Doi:10.1016/j.ajog.2005.05.074

American Journal of Obstetrics and Gynecology (2006) 194, 138–43 Selective use of fetal fibronectin detection after cervicallength measurement to predict spontaneous pretermdelivery in women with preterm labor Thomas Schmitz, MD,a,* Franc¸oise Maillard, MSc,b Sandrine Bessard-Bacquaert, MD,aGilles Kayem, MD,c Yvonne Fulla, MD,d Dominique Cabrol, MD, PhD,aFranc¸ois Goffinet, MD, PhDa,b Maternite´ Port-Royal,a De´partement de Me´decine Nucle´aire,d Hopital Cochin, AP-HP, Universite´ Rene´ DescartesParis V, INSERM U149,b Pavillon Baudelocque, Paris, France; Service de Gyne´cologie et Obste´trique,c CentreHospitalier Intercommunal de Cre´teil, Universite´ Paris XII, Cre´teil, France Received for publication January 11, 2005; revised April 18, 2005; accepted May 18, 2005 Objective: The purpose of this study was to determine whether selective use of fetal fibronectin detection after ultrasound measurement of cervical length predicts preterm delivery in symptomatic patients better than either indicator alone.
Study design: This prospective blinded study performed both tests on 359 women hospitalized for preterm labor between 18 and 34 completed weeks’ gestation. The primary outcome waspreterm delivery before 35 weeks’gestation.
Results: Among the 359 women included, 48 (13.4%) delivered before 35 weeks’ gestation. Thesensitivity, specificity, and positive and negative predictive values of cervical length %25 mm were75%, 63%, 24%, and 94%, respectively, and of fetal fibronectin R50 ng/mL, 63%, 81%, 33%,and 93%. Fetal fibronectin detection was significantly (P ! .001) more specific than cervicallength measurement. For selective use of fetal fibronectin detection after cervical lengthmeasurement, the test was considered positive if cervical length was %15 mm or if cervical lengthwas between 16 and 30 mm with fetal fibronectin R50 ng/mL. The predictive values of this testwere not significantly different from those of fetal fibronectin detection (67%, 81%, 36%, and94%). This strategy could have avoided 200 fibronectin tests.
Conclusion: Selective use of fetal fibronectin detection after cervical length measurement is morespecific than cervical length and as effective as fetal fibronectin assays in the entire population ofwomen in preterm labor for predicting preterm birth.
Ó 2006 Mosby, Inc. All rights reserved.
Preterm birth is the main cause of perinatal morbidity and mortThe rate of preterm delivery has notdecreased significantly the last 20 years and preterm * Reprint requests: Dr Thomas Schmitz, Maternite´ Port-Royal, birth still accounts for 6.8% of all deliveries in Franc 123 Boulevard de Port-Royal, 75014 Paris, France.
This situation can be explained, at least in part, by the 0002-9378/$ - see front matter Ó 2006 Mosby, Inc. All rights reserved.
doi:10.1016/j.ajog.2005.05.074 lack of reliable criteria for selecting populations at high General characteristics and pregnancy outcomes of risk for preterm delivery to provide them with special monitoring and prevention. Indeed, both cervical clin- ical examinationand uterine contraction frequency have poor sensitivity and specificity for predicting pre-term birth in patients with preterm labor and, thus, produce high rates of false-negatives and false-positives.
