Uncomplicated oocyte donation pregnancies display an elevated CD163‐positive type 2 macrophage load in the decidua, which is associated with fetal‐maternal HLA mismatches

Abstract Problem The embryo of an oocyte donation (OD) pregnancy is completely allogeneic to the mother, which may challenge the maternal immune system to tolerize the fetus. Decidual macrophages are essential in maintaining a healthy pregnancy, and type 2 macrophages may exhibit immune suppressive activity. We hypothesized that the composition of decidual macrophages is different between uncomplicated OD pregnancies and non‐OD in vitro fertilization (IVF) pregnancies, and is related to fetal‐maternal incompatibility. Method of study Women with uncomplicated pregnancy were enrolled: 25 singleton OD pregnancies and 17 non‐OD IVF pregnancies. The extent of immunohistochemical staining of CD14 (pan‐macrophage marker) and CD163 (type 2 macrophage marker) in both decidua basalis and parietalis was quantitated by digital image analysis. Maternal and fetal DNA was typed for human leukocyte antigen (HLA)‐A, ‐B, C, ‐DRB1, and ‐DQB1, and fetal‐maternal HLA mismatches were calculated. Results OD pregnancies showed a higher percentage of CD163+ staining (P = .040) and higher CD163/CD14 ratio (P = .032) in the parietalis than non‐OD IVF. The OD group was separated into a semi‐allogeneic group (≤5 fetal maternal HLA mismatches) and a fully allogeneic group (> 5 mismatches). The HLA‐fully‐allogeneic OD group, but not the HLA‐semi‐allogeneic OD group, showed significantly elevated CD163/CD14 ratio in the parietalis compared with the non‐OD IVF group (P < .05). Conclusions Uncomplicated OD pregnancies display a higher CD163‐positive cell fraction in the total decidual macrophage population compared to autologous pregnancies, which may suggest that a local type 2 macrophage‐related mechanism is needed to compensate for the higher fetal‐maternal HLA mismatch load.


INTRODUCTION
During pregnancy, fetal and maternal tissue come into contact as soon as the blastocyst implants into the endometrium and thereafter through the whole gestation period. 1 Since the fetus is genetically different from the mother, the fetal cells are exposed to a potential attack of the maternal immune system. 2 To maintain an uncomplicated pregnancy until term, immune modulatory mechanism are at play at the fetal-maternal interface. 3 Maternal immune cells contact with fetal cells at multiple locations during gestation, 4 in which the decidua basalis is the maternal part of the placenta and invaded by the extravillous trophoblast; the decidua parietalis is the maternal part of the membrane and confronts the trophoblast cells of the chorion.
In oocyte donation (OD) pregnancy, the fetus has inherited genes from both the father and the donor: as a consequence the fetus usually is fully allogeneic to the pregnant mother. 5 OD pregnancies are related to a higher extent of human leukocyte antigen (HLA) mismatches between the fetus and the mother compared to non-OD pregnancies. It is reported that women with OD pregnancy have higher risk of pregnancy complications than women with non-OD pregnancies, which include hypertensive disorders, preeclampsia, preterm birth, and low birth weight. 6,7 In successful uncomplicated OD pregnancies, HLA matching between fetus and mother was observed to be significantly higher than expected by chance. 8 These observations suggest that a larger number of HLA mismatches is an essential factor that accounts for women with pregnancy after OD to have a higher chance of developing complications. It is hypothesized that due to this higher gene dissimilarity, the demand for immune cells in the fetal-maternal interface of OD pregnancies is higher, in order to maintain tolerance during pregnancy. 9 Among all immune cells in the placental bed, macrophages have shown their importance from multiple aspects. In the total decidual immune cell population, 20-25% is represented by macrophages. 10 This quantity stays constant over the whole period of pregnancy, 11 suggesting that macrophages have a role in each trimester. Moreover, macrophages have been suggested to be functional in multiple maternal adaptation processes. 12 During implantation, a large amount of macrophages migrate towards the invading trophoblasts and spiral arteries, implying a role for decidual macrophages in trophoblast invasion, vascular remodeling, and placentation. 13,14 Macrophages also act as phagocytes to engulf apoptotic cells, thereby not only preventing proinflammatory reactions but also regulating the extent of fetal cells invading into the uterine wall. 15,16 The capability of macrophages to provide multiple functions during pregnancy is related to their phenotypical plasticity. Macrophages are often classified into two categories based on phenotype and characteristics, namely classically activated macrophages (M1) and alternatively activated macrophages (M2). 17 M1 macrophages are proinflammatory and regarded to be defensive against infections and tumors, whereas M2 macrophages are anti-inflammatory exhibiting immune suppressive activity. 18,19 Cluster of differentiation (CD) 14 is a glycosylphosphatidylinositol anchored membrane protein, expressed on monocytes and macrophages, while CD163 is a glycoprotein antigen expressed on M2 macrophages. 10,20 As macrophages are essential in maintaining healthy pregnancy we hypothesize that the quantity and composition of decidual macrophages are different between uncomplicated OD pregnancies and non-OD in vitro fertilization (IVF) pregnancies, and that these differences in macrophages are related to fetal-maternal incompatibility.
Therefore, in this study we investigated the quantity and composition of the macrophage population in those groups by immunohistochemical staining and related the macrophage load to fetal-maternal HLA mismatches.

