Pre-eclampsia is a pregnancy complication affecting approximately 5–8% of pregnant women and is capable of causing both maternal and fetal morbidity and mortality. Maternal death rates from pre-eclampsia have been significantly reduced by careful patient management in the developed world. However, significant maternal death from hypertension still occurs in developing countries, which account for 99% (533 000) of total annual global maternal deaths 1, 2. Pre-eclampsia is characterized by gestational hypertension, proteinuria, systemic endothelial cell activation and an exaggerated inflammatory response. Whilst the pathophysiology of pre-eclampsia is not completely understood, it is clear that the presence of the placenta is both necessary and sufficient to cause the disorder, and the delivery of the baby (and thus the removal of the placenta) remains the only current cure. The pathophysiology of pre-eclampsia involves a complicated web of interacting maternal and fetal factors. Slowly we are gaining an understanding of how these factors may fit together to result in the range of clinical presentations of pre-eclampsia, and it is becoming clear that the aetiology of early-onset and late-onset pre-eclampsia, as well as intra-uterine growth restriction (IUGR) may arise from overlapping but distinct underlying causes.
The placenta consists of a branching villous structure. Each villus contains fetal blood vessels and occasional macrophages in a core of mesenchymal connective tissue surrounded by two layers of specialized placental cells termed trophoblast. These two layers consist of an inner layer of mononuclear cytotrophoblast stem cells and an overlying continuous layer of multinucleated syncytiotrophoblast, which covers the entire surface of the villous placenta and provides an impermeable barrier to maternal blood (Figure 1). Cytotrophoblast fuse into the overlying syncytiotrophoblast to support its expansion and function as the site of nutrient exchange and steroid and hormone synthesis. Cytotrophoblast in villous tips also differentiate into extravillous trophoblast that grow out from the placenta in columns that merge with those from neighbouring villi to form a layer known as the cytotrophoblast shell, from which they invade into the decidua, migrate down and remodel the decidual spiral arteries (Figure 1).
Whilst the symptoms of pre-eclampsia do not manifest until approximately 18 weeks of gestation onwards, the pathogenesis is established much earlier, at the time of trophoblast invasion and remodelling of the spiral arteries during the first 12 weeks of pregnancy. This has led to pre-eclampsia being described as a ‘two-stage’ disease, where Stage I refers to the inadequate placentation early in pregnancy, whilst Stage II refers to the events that directly result in the onset of systemic endothelial activation and other clinical manifestations in the third trimester 3. Placental development is important from the very beginning of the first trimester, with trophoblast differentiation in the placenta, invasion of trophoblast into the decidua and trophoblast-induced remodelling of the spiral arteries, all crucial processes for the success of pregnancy. At the end of the first trimester, trophoblast plugs that have until now occluded the spiral arteries gradually dislodge, and oxygenated maternal blood can directly perfuse the placenta. As pregnancy progresses, inadequate placental perfusion as a result of insufficient spiral artery remodelling is believed to result in hypoxia–reperfusion-type injuries to the placenta, which in turn stimulates an increase in the release of material from the syncytiotrophoblast that is capable of activating the maternal endothelium 4, 5. It is unclear why some mothers are able to tolerate and clear this placental debris from the circulation better than others, and the maternal clearance and handling of this material may play an equally important role as the volume and nature of material released in the development, timing of onset and severity of the disease.
As is evident in the numerous processes briefly summarized above, the volume and detail of the work in the field of pre-eclampsia is far too great to cover adequately in a single review; thus, this review aims not to cover each aspect that may contribute to pre-eclampsia comprehensively, but to provide an overview that ties our current understanding of the contributing factors together.
