Hot or not: defining trophoblast PD-L1 expression and lymphohistiocytic density in gestational trophoblastic neoplasia

Abstract

<h3>Objectives:</h3> PD-L1 is an important axis of immunotolerance in normal pregnancy and this may be perpetuated as a mechanism of immune evasion in the pathobiologic development of gestational trophoblastic neoplasia (GTN). There are early efficacy signals for anti-PD-1 immunotherapy in chemoresistant GTN, however response rates are variable. The characteristics of predictive biomarkers used for triaging anti-PD-1 immunotherapy are not well-known in GTN. The aim of this study was to quantify trophoblast PD-L1 expression in GTN using the FDA approved PD-L1 22C3 immunohistochemical (IHC) companion assay and to profile GTN-infiltrating immune cells. <h3>Methods:</h3> Choriocarcinoma (CC), epithelioid trophoblastic tumor (ETT) and placental site trophoblastic tumor (PSTT) cases with available tissue blocks in the Johns Hopkins GTN library were used for this investigation. IHC staining for PD-L1 and each immune marker (CD68, CD8, CD4, FOXP3) was performed on paired, fresh cut 4 micrometer tissue sections. Tumor proportion scores (TPS, percentage of viable tumor cells with any PD-L1 membrane staining) were assigned by 2 independent gynecologic pathologists. Discrepancies in scoring were adjudicated by a third gynecologic pathologist. Digital image analyses using QuPath v0.2.3 to average CD68, CD8, CD4 and FOXP3 immune cell density in three 40x hotspots per case was performed. Hotspots were targeted to the tumor-immune cell interface. Descriptive statistics and the Wilcoxon Rank Sum Test were performed for analysis. <h3>Results:</h3> A large proportion of PSTT (n=6/8,75%) and CC (n=22/32, 69%) and approximately half of ETT (n=9/19, 47%) cases have highpositive PD-L1 expression defined as the clinically relevant TPS cutoff of >50%. These exceed the rates of high PD-L1 positivity observed in all other solid tumor sites, which ranges from 2-31%. PSTT and CC have significantly higher trophoblast PD-L1 expression than ETT; the median TPS for PSTT and CC is 90% and 85%, compared to 35% for ETT (p=0.02). CD68 monocytes were more abundant in CC and PSTT compared to ETT. However, CD8 cytotoxic T (p<0.01) and CD4 T (p=0.04) regulatory cell densities are significantly higher in ETT. ETT has a wider spectrum of tumor PD-L1 expression than PSTT and CC. ETT cases with low/negative trophoblast PD-L1 (TPS <5%) have fewer immune infiltrating lymphocytes than ETT cases with TPS >5% (Table). <h3>Conclusions:</h3> PD-L1 expression in GTN is heterogeneous, yet disproportionately high compared to other primary tumor sites. Of the 3 GTN subtypes, ETT has the lowest and broadest range of PD-L1 expression; this is associated with lower CD68 monocyte infiltration and correlates with degree of lymphocytic infiltration. How these expression patterns predict response to anti-PD-L1/PD-1 agents in GTN will be need to be explored. In a tumor type with such lineage inherent PD-L1 expression, characterization of tumor cell PD-L1 expression in conjunction with the tumor immune-microenvironment may provide a unique opportunity to study mechanisms of immune escape.Funding was provided by the Bruce Patsner Research Grant for Innovative Research and Technology through the Foundation for Women's Cancer.