Introduction
Preterm birth (PTB), defined as birth before 37 + 0 weeks of gestation, occurs in approximately 5 to 13% of all pregnancies worldwide [
1,
2]. It is the leading cause of neonatal death in most industrial countries and determinates a high rate of short and long-term morbidity [
3‐
5]. About 70% of all PTBs result from spontaneous premature labor with or without prior rupture of membranes [
6]. Despite its importance, the prediction of spontaneous PTB (sPTB) is a challenge. Prediction by clinical symptoms or measurement of the cervical ripening by digital examination is imprecise because of low specificity [
7,
8]. There are efforts to improve predictive accuracy by additional measurement of biomarkers from cervicovaginal fluid, which are released during parturition [
9]. Of these biomarkers fetal fibronectin (fFN), phosphorylated insulin-like growth factor binding protein-1 (pIGFBP-1) and placental alpha microglobulin-1 (PAMG-1) have been widely studied and are commercially available as point of care tests (POCT) [
10‐
15]. However, PAMG-1 with the best test characteristics of these biomarkers [
16], did not met the expectations in the routine care use [
17].
Cytokine-induced inflammatory response appears to play a significant role in the initiation of parturition in term as well preterm birth [
18‐
21] and are therefore promising candidates as predictive biomarkers [
22]. Chemokines like CC-chemokine-ligand 2 (CCL2) and CXC-chemokine-ligand 10 (CXCL10) are widely expressed, small secretory proteins of 8–10 kDa which are involved in the regulation of parturition [
23]. CCL2, also known as Monocyte Chemoattractant Protein-1 (MCP-1), belongs to the group of CC-chemokines and is secreted by numerous cell types including activated monocytes, endothelial cells, fibroblasts or smooth muscle cells [
24]. CCL2 is in particular induced by IL-6 or TNF-alpha. The corresponding receptor CCR2 is mainly expressed on monocytes/macrophages, natural killer (NK) cells and vascular smooth muscle cells. Secretion of CCL2 results mainly in attracting monocytes/macrophages and, in a lesser extent, of T and B lymphocytes as well as NK cells [
24]. CXCL10, also named interferon-gamma induced protein 10 (IP-10), belongs to the group of CXC-chemokines and is produced by neutrophils, monocytes, T leucocytes, fibroblasts, endothelial cells and keratinocytes under the influence of INF-y and participates in the activation of peripheral immune cells, especially stimulated CD4
+-T lymphocytes, monocytes and NK cells [
25,
26]. Previous studies have been showing increased levels of both chemokines at birth in the maternal cervix, the amniotic fluid, the myometrium as well as the placenta and membranes [
27,
28]. Decidual stromal cells strongly express numerous chemokines and its receptors including CCL2 and CXCL10, which are involved in the accumulation of immune cells in the decidua and may be important in the regulation of immune tolerance at the fetal maternal interface [
25,
29]. Altered chemokine expression can be observed in term and preterm births [
23,
25,
29].
CRP and the leukocyte count are diagnostic parameters of inflammation, which are determined in routine clinical laboratories. Both, CRP [
30‐
32] as well as the leucocyte count [
33‐
35] appear to be useful in predicting parturition at different stages of pregnancy.
Moreover, there has been found an association between maternal serpin B7 levels and early spontaneous preterm birth, which we wanted to integrate in our study [
36].
The identification of predictive biomarkers in the maternal serum instead of cervicovaginal fluid shows several potential advantages, as it is an easy-to-perform and cost-effective diagnostic procedure.
The aim of this study was to analyze, whether CXCL10, CCL2 and Serpin B7 in maternal serum of symptomatic woman could predict sPTB in women at risk. The results were compared to the predictive properties of CRP and leucocytes.
Material and methods
Study participants
The prospective cohort study was conducted at a tertiary care center (Rostock, Germany) between January 2017 and April 2019 with recruitment of 109 participants.