High false-negative rates result in unpredicted preterm birth of infants delivered without the benefits of corti- costeroids and tertiary neonatal care, whereas high rates of false-positives are responsible for unnecessary and potentially hazardous treatments.New markers for the prediction of preterm birth have therefore been devel- oped. Among them, transvaginal ultrasonographic mea- surement of cervical length and detection of fetal fibronectin in cervicovaginal fluids both show particular promise for improving the prediction rate for preterm delivery.However, despite a considerable literature, no clear, rational, and evaluated strategy for a daily practical use of these new markers has emerged, mainly because too few studieshave assessed the combined or selective use of fetal fibronectin detection after cervical length measurement for evaluating the risk of In the present prospective study, our goal was to Mean interval between inclusion and delivery determine whether the selective use of fetal fibronectin detection after ultrasound cervical length measurement was more effective than either indicator alone for predicting preterm delivery in patients with preterm eclampsia, or medically indicated preterm delivery be- fore 35 weeks’ gestation. Gestational age was establishedby the date of the last menstrual period and confirmed This prospective blinded study was conducted between by ultrasonography performed during the first trimester January 1997 and May 2000 at the tertiary university of pregnancy. If menstrual dates were unreliable or Port-Royal Maternity Hospital. The protocol was discordant by more than 5 days, gestational age was set approved by the local ethics committee (Conseil Consul- tatif Pour la Participation a` la Recherche Biome´dicale– Vaginal sampling for fetal fibronectin and cervical CCPPRB) of Cochin Hospital. All subjects gave length measurement were performed at admission dur- informed consent at the time of screening in accordance ing the same consultation in the emergency room of the with our institutional review board policies. Women maternity. Each subject was first examined with a were eligible for the study if they were hospitalized for vaginal speculum. A Dacron swab was rotated in the preterm labor at the maternity between 18 and 34 weeks posterior fornix of the vagina and sent to the labo- and 6 days of gestation. Preterm labor was defined by ratory. Fetal fibronectin concentrations were measured regular uterine contractions 30 seconds in duration at a by enzyme-linked immunosorbent assay (ELISA) with rate of at least 4 contractions per 30 minutes, confirmed the specific monoclonal antibody FDC-6. Results were by external uterine tocodynamometry, and cervical dila- tation of 0 to 3 cm (nulliparous women) or 1 to 3 cm Ultrasound examination of the cervix was performed (primiparous or multiparous women) and 50% cervical as soon as the uterine activity recording ended. Cervical effacement. Patients were excluded in case of cervical length was measured in the sagittal plane by the follow- manipulation or sexual intercourse within the previous ing standardized technique used in our maternity.
24 hours, multiple pregnancy, confirmed rupture of A Hitachi (Tokyo, Japan) EUB-405 ultrasonography membranes, cervical dilatation R3 cm, cervical cerclage, machine with a 6.5 MHz transvaginal transducer was uterine anomalies, vaginal bleeding, placenta previa, used. Each examination, performed with an empty abruptio placentae, intrauterine growth restriction, pre- bladder to avoid a deceptively elongated image, began Comparison of receiver-operator characteristic Comparison of receiver-operator characteristic curves constructed for cervical length and fetal fibronectin curves constructed for cervical length and fetal fibronectin levels in prediction of preterm delivery before 35 weeks’ levels in prediction of preterm delivery within 7 days. Areas gestation. Areas under the cervical length (0.758) and fetal under the cervical length (0.826) and fetal fibronectin (0.868) fibronectin (0.797) curves did not differ significantly (P = .47).
curves did not differ significantly (P = .28). Areas under both Areas under both curves differed significantly from the area curves differed significantly from the area (0.500) under the 45- (0.500) under the 45-degree diagonal line of unity (P ! .001 degree diagonal line of unity (P ! .001 and P ! .001, by placing the transducer on the cervix and identifying algorithm with Stata Software (College Station, TX) the internal cervical os, the cervical canal, and the and compared using the c2 test. For selective use of fetal external os. Pressure on the transducer was then relaxed fibronectin detection after cervical length measurement, to the point at which the image blurs before enough we retrospectively analyzed the predictive values of fetal gentle pressure to recreate a clear cervical image was fibronectin of ultrasonography-selected women. The cervical length cutoffs were chosen to provide high During the study period, results of ultrasound exam- sensitivity and specificity values. Predictive values and ination of the cervix and fetal fibronectin assay were likelihood ratios with their 95% confidence interval were blinded to the obstetric team and did not influence first calculated for each test considered separately, then subsequent patient management. Hospitalization was for the 2 markers used selectively and compared with decided exclusively on clinical grounds, ie, syste- matic digital examination of the cervix and uterinecontractions. Tocolytic therapy and corticosteroidswere administered on the discretion of the attending physician. Salbutamol was the main tocolytic drugprescribed and was maintained until contractions dis- Between January 1997 and May 2000, 815 women appeared. The infusion flow was then progressively consulted at Port-Royal Maternity Hospital for preterm reduced and finally withdrawn when judged possible.