Spiral artery remodelling (Stage I)
Many studies support the fact that the first major step in the pathophysiology of pre-eclampsia is inadequate spiral artery remodelling in the first trimester (Figure 2), although this has been questioned 6. Spiral artery remodelling occurs due to a subset of extravillous trophoblast termed endovascular trophoblast that migrate down these arteries and interdigitate between and replace the arterial endothelial cells in these vessels. This is achieved in part by mimicking some of the endothelial cell adhesion molecule repertoire 7, and by trophoblast induction of endothelial cell apoptosis 8, 9. This process may also be facilitated by priming of the spiral arteries by uterine natural killer cells and/or interstitial trophoblast that have invaded into the decidua and migrated to the arteries, where they may induce arterial smooth muscle cell apoptosis 10–15. This combination of events produces the ‘physiological changes of pregnancy’ described historically by Robertson et al(1967), resulting in a loss of smooth muscle and the elastic and collagenous extracellular matrix from the spiral arteries, and its replacement with a fibrin-based deposit (fibrinoid), such that by the end of the first trimester the spiral arteries are large, flaccid, non-vasoactive tubes 16. Endovascular trophoblast invasion continues until the middle of the second trimester, by which time trophoblast invasion has reached the inner one-third of the myometrium 17. The importance of spiral artery remodelling for the success of pregnancy and its role in pre-eclampsia will be discussed in more detail later in this review, but first let us briefly examine why insufficient spiral artery remodelling may occur.
The major placental defects that affect spiral artery remodelling in the first trimester of pregnancy involve extravillous trophoblast differentiation and invasion. Deficits in the process of cytotrophoblast differentiation into extravillous trophoblast may lead to a reduced pool of extravillous trophoblast, which could affect both the number of spiral arteries remodelled and the depth of this remodelling. The exact mechanisms that direct cytotrophoblast differentiation into either the syncytiotrophoblast or extravillous trophoblast lineages are not clearly understood, although a large number of tightly regulated and specific molecular changes have been identified, involving adhesion molecules, cell cycle regulators and proteases, as well as autocrine and paracrine signalling pathways (Figure 1, Table 1). Trophoblast differentiation is also regulated at the transcriptional level by a number of factors, including activator protein-1 family members Jun and Fos and basic helix–loop–helix transcription factors, such as Hash-2 and Id, which all play roles in trophoblast differentiation into an invasive extravillous phenotype 18, 19. Many of the same mechanisms that regulate cytotrophoblast differentiation into extravillous trophoblast continue to regulate the progressive differentiation of extravillous trophoblast from a proliferative to an invasive phenotype as they migrate away from the placenta, and the large network of interacting regulatory factors derived from both the placenta and decidua mean that none of these factors acts in isolation (Table 1) (for review, see 20). Aberrations in any of these regulatory pathways have the potential to affect spiral artery remodelling in a manner that, either alone or in association with other placental defects, may result in pre-eclampsia. Furthermore, the number of pro- and anti-inflammatory cytokines involved in these processes ties them closely to the maternal immune system.
Table 1. Factors that affect trophoblast differentiation, invasion and spiral artery remodelling
Cytokines and growth factors
Other factors (including cell cycle regulators and transcription factors)
As half the fetal genes are paternally derived, the implanting placenta and embryo are a ‘semi-allograft’ to the maternal immune system. However, the maternal immune system manages to recognize the placenta as a ‘temporary self’ and in normal pregnancy does not mount an immune response against it 21. This is partially due to: (a) a lack of trophoblast expression of class I (HLA-A and B) and II MHC, which ‘hides’ the placenta from the maternal immune system; and (b) trophoblast expression of a unique combination of HLA-C (class Ia), HLA-G and HLA-E (class Ib) MHC molecules, which aid the trophoblast in actively avoiding immune attack 22. HLA-G can interact with receptors on cytotoxic T lymphocytes and natural killer cells and inhibit their ability to induce cell lysis, thus allowing the invading trophoblast to coexist with a large number of maternal immune cells in close proximity 23. Evidence is also accumulating to suggest that these decidual immune cells may play their own vital roles in ensuring the success of pregnancy. A specialized population of natural killer cells, termed decidual natural killer (dNK) cells, infiltrate the decidualized uterus prior to implantation and remain in large numbers (approximately 70% of decidual leukocytes) throughout the first trimester of pregnancy 24. dNK cells differ greatly from peripheral blood NK cells, exhibiting a CD56-bright and CD16-negative phenotype, and only a few dNK cells express the CD160-activating receptor present on the majority of peripheral NK cells 25. In line with this observation, dNK cells do not appear to be naturally cytotoxic, and may have an altogether different function in the uterine setting 25, 26.
