In lung cancer patients presenting with malignant pleural effusion (MPE), cytology might represent the only source of tumor tissue for diagnosis and predictive biomarker testing. Programmed death ligand 1 (PD-L1) expression in tumor cells is a predictive biomarker for immunotherapy in non-small cell lung carcinomas and is tested using immunohistochemistry. However, knowledge of the validity of PD-L1 testing on MPE samples is limited. We evaluated the feasibility of immunocytochemistry (ICC) for PD-L1 in MPE cell blocks (CBs) and assessed the concordance in expression with patient-matched histologic samples. ICC for PD-L1 was performed on formalin-fixed paraffin-embedded CBs of MPE and patient-matched histologic samples, if available, using the automated Ventana PD-L1 SP263 assay. The tumor proportion score (TPS), based on partial or complete membranous tumor cell staining, was categorized as negative (<1%), low (≥1% to <50%), and high (≥50%). In CBs with any degree of PD-L1 expression, ICC for CD163 highlighting macrophages was performed to exclude nonspecific PD-L1 expression in macrophages. The CB PD-L1 TPS was compared with the TPS obtained from the patient-matched histologic samples. Of 43 MPE CBs available, 25 were positive for PD-L1 (25 of 42; 59%), and 1 sample was inadequate. Of the 11 patient-matched histologic samples tested, the PD-L1 TPS categories were concordant for 10 of the 11 (91% concordance) cases. PD-L1 expression in MPE CBs showed good concordance with expression in histologic samples and is feasible as a source for PD-L1 testing. The concurrent use of CD163 immunostains will aid in the manual assessment of PD-L1 TPS.
Read full abstract