed that, despite the high OPA, there was discordance in the PPA between 22C3 as a standard assay and 28-8 as a comparator assay at 25% and 50% PD-L1 cutoff levels, indicating that the results of 28-8 could be translated to those of 22C3 but not vice versa.
Keywords: 22C3; 28-8; Immunohistochemistry; Negative percent agreement; Overall percent agreement; Positive percent agreement; Programmed cell death ligand-1; Programmed cell death-1.
Copyright © 2018 Elsevier B.V. All rights reserved.