- Email: [email protected]
Changes in rates of salivary estriol increases before parturition at term
Changes in rates of salivary estriol increases before parturition at term Herman L. Hedriana, MD, Coralie J. Munro, PhD, Elaine M. Eby-Wilkens, BS, and Bill L. Lasley, PhD Davis, California OBJECTIVE: The aim of this study was to characterize the increases of salivary estriol concentrations before the onset of labor at term. STUDY DESIGN: Salivary estriol concentrations were measured in weekly patient-collected samples by means of a sensitive (mean ± SD threshold, 0.025 ± 0.001 ng/mL; coefficient of variation, 3.8%) direct enzyme immunoassay in a microtiter plate format. The salivary estriol concentrations in 16 healthy pregnant women were characterized from 30 weeks’ gestation until the time of parturition and delivery. Samples were stored frozen at collection and analyzed in batches after delivery. RESULTS: The median salivary estriol concentration profile revealed a nonlinear rise beginning from 30 weeks’ gestation (0.89 ng/mL) until term (2.70 ng/mL, an increase of 201%). At 35 weeks’ gestation the salivary estriol concentration median value increased sharply (positive inflection point, 50%-93% increase) at a demarcation between a slower increase during early pregnancy and a more rapid increase during late pregnancy. This positive inflection point associated with a late pregnancy increase characterized subgroups of pregnancies according to the lengths of gestation as follows: early term (delivered at <38 weeks 1 day’s gestation), middle term (delivered at 38 weeks 1 day–40 weeks’ gestation), and late term (delivered at >40 weeks’ gestation). Five weeks before delivery the mean (±SEM) rate of rise in salivary estriol concentration was 0.50 ± 0.13 ng/mL per week to 0.84 ± 0.26 ng/mL per week in the early term group. The increase in rate for the middle term group was 0.32 ± 0.06 ng/mL per week to 0.37 ± 0.26 ng/mL per week, whereas in the late term group the rate of salivary estriol concentration rise was 0.37 ± 0.03 ng/mL per week to –0.03 ± 0.25 ng/mL per week. CONCLUSION: These data demonstrate in normal pregnancies (1) that a direct, nonradiometric measure of salivary estriol concentration can be used to monitor the late pregnancy increase in estriol production, (2) that 35 weeks’ gestation marks a positive inflection point of the onset of increased estriol production, and (3) that the late pregnancy rise in salivary estriol concentration shows distinct patterns that tend to be characteristic of the length of pregnancy. These data support the concept that the rate of increase of estriol production is related to the timing of the onset of labor. (Am J Obstet Gynecol 2001;184:123-30.)
Key words: Labor, length of pregnancy, salivary estriol
The sequence of events leading to the onset of spontaneous labor in ewes has been well defined as a rapid increase in maternal plasma estrogen concentration and a slower rise in plasma progesterone concentration.1 The increased production of estrogen effects an increase in prostaglandin synthesis, facilitates expression of gap junctions between myometrial cells, ripens the cervix, and ultimately leads to labor and delivery. Similarly, the increase in estrogen production has been shown to precede From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, and the Department of Health, Population, and Reproduction, School of Veterinary Medicine, University of California at Davis. Received for publication November 24, 1999; revised February 2, 2000; accepted April 28, 2000. Reprint requests: Herman L. Hedriana, MD, 5301 F St, Suite 110, Sacramento, CA 95819. Copyright © 2001 by Mosby, Inc. 0002-9378/2001 $35.00 + 0 6/1/108338 doi:10.1067/mob.2001.108338
the onset of labor in lower primates, which has prompted the suggestion of an analogous phenomenon in human beings.2 Estriol is a placental aromatase conversion product of circulating 16-hydroxylated dehydroepiandrosterone produced in the fetal liver from fetal adrenal dehydroepiandrosterone sulfate.2 Unlike estradiol, the estriol in maternal circulation is almost entirely derived from fetal adrenal and liver steroid precursors. Serum unbound, unconjugated estrogen concentrations increase progressively with advancing gestation, with a zenith at or near spontaneous delivery.3 Serum assays for unconjugated estrogens measure both the bioactive (free) and the protein-bound forms of the steroid.4 Maternal salivary levels of estrone, estradiol, and estriol have been shown to reflect the unbound, unconjugated serum stores of these estrogens during pregnancy.3, 4 Among the salivary estrogen levels measured, salivary estriol concentration has been seen to increase rapidly 123
124 Hedriana et al
and copiously a few weeks before the onset of spontaneous delivery, unaffected by progesterone production.3, 4 Measurement by radioimmunoassay is relatively expensive, requires special equipment, and requires licensing because of the need for appropriate disposal of the radioisotopes to avoid contamination. Enzyme immunoassay, a practical and easier assay, has been shown to measure the rapid increase of salivary estriol concentration before the onset of spontaneous labor as well as does radioimmunoassay.5 At an arbitrary serum cutoff value, the salivary estriol concentration had limited sensitivity for spontaneous labor prediction. This may suggest that at a given extremely preterm gestational age the rate of salivary estriol production may be modest3, 4 but unique to the size of the fetal adrenals and liver. A relatively high cutoff value may therefore, it is speculated, fail to assign a risk of approaching labor. The purpose of this study was to characterize the rates of salivary estriol production during the last 10 weeks of gestation among healthy women and the association of these production rates with delivery at term (37-42 weeks’ gestation). Specifically, this study was conducted (1) to test the ability of a simple and economical assay to measure changes in salivary estriol production in pregnant women toward term, (2) to determine whether the changes in salivary estriol production near term are consistent among women and are predictive of the onset of parturition, and (3) to determine at what time point the basal levels of salivary estriol production change to indicate that the physiologic processes of parturition have been evoked. Material and methods Subjects. Women who attended the University of California, Davis, Medical Center obstetric clinics for prenatal care were recruited into the study after 30 weeks’ gestation. The dating criteria of all enrolled subjects were verified by dates of last menstrual period corroborated by a mid second-trimester ultrasonographic examination. Twenty women with normal pregnancies were enrolled in the study. Normal pregnancy was defined as follows: (1) no coexisting or preexisting medical problems that might change the course of otherwise routine obstetric care, (2) no risk factor associated with preterm delivery, (3) no severe hyperemesis or vaginal bleeding at <16 weeks’ gestation, (4) appropriate maternal weight gain, (5) singleton fetus, (6) appropriate fetal growth, (7) no fetal anomaly according to ultrasonography, and (8) normal placental architecture and location according to ultrasonography. All prenatal care began in the early second trimester. Delivery at term was defined as birth between 37 and 42 weeks’ gestation. The human subjects committee of the Institutional Review Board at the University of California, Davis, approved this study. The subjects were instructed to collect ≥3 mL saliva in a provided clean vial. The collection was done on a
January 2001 Am J Obstet Gynecol
