Women with diabetes have increased stillbirth risk. Although the underlying pathophysiological processes are poorly understood, stillbirth is frequently related to abnormal placental structure and function.
To investigate placental morphology and cellular characteristics in the placentas of women with diabetes who had stillbirths and stillbirths of unexplained cause.
Placentas from women with uncomplicated live births, live births in women with diabetes, unexplained stillbirths, and stillbirths related to diabetes (n = 10/group) underwent clinical histopathologic assessment and were also investigated using immunohistochemical staining to quantify syncytial nuclear aggregates, proliferation, trophoblast area, vascularization, T cells, placental macrophages (Hofbauer cells), and the receptor for advanced glycation end products.
Ki67+ cells were decreased in unexplained stillbirths compared with live births in women with diabetes. Both stillbirth groups had increased cytokeratin 7+/nuclear area compared with controls. Blood vessels/villi were decreased in unexplained stillbirth compared with live births from women with diabetes. Compared with uncomplicated controls, CD163+ macrophages were increased in live births in women with diabetes and unexplained stillbirths, and further increased in stillbirths related to diabetes. There was no change in CD3+ T cells or syncytial nuclear aggregates. Receptor for advanced glycation end products–positive cells were decreased in both stillbirth groups compared with diabetes-related live births. Co-localization of receptor for advanced glycation end products in macrophages was increased in both stillbirth groups compared with live birth groups.
Stillbirths related to diabetes exhibit placental phenotypic differences compared with live births. Further investigation of these parameters may provide understanding of the pathologic mechanisms of stillbirth and aid the development of stillbirth prevention strategies.
Stillbirth is defined in the United Kingdom as intrauterine fetal death after 24-weeks gestation and in 2017 the stillbirth rate was 4.2 per 1000 total births.1 The stillbirth rate in women with pre-existing diabetes is increased around 4-fold.2 The impact of gestational diabetes (GDM) on stillbirth is less clear, with some studies reporting increased risk and others suggesting the risk is mitigated by appropriate screening and management of the condition.3 All forms of maternal diabetes are characterized by hyperglycemia, but Type 1 diabetes is caused by a lack of insulin whereas Type 2 diabetes and GDM are caused by long-term and transient insulin resistance, respectively. The greater stillbirth risk associated with pre-existing diabetes compared with GDM may be due to the relative duration or severity of exposure to hyperglycemia.4
Histologic examination of the placenta is used to provide evidence to determine the cause of stillbirth. Placental abnormalities are found in around one-third of stillbirths and placental examination contributes to the classification of stillbirth cause in 47% of cases, and is the primary source of evidence in 16%.5 Our previous study of placental morphology in cases of stillbirth found that placentas from women with diabetes were associated with reduced proliferation, but no changes in trophoblast area, syncytial nuclear aggregates (SNA), or CD45+ leukocyte infiltration.6 Further research is needed to confirm these findings and also to compare placental morphology between women with diabetes who had a stillbirth and those who had a live birth.
Controlled or temporary inflammation plays a physiological role in tissue homeostasis; however, sustained or exaggerated inflammation can be pathogenic. In women with diabetes, the placenta has a chronic elevated inflammatory state.7 Abnormal inflammation is associated with adverse outcomes in pregnancy, including preeclampsia, preterm birth, miscarriage, fetal growth restriction, and stillbirth.8 Inflammation is characterized by the accumulation of immune cells, which is directed by cell signaling. One such signaling pathway signals through the receptor for advanced glycation endproducts (RAGE). Activation of RAGE signaling is generated by binding with advanced glycation endproducts (AGE) or other ligands, resulting in a proinflammatory response. The formation of AGE is accelerated in a hyperglycemic environment and is involved in the progression of secondary complications.9 We hypothesized there may be differences in the placental structure, presence of immune cells, and RAGE in the placentas of diabetes-related stillbirths compared with live births.
Placental Tissue Samples
Placenta samples were obtained from women who consented for postmortem examination and tissue to be used for research. Cases were identified by a consultant histopathologist (GB) using information from postmortem reports, obstetric review, and placental histopathology reports. Cases with congenital abnormalities, reports of clinical infection, ischemia, and intrapartum asphyxia were excluded from the study. Cases of unexplained stillbirths were excluded if placental changes (eg, villous immaturity and fetal vascular malperfusion) were identified.
