The Interplay of the Interleukin 1 System in Pregnancy and Labor

Abstract

This work assessed the temporal coexpression of interleukin 1 (IL-1) and its inhibitor, IL-1 receptor antagonist (IL-1ra), in the cervicovaginal fluid (CVF) beyond 24 weeks gestation including women in spontaneous term labor. Two cohorts of women were recruited at 24 to 35 weeks’ gestation (n = 65) and in late pregnancy (>36 weeks’ gestation; n = 88). The CVF was serially collected either every 4 weeks between 24 and 35 weeks’ gestation (n = 123 samples) or weekly during late pregnancy (n = 240 samples). The IL-1 and IL-1ra were quantitated by enzyme-linked immunosorbent assay, and the effect of vaginal microflora and unprotected sexual intercourse were also investigated. The IL-1β and IL-1ra remain unaltered between 24 and 35 weeks’ gestation. At late pregnancy, IL-1α and β concentrations peak at 4 to 14 days prior to labor onset, while IL-1ra decreases with approaching spontaneous term labor (P < .05, 2-way analysis of variance). The IL-1 and IL-1ra were significantly correlated (P < .001, Pearson r). A combined biomarker model of IL-1α, IL-1β, and IL-1ra can predict term labor with 86% sensitivity and 92% specificity. This study indicates a shifting inflammatory balance in the gestational tissues prior to labor onset.

Introduction

The cervicovaginal fluid (CVF) proteome reflects the local biochemical milieu and is influenced by the physical changes occurring in the cervix and adjacent overlying fetal membranes.^1,2 The current gold standard test to predict preterm birth (PTB) is the detection of fetal fibronectin (fFN > 50 ng/mL) in the CVF with low sensitivity and a high negative predictive value (NPV).³ As preterm delivery requires medical intervention, an alternative biomarker to predict PTB with improved sensitivity and a high positive predictive value (PPV) is warranted. The development of a PTB diagnostic test faces many challenges, including the difficulty in obtaining a sufficient number of prospective samples from women who subsequently deliver preterm and to discover and validate biomarkers.

As preterm and term labor share a final common pathway of cervical ripening, myometrial activation, and membrane rupture leading to birth, it is hypothesized that biomarkers predictive of term labor could potentially be utilized in a preterm setting. In addition, it is worth considering the benefits of predicting postterm labor as postterm pregnancy is associated with considerable health risks to the fetus and mother.⁴ The ability to predict postterm labor may permit the tailoring of obstetric management to reduce costs and minimize perinatal and maternal morbidity associated with unnecessary induction of labor or cesarean section delivery.^5,6

Using 2-dimensional polyacrylamide gel electrophoresis, we discovered that the naturally occurring inhibitor of interleukin 1 (IL-1), IL-1 receptor antagonist (IL-1ra), was significantly decreased in the human CVF in association with spontaneous term labor.⁷ Our IL-1ra finding was subsequently validated using enzyme-linked immunosorbent assay (ELISA).⁸ It is widely acknowledged that human labor is an acute inflammatory process. Elevated levels of maternal circulating leukocytes,⁹ influx of leukocytes, and expression of proinflammatory cytokines, including IL-1, in the cervix, myometrium, and fetal membranes before or during parturition have been reported.^10–13 The IL-1ra inhibits the biologic responses induced by IL-1 by competing for cell surface receptors (IL-1 receptor types I and II) without activating target cells.¹⁴

The IL-1 and IL-1ra concentrations in the CVF have been investigated by others to determine their utility as predictors of preterm^15,16 and/or term labor.^16–19 Most of these previous studies quantified only IL-1β^16–18; while 2 other studies measured IL-1α and IL-1β¹⁹ or IL-1β and IL-1ra.¹⁵ This current study aims to provide a comprehensive screen of IL-1 and IL-1ra in serially collected CVF beyond 24 weeks’ gestation. Therefore, the specific aims of our study are to investigate (1) the quantitative temporal expression IL-1β and IL-1ra between 24 and 35 weeks of gestation; (2) the quantitative temporal expression of IL-1α, IL-1β, and IL-1ra in late pregnancy (>36 weeks’ gestation) until spontaneous term labor; and (3) the utility of the IL-1 system to predict spontaneous term labor.

