Preeclampsia gene signatures and PD-L1 in endometrial cancer.

Abstract

e22589 Background: Aggressive endometrial cancers (EC), particularly p53-abnormal (p53abn), have poor outcomes despite immune-active features. Preeclampsia (PE), a pregnancy-specific hypertensive disorder, involves interferon-driven inflammation similar to anti-tumor immunity, yet whether PE-associated signatures identify immunotherapy-responsive EC phenotypes remains unexplored. We hypothesized that PE gene signatures might distinguish aggressive EC with distinct immune profiles relevant to immunotherapy selection. Methods: Transcriptomic and PD-L1 (22C3 IHC) data from 730 EC tumors (subtypes: p53abn, POLE, MSI-H, NSMP) were analyzed. Differential gene expressions between tumor subtypes were assessed using linear modeling with limma-voom. Gene set enrichment (GSEA) and over-representation analyses were conducted using Hallmark, GO Biological Process, KEGG_Medicus, and Reactome databases. Single-sample GSEA was performed to generate per-tumor PE scores from four published PE gene sets and a combined signature. Bipartite network clustering of PE scores defined PE-high/low groups among samples with detectable PE enrichment. Associations with PD-L1 were assessed via multivariable negative binomial regression adjusting for age, ancestry, stage, and molecular subtype. Results: Among 730 tumors analyzed, patient demographics included 45% aged 50–69 years; ancestry distribution of 41% European, 14% African, 16% other, and 29% missing; and molecular subtypes of p53abn (39%, aggressive), POLE (16%), MSI-H (13%), and NSMP (32%). Aggressive tumors showed immune activation with enrichment of interferon signaling (normalized enrichment score [NES] 1.42–1.56), antigen presentation (NES 1.45–1.50), and T-cell/leukocyte pathways (NES 1.40–1.51, GSEA FDR < 0.25). Over-representation analysis identified 290 immune pathways (FDR < 0.05), with T-cell activation, leukocyte adhesion, interferon response, and antigen processing among the most significant. PE signatures were enriched in aggressive tumors (combined PE OR 1.92, p=0.002), with 81/125 overlapping upregulated genes. PE enrichment correlated with PD-L1 expression (ρ=0.26, p=2.7×10⁻¹²) and independently predicted PD-L1 positivity (IRR 1.63 per SD, 95% CI 1.36–1.95, p=6.8×10⁻⁷), superseding molecular subtype (all p>0.21). Bipartite clustering identified PE-high (n=350) and PE-low (n=380) groups. PE-high tumors showed higher PD-L1 positivity (44% vs 28%, p=1.6×10⁻⁵), though expression intensity among positives was comparable (p=0.36). Conclusions: PE-associated gene signatures define an immune-active subset of aggressive EC characterized by interferon signaling and antigen presentation. These signatures correlate with PD-L1 expression and outperform molecular subtype in predicting PD-L1 positivity, suggesting potential utility for refining immunotherapy selection and warranting further clinical validation.