Association of placental inflammation with fetomaternal hemorrhage and loss of placental mucin-1
- First Online:
- Cite this article as:
- Scholz, C., Hermann, C., Kachler, A. et al. Arch Gynecol Obstet (2012) 285: 605. doi:10.1007/s00404-011-2028-1
Fetomaternal hemorrhage (FMH) poses an immediate risk to the fetus and, in case of Rhesus-immunization, to future pregnancies. Given that altered endothelial permeability is part of the pathophysiology of inflammation, in this study we investigated whether placental inflammatory processes like chorioamnionitis (ChoA) or preeclampsia (PE) lead to increased rates of FMH compared to the established risk factor of placenta previa (PP). Putative accompanying markers of trophoblastic damage were also explored.
40 patients (14 PE; 6 ChoA; 9 PP; 11 normal controls) were evaluated for FMH using a flowcytometric test kit, which is able to quantify FMH of 0.06% fetal cells. Placental tissue samples were immunostained for human placental lactogen (hPL), human chorionic gonadotropin (hCG), and mucin-1 (MUC1). MUC1 was evaluated as a potential serum marker of FMH.
Patients with ChoA had a mean calculated FMH volume of 29 ml, compared to 4 ml in PE and 1 ml in PP and controls. MUC1 staining was reduced in PE and ChoA placenta samples, while elevated MUC1 serum concentration correlated positively with FMH.
Diseases of placental inflammation are associated with FMH. Placental MUC1 staining is reduced and serum concentrations are increased in cases of FMH.
KeywordsFetomaternal hemorrhage Chorioamnionitis Preeclampsia Inflammation MUC1 Ca27.29
In 1.1 out of 1000 deliveries, a loss of more than 20 ml of fetal blood volume into maternal circulation occurs and defines fetomaternal hemorrhage (FMH), which is associated with an unfavorable long-term neonatal outcome . FMH can result in immediate fetal compromise and poses a risk of Rhesus (Rh)-immunization for Rh-negative mothers . Antenatal risk factors for FMH include cases of placental manipulation or trauma (cesarean section, blunt abdominal trauma) as well as placenta previa (PP) . However, roughly half of FMH cases occur in the absence of any known risk factors . Therefore, numerous case reports describe unsuspected cases of FMH among patients outside of the defined at-risk population [5, 6]. Inflammatory reactions are an integral part of the pathologies of chorioamnionitis (ChoA) and preeclampsia (PE), and are known to be associated with placental damage . Whether these inflammatory changes translate into an increased vascular leakage and might contribute towards the etiology of FMH remains an open question.
In the present study we asked, whether there is a link between inflammation and FMH. First, we tested this hypothesis by testing patients with above mentioned placental pathologies for FMH. In the present study, the amount of fetal erythrocytes in maternal peripheral blood among patients with PP was compared with that of patients with inflammatory conditions namely, PE and ChoA. To quantify FMH, a flow-cytometric test kit was used that proved to reliably quantify fetal cells down to a minimal concentration of 0.06%.
Second, in the search of a morphological basis of FMH, we investigated whether alterations in trophoblast-markers (human placental lactogen (hPL), human chorionic gonadotropin (hCG) or mucin-1 (MUC1)), were associated with placental damage. More knowledge of associations between these markers and the development of FMH, might aid in developing a set of predictive markers for the future. Finally, in a proof of concept approach, we tested serum markers that might accompany FMH in the context of inflammation.
Two abundantly secreted trophoblastic proteins, hPL and hCG, have been shown to be differentially regulated in villous trophoblasts of preeclamptic patients and pregnancies complicated by intrauterine growth restriction (IUGR) . While hPL has a regulatory effect on syncytialization and differentiation, hCG additionally promotes trophoblast invasion, and seems to be regulated by peroxisome proliferator-activated receptors (PPAR) until parturition [9, 10]. MUC1 is a multifunctional glycoconjugate also expressed by human trophoblasts . Under the name Ca27.29 it is evaluated as a bone-related biomarker for breast cancer . MUC1 is part of the family of membrane-bound mucins that consist of a single membrane-spanning domain and a highly conserved cytoplasmatic tail. These membrane-bound mucins sense the extracellular environment and take part in cellular signaling, proliferation, apoptosis, adhesion, and invasion . While MUC1 is detectable in trophoblastic tissue even in early gestation, and seems to have a central role in blastocyst attachment and implantation, its placental expression increases during the course of pregnancy and it is highly expressed throughout the third trimester .
