Programmed Death Ligand 1 (PD-L1) as a Therapeutic Target in Gestational Trophoblastic Neoplasia (GTN) (098)

Abstract

<b>Objectives:</b> Despite early efficacy signals for anti-PD-L1/PD-1 immunotherapy in chemotherapy-resistant GTN, response rates are variable. Lack of consistent evaluation and reporting of trophoblast PD-L1 expression and other candidate predictive biomarkers makes developing a precise strategy for PD-L1/PD-1 blockade in the treatment of GTN challenging. We characterize trophoblast PD-L1 expression using the clinically applicable PD-L1 IHC 22C3 assay, tumor-infiltrating immune cell (TIIC) density and clinical outcomes in one of the largest GTN cohorts to date. <b>Methods:</b> Cases of paraffin-embedded primary tumors in our GTN library were used for this investigation (<i>n</i>=60). Immunohistochemistry (IHC) staining for PD-L1, CD68 (macrophages), CD8, CD4, and CD56 (NK cells) was performed on fresh-cut 4-mm tissue sections. Tumor proportion scores (TPS) were independently assigned by two gynecologic pathologists. Digital image analysis using QuPath v0.2.3 to average TIIC density in 3 40x hotspots per case was performed. Complete clinical data was possible for 18 patients. Descriptive statistics, Wilcoxon rank-sum test, and simple linear regression were used. <b>Results:</b> PD-L1 expression and TIIC density were quantified for all 60 cases of choriocarcinoma (CC, <i>n</i>=32), placental site trophoblastic tumor (PSTT, <i>n</i>=9), and epithelioid trophoblastic tumor (ETT, <i>n</i>=19). CC and PSTT have higher trophoblast PD-L1 expression than ETT; the median TPS for CC and PSTT was 85% (range: 0-98%) and 90% (range: 0-90%), compared to 35% (range: 0-100%) for ETT (p=0.02). CD68 monocytes were more abundant in CC and PSTT than ETT. CD8 cytotoxic T (p<0.01) and CD4 regulatory T (p=0.04) cell densities were higher in ETT. CD56 expression was associated with PD-L1 expression (slope=0.36, p=0.03). The Table summarizes the clinically annotated subset of patients. Similar patterns of PD-L1 expression/ TPS and TIICs were observed for CC and ETT; complete data were available for only two PSTT cases. Of the 15 patients who did not see anti-PD-L1/PD-1 treatment, seven (47%) would have been eligible based on a TPS of >0 and might have been offered it as an alternative to multi-agent chemotherapy or fertility loss. Two (PSTT case 16 and ETT case 17) of the three patients who received salvage anti-PD-1 therapy had durable complete responses. The third case (ETT case 18) had a complete response but then progressed after two years. All three had above-median TPS scores for their histologic subtypes and had moderate to high levels of CD56 expression. The two patients without evidence of disease had above median levels of CD8 and CD68 for their subtype, while the patient who recurred had below-median levels of both immune markers. <b>Conclusions:</b> Contrary to the published literature, ubiquitously high trophoblastic PD-L1 expression was not observed using the clinically applicable PD-L1 IHC 22C3 assay. Rather, there is a wide range of PD-L1 expression, including some cases with a TPS score of 0%. We added three cases of non-CC treated with salvage anti-PD-L1/PD-1 therapy to the four reported to date. In these cases, tumor-infiltrating CD8, CD68, and CD56 cell density may influence anti-PD-1 efficacy. There may be an opportunity to spare GTN patients from toxicity and/or adverse outcomes by prescribing anti-PD-L1/PD-1 therapy earlier in their disease course.