Stillbirths were attributed to diabetes if they occurred in women with pre-existing Type 1 or 2 diabetes or GDM, and this was deemed to be the cause of death following multidisciplinary review using the ReCoDe classification system.10 Unexplained stillbirth cases lacked any cause or abnormality, which was not deemed causative by multidisciplinary review. Findings from histopathological examination from cases of stillbirth were classified using Turowski classification11 to determine if there were any differences between the stillbirth groups. Cases of livebirths from uncomplicated pregnancies and livebirths from women with diabetes were obtained from the Maternal and Fetal Health Research Centre placental tissue biobank (Ref 08/H1010/55+5). Cases were matched to the respective stillbirth groups based on gestation, body mass index, ethnicity, and parity. The current study was approved by the Greater Manchester South Research Ethics Committee (09/H1012/11).
Histochemical and immunohistochemical analysis was carried out as previously described to assess the cellular composition and structural morphology of the placental samples.6 The placenta samples had not been used in previous studies from our laboratory. Samples were classified into the following 4 groups: live birth from uncomplicated pregnancy (n = 10), live birth from woman with diabetes (n = 10), unexplained stillbirth (n = 10), and stillbirth from woman with diabetes (n = 10). The tissue was stained with hematoxylin and eosin and using antibodies against cytokeratin 7 (CK7), Ki67, CD31, CD3, CD163, and RAGE.
Briefly, 3 villous tissue samples (1–2 cm3) from the center, middle, and edge of each placenta were fixed in 4% neutral buffered formalin and embedded in wax. Immunohistochemical staining was performed on 5-μm paraffin sections on poly-L-lysine (Sigma Chemical Company, St Louis, Missouri) coated slides. The immunohistochemistry protocol was carried out on an IntelliPATH FLX autostainer (A.Menarini Diagnostics LTD, Berkshire, United Kingdom) or by hand. Antigen retrieval was undertaken in either citrate buffer pH 6.0, ethylenediaminetetraacetic acid solution pH 8.0, or Menapath Access super pH 9.5 (MP-606-PG1; A.Menarini diagnostics LTD). Endogenous peroxidase quenching was carried out with MenaPath Peroxidase block or 3% hydrogen peroxide (Sigma). Background block was performed with MenaPath Casein block or serum block (10% goat [Sigma] or 10% rabbit [Sigma] with 2% human [Sigma] serums in tris-buffered saline with 0.1% tween). Primary antibodies Ki67 (MP-325-CRM1, 250 μg/mL, Menapath), CK7 (MP-061-CM1, 250 μg/mL, MenaPath), CD31 (M082301-2, 214 μg/mL, Dako, Carpinteria, California), CD163 (MCA1853, 5 μg/mL, Bio-Rad, Hercules, California), CD3 (M725, 1.4 μg/mL, Dako), and RAGE (ab37647, 16 μg/mL, Abcam) were diluted in Menapath universal antibody diluent or serum block and incubated for 30 minutes or overnight at 4°C. Negative controls were carried out using nonimmune mouse or rabbit isotype controls (Dako) at matching concentrations to the primary antibody. Autostained slides were incubated in universal probe, HRP-Polymer, and diaminobenzoate (MenaPath), and then counterstained in modified Lillie-Mayer Hematoxylin (MenaPath). Hand-stained slides were incubated in either goat anti-mouse (3.3 μg/mL, Dako) or goat anti-rabbit (5 μg/mL, Dako) biotinylated secondary antibodies, then avidin–peroxidase (Sigma), diaminobenzoate (Sigma), and counterstained with Harris modified hematoxylin (Sigma). Autostained and hand-stained diaminobenzoate slides were dehydrated and mounted using DPX (Sigma).
Fluorescent-stained slides were treated with the same protocol as the immunoperoxidase-stained slides up to the application of the secondary antibody. Slides were incubated in either donkey anti-mouse Alexa Fluor 488 or donkey anti-rabbit Alexa Fluor 568 secondary antibodies (10 μg/mL; Thermo Scientific). Fluorescent-stained slides were treated to quench autofluorescence (Trueview, Vector), counterstained with 4′,6-diamidino-2-phenylindole (Sigma), and mounted with HardSet Mounting Medium (Vector).
Nonfluorescent images for analysis were captured by either using a light microscope to capture 5 images taken at random over the section area, or by capturing the whole section using a slide scanner. Light microscope images were captured using an Olympus BX41 light microscope (Southend-on-Sea, United Kingdom) with a mounted QIcam Fast 1394 (QImaging, British Columbia, Canada) microscope camera at ×20 magnification using Image Pro Plus 7.0 (Media Cybernetics Inc., Maryland) software. Whole slide images were captured with a Pannoramic 250 slide scanner (3D Histec). For fluorescent slides 10 images were captured at ×40 using a Zeiss Axiophot microscope with Zeiss Zen software. The exposure was set on each slide by running auto exposure on the most intense area for each fluorophore.