Materials and methods
Blood samples and placental tissues were obtained between 2006 and 2008 from 40 women delivering at the Department of Obstetrics and Gynecology—Innenstadt of Ludwig-Maximilians University, Munich. In total, 9 patients with PP, 14 patients with severe PE, 6 patients with ChoA, and 11 normal pregnancies were included in the study. Prior to delivery or any surgical procedure, 5 ml of maternal total venous blood was collected into ethylenediaminetetraacetate (EDTA) anticoagulant, as well as 8 ml into the appropriate collection tubes (Sarstedt, Nümbrecht, Germany) for serum samples. Serum samples were spun down at 400×g and stored at −80°C until final testing. EDTA blood was stored at room temperature on a shaker and processed within 72 h. Placental samples were taken from three central parts of the placenta, laid out in an isosceles triangle, following the principles of systematic uniform random sampling laid out by Mayhew . The obtained tissue samples were fixed immediately after delivery in 4% buffered formalin for 20–24 h and then embedded in paraffin.
Mothers hospitalized for PP (mean gestational age at delivery: 34 + 3 weeks; Standard Deviation (SD) 4.3)) had blood drawn prior to cesarean section. Ultrasound scanning of one of these patients revealed additional placenta percreta and placental tissue was left in situ after delivery; the patient’s uterine arteries were then coiled via a femoral artery catheter by interventional radiologists. No placental tissue for histological evaluation could be obtained from this patient. PP patients had a median of 3 episodes of vaginal bleeding. Six patients were hospitalized for ChoA (30 + 2 weeks; SD 2.8). Diagnosis was based on clinical criteria as well as cellular and serological markers of inflammation. Women defined as suffering from ChoA had fever >38° C and at least two of the following conditions: leukocytosis (>12 G/µl), C-reactive protein (CRP) >0.50 mg/dl, maternal tachycardia >100 bpm, fetal tachycardia >160 bpm, uterine tenderness or foul odor of amniotic fluid . Women were not in labor at the time of the blood test. After delivery, diagnosis was confirmed with standard histological assessment performed by a gynecological pathologist. Thirteen mothers with severe PE were included in the study (33 + 5 weeks; SD 1.9). Inclusion criteria were set to maternal hypertension (BP >140/90 mmHg) without previous hypertensive history and/or proteinuria >5 g/24 h without any prior history of renal disease. Women delivering spontaneously after an uneventful pregnancy were recruited as normal controls (38 + 5; SD 1.4). The ethics committee of the faculty of the Ludwig-Maximilians-University, Munich, Germany approved the study, and written consent was obtained from each patient.
Assessment of fetomaternal hemorrhage
Detection and quantification of fetal erythrocytes was performed with the Fetal Cell Count Kit II (IQ-Products, Groningen, The Netherlands) according to manufacturer’s instructions . This flow-cytometric test kit distinguishes between maternal and fetal erythrocytes by staining for fetal hemoglobin (HbF) and carbonic anhydrase (CA). While there is a significant proportion of maternal cells that physiologically contain HbF, CA is an enzyme of the respiratory chain and is expressed only postnatally. Therefore, this test distinguishes between maternal HbF-containing cells (HbF+/CA+) and fetal HbF-containing cells (HbF+/CA−) and there is no need to determine threshold levels for physiologic HbF .