Women were eligible for inclusion if they were pregnant between 20 + 0 to 31 + 6 weeks of gestation and complained at least one of the following risk factors of PTB: regular (< 3/30 min) or painful contractions, a cervical length < 25 mm measured by transvaginal ultrasound or a history of previous preterm birth. Also, a study inclusion was possible solely based on a personal history of PTB all study participants showed at least one further criterion of threatened PTB. Criterions of exclusion were a cervical dilatation > 3 cm, maternal temperature above 37.5 °C, premature rupture of membranes or vaginal bleeding at admission. We also excluded patients who had undergone tocolysis within the last seven days, who prior received a cervical cerclage or cerclage pessary, who required an iatrogenic delivery within the next seven days and patients with known hypertensive pregnancy disorders. Further clinical treatment was based on physician’s choice in accordance to the national guideline recommendations and depending on the clinical situation. Primary study outcome was spontaneous PTB < 34 weeks, secondary outcome was PTB < 37 weeks or delivery within seven days after study inclusion. Gestational age was calculated from the first day of the last menstrual cycle and was corrected based on ultrasound findings if measurements of the crown-rump length in the first trimester differed by more than seven days.
Sample collection and processing
Peripheral venous blood samples were obtained upon admission using a serum collection tube (7.5 mL, Sarstedt, Nümbrecht, Germany). The blood clot was immediately separated by centrifugation at 2000 × g for 15 min at 15 °C and serum was stored in aliquots at -80 degrees until use. The serum concentrations of the biomarkers were determined using sandwich immunoassays: Human CXCL10/IP-10 ELISA Kit PicoKine® (Boster Biological Technology, Pleasantan, CA, USA, Catalog #EK0735), Human MCP-1/CCL2 ELISA Kit PicoKine® (Boster Biological Technology, Pleasanton CA, USA, Catalog #EK0441) and Human serpin peptidase inhibitor, clade B (ovalbumin), member 7 (SERPINB7) ELISA Kit (MyBioSource, San Diego, CA, USA, #MBS2610991) with pre-coated monoclonal antibodies. Undiluted samples were measured in duplicates following the manufacturer´s instructions (Table S
1). Serum concentrations below the range specified were defined as negative (0).
CRP was automatically measured as part of the clinical diagnostic by latex-enhanced nephelometric immunoassays (CardioPhase® hsCRP) performed on a Siemens BN ProSpec® System (Siemens Healthcare Diagnostics GmbH, Eschborn, Germany).
Statistical analysis
All data were stored and statistically analyzed by using the IBM SPSS statistical package 27 (SPSS Inc., Chicago, IL, USA), Excel 2016 (Microsoft Corporation, Redmond, WA, USA) and the statistical software R 4.2.2 with R-studio (2022.12.0) and the packages ggplot2 version 3.4.0 and plotRoc version 2.3.0. Testing for differences in continuous variables between groups was done using Studentʼs t‑test, Mann–Whitney U-test or Kruskal–Wallis test as appropriate; comparisons of categorical variables between groups was done with Fisherʼs exact test. For description of the relationship between investigated biomarkers and the remaining gestation period a correlation analysis according to Pearson with specification of the correlation coefficient (r) was performed. Survival analysis (time to delivery) was carried out with the Kaplan–Meier method, with statistical comparison of groups done using log-rank test. The impact of multiple gestations on CCL2, CXCL10 and the CXCL10/CCL2-ratio depending on gestational age at delivery was computed with two-factorial ANOVA. All p-values were obtained using two-sided statistical tests, and values < 0.05 were considered statistically significant. Receiver operating characteristics (ROC) curves and the area under the curve (AUC) were computed, and the optimal cut-off value (minimal distance to sensitivity and specificity of 1) was calculated using the following equation: (1-sensitivity)2 + (1-specificity)2. To calculate the diagnostic criteria, a second diagnostic cutoff with a fixed specificity of 0.9 was selected. A logistic regression model was used to assess the independence of specific outcome parameters. We used a sequential method with the addition of variables in order of magnitude of the crude odds ratios (ORs) and starting with the largest estimate. In the adjusted model for the CXCL10/CCL2-ratio, the covariates “multiple gestation” and “body mass index” were included. The covariates “maternal age” and “gestational age at study inclusion” did not improve the predictive model and were excluded. Due to the importance of the cervical length for estimating the risk of preterm birth, this parameter was also integrated into the regression model.
