background: Treatment with antibodies directed against programed-cell death ligand 1 (PD-L1) is a novel therapy for patients with gestational trophoblastic disease. Assessment of PD-L1 expression in tumor tissue is commonly used to identify patients who might benefit from anti-PD-L1 treatment. Multiple antibodies are available to detect PD-L1-expressing cells, and percentages of PD-L1-expressing cells in samples of patients with gestational trophoblastic disease indicated by these antibodies differ substantially. This raises the question which PD-L1 antibody best reflects PD-L1 expression to select patients for treatment.
methods: Seven commercially available antibodies for PD-L1 staining (E1L3N, 73-10, 22C3, C... More
background: Treatment with antibodies directed against programed-cell death ligand 1 (PD-L1) is a novel therapy for patients with gestational trophoblastic disease. Assessment of PD-L1 expression in tumor tissue is commonly used to identify patients who might benefit from anti-PD-L1 treatment. Multiple antibodies are available to detect PD-L1-expressing cells, and percentages of PD-L1-expressing cells in samples of patients with gestational trophoblastic disease indicated by these antibodies differ substantially. This raises the question which PD-L1 antibody best reflects PD-L1 expression to select patients for treatment.
methods: Seven commercially available antibodies for PD-L1 staining (E1L3N, 73-10, 22C3, CAL10, SP142, 28-8, SP263) were validated on Chinese hamster ovarian (CHO) cells transfected with PD-L1, PD-L2, wildtype CHO cells and tonsil tissue. Next, four complete hydatidiform moles and four choriocarcinomas were stained. Samples were independently assessed by two pathologists.
results: All seven antibodies showed membranous staining in the PD-L1-transfected CHO cells. E1L3N and 22C3 scored the highest percentages of PD-L1-positive cells (70%-90% and 60%-70%, respectively). E1L3N stained the cytoplasm of non-transfected CHO cells and was excluded from analysis. The remaining six antibodies predominantly stained syncytiotrophoblast cells of both complete hydatidiform moles and choriocarcinomas. The percentage of PD-L1-stained trophoblast cells and staining intensity varied substantially per used PD-L1 antibody and between complete hydatidiform moles and choriocarcinomas. Agreement between pathologists was best with 22C3 (intraclass correlation coefficient 0.94-0.96).
conclusions: Based on staining results of the CHO cells, gestational trophoblastic disease samples and intraclass correlation coefficient, 22C3 seems the most suitable for adequate detection of PD-L1-expressing trophoblast cells. All antibodies detected PD-L1-expressing cells in the gestational trophoblastic disease samples, though with great variability, hampering comparison of results between studies in this rare disease and emphasizing the need for uniformity in detecting PD-L1-expressing cells.