Materials and Experimental Design

Patient Recruitment

This longitudinal observational study of CVF in pregnancy was approved by the Mercy Health, Research Ethics Committee. All participants provided written informed consent. Pregnant women who attended the Antenatal Clinic at the Mercy Hospital for Women (Heidelberg, Victoria, Australia) were prospectively recruited by a research midwife beyond 24 weeks gestation. Detailed inclusion and exclusion criteria are described in the following sections. After obtaining the participants’ delivery outcomes, women were retrospectively selected, focusing on those who experienced spontaneous onset of labor. Two experimental groups were investigated and for convenience are referred to as cohort A and cohort B. Temporal changes in the concentration of IL-1β and IL-1ra in the CVF between 24 and 35 weeks’ gestation were investigated in samples from cohort A, while temporal changes in IL-1α, IL-1β, and IL-1ra in the CVF during late pregnancy (>36 weeks) until spontaneous onset of term labor were investigated in samples from cohort B (Table 1). Women were questioned at each CVF sampling as to whether they had unprotected sexual intercourse in the preceding 48 hours, an internal ultrasound within the preceding 6 hours, experienced any recent vaginal bleeding, or were taking any medication.

Table 1.

Demographic and Obstetric Characteristics of Women in Cohorts A and B.

	Cohort A (24-35 Weeks’ Gestation)	Cohort B (≥36 Weeks’ Gestation)
Women (n)	65	88
Maternal age, years (mean ± SD)	31.9 ± 4.1	32.0 ± 5.2
Gravidity (mean ± SD)	3.7 ± 2.0	3.1 ± 1.3
Parity (mean ± SD)	1.1 ± 1.4	1.5 ± 0.8
Gestational age at delivery, weeks (mean ± SD)	39.5 ± 1.2	40.1 ± 0.9
Fetal birth weight, g (mean ± SD)	3436.5 ± 505.2	3626.5 ± 422.5
Fetal gender, n (%)
Male	27 (42)	50 (57)
Female	38 (58)	38 (43)

Abbreviation: SD, standard deviation.

Cohort A Participants

Women in cohort A were recruited at approximately 24 weeks gestation. Serial CVF swabs were obtained every 4 weeks where possible until 35 completed weeks of gestation. Inclusion criteria were healthy women with a singleton pregnancy who delivered spontaneously between 37 and 42 weeks’ gestation. Exclusion criteria were multifetal gestation, PTB (delivery <37 weeks’ gestation), prelabor rupture of fetal membranes (PROMs), elective cesarean section delivery, induction of labor, bacterial vaginosis, cervical cerclage, and progesterone pessary treatment. Samples were not collected if the women had a digital vaginal examination or transvaginal ultrasound within the preceding 6 hours. Two women reported an isolated incidence of mild transient vaginal bleeding. Neither woman was actively bleeding at the time of CVF sampling.

Cohort B Participants

Cohort B consisted of women recruited at 36 weeks’ gestation. Serial CVF swabs were obtained weekly starting from 36 weeks’ gestation and, when possible, an in-labor swab was obtained before the rupture of fetal membranes. Labor was defined as regular painful uterine contractions that lead to the effacement and dilation of the cervix (>3 cm). Inclusion criteria were healthy women with a singleton pregnancy and spontaneous onset of labor at 37 to 42 weeks’ gestation. Exclusion criteria were multifetal gestation, <37 or >42 weeks’ gestation at delivery, PROM, elective cesarean section, induction of labor, and bacterial vaginosis. Likewise, CVF was not collected if the women had a digital vaginal examination or transvaginal ultrasound within the preceding 6 hours. One woman reported an incident of mild transient vaginal bleeding but was not actively bleeding at the time of CVF sampling. Two women were taking antibiotics for mild respiratory tract infections.