Technically, red blood cells (RBC) were washed three times in phosphate-buffered saline (PAA-Laboratories, Pasching, Austria). RBC were fixed at room temperature for 30 min in a formaldehyde containing solution, washed once more, and then permeabilized in sodium dodecyl sulfate solution for 4 min at room temperature. The fixed and permeabilized RBC were stained with fluorescein isothiocyanate-labeled monoclonal mouse anti-human HbF antibody (recognizing the HbF alpha-chain) and a phycoerythin-labeled polyclonal rabbit anti-human CA antibody (recognizing the CA-II isoform). Prior to flow cytometry, optimal detector amplifications and compensations were determined using unstained or single-stained controls as well as cord blood samples. Data acquisition was performed on a FACSCalibur flow cytometer (Becton–Dickinson, Heidelberg, Germany). For quality control, spiking experiments were performed using a mixture of peripheral blood of nulliparous non-pregnant individuals and cord blood in various dilutions. In serial dilutions, the peripheral blood of non-pregnant adults was mixed with different numbers of freshly isolated fetal cord blood cells or commercially available pre-spiked fixed RBC.
Concomitant test controls to validate and monitor the quality of flow-cytometric procedures were performed using FetalTrol test samples (IQ-Products, Groningen, The Netherlands). For the calculation of transfusion volume, we used the following formula:Volume of FMH = (Maternal blood volume × maternal hematocrit × % fetal RBC)/(Newborn hematocrit) .
Paraffin-embedded specimens (2–3 µm) were mounted on SuperFrost/Plus microscope slides (Menzel, Walldorf, Germany). After deparaffinization and rehydration, the slides of different placental tissue specimens were incubated in methanol/H2O2 for 30 min in order to inhibit endogenous peroxidase activity; this was followed by a 5 min wash in phosphate-buffered saline (PBS, pH 7.4) and treatment with rabbit or horse serum (20 min, 22°C, Vector laboratories, Burlingame, USA) to reduce non-specific background staining. Tissues were incubated with the primary antibodies, polyclonal rabbit anti-human chorionic gonadotropin (hCG) (A-0231; dilution 1:400; Dako, Hamburg, Germany), polyclonal rabbit anti-human placental lactogen (hPL) (PU040-UP; 1:75; BioGenex, Hamburg, Germany) and two monoclonal mouse anti-human MUC1 antibodies (PankoMab; 1:1000; Glycotope, Berlin, Germany and VU-4H5; 1:500; Invitrogen, Karlsruhe, Germany) for 1 h at room temperature. PankoMab detects a proprietary conformational MUC1 target, whereas VU-4H5 detects a triple tandem repeat on the MUC1 peptide. Sections were incubated with the biotinylated secondary antibody (1 h, room temperature, Vector laboratories, Burlingame, USA) and then avidin-biotinylated peroxidase (45 min, room temperature, Vector laboratories, Burlingame, USA). Between each step, sections were washed with PBS. The peroxidase staining reaction was performed with diaminobenzidine/H2O2 (1 mg/ml; 5 min, Vector laboratories, Burlingame, USA) and halted in tap water (10 min). Sections were counterstained in hemalaun (1 min, Vector laboratories, Burlingame, USA) and then coverslipped. In negative controls, the primary antibody was replaced with pre-immune mouse serum or rabbit serum IgG at the same concentrations as the primary antibodies. For positive controls we used MUC1 positive breast cancer specimen. The intensity and distribution patterns of the staining reaction were evaluated using an adaptation of the semi-quantitative immunoreactivity score (IRS) as previously described . IRS was determined by three independent examiners. The IRS was calculated by multiplication of staining intensity (graded as 0 = non-, 1 = weak-, 2 = moderate- and 3 = strong-staining) with the percentage of positive stained cells (0 = no staining, 1 = <10% of the cells, 2 = 11–50% of the cells, 3 = 51–80% of the cells and 4 = >81% of the cells), without knowledge of the pathological evaluation or the diagnosis of the specimen. Formation of a glycocalix was not incorporated into the semi-quantitative evaluation, as it emerged as an additional parameter only during the process of histological evaluation. As MUC1 staining can be confounded by gestational age, we controlled all immunhistochemical stainings with healthy gestational age-matched placental specimens.
Detection of MUC1
MUC1 (also known as Ca27.29) was assessed on a fully automated immunoassay analyzer (AIA-600 II, Tosoh Bioscience, Tessenderlo, Belgium) using the specific ST AIA-PACK Ca27.29 according to the manufacturer’s instructions. None of the specimens was subject to more than one freeze-thaw cycle.
