Short communication: Can we improve the diagnosis of rupture of membranes? The value of insulin-like growth factor binding protein-1
Dr M Boulvain, Division of Obstetrics, Department of Obstetrics and Gynaecology, University Hospitals of Geneva, Geneva CH-1211, Switzerland. Email email@example.com
Our objective was to assess the value of insulin-like growth factor binding protein-1 (IGFBP-1) and other tests for the diagnosis of rupture of the membranes (ROM). We included 49 women with suspected ROM. The gold standard for membranes status was defined based on clinical examination, ultrasonography, tests results (except IGFBP-1) and labour information. Sensitivity, specificity, positive predictive value and negative predictive value of each test were as follows, respectively: IGFBP-1 (86, 74, 73 and 87%); bromothymol (64, 100, 100 and 77%); fern test (62, 96, 93 and 75%) and ultrasonography (19, 100, 100 and 61%). The detection of IGFBP-1 in vaginal secretions has high sensitivity for the diagnosis of ROM.
Prelabour rupture of membranes (PROM) complicates about 10% of all pregnancies, and at least 60% of cases occur at term. Despite the recent advances in perinatal care, ROM before labour continues to lead to important morbidity in the mother and the infant. Some of these complications are associated with the PROM itself (infection, prematurity, cord prolapse) and others are associated with the clinical management (failed labour induction, prolonged labour, induced prematurity). Thus, a correct diagnosis of the state of fetal membranes is crucial.
Diagnosis of ROM is easy when the rupture is obvious, but it can be extremely difficult. Various tests have been used. Crystallisation test, with ferning pattern, and nitrazine test, a dye test for alkaline pH, have been criticised for their high number of false-positive and false-negative results. Ultrasonographic measurement of the amount of amniotic fluid is not likely to detect minor leakage, and the presence of oligohydramnios is not always caused by a ROM. Therefore, a rapid, reliable and noninvasive test would be of value to the clinician to diagnose ROM.
Insulin-like growth factor binding protein-1 (IGFBP-1) has been promoted as a reliable marker in this context of ROM. IGFBP-1 is a major protein produced by human decidua, and its concentration in the amniotic fluid is 100 to 1000 times higher than in the serum.1 The phosphorylation status of IGFBP-1 in the decidua differs from that in the amniotic fluid.2 Amniotic fluid contains nonphosphorylated and lesser phosphorylated isoforms whereas the decidua contains highly phosphorylated isoforms. Because of this difference, the origin of IGFBP-1 can be identified through the use of monoclonal antibodies. The presence of amniotic fluid isoforms of IGFBP-1 in cervicovaginal samples has been proposed by several authors for the diagnosis of ROM.3–6 In one of the studies, the results of the IGFBP-1 test were included in the definition of ROM, which may falsely increase the sensitivity and specificity of the test.3 Also, if women with certain diagnosis of membrane status are included, the identification is easy and the evaluation is biased.4,6 Furthermore, studies on the diagnosis of ROM are limited by the absence of a true ‘gold standard’.
The present study was undertaken to assess which of the tests (vaginal alkaline pH, ferning, ultrasonographic assessment or IGFBP-1) is the best diagnostic tool to detect ROM using the ‘best gold standard’ we could define.
Material and methods
From March to September 2000, 49 women, between 24 and 39 weeks of gestation, consulting the emergency obstetric unit with the complaint of vaginal fluid leakage were included in the study. We also performed the tests in a group of 20 women with certain ROM (women in labour between 36 and 41 weeks of gestation with artificially ruptured membranes) and in 14 women with intact membranes (women at their routine antenatal visit between 24 and 40 weeks of gestation without any complaint of vaginal leakage). The study was approved by the departmental Ethics Committee and all participants gave written informed consent.
Cervical digital exam was not performed to avoid infectious complications. A sterile speculum was used to expose the cervix and the posterior fornix. Clinical examination was considered as positive when the presence of fluid in the vagina and/or of cervical fluid leakage was observed. In case of total absence of vaginal secretions, ROM was also considered, as this might have been related to ‘washing’ of the vagina by the amniotic fluid leak. Three vaginal swabs were taken from the posterior fornix. Methods for each test are described below:
- • To measure pH, the swab was soaked with a bromothymol blue alcohol solution. The reaction was colorimetric and was considered as positive if the yellow colour changed to dark green. The result of this test was seen immediately and available to the clinician.