We performed a pre-study sample size calculation with a logistic regression model using the statistical G*Power 3.1.9.2 software. The estimated a priori probability for delivery < 34 weeks was set to 15%, which resulted in a calculated sample size of 139 patients for an assumed post-test probability of 30%, an α-error of 0.05, and a power of 90%. Study recruitment was limited a priori to a period of two years. At this time point, the sample size was 109 patients, which corresponds to a statistical power of > 99% in a two-tailed post-hoc power analysis.
Ethical approval
The local ethics committee of the University of Rostock (IRB-No. A2016-0162) approved the protocol of the study. The study was also registered within the German Clinical Trials Register (DRKS‑ID: DRKS00010763). Written informed consent was obtained from all participants.
The study was conducted as part of the “Analysis of diagnostic accuracy of predictive biomarkers in risk assessment of threatening preterm birth” (ADAPROB) study, parts of which have already been published [
37,
38].
Discussion
There are a myriad of different biomarkers that have been analyzed for the prediction of sPTB in the past decades with contradictory results [
39‐
41]. In particular, it is difficult to predict sPTB over a period of more than 14 days. In accordance to the present guidelines, biomarkers already established on the market (PAMG-1, fFN, and phIGFPB-1) are useful for the short term prediction of an imminent PTB within 7 days [
3]. Even here, in a real life situation in women at high risk, these biomarkers did not reveal a satisfactory prediction of a subsequent preterm birth before 34 weeks of gestation [
17,
42]. In addition, due to the local release in the area of the maternal fetal interface, a determination in the cervicovaginal fluid is required [
43‐
48].
The initiation of parturition is characterized by an inflammatory process, which is accompanied by the release of numerous chemokines in the surrounding tissue as well as the circulation [
18,
49]. Chemokines are involved in the regulation of the inflammatory response and can be measured in maternal serum samples [
49]. In our study the chemokines CXCL10 and to a lesser extend CCL2 showed predictive properties in relation to prematurity. Various studies examined the connection between CXCL10 and preterm birth. Levels of CXCL10 were increased in cases of preterm birth in the amnion fluid [
50,
51], the cervicovaginal fluid [
52], the fetal membranes as well as the placenta [
53]. The increase of CXCL10 at the maternal fetal interface is not specific for PTB and higher levels were detected in cases of preeclampsia [
54], chorioamnionitis [
55] and villous inflammation [
56]. However, all of these can be regarded as a manifestation of chronic chorioamnionitis, which may define a common pathological lesion among preterm labor [
53,
56]. Along a concentration gradient, CXCL10 acts as a chemoattractant mainly for maternal T lymphocytes, which is thought to mediate the breakdown of the maternal–fetal homeostasis with anti-fetal rejection and labor induction [
57]. This fits with the observation that the concentrations measured in the amniotic fluid were 5 to 10-fold higher compared to our serum levels [
50,
51]. CXCL10 expression in the fetal membranes is activated in particular by interleukin 27 through multiple signalling pathways [
58]. Elevated plasma levels of CXCL10 in PTB have also previously been observed [
59]. Interestingly, CXCL10 concentrations in amniotic fluid from women undergoing genetic midtrimester amniocentesis were higher in women with sPTB after 32 weeks [
51]. However, in asymptomatic women blood samples at 16–20 weeks of gestation did not reveal differences of the CXCL10 level between women with and without subsequent sPTB [
60]. These results suggest that while CXCL10 rises early in symptomatic women with subsequent PTB, it is not predictive of asymptomatic women. Some study results where contradictory compared to ours. In women with premature prelabor rupture of membranes (PPROM) CXCL10 plasma level at admission were significantly lower compared to uncomplicated pregnancies [
61]. Another study did not found an association between the CXCL10 serum concentrations and the occurrence of PTB in women with preterm labor [
62].