SNA were quantified from light microscope images of hematoxylin and eosin staining. A syncytial knot was defined as a multilayered aggregation of at least 10 syncytiotrophoblast nuclei protruding from the villous surface that was not in direct contact with adjacent villi. The number of syncytial knots was counted manually, and the villous tissue area was determined by red-green-blue masking using Image Pro Plus 7.0 software, permitting calculation of the number of SNA per millimeters squared. The proportion of Ki67+ nuclei was determined by an automated image analysis protocol in HistoQuest software (v3.5.3, Tissue Gnostics). The trophoblast area was determined as the CK7+ tissue area normalized to the nuclear hematoxylin area in light microscope images using a HistoQuest protocol. The average number of vessels per villus was determined by identifying the number of whole villi and the number of CD31-stained vessels within those villi by manual counting. The proportion of CD163+ and RAGE+ cells was determined by running an automated positive cell detection script for the whole tissue section in Qupath software (v0.2.0-m2).12 The proportion of CD3+ cells was determined from scanned images by selecting 10 areas of 0.25 mm2, manually counting the number of diaminobenzoate-positive cells, and running an automated cell count using Qupath. The proportion of fluorescent CD163+ and RAGE+ co-localized cells was determined by running an automated co-localization script on microscope images in Qupath. The thresholds were set for each slide to minimize background staining for each fluorophore.
All statistical analysis was performed using GraphPad Prism software (Version 7, La Jolla, California). Outliers were identified, investigated, and removed if necessary. Groups were compared using non-parametric (Kruskal-Wallis) analyses with Dunn's post hoc test.
Study Population and Placental Classification
The following 4 groups were investigated: live births without complications, live births in women with diabetes, otherwise uncomplicated unexplained stillbirths, and stillbirths in women with diabetes. By design, there were no significant differences between gestation, body mass index, ethnicity, parity, and diabetes type in the study population (Table 1). Placental histopathology using Turowski's classification confirmed the majority of placentas from diabetes-related stillbirths had villous immaturity and the majority of unexplained stillbirth group placentas were classified as normal (Table 2). The remaining placentas from the unexplained stillbirths had abnormalities that were not thought to be causative (mild chorioamnionitis and a circumvallate placenta with a small peripheral hemorrhage).
Placental Morphology and Characterization of Cell Type
Histologic analyses were undertaken to define the morphologic and cellular characteristics of placentas. There was no difference in the prevalence of SNAs between the groups (Figure 1, A through C). The proportion of Ki67+ cells was reduced in unexplained stillbirths compared with livebirths (P = .04; Figure 1, D through F) and stillbirths from women with diabetes (P = .04; Figure 1, D through F). CK7+ trophoblast/nuclear area was increased in unexplained stillbirths compared with uncomplicated live births (P = .01; Figure 1, G through I) and diabetes-related live births (P < .001; Figure 1, G through I). Diabetes-related stillbirths also had higher CK7+ trophoblast/nuclear area compared with live births from women with diabetes (P = .002; Figure 1, G through I). There was a decrease in the number of CD31+ blood vessels per villus in the unexplained stillbirths compared with live births from women with diabetes (P = .01; Figure 1, J through L).
Histologic analysis of immune cells and inflammatory marker RAGE was undertaken. Compared with uncomplicated livebirths the proportion of placental macrophages (Hofbauer cells) was significantly increased in live births from women with diabetes (P = .02; Figure 2, A through C) and unexplained stillbirths (P = .03; Figure 2, A through C), and further increased in stillbirths from women with diabetes (P < .001; Figure 2, A through C). There was no difference in the number of CD3+ T cells between the groups (Figure 2, D through F). The number of RAGE+ cells was lower in stillbirths of women with diabetes compared with uncomplicated livebirths (P = .04; Figure 2, G through I) and live births from women with diabetes (P = .002; Figure 2, G through I). The number of RAGE+ cells was also lower in unexplained stillbirths compared with livebirths of women with diabetes (P < .001; Figure 2, G through I). The proportion of cells that were both CD163+ and RAGE+ was increased in both of the stillbirth groups compared to both of the live birth groups (P < .001; Figure 2, J through L).