- • For the fern test or amniotic fluid crystallisation test, the absorbed fluid from the swab was put on a slide, dried and transported to the lab. All slides were read under a microscope by two independent investigators who were not aware of the clinical examination, the results of the other tests and the reading of the colleague. When results were discordant, samples were read and discussed with a third examinator to achieve agreement. Results were not available to the attending clinician.
- • For IGFBP-1 detection, we used the actim PROM test® (Medix Biochemica, Kuopio, Finland) according to manufacturer instructions. The absorbed fluid was removed into the extraction solution and tested with the dipstick. The result was read after 5 minutes and was considered as positive when two blue lines (sample and a positive control) appeared on the stick. Results of this test were not available to the attending obstetrician.
When ROM was diagnosed by the attending physician (based on clinical examination, vaginal pH and result of ultrasound), women were hospitalised. Women after 34 weeks of gestation were managed expectantly for 12 hours before induction of labour. For women admitted before 34 weeks of gestation, labour was induced when they reached 34 weeks of gestation or if signs of infection appeared. Maternal age, gestational age at inclusion, interval between suspected ROM and testing and interval between testing and delivery (hours) were recorded.
For study purposes, women were re-classified, after delivery, as having ruptured or nonruptured fetal membranes. This new classification was made by an investigator who had not been involved in the clinical examination. The following criteria were considered in this ‘gold standard’ definition: presence of amniotic fluid in the vagina, or total absence of vaginal secretions, alkaline pH (dark green colour), positive fern test (crystallisation), oligohydramnios (amniotic fluid index [AFI] less than 5 cm), chorioamnionitis, absence or very little amount of amniotic fluid leakage during labour and delivery. The results of IGFBP-1 were not included in the definition. It was the newest test and we did not want to artificially increase its sensitivity and specificity.
Sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) and likelihood ratio for a positive result (LR+) with their 95% confidence intervals were calculated for each test. We reported means and SDs for variables with a normal distribution and medians and ranges for other cases. Differences between groups were tested using the Student’s t test for the former and the Mann–Whitney U test for the latter. Analyses were performed using StatExact 4 (Cytel Software, Cambridge, MA) and StatsDirect (StatsDirect Ltd, Cheshire, UK).
In women with either definite ROM or known intact membranes (certain membrane status), sensitivity, specificity, PPV and NPV of each test were as follows, respectively: bromothymol (75, 100, 100, 74%); fern test (79, 92, 94, 75%); ultrasonography (18, 100, 100, 55%) and IGFBP-1 (100, 92, 95, 100%).
In the group of women with suspected ROM, median gestational age at admission was 39 weeks (range: 28–41), median interval from the suspected moment of ROM until clinical evaluation was 10 hours (range: 1–408) and the median interval from evaluation until delivery was 24 hours (range: 12–1152). Fluid was present in the vagina in 12 women. We considered, based on our gold standard definition, that 22 women (45%) had ruptured the fetal membranes and 27 women (55%) had not. The median interval between clinical evaluation and delivery was 24 hours in the group considered as with ruptured membranes, compared with 144 hours in women with intact membranes. This is explained by induction of labour in case of ROM after 34 weeks of gestation.
Three results for the fern test (slides lost or broken) and two for the AFI (ultrasound not performed) were not available for analysis. Results were doubtful in four cases for the bromothymol test and in one for the IGFBP-1 test. They were considered as negative in the analysis. Sensitivity, specificity, PPV, NPV and LR+ were calculated for each test (Table 1). Women who had a positive IGFBP-1 had shorter intervals from the suspected moment of rupture until consultation than women who had a negative result (median: 6 versus 23 hours; P= 0.007).