weekly basis at approximately the same time between 9 AM and 9 PM. The gestational age at which the sample was collected was calculated from a confirmed estimated date of delivery. The subjects were asked to store the samples frozen in the freezer of the household refrigerator. The samples were collected from the subject’s home and transported to the laboratory to be stored in a –20°C freezer. All samples were analyzed at our laboratory as described here. Assay reagents. Antiserum against the immunogen estriol-6-carboxymethyloxime/bovine serum albumin (Steraloids, Inc, Newport, RI) was produced in 4 male New Zealand White rabbits 3 to 5 months old by an immunization protocol similar to that described by Munro and Lasley.6 Each rabbit received 1.0 mL (1.0 mg/mL) of the emulsified immunogen at 0, 2, and 4 weeks, with a booster administered approximately every 3 to 4 weeks thereafter until no further increase in titer was obtained. All 4 rabbits produced antiserum to estriol-6-carboxymethyloxime/bovine serum albumin; however, 2 rabbits, R4834 and R4835, had the highest titers, and these antisera were purified for further testing. The gamma globulin fraction of the antiestriol antisera was purified by ammonium sulfate precipitation, and the nonspecific anti–bovine serum albumin antisera were removed by equivalence zone adsorption as described by Munro and Stabenfeldt.7 Both antisera were specific for estriol, with all steroids and steroid metabolites structurally related to estriol showing <0.1% cross-reactivity in the enzyme immunoassay. The antiserum R4835 was chosen for use in this study. Horseradish peroxidase was coupled to the same derivative (carboxymethyloxime) at 2 different sites (3 and 6 positions) on the estriol molecule by the carboxylic acid group on the carboxymethyloxime group by means of the mixed anhydride procedure.7 The estriol-3carboxymethyloxime/horseradish peroxidase enzyme conjugate yielded a much more sensitive standard doseresponse curve than did the estriol-6-carboxymethyloxime/horseradish peroxidase. This indicated that a heterologous site of conjugation provided the more sensitive configuration for the estriol enzyme immunoassay. All studies were therefore conducted with the estriol-3carboxymethyloxime/horseradish peroxidase enzyme conjugate. After elution of the conjugate by Sephadex G25 chromatography (Amersham Pharmacia Biotech, Inc, Piscataway, NJ) in a 0.05-mol/L phosphate buffer, pH 7.5, the steroid-enzyme conjugate was divided into aliquots and stored at –15°C. The competitive solid-phase microtiter plate enzyme immunoassay for the measurement of unbound, unconjugated estriol was similar in configuration to the enzyme immunoassay for progesterone described by Munro and Stabenfeldt.7 Flat-bottomed Immulon 1 microtiter plates (DYNEX Technologies, Inc, Chantilly, Va) were coated
Hedriana et al 125
Volume 184, Number 2 Am J Obstet Gynecol
with 50 µL per well of the purified antiserum R4835 at a dilution of 1:12,000 in 0.05-mol/L bicarbonate buffer, pH 9.6. The plates were sealed tightly with waterproof plate sealer covers and incubated overnight at 4°C or stored at 4°C until use. On the day of the assay the enzyme conjugate (estriol-3-carboxymethyloxime/horseradish peroxidase) was diluted to 1:80,000 in assay buffer (0.1-mol/L phosphate-buffered saline solution containing 0.1% bovine serum albumin, pH 7.8). Estriol standards were diluted in assay buffer. The range of the standard curve was 0.5 to 250 pg/well. Saliva samples were diluted 1:1 with assay buffer, mixed in a vortex appliance for 1 minute, and centrifuged at 2000g for 30 minutes to remove insoluble materials. To compensate for the effects of saliva in the estriol enzyme immunoassay, the pH of the normal steroid enzyme immunoassay buffer (7.0) used in all other steroid enzyme immunoassay systems was changed to a pH of 7.8. At this higher pH, saliva samples from nonpregnant and male subjects showed a true 0 value for concentration of estriol (the 0 concentration optical density and the optical density readings for the male or nonpregnant saliva samples were the same). Raising the pH to 7.8 in the estriol enzyme immunoassay eliminated the need to compensate for the blank values that occurred in the presence of saliva when the standard enzyme immunoassay buffer (pH 7.0) was used. Assay procedure. Before the assay unbound antibody was removed from the wells with wash solution (0.15mol/L saline solution containing 0.05% [vol/vol] polysorbate 20). Each microtiter plate was considered a separate assay, with 2 standard curves and internal control runs formatted within the same plate. The assay procedure was as follows: 30 µL of assay buffer was pipetted across the entire plate, followed by 20 µL of standard estriol (Sigma, St Louis, Mo), then by diluted saliva sample or internal control preparation. This step was immediately followed by the addition of 50 µL of the estriol-3carboxymethyloxime/horseradish peroxidase conjugate solution (1:80,000) to all plate wells. The plates were covered tightly with plate sealer covers, and the competitive reaction was allowed to proceed for 2 hours. To separate free from bound hormone the plates were emptied, rinsed 5 times with wash solution, and blotted dry. Freshly prepared substrate solution (100 µL of 0.05mol/L citrate, 1.6-mmol/L hydrogen peroxide, 0.4mmol/L 2,2´-azino-di-[3-ethylbenzthiazoline sulfonic acid] diammonium salt, pH 4.0) was then added to all wells, and color was allowed to develop to determine the amount of conjugate (estriol-3-carboxymethyloxime/ horseradish peroxidase) bound to the solid-phase antibody. The color change was stopped after approximately 60 minutes by the addition of 100 µL stop solution (0.15mol/L hydrofluoric acid containing 0.006-mol/L sodium hydroxide, and 0.001-mol/L ethylenediaminetetraacetic acid, pH 3.3). Absorbance was measured at 405 nm with
an automatic microtiter plate spectrophotometer (Dynatech MR600; DYNEX), and the data were transferred to an interfaced computer for analysis. Calibration curves were constructed with weighted least-squares linear regression according to the method of Rodbard and Lewald,8 and the raw data were reduced by log transformation to yield concentrations of estriol in nanograms per milliliter. Assay sensitivity. Assay sensitivity was determined as the least amount of hormone that could be distinguished from 0 concentration of standard defined by the mass at 2 SD above the mean of the 0 absorbance. Sensitivity was 0.25 ± 0.044 pg/well, with a 50% displacement at 11.7 ± 2.1 pg estriol and a linear response of the standard curve through 250 pg/well. Serial dilutions of selected saliva samples from pregnant women produced displacement curves parallel to that of the estriol standard doseresponse curve up to 5 halving dilutions (25-0.78 µL saliva per well) in assay buffer (pH 7.8). A recovery of 96.25% ± 5.49% was achieved from spiked saliva samples. The intraplate coefficient of variation for salivary estriol concentrations averaged 3.3% to 5.4%, whereas the average interplate (intra-assay) coefficient of variation was 5.3% to 12.5%. The average interassay coefficient of variation for salivary estriol concentration was 7.5% to 15.5%. Data analysis. Because the whole data set had a skewed distribution along the y axis, median values for salivary estriol concentration were calculated for each week’s measurement. The mean and SD for each week were also calculated for comparison and to demonstrate the dispersion of data around the mean. Regression techniques were applied to the data from the last 5 weeks before delivery to calculate weekly rates of salivary estriol production (slopes), which were then compared. This subset of data (slopes) was tested to determine the degree of deviation from the predicted normal distribution by means of Kolmogorov-Smirnov statistics. This test validated that the slope data met the assumptions to use the robust parametric 2-way analysis of variance with repeated measures test to compare the changes in salivary estriol production through the last 5 weeks before delivery. Results All women enrolled in the study were delivered of live infants between 37 and 41 weeks’ gestation. There were 12 nulliparous and 8 multiparous women. The mean (±SD) age was 28 ± 6 years, with a range of 21 to 38 years. Of these 20 women 16 were included for analysis of salivary estriol concentration after delivery and 4 were excluded because of the unexpected appearance of obstetric complications necessitating cesarean delivery for nonreassuring fetal well-being before the onset of labor. Of the 4 women excluded, severe preeclampsia complicated the pregnancies of 2 women, 1 had poorly controlled gestational diabetes mellitus, and 1 had cholesta-
126 Hedriana et al
January 2001 Am J Obstet Gynecol
Fig 1. Salivary estriol production increase was nonlinear, almost in stepwise fashion, starting 10 weeks before delivery. Solid line, Smoothed weekly median salivary estriol concentration. Surge is seen 5 weeks before delivery. Correlation coefficient was 0.85 (P < .0001). To demonstrate scatter of salivary estriol concentration values (open circles) at each gestational week, mean weekly salivary estriol concentrations before delivery (broken line with filled squares) were plotted for comparison. Error bars, SD.
