Objective
The purpose of this study was to determine whether quantification of cervicovaginal fluid fetal fibronectin (fFN) improves diagnostic accuracy of spontaneous preterm birth (sPTB) in symptomatic women.
Study Design
A prospective blinded predefined secondary analysis of a larger study of cervicovaginal fluid fFN concentration (nanograms per milliliter) in women symptomatic of preterm labor (n =300 women; 22-35 weeks’ gestation) with a Hologic 10Q system (Hologic, Marlborough, MA). Clinicians were blinded to the result until after the delivery, but the qualitative Hologic TLI IQ fFN result was made available.
Results
The positive predictive value for sPTB (<34 weeks’ gestation) increased from 19%, 32%, 61%, and 75% with increasing thresholds (10, 50, 200, and 500 ng/mL, respectively). Compared with <10 ng/mL fFN, the relative risk of delivery was 5.6 (95% confidence interval [CI], 1.05–29.57), 7.9 (95% CI, 1.38–45.0), 22.8 (95% CI, 3.84–135.5), and 51.3 (95% CI, 12.49–211.2; P < .01).
Conclusion
Quantitative fFN provides thresholds (10 and 200 ng/mL) in addition to the qualitative method (50 ng/mL) to discriminate the risk of sPTB in symptomatic women.
Spontaneous preterm birth (sPTB) remains a challenging problem in obstetrics and is associated with significant neonatal morbidity and death. Contractions and preterm labor symptoms are poor indicators in symptomatic women as to who will actually deliver preterm. Accurate discrimination of those women who are at greatest risk would facilitate treatment, including targeted administration of steroids or in utero transfer, thereby potentially reducing neonatal death. There is a paucity of accurate tests to identify symptomatic pregnant women who are at highest risk of delivery, but the most commonly used test is a qualitative bedside test (Rapid fFN TLI IQ ; Hologic, Marlborough, MA) that analyses cervicovaginal fetal fibronectin (fFN). FFN is a glycoprotein found in amniotic fluid, placental tissue, and the extracellular component of the decidua basalis adjacent to the placental intervillous space. It is released after mechanical- or inflammatory-mediated damage to the membranes or placenta before birth. Cervicovaginal fluid fFN has a high negative predictive value for delivery in symptomatic women within 2 weeks of testing and has been established as a useful test in women who have symptoms that are suggestive of preterm labor. Positive prediction, however, is modest (<20%), and because most women have a good outcome, many are over treated.
The current qualitative test provides positive or negative result based on a threshold of 50 ng/mL. Qualitative tests that are based on a single threshold are prone to increasing false-positive/-negative results around the threshold, and clinicians do not equally weigh false-positive and false-negative tests. Studies that have used enzyme-linked immunosorbent assay (ELISA)–based quantification of fFN suggest that concentrations of fFN within cervicovaginal secretions correlate with the risk of sPTB and that a knowledge of the fFN concentration may improve prediction. The aim of this study was to evaluate a novel bedside quantitative system and to ascertain the predictive potential of prespecified threshold concentrations of fFN for the prediction of sPTB.
Materials and Methods
This was a prospective observational blinded study that involved a consecutive series of 300 symptomatic women with singleton pregnancies who underwent fFN sampling between 22 +0 and 35 +6 weeks’ gestation for symptoms that are suggestive of threatened preterm labor. Such women had all presented themselves to an emergency assessment unit with symptoms of threatened preterm labor that the attending clinician believed warranted an fFN test. According to hospital protocol and the licensing recommendations on fFN testing, women with previous vaginal examination, sexual intercourse (within 24 hours), cervical dilation >3 cm, frank bleeding, or rupture of membranes (on speculum examination) were excluded from the study.
