Context.—: The separation of benign from malignant mesothelial proliferations is a difficult morphologic problem. Some mesotheliomas stain for programmed death ligand-1 (PD-L1).
Objective.—: To determine whether PD-L1 staining can separate mesotheliomas from reactive mesothelial proliferations (RMPs).
Design.—: We stained 2 tissue microarrays containing in toto 62 malignant mesotheliomas and 88 RMPs, using anti-PD-L1 antibody 22C3. Staining was graded by using an immunoreactive score encompassing intensity/distribution and was divided into negative, weak, moderate, and strong. Because PD-L1 staining can be heterogeneous, we also stained 10 whole sections of sarcomatoid/desmoplastic mesotheliomas (SMMs) and 10 whole sections of spindled RMPs in the form of organizing pleuritis, the major morphologic differential of SMM. These 20 cases were not on the tissue microarrays.
Results.—: RMPs showed generally negative, or occasionally weak staining in either the epithelioid or spindle cell compartments, whereas moderate to strong staining was seen in 10 of 14 SMMs but only in 6 of 48 epithelioid mesotheliomas (EMMs). The difference between SMM and RMP staining was statistically significant (P < .001), but between EMM and RMP was not significant. Whole section cases of organizing pleuritis showed no staining (N = 8) or weak staining (N = 2), whereas moderate/strong staining was seen in 9 of 10 whole sections of SMMs, a statistically significant difference (P = .02). There were no "hot spots" of RMP staining, suggesting that heterogeneous staining of RMPs is not a confounder.
Conclusions.—: Strong diffuse PD-L1 staining using antibody 22C3 supports a diagnosis of SMM when the differential diagnosis is RMPs, but PD-L1 staining is not useful for separating EMMs from RMPs.