Sample Collection

The CVF was collected adhering to our previously published protocol.^{1,7,8,20–22} Briefly, CVF was collected using a sterile speculum and a DUO-TRANSTUBE double-tipped rayon swab (Medical Wire & Equipment Co Ltd, Corsham, Wilts, England). The swab was placed into the posterior vaginal fornix for 30 s and immersed in 1 mL of chilled “extraction” buffer (50 mmol/L 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 150 mmol/L NaCl, 0.1% sodium dodecyl sulfate, 1 mmol/L EDTA, and 1 mmol/L Pefabloc SC 4-(2-aminoethyl)benzene sulfonyl fluoride; Roche Diagnostics GmbH, Mannheim, Germany). The tubes were left on ice and transported to the laboratory for processing within 2 hours. Samples were centrifuged, and the supernatant fluid was collected and stored at −80°C.

Vaginal Microflora

Upper vaginal microbiology assessment and culture were performed by the clinical microbiology laboratories at St Vincent’s Hospital and Austin Pathology, Austin Health (Melbourne, Victoria, Australia). The presence of group B Streptococcus (GBS), Candida spp, and Ureaplasma spp in the vaginal tract are generally not considered as pathogens; their detection is termed as “colonization.” Microbiology reports were divided into 5 groups, no significant pathogens, GBS colonization, Candida spp colonization, Ureaplasma spp Colonization, and mixed colonization (consisting of 2 or more of these groups). Microbiologic results were matched to the IL-1α, IL-1β, and IL-1ra concentrations of the CVF sample collected on the same day of testing.

For cohort A, microbiologic analysis was generally performed at every sampling. A total of 91 microbiology results were obtained from the 65 women recruited. One woman was treated with clindamycin for abnormal vaginal discharge, and 9 women were prescribed erythromycin for Ureaplasma spp colonization. For cohort B, microbiologic assessment was performed only once, usually on the day of recruitment (n = 59 women). One woman allocated into the mixed colonization group had Mycoplasma sp and Candida spp One woman diagnosed with candidiasis was treated with clotrimazole pessaries.

Quantitation of IL-1 and IL-1ra

The CVF samples in cohort B were quantitated prior to the completion of sample collection in cohort A. The ELISA kit used to analyze samples from cohort B was discontinued by the manufacturer; therefore, a different ELISA kit was used for sample analysis for cohort A. For cohort A, IL-1β was quantitated using the IL-1β/IL-1F2 DuoSet ELISA (R&D Systems, Minneapolis, Minnesota). The sensitivity, interassay, and intraassay coefficient of variations (CVs) were 3.91 pg/mL, 6.5%, and 6.2%, respectively. Samples below the detectable IL-1β assay range were assigned as 0. The IL-1ra was quantitated using the IL-1ra/IL-1F3 DuoSet ELISA (R&D Systems). The sensitivity, interassay, and intraassay CVs for the IL-1ra assay were 3.6 pg/mL, 6.0%, and 5.6%, respectively.

For cohort B, IL-1α was quantitated using the IL-1α CytoSet ELISA (BioSource International, Camarillo, California). The sensitivity, interassay, and intraassay CVs for the IL-1α assay were 2.49 pg/mL, 3.5%, and 5.7%, respectively. The IL-1β was quantitated using the IL-1β singleplex express assay (Bio-Rad, Hercules, California). The linear range of the assay was 3.2 to 3261 pg/mL; the sensitivity of the assay was 0.6 pg/mL, and the interassay and intraassay CVs were 8% and 6%, respectively. The IL-1ra was quantitated using the IL-1ra CytoSet ELISA (BioSource International). The sensitivity of the assay was 4.0 pg/mL; interassay and intraassay CVs were 5.2% and 4.1%, respectively. Samples below the detectable range were assigned as 0.

Data presented in the results section were corrected for total protein and expressed as per milligram of protein. The total protein of all CVF samples was determined using the Bicinchoninic acid protein assay (Pierce, Rockford, Illinois).

Statistical Analyses

Statistical analyses were performed using SPSS (v17.0; SPSS Inc, Chicago, Illinois). Data for each cohort were assessed for homogeneity of variance and were not normally distributed (P < .05, Kolmogorov-Smirnov test). In order to utilize the parametric 2-way analysis of variance (ANOVA), data were transformed to normality. The IL-1α and IL-1β were cube root transformed, while IL-1ra was square root transformed. All statistical comparisons between the groups were performed using 2-way ANOVA with post hoc testing using Tukey honestly significant difference (HSD), where appropriate. To account for multiple sampling from the same woman, each patient was included as a random factor in all statistical models. The presence of specific vaginal microflora and the potential influence of unprotected sexual intercourse on the concentrations of IL-1 or IL-1ra were investigated with “gestational age” or “days from labor” included as a covariate factor in statistical analyses.