The morphology and cellular composition of the placenta after birth are thought to be indicative of pathological processes present in pregnancy. From routine histopathological examination we observed villous immaturity characterized by larger terminal villi with more centralized vascular channels, increased stroma, and poor vasculosyncytial membrane formation in diabetes-related stillbirth cases. This agrees with a review which identified villous immaturity as one of the most commonly observed characteristics of placentas in women with diabetes.13 Trophoblast area was also increased in stillbirths related to diabetes; ultrastructural studies have identified that this observation is indicative of thickening of the vasculosyncytial membrane. Thus, the structural abnormalities associated with immature villi could result in inhibition of maternal–fetal exchange.14
Stillbirth cases were classified as unexplained by multidisciplinary case review and in 8 of 10 cases the placentas were classified as normal using Turowski's Classification System. In 2 of 10 cases placental abnormalities were observed; however, they were deemed insufficient to constitute the cause of stillbirth in multidisciplinary case review. Indeed, placental abnormalities that can be insufficient to indicate stillbirth cause, such as chorioamnionitis, can be observed in otherwise uncomplicated pregnancies.15 Although the histopathological examination of the unexplained stillbirths were classified as normal in 8 of 10 placentas, the trophoblast area was increased in common with the diabetes-related stillbirths. In contrast, 1 study has reported the trophoblast area is decreased in stillbirth16 and others have reported no difference from control cases in unexplained6 and diabetes-related stillbirths.6,17
Increases in SNA have been associated with adverse outcomes, such as fetal growth restriction,18,19 stillbirth,16 and unexplained stillbirth6 ; however, we did not observe an increase in SNA in the unexplained stillbirth group in the current study. One study of livebirth cases reported an increase in SNA in subclinical hyperglycemia and overt diabetes mellitus but not GDM.20 Despite the majority of our diabetes-related stillbirth cases being pre-existing diabetes mellitus we did not observe an increase in SNA in the current study, which is also in line with our previous observations.6 The observed decrease in proliferation measured by Ki67 expression in the unexplained stillbirth group compared with livebirth controls was also concordant with our previous study; however, a reduction in proliferation in the diabetic stillbirth group was also previously observed.6 Another study has also suggested a decrease in placental proliferation in maternal diabetes live birth cases.21 A decrease in proliferation would suggest the capability of the villi to maintain tissue homeostasis is reduced, which could be a factor in stillbirth. A reduction in villous vasculature has previously been associated with fetal growth restriction,6,19 stillbirth,16,22 and unexplained stillbirth cases6 ; the observed decrease in blood vessel number in unexplained stillbirths could indicate compromised maternal–fetal exchange. Our observations suggest changes to SNA and villous vascularity may not have a role in stillbirths related to diabetes but may have a role in unexplained stillbirths. Trophoblast area and proliferation may have a role in both diabetes-related and unexplained stillbirth.
We observed an increase in placental CD163+ cells in both diabetes-related groups compared with uncomplicated live births. There was no change in CD3+ cells, which contrasts with the increase in T cells routinely observed in villitis of unknown etiology.8 This suggests the adaptive immune T-cell response or the pathological mechanisms in villitis of unknown etiology may not be important factors in diabetes-related or unexplained stillbirths. The increase in the number of CD163+ cells may be caused by the hyperglycemic environment, as maternal diabetes is associated with sustained inflammatory environment in the placenta.7,23 An increase in CD163+ cells has also been observed in the chorion and decidua in GDM; however, in contrast to our findings, this study found no increase in the villous core.24 The increase in macrophages in unexplained stillbirths observed in the current study demonstrates this observation is not limited to diabetes. Indeed, there are several reports of increased macrophages from cases of stillbirth relating to villitis of unknown etiology and fetal growth restriction.8 Although CD163+ macrophages are generally considered to have an M2 phenotype producing anti-inflammatory cytokines, M1-M2 macrophage phenotype is now regarded as a spectrum rather than distinct groupings. This is supported by evidence that CD163+ macrophages can produce proinflammatory cytokines, such as interleukin (IL)-1β, IL-6,25 and IL-12.8 Other studies have made a case for a proinflammatory placental macrophage role in pathological environments, including an increase in M1 macrophage markers in villitis of unknown etiology26 and GDM.27 Therefore, further investigation is required to explore whether increases in the number of CD163+ cells represents a pro- or anti-inflammatory response and if they are involved in pathological processes in the placenta relating diabetes and stillbirth.
