Quantitative Fetal Fibronectin at 18 Weeks of Gestation to Predict Preterm Birth in Asymptomatic High-Risk Women

To compare quantitative fetal fibronectin measurement from 18 to 21 weeks of gestation to measurement at 22-27 weeks of gestation for the prediction of spontaneous preterm birth. In a prospective cohort study, we studied the accuracy of cervicovaginal fluid quantitative fetal fibronectin concentrations measured between 18 0/7 weeks of gestation and 21 6/7 weeks of gestation in high-risk asymptomatic women to predict spontaneous preterm birth before 34 weeks of gestation. Predefined fibronectin thresholds were 10 or greater, 50 or greater, and 200 ng/mL or greater. Diagnostic accuracy of the early test (n=898) was compared with the standard test performed between 22 0/7 and 27 6/7 weeks of gestation (n=691) in the same cohort. Subgroup analysis was performed according to cervical length measurement. Of 898 women, 8.7% delivered spontaneously before 34 weeks of gestation. Only 3.8% of the women with concentrations less than 10 ng/mL (65% of test results) delivered before 34 weeks of gestation. A concentration threshold of 10 ng/mL measured at 18 and 22 weeks of gestation had comparably high sensitivity (early 0.71, 95% confidence interval 0.60-0.81; standard 0.76, 0.63-0.87) and negative predictive value (early 0.96, 0.94-0.98; standard 0.97, 0.95-0.99) for delivery before 34 weeks of gestation. Specificity was also comparable (early 0.69, 0.65-0.72; standard 0.70, 0.66-0.74). A threshold of 200 ng/mL had high specificity (early 0.96, 0.94-0.98; standard 0.96, 0.94-0.97) with lower sensitivity (early 0.26, 0.17-0.37; standard 0.35, 0.22-0.49). Consideration of cervical length strengthened prediction. Quantitative cervicovaginal fetal fibronectin measured from 18 to 21 weeks of gestation has similar predictive value as measurement at 22-27 weeks of gestation for prediction of spontaneous preterm birth. Low fibronectin concentrations are associated with spontaneous preterm birthrates approaching population background levels.