Choice of fixative affects programmed death-ligand 1 expression in cell blocks from pleural effusions with metastatic pulmonary adenocarcinoma.

[BACKGROUND] Programmed death-ligand 1 (PD-L1) immunocytochemical (ICC) analysis of cell blocks (CBs) has recently emerged in clinical practice. Unlike standardized immunohistochemistry on formalin-fixed, paraffin-embedded tissues, cytology involves various preparation methods and fixatives. This study investigated how various fixatives influence PD-L1 immunoreactivity in CBs from malignant pleural effusions (MPEs) with metastatic pulmonary adenocarcinoma (AC).

[METHODS] Thirty-three MPEs from patients with pulmonary AC were prospectively included. Four matched CBs per case were fixed in four different fixatives and immunostained with three PD-L1 antibodies. Tumor proportion score and staining intensity were evaluated at multiple cutoffs.

[RESULTS] The cytology-cytology correlation of PD-L1 expression with the antibodies 28-8, 22C3, and SP263 was assessed in matched CBs fixed in either formalin, PreservCyt, CytoLyt, or CytoRich Red (the latter only in 26 cases). Compared to formalin, PreservCyt and CytoLyt showed moderate concordance at the ≥1% cutoff (Cohen kappa [κ], 0.463-0.535 and 0.57-0.586, respectively), except SP263 with CytoLyt, which demonstrated only fair concordance (κ, 0.382). The corresponding figures for CytoRich Red indicated substantial concordance for 28-8 and SP263 (κ, 0.601 and 0.669) and very good concordance for 22C3 (κ, 0.806). At the ≥50% cutoff, concordance improved for 28-8 and 22C3 but remained largely unchanged for SP263. All alcohol-based fixatives produced significantly weaker PD-L1 staining intensity than formalin across all antibodies (p < .001-.007).

[CONCLUSIONS] PD-L1 ICC expression in CBs depends on the fixative and antibody used. Alcohol-based fixatives, particularly with low cutoffs, risk underestimating PD-L1 positivity, and may contribute to false-negative results. ICC protocol optimization is essential before diagnostic use.