Temporal changes in the concentration of IL-1α, IL-1β and IL-1ra were analyzed between the groups (cohort A: 24-25, 26-27, 28-29, 30-31, 32-33, and 34-35 weeks; cohort B: in-labor, 0-3, 4-14, and >14 days before labor). Median values were used to calculate the fold changes. The box and whiskers plots were produced using GraphPad Prism v5.02 (GraphPad Software, San Diego, California); the boxes represent the median and interquartile range, while the whiskers represent the 5th and 95th centiles.

Linear regression analysis (2-way ANOVA) was used to examine the linear relationship of IL-1 and IL-1ra with advancing gestational age or sampling-to-labor intervals. Scatterplots were graphed using square- or cube root-transformed values and GraphPad Prism v5.02.

Correlation was performed using Pearson r. Receiver–operator characteristic (ROC) curves were plotted for IL-1α, IL-1β and IL-1ra, and the threshold values were determined for each variable to evaluate their utility as individual predictors of labor. Binary logistic regression (patients included as a categorical variable for multiple sampling) provided the sensitivity, specificity, PPV, and NPV of a 3-cytokine model to predict labor onset. Statistical significance was assumed when P < .05.

Results

Cohort A

Each woman contributed 1 to 4 CVF samples ranging from 16 to 117 days prior to spontaneous labor onset. As the samples were collected remote from labor onset, they were grouped according to the gestational age as follows: 24 to 25 (22 women, n = 22 samples), 26 to 27 (18 women, n = 18), 28 to 29 (27 women, n = 27), 30 to 31 (22 women, n = 22), 32 to 33 (25 women, n = 25), and 34 to 35 (9 women, n = 9) weeks’ gestation.

In cohort A, IL-1β concentrations were significantly different between women with no pathogens isolated, colonized by GBS, Candida spp, Ureaplasma spp, or mixed colonization (Table 2; P = .036). The IL-1β concentrations associated with Candida spp colonization were significantly higher than no pathology (6.1-fold, P < .001), GBS (7.3-fold, P = .008) and Ureaplasma spp colonization (14.3-fold, P < .001, Tukey HSD). The IL-1ra concentration were not affected by the different pathogens (Table 2; P = .157).

Table 2.

Microbiology Results for Cohorts A and B.^a

	No Pathogens Isolated	Group B Streptococcus Colonization	Candida spp Colonization	Ureaplasma spp Colonization	Mixed Colonization	Two-Way ANOVA P Value
Cohort A (total of 91 samples from 65 women)
Samples (n)	57	13	6	13	2
Women (n)	47	11	4	11	2
IL-1β median, pg/mg protein	52.21	185.80	1615.09	0.00	227.82	.036b
IL-1ra median, μg/mg protein	2.97	7.73	2.46	1.97	1.78	.157
Cohort B (total of 59 samples from 59 women)
Samples (n)	29	4	7	10	9
Women (n)	29	4	7	10	9
IL-1α median, pg/mg protein	145.32	232.43	0.00	219.95	207.27	.243
IL-1β median, pg/mg protein	132.11	397.42	65.47	567.81	334.05	.227
IL-1ra median, μg/mg protein	0.92	2.14	0.60	1.18	0.99	.178

Abbreviations: ANOVA, analysis of variance; HSD, honestly significant difference; IL-1, interleukin 1; IL-1ra, IL-1 receptor antagonist.

^a For cohort A, microbiologic assessment was performed at each sampling, where possible. For cohort B, microbiologic assessment was performed once at 36 weeks’ gestation.

^b Tukey HSD post hoc tests are reported in Results section.

There was no significant difference in IL-1β and IL-1ra concentrations in samples collected from women who reported having unprotected sexual intercourse in the preceding 48 hours (n = 14) versus those that did not (n = 55; IL-1β P = .576, IL-1ra P = .767).