labor. After exclusion of 137 women with multiple pregnancies, 123 with preterm premature rupture of The primary outcome measure was delivery before 35 membranes, 34 with both multiple pregnancy and pre- weeks’ gestation. The secondary outcome measure was term premature rupture of membranes, and 16 women delivery within 7 days following inclusion. To determine at a term equal to 35 weeks, 505 women were eligible for the most useful cutoff point for cervical length and fetal the study, 445 (88%) of whom had cervical length fibronectin, we constructed receiver-operator character- measured with ultrasound and 403 (80%) of whom istic (ROC) curves. The areas under the ROC curves had fetal fibronectin assayed. Reasons that 1 or both were calculated following the Delong and Clark-Pearson tests were not always performed for every eligible Predictive values for preterm birth of cervical length, fetal fibronectin, and selective use of fetal fibronectin in ultrasonography-selected patients (n = 359) Se (%) (95% CI) Sp (%) (95% CI) PPV (%) (95% CI) NPV (%) (95% CI) LRC (95% CI) To analyze selective use of fetal fibronectin in ultrasonography-selected patients (selective test), we considered the test positive if cervical length was%15 mm or between 16 and 30 mm with fetal fibronectin R50 ng/mL. The test was considered negative if cervical length was O30 mm or between 16and 30 mm with fetal fibronectin !50 ng/mL. CL, Cervical length; fFn, fetal fibronectin; Se, sensitivity; Sp, specificity; PPV, positive predictive value;NPV, negative predictive value; LRC, likelihood ratio for a positive result; LRÿ, likelihood ratio for a negative result.
* n = 48.
y n = 23.
z P = .24 vs CL.
x P = .29 vs CL.
{ P = 1.0 vs CL.
Predictive values for preterm birth of various cervical length cutoff values (n = 359) CL, Cervical length; Se, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictive value.
women included: no available ultrasound operator, were the best cutoff points for predicting preterm birth cervical manipulation or sexual intercourse within the previous 24 hours, or vaginal bleeding. Finally, 370 patients (73%) had both tests, but 2 were respectively sensitive but less specific than fetal fibronectin O50 ng/ excluded from the analysis because of labor induction mL for predicting preterm birth both before 35 weeks’ and 9 for planned cesarean section before 35 weeks.
Therefore, the final study population consisted of the We decided to evaluate fetal fibronectin detection in cervical ultrasonography-selected patients based firstly The general characteristics of the population are on the high sensitivity and negative predictive value of given in Preterm delivery rates before 35 weeks’ cervical length %30 mm for predicting preterm birth gestation and within 7 days following inclusion were and secondly on the high specificity of cervical length 13.4% (48/359) and 6.4% (23/359), respectively. Cervi- %15 mm. Indeed, when a 30 mm cutoff was chosen, the cal length and fetal fibronectin ROC curves were sensitivity of the test for predicting preterm birth before significantly above the 45-degree diagonal line of unity 35 weeks’ and within 7 days was 90% and 100%, but areas under the curves did not differ significantly respectively, with negative predictive values of 97% (). ROC curve analysis of cervical length and fetal fibronectin showed that 25 mm and 50 ng/mL mm, specificity was 89% for delivery before 35 weeks Predictive values for preterm birth of fetal fibronectin for women with cervical length between 16 and 30 mm (n = 159) fFn, Fetal fibronectin, Se, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictive value.
these new markers, used separately or in combination, in cervical length values between 16 and 30 mm ( any single population.Indeed, current protocols for fetal fibronectin detection was more specific than cervi- the management of preterm laborrely on data cal length measurement with a 25 mm cutoff (P ! .001) extracted from heterogenous populations. Second, se- (Therefore, to analyze selective use of fetal lective use of these new markers, by combining the fibronectin after cervical length measurement, we con- strength of the 2 tests, could further reduce the high sidered that patients were at high risk of preterm birth rates of false-positives resulting from the poor specificity if their cervical length was %15 mm or between 16 and of the clinical indicators. This false-positive rate results 30 mm with a fetal fibronectin concentration R50 ng/mL.