dNK cells have been reported to be associated with endometrial vessels and glands and have the potential to affect the success of pregnancy by influencing a range of processes that are essential for implantation, but research has focused on two main processes: (a) trophoblast invasion and spiral artery remodelling; and (b) immunotolerance of the allogenic placenta and fetus 27. dNK cells express cytokines not normally produced by peripheral NK cells, including angiogenic cytokines, such as vascular endothelial growth factor (VEGF), placental growth factor (PlGF) and angiopoietin-2 25. The influence of dNK on spiral artery remodelling may begin even before the presence of endovascular trophoblast, with recent work demonstrating an infiltration of dNK and macrophages into the vascular smooth muscle cell layers surrounding the spiral arteries, whereby they act to disrupt these layers, potentially as a priming step for the following phase of trophoblast-induced spiral artery remodelling to follow 15, 28. Direct interactions of dNK with the invading trophoblast have also been observed in vitro, and have been shown to result in the production of a range of cytokines, including tumour necrosis factor-α (TNFα), interferon-γ (IFNγ) and macrophage colony stimulating factor (M-CSF), that may affect trophoblast apoptosis, invasion and remodelling 29–31. While this evidence indicates that dNK cells play a direct role in spiral artery remodelling and trophoblast invasion, only limited data are available to demonstrate how dNK cells may be involved in cases of pre-eclampsia, and alterations in the populations of dNK cells present in pre-eclampsia remain unclear, with studies in humans providing conflicting results 32–34.
Macrophages constitute approximately 20–25% of decidual leukocytes in the first trimester, although whether these numbers remain constant or decline with gestation remains controversial 32, 35–37. Decidual macrophages may play an important role in promoting immune tolerance via the production of anti-inflammatory substances, such as IL-10 and indoleamine 2,3-dioxygenase (IDO), as well as interactions with dNK cells via the production of IL-15 38–39. Trophoblast may modulate macrophages via HLA-G binding to leukocyte immunoglobulin-like receptors (LILR) on decidual macrophages and stimulating production of IL-6 and IL-10 40. Trophoblast may also interact with macrophages and other components of the immune system via the expression of Toll-like receptors, a family of pattern-recognition receptors that are central to the innate immune response. These may give the placenta itself the ability to recognize antigenic ‘danger signals’, including host-derived molecules and non-infectious agents, and subsequently lead to immune cell activation 41. Finally, macrophages appear to disrupt vascular smooth muscle in the spiral arteries prior to extravillous trophoblast invasion, and activated macrophages have been shown to inhibit trophoblast invasion, demonstrating that they may have a direct influence on spiral artery remodelling 42. Both the cytokine imbalances and the direct effects on spiral artery remodelling described have the potential to contribute to pre-eclampsia.
The risk of pre-eclampsia is greatest in primiparous women, with subsequent pregnancies from the same partner carrying a lower risk of pre-eclampsia 43–46. However, the risk of pre-eclampsia also appears to be partner-dependent. Subsequent pregnancies with new partners may raise the risk of pre-eclampsia and prolonged exposure to a partner's semen may reduce the risk of pre-eclampsia, suggesting that exposure to paternal antigens may induce a form of immunological memory 44. In contrast, in women who have had a previous pre-eclamptic pregnancy, a change of partner has been shown to reduce the risk of pre-eclampsia, potentially as a result of a more compatible combination of the polymorphic HLA-C present on the trophoblast surface and its receptors present on maternal NK cells 30, 47. Intriguingly, it has recently been reported that systemic NK cells can demonstrate immunological memory 48. If dNK cells were also capable of immunological memory, then this may provide a further potential mechanism for this partner-specific pattern of pre-eclampsia. These data strongly suggest an immunological contribution to the pathogenesis of pre-eclampsia, and that the vast number of immune cells, in particular dNK cells, present at the materno–fetal interface in the first trimester may play an important role in trophoblast invasion and spiral artery remodelling.