The IL-1β and IL-1ra were assayed in 123 samples collected from 65 women. There was no linear trend for IL-1β with advancing gestational age (P = .234, Figure 1B) nor between the gestational age groups (P = .799, Figure 1A). The IL-1ra displayed a weak significant increasing linear trend with advancing gestational age (P = .032, Figure 1D) but was not different between the groups (P = .145, Figure 1C). The IL-1β and IL-1ra were significantly correlated (P < .001, Pearson r = .460).

Figure 1.

In cohort A (65 women, n = 123), IL-1β displayed no differential expression between the groups (A; P = .799, 2-way ANOVA) nor linear trend (B; P = .234, 2-way ANOVA) with advancing gestational age from 24 to 35 weeks. There was no differential expression in IL-1ra concentration between the groups (C; P = .145, 2-way ANOVA) despite a mild increasing linear trend for IL-1ra was observed (D; P = .032, 2-way ANOVA). ANOVA indicates analysis of variance; IL-1β, interleukin 1β; IL-1ra, IL-1 receptor antagonist.

Cohort B

Each woman contributed 1 to 6 CVF samples. Samples collected ranged from in-labor to 47 days prior to spontaneous labor onset and were stratified into 4 groups: in-labor (20-25 women, n = 20-25 samples), 0 to 3 days (17-22 women, n = 17-22), 4 to 14 days (52-58 women, n = 74-83), and >14 days (58-66 women, n = 98-110) before spontaneous labor.

The IL-1α, IL-1β, and IL-1ra concentrations were not different between women with no pathogens isolated, colonized by GBS, Candida spp, Ureaplasma spp, or with mixed colonization (Table 2; IL-1α P = .243, IL-1β P = .227, and IL-1ra P = .178).

There was no difference in IL-1α, IL-1β, and IL-1ra concentrations in samples collected from women who reported having unprotected sexual intercourse in the preceding 48 hours (n = 26-29) versus those that did not (n = 122-129; IL-1α P = .408, IL-1β P = .802, and IL-1ra P = .443).

A total of 216 IL-1α measurements were performed on 84 women. The IL-1α displayed no linear trend with approaching labor (P = .378, Figure 2B). However, the IL-1α concentration in the CVF was significantly different between the groups (P = .001). The IL-1α concentration at 4 to 14 days was significantly higher (↑1.45-fold) than 0 to 3 days before labor onset (P = .024, Tukey HSD; Figure 2A).

Figure 2.

IL-1α, IL-1β, and IL-1ra concentrations obtained from cohort B during late pregnancy. A, IL-1α at 4 to 14 days was 1.45-fold significantly higher than 0 to 3 days before labor but did not display any linear trend with approaching labor onset (B; 84 women, n = 215, P = .378, 2-way ANOVA). C, IL-1β at 4 to 14 days was 4.55- and 1.91-fold significantly higher than both 0 to 3 days and >14 days but did not display any linear trend with approaching labor onset (D; 81 women, n = 211, P = .145, 2-way ANOVA). E, IL-1ra was significantly decreased in labor and 0 to 3 days before labor onset compared with 4 to 14 days (3.0- and 1.6-fold, respectively) and >14 days prior to labor (3.7- and 2.0-fold, respectively). IL-1ra demonstrated a significant linear decreasing trend toward labor (F; 88 women, n = 240, P = .005, 2-way ANOVA). P values in (A), (C), and (E) represent Tukey HSD post hoc tests. ANOVA indicates analysis of variance; HSD, honestly significant difference; IL-1, interleukin 1; IL-1ra, IL-1 receptor antagonist.

The IL-1β was assayed in 211 samples collected from 81 women. There was no linear trend with approaching labor (P = .145, Figure 2D). The IL-1β concentration in the CVF was significantly different between the groups (P = .050). The IL-1β concentration at 4 to 14 days was significantly higher than 0 to 3 days (↑4.55-fold, P = .012) and >14 days (↑1.91-fold, P = .005, Tukey HSD) prior to labor onset (Figure 2C).