AGE are macromolecules that react nonenzymatically with reducing carbohydrates, such as glucose. RAGE binds AGE among other ligands triggering a proinflammatory response.28 AGE are elevated when blood glucose levels are increased in diabetes and are involved in the development of secondary complications.29 We observed a trend toward increased RAGE in livebirths in women with diabetes compared with uncomplicated controls while other reports suggest a decrease in RAGE gene expression in live births from women with GDM.30 It has previously been suggested RAGE is localized in trophoblasts of the villus31 ; however, our observations show that RAGE is variably localized in endothelial cells, trophoblast, and the villous parenchyma, including macrophages. Each live birth group had higher levels of RAGE compared with at least 1 of the stillbirth groups. This suggests RAGE may have a protective effect against stillbirth, which is counterintuitive considering the commonly reported pathogenic inflammatory role of the RAGE pathway. Indeed, in reproductive research increased levels of AGE and RAGE have been implicated in oxidative damage in a trophoblast cell line,32 inhibition of embryo implantation,33 and pre-eclampsia and decreased trophoblast invasion.30 Another explanation for the reduction in RAGE observed may be related to a decline in RAGE production after fetal death. RAGE can also play an anti-inflammatory role as a decoy receptor quenching RAGE ligand in its nonmembrane-bound secretory isoform.34 However, our staining is focal, suggesting our observations are membrane-bound RAGE rather than its nonmembrane-bound secretory isoform. We have demonstrated that RAGE co-localizes with CD163+ macrophages to a greater degree in the stillbirth groups. This suggests CD163+RAGE+ cells may either play a role in stillbirth or that macrophages take up RAGE after fetal death. In support of the former hypothesis, RAGE has been implicated as a factor in the function of macrophages relating to the pathogenesis of atherosclerosis.35 Furthermore, RAGE has been associated with proinflammatory M1 polarization36 and macrophage migration/accumulation.37 Other authors suggest RAGE signaling in M2 macrophages does not alter polarization and produces a response typical of an anti-inflammatory M2 phenotype.38 Further study is needed to understand what effect RAGE expression has in macrophages or how this might impact birth outcome.
Due to the limited availability of placental tissue from diabetes-related stillbirths the cases were a mixture of Type 1 and 2 diabetes and GDM. Although the severity and mechanism of pathology are varied, all of these conditions share hyperglycemia during pregnancy as a defining characteristic. It is therefore suggested measuring the placental characteristics of these conditions as a single group should lead to valid observations. However, future studies with sufficient numbers of cases for each type of diabetes would be preferable as it would allow any differences between these conditions to be investigated. It is also important to be able to differentiate the changes observed here from artifacts caused by in utero retention between fetal death and birth and from placental storage or processing after delivery. Several of the morphologic characteristics we investigated in the current study are stable for up to 48 hours after delivery39 but the effects of in utero retention could contribute to the changes we observed. As the time of fetal death was unknown we cannot reliably estimate the duration of in utero retention in our study; we did not see evidence of gross devascularization or fibrosis of villi, which may have suggested prolonged in utero retention.40 Nevertheless, further study is required to determine the stability of the morphologic characteristics examined here across different times of in utero retention. Unexplained stillbirths were defined by an absence of known cause, which means they could be a homogenous group as defined by a single, yet discovered cause or they could have a variety of causes. Here, we excluded cases of villous immaturity from the unexplained stillbirth group, as this is the most commonly seen placental lesion in cases of diabetes. However, this lesion is not specific to diabetes, so we may have excluded other causes of stillbirth. The inclusion/exclusion criteria may explain some discrepancies observed between the current study and a previous study that investigated many of the same parameters.6 Nevertheless, there were also observations consistent between the studies. The current study adds to the characterization of stillbirths of unknown cause and further investigations will allow us to discover the nature of this group of stillbirth cases.
Histologic investigation of the placenta is vital to provide information to clinicians to determine the cause of adverse outcomes, including stillbirth. Current stillbirth classification systems recording the findings of placental examination are unable to identify all causes of stillbirth and therefore exploration of novel histologic parameters of the placenta is necessary. Some of our preliminary findings are in line with previous publications, including collated observations41 (Table 3). Observations confirmed in at least 2 studies suggest the number of SNA remain unchanged in stillbirths from women with diabetes, there is a reduction in proliferation in unexplained stillbirths, the number of blood vessels is unchanged in stillbirths from women with diabetes but decreased in unexplained stillbirths, and placental macrophages are increased in complicated live birth and stillbirth cases of variable causes. Our novel measurements demonstrate placentas from diabetes-related stillbirths had an increase in CD163+ cells, a decrease of RAGE+ cells, but an increase in CD163+RAGE+ cells. Further studies are required to confirm the significance and reproducibility of our findings. We anticipate these markers may be useful in determining the cause of stillbirth, indicating pathological processes, and identifying potential therapeutic intervention in cases of stillbirth.
The authors wish to acknowledge the contribution of parents who at the time of their bereavement gave consent for the use of samples for research purposes.
The authors have no relevant financial interest in the products or companies described in this article.