Secondary Logo

Journal Logo

Contents: Original Research

Alpha-Fetoprotein as a Tool to Distinguish Amniotic Fluid From Urine, Vaginal Discharge, and Semen

Mor, Amir MD, PhD; Tal, Reshef MD, PhD; Haberman, Shoshana MD, PhD; McCalla, Sandra MD; Irani, Mohamad MD; Perlman, Jaqueline MD; Seifer, David B. MD; Minkoff, Howard MD

Author Information
doi: 10.1097/AOG.0000000000000635
  • Free

Rupture of fetal membranes is a significant event at any gestational age because it increases the likelihood of intra-amniotic infection and preterm birth. Premature rupture of membranes complicates approximately 3% of all pregnancies in the United States.1 Early diagnosis of rupture of membranes is critical, because there are several interventions that have been shown to decrease these risks.2

The diagnosis of rupture of membranes can be confirmed by several methods: the visualization of amniotic fluid passing from the cervical canal and pooling in the vagina; a basic pH test of vaginal fluid (the Nitrazine test); or arborization (ferning) of dried vaginal fluid.

The Nitrazine test is known to have low specificity and a low positive predictive value.3 A more accurate way to differentiate amniotic fluid from other bodily fluids might be quantification of proteins absorbed in a pad. Assuming that amniotic fluid contains a different protein repertoire than other bodily fluids,4 sampling the pad for specific protein(s) may assist in the definitive diagnosis of rupture of membranes.

We performed a prospective cohort study with a primary goal of characterizing and comparing alpha-fetoprotein (AFP) concentrations in amniotic fluid, semen, urine, and vaginal discharge. The secondary goals were: 1) to examine whether AFP can be extracted and quantified from sanitary pads and 2) to estimate whether the concentrations of AFP extracted from the pads can distinguish amniotic fluid from the other bodily fluids such as urine, semen, and normal vaginal discharge and therefore serve as a marker of rupture of membranes.


The study protocol was approved by the Maimonides Medical Center institutional review board. Written consent was obtained from all study participants before participation. Seventy-nine pregnant women at gestational age of 34 weeks or greater were recruited (Table 1) at the Maimonides Medical Center labor and delivery department from July 18, 2014, to September 2, 2014. Exclusion criteria were a history of malignant or benign gastrointestinal or urogenital tumors, abnormal maternal serum analytes, and fetal anatomical abnormalities that could affect AFP concentrations in bodily fluids. All women had intact membranes on admission. Fifty-two of the 79 women were admitted in labor. Those 52 women provided a urine specimen on admission. During the course of labor artificial rupture of membranes was performed on each of those 52 participants, and their amniotic fluid specimens were collected into cups. All urine specimens were collected before ruptures of membranes occurred. The other 27 pregnant participants with intact membranes were admitted for observation and antepartum management. These women denied leakage of fluid, had amniotic fluid index greater than 5 cm, and negative Nitrazine and pooling tests; normal vaginal discharge samples were collected from pads they wore for 4–6 hours. Seventeen semen specimens were collected from men at the hospital's fertility center. Alpha-fetoprotein concentrations in the urine, amniotic fluid, and semen specimens were quantified by a fully automated immunoassay. Samples from these specimens were also instilled in sanitary pads. One-square-centimeter patches were cut from pads that had been instilled with 1 mL urine, amniotic fluid, or semen. The patches were then immersed in 1 mL normal saline. Test tubes containing the 1 mL saline were sent for AFP concentrations quantification (using the same assay as described previously). One-square-centimeter patches were also cut from the 27 pads absorbed with normal vaginal discharge and underwent the same process of AFP extraction and quantification. As a result of limited sample volumes, AFP concentrations could not be measured directly from normal vaginal discharge and were measured only from the pads.

Table 1
Table 1:
Demographic Characteristics of Laboring Patients Who Had Artificial Rupture of Membranes Compared With Patients Who Were Admitted for Antepartum Management and Remained With Intact Membranes

Continuous variables were expressed as mean±1 standard deviation or ±1 standard error of the mean. The statistical analyses were performed using the Mann-Whitney test. Differences were considered statistically significant when P value <.05 (two-tailed analyses). A receiver operating characteristic (ROC) curve was generated for assessment of AFP as a marker of amniotic fluid. The sensitivity and specificity were calculated for the optimal AFP cutoff concentration determined by ROC curve analysis. Confidence intervals were provided where applicable. Statistical data were analyzed and calculated using SPSS software. Graphical illustrations were constructed using MedCalc software.