A subset of IL-1ra data presented in this study has been previously published.⁸ We corrected the IL-1ra data for total protein in this current study and have included additional samples. In total, IL-1ra was assayed in 240 samples collected from 88 women. There was a significant linear decreasing trend toward labor onset (P = .005, Figure 2F). The IL-1ra concentrations were significantly different between the groups (P < .001). The IL-1ra was significantly decreased in labor compared with 4 to 14 days (↓2.96-fold, P < .001) and >14 days (↓3.66-fold, P < .001, Tukey HSD) prior to labor onset. The IL-1ra concentration at 0 to 3 days before labor onset was also significantly lower than at 4 to 14 days (↓1.62-fold, P = .023) and >14 days (↓2.01-fold, P = .004, Tukey HSD; Figure 2E). All correlations were significant (P < .001, Pearson r): IL-1α and IL-1β (r = .546, n = 204); IL-1α and IL-1ra (r = .491, n = 216); and IL-1β and IL-1ra (r = .543, n = 211).

Predicting Term Labor

Individual ROC curves IL-1α, IL-1β, and IL-1ra as well as the combination of all 3 cytokines (using predicted probabilities derived from binary logistic regression) were generated to predict spontaneous onset of term labor within 3 days of sampling (Figure 3). A total of 204 samples from 80 women in cohort B screened for all 3 cytokines were used in this part of the analysis. The combined cytokine model yielded 86% sensitivity, 92% specificity, 69% PPV, and 97% NPV to predict labor (Table 3).

Figure 3.

Receiver–operator characteristic curves for IL-1α (AUC = 0.649), IL-1β (AUC = 0.533), and IL-1ra (AUC = 0.693), and the combination of all the 3 biomarkers (AUC = 0.963, 80 women, n = 204) to predict the onset of term labor within 3 days. AUC indicates area under the curve; IL-1, interleukin 1; IL-1ra, IL-1 receptor antagonist.

Table 3.

Efficiency of IL-1α, IL-1β, and IL-1ra and the Combination of All the 3 Biomarkers to Predict Spontaneous Term Labor Within 3 days of Onset in 80 Women From Cohort B (Total n = 204 Samples; Samples Ranged From In-Labor and up to 47 Days Prior to Spontaneous Labor Onset).

	IL-1α	IL-1β	IL-1ra	IL-1α, IL-1β, and IL-1ra
Concentration thresholds	95 pg/mg protein	120 pg/mg protein	0.37 μg/mg protein	–
Probability cutoff value	–	–	–	0.7
Area under the curve	0.649	0.533	0.693	0.963
95% confidence interval	0.551-0.747	0.428-0.637	0.598-0.788	0.938-0.989
P value	.005	.540	<.001	<.001
Sensitivity, %	52.8	52.8	52.8	86.1
Specificity, %	77.4	66.1	76.2	91.7
Positive predictive value, %	33.3	25.0	32.2	68.9
Negative predictive value, %	88.4	86.7	88.3	96.9
False positive rate, %	22.6	33.9	23.8	8.3
False negative rate, %	47.2	47.2	47.2	13.9

Abbreviations: IL-1, interleukin 1; IL-1ra, IL-1 receptor antagonist.

Discussion

This study has examined the interplay of the IL-1 system beyond 24 weeks’ gestation in the human CVF. From 24 to 35 weeks’ gestation, IL-1β and IL-1ra appear to maintain a relatively consistent baseline expression. The IL-1ra may play a role in the dampening of the proinflammatory response of IL-1 during this period of pregnancy to maintain uterine quiescence. It should be emphasized that the bioavailability of IL-1ra is several thousand times greater than the combined bioavailability of IL-1α and IL-1β. The IL-1α and IL-1β concentrations remained unaltered in late pregnancy, while IL-1ra displayed a significant linear decrease with approaching labor. The significant correlations between these 3 cytokines provide evidence of their well-established biologic interdependence.