sis of pregnancy. Of the 16 women included in the study, 9 had spontaneous vaginal deliveries and 7 were delivered abdominally for obstetric indications during the active phase of labor. Although 10 women had spontaneous labor, 6 women had labor induced at >40 weeks’ gestation as follows: 3 for oligohydramnios, 1 for significantly decreased perception of fetal movement, and 2 for nonreactive nonstress tests. These indications have not been shown to affect steroid metabolism2 and were therefore not anticipated to affect analysis of the salivary estriol concentration data. The salivary estriol enzyme immunoassay had a mean (±SD) sensitivity of 0.025 ± 0.001 ng/mL saliva. The baseline median salivary estriol production at 30 weeks’ gestation was 0.90 ng/mL, and the highest median at term was 2.70 ng/mL, a percentage increase of 201%. Increase in salivary estriol production was nonlinear, almost in a stepwise fashion, starting 10 weeks before delivery (Fig 1). Because the distribution of the raw data was skewed on the y axis, the profile for the smoothed median concentration for salivary estriol for each weekly interval was compared side by side with the mean (±SD) concentration for the same week (Fig 1). Both the mean and median profiles revealed the same general trend of increasing salivary estriol concentrations with each progressive week of gestation. Both profiles also revealed a positive inflection point between 4 and 6 weeks before delivery. According to the median weekly levels the salivary estriol rate increases were 50% (from 1.38 to 2.06 ng/mL) from week
35 to week 36 and another 30% (from 2.06 to 2.67 ng/mL) from week 36 to week 37, for a total 93% increase between weeks 35 and 37 (Fig 2). Because of the salivary estriol concentration surge from 35 weeks’ gestation, we wanted to know whether changes in salivary estriol production were consistent among women who were delivered at various gestational ages at term. The salivary estriol concentration data points were divided into 3 groups: early term (delivery at <38 weeks 1 day’s gestation), middle term (delivery at 38 weeks 1 day–40 weeks’ gestation), and late term (delivery at >40 weeks’ gestation). Because of the bias in distribution of data points and the small number of women in each group, the median was again used to characterize salivary estriol production. As can be seen in Fig 3, salivary estriol production rates were different between groups from 35 weeks’ gestation to delivery. Women who were delivered by 38 weeks’ gestation had a surge in salivary estriol concentration from 1.78 ng/mL at 35 weeks’ gestation to 4.51 ng/mL at 38 weeks’ gestation (increase of 154%). In the middle term group the salivary estriol concentration median increased by 152% from 1.38 ng/mL to 3.47 ng/mL at 40 weeks’ gestation. Those who were delivered at >40 weeks’ gestation had a salivary estriol concentration increase of 94% (2.68 ng/mL). To evaluate the variation in rates of salivary estriol production among these groups of term deliveries, the slopes generated between weekly measurement of salivary estriol concentrations were compared. Because of
Volume 184, Number 2 Am J Obstet Gynecol
Hedriana et al 127
Fig 2. Salivary estriol production across last 10 weeks of pregnancy displayed as median salivary estriol production at each week of gestation. Surge of increased salivary estriol production is apparent at 35 weeks’ gestation.
Fig 3. Salivary estriol production across last 10 weeks of pregnancy displayed as median salivary estriol production increases as function of gestational age at delivery. Open circles, Patients who were delivered at <39 weeks’ gestation (early term). Open squares, Patients who were delivered at >40 weeks’ gestation (late term). Open triangles, Patients who were delivered between 39 and 40 weeks’ gestation (middle term). Note differences in salivary estriol production median rate curves.
the surge seen in Fig 1 we chose to compare the 5 weeks of salivary estriol production rates before delivery. This subset of data met the assumption of normality for use of the parametric 2-way analysis of variance with repeated measures after validation that the sample did not deviate
from the predicted normal distribution within the limits of the Kolmogorov-Smirnov test. The assumptions for variance and independence between groups were likewise met for the 2-way analysis of variance test. The mean slopes of the 3 groups were significantly different
128 Hedriana et al
January 2001 Am J Obstet Gynecol
Table I. Weekly rates of salivary estriol production 5 weeks before delivery Time before delivery (wk)
Early term (n = 7)
Middle term (n = 3)
Late term (n = 6)
5 4 3 2 1
0.50 ± 0.13 0.56 ± 0.13 0.68 ± 0.15 0.69 ± 0.17 0.84 ± 0.26
0.32 ± 0.06 0.35 ± 0.13 0.41 ± 0.10 0.43 ± 0.19 0.37 ± 0.26
0.37 ± 0.03 0.30 ± 0.05 0.25 ± 0.06 0.00 ± 0.27 –0.03 ± 0.25
The weekly production rates (slopes) in nanograms per milliliter per week were calculated between weekly mean salivary estriol concentrations and are presented as mean ± SEM. There were significant differences in mean weekly production rates as the groups approached delivery (P < .05, 2-way analysis of variance with repeated measures).
through the last 5 weeks before delivery (Table I; P < .05 analysis of variance). The women who were delivered at early term had the greatest rates of salivary estriol production, followed by those in the middle term group and then by those women who were delivered late. In the last 5 weeks before delivery the rates of salivary estriol production were significantly greater in the early and middle term groups but differed significantly between groups, whereas in the late term group salivary estriol production remained unchanged during the last 3 weeks after an initial 2-week surge (Fig 3). The median salivary estriol concentration in both the early and middle term groups 3 weeks before delivery was 1.78 ng/mL, whereas that in the late group was 2.65 ng/mL. In contrast to the strong association of the rise of salivary estriol rates with the time of delivery, no strong association was found between the absolute concentrations of salivary estriol and impending delivery in 3 weeks. Although the small number of women in the study was a limitation, the number of data points was adequate for the purpose of describing the behavior of salivary estriol concentrations before delivery at term. Comment Increases in unbound, unconjugated estriol production at 30 to 42 weeks’ gestation were reliably measured by means of a direct nonradiometric enzyme immunoassay from patient-collected saliva samples, which allowed us the opportunity to study changes in estriol production though pregnancy. We have described the development and validation of a sensitive and rapid solid-phase microtiter plate enzyme immunoassay for salivary estriol concentration in pregnant women. In this study the distinctive patterns of increasing salivary estriol production were associated with the different lengths of pregnancy, observations that have not been previously reported. Despite advances in parturition research, few facts are known about the mechanisms responsible for the initiation of labor resulting in subsequent delivery. In studies
with sheep and monkeys9, 10 the onset of uterine contractions associated with labor seemed to depend on changes in the concentrations of estrogens and progesterone in the fetal and maternal circulations. However, extrapolations of these data remain problematic in explaining the series of events before the onset of human labor. Adding to the controversy, many studies of human labor at term have not found associations with consistent significant changes in both bound and unbound concentrations of estrogens and progesterone and in the estrogen/progesterone ratio.11-13 Until recently, human studies have measured these forms of steroid hormones in plasma. In separate studies Lachelin and McGarrigle3 and Vining et al14 showed that salivary estrogen and progesterone concentrations were reflective of concentrations of the free or unbound biologically active hormones in the plasma pool, in contrast to plasma and urine assays, which measured both the bound and unbound forms of these steroids.3, 14 Because these steroids varied in their percentage binding with protein in plasma, important fluctuations in free steroid activity were not detected. The unbound form of these hormones can filter into secreting glands, such as the salivary glands. By measuring salivary steroid hormone production across gestation,3 estriol production was noted to increase exponentially 5 to 6 weeks before term delivery.4 At this period in gestation salivary estriol concentration was found to be 15 times greater than that of estradiol in healthy pregnant women.15 Similar patterns were also observed at this period among patients who were delivered preterm (delivery at <37 weeks’ gestation).16 The estriol increase was parallel with but independent of the slow increase in progesterone production. These observations3, 4, 15, 16 suggested to us that the rate of rise of estriol production, rather than the absolute concentration, might be more predictive of impending labor and delivery. This concept became apparent when subjects’ weekly salivary estriol concentrations were traced side by side in this study. Our subjects who demonstrated an upsurge in salivary estriol production 4 to 5 weeks before delivery followed by a continued increase tended to be delivered earlier than the subjects who did not show a persistent increase in salivary estriol production. In addition, the median salivary estriol concentrations in women who were in labor early were lower than those in the subset of women who were delivered at >40 weeks’ gestation. In contrast to the strong association of the rise in salivary estriol production rate with the time of labor onset and delivery, no strong association was found between the absolute concentration of salivary estriol and impending labor and delivery. However, quite unexpectedly, the median salivary estriol concentration values plateaued at 37 weeks’ gestation in those subjects who had labor induced after 40 weeks’ gestation for common obstetric indications.