The low oxygen environment of the first trimester
Trophoblast that migrate down spiral arteries from the cytotrophoblast shell form plugs that completely occlude the vessel lumen, preventing maternal blood flow into the intervillous space until 10–12 weeks of pregnancy 49. Therefore, placental development and the initial growth and organogenesis of the embryo occur under low-oxygen (approximately 2–3%) conditions during this time 50. These low-oxygen conditions are physiologically normal for the placenta and embryo at this time, and have been shown to play important roles in regulating trophoblast differentiation and invasion 51. Trophoblast differentiation is regulated by the expression of redox-sensitive transcription factors, such as the hypoxia-inducible factor (HIF) family of transcription factors. HIF-1 is strongly expressed from 5 weeks of gestation in the extravillous trophoblast, cytotrophoblast and syncytiotrophoblast, and is able to regulate the expression of a large number of cytokines important for trophoblast differentiation and invasion, including TGFβ3, insulin growth factor II (IGF-II) and vascular endothelial growth factor (VEGF) 52–54 Furthermore, premature exposure of the placenta and fetus to oxygen may be detrimental to the pregnancy 55. Thus, the establishment and maintenance of first-trimester hypoxia is important for normal placental development in the first trimester.
Spiral artery blood flow
In addition to the hypoxic effects of trophoblast plugs, ultrasound data show that these plugs alter the rate and pattern of blood flow in the spiral arteries 49, 55–57. Therefore, we must also consider how these plugs may alter the blood flow rate and wall shear stress, and the role this may play in facilitating trophoblast-mediated spiral artery remodelling. Laminar shear stress inhibits endothelial cell apoptosis induced by a variety of stimuli in vitro58–61. The inhibition of endothelial cell apoptosis with shear stress has been reported to involve many mechanisms, including Fas–FasL interactions, endothelial nitric oxide synthase (eNOS) up-regulation, induction of inhibitors of apoptosis protein (IAP) 1 and 2 expression, and activation of the PI3k–Akt survival pathway 58–59, 61–63. Furthermore, the degree of inducible endothelial cell apoptosis is related to the flow rate, with higher levels of apoptosis occurring at lower flow rates in vitro61. In comparison to human umbilical vascular endothelial cells (HUVECs), trophoblast show increased survival in response to apoptotic stimuli at low shear stresses 64. Therefore, decreased blood flow in the spiral arteries during the first 10 weeks of pregnancy may reduce the resistance of endothelial cells to trophoblast-induced apoptosis, while also providing minimal physical disruption for the trophoblast–endothelial cell interactions necessary for remodelling to occur.
Endovascular trophoblast migrate down the lumen of the spiral arteries retrograde to flow, but little is known about mechanisms explaining their movement against the blood stream. Whilst a number of cytokines and growth factors are know to affect trophoblast migration 65–68, the fact that endovascular invasion only occurs within arteries, and never into veins, suggests a further potential contribution of oxygen content or haemodynamics. Flow-induced shear stress has been shown to promote the migration of macaque trophoblast retrograde to the direction of flow, but only in single cell type cultures and not in trophoblast–endothelial co-cultures 69–70.
Thus, by the end of the first trimester, placental, immune and haemodynamic factors have all played overlapping roles in regulating spiral artery remodelling. As the trophoblast plugs dissipate, maternal blood comes into direct contact with the syncytiotrophoblast for the first time. This blood flow is now able to directly affect the developing placenta and fetus, and any inadequacies in the process of spiral artery remodelling may begin to affect the pregnancy.
Syncytiotrophoblast microparticles and knots (stage II)
The outer syncytiotrophoblast layer of the placenta is a continuous multinucleated cell layer that is in direct contact with maternal blood and undergoes constant turnover throughout pregnancy, with underlying cytotrophoblast fusing into this layer, and debris including syncytiotrophoblast knots, microparticles and mononuclear trophoblast continuously being shed off into the intervillous space 71, 72. The evidence increasingly suggests that this syncytiotrophoblast debris forms a crucial link in the interaction of the placenta with the maternal immune and cardiovascular systems, and aberrations in this process have been associated with the onset of pre-eclampsia (Figure 2).
