Poster #9 - Yuheng Du

Cell-Type-Resolved Placental Epigenomics Identifies Clinically Distinct Subtypes of Preeclampsia

Yuheng Du, Ph.D. Student, Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI, USA; Paula A. Benny, Research Scientist, Department of Obstetrics and Gynecology, University of Hawaii, Honolulu, HI, USA; Fadhl M. AlAkwaa, Assistant Research Scientist, Department of Neurology, University of Michigan, Ann Arbor, MI, USA; Cameron B. Lassiter, Research Laboratory Manager, University of Hawaii Cancer Center, Honolulu, HI, USA; Yuansen (Jonathan) Liu, Undergraduate Student, Department of Mathematics,Washington University St Louis, USA; Yu Liu, Postdoctoral Researcher, Fox Chase Cancer Center, Philadelphia, PA, USA; Annette Lum-Jones, Lab Specialist, University of Hawaii Cancer Center, Honolulu, HI, USA; Joshua Astern, Manager, University of Hawaii Biorepository, John A. Burns School of Medicine, Honolulu, HI, USA; Maarit Tiirikainen, Genomics Lab Director, University of Hawaii Cancer Center, Honolulu, HI, USA; Lana X. Garmire, Associate Professor, Department of Computational Medicine and Bioinformatics, University of Michigan, Ann Arbor, MI, USA.

Preeclampsia is a heterogeneous pregnancy complication with gestational age-dependent manifestations and profound cellular complexity. Epigenome-wide association studies (EWAS) of preeclampsia have been hampered by both confounding clinical factors and cellular heterogeneity. To address these challenges, we analyzed placentas from the RMATRIX Hawaii Biorepository (HiBR), including 169 severe preeclampsia cases and 277 matched controls (106 preterm and 171 term deliveries). To adjust for confounding from cell composition that are often overlooked, we developed HOMED (Hierarchical Optimization Methylation Deconvolution), the first placenta-specific DNA methylation reference panel that incorporates hierarchical deconvolution of both major lineages and trophoblast subtypes. HOMED revealed significant differences in placental cell compositions, with increased villous cytotrophoblasts and extravillous trophoblasts, but lower proportions of Hofbauer cells, stromal cells, and syncytiotrophoblasts in cases. These trends are consistent with single-cell placental data. Subsequent stringent EWAS after adjustment of cell type composition, gestational age and clinical covariates identified thousands of differentially methylated probes, enriched for HIF1A-mediated hypoxia response, angiogenesis, and insulin resistance pathways in preeclampsia cases. Consensus clustering of these severe preeclampsia cases further identified two distinct epigenetic subtypes: Subtype 1 which resembles controls, and Subtype 2 which showed marked divergence from the controls and a significantly higher incidence of small-for-gestational-age (SGA) births. Application of a placental gestational age clock model revealed accelerated placental aging in Subtype 2 compared to both Subtype 1 and controls, corresponding to pre-term deliveries clinically. In summary, our study advances pathological and physiological changes of placentas under preeclampsia and highlights clinically relevant heterogeneity with implications for further stratification of severe preeclampsia.