Abstract
BACKGROUND: Multifetal pregnancies have increased preeclampsia risk, but the underlying pathogenesis may differ from that of singletons. It remains unclear whether twin placentas show molecular signs of preeclampsia synchronously. METHODS: We performed RNA sequencing on 32 individual placental samples from twin gestations grouped by preeclampsia status: 24 dichorionic twin (DT) and 8 monochorionic twin gestations. Ten singleton placentas from preeclamptic pregnancies were also analyzed. A benchmark data set (GSE203507, GSE114691, and GSE1482410) and a test data set (GSE190973) comprised 71 early onset preeclampsia and 69 control singleton placentas. Differential abundance analysis was conducted, and machine learning was used to derive a novel 98-transcript classification signature (accuracy >0.97 in benchmark and test data sets). RESULTS: Across 7 groups, 2946 transcripts were differentially modulated (likelihood-ratio test; false discovery rate <0.05). Placental signature scoring distinguished normotensive from early onset preeclampsia in GSE203507 singletons (P<0.0001) although normotensive DTs did not differ from DTs with preeclampsia (Kruskal-Wallis/Dunn). Notably, some twin placentas without clinical preeclampsia exhibited preeclampsia-like profiles. Linear mixed-effects regression, which accounted for intertwin correlation structure, revealed increasing signature scores across singleton and DT groups (all P<0.01): normotensive singletons