On behalf of my co-authors, I'd like to thank the organizers for selecting our study for a featured poster presentation today. Our study investigated PD-L1 assessment in cytology compared to histology samples in predicting treatment response to PD-L1 inhibitors. These are my disclosures. To start off with some background, tumor PD-L1 expression is key in determining first-line therapy in advanced non-small cell lung cancer. In this population, diagnosis and molecular biomarker testing are often done on cytology specimens obtained through minimally invasive procedures. However, PD-L1 testing was developed and validated on histology samples. Prior studies have indicated a similar PD-L1 distribution in cytology samples and a high concordance between paired samples, at least for the high expressors greater than 50%. The goal of our study is to further examine the response and survival to PD-L1 inhibitors in patients with PD-L1 testing done on cytology compared to histology specimens. We reviewed all non-small cell lung cancer cytology and histology samples that were tested for PD-L1 using the 22C3 assay at the University Health Network between 2013 and 2020. PD-L1 assessment was conducted on FFPE cell blocks and samples were received fresh or prefixed with cytolite. A subset of patients treated with PD-L1 inhibitors at our center were reviewed for their treatment outcomes. Between 2013 and 2020, we tested 487 cytology and 1,683 histology samples for PD-L1. Rates of informative testing were high and were similar between cytology and histology samples. The PD-L1 distribution among lung core biopsies, lung or lymph node FNAs, and pleurofusions were similar. Within this cohort, 117 patients were treated at our center with single agent PD-L1 inhibitors. The majority of patients in both groups have PD-L1 expression over 50%. This is reflective of practice patterns over time where patients with PD-L1 less than 50% are now receiving combination chemoimmunotherapy up front. All other baseline patient characteristics, including the presence of driver mutations, were similar between the two groups. The overall response rates were 39% in patients with cytology-assessed samples and 35% in those with histology-assessed samples and were not statistically different. Progression-free survival was also similar between cytology and histology samples. And the levels of PD-L1 expression continue to be predictive of response to checkpoint inhibitors. In a subset of patients with PD-L1 expression over 50% who were treated first line with pemulizumab, the median PFS was 6.1 months in the cytology group and 7.4 months in the histology group. These numbers are comparable to each other and are comparable to the survivals that were reported in clinical trials. In conclusion, PD-L1 test success rates were high for cytology specimens and the PD-L1 distribution was similar to that at testing by histology. Response rates and PFS was also comparable whether PD-L1 was assessed on cytology or histology specimens and the treatment outcomes were similar to those reported in clinical trials. Overall, our results support the utility of biomarker testing on cytology samples. Thank you very much for listening to our presentation today. If there are any questions about our study, please do not hesitate to reach out at the email listed at the bottom of the slide.

PD-L1 assessment

cytology samples

histology samples

treatment response

PD-L1 inhibitors