Syncytiotrophoblast turnover and trophoblast deportation
Syncytial knots are large multinucleated fragments that are shed from the syncytiotrophoblast into the maternal circulation in a process termed trophoblast deportation, and were first identified more than 100 years ago in the capillary beds of the lungs of women who had died of eclampsia 73. Older ultrastructural and enzyme histochemical studies have suggested that ageing areas of the syncytiotrophoblast are preferentially involved in fusion, indicating that fusion could be driven by areas of the syncytiotrophoblast that require replenishment 74–77. Indeed, the turnover of the syncytiotrophoblast appears to be closely linked to the regulation of cytotrophoblast proliferation, differentiation and fusion into the overlying syncytial layer 71, 78. Prior to fusion, cytotrophoblast are believed to initiate the early stages of the apoptotic cascade in order to facilitate fusion 72. Upon fusion, this cascade is arrested until the syncytiotrophoblast is shed 72, 79. A large number of factors have been implicated in the regulation of cytotrophoblast differentiation into syncytiotrophoblast, including ADAM12 (a disintegrin and metalloprotease), GCM-1 (a transcription factor), endogenous retroviral proteins, such as syncytin-1 and -2, as well as autocrine and paracrine signalling molecules, such as colony-stimulating factor-1 (CSF-1), granulocyte macrophage-colony stimulating factor (GM-CSF), epidermal growth factor (EGF), human chorionic gonadotropin (hCG) and human placental lactogen (hPL) (Table 1) 80, 81. Thus, there are a number of potential external and intrinsic factors that could affect cytotrophoblast differentiation into syncytiotrophoblast, and subsequently trophoblast deportation.
Trophoblast deportation is increased in pre-eclamptic pregnancies, but it is becoming apparent that both the volume and properties of placental debris shed into the maternal circulation may be important in maintaining a healthy pregnancy 79, 82, 83. In other cell systems the phagocytosis of apoptotic material does not generally result in inflammation and rather may promote active immunosuppression. The majority (90%) of trophoblast are shed from the placenta by apoptotic mechanisms, and interrupting the normal process of syncytiotrophoblast turnover by caspase inhibition results in a shift from predominantly apoptotic to predominantly necrotic trophoblast shedding 79. An increase in the ratio of necrotic to apoptotic material shed from the placenta may be important in the pathophysiology of pre-eclampsia, as phagocytosis of necrotic but not apoptotic trophoblast results in endothelial cell activation 84. Furthermore, following phagocytosis, endothelial cell activation may be spread to endothelial cells not directly exposed to trophoblast debris via the secretion of factors such as IL-6 84.
In addition to multinucleated syncytiotrophoblast knots, the syncytial layer releases syncytiotrophoblast microparticles (STMPs) throughout pregnancy. Microparticles are subcellular particles of 100–200 nm, produced by necrotic cells or as a result of apoptotic blebbing 85. The syncytiotrophoblast also secretes smaller particles of 20–40 nm, termed nanoparticles or exosomes, although these are likely to have different functions 86. Both of these types of particles are rich in lipid rafts and retain cell surface molecules and, potentially, some cytoplasmic contents of their cell of origin, along with fetal DNA and miRNA 5, 87. Microparticles and endosomes mediate cell–cell communication and have been shown to have roles in haemostasis, thrombosis, inflammation and angiogenesis 85, 86, 88. During pregnancy, microparticles and exosomes are continuously shed from the syncytiotrophoblast into the maternal circulation as part of the normal turnover and renewal of the syncytial surface by apoptosis 71, 89. The level of these shed STMPs increases with gestation as the placenta grows in size, but STMPs are normally cleared by phagocytes, such that they do not adversely affect the maternal system to a significant degree in normal pregnancy 5. Exosomes from dendritic and other cells express MHC class I and II and present antigens to T cells in a manner that can be either stimulatory or tolerogenic, depending on the conditions 86. STMPs may also play an immunosuppressive role in normal pregnancy, with STMPs from normal but not pre-eclamptic placentae inhibiting the production of IFNγ by monocytes 90. The presence of miRNA in trophoblast exosomes is also intriguing, as the ability of these particles to modulate gene expression at the posttranscriptional level has recently been implicated in the pathogenesis of several cardiovascular disorders, including hypertension 91. Thus, the presence of STMPs may play an important role in the homeostasis of normal pregnancy and placental interactions with the immune and cardiovascular systems.