CCL2, which is strongly expressed in the stromal and immune cells of the decidua, seem to play an important role in mediating the fetal immune tolerance via attracting maternal macrophages and CD4
+ T lymphocytes [
29]. A particular strong increase of expression can be found in the myometrium in both term and preterm labor [
63]. In cases of infection-related PTB an increase in CCL2 expression was observed in uterine smooth muscles cells, in the placenta and amniotic fluid [
29,
64,
65]. In line with this, CCL2 was most strongly expressed in the fetal membranes of PPROM compared to rupture of membranes at term with the highest concentrations in the area of the rupture zone [
66]. In patients with PPROM, plasma concentrations of CCL2 were higher at delivery compared to admission, but in contrast to CXCL10 did not differ to unaffected controls at admission [
61]. In contrast, an increase of CCL2 in the amniotic fluid preceded a subsequent PTB in women with shortened cervix [
67]. The inverse correlation in CCL2 with decreased serum levels in women with subsequent preterm birth in our study cannot be explained so far. Studies in rats suggest that this effect could be due to a redistribution of CCL2-secreting monocytes into the reproductive tissue [
68,
69]. Endocervical smears collected between 12 and 25 weeks of gestation from women with a history of sPTB revealed an absence of cervical macrophages and low levels of CCL2 in women with subsequent sPTB below 34 weeks [
70]. This immune phenotype revealed better predictive performance than the history of early sPTB in combination with a short cervix before 18 weeks.
An increase of CCL2 in the cervicovaginal fluid was in combination with cervical length, interferon-gamma and interleukin-6 predictive for PTB within seven days in symptomatic women with preterm labor [
71].
In this study, the increase of CXCL10 in maternal serum was strongly predictive for subsequent PTB, especially below 34 weeks of gestation. This increase appears to be an early event of threatened PTB and apparently preceded the onset of delivery for more than seven days. Serum levels of CCL2 decreases shortly before delivery and allowed prediction of PTB within seven days, but works worse at a longer interval. Combining both parameters further improved the predictive properties with an AUC of 0.83 (95% CI 0.73–0.93) with a sensitivity of 0.67 and a PPV of 0.43 for delivery < 34 weeks of gestation. The results remained statistically significant even after logistic regression and excluding multiple pregnancies. For delivery within 7 days, the CXCL10/CCL2-ratio even showed an AUC of 0.87 (95% CI: 0.7–1.0) with a sensitivity of 0.83 and a PPV of 0.25. This is worse than the results of a meta-analysis with PAMG-1 and PPVs ranging between 0.34 in the low-risk-group and 0.83 in the high-risk-group [
16]. However, data on symptomatic women with threatened PTB did not get such convincing results with detection rates of only 50% and a PPV of 23% for prediction of PTB within seven days in singleton pregnancies [
17]. Interestingly, simple performance of a leucocyte count revealed comparably good predictive properties within seven days. Interestingly, in our study cohort the cervical length was worse predictive as the inflammatory markers (AUC 0.68 [0.54–0.83],
p = 0.021) [
38].
It is a strength of our prospective study that used criterions of inclusion corresponds to a real-life situation. According to the study protocol, blood samples for analysis of putative biomarkers were taken before antenatal steroid prophylaxis was applied, which exclude an influence on our results. Steroid prophylaxis was indicated solely on the basis of clinical assessment. The inclusion of multiple pregnancies could be a limitation, but did not influence the results. Using serum values to calculate the CXCL10/CCL2-ratio requires accurate determination of the concentrations by immunoassays. Establishing it in a routine care setting could be a cost-effective and minimally time-consuming laboratory procedure.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.