P2.09-14 Evaluation on Inter-Assay Consistency and Inter-Reader Precision for PD-L1 Assays: Ventana SP263 and Dako 22C3 in Non-Small Cell Lung Cancer

Abstract

FDA has approved several anti-PD1/PD-L1 checkpoint inhibitors for the treatment of a variety of cancers. It has been shown the expression level of PD-L1 in NSCLC tumor cells is associated with the response to anti-PD-1/PD-L1 treatment. Currently, five anti-PD1/PD-L1 inhibitors have been approved for either as a companion diagnostic (e.g., pembrolizumab) or as complementary diagnostic (e.g., nivolumab, atezolizumab and durvalumab). The blueprint studies and several others have reported the concordance on different PD-L1 IHC assays, however, few Chinese samples were included for assay consistence study; in addition, it remains unclear how these two assays perform in inter-reader variability of pathologists in China. Surgical specimens from 48 NSCLC patients were selected for PD-L1 IHC by Ventana SP263 and Dako 22C3. Tumor proportion scores (TPS) were interpreted by three pathologists from three centers. Inter-assay consistency were evaluated by Kappa test at different cut-off value (1%, 25%, and 50%) for each pathologist; and inter-reader precision were performed by Kendall’s W test for Ventana SP263 and Dako 22C3, respectively. 1) Inter-assay consistency: When TPS were categorized as <1%, 1% to 24%, ≥25%, the overall agreement percentage (OPA) between two IHC assays was 81.25% observed by Pathologist A (κ=0.710), 89.6% observed by Pathologist B (κ=0.84) and 79.2% observed by Pathologist C (κ=0.684); If the TPS were categorized as <1%, 1% to 50%, ≥50%, the OPA were 81.25%, 93.75% and 85.4% (κ= 0.675, 0.899 and 0.760), respectively. 2) Inter-reader precision: The interpretation results on Ventana SP263 by three pathologists were consistent, when TPS was categorized as <1%, 1% to 49%, ≥50% (p=0.017) or <1%, 1% to 24%, ≥25% (p=0.047) respectively; the consistency was not optimal, when TPS was further categorized with narrower groups as <1%, 1% to 24%, 25% to 49%, ≥50% (p=0.065). The interpretation results of Dako 22C3 were inconsistent among three observers in all categories above, with all p values greater than 0.05. 1) The inter-assay consistency of PD-L1 assays is high between Ventana SP263 and Dako 22C3; 2) The interpretation results of Ventana SP263 is more consistent among different observers.