P2.01-048 Paired Comparison of PDL1 Assessment on Cytology and Histology from Malignancies in the Lung: Topic: Immune Mechanisms in Thoracic Cancer and Targeted Therapy

PD-L1 tests have been approved on histological specimens only. More than 1/3 of NSCLC patients are diagnosed on cytology alone. The hypothesis of this study is that cytological cell block material is as good as histological material for PD-L1 analysis. 86 paired samples of malignancies from the lung (NSCLC = 72, other primary malignancies and metastases = 14), where cytological cell block and histological material were available from the same lesion within 6 weeks were stained for PD-L1 on serial sections with PD-L1 IHC 28-8 pharmDx (28-8pharmDx) and PD-L1 IHC 22C3 pharmDx (22C3pharmDx). The partial or complete membrane staining on malignant cells regardless of intensity was assessed as ≥1%, ≥5%, ≥10% and hereafter in 10% increments. Number of malignant cells (< or > 100) and staining heterogeneity were recorded. The pathologist was blinded to previous evaluations of pair members. Overall (OA) and Average Positive (APA) and Negative (ANA) agreement were calculated as well as Positive Percent Agreement (PPA) and Negative Percent Agreement (NPA) with histology as the non-reference standard. CIs were bootstrapped. Agreement statistics for PD-L1 positivity with different cutoffs in cytology and histology specimens, % (95% CI)Tabled 122C3pharmDx≥ 1% positive≥ 50% positiveOA85 (77–92)94 (88–99)APA83 (73–91)82 (61–96)ANA86 (78–93)97 (93–99)PPA80 (67–92)100NPA89 (79–98)93 (87–99) Open table in a new tab Tabled 128-8pharmDx≥ 1% positive≥ 5% positive≥ 10 % positiveOA87 (80–94)95 (91–99)90 (83–95)APA86 (77–94)94 (86–99)84 (71–93)ANA88 (80–94)96 (92–99)92 (87–97)PPA81 (69–92)91 (79–100)79 (63–93)NPA93 (85–100)98 (94–100)95 (88–100)There was high agreement between the cytological and histological scores, applying to both pharmDx kits and all cutoffs (Table), not changing by exclusion of cytological cell blocks containing <100 cells nor by exclusion of non-NSCLC. Pearson r2 were 0.87-0.89. Paired samples with PD-L1 staining heterogeneity on histological material seemed to have lower agreement than samples with homogenous staining. Open table in a new tab There was high agreement between the cytological and histological scores, applying to both pharmDx kits and all cutoffs (Table), not changing by exclusion of cytological cell blocks containing <100 cells nor by exclusion of non-NSCLC. Pearson r2 were 0.87-0.89. Paired samples with PD-L1 staining heterogeneity on histological material seemed to have lower agreement than samples with homogenous staining. High correlation between staining on cytological cell block material and histological specimens was observed using PD-L1 IHC 28-8pharmDx and PD-L1 IHC 22C3pharmDx, suggesting that cytological material is as good as histological material for PD-L1 IHC analysis. Intra-tumor heterogeneity may contribute to disagreement and more research on the influence of staining heterogeneity is warranted.