Hedriana et al 129
Volume 184, Number 2 Am J Obstet Gynecol
Together with salivary estriol production, estrone and estradiol values, including the estriol/progesterone ratio, increased 5 to 6 weeks before the onset of labor at term.4, 5, 15, 17 This enormous production of salivary estriol was consistent before preterm deliveries,5, 16 but neither estriol nor progesterone production increased significantly among women whose pregnancies were complicated by premature rupture of membranes at term and preterm who were not in labor.15, 16 The initiation of labor after rupture of the fetal membranes may not be estrogen driven but may result from another pathway initiated by disruption of the fetal membranes.5 In the same way increases in estriol and progesterone production were not seen among subjects who had continuation of the pregnancy to >40 weeks’ gestation.17 Although the fetal and placental hypothalamic-pituitary-adrenal axis has been implicated in pathways leading to labor in recent reports,18-20 the physiogenesis of this phenomenon has remained largely unexplained. In cases in which the continued increase of salivary estriol production with advancing gestation was not observed, the ability to produce estriol through placental regulation may be affected by decreased capacity to aromatize circulating fetal, adrenal-derived 16-hydroxylated substrates in aging placenta.2, 6, 21, 22 The limiting factor in this condition could be the inadequate supply of essential precursors from the fetal liver and fetal adrenals, much the same as was seen in the initiation of preterm labor in rhesus monkeys infused with extraneous androstenedione, an adrenal-derived placental precursor.23 In these monkeys the increase in serum androstenedione level paralleled the rise of serum dehydroepiandrosterone sulfate, which led to increased frequency of laborassociated uterine contractions. We speculate that in a normal pregnancy the availability of enormous amounts of placental estriol precursors may begin at 35 weeks’ gestation, as evidenced by the positive inflection point in this study. That supply of precursors for unknown reasons levels off, resulting in the delay of labor onset and delivery. No physical or biochemical changes in the placentas were demonstrated in our subjects who had labor induced at >40 weeks’ gestation. All 6 placentas appeared grossly normal, and all 4 of the placentas examined histopathologically were considered normal. The 6 neonates had normal liver and adrenal functions. In a series of studies with pregnant women at >40 weeks’ gestation Moran et al17 showed no demonstrable significant increases in salivary estriol concentration and in estriol/progesterone ratio after 40 weeks’ gestation. All their subjects had labor induced at 42 weeks’ gestation. In another study Moran et al24 administered estriol rectally to women with normal term pregnancies at 41 to 42 weeks’ gestation and reported an increase in plasma estriol level and an associated decrease in plasma progesterone level. The onset of spontaneous labor and sub-
sequent delivery was achieved in 5 of 17 subjects after 48 hours. Compared with the radioimmunoassay techniques reported earlier3, 4, 15, 16 the enzyme immunoassay used in this study has similar qualities but has the advantage of not requiring radioactive reagents. The enzyme immunoassay used and described here is not proprietary and is not the same as the SalEst system (Biex, Inc, Dublin, Calif).5, 25 The salivary estriol concentration enzyme immunoassay in this study has not been previously reported, which is why a relatively lengthy Methods section was required in this report. This salivary estriol concentration enzyme immunoassay has adequate specificity, sensitivity, and robustness to make it useful for research protocols in which processing a large quantity of samples and economy are high priorities. Recent studies5, 25 have shown the versatility of the clinical applications of a salivary estriol concentration enzyme immunoassay. When used as an aid to assess the likelihood of preterm labor and delivery, an increase in estriol production was demonstrated 3 to 4 weeks before delivery in a heterogeneous population from multiple medical centers.5 With an optimal salivary estriol concentration cutoff value of 2.3 ng/mL between 22 and 36 weeks’ gestation, McGregor et al5 correctly discriminated 73% of women who were subsequently delivered. If the test result was negative (salivary estriol concentration <2.3 ng/mL), there was a 95% likelihood that delivery would not occur within 2 to 3 weeks. When tested against the Creasy traditional risk factor assessment test, the salivary estriol concentration correctly predicted 87% of preterm births versus 7.2%.25 Despite these clinical applications, the positive prediction for preterm birth is low when a mean cutoff value is used. This may be due to the large variation in salivary estriol concentration values among patients at the same gestational age, as previously explained here and observed in other studies.3, 26 Despite the limitation of inadequate sample size in this study, the behavior of salivary estriol concentration production weeks before delivery that we have described is consistent with various studies that have used different assays.3-5, 15-17, 23-26 Although firm conclusions are beyond the scope of this study, improvement in prediction of preterm births could possibly be obtained by determining the rate of salivary estriol concentration production across gestational week intervals after correct identification of the baseline levels. We are indebted to Jiangang Chen for his assistance in statistical analysis. REFERENCES
1. Liggins GC, Fairclough RG, Grieves SA, Kendall JZ, Knox BX. The mechanism of initiation of parturition in the ewe. Recent Prog Horm Res 1973;29:111-59. 2. Word RA. Parturition. In: Carr BR, Blackwell RE, editors. Textbook of reproductive medicine. East Norwalk (CT): Appleton and Lange; 1993. p. 41-8.
130 Hedriana et al
3. Lachelin GC, McGarrigle HH. A comparison of saliva, plasma unconjugated and plasma total oestriol levels throughout normal pregnancy. Br J Obstet Gynaecol 1984;91:1203-9. 4. Darne J, McGarrigle HH, Lachelin GC. Saliva oestriol, oestradiol, oestrone and progesterone levels in pregnancy: spontaneous labour at term is preceded by a rise in salivary oestriol: progesterone ratio. Br J Obstet Gynaecol 1987;94:227-35. 5. McGregor JA, Jackson GM, Lachelin GC, Goodwin TM, Artal R, Hastings C, et al. Salivary estriol as risk assessment for preterm labor: a prospective trial. Am J Obstet Gynecol 1995;173: 1337-42. 6. Munro CJ, Lasley B. Non-radiometric methods for immunoassay of steroid hormones. In: Albertson BD, Haseltine FP, editors. Non-radiometric assays: technology and application in polypeptide and steroid hormone detection. New York: Alan R Liss; 1988. p. 289-329. 7. Munro C, Stabenfeldt G. Development of a microtitre plate enzyme immunoassay for the determination of progesterone. J Endocrinol 1984;101:41-9. 8. Rodbard D, Lewald JE. Computer analysis of radioligand assay and radioimmunoassay data. Acta Endocrinol Suppl (Copenh) 1970;147:79-103. 9. Csapo AI. Progesterone “block.” Am J Anat 1956;98:273-91. 10. Challis JR, Olson DM. Parturition. In: Knobil E, Neill J, editors. The physiology of reproduction. New York: Raven Press; 1988. p. 2177-216. 11. Cousins LM, Hobel CJ, Chang RJ, Okada DM, Marshall JR. Serum progesterone and estradiol-17β levels in premature and term labor. Am J Obstet Gynecol 1977;127:612-5. 12. Thornburg GD, Challis JR. Control of parturition. Physiol Rev 1979;59:863-918. 13. Willcox DL, Yovich JL, McColm SC, Phillips JM. Progesterone, cortisol and oestradiol-17 beta in the initiation of human parturition: partitioning between free and bound hormone in plasma. Br J Obstet Gynaecol 1985;92:65-71. 14. Vining RF, McGinley R, Rice BV. Saliva estriol measurements: an alternative to the assay of serum unconjugated estriol assessing feto-placental function. J Clin Endocrinol Metab 1983;56: 454-60.
