P2.09-16 Assessment of PD-L1 and CD8 Expression in Lung Cancer Using RNA in Situ Hybridisation

Abstract

PD-L1 is routinely assessed using immunohistochemistry (IHC) as a companion-diagnostic test. However, patients may be missing out on therapy as issues exist in determining PD-L1 positivity. This is partly due to the use of different antibodies, staining platforms and positivity cut-offs. The use of RNA in-situ hybridisation (RISH) for the detection of PD-L1 levels may be a way to address this; particularly as PD-L1 mRNA expression levels are associated with better overall survival in NSCLC. It is well established that PD-L1 expression is associated with increased numbers of Tumour Infiltrating Lymphocytes (TILs) including those which are CD8+. Therefore, it may also be worthwhile to combine PD-L1 testing with additional marker testing such as CD8. The aim of this study is to investigate the expression of PD-L1 and CD8 by RISH in a cohort of NSCLC samples and correlate with clinical outcome. RISH protocols (RNAScope® 2.5 HD assay, Advanced Cell Diagnostics, Inc.) were initially optimised on a well annotated cell line TMA (low, moderate and high expression) and compared with standard IHC (DAKO - clone 22C3). Following successful optimisation of the technique, a test cohort of 35 full-face FFPE NSCLC samples were examined for PD-L1 expression by both RISH and IHC. All slides were scored independently by a pathologist. RISH slides were scored as per ACD guidelines (0-4; based on number of positive dots per cell and number of dot clusters). IHC was scored using the standard tumour proportion score system. Image analysis (IA) was subsequently performed by Visiopharm®. Currently, a further cohort of 200 FFPE samples is being assessed, in addition to 50 FFPE samples from patients who received an anti-PD-1 therapy. Optimisation is on-going for dual CD8/PD-L1 RISH staining. In the initial 35 samples, PD-L1 was detected in 20% of cases by IHC (≥50% positivity) with mRNA expression identified in 14% of cases by RISH (≥2). In contrast with other recently published studies, issues were identified between IHC and RISH. There were 3 cases which scored positive by IHC, which were negative by RISH. Additionally, 1 case was positive by RISH but negative by IHC. The comparison between both assays produced a ‘moderate agreement’ as assessed using Cohen’s kappa coefficient (κ=0.528, p=0.002). IA undertaken on both assays could not be statistically compared using Cohen’s kappa coefficient due to the use of a composite scoring matrix to quantify RISH mRNA signals. However, IHC stained sections scored by IA showed similarity to IHC scored by a pathologist. Further optimisation is required on the RISH IA scoring algorithm. A comparison of PD-L1 IHC with RISH, combined with dual CD8/PD-L1 staining in a larger cohort of clinical samples is on-going, which will further determine the clinical utility of this technique. Data from our test cohort of samples have shown that PD-L1 detection by RISH is feasible and compares well to IHC. Preliminary data would also suggest that IA may be better than a pathologist alone, particularly in borderline cases. RISH may be a suitable method for improved detection of PD-L1 in NSCLC.