Abstract 131: Concordance of mRNA expression (nCounter) and protein expression (IHC) for the detection of PD-L1 in patients with advanced non-small cell lung cancer (NSCLC)

• Published in: Cancer research (Chicago, Ill.) Ill.), 2019-07, Vol.79 (13_Supplement), p.131-131
• Main Authors: Teixido, Cristina, Marin, Elba, Aguado, Cristina, Pare, Laia, Gimenez-Capitan, Ana, Lopez-Prades, Sandra, Cardona, Andres Felipe, Cabrera, Carlos, Gonzalvo, Elena, Lopez, Laura, Roman, Ruth, Martinez, Daniel, Sullivan, Ivana, Jares, Pedro, Prat, Aleix, Molina-Vila, Miguel Angel, Reguart, Noemi
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2019-131

Background: PD-L1 immunohistochemistry (IHC) staining is currently accepted as the gold-standard biomarker for immune therapy in advanced non-small cell lung cancer (NSCLC). However, the use of various antibodies and cut-offs as well as certain degree of subjectivity in pathological evaluation has overshadowed the clear-cut predictive performance of PD-L1 expression. Multiplexed technologies can be of help in this setting providing an objective measurement of PD-L1 levels. On the other hand, gene expression signatures incorporating not only PD-L1 but also other components of the stroma might better capture the immune-context of the molecular heterogeneity of NSCLC tumors. nCounter gene expression technology is an alternative method to measure PD-L1 gene expression by digital counting proving a direct measurement of mRNA levels. Methods: A 7-gene ‘immune signature’ comprising CD4, CD8, programmed cell death-1 (PD-1), programmed death-ligand 1 (PD-L1), interferon gamma (IFNG), granzyme M (GZMM) and forkhead box P3 (FOXP3) were included in a customized nCounter panel (NanoString Technologies), used in our institution on a routine basis to simultaneously screen for relevant oncogenic-drivers (ALK, ROS1, RET, NTRK1 gene fusions and METΔ14 mutation). Total RNA obtained from formalin-fixed paraffin embedded (FFPE) samples was used for PD-L1 digital counting (nCounter) which was normalised with six housekeeping genes (ACTB, MRPL19, PSMC4, RPLP0, SF3A1, GAPDH) and compared with PD-L1 protein IHC evaluation using whole tissue section with 22C3 monoclonal mouse anti-PD-L1 antibody measured on tumor cells. Results: A total of 425 FFPE samples from advanced NSCLC were analyzed with the nCounter panel. Among them, 25 samples were not evaluable (5.9%). PD-L1 IHC was available for 163 FFPE samples and were compared with nCounter PD-L1 expression results. By IHC, 63/163 samples (38.65%) were scored as negative for PD-L1 protein expression, whereas 100/163 (61.35%) were evaluated as positive. Among positive, 62 (38.04%) and 38 (23.31%) presented a moderate (≥ 1-49%) and high PD-L1 staining (≥50%) respectively. Using an appropriate cut-off value (IHC≥1%), PD-L1 mRNA expression levels correlated with PD-L1 IHC evaluation with a 76% of concordance and a 0.755 Cohen’s kappa (confidence interval 95% 0.651- 0.858). Unsupervised clustering across of mRNA expression data from 395 samples using the seven-immune-related genes and correlations between each immune gene were performed a