Alpha-protein concentrations were significantly higher in amniotic fluid than in either of the other substances (Table 2, left; P<.001 for amniotic fluid compared with urine and for amniotic fluid compared with semen). Furthermore, the distribution of AFP concentrations in amniotic fluid did not overlap with its distributions in urine or semen (Fig. 1). Because there was no overlap, ROC analyses were not performed in these cases.

Table 2
Table 2:
Alpha-Fetoprotein Concentrations (ng/mL) Measured Directly From Bodily Fluids or After Absorption Onto a Pad
Fig. 1
Fig. 1:
Alpha-fetoprotein concentrations measured directly from bodily fluids. Distributions of alpha-fetoprotein concentrations in amniotic fluid, urine, and semen. Amniotic fluid distribution does not overlap with urine or semen distributions.Mor. AFP as a Marker for Amniotic Fluid. Obstet Gynecol 2015.

We next examined AFP concentrations in amniotic fluid absorbed in pads and compared those results with AFP concentrations in urine, semen, and normal vaginal discharge also absorbed in pads. Pads absorbed with amniotic fluid contained significantly greater AFP concentrations than pads absorbed with normal vaginal discharge (Table 2, right; P<.001). Because the distributions had a minor degree of overlap (Fig. 2, top), ROC analysis was performed: with a cutoff concentration of 3.88 ng/mL, the sensitivity was 96.2%, specificity was 100%, and the area under the curve was 0.99 (Fig. 2, bottom). AFP was undetectable in the absorbed urine and semen samples.

Fig. 2
Fig. 2:
Alpha-fetoprotein concentrations from pads absorbed with amniotic fluid compared with pads absorbed with normal vaginal discharge. A. Distribution of alpha-fetoprotein concentrations in pads absorbed with vaginal discharge and amniotic fluid (logarithmic scale). Horizontal bars represent mean levels±1 standard deviation. The determined cutoff level is 3.88 ng/mL, for which the corresponding sensitivity is 96.2% (confidence interval [CI] 86.8–99.5%) and the corresponding specificity is 100% (CI 87.2–100%). B. Receiver operating characteristic curve analyses. The calculated area under the curve is 0.99 (CI 0.94–1.0).Mor. AFP as a Marker for Amniotic Fluid. Obstet Gynecol 2015.


The nature of this study is a proof of concept. We found that AFP concentrations in amniotic fluid are significantly higher than AFP concentrations in maternal urine or semen. The distributions of AFP concentrations in amniotic fluid do not overlap with those of semen or urine, suggesting that an assay for rupture of membranes based on direct measurement of AFP from fluids may be feasible.

When pads were absorbed with semen or urine, AFP could not be detected. Comparison of the mean AFP concentration in amniotic fluid collected in a cup with the mean concentration in amniotic fluid absorbed in pads showed approximately a 12-fold dilution. This was the result of the use of pad patches that were immersed in 1 mL saline. Alpha-fetoprotein concentrations from urine and semen samples that were originally low when directly measured were below the detection threshold when extracted from pads (Table 2, right).

When comparing AFP concentrations in pads absorbed with amniotic fluid with pads absorbed with normal vaginal discharge, there was a minimal overlap between the distributions. Based on our results, a “pad test” for rupture of membranes that compares AFP in amniotic fluid with AFP in normal vaginal discharge would have a sensitivity of 96.2% and a specificity of 100% when using a cutoff concentration of 3.88 ng/mL.

In our institution, the cost of a single quantitative AFP immunoassay is $9. Therefore, the pad test may be a relatively inexpensive technique to use when rupture of membranes is in doubt. Moreover, this pad test technique can be used by hospitals and outpatient facilities without the need for any new equipment or commercial rupture of membrane assay kits. Another advantage is the ability of a pad to absorb fluid over several hours. After rupture of membranes, amniotic fluid will continue to leak as long as there is a healthy fetus producing urine. However, in most cases, the leakage of fluid is intermittent rather than continuous. This is one of the reasons why commercial tests based on vaginal and cervical swab sampling may be less sensitive. For example, sampling the posterior fornix with the Nitrazine swab may be negative in the absence of pooling, even when rupture of membranes has already occurred (a false-negative result). A sanitary pad worn for some period of time is more likely to “catch” an episode of amniotic fluid leakage than are Nitrazine swabs that are inserted into the vagina for only a few seconds.