Cumulative data are presented as mean and standard error of mean (SEM) for each group. The results were evaluated using the non-parametric Mann–Whitney U Test for comparison and assessment of significant differences of the means (p < 0.05), indicated by asterisk in figures. The Positive Predictive Value (PPV) was calculated as the number of True Positives (TP) divided by the sum of TP and False Positives (FP). The Negative Predictive Value (NPV) was calculated as the number of True Negatives (TN) divided by the sum of TN and False Negatives (FN). Sensitivity equaled TP divided by (TP + FN) and Specificity was calculated using TN divided by (FP + TN).
Pregnancy pathologies associated with inflammation are also associated with FHM
Inflamed placental tissue show an alteration in MUC1 expression profile
Only serum level of MUC1—and not leukocyte count or CRP—correlates to FMH
The leukocyte count (L) (normal value = 4.0–11.0 G/l) and concentration of C-reactive protein (CRP) (normal value <0.50 mg/dl) were assessed for all patients. Neither of these markers reliably predicted the occurrence or extent of FMH. The positive predictive values (PPV) were calculated and retrieved for CRP: 0.17 and for L: 0.25. Sensitivity for these two tests with regard to occurrence of FMH was 0.75 for both CRP and L. Specificity was determined to be 0.38 for CRP and 0.59 for L. The absence of elevated CRP values predicted an absence of FMH, with a negative predictive value (NPV) of 0.88; this was improved to 0.92 when normal leukocyte count was considered as a negative prediction marker for FMH.
In the present study, inflammation that is associated with placental damage was found to be a risk factor for FMH, whereas the established risk factor of PP did not retrieve elevated FMH levels when compared with normal controls. PE had a minor yet significant effect on FMH volumes. In immunohistochemical analysis, placentas corresponding to inflammatory processes such as ChoA or PE had significantly reduced trophoblastic staining of MUC1, yet remained positive for hPL and hCG. Serum MUC1 levels correlated with the amount of FMH and were better predictors of FMH than leukocyte count or concentration of C-reactive protein.
So far, conflicting results for the sensitivity and specificity of the traditional Kleihauer–Betke test (KBT) in comparison with flow cytometry (FACS) in detecting HbF positive cells, have been published [22, 23, 24]. Based on a review of the current data, skillful FACS seems to be equivalent to skilful KBT at detecting HbF+ cells in maternal peripheral blood. Only recently, however, a two-color approach using HbF and CA to discriminate between fetal and adult RBC has been introduced and evaluated . In previous FACS studies, HbF has been used as the sole marker for fetal RBC. This approach may lead to an overestimation of the number of RBC in both FACS as well as KBT analyses, as there are varying numbers of persistent adult HbF+ cells present in maternal peripheral blood .
In our study, we used a commercially available test kit for HbF/CA two-color flow cytometry, which screened 2 × 106 maternal peripheral venous RBCs. Given the availability of a flow cytometer, this approach allows for a timely and stable quantification of FMH. In preliminary experiments, a linearity of results from 0.03 to 1.00 percent fetal RBC was demonstrated after performing serial dilutions. This corresponded with published experiments that have evaluated the test kit in a total of 579 samples . Fetal RBC are fairly stable within the maternal circulation and are detectable for up to 119 days . Given these findings, the test applied in this study allowed for a very sensitive and specific quantification of the fetomaternal transfusion of RBC that had taken place during the previous 3–4 months.
A higher incidence of cellular components of fetal origin has been described for preeclamptic patients in many of the studies [27, 28]. As early as 1893, Georg Schmorl described trophoblastic pulmonary emboli in patients that had died of PE, implying that fetal cells deported into the maternal circulation might participate in the etiology of this disease . In search for predictive markers, the detection of fetal cellular content within the maternal circulation of PE patients was mainly based on detecting fetal (male) DNA. The number of male fetal cells found in patients with PE was up to 30 times greater than that found in controls . While these molecular markers indicate a significant disturbance in cellular trafficking in PE patients, only one study has been undertaken to evaluate RBC volumes in the maternal circulation of PE patients . In a matched pair analysis of 31 PE patients, Hsu et al.  were unable to find significant differences between PE pregnancies and matched normal controls using a regular Kleihauer–Bethke test. Given the rather subtle yet significant increase in FMH in our study population, and the different test method with regular screening of 2 × 106 cells with two markers, these findings may in fact simply reflect the increased sensitivity of the test method, rather than differing biological behavior. While there is solid data supporting the contribution of increased apoptotic breakdown and reduced invasiveness of trophoblastic cells towards the etiology of PE, this seems to translate into rather subtle changes in permeability for cellular components, such as RBCs.