A subset of the IL-1ra data obtained from women in late pregnancy and included in the current study has been published previously.⁸ In that publication, IL-1ra was similarly found to significantly decrease with approaching spontaneous labor and demonstrated a similar 3- to 4-fold decrease in labor. Data from the previous publication were not corrected for total protein. In our experience, the total yield of CVF protein can vary widely between women. To accurately quantify the measurement of any biomarker, we have normalized the data by correcting for total protein. We found that correcting the IL-1ra data for total protein in the current study did not differ from our initial IL-1ra interpretation in late pregnancy.⁸

Numerous studies have investigated the biologic effects of IL-1 during parturition and its increase in serum, gestational tissues, and amniotic fluid.^23–29 Similarly, previous studies of the CVF have described impending parturition and labor onset to be associated with an increase in IL-1β concentration.^16–19 We failed to observe an increase in IL-1β in the CVF in our study. This conflicting IL-1β result may partly explain why IL-1β is yet to be translated into clinical use to predict spontaneous labor or the successful induction of labor. In contrast, our IL-1α data are consistent with Imseis et al,¹⁹ where they also reported no change in CVF IL-1α with labor onset.

The constant level of IL-1 and the decrease in IL-1ra with impending term labor may augment the actions of IL-1 in the myometrium, cervix, and fetal membranes as the bioavailability of IL-1ra is reduced.^26,30–33 We previously reported that IL-1ra appears to be in lower concentrations in women with PROM and regular contractions occurring ≥1 hour after compared with women with spontaneous onset of labor with intact membranes.⁸ In addition, we recently found that IL-1ra was significantly decreased by 40% in the CVF of women who subsequently experienced preterm PROM compared with gestation-matched controls.³⁴ Future studies elucidating the direct role of IL-1ra in fetal membrane physiology could be of interest.

The complex physiology of parturition precludes the likely opportunity of finding a single biomarker that reliably predicts human labor. A multiple biomarker diagnostic test with improved sensitivity and specificity would therefore be more reliable and efficacious in predicting labor.^20,35,36 In addition, the final pathway of cervical remodeling, myometrial activation, and membrane rupture common to term and preterm labor suggests the potential translation of term labor biomarker(s) to predict preterm labor. Compared with each individual cytokine, the combination of all the 3 cytokines to predict term labor improved the area under the curve and predictive values. For term pregnancies where labor is expected, a highly specific test with a high NPV may be a useful indicator of “failure to labor,” which would invariably require medical intervention.

The fFN predicts term labor within 2 to 7 days of sampling with a wide range of sensitivity, specificity, PPV, and NPV,^5,17,37–41 while IL-1β predicts term labor within 7 days of sampling with 76% sensitivity, 55% specificity, 50% PPV, and 79% NPV.¹⁷ Although the predictive properties of IL-1 and IL-1ra in this current study (either individually or in combination) cannot be directly compared with these previous fFN and IL-1β studies due to different sample-to-delivery time frames and study populations, we believe the combination of the 3 biomarkers (Il-1α, IL-1β, and IL-1ra) to predict spontaneous labor compares favorably to fFN with improved PPV.

An ideal diagnostic test should be robust to the high phenotypic variation between women caused by genetic, environmental, health condition, and lifestyle factors. Moreover, with regard to human CVF, any diagnostic test should also be robust to the influence of vaginal microflora, semen, and vaginal bleeding. The wide range in upper vaginal microflora did not influence CVF’s IL-1 and IL-1ra concentrations at term. The microflora between 24 and 35 weeks gestation did not affect IL-1ra concentration, whereas IL-1β was significantly increased when Candida spp was detected. These data must be interpreted with caution due to small sample size. It is also possible that erythromycin treatment may have influenced IL-1 or IL-1ra CVF concentrations associated with Ureaplasma spp. colonization between 24 and 35 weeks. Unprotected sexual intercourse did not influence CVF IL-1 and IL-1ra concentrations in this and other studies,^42,43 but it is known to influence fFN testing.⁴⁴

In conclusion, this study describes the differential expression of IL-1α, IL-1β, and IL-1ra beyond 24 weeks’ gestation and until spontaneous term labor onset. Although IL-1α and IL-1β showed little or no change, IL-1ra concentration remained fairly constant throughout gestation followed by a rapid decline in the final days before spontaneous labor. These changes may be indicative of a shifting inflammatory balance in the gestational tissues prior to labor onset. In the future, it may also be beneficial to include fFN and/or insulin growth factor-binding protein 1⁴⁵ in a multiple marker model to predict labor.