False-positive Nitrazine test results may occur in the presence of semen, alkaline antiseptics, or bacterial vaginosis because the pH of those fluids and secretions can be higher than 7.0 and in the range of amniotic fluid pH. Another advantage over the Nitrazine test is that the AFP immunoassay is not sensitive to a relatively high pH and therefore will not have false-positive results in the presence of semen.

On the other hand, maternal blood contains high concentrations of AFP5 so that vaginal bleeding will result in high concentrations of AFP extracted from a pad. Thus, the pad test may have false-positive results in the presence of blood. “Bloody show” is viscous and less likely to penetrate the pad. Therefore, the inner layers of the pad will remain dry in the absence of rupture of membranes (ie, a false-positive result is less likely). Proteins such as insulin-like growth factor binding protein 1 and placental α microglobulin-1 may be more specific to amniotic fluid,3 although this claim has been challenged recently.6 Assuming that maternal blood contains significantly lower concentrations of insulin-like growth factor binding protein 1, placental α microglobulin-1, or both, sampling pads for these proteins may be advantageous in cases of suspected rupture of membranes in the presence of vaginal bleeding.

Another potential problem in the clinical setting may be the sampling of low amniotic fluid volume that has been sitting in the vagina for a few hours. This sample might contain different protein properties than “freshly collected” samples. It is possible that AFP might degrade or become diluted and therefore undetectable. This problem can be overcome by continuing to use the pads, which will absorb fluids over time, capturing further events of leakage of fluid. True premature rupture of membranes with a healthy fetus will inevitably result in recurrent episodes of leakage of fluid. Sampling the fluids in standing or sitting up positions will avoid accumulation of fluid in the vagina for long periods of time.

Vaginal infection (eg, bacterial vaginosis, candidiasis, trichomoniasis) may also affect AFP concentrations and may, theoretically, cause a false-positive result. Because we did not measure vaginal secretion AFP concentrations in pregnant women known to be diagnosed with vaginal infection, we cannot comment on this potential challenge.

Our preliminary data suggest that extraction of AFP from absorbing pads can differentiate between amniotic fluid and other common secretions. If confirmed this would allow the use of a simple, relatively inexpensive rapid test to confirm the diagnosis of rupture of membranes. This pad test method could be used easily in any medical facility with standard equipment and without any need to purchase specific kits or extra equipment.

The pad test method may be particularly beneficial in the absence of vaginal bleeding and when rupture of membranes cannot be confirmed with the standard approach that includes the Nitrazine and ferning tests.


1. Waters TP, Mercer B. Preterm PROM: prediction, prevention, principles. Clin Obstet Gynecol 2011;54:307–12.
2. Premature rupture of membranes. Practice Bulletin No. 139. American College of Obstetricians and Gynecologists. Obstet Gynecol 2013;122:918–30.
3. Di Renzo GC, Roura LC, Facchinetti F, Antsaklis A, Breborowicz G, Gratacos E, et al.. Guidelines for the management of spontaneous preterm labor: identification of spontaneous preterm labor, diagnosis of preterm premature rupture of membranes, and preventive tools for preterm birth. J Matern Fetal Neonatal Med 2011;24:659–67.
4. Thomasino T, Levi C, Draper M, Neubert AG. Diagnosing rupture of membranes using combination monoclonal/polyclonal immunologic protein detection. J Reprod Med 2013;58:187–94.
5. Mizejewski GJ. Levels of alpha-fetoprotein during pregnancy and early infancy in normal and disease states. Obstet Gynecol Surv 2003;58:804–26.
6. Rutanen EM. Comment on: guidelines for the management of spontaneous preterm labor: identification of spontaneous preterm labor, diagnosis of preterm premature rupture of membranes and preventive tools for preterm birth. J Matern Fetal Neonatal Med 2012;25:546–9.
© 2015 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.