P1.09-10 Molecular Immunocytochemistry in Various NSCLC Samples Prepared as Non-Cell Block Cytology

Molecular prognostic-predictive immunocytochemistry is obligatory for therapy decision in NSCLC patients. Immunocytochemistry on cell blocks FFPE cytology is preferred because of some validation advantages, but is not always sufficient in terms of samples acquisition and cell number adequacy. Aim of the study is to compare molecular immunocytochemistry expression between cytology samples obtained with bronchoscopy (bronchial washing/brushings and transbronchial fine needle aspirations) and other cytology samples such as pleural effusion, FNA of peripheral lymph nodes and skin nodules and transthoracic FNA/biopsy; all prepared as non-cell block cytology. We compared archive records of the 1109 immunocytochemistry (ICC) results of ALK, ROS1 and PD-L1 expression in carcinoma cells in relation to specimen type (bronchoscopic and non-bronchoscopic samples) prepared as smears and cytospins at our institution over a one-year period. Air dried cytology smears and cytospins of NSCLC samples were stained with Anti-ALK Clone D5F3, Cell Signaling, appendix positive control; Anti-ROS1 Clone D4D6 Cell Signaling, HCC-78 cell line positive control and Anti-PD-L1, Clone 22C3, Dako, placenta imprint positive control; EnVision detection system on Immunocytochemistry Autostainer. PD-L1 protein expression was scored using Tumor Proportion Score (TPS) with positive cut-off of ≥1%, membrane staining. The slides were routinely examined and scored by two cytologists. Internal and external quality control were performed on FFPE cell blocks and histology slides with corresponding Ventana antibodies and staining systems. Out of 1109 results, 440 were ALK ICC results, 111 ROS1 and 558 were PD-L1 ICC staining results. Among them 314/440 (71, 36%) ALK, 75/111 (67,57%) ROS1 and 392/558 PD-L1 ICC (70,25%) were in bronchoscopic samples and 126/440 (28,64%) ALK, 36/111 (32,43%) ROS1 and 166/558 (29,75%) PD-L1 ICC were in various non-bronchoscopic samples. Comparison in ALK, ROS1 and PD-L1 ICC distribution between two groups of samples showed no statistical significant difference among groups (X2 test, df 2, p=0,730). Positive ALK and ROS1 numbers were in observed range, but insufficient for statistical analysis. PD-L1 ICC scored results were in total of 392 samples collected during bronchoscopy and 166 of various other non-bronchoscopic samples: 88 were obtained by FNA of peripheral lymph nodes and skin nodules, 38 by transthoracic FNA/biopsy and 40 pleural effusions. PD-L1 ICC scored negative in total of 281/558 (50,36%) samples, 191 bronchoscopic and 90 non-bronchoscopic. PD-L1 ICC scored positive in total of 277/558 (47,11%) samples of which 201 were obtained with bronchoscopy and 76 of samples collected with other methods. Among of total PD-L1 positive smears, 128 were PD-L1 positive ≥ 50%; 90/128 (70,31%) were bronchoscopical samples and 38/128 (29,7%) other samples. Comparison in PD-L1 expression between two groups of samples showed no statistical significant difference among groups (X2 test, df 1, p=0,236). Comparison between two groups of PD-L1 positive ≥50% showed no statistical significant difference among sample groups (X2 test, df 1, p=0,436). There were no statistical significant differences in molecular ALK, ROS1 and PD-L1 immunocytochemistry results between samples collected during bronchoscopy and non-bronchoscopic samples prepared as non-cell block cytology. PD-L1 scoring results were also independent of cytology sample type in our study.