The study was conducted from October 2010 through April 2012 at 5 hospitals in the United Kingdom. This investigation was a predefined substudy of evaluation of a quantitative instrument for prediction of preterm birth, which is a study of quantitative fFN for the prediction of sPTB and which was undertaken over the same period of recruitment. The study was approved by the South East London Research Ethics Committee, and all local research ethics committees that were associated with participating centers. Informed written consent was obtained from all participants. Gestational age was confirmed by early obstetric ultrasound examination.
Women with blood-stained swabs were excluded from analysis because of interference with fFN measurement. One aliquot (200 μL) of the sample was analyzed with the conventional qualitative Rapid fFN TLI IQ analyzer (Hologic), and another sample was analyzed with the quantitative Rapid fFN 10Q analyzer (Hologic) according to manufacturer’s instructions.
The reliability of the Rapid 10Q analyzer that was measured at 2 concentrations of fFN (53 and 156 ng/mL) demonstrates precision as 5.9% and 7.5% and accuracy as 98.1% and 93.1%, respectively. Experiments that were performed during product development confirmed a good correlation between ELISA and 10Q tests (slope = 0.97; r 2 = 0.82; personal communication, Hologic with Jerome Lapointe, September 2010).
All clinicians were trained in the use of the fFN analyzers, and the 2 tests were run concurrently. Categoric TLI IQ data (positive/negative) were provided to clinicians, but 10Q results remained double-blinded to the patient and clinician (a random result code was generated by the analyzer) until after delivery. Thresholds of 10, 50, 200, and 500 ng/mL were predefined before analysis, based on the literature. Statistical analysis was performed with STATA software (version 11.2; StataCorp LP, College Station, TX). Descriptive characteristics were calculated for baseline demographics. Thresholds for fFN were used to establish sensitivity, specificity, positive predictive value, and negative predictive value for spontaneous delivery within 14 days (primary endpoint) and a predefined outcome of delivery at <34 completed weeks’ gestation; receiver operator characteristic curves were generated. Results of fFN quantification were then grouped into the 5 prespecified incremental categories (0-9, 10-49, 50-199, 200-499, and ≥500 ng/mL), and the corresponding sPTB rates were calculated. The relative risk (relative to fFN of 0-9 ng/mL category) and exact 95% confidence intervals were calculated. A χ 2 test was used to determine statistical significance between fFN categories.
The category (0-9 ng/mL) was selected because the 10Q analyzer displayed a lower limit of detection than ELISA and was therefore equivalent to previous reports of “zero” with ELISA in terms of prevalence. Greater than 500 ng/mL was defined as the highest category because this was the upper limit to be reported by the 10Q analyzer and was anticipated to be prevalent in a symptomatic population.
Secondary outcomes that were analyzed included steroid administration and particularly evaluated steroid administration within 2 weeks of delivery. When the predictive value of each fFN concentration category for outcome <34 weeks’ gestation was assessed, all tests between 33 +0 and 33 +6 weeks’ gestation were excluded from the analysis to reduce false-positive test results because of proximity of test to endpoint. Data from women with iatrogenic delivery before the gestation of interest were also excluded. This analysis is of a subgroup of symptomatic patients from the larger study of quantitative fFN, which was powered on asymptomatic patients ( Figure 1 ). No formal power calculation was performed therefore for this subgroup analysis. Because this is the first study to evaluate the quantitative fFN system, the study was carried out in accordance to guidelines for the evaluation of the performance of a new diagnostic test, and standards for the reporting of diagnostic accuracy studies criteria were adhered to throughout ( Figure 1 ).