January 2001 Am J Obstet Gynecol
15. McGarrigle HH, Lachelin GC. Increasing saliva (free) oestriol to progesterone ratio in late pregnancy: a role for oestriol in initiating spontaneous labour in man? BMJ 1984;289:457-9. 16. Darne J, McGarrigle HH, Lachelin GC. Increased saliva oestriol to progesterone ratio before idiopathic preterm delivery: a possible predictor of preterm labour? BMJ 1987;294:270-2. 17. Moran DJ, McGarrigle HH, Lachelin GC. Lack of normal increase in saliva estriol/progesterone ratio in women with labor induced at 42 weeks’ gestation. Am J Obstet Gynecol 1992; 167:1563-4. 18. Chrousos GP. Reproductive placental corticotropin-releasing hormone and its clinical implications. Am J Obstet Gynecol 1999;180(1 Pt 3):S249-50. 19. Hobel CJ, Dunkel-Schetter C, Roesch SC, Castro LC, Arora CP. Maternal plasma corticotropin-releasing hormone associated with stress at 20 weeks’ gestation in pregnancies ending in preterm delivery. Am J Obstet Gynecol 1999;180(1 Pt 3):S257-63. 20. Lockwood CJ. Stress-associated preterm delivery: the role of corticotropin-releasing hormone. Am J Obstet Gynecol 1988;180(1 Pt 3):S264-6. 21. Nathanielsz PW. The role of basic science in preventing low birth weight. Future Child 1995;5:57-70. 22. Olson DM, Mijovic JE, Sadowsky DW. Control of human parturition. Semin Perinatol 1995;19:52-63. 23. Mecenas CA, Giussani DA, Owiny JR, Jenkins SL, Wu WX, Honnebier BO, et al. Production of premature delivery in pregnant rhesus monkeys by androstenedione infusion. Nat Med 1996;2:443-8. 24. Moran DJ, McGarrigle HH, Lachelin GC. Maternal plasma progesterone levels fall after rectal administration of estriol. J Clin Endocrinol Metab 1994;78:70-2. 25. Heine RP, McGregor JA, Dullien VK. Accuracy of salivary estriol testing compared to traditional risk factor assessment in predicting preterm birth. Am J Obstet Gynecol 1999;180(1 Pt 3): S214-8. 26. Darne J, McGarrigle HH, Lachelin GC. Diurnal variation of plasma and saliva oestrogen, progesterone, cortisol and plasma dehydroepiandrosterone sulfate in late pregnancy. Eur J Obstet Gynecol Reprod Biol 1989;32:57-66.
Key words: Labor, length of pregnancy, salivary estriol
The sequence of events leading to the onset of spontaneous labor in ewes has been well defined as a rapid increase in maternal plasma estrogen concentration and a slower rise in plasma progesterone concentration.1 The increased production of estrogen effects an increase in prostaglandin synthesis, facilitates expression of gap junctions between myometrial cells, ripens the cervix, and ultimately leads to labor and delivery. Similarly, the increase in estrogen production has been shown to precede From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, and the Department of Health, Population, and Reproduction, School of Veterinary Medicine, University of California at Davis. Received for publication November 24, 1999; revised February 2, 2000; accepted April 28, 2000. Reprint requests: Herman L. Hedriana, MD, 5301 F St, Suite 110, Sacramento, CA 95819. Copyright © 2001 by Mosby, Inc. 0002-9378/2001 $35.00 + 0 6/1/108338 doi:10.1067/mob.2001.108338
the onset of labor in lower primates, which has prompted the suggestion of an analogous phenomenon in human beings.2 Estriol is a placental aromatase conversion product of circulating 16-hydroxylated dehydroepiandrosterone produced in the fetal liver from fetal adrenal dehydroepiandrosterone sulfate.2 Unlike estradiol, the estriol in maternal circulation is almost entirely derived from fetal adrenal and liver steroid precursors. Serum unbound, unconjugated estrogen concentrations increase progressively with advancing gestation, with a zenith at or near spontaneous delivery.3 Serum assays for unconjugated estrogens measure both the bioactive (free) and the protein-bound forms of the steroid.4 Maternal salivary levels of estrone, estradiol, and estriol have been shown to reflect the unbound, unconjugated serum stores of these estrogens during pregnancy.3, 4 Among the salivary estrogen levels measured, salivary estriol concentration has been seen to increase rapidly 123
124 Hedriana et al
and copiously a few weeks before the onset of spontaneous delivery, unaffected by progesterone production.3, 4 Measurement by radioimmunoassay is relatively expensive, requires special equipment, and requires licensing because of the need for appropriate disposal of the radioisotopes to avoid contamination. Enzyme immunoassay, a practical and easier assay, has been shown to measure the rapid increase of salivary estriol concentration before the onset of spontaneous labor as well as does radioimmunoassay.5 At an arbitrary serum cutoff value, the salivary estriol concentration had limited sensitivity for spontaneous labor prediction. This may suggest that at a given extremely preterm gestational age the rate of salivary estriol production may be modest3, 4 but unique to the size of the fetal adrenals and liver. A relatively high cutoff value may therefore, it is speculated, fail to assign a risk of approaching labor. The purpose of this study was to characterize the rates of salivary estriol production during the last 10 weeks of gestation among healthy women and the association of these production rates with delivery at term (37-42 weeks’ gestation). Specifically, this study was conducted (1) to test the ability of a simple and economical assay to measure changes in salivary estriol production in pregnant women toward term, (2) to determine whether the changes in salivary estriol production near term are consistent among women and are predictive of the onset of parturition, and (3) to determine at what time point the basal levels of salivary estriol production change to indicate that the physiologic processes of parturition have been evoked. Material and methods Subjects. Women who attended the University of California, Davis, Medical Center obstetric clinics for prenatal care were recruited into the study after 30 weeks’ gestation. The dating criteria of all enrolled subjects were verified by dates of last menstrual period corroborated by a mid second-trimester ultrasonographic examination. Twenty women with normal pregnancies were enrolled in the study. Normal pregnancy was defined as follows: (1) no coexisting or preexisting medical problems that might change the course of otherwise routine obstetric care, (2) no risk factor associated with preterm delivery, (3) no severe hyperemesis or vaginal bleeding at <16 weeks’ gestation, (4) appropriate maternal weight gain, (5) singleton fetus, (6) appropriate fetal growth, (7) no fetal anomaly according to ultrasonography, and (8) normal placental architecture and location according to ultrasonography. All prenatal care began in the early second trimester. Delivery at term was defined as birth between 37 and 42 weeks’ gestation. The human subjects committee of the Institutional Review Board at the University of California, Davis, approved this study. The subjects were instructed to collect ≥3 mL saliva in a provided clean vial. The collection was done on a
January 2001 Am J Obstet Gynecol
weekly basis at approximately the same time between 9 AM and 9 PM. The gestational age at which the sample was collected was calculated from a confirmed estimated date of delivery. The subjects were asked to store the samples frozen in the freezer of the household refrigerator. The samples were collected from the subject’s home and transported to the laboratory to be stored in a –20°C freezer. All samples were analyzed at our laboratory as described here. Assay reagents. Antiserum against the immunogen estriol-6-carboxymethyloxime/bovine serum albumin (Steraloids, Inc, Newport, RI) was produced in 4 male New Zealand White rabbits 3 to 5 months old by an immunization protocol similar to that described by Munro and Lasley.6 Each rabbit received 1.0 mL (1.0 mg/mL) of the emulsified immunogen at 0, 2, and 4 weeks, with a booster administered approximately every 3 to 4 weeks thereafter until no further increase in titer was obtained. All 4 rabbits produced antiserum to estriol-6-carboxymethyloxime/bovine serum albumin; however, 2 rabbits, R4834 and R4835, had the highest titers, and these antisera were purified for further testing. The gamma globulin fraction of the antiestriol antisera was purified by ammonium sulfate precipitation, and the nonspecific anti–bovine serum albumin antisera were removed by equivalence zone adsorption as described by Munro and Stabenfeldt.7 Both antisera were specific for estriol, with all steroids and steroid metabolites structurally related to estriol showing <0.1% cross-reactivity in the enzyme immunoassay. The antiserum R4835 was chosen for use in this study. Horseradish peroxidase was coupled to the same derivative (carboxymethyloxime) at 2 different sites (3 and 6 positions) on the estriol molecule by the carboxylic acid group on the carboxymethyloxime group by means of the mixed anhydride procedure.7 The estriol-3carboxymethyloxime/horseradish peroxidase enzyme conjugate yielded a much more sensitive standard doseresponse curve than did the estriol-6-carboxymethyloxime/horseradish peroxidase. This indicated that a heterologous site of conjugation provided the more sensitive configuration for the estriol enzyme immunoassay. All studies were therefore conducted with the estriol-3carboxymethyloxime/horseradish peroxidase enzyme conjugate. After elution of the conjugate by Sephadex G25 chromatography (Amersham Pharmacia Biotech, Inc, Piscataway, NJ) in a 0.05-mol/L phosphate buffer, pH 7.5, the steroid-enzyme conjugate was divided into aliquots and stored at –15°C. The competitive solid-phase microtiter plate enzyme immunoassay for the measurement of unbound, unconjugated estriol was similar in configuration to the enzyme immunoassay for progesterone described by Munro and Stabenfeldt.7 Flat-bottomed Immulon 1 microtiter plates (DYNEX Technologies, Inc, Chantilly, Va) were coated
