The aim of this study was to evaluate the efficacy of the test for the decidual phosphorylated isoform of insulin-like growth factor binding protein-1 (phIGFBP-1) in endocervical secretions in predicting preterm delivery in women with uterine contractions.
The study included 210 women with a singleton pregnancy with documented uterine contractions and intact membranes at between 24 and 34 weeks' gestation who underwent the cervicovaginal phIGFBP-1 test and transvaginal sonographic measurement of cervical length. A receiver-operating characteristics (ROC) curve was used to determine the most useful cut-off point for cervical length. A multivariate logistic regression model was used in order to analyze the combination of significant predictive variables for preterm delivery following univariate analysis.
ROC curves indicated that 26 mm was the optimal cut-off value for cervical length in predicting preterm delivery. A cervical length of < 26 mm and the presence of phIGFBP-1 were statistically significant in univariate logistic regression analyses (P < 0.0001) with odds ratios of 16.18 and 9.29 for preterm delivery, respectively. Multivariate analysis of cervical length and phIGFBP-1 showed that they were independent and therefore useful in combination for predicting preterm delivery.
One of the most challenging unsolved problems in obstetrics worldwide is that of preterm births, which occur before 37 weeks' gestation and account for 5–9% of deliveries in Europe and 12–13% in the USA1, 2; 40–45% of preterm births follow spontaneous preterm labor1. About 5% of preterm births occur before 28 weeks' gestation, 15% at 28–31 weeks, 20% at 32–33 weeks, and c. 60% at 34–36 weeks1.
Recent attempts to predict preterm delivery have included the use of sonographic measurements of the cervix and biochemical markers. Considerable interest has been shown in developing biomarker assays for the prediction of preterm birth3–10, including one rapid test for determining the phosphorylated isoform of insulin-like growth factor binding protein-1 (phIGFBP-1) in cervical secretions. The non-phosphorylated isoform of IGFBP-1 is secreted by the fetal and adult liver, and is contained in amniotic fluid, fetal serum and maternal plasma5, with a concentration in amniotic fluid 100–1000 times higher than that in fetal serum7, 9. However, phIGFBP-1 is mainly secreted by maternal decidual cells, and may be an indicator of tissue damage of the choriodecidual interface5, 7. In the early stages of labor the fetal membrane begins to detach from the decidua and a small amount of phIGFBP-1 may be found in cervical secretions5–10. Kekki et al. have reported that a phIGFBP-1 concentration of at least 10 µg/L in a cervical swab sample indicates a 10-fold greater risk of preterm delivery9.
Four studies have compared the value of cervical phIGFBP-1 with measurements of cervical length5, 8, 11, 12. Bittar et al. evaluated the effectiveness of combining cervical length at 22–24 weeks' gestation with phIGFBP-1 tests in 105 asymptomatic pregnant women with a history of preterm birth, and concluded that both are useful in predicting preterm delivery in such patients11.
The aims of this study were to compare the performance of phIGFBP-1 and the sonographic measurement of cervical length in predicting preterm delivery in symptomatic patients and to determine whether they are independent of one another.
This prospective study involved 210 Caucasian women who received prenatal care between January 2006 and December 2006 in our departments of obstetrics in Padua and Milan. All the patients signed an informed consent form approved by the local Health Sciences in Human Subjects Committee. The inclusion criteria were a singleton pregnancy at 24–34 weeks' gestation, with documented uterine contractions (at least 10 per hour) and intact membranes. The exclusion criteria were > 2 cm dilatation of the cervix, having undergone a cervical examination or sexual intercourse less than 24 h previously, vaginal bleeding, placenta previa, multiple gestations, fetal abnormalities and uterine anomalies.
Gestational age was based on menstrual data, confirmed by an early first-trimester ultrasound scan; the other recorded data included demographic information, preterm labor management and delivery outcomes.
All the women were admitted to hospital and received an intravenous infusion of ritodrine; fetal lung maturation was accelerated by administering 24 mg of betamethasone (12 mg intramuscularly on two consecutive days) to the women with fetuses at less than 34 weeks' gestation.