in prescription of unnecessary tocolytic and glucocorti- Patients were at low risk if their cervical length was coid therapies and in utero transfer. Third, performing O30 mm or between 16 and 30 mm with a fetal fetal fibronectin detection only in ultrasonography- fibronectin !50 ng/mL. In this 2-step strategy, the first selected patients could be helpful in reducing the num- step, based on cervical length measurement, is very sensitive (90% and 100%), and the second step, based We evaluated the selective use of fetal fibronectin on fetal fibronectin detection in patients with cervical detection after cervical length measurement for predict- length between 16 and 30 mm, is specific, with spec- ing preterm birth before 35 weeks’ gestation and within ificity values of 83% and 82% (The spec- 7 days following inclusion. By constructing ROC curves, ificity of fetal fibronectin detection used selectively as we first determined that 25 mm and 50 ng/mL were the mentioned here was significantly greater (P ! .001) best cutoff points for these 2 markers. These values are than that of cervical length measurement ( in accordance with what is commonly reported in Predictives values of the selective test and of fetal fibronectin detection in the entire population of women Fetal fibronectin detection in cervical ultrasonogra- in preterm labor did not differ significantly ( phy-selected patients provided excellent negative predic- However, this 2-step strategy could have avoided 200 tive values (94% and 99%) for a delivery both before 35 fibronectin tests thus reducing their number weeks’ and within 7 days. Fetal fibronectin used selec- tively in a daily obstetric practice could thus provide the Further analysis according to patient origin (trans- clinician with a specific test on which a decision not to ferred patients or not) or gestational age at admission treat may be based. Furthermore, the positive predictive (before or after 28 weeks’ gestation) did not modify the values, although less impressive, seem higher than those of cervical examination or uterine contraction fre-quency.Accordingly, this selective test may lead to areduction in the number of unnecessary or potentially hazardous treatments now administered because of thepoor capabilities of the clinical indicators. Fetal fibro- The aim of this study was to evaluate in the same nectin detection in patients selected by cervical ultraso- population a practical strategy for selective use of fetal nography predicts preterm birth before 35 weeks and fibronectin detection after cervical length measurement within 7 days with greater specificity than cervical length in women with preterm labor. Such studies are needed measurement and as effectively as fetal fibronectin in the for several different reasons. First, although use of entire preterm labor population but decreases the num- cervical length measurement and fetal fibronectin detec- ber of fetal fibronectin tests by 55%. This 2-step strategy tion in the management of women with preterm labor therefore requires fewer than half as many fetal fibro- has decreased false-positive and false-negative rates, nectin detection tests as the 1-step combination we there are very few reports of the predictive values of actually used, when we prescribed both tests to every patient. We must note, however, that selective use of detection in the entire preterm labor population for fetal fibronectin is not possible with the current test predicting preterm birth, and it reduces the number of because fetal fibronectin detection by ELISA provides fetal fibronectin ELISA tests performed by 55%. There- delayed results, and decisions about tocolysis and cor- fore, once an effective fetal fibronectin bedside test is ticoid administration cannot yet be postponed until available, fetal fibronectin might usefully be assayed only these results are ready. The benefits from the strategy in women with cervical lengths between 16 and 30 mm.
described above depend on the availability of a reliablebedside test as effective as ELISA. The practical interestof our results thus lies in their use in planning future studies to evaluate the impact of a bedside fibronectin 1. Stevenson DK, Wright LL, Lemons JA, Oh W, Korones SB, test performed only in ultrasonography-selected patients Papile LA, et al. Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Previous studies have compared the predictive values Research Network, January 1993 through December 1994. Am J of cervical length measurement and fetal fibronectin 2. Blondel B, Norton J, Mazaubrun C, Breart G. Evolution des detection within a single population. Rizzo et al, principaux indicateurs de la sante´ pe´rinatale en France me´tropol- however, did not present results regarding predictive itaine entre 1995 et 1998. Re´sultats des enqueˆtes nationales values for the combined use of sonography and fibro- pe´rinatales. J Gynecol Obstet Biol Reprod 2001;30:552-64.