Hedriana et al 125
Volume 184, Number 2 Am J Obstet Gynecol
with 50 µL per well of the purified antiserum R4835 at a dilution of 1:12,000 in 0.05-mol/L bicarbonate buffer, pH 9.6. The plates were sealed tightly with waterproof plate sealer covers and incubated overnight at 4°C or stored at 4°C until use. On the day of the assay the enzyme conjugate (estriol-3-carboxymethyloxime/horseradish peroxidase) was diluted to 1:80,000 in assay buffer (0.1-mol/L phosphate-buffered saline solution containing 0.1% bovine serum albumin, pH 7.8). Estriol standards were diluted in assay buffer. The range of the standard curve was 0.5 to 250 pg/well. Saliva samples were diluted 1:1 with assay buffer, mixed in a vortex appliance for 1 minute, and centrifuged at 2000g for 30 minutes to remove insoluble materials. To compensate for the effects of saliva in the estriol enzyme immunoassay, the pH of the normal steroid enzyme immunoassay buffer (7.0) used in all other steroid enzyme immunoassay systems was changed to a pH of 7.8. At this higher pH, saliva samples from nonpregnant and male subjects showed a true 0 value for concentration of estriol (the 0 concentration optical density and the optical density readings for the male or nonpregnant saliva samples were the same). Raising the pH to 7.8 in the estriol enzyme immunoassay eliminated the need to compensate for the blank values that occurred in the presence of saliva when the standard enzyme immunoassay buffer (pH 7.0) was used. Assay procedure. Before the assay unbound antibody was removed from the wells with wash solution (0.15mol/L saline solution containing 0.05% [vol/vol] polysorbate 20). Each microtiter plate was considered a separate assay, with 2 standard curves and internal control runs formatted within the same plate. The assay procedure was as follows: 30 µL of assay buffer was pipetted across the entire plate, followed by 20 µL of standard estriol (Sigma, St Louis, Mo), then by diluted saliva sample or internal control preparation. This step was immediately followed by the addition of 50 µL of the estriol-3carboxymethyloxime/horseradish peroxidase conjugate solution (1:80,000) to all plate wells. The plates were covered tightly with plate sealer covers, and the competitive reaction was allowed to proceed for 2 hours. To separate free from bound hormone the plates were emptied, rinsed 5 times with wash solution, and blotted dry. Freshly prepared substrate solution (100 µL of 0.05mol/L citrate, 1.6-mmol/L hydrogen peroxide, 0.4mmol/L 2,2´-azino-di-[3-ethylbenzthiazoline sulfonic acid] diammonium salt, pH 4.0) was then added to all wells, and color was allowed to develop to determine the amount of conjugate (estriol-3-carboxymethyloxime/ horseradish peroxidase) bound to the solid-phase antibody. The color change was stopped after approximately 60 minutes by the addition of 100 µL stop solution (0.15mol/L hydrofluoric acid containing 0.006-mol/L sodium hydroxide, and 0.001-mol/L ethylenediaminetetraacetic acid, pH 3.3). Absorbance was measured at 405 nm with
an automatic microtiter plate spectrophotometer (Dynatech MR600; DYNEX), and the data were transferred to an interfaced computer for analysis. Calibration curves were constructed with weighted least-squares linear regression according to the method of Rodbard and Lewald,8 and the raw data were reduced by log transformation to yield concentrations of estriol in nanograms per milliliter. Assay sensitivity. Assay sensitivity was determined as the least amount of hormone that could be distinguished from 0 concentration of standard defined by the mass at 2 SD above the mean of the 0 absorbance. Sensitivity was 0.25 ± 0.044 pg/well, with a 50% displacement at 11.7 ± 2.1 pg estriol and a linear response of the standard curve through 250 pg/well. Serial dilutions of selected saliva samples from pregnant women produced displacement curves parallel to that of the estriol standard doseresponse curve up to 5 halving dilutions (25-0.78 µL saliva per well) in assay buffer (pH 7.8). A recovery of 96.25% ± 5.49% was achieved from spiked saliva samples. The intraplate coefficient of variation for salivary estriol concentrations averaged 3.3% to 5.4%, whereas the average interplate (intra-assay) coefficient of variation was 5.3% to 12.5%. The average interassay coefficient of variation for salivary estriol concentration was 7.5% to 15.5%. Data analysis. Because the whole data set had a skewed distribution along the y axis, median values for salivary estriol concentration were calculated for each week’s measurement. The mean and SD for each week were also calculated for comparison and to demonstrate the dispersion of data around the mean. Regression techniques were applied to the data from the last 5 weeks before delivery to calculate weekly rates of salivary estriol production (slopes), which were then compared. This subset of data (slopes) was tested to determine the degree of deviation from the predicted normal distribution by means of Kolmogorov-Smirnov statistics. This test validated that the slope data met the assumptions to use the robust parametric 2-way analysis of variance with repeated measures test to compare the changes in salivary estriol production through the last 5 weeks before delivery. Results All women enrolled in the study were delivered of live infants between 37 and 41 weeks’ gestation. There were 12 nulliparous and 8 multiparous women. The mean (±SD) age was 28 ± 6 years, with a range of 21 to 38 years. Of these 20 women 16 were included for analysis of salivary estriol concentration after delivery and 4 were excluded because of the unexpected appearance of obstetric complications necessitating cesarean delivery for nonreassuring fetal well-being before the onset of labor. Of the 4 women excluded, severe preeclampsia complicated the pregnancies of 2 women, 1 had poorly controlled gestational diabetes mellitus, and 1 had cholesta-
126 Hedriana et al
January 2001 Am J Obstet Gynecol
Fig 1. Salivary estriol production increase was nonlinear, almost in stepwise fashion, starting 10 weeks before delivery. Solid line, Smoothed weekly median salivary estriol concentration. Surge is seen 5 weeks before delivery. Correlation coefficient was 0.85 (P < .0001). To demonstrate scatter of salivary estriol concentration values (open circles) at each gestational week, mean weekly salivary estriol concentrations before delivery (broken line with filled squares) were plotted for comparison. Error bars, SD.
sis of pregnancy. Of the 16 women included in the study, 9 had spontaneous vaginal deliveries and 7 were delivered abdominally for obstetric indications during the active phase of labor. Although 10 women had spontaneous labor, 6 women had labor induced at >40 weeks’ gestation as follows: 3 for oligohydramnios, 1 for significantly decreased perception of fetal movement, and 2 for nonreactive nonstress tests. These indications have not been shown to affect steroid metabolism2 and were therefore not anticipated to affect analysis of the salivary estriol concentration data. The salivary estriol enzyme immunoassay had a mean (±SD) sensitivity of 0.025 ± 0.001 ng/mL saliva. The baseline median salivary estriol production at 30 weeks’ gestation was 0.90 ng/mL, and the highest median at term was 2.70 ng/mL, a percentage increase of 201%. Increase in salivary estriol production was nonlinear, almost in a stepwise fashion, starting 10 weeks before delivery (Fig 1). Because the distribution of the raw data was skewed on the y axis, the profile for the smoothed median concentration for salivary estriol for each weekly interval was compared side by side with the mean (±SD) concentration for the same week (Fig 1). Both the mean and median profiles revealed the same general trend of increasing salivary estriol concentrations with each progressive week of gestation. Both profiles also revealed a positive inflection point between 4 and 6 weeks before delivery. According to the median weekly levels the salivary estriol rate increases were 50% (from 1.38 to 2.06 ng/mL) from week
35 to week 36 and another 30% (from 2.06 to 2.67 ng/mL) from week 36 to week 37, for a total 93% increase between weeks 35 and 37 (Fig 2). Because of the salivary estriol concentration surge from 35 weeks’ gestation, we wanted to know whether changes in salivary estriol production were consistent among women who were delivered at various gestational ages at term. The salivary estriol concentration data points were divided into 3 groups: early term (delivery at <38 weeks 1 day’s gestation), middle term (delivery at 38 weeks 1 day–40 weeks’ gestation), and late term (delivery at >40 weeks’ gestation). Because of the bias in distribution of data points and the small number of women in each group, the median was again used to characterize salivary estriol production. As can be seen in Fig 3, salivary estriol production rates were different between groups from 35 weeks’ gestation to delivery. Women who were delivered by 38 weeks’ gestation had a surge in salivary estriol concentration from 1.78 ng/mL at 35 weeks’ gestation to 4.51 ng/mL at 38 weeks’ gestation (increase of 154%). In the middle term group the salivary estriol concentration median increased by 152% from 1.38 ng/mL to 3.47 ng/mL at 40 weeks’ gestation. Those who were delivered at >40 weeks’ gestation had a salivary estriol concentration increase of 94% (2.68 ng/mL). To evaluate the variation in rates of salivary estriol production among these groups of term deliveries, the slopes generated between weekly measurement of salivary estriol concentrations were compared. Because of
Volume 184, Number 2 Am J Obstet Gynecol
Hedriana et al 127
Fig 2. Salivary estriol production across last 10 weeks of pregnancy displayed as median salivary estriol production at each week of gestation. Surge of increased salivary estriol production is apparent at 35 weeks’ gestation.
