Interobserver agreement of PD-L1 (SP263) assessment in advanced NSCLC on cytological smears and histological samples

Background: The PD-L1 assessment is mandatory for the selection of patients affected by advanced non-small-cell lung cancer (NSCLC) who can benefit from the PD-1/PD-L1 checkpoint inhibitors therapy. Previous studies tested PD-L1 on cytological smears to evaluate this sample as an alternative to formalin-fixed paraffin-embedded (FFPE) ones, but several critical issues needed to be clarified. Aim: We evaluated the cyto-histological agreement (CHA) and the PD-L1 interobserver agreement (IrOA) among three different pathologists (Path1, Path2, Path3) on 160 paired cytological smears and histological samples of advanced NSCLC. Results: With the cut-off of < 50%/≥ 50%, CHA resulted good for Path1 (Cohen's k: 0.702) and Path3 (Cohen's k: 0.731), moderate for Path2 (Cohen's k: 0.576) adopting the same cut-off, the IrOA was moderate (ICC 0.72 [95% CI: 0.63–0.78]) for smears and good for histological samples (ICC 0.85 [95% CI: 0.80–0.85]). Conclusion: With a cut-off system of < 50%/≥ 50%, PD-L1 assessment shows moderate to good CHA and exhibited moderate IrOA on smears and good IrOA on FFPE. As result, PD-L1 assessment should be improved on cytological smears as well as could be a suitable alternative for patients without FFPE samples and not eligible for pembrolizumab, adopting a cut-off of < 50%/≥ 50%; presumably, an appropriate pathologist training could further improve the reproducibility.