The usefulness of various cytologic specimen preparations for PD-L1 immunostaining in non-small cell lung carcinoma

Introduction: Evaluation of programmed cell death ligand-1 (PD-L1) on formalin-fixed paraffin-embedded (FFPE) histologic specimens is required for immunotherapy with pembrolizumab in non-small cell lung carcinoma (NSCLC). A significant percent of patients may be diagnosed on cytology samples alone; however, FFPE tissue may not always be available. Here, we evaluated the feasibility of PD-L1 staining on a variety of cytologic preparations including conventional cell blocks (CB), cell-transfer cell blocks (CTCB), and direct smears. Materials and Methods: We identified 30 NSCLC cytology cases that had PD-L1 status evaluated on a concurrent core needle biopsy. Papanicolaou-stained smears (PS) were reviewed and a moderately cellular smear was selected per case to prepare a CTCB. CTCB, CB, and PS (when available) were stained with PD-L1 22C3 and tumor proportion scores (TPS) were compared across all preparations with the concurrent core biopsies. Results: Concordance of PD-L1 positivity in CB and PS versus core biopsy was high, 80% (12 of 15) and 94.4% (17 of 18), respectively. CTCB versus core biopsy concordance was lower in comparison, 62% (13 of 21) for PD-L1 positivity. Overall, TPS was lower in CTCB compared with CB and PS. Among the 3 preparations, CTCB and CB sections were easier to interpret as PS displayed various artifactual staining patterns. Conclusions: In summary, our study demonstrates that CB and PS preparations are suitable surrogates for histologic FFPE tissue for determining PD-L1 status in NSCLC patients. CTCBs, on the other hand, show high discordance and are not optimal specimens. Pre-analytic factors in unconventional cytology preparations may need optimization and negative results should be interpreted with caution.