Table 1. Sensitivity, specificity, PPV, NPV and LR+ of the bromothymol and fern tests, AFI less than 5 cm and IGFBP-1 in women with suspected ROM
|Bromothymol||14/22 (64)||27/27 (100)||14/14 (100)||27/35 (77)||∞ (5.7 to +∞)|
|Fern test*||13/21 (62)||24/25 (96)||13/14 (93)||24/32 (75)||15.5 (3.0–88.9)|
|AFI < 5 cm**||4/21 (19)||26/26 (100)||4/4 (100)||26/43 (61)||∞ (1.4 to +∞)|
|IGFBP-1||19/22 (86)||20/27 (74)||19/26 (73)||20/23 (87)||3.3 (1.8–6.7)|
We analysed the cases of seven women considered as having intact membranes with the gold standard definition but who had a positive IGFBP-1. Four of them were at term, and they all went into labour spontaneously within 4 days. Another woman was in preterm labour at 32 weeks of gestation and delivered 27 days later. The remaining two women presented with a complete absence of vaginal secretions and no pooling of fluid at clinical examination. One had a story strongly suggesting ROM and the other had uterine contractions on admission. They were considered by the clinician in charge as having ruptured membranes, but they were re-classified as having intact membranes based on the fact that all criteria included in the definition of the gold standard were negative. There were also three false negatives for the IGFBP-1 test. Two women consulted after more than 4 days from the putative rupture and had severe oligohydramnios (AFI less than 2 cm) at 39 and 41 weeks of gestation. The third had a positive clinical examination, positive fern and bromothymol tests and normal AFI. All had labour induction.
In diagnosing the rupture of the fetal membranes, history and physical examination alone are often insufficient. For this reason, multiple cytological, biochemical, colorimetric and sonographic methods have been developed. However, most of these tests lack the necessary sensitivity and specificity in cases in which the diagnosis of ROM is clinically doubtful.
In many studies, definition of ROM is based on the presence of amniotic fluid at clinical examination.4,6 Jeurgens-Borst et al.4 used continuing vaginal fluid loss and/or ‘ruptured fetal membranes’ at first digital examination as gold standard. Therefore, there is a potential to overestimate both sensitivity and specificity of the clinical examination. A unique feature of our study is that we defined a ‘gold standard’ that included the clinical exam, two vaginal tests (amniotic fluid crystallisation and vaginal pH), the sonographic assessment of the amount of amniotic fluid and labour and delivery information. This last piece of information is not known at the time of admission and cannot be used for diagnosis in normal conditions. However, it is valuable for evaluating the accuracy of the available tests. We have not included the IGFBP-1 test results in the gold standard to avoid bias in the evaluation of its sensitivity. Nevertheless, the IGFBP-1 showed the highest sensitivity both in obvious (known membrane status) and in doubtful situations, as it has been shown in previous studies using different definitions of ROM.3
In our study, only half of the women with ruptured membranes had a positive clinical examination. The fern and the bromothymol tests also had lower sensitivities. This confirms the results from other studies. Results of the fern test were discordant between examinators in 8% (6/78) of the cases. This means that the fern test is dependent on the clinician’s experience and subjectivity. Interpretation of a dubious test is time consuming, which makes it difficult to perform in the emergency room. When comparing the prediction of these tests with those of IGFBP-1, it should be taken into account that the results of the former were included in the definition of the ‘gold standard’, thus overestimating their efficacy.
The IGFBP-1 test was not perfect either. We had seven false-positive and three false-negative cases. Among the former, there were two women with uterine contractions and four who went into labour within 4 days. Since phosphorylated isoforms of IGFBP-1 increase during labour, we speculate that high concentrations of these forms might have cross-reacted with the actim PROM test®. Therefore, in order to avoid false-positive results, one should probably not use this test in women with uterine contractions. Another explanation is that some of these women might have truly had a ROM but we were not able to correctly diagnose the rupture with our ‘gold standard’ definition. Among the false negatives, two women had isolated oligohydramnios and consulted after more than 4 days of the suspected rupture. This long interval might explain the negative results, as we have observed that women with positive IGFBP-1 results had shorter intervals between rupture and consultation. Another issue is the validity of diagnosis of ROM based solely on the presence of oligohydramnios. A reduced amount of amniotic fluid might represent other pregnancy complications, such as placental insufficiency. Whether these cases represent true false negatives of the IGFBP-1 test or wrong diagnosis of the ‘gold standard’ definition cannot be determined.
A weakness of our study is the small number of women included. Nevertheless, they all had several tests performed to diagnose the ROM and were classified using a gold standard definition that included more information than any other study used before.
We believe that the IGFBP-1 test shows promise where diagnosis of ROM is uncertain. Careful interpretation of IGFBP-1 results needs to be carried out when used in women with uterine contractions or prolonged intervals from rupture to consultation. Its place relative to other available tests needs examination in other larger studies of test accuracy.