(514) Role of placental growth factor and placental T2* in predictive models of adverse pregnancy outcomes

Thursday, February 9, 2023

4:00 PM – 5:30 PM

Objective: Placental T2* (PT2*) measurements and serum placental growth factor (PlGF) are associated with adverse pregnancy outcomes (APOs) due to placental insufficiency. We aimed to determine the correlation between these two predictors and their performance in predictive models of APOs.

Study Design: Secondary cohort analysis of the Placenta Imaging Study, a prospective study of pregnant individuals enriched for risk factors for placental insufficiency. Participants had serial blood draws and up to three MRIs. Severe placental mediated APO composite (cAPO) included preeclampsia with severe features, perinatal death, delivery due placental insufficiency, and 5 min Apgar < 7. We used Pearson correlation to evaluate PlGF and PT2*. Logistic regression models were built containing clinical variables only, PlGF, PT2* z-score, and combinations of these. Measurements from T2 (22-29 6/7 wks) and T3 (30+ wks) were used. Model performance and test characteristics were reported after bootstrapping to reduce optimism and optimized model was chosen using these. Predicted probability of 10% was considered a positive test.

Results: 316 people were in the clinical model, 146 in T3 PlGF, and 207 in T3 PT2*. 65 (20.6%) experienced cAPO. Those with cAPOs were more likely to smoke, have chronic hypertension (cHTN), pregestational diabetes (DM), or asthma, have less education, be unpartnered, and use government insurance. Correlation between PlGF and PT2* was r=0.3473 (medium strength), p= < 0.001. PlGF and PT2* both had higher specificity for cAPOs than the clinical model but lower sensitivity (Table 1). The optimized model included cHTN, DM, tobacco use, body mass index < 18.5 kg/m², post-term, T2 PT2*, and T3 PlGF measurements, with c-statistic of 0.787 after shrinkage, sensitivity of 83.3%, specificity of 55.7%, positive predictive value of 29.9%, and negative predictive value of 93.7% (Fig. 2).

Conclusion: Incorporating PlGF and PT2* into predictive models of APOs improves discrimination, calibration, and test characteristics, particularly with regards to specificity. Such models may prove useful in risk stratification.

Primary & Presenting Author(s)

Karen J. Gibbins, MD, MSCI (she/her/hers)
Assistant Professor
Oregon Health and Science University
Portland, Oregon, United States

Coauthor(s)

VR
Victoria H. Roberts, PhD,BS
Assistant Professor
Oregon National Primate Research Center
Beaverton, Oregon, United States
JL
Jamie Lo, MD
Oregon Health and Science University
Portland, Oregon, United States
MS
Matthias Schabel, PhD
Oregon Health and Science University
Portland, Oregon, United States
BS
Bob M. Silver, MD
Chair OB/GYN
Univ. of Utah School of Medicine
Salt Lake City, Utah, United States
AF
Antonio E. Frias, MD
Professor
Oregon Health and Science University
Portland, Oregon, United States