Fig 3. Salivary estriol production across last 10 weeks of pregnancy displayed as median salivary estriol production increases as function of gestational age at delivery. Open circles, Patients who were delivered at <39 weeks’ gestation (early term). Open squares, Patients who were delivered at >40 weeks’ gestation (late term). Open triangles, Patients who were delivered between 39 and 40 weeks’ gestation (middle term). Note differences in salivary estriol production median rate curves.
the surge seen in Fig 1 we chose to compare the 5 weeks of salivary estriol production rates before delivery. This subset of data met the assumption of normality for use of the parametric 2-way analysis of variance with repeated measures after validation that the sample did not deviate
from the predicted normal distribution within the limits of the Kolmogorov-Smirnov test. The assumptions for variance and independence between groups were likewise met for the 2-way analysis of variance test. The mean slopes of the 3 groups were significantly different
128 Hedriana et al
January 2001 Am J Obstet Gynecol
Table I. Weekly rates of salivary estriol production 5 weeks before delivery Time before delivery (wk)
Early term (n = 7)
Middle term (n = 3)
Late term (n = 6)
5 4 3 2 1
0.50 ± 0.13 0.56 ± 0.13 0.68 ± 0.15 0.69 ± 0.17 0.84 ± 0.26
0.32 ± 0.06 0.35 ± 0.13 0.41 ± 0.10 0.43 ± 0.19 0.37 ± 0.26
0.37 ± 0.03 0.30 ± 0.05 0.25 ± 0.06 0.00 ± 0.27 –0.03 ± 0.25
The weekly production rates (slopes) in nanograms per milliliter per week were calculated between weekly mean salivary estriol concentrations and are presented as mean ± SEM. There were significant differences in mean weekly production rates as the groups approached delivery (P < .05, 2-way analysis of variance with repeated measures).
through the last 5 weeks before delivery (Table I; P < .05 analysis of variance). The women who were delivered at early term had the greatest rates of salivary estriol production, followed by those in the middle term group and then by those women who were delivered late. In the last 5 weeks before delivery the rates of salivary estriol production were significantly greater in the early and middle term groups but differed significantly between groups, whereas in the late term group salivary estriol production remained unchanged during the last 3 weeks after an initial 2-week surge (Fig 3). The median salivary estriol concentration in both the early and middle term groups 3 weeks before delivery was 1.78 ng/mL, whereas that in the late group was 2.65 ng/mL. In contrast to the strong association of the rise of salivary estriol rates with the time of delivery, no strong association was found between the absolute concentrations of salivary estriol and impending delivery in 3 weeks. Although the small number of women in the study was a limitation, the number of data points was adequate for the purpose of describing the behavior of salivary estriol concentrations before delivery at term. Comment Increases in unbound, unconjugated estriol production at 30 to 42 weeks’ gestation were reliably measured by means of a direct nonradiometric enzyme immunoassay from patient-collected saliva samples, which allowed us the opportunity to study changes in estriol production though pregnancy. We have described the development and validation of a sensitive and rapid solid-phase microtiter plate enzyme immunoassay for salivary estriol concentration in pregnant women. In this study the distinctive patterns of increasing salivary estriol production were associated with the different lengths of pregnancy, observations that have not been previously reported. Despite advances in parturition research, few facts are known about the mechanisms responsible for the initiation of labor resulting in subsequent delivery. In studies
with sheep and monkeys9, 10 the onset of uterine contractions associated with labor seemed to depend on changes in the concentrations of estrogens and progesterone in the fetal and maternal circulations. However, extrapolations of these data remain problematic in explaining the series of events before the onset of human labor. Adding to the controversy, many studies of human labor at term have not found associations with consistent significant changes in both bound and unbound concentrations of estrogens and progesterone and in the estrogen/progesterone ratio.11-13 Until recently, human studies have measured these forms of steroid hormones in plasma. In separate studies Lachelin and McGarrigle3 and Vining et al14 showed that salivary estrogen and progesterone concentrations were reflective of concentrations of the free or unbound biologically active hormones in the plasma pool, in contrast to plasma and urine assays, which measured both the bound and unbound forms of these steroids.3, 14 Because these steroids varied in their percentage binding with protein in plasma, important fluctuations in free steroid activity were not detected. The unbound form of these hormones can filter into secreting glands, such as the salivary glands. By measuring salivary steroid hormone production across gestation,3 estriol production was noted to increase exponentially 5 to 6 weeks before term delivery.4 At this period in gestation salivary estriol concentration was found to be 15 times greater than that of estradiol in healthy pregnant women.15 Similar patterns were also observed at this period among patients who were delivered preterm (delivery at <37 weeks’ gestation).16 The estriol increase was parallel with but independent of the slow increase in progesterone production. These observations3, 4, 15, 16 suggested to us that the rate of rise of estriol production, rather than the absolute concentration, might be more predictive of impending labor and delivery. This concept became apparent when subjects’ weekly salivary estriol concentrations were traced side by side in this study. Our subjects who demonstrated an upsurge in salivary estriol production 4 to 5 weeks before delivery followed by a continued increase tended to be delivered earlier than the subjects who did not show a persistent increase in salivary estriol production. In addition, the median salivary estriol concentrations in women who were in labor early were lower than those in the subset of women who were delivered at >40 weeks’ gestation. In contrast to the strong association of the rise in salivary estriol production rate with the time of labor onset and delivery, no strong association was found between the absolute concentration of salivary estriol and impending labor and delivery. However, quite unexpectedly, the median salivary estriol concentration values plateaued at 37 weeks’ gestation in those subjects who had labor induced after 40 weeks’ gestation for common obstetric indications.