Massive inflammation of the fetomaternal interface is the pathologic basis of ChoA. We asked whether cellular trafficking was altered in the context of these inflammatory processes. The volume of HbF+/CA− fetal cells in maternal circulation was significantly increased in the six patients of our study population. Fortunately, ChoA is a rare event; in most of the instances, delivery is pursued before overt ChoA can occur. Nevertheless, in our small sample size, the amount of fetal cells found in maternal peripheral blood was surprising. The mean FMH volume was calculated at 29 ml, and thus was greater than the threshold of 20 ml that has been shown to define a poorer long-term neonatal prognosis . The reduced trophoblastic expression of MUC1 might be seen in the context of the breakdown of placental defensive barriers .
In our study population, we found uniformly strong staining of hCG and hPL in all placental tissues, independent of an underlying disease. In line with our findings, Newhouse et al.  found a significant increase of syncytialization and hCG and hPL expression only for IUGR pregnancies, whereas PE patients were found to have a non-significantly reduced expression of these markers.
While all our samples are from third-trimester placenta, some effect due to the lower gestational week (30 + 2) of ChoA samples cannot be ruled out. It does not explain, however, the virtual absence of MUC1 in those placentas. The functional role of MUC1 in the third-trimester placenta remains unclear. We found a strong MUC1 expression in normal third-trimester placental tissue, with strong intracellular staining and an intact apical glycocalix throughout the villous trophoblast. In PE placental tissue, however, this expression was reduced intracellularly and areas of missing glycocalix were present. These findings were even more pronounced in the cases of ChoA. Among patients with >20 ml of FMH, intracellular staining of placental tissue for MUC1 was weak and an apical glycocalix was missing with the exception of small islands.
We are perfectly aware that our sample size is as yet too small to establish MUC1 as a predictive marker of FMH. Nevertheless, this association seems to us an intriguing finding that warrants further investigation. When interpreting these descriptive findings, one must contextualize MUC1 within the inflammatory response. MUC1 is a member of the larger family of o-linked glycans that are taken up by antigen-presenting cells, control antigen processing, and influence subsequent immune responses . Most importantly, MUC1 has been recently shown to be a negative regulator of pro-inflammatory toll-like receptor (TLR) signaling . TLR are the most important pattern recognition receptors of the innate immune system and are expressed by trophoblastic cells [34, 35]. We speculate, that trophoblast cells reduce intracellular and extracellular MUC1-expression, as part of the inflammatory process. This leaves their altered apical trophoblastic surface exposed to maternal immunocompetent cells. Further studies are needed to evaluate trophoblastic behavior in the face of immune stimulation.
Finally, we tested the hypothesis that increased MUC1 cleavage might be detectable in the peripheral blood of affected patients. Indeed, by testing the sera of our study population with an automated MUC1 (Ca27.29) immunoenzymatic assay, MUC1 concentrations were significantly increased in those patients that had a concurrent FMH of >20 ml, implying MUC1 shedding in those patients. To our knowledge, this is the first time that a correlation between serological concentration of MUC1 and FMH volume could be demonstrated. MUC1 testing was as sensitive and had a roughly equal NPV as compared to tests for standard markers of inflammation. Specificity (0.85) and PPV (0.50), however, were markedly improved with regards to FMH. Due to the small study population, these findings are not sufficient to establish increased MUC1 levels as predictive markers of ensuing FMH. Nevertheless, our data support MUC1 as a key player in the fetomaternal interface that is directly linked to trophoblast behavior, and may, therefore, serve as a serological marker of trophoblastic damage.
The authors are very grateful to Prof. S. Endres and his team at the Division of Clinical Pharmacology, University of Munich for their generous assistance with flow-cytometric experiments.
Conflict of interest