Cell-free DNA fragmentome analysis informs pregnancy outcome in patients with immune-mediated disease

Pregnancy complications contribute substantially to maternal and fetal morbidity and mortality. If complications are suspected early in pregnancy, individuals could be triaged to escalate care and ideally reduce adverse outcomes. However, clinical risk factors are poor predictors early in gestation. Cell-free DNA (cfDNA) fragmentomics has emerged as a promising biomarker in multiple disease states but has yet to be implemented in the prenatal setting. Here, we analyzed 1910 first-trimester blood cfDNA samples from pregnant individuals at high risk of an adverse pregnancy outcome (APO) because of twinning, a preexisting immune-mediated disease (IMD), or cytomegalovirus infection and from low-risk individuals. We extracted multimodal cfDNA fragmentome features, including transcriptome-aware cell-type composition estimates, end-motif frequencies and fragment length profiles, and trained models that could discriminate each high-risk group from controls, except diabetes. In IMDs, we further identified a cfDNA fragmentome profile associated with APOs in patients with clinically quiescent and serologically inactive disease. At a fixed false-positive rate of 10%, we could correctly discriminate 54, 38, 73, and 90% of patients with systemic IMD, thyroid-related IMD, Crohn's disease, and psoriasis, respectively, who went on to have an APO from those who did not. These results were validated in an independent cohort of pregnant patients with IMD from an external center. The fragmentome was additive to and independent of clinical risk factors and fetal fraction estimates in stratifying APO risk in IMD. Overall, these findings support the further investigation of fragmentomics for triage of high-risk pregnant patients to improve prenatal care.