Hedriana et al 129
Volume 184, Number 2 Am J Obstet Gynecol
Together with salivary estriol production, estrone and estradiol values, including the estriol/progesterone ratio, increased 5 to 6 weeks before the onset of labor at term.4, 5, 15, 17 This enormous production of salivary estriol was consistent before preterm deliveries,5, 16 but neither estriol nor progesterone production increased significantly among women whose pregnancies were complicated by premature rupture of membranes at term and preterm who were not in labor.15, 16 The initiation of labor after rupture of the fetal membranes may not be estrogen driven but may result from another pathway initiated by disruption of the fetal membranes.5 In the same way increases in estriol and progesterone production were not seen among subjects who had continuation of the pregnancy to >40 weeks’ gestation.17 Although the fetal and placental hypothalamic-pituitary-adrenal axis has been implicated in pathways leading to labor in recent reports,18-20 the physiogenesis of this phenomenon has remained largely unexplained. In cases in which the continued increase of salivary estriol production with advancing gestation was not observed, the ability to produce estriol through placental regulation may be affected by decreased capacity to aromatize circulating fetal, adrenal-derived 16-hydroxylated substrates in aging placenta.2, 6, 21, 22 The limiting factor in this condition could be the inadequate supply of essential precursors from the fetal liver and fetal adrenals, much the same as was seen in the initiation of preterm labor in rhesus monkeys infused with extraneous androstenedione, an adrenal-derived placental precursor.23 In these monkeys the increase in serum androstenedione level paralleled the rise of serum dehydroepiandrosterone sulfate, which led to increased frequency of laborassociated uterine contractions. We speculate that in a normal pregnancy the availability of enormous amounts of placental estriol precursors may begin at 35 weeks’ gestation, as evidenced by the positive inflection point in this study. That supply of precursors for unknown reasons levels off, resulting in the delay of labor onset and delivery. No physical or biochemical changes in the placentas were demonstrated in our subjects who had labor induced at >40 weeks’ gestation. All 6 placentas appeared grossly normal, and all 4 of the placentas examined histopathologically were considered normal. The 6 neonates had normal liver and adrenal functions. In a series of studies with pregnant women at >40 weeks’ gestation Moran et al17 showed no demonstrable significant increases in salivary estriol concentration and in estriol/progesterone ratio after 40 weeks’ gestation. All their subjects had labor induced at 42 weeks’ gestation. In another study Moran et al24 administered estriol rectally to women with normal term pregnancies at 41 to 42 weeks’ gestation and reported an increase in plasma estriol level and an associated decrease in plasma progesterone level. The onset of spontaneous labor and sub-
sequent delivery was achieved in 5 of 17 subjects after 48 hours. Compared with the radioimmunoassay techniques reported earlier3, 4, 15, 16 the enzyme immunoassay used in this study has similar qualities but has the advantage of not requiring radioactive reagents. The enzyme immunoassay used and described here is not proprietary and is not the same as the SalEst system (Biex, Inc, Dublin, Calif).5, 25 The salivary estriol concentration enzyme immunoassay in this study has not been previously reported, which is why a relatively lengthy Methods section was required in this report. This salivary estriol concentration enzyme immunoassay has adequate specificity, sensitivity, and robustness to make it useful for research protocols in which processing a large quantity of samples and economy are high priorities. Recent studies5, 25 have shown the versatility of the clinical applications of a salivary estriol concentration enzyme immunoassay. When used as an aid to assess the likelihood of preterm labor and delivery, an increase in estriol production was demonstrated 3 to 4 weeks before delivery in a heterogeneous population from multiple medical centers.5 With an optimal salivary estriol concentration cutoff value of 2.3 ng/mL between 22 and 36 weeks’ gestation, McGregor et al5 correctly discriminated 73% of women who were subsequently delivered. If the test result was negative (salivary estriol concentration <2.3 ng/mL), there was a 95% likelihood that delivery would not occur within 2 to 3 weeks. When tested against the Creasy traditional risk factor assessment test, the salivary estriol concentration correctly predicted 87% of preterm births versus 7.2%.25 Despite these clinical applications, the positive prediction for preterm birth is low when a mean cutoff value is used. This may be due to the large variation in salivary estriol concentration values among patients at the same gestational age, as previously explained here and observed in other studies.3, 26 Despite the limitation of inadequate sample size in this study, the behavior of salivary estriol concentration production weeks before delivery that we have described is consistent with various studies that have used different assays.3-5, 15-17, 23-26 Although firm conclusions are beyond the scope of this study, improvement in prediction of preterm births could possibly be obtained by determining the rate of salivary estriol concentration production across gestational week intervals after correct identification of the baseline levels. We are indebted to Jiangang Chen for his assistance in statistical analysis. REFERENCES
1. Liggins GC, Fairclough RG, Grieves SA, Kendall JZ, Knox BX. The mechanism of initiation of parturition in the ewe. Recent Prog Horm Res 1973;29:111-59. 2. Word RA. Parturition. In: Carr BR, Blackwell RE, editors. Textbook of reproductive medicine. East Norwalk (CT): Appleton and Lange; 1993. p. 41-8.
130 Hedriana et al
3. Lachelin GC, McGarrigle HH. A comparison of saliva, plasma unconjugated and plasma total oestriol levels throughout normal pregnancy. Br J Obstet Gynaecol 1984;91:1203-9. 4. Darne J, McGarrigle HH, Lachelin GC. Saliva oestriol, oestradiol, oestrone and progesterone levels in pregnancy: spontaneous labour at term is preceded by a rise in salivary oestriol: progesterone ratio. Br J Obstet Gynaecol 1987;94:227-35. 5. McGregor JA, Jackson GM, Lachelin GC, Goodwin TM, Artal R, Hastings C, et al. Salivary estriol as risk assessment for preterm labor: a prospective trial. Am J Obstet Gynecol 1995;173: 1337-42. 6. Munro CJ, Lasley B. Non-radiometric methods for immunoassay of steroid hormones. In: Albertson BD, Haseltine FP, editors. Non-radiometric assays: technology and application in polypeptide and steroid hormone detection. New York: Alan R Liss; 1988. p. 289-329. 7. Munro C, Stabenfeldt G. Development of a microtitre plate enzyme immunoassay for the determination of progesterone. J Endocrinol 1984;101:41-9. 8. Rodbard D, Lewald JE. Computer analysis of radioligand assay and radioimmunoassay data. Acta Endocrinol Suppl (Copenh) 1970;147:79-103. 9. Csapo AI. Progesterone “block.” Am J Anat 1956;98:273-91. 10. Challis JR, Olson DM. Parturition. In: Knobil E, Neill J, editors. The physiology of reproduction. New York: Raven Press; 1988. p. 2177-216. 11. Cousins LM, Hobel CJ, Chang RJ, Okada DM, Marshall JR. Serum progesterone and estradiol-17β levels in premature and term labor. Am J Obstet Gynecol 1977;127:612-5. 12. Thornburg GD, Challis JR. Control of parturition. Physiol Rev 1979;59:863-918. 13. Willcox DL, Yovich JL, McColm SC, Phillips JM. Progesterone, cortisol and oestradiol-17 beta in the initiation of human parturition: partitioning between free and bound hormone in plasma. Br J Obstet Gynaecol 1985;92:65-71. 14. Vining RF, McGinley R, Rice BV. Saliva estriol measurements: an alternative to the assay of serum unconjugated estriol assessing feto-placental function. J Clin Endocrinol Metab 1983;56: 454-60.
January 2001 Am J Obstet Gynecol
15. McGarrigle HH, Lachelin GC. Increasing saliva (free) oestriol to progesterone ratio in late pregnancy: a role for oestriol in initiating spontaneous labour in man? BMJ 1984;289:457-9. 16. Darne J, McGarrigle HH, Lachelin GC. Increased saliva oestriol to progesterone ratio before idiopathic preterm delivery: a possible predictor of preterm labour? BMJ 1987;294:270-2. 17. Moran DJ, McGarrigle HH, Lachelin GC. Lack of normal increase in saliva estriol/progesterone ratio in women with labor induced at 42 weeks’ gestation. Am J Obstet Gynecol 1992; 167:1563-4. 18. Chrousos GP. Reproductive placental corticotropin-releasing hormone and its clinical implications. Am J Obstet Gynecol 1999;180(1 Pt 3):S249-50. 19. Hobel CJ, Dunkel-Schetter C, Roesch SC, Castro LC, Arora CP. Maternal plasma corticotropin-releasing hormone associated with stress at 20 weeks’ gestation in pregnancies ending in preterm delivery. Am J Obstet Gynecol 1999;180(1 Pt 3):S257-63. 20. Lockwood CJ. Stress-associated preterm delivery: the role of corticotropin-releasing hormone. Am J Obstet Gynecol 1988;180(1 Pt 3):S264-6. 21. Nathanielsz PW. The role of basic science in preventing low birth weight. Future Child 1995;5:57-70. 22. Olson DM, Mijovic JE, Sadowsky DW. Control of human parturition. Semin Perinatol 1995;19:52-63. 23. Mecenas CA, Giussani DA, Owiny JR, Jenkins SL, Wu WX, Honnebier BO, et al. Production of premature delivery in pregnant rhesus monkeys by androstenedione infusion. Nat Med 1996;2:443-8. 24. Moran DJ, McGarrigle HH, Lachelin GC. Maternal plasma progesterone levels fall after rectal administration of estriol. J Clin Endocrinol Metab 1994;78:70-2. 25. Heine RP, McGregor JA, Dullien VK. Accuracy of salivary estriol testing compared to traditional risk factor assessment in predicting preterm birth. Am J Obstet Gynecol 1999;180(1 Pt 3): S214-8. 26. Darne J, McGarrigle HH, Lachelin GC. Diurnal variation of plasma and saliva oestrogen, progesterone, cortisol and plasma dehydroepiandrosterone sulfate in late pregnancy. Eur J Obstet Gynecol Reprod Biol 1989;32:57-66.