Preterm Birth (PTB), delivery of an infant prior to 37 weeks of gestation, causes significant morbidity and mortality among newborns.  Approximately 12% of infants were born prematurely in 2010, a 20% increase from 1990.  A major cause for morbidity and mortality among premature infants is respiratory distress syndrome (RDS) due to immature fetal lungs.  Fetal lung maturity can be hastened with antenatal corticosteroid injections at least 48 hours prior to a premature delivery and/or with intratracheal surfactant therapy in a preterm infant.  Therefore, identification of those women who will deliver imminently is important to reduce morbidity and mortality in the neonate.  Fetal Fibronectin (fFN) is the only test recommended by the American College of Obstetricians and Gynecologists for the prediction of preterm birth in women with and without symptoms of preterm labor.  So why are more and more hospitals discontinuing the use of fFN testing?

In order to understand this trend, I’d like to first describe fFN and discuss its diagnostic utility.  FFN is an extracellular matrix molecule located at the interface between the fetal chorionic and the maternal decidual membranes, where it is thought to be critical for implantation.  Just before full term delivery of an infant, fFN is shed into the cervicovaginal fluid (CVF).  The presence of fFN in CVF between 22 – 35 weeks gestation is associated with a higher risk for preterm delivery, but the diagnostic utility of FFN is really in its high negative predictive value (NPV).  In women with symptoms of preterm labor (contractions, abdominal pain and/or cramping), a negative fFN result confers a < 1% chance of delivery within 7 days and about a 1 – 3% chance of delivery within 2 weeks of sampling (i.e. NPVs of >99% and 97-99% for delivery in 7 and 14 days respectively).  A positive fFN result is far less useful with positive predictive values in the 15 – 40% range depending on the patient population.  The high NPVs for fFN testing are a bit misleading because, even in symptomatic women, only about 5% will actually deliver preterm.  Any diagnostic marker would demonstrate a similar NPV in a low disease prevalence.  The NPV of a coin flip in a disease with 5% prevalence is 95%.  Likelihood ratios are a better estimate of diagnostic accuracy.  In this case, the LR+ depicts how likely a positive fFN is associated with PTB, while LR- describes how likely a negative fFN result is associated with a full term birth.  A test with strong diagnostic accuracy in a rule out situation has a LR- of <0.1, while a LR+ >10 indicates a strong “rule in” test.  In a metaanalysis of over 5000 patients from 32 trials, the authors demonstrate that fFN does not have sufficient strength to rule in or out PTB within 7 days of testing.  Despite lack of diagnostic strength, physicians have used fFN testing for years to decide which high risk women to send home.  Even without fFN testing few women with symptoms of labor actually deliver preterm.  In those who do deliver preterm, antenatal corticosteroids and intratracheal surfactant therapies reduce the risk of significant morbidities associated RDS.  Given lack of diagnostic strength, low pretest probability for PTB and improved treatments for RDS, is it time to discontinue fFN testing in clinical laboratories?   

Suggested References:
1. Woodworth A et al.  Clin Chem 2007; 53: 1534 – 1540
2. Sanchez-Ramos et al.  Obstet Gynecol 2009;114:631-40