Results
Demographic and obstetrics characteristics for the study participants are described in Table 1 . A total of 300 women with singleton pregnancies were eligible for analysis (21 women were removed according to the listed exclusion criteria; Figure 1 ). The mean gestational age of testing was 29 +4 weeks’ gestation (SD, 3.82 weeks), and 5.7% of the study population delivered spontaneously within 14 days of testing. There were no adverse events related to the test. Overall, there was a sPTB rate of 8.7% at <34 weeks’ gestation and 12% at <37 weeks’ gestation.
| Characteristic | Value |
|---|---|
| Age, y a | 29 ± 6 |
| Body mass index, kg/m 2 | 26 ± 6 |
| Ethnicity, n (%) | |
| White | 153 (53) |
| Black | 94 (31) |
| Other | 53 (16) |
| Previous preterm birth, n (%) | 55 (18) |
| Previous premature prelabor rupture of membranes, n (%) | 12 (4) |
| Previous second-trimester miscarriage, n (%) | 15 (5) |
| Previous cervical surgery, n (%) | 15 (5) |
| Smoking history, n (%) | |
| Current | 37 (12) |
| Exsmoker | 38 (13) |
| Never | 223 (76) |
| History of domestic violence, n (%) | 29 (14) |
| History of recreational drugs, n (%) | 5 (2) |
The median gestational age at delivery was 38 +5 weeks’ gestation (SD, 3.22). Of the 300 participants, 209 of the participants (70%) had a spontaneous onset of labor; 91 women (30%) underwent induction or had a prelabor cesarean delivery. Concentrations of fFN were obtained for all women with the use of the 10Q system. A TLI IQ test was unsuccessful for 1 case; of the 299 TLI IQ tests that were performed, 67 patients (22%) had a positive cervicovaginal fluid fFN result.
Stepwise logistic regression confirmed which of the preselected cutoffs had the greatest predictive power ( R 2 = 0.13; P = .02) for delivery at <34 weeks’ gestation. These were <10 ng/mL (very-low risk) and ≥200 ng/mL (high-risk). The ≥50 ng/mL threshold was retained because it is the standard threshold that is used for the TLI IQ test. The threshold of ≥500 ng/mL was included because it was the upper limit of the device. Tables 2 and 3 summarize the data for the prediction of delivery at <14 days and sPTB at <34 weeks’ gestation for each of the 10Q fFN thresholds. The number of women with results falling within in each of the prespecified fFN categories is shown in Table 4 . In most participants (57%), the fFN concentration was <10ng/mL.
| Predictive variable | Fetal fibronectin threshold | |||
|---|---|---|---|---|
| 10 ng/mL | 50 ng/mL | 200 ng/mL | 500 ng/mL | |
| Sensitivity, % | 82.4 | 76.5 | 58.8 | 35.3 |
| Specificity, % | 59.3 | 81.1 | 93.9 | 97.5 |
| Negative predictive value, % | 98.2 | 98.3 | 97.4 | 96.1 |
| Positive predictive value, % | 10.9 | 19.7 | 37.0 | 46.2 |
| Likelihood ratio | ||||
| Plus | 2.02 | 4.04 | 9.69 | 14.12 |
| Minus | 0.30 | 0.29 | 0.44 | 0.66 |
| Receiver operator characteristic curve area | 0.71 | 0.79 | 0.76 | 0.66 |
| Relative risk (relative to fetal fibronectin 0-9 ng/mL) | 0.9 | 4.3 | 16.1 a | 26 a |
| Predictive variable | Fetal fibronectin threshold | |||
|---|---|---|---|---|
| 10 ng/mL | 50 ng/mL | 200 ng/mL | 500 ng/mL | |
| Sensitivity, % | 90.0 | 70.0 | 55.0 | 45.0 |
| Specificity, % | 64.0 | 85.7 | 96.7 | 98.6 |
| Negative predictive value, % | 98.5 | 96.8 | 95.8 | 95.0 |
| Positive predictive value, % | 19.4 | 31.8 | 61.1 | 75.0 |
| Likelihood ratio | ||||
| Plus | 2.52 | 4.90 | 16.5 | 31.5 |
| Minus | 0.16 | 0.35 | 0.47 | 0.56 |
| Receiver operator characteristic curve area | 0.77 | 0.78 | 0.76 | 0.72 |
| Relative risk (relative to fetal fibronectin 0-9 ng/mL) | 5.6 a | 7.9 b | 22.8 b | 51.3 b |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