If shedding of placental debris is a part of normal pregnancy throughout gestation, how could this play a role in the pathogenesis of pre-eclampsia? The number of microparticles released from the placenta is significantly greater in pre-eclampsia, in particular in early-onset cases, and early findings indicate that STMPs from pre-eclamptic patients differ in quality from those from normal pregnancies 5, 92. In vitro studies have shown that STMPs affect both endothelial cell and lymphocyte proliferation 93. STMPs are able to bind in vivo to circulating monocytes and stimulate production of the inflammatory cytokines TNFα and interleukin (IL)-12 and -18, with the production of these cytokines significantly increased in STMPs from pre-eclamptic patients 90, 94. Thus, while a progressive increase in serum inflammatory profile is a physiologically normal feature with increasing gestation, the exacerbation of this response in pre-eclampsia is likely to further contribute to the hypertensive effects of this condition, as increasing evidence suggests that hypertension-associated vascular disease is an inflammatory process 95. In order to understand the processes that may result in altered STMP release and subsequent activation of the maternal endothelium in pre-eclampsia, we must once again return to the key process of spiral artery remodelling.
The onset of pre-eclampsia
Spiral artery remodelling can affect the delivery of blood to the placenta in two ways: by affecting the flow rate of blood into the intervillous space; and by affecting the consistency of this flow, and thus the delivery of a steady supply of oxygen. In cases of inadequate spiral artery remodelling, insufficiencies can be found in both the extent of arterial remodelling and the number of converted spiral arteries (reducing the diameter of the artery openings, and hence the rate of blood flow into the intervillous space) and/or the depth of trophoblast invasion, such that the junctional zone of the spiral arteries is not modified and retains its contractile potential (resulting in potential fluctuations in oxygen delivery to the placenta).
Blood flow in the intervillous space
The association between spiral artery remodelling and pre-eclampsia was previously thought to be due to inadequate conversion of the spiral arteries, resulting in an insufficient blood supply to the developing placenta and fetus, leading to both nutritional and oxygen deficiencies. However, in recent years closer investigations of the mechanisms regulating uterine blood flow, and the effect of flow characteristics in the intervillous space, have given us a more accurate idea of the role these remodelled arteries may play in successful pregnancies.
It is becoming evident that remodelled spiral arteries have a significant effect on pregnancy, not so much by increasing the volume of blood perfusing the placenta, but the manner in which this blood is delivered 96. The wide-bore openings of the spiral arteries into the intervillous space may serve an important role in reducing the velocity of blood flowing into the intervillous space. This would be advantageous for several reasons:
1 Blood entering the intervillous space at an increased velocity or with a turbulent flow pattern may affect the developing villous architecture, resulting in the formation of intervillous lakes that are often found lined with thrombotic material 96. In the most severe cases this may result in gross morphological changes as a result of the chorionic plate being pushed up by jet-like streams that can be detected by ultrasound from 14–15 weeks in some pregnancies 97. These jet-like streams may also create channels in the villous structure, through which blood could flow preferentially by the path of least resistance, thus decreasing perfusion throughout the entire intervillous space.
2 To maximize diffusional exchange between the maternal and fetal circulations, a thin barrier needs to be maintained. This requires the transmembrane pressure differential to be positive in a fetal-to-maternal direction, such that the pressure in the intervillous space is less than that in the fetal capillaries, to prevent compression of the latter 98. High-velocity blood flow would negatively affect this pressure differential.
3 Excessive shear stresses resulting from increased or turbulent blood velocity in the intervillous space have the potential to create morphological damage and increased shedding of the syncytiotrophoblast, which may contribute to the maternal endothelial cell activation featured in pre-eclampsia 99, 100.