nectin, while Rozenberg et alused a 1-step combina- 3. Iams JD, Casal D, McGregor JA, Goodwin TM, Kreaden US, tion of the 2 techniques (both tests performed for every Lowensohn R, et al. Fetal fbronectin improves the accuracy ofdiagnosis of preterm labor. Am J Obstet Gynecol 1995;173:141-5.
patient) and concluded that the additional information 4. King JF, Keirse MJNC, Chalmers I. Betamimetics in preterm about the fibronectin level provided only slight benefits.
labor: an overview of the randomised clinical trials. BJOG Gomez et al, on the contrary, reported recently that adding consideration of fetal fibronectin results to those 5. Gomez R, Galasso M, Romero R, Mazor M, Sorokin Y, of cervical length measurement significantly improved Goncalves L, et al. Ultrasonographic examination of the uterinecervix is better than cervical digital examination as a predictor of the prediction of preterm delivery.Their report did the likelihood of premature delivery in patients with preterm labor not, however, provide information about the predictive and intact membranes. Am J Obstet Gynecol 1994;171:956-64.
values of a combined test with cervical length and fetal 6. Leitich H, Brunbauer M, Kaider A, Egarter C, Husslein P.
fibronectin. The first group to evaluate a 2-step strategy Cervical length and dilatation of the internal cervical os detected was Hincz et al.They assayed fetal fibronectin in cases by vaginal ultrasonography as markers for preterm delivery:a systematic review. Am J Obstet Gynecol 1999;181:1465-72.
where the cervical length was between 21 and 31 mm 7. Lockwood CL, Senyei AE, Dische M. Fetal fibronectin in cervical and concluded that sequential use of cervical sonogra- and vaginal secretions as a predictor of preterm delivery. N Engl J phy and fetal fibronectin testing produced higher sensi- tivity and a better negative predictive value than either 8. Honest H, Bachmann LM, Gupta JK, Kleijnen J, Khan KS.
of these methods alone. Nevertheless, the primary end Accuracy of cervicovaginal fetal fibronectine test in predicting riskof spontaneous preterm birth: systematic review. BMJ 2002;325: point of their study was delivery within 28 days, which raises significant doubts about the clinical interest of 9. Rizzo G, Capponi A, Arduini D, Lorido C, Romanini C. The their results: because patients were included through 34 value of fetal fibronectin in cervical and vaginal secretions and of weeks, they may have delivered at term. The Ohio State ultrasonic examination of the uterin cervix in predicting premature protocol has also proposed performing fetal fibronectin delivery for patient with preterm labor and intact membranes.
Am J Obstet Gynecol 1996;175:1146-51.
detection only when clinical and sonography data are 10. Rozenberg P, Goffinet F, Malagrida L, Giudicelli Y, Perdu M, equivocal (eg, dilatation 2 cm and cervical length Houssin I. Evaluating the risk of preterm delivery: a comparison of between 20 and 30 mm).This strategy is based on fetal fibronectin and transvaginal ultrasonographic measurement findings in the literature from a variety of populations.
of cervical length. Am J Obstet Gynecol 1997;176:196-9.
Because the sensitivity and specificity of the selective test 11. Gomez R, Romero R, Medina L, Nien JK, Chaiworapongsa T, Carstens M, et al. Cervicovaginal fibronectin improves the predic- did not fall when we assayed fetal fibronectin only in tion of preterm delivery based on sonographic cervical length in ultrasonography-selected patients, our results confirm patients with preterm uterine contractions and intact membranes.
the strategy suggested by Iams et al.Its impact on preterm birth rates, duration of hospitalization, and 12. Hincz P, Wilczynski J, Kosarzewski M, Szaflik K. Two-step test: tocolytic treatments now needs to be evaluated.
the combined use of fetal fibronectin and sonographic examinationof the uterine cervix for prediction of preterm delivery in symp- In conclusion, fetal fibronectin in cervical ultraso- tomatic patients. Acta Obstet Gynecol Scand 2002;81:58-63.
nography-selected patients is more specific than cervical 13. Iams JD. Prediction and early detection of preterm labor. Obstet length measurement and as effective as fetal fibronectin

Source: http://www.ginecologiaeprevencao.com.br/pdf/fFN%20Am%20J%20Obstet%20Gynecol%202006.pdf