Immediately upon admission, a rapid cervical sample for phIGFBP-1 determination (Actim Partus Test, Medix Biochemica, Kauniainen, Finland) was taken by means of a polyester-tipped swab during a speculum examination of the cervix, and extracted with specimen-extraction solution. The lower end of the swab was inserted into the external cervical orifice and left in place for about 10 s, after which it was placed in a test tube containing 0.5 mL of buffer solution. The dipstick was dipped into the samples and left for 15 s to allow the liquid front to enter the results area. After removing the dipstick from the solution and holding it for 5 min in a horizontal position, the test was interpreted as being positive, negative or invalid, respectively, when two, one or no blue lines appeared in the result area. The test is based on immunochromatography and has a detection limit of 10 µg/L.
Transvaginal sonography was also performed upon admission (after the phIGFBP-1 test) using a 5-MHz transvaginal probe placed in the anterior fornix, and cervical length was defined as the distance between the internal and external cervical ora. The cervix was measured three times and the mean value was calculated. A vaginal examination was also performed. The clinicians were informed of the test results, but no changes were made in the treatment of the positive or negative patients.
Student's t-test was used to compare the average values of the continuous variables, with the values being expressed as the mean ± SD. Pearson's chi-square test or Fisher's exact test was used to analyze the categorical variables, whose values are expressed as percentages. A receiver–operating characteristics (ROC) curve was constructed and used to determine the most useful cut-off point (with highest overall accuracy) for cervical length in predicting preterm delivery. Univariate logistic regression analysis was performed to assess the capability of the dichotomized cervical length and the phIGFBP-1 test to predict preterm birth, with positive and negative likelihood ratios (LR+ and LR−, respectively) also calculated for these variables. A multivariate logistic regression model was then developed using the forward stepwise method and 95% confidence levels in order to analyze the significant predictive variables in combination.
Two hundred and ten symptomatic women with singleton pregnancies were included in the study. The average maternal age of the patients delivering preterm was 30.7 ± 5.1 years, and that of the patients delivering at term was 30.4 ± 5.6 years (non-significant (NS) difference); their average body mass index was 22.72 ± 3.7 and 23.36 ± 3.99 kg/m2, respectively (NS). The prevalence of preterm delivery in the symptomatic patients was 16%.
The ROC curve showed that 26 mm was the best cut-off value for cervical length in terms of predicting preterm delivery (LR+, 3.69; LR−, 0.22) (Figure 1), with a sensitivity of 86.4%, specificity of 71.9%, positive predictive value (PPV) of 34.5% and negative predicting value (NPV) of 96.8% (Table 1). Preterm delivery occurred in 34.5% of the patients with a cervical length of < 26 mm as against only 3.2% of those with a cervical length of > 26 mm (P < 0.0001). The logistic regression of a cervical length of < 26 mm was statistically significant in predicting preterm delivery (P < 0.0001), and the odds ratio (OR) was very high (16.18; 95% CI, 4.46–58.66).
Table 1. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of cervical length < 26 mm and positive test for phosphorylated insulin-like growth factor binding protein-1 (phIGFBP-1) in predicting preterm delivery
Cervical length +
The qualitative phIGFBP-1 test was positive in 48.6% patients who delivered before 37 weeks' gestation (P < 0.0001). The mean gestational age at delivery was 36.4 weeks for patients with a positive phIGFBP-1 test and 38.4 weeks for patients with a negative phIGFBP-1 test (P = 0.002). The sensitivity, specificity, PPV and NPV of phIGFBP-1 were 52.9%, 89.2%, 48.7% and 90.8%, respectively (Table 1). The logistic regression of phIGFBP-1 was statistically significant in predicting pre-term delivery (P < 0.0001), with an OR of 9.29 (95% CI, 4.05–21.3), an LR+ of 4.89 and an LR− of 0.52.