Whilst trophoblast invasion and subsequent spiral artery remodelling reaches the inner one-third of the myometrium in normal pregnancy, it does not reach the radial or arcuate arteries, and it is this portion of the artery that provides the rate-limiting step for blood flow 17, 101. However, these arteries have been shown to undergo significant dilation underneath the implantation site in pregnancy 101, 102. This dilation is believed to arise from a combination of endocrine stimulation (including placenta-derived factors, such as PlGF and VEGF) and nitric oxide-mediated dilation 103, 104, and results in an increase in flow from approximately 45 ml/min in the non-pregnant uterus to 750 ml/min at term 105, 106. Indeed, sonographic evidence demonstrates that arterial diameter increases with arterial branching through the radial, arcuate and finally spiral arteries that increase in diameter approximately four-fold in pregnancy to reach a average size of 2.4 mm where they open into the intervillous space 102. Thus, by mid-pregnancy the narrower uterine artery forms the rate-limiting step for blood flow to the intervillous space 96. This has implications for the traditional model of blood flow to the intervillous space being limited by inadequate spiral artery remodelling, and further suggests that the progressively increasing arterial diameters act to slow the velocity of blood and result in uniform perfusion of the intervillous space.
Placental hypoxia–reperfusion in pre-eclampsia
Oxidative stress has for some time been postulated as the link between the early and late stages of the pathophysiology of pre-eclampsia 107. The dogma has been that this oxidative stress resulted from prolonged placental hypoxia (approximately 6% oxygen) as a result of an insufficient volume of blood perfusing the intervillous space, due to inadequate spiral artery remodelling. However, it is becoming apparent that, rather than consistently low oxygen levels, inadequate spiral artery remodelling may lead to intermittent perfusion of the intervillous space and fluctuating oxygen levels 4. Spiral artery conversion to the inner one-third of the myometrium may be essential to remove the smooth muscle from the highly contractile segment of the spiral artery in the junctional zone, thus reducing spontaneous vasoconstriction, which may lead to intermittent placental perfusion 96.
Oxidative stress arising from such hypoxic reoxygenation injuries results in widespread placental lipid and protein oxidative modifications, mitochondrial and endoplasmic reticulum stress and tissue apoptosis and necrosis 4, 108. Such conditions can induce HIF-1-mediated degradation of the transcription factor GCM1, which is down-regulated in pre-eclamptic placentae and plays an important role in syncytiotrophoblast formation and turnover via its downstream regulation of syncytin and PlGF 109–111. Along with an increase in the volume of debris shed from the placenta, hypoxic-reperfusion studies have been shown to favour necrotic rather than apoptotic death 78, 108. The haemodynamic changes that result in oxidative stress may also have mechanical implications for the syncytiotrophoblast, as increased or turbulent velocities can increase the volume and alter the profile of the shed debris 99. Indeed, the ability of STMPs to generate a pro-inflammatory response appears to be dependent on their mode of preparation, with STMPs derived from perfused placental lobules, as opposed to mechanical stimulation, demonstrating an immunosuppressive effect via the suppression of T cell responses 94. Thus, the quality, rather than the absolute quantity, of placental perfusion may be more important in the pathophysiology of pre-eclampsia.
The placenta expresses a number of antioxidants, including glutathione peroxidase, catalase and superoxide dismutase, which would act to reduce the oxidative stress generated by inadequate placental perfusion by mopping up oxygen free radicals 112–113. However, data concerning the expression and activity of these enzymes in pre-eclampsia are conflicting; thus, it remains unclear whether decreased levels of antioxidants predispose the placenta to oxidative stress, or increased levels of antioxidants are observed as a response to the oxidative stress to which pre-eclamptic placentae are exposed 114–116. Surprisingly, the large multicentre vitamins in pre-eclampsia (VIP) trial demonstrated no protective effect of vitamin C + E supplementation from the second trimester of pregnancy, and conversely this supplementation was shown to be associated with lower-birthweight babies 117.
Soluble angiogenic factors
In addition to the release of syncytiotrophoblast knots and microparticles, a number of circulating factors that contribute to endothelial dysfunction in pre-eclampsia have been identified, which are believed to play synergistic