Clinical validation of novel second-trimester preterm preeclampsia risk predictors combining clinical variables and serum protein biomarkers

Abstract Objective To develop and validate mid-trimester preterm preeclampsia (PE) risk predictors combining clinical factors and serum protein analytes, and to compare their performance with those of widely used clinical and risk assessment algorithms endorsed by professional societies. Methods This was a secondary analysis of data from two large, multicenter studies in pregnant individuals (PAPR, NCT01371019 ; TREETOP, NCT02787213 ), originally conducted to discover, verify, and validate a serum proteomic predictor of preterm birth risk. Serum protein abundances were determined by mass spectrometry. Classifier models combined one or two novel protein ratio(s) with a composite clinical variable, denoted as ClinRisk3, which included prior PE, pre-existing hypertension, or pregestational diabetes. Predictive performance was assessed for the full validation cohort and for a subset that had early gestational age (GA) dating via ultrasound. Classifier performance was compared directly to the U.S. Preventive Services Task Force (USPSTF) algorithm for identification of pregnancies that should receive low-dose aspirin (LDASA) for PE prevention. Results Nine of nine prespecified classifier models were validated for risk of preterm PE with delivery <37 weeks’ gestation. Areas under the receiver operating characteristic curve ranged from 0.72-0.78 in the full validation cohort, compared to 0.68 for both ClinRisk3 alone and for the USPSTF algorithm. In the early GA dating subcohort, an exemplar predictor, ClinRisk3 + inhibin subunit beta C chain/sex hormone binding globulin (ClinRisk3+INHBC/SHBG) showed a markedly lower screen positive rate (11.1% vs 43.3%) and higher positive predictive value (13.0% vs 5.0%) and odds ratio (9.93 vs 5.24) than USPSTF. Its performance was similar in nulliparas and all parities. Conclusion Nine preterm PE risk predictors were identified, validated in an independent cohort, and shown to be more predictive than the USPSTF-endorsed algorithm. Our results indicate that a single blood test performed in the first half of pregnancy can be used for personalized PE risk assessment, particularly for pregnancies with minimal or no identified clinical risk factors, including nulliparas. Results can be used to guide personalized pregnancy management, including but not restricted to LDASA for PE prophylaxis, and serve as a basis for developing new prevention strategies. Funding Source Sera Prognostics, Inc. Précis Combining novel serum protein biomarkers and selected clinical variables for preterm preeclampsia prediction outperforms widely used clinical risk assessment algorithms currently recommended by practice guidelines.