Multivariate analysis of the dichotomized cervical length and the phIGFBP-1 test showed that the two variables were independent and therefore useful in combination to predict preterm delivery (χ2 = 39.25; P < 0.0001). The specificity of cervical length < 26 mm and presence of phIGFBP-1 was 96.1% (cervical length: P < 0.0001, OR 15.17; phIGFBP-1: P = 0.001, OR 7.37; Tables 1 and 2). Confounding factors such as parity, prior preterm deliveries, education and tobacco use were tested, but were not included in the final logistic model.
Table 2. Results of univariate and multivariate logistic regression analyses of cervical length < 26 mm and positive test for phosphorylated insulin-like growth factor binding protein-1 (phIGFBP-1) in the prediction of preterm delivery before 37 weeks' gestation
Odds ratio (95% CI)
Odds ratio (95% CI)
Recent attempts to predict preterm birth have included the use of ultrasonographic measurements of the cervix and/or biochemical markers. The use of transvaginal sonography to visualize the cervix has shown that cervical shortening is predictive of preterm birth13, 14. Kurkinen-Räty et al. found that a cervical length of 29.3 mm was the best cut-off point for predicting preterm births8, and their findings are in line with those of Iams et al., who found that a cervical length of < 30 mm had a sensitivity of 70% and a specificity of 69% in predicting birth before 35 weeks' gestation14.
Our own ROC curves showed that 26 mm was the best cut-off value for cervical length in terms of predicting preterm delivery, with a sensitivity and NPV of 86% and 97%, respectively.
The recent introduction of a new cervicovaginal test to detect phIGFBP-1 may improve the accuracy of predicting preterm delivery as we found that it had a high NPV of 91%. As already mentioned only four previously published studies have compared the value of cervical phIGFBP-1 with that of measuring cervical length5, 8, 11, 12. Bittar et al. investigated the efficacy of combining cervical length with an assessment of phIGFBP-1 in predicting preterm delivery in 105 asymptomatic pregnant women with a history of preterm birth, and found that measuring cervical length at 22–24 weeks' gestation and phIGFBP-1 at 30 weeks' gestation improved the prediction of preterm delivery over either method used alone (OR, 7.41)11. Eroglu et al. analyzed the prediction of preterm delivery in 51 symptomatic women (i.e. with uterine contractions) at between 24 and 35 weeks' gestation, and their univariate analysis showed that the NPV of phIGFBP-1 was 92.3%, with cut-offs of sonographic cervical length of < 20 mm and < 25 mm giving values of 91.1% and 90.5%, respectively12.
A prospective study of 57 symptomatic women (with signs and symptoms of preterm labor) at 24–36 weeks by Akercan et al. confirmed the high NPV of phIGFBP-1, obtaining a value of 90%5. Kurkinen-Räty et al. analyzed 77 symptomatic patients (i.e. with uterine contractions or shortening, opening or softening of the cervix) and found that the best cut-off value for cervical length was 29.3 mm at first examination (25–31 weeks), with a rate of prematurity of 21% if the canal was shorter than 29.3 mm, and a sensitivity of 82%8. At the same examination they found that a cervical phIGFBP-1 concentration exceeding 6.4 mg/L predicted preterm birth with a sensitivity of 19.2% and a specificity of 90.2% (NPV, 72%).
We assessed phIGFBP-1 in cervical secretions and the sonographic measurement of cervical length in 210 symptomatic patients, and found that their combination had an NPV of 90%, greater specificity and a better PPV than either method alone. The ORs of cervical length < 26 mm and presence of phIGFBP-1 for preterm birth were 15 and 7, respectively, with a χ2 of 39 (P < 0.0001), thus indicating that cervical length and the phIGFBP-1 test are independent variables that can be used together to predict preterm delivery.
The many proteins produced by the chorion-decidua include fetal fibronectin, analysis of whose levels is widely used clinically. However, we believe that analysis of phIGFBP-1 levels has many advantages because it is cheaper to assess and unaffected by vaginal bleeding, urine, seminal plasma, vaginal examination or transvaginal sonography6.
We conclude that a sonographically measured cervical length of > 26 mm with a negative phIGFBP-1 test in a patient with regular uterine contractions before 37 weeks' gestation seems to indicate a low risk of preterm delivery and may therefore allow the avoidance of unnecessary interventions with potentially hazardous medications.