Patient selection
This retrospective case-control study was performed at the Leiden Uni-

Immunohistochemistry
Placental tissues were embedded in paraffin. From each placenta two paraffin blocks were selected, one from the decidua basalis and one from the decidua parietalis. Most tissue blocks were selected from the lateral part of the placenta; central blocks were used when the lateral blocks were of low quality. Sequential serial sections (4 µm)

Double label immunofluorescence of CD14 and CD163
Three basalis samples and three parietalis samples with various  All basalis slides were semi-quantitatively and quantitatively scored by KA. All parietalis slides were semi-quantitatively and quantitatively scored by XT. Fifteen slides for decidua basalis and twenty slides for decidua parietalis were randomly selected and assessed by two investigators individually, to determine the discrepancy between two observers for both semi-quantitative and quantitative scores. All the investigators were blinded to the clinical history of the patients.

Quantification of double label immunofluorescence staining
On each slide, five .5mm 2 regions containing basalis or parietalis were selected randomly to perform cell quantification analysis, using cell All HLA antigens were evaluated at split level. The HLA class I mismatches were defined as the summary of HLA-A, -B, -C mismatches; the HLA class II mismatches were defined as the summary of HLA-DRB1 and -DQB1 mismatches; the total HLA mismatches were defined as the summary of all these five HLA loci mismatches.
To determine the association between fetal-maternal HLA mismatches and the percentage of CD14 and CD163 positive cells, OD pregnancies were divided into two groups according to the number of HLA mismatches. In the semi-allogeneic group, the number of mismatches was not higher than half of the antigens per HLA locus, which is similar to the maximum fetal-maternal HLA mismatches in the non-OD IVF pregnancies. In the fully allogeneic group, the number of mismatches was higher than half of the antigens per HLA locus, which is a situation unique for OD pregnancies.

Data analysis
Statistical analyses were performed using SPSS statistics 25

Patient characteristics
Clinical characteristics are summarized in Table 1. There were no significant differences in any of the maternal or fetal parameters between the OD group and non-OD IVF group.

The extent of macrophage positive surface staining in the decidua
The expression of CD14 and CD163 in the decidua basalis and parietalis was analyzed and compared quantitatively and semiquantitatively between the OD group and non-OD IVF group.
First, the reproducibility of semi-quantitative and quantitative analyses of the extent of positive staining was evaluated. High correlations were found for intra-observer variation and inter-observer variation.
High similarity was also shown between semi-quantitative and quantitative results (Supplementary Material 2).
A significantly lower extent of CD14+ staining in basalis was observed in the OD group (median 4.55%) compared to the non-OD IVF group (median 7.55%) (P = .030; Figure 1A). Semi-quantitative scoring of CD14+ staining gave similar results ( Figure S3A). The extent of CD163+ staining in the basalis was not different between groups.
The CD163/CD14 ratio in the OD group (median .84) was higher than in the non-OD IVF group (median .59), but this difference was not significant ( Figure 1B, C).
Next, the surface staining in the decidua parietalis of CD14+ and CD163+, along with the CD163/CD14 ratio, was analyzed. In computerized quantitative analysis, CD163-positive staining was significantly higher in the OD group (median 1.04%) compared to the non-OD IVF group (median .39%, P = .040), but no significant difference was detected in the extent of CD14 positive staining in the parietalis between groups (Figure 1D, E).
When comparing CD163 and CD14, we found that the intensity of CD14 staining in parietalis was generally lower than CD163 staining in parietalis. Thus, the thresholds for CD14 positive signals were set higher than those for CD163 positive signals, to only measure specific staining signal and avoid including measurement of background staining in the tissue sections. As ratio between extent of CD163 staining and extent of CD14 staining was peculiarly high (> 3) in some cases, we suspected the difference in intensity of the two stainings might be explaining these high ratios. Indeed we found a significant correlation between staining intensity ratio and staining extent ratio (Spearman's r = −.47, R 2 = .22, Spearman's P = .002). Therefore, four samples with extreme intensity ratios (either < .5 or > 1.5) were excluded from the analyses altogether (see detailed methods in Supplementary Material 4 and results in Figure S4), resulting in disappearance of the significant correlation between staining intensity ratio and staining extent ratio ( Figure S4b). After this correction, the ratio of extent of CD163 to CD14 surface staining was significantly higher in the OD group (median .84) compared to the non-OD IVF group (median .45, P = .032) ( Figure 1F).
We wanted to test if the CD163/CD14 ratios measured in the decidua reflect the cells that are double positive for CD14 and CD163.
Therefore, a double label immunofluorescence staining was performed for three decidua basalis and three decidua parietalis samples, showing either a high, medium or low ratio. In total, 4255 single or double showed positivity for CD163 (Figure 2A-C). In the total sample set, the immunohistochemical CD163/CD14 ratio highly correlated with the proportion of CD163/CD14 double-stained cells in immunofluorescence (Spearman's R 2 = .89, Spearman's P = .017; Figure 2D).
Since the sampling time has significant difference between the two groups (P = .028) and the median of the sampling time in the OD group is higher than that in the non-OD IVF group, we performed the correlation analysis (Spearman's correlation) between outcomes and sampling time within the groups: none of them showed significant correlation.
Although the comparison of clinical parameters showed no significant differences between the two groups, we still performed the correlation analysis between the outcomes (CD14, CD163 and CD163/CD14 ratio in decidua basalis and parietalis) and patient characteristics (maternal age and gestation age), which all showed no significant correlation (Spearman's correlation).

Association between the ratio of CD163/CD14 positive staining and fetal-maternal HLA mismatches
We wanted to test whether differences in macrophage composition are related to fetal-maternal HLA incompatibility. Therefore, we divided the OD cohort into a semi-allogeneic group (≤5 fetal-maternal HLA mismatches) and a fully allogeneic group (> 5 fetal-maternal HLA mis-matches). Clinical data of the three groups, namely non-OD IVF group, semi-allogeneic OD group and fully allogeneic OD group, showed no significant differences (data not shown).
Since the CD163/CD14 ratio showed significant difference between OD and non-OD IVF group in decidua parietalis, but not in decidua basalis, here we concentrate on the association between the CD163/CD14 ratio in decidua parietalis and fetal-maternal HLA mismatching. First, at individual locus level, we found that the HLA-DR-fully-allogeneic OD group showed a higher CD163/CD14 ratio than the non-OD IVF group (P = .047) ( Figure 3A). This significantly higher CD163/CD14 ratio was also found when comparing HLAclass-II-fully-allogeneic OD pregnancies with non-OD IVF pregnancies (P = .047) ( Figure S5A).
There was also a significantly higher CD163/CD14 ratio in the HLA-B-fully-allogeneic and HLA-C-fully-allogeneic OD group than in the non-OD IVF group (P = .037 and P = .027, respectively) ( Figure 3B, C). In line with this finding, a significantly higher percentage of CD163 positive staining in decidua parietalis was observed when comparing HLA-B-or HLA-C-fully-allogeneic OD group with non-OD IVF group (P = .010 and P = .023, respectively) ( Figure S5B, C). This significantly higher CD163 positive staining was also found in the HLA-class-I-fullyallogeneic OD group compared to non-OD IVF group (P = .047) (Figure S5D).
In addition, when comparing total-HLA (class I+II)-fully-allogeneic group with non-OD IVF group, a significantly higher CD163/CD14 ratio and CD163 positive staining were observed (P = .019 and P = .031, respectively) ( Figure S5E, F). and CD163 macrophages markers are significantly upregulated. 28 Nakabayashi hypothesized that the decrease of macrophages in the decidua basalis and the increase of macrophages in the chorionic plate might be due to migration of macrophages from one layer to the other. 25 However, both these studies did not calculate the percentage of CD163 positive macrophages (M2 macrophages) in the total number of macrophages, which is the main finding of our study. CD163 has been suggested as a marker of anti-inflammatory macrophages, 13,18 and CD14 is considered to be present on all macrophages, irrespective whether they are pro-inflammatory or anti-inflammatory. Thus the CD163/CD14 ratio was calculated to determine the polarization of the decidual macrophages. Since the chorionic plate, villi, and basal plate are all directly in contact with maternal tissues or maternal blood in the intervillous space, fetal-maternal interactions may be occurring at each of these layers. 29 Thus, immune responses would be expected to achieve fetal-maternal tolerance in all these locations. Therefore, besides the migration of immune cells from one place to the other, the polarization of macrophages might be an explanation for the changes of general and M2 macrophage markers in different layers of the placenta.
As all OD pregnancies included in our study were maintained until term delivery without complications, the higher ratio of decidual M2 macrophages, which are generally believed to have anti-inflammatory and immunoregulatory effect, 19,30 may reflect a local compensatory mechanism to inflammatory conditions to make sure that the fetus is retained until the end of pregnancy. Although decidual macrophages were reported to mediate a pro-inflammatory environment at the very beginning of implantation 31 and at term, 32 this argument does not refute our hypothesis as the sampling time between the two study F I G U R E 3 Association of ratio of CD163/CD14 and CD163 positive staining in decidua parietalis with the extent of HLA mismatching. The numbers in parentheses indicate the range of HLA mismatches for each group: IVF (0-1) refers to non-OD IVF group, OD (0-1) refers to semi-allogeneic OD group, OD (2) refers to fully allogeneic OD group. The middle horizonal line within the box indicates the median, the ends of the box correspond to the upper and lower quartiles of the data, and the whiskers indicate minimum and maximum values. Mann-Whitney U tests were performed to identify differences between two groups. Significant differences are shown with p value, ns means non-significant groups was similar. Admittedly, in this study we only used CD163 as the surface marker for M2 macrophages, and the exact functional role of these CD163 positive cells needs to be further defined. It is reported that a high CD163 expression in macrophages is a characteristic of antiinflammatory responses, accounted to the potential to scavenge cellular debris, produce anti-inflammatory cytokines such as IL-10, and limit progression of inflammatory reactions. 33 It is expected that in the double staining all CD163+ cells are positive for CD14+ and that consequently the CD163/CD14 ratio is lower than 1, as CD14 is present on all macrophages and CD163 is only present on a subset of anti-inflammatory macrophages. 10,13 Yet, several samples in our study had a ratio higher than 1 and a few single CD163 positive cells were also observed, which might suggest that not all CD163+ cells are CD14+. An earlier study of patients with preeclampsia and preterm birth found a unique subset of CD14−/CD68+ cells in the placenta, and concluded that these cells were a subpopulation of macrophages based on their immunophenotypic characteristics. 37 However, another study found a CD14-CD163+ dendritic cell subset, secreting intermediate quantities of pro-inflammatory mediators. 38 The presence of a CD14-group is hinted to but not proven in our study. Future study with larger sample size might be more suitable to identify a new subset in placenta.
Moreover, functional tests will be needed to determine the characteristics of such subset. An alternative explanation of some samples exceeding 1 or even 2 might be that the CD14 staining was less intense than the CD163 staining, which became more apparent in the parietalis sections. Although both stainings worked optimal and were applied at the same time on the whole series of slides from both study groups, elevating the threshold for detecting CD14 specific signal may have led to loss of weak but specific CD14 signal in some samples.
The association between macrophages and HLA mismatches was analyzed as well. For evaluating associations we concentrated on CD163 positive staining and CD163/CD14 ratio in decidua parietalis, since they showed the largest differences between groups. To eliminate confounding variables and to only focus on HLA mismatches, we separated the OD group into a semi-allogeneic and a fully allogeneic group.
Clinical data of these two groups and the non-OD IVF group showed no significant differences. Therefore, only the number of HLA mismatches was the independent variable between semi-allogeneic and fully allogeneic OD group. Unfortunately, between the semi-allogeneic and fully allogeneic OD group no significant difference was found for CD163 positive staining or CD14/CD163 ratio in the decidua parietalis, possibly due to the limited sample size in our study. However, when comparing fully allogeneic OD group with non-OD IVF group, a significantly higher percentage of CD163 positive cells and higher CD163/CD14 ratio were observed in the decidual parietalis. These differences were not found when comparing semi-allogeneic OD group with non-OD IVF group, suggesting that the extent of HLA mismatching between mother and fetus remains a possible factor associated with the quantity and the composition of macrophages in the placenta of OD pregnancies. A significant correlation between HLA mismatches and macrophages in OD pregnancies might be found if sample size would be enlarged.
Among all the five HLA loci that we calculated in our study, HLA-DR incompatibility was related to the most prominent difference in macrophage composition between groups. We found that the association not only exist between macrophages composition and HLA-DR The exact mechanism of how macrophages would respond to larger genetic dissimilarity is not yet clear and needs to be further estab- become impaired and abnormal when fetal-maternal genetic dissimilarity is larger. But macrophages gathering around the spiral arterials might be able to compensate for this impairment and dissimilarity. 13 In conclusion, we found a higher extent of CD163 positive M2 macrophages within the total decidual macrophage load of uncomplicated OD pregnancy than in non-OD IVF pregnancies. The observations may demonstrate that a compensatory mechanism is needed in order to deal with the higher HLA incompatibility between fetus and mother in OD pregnancies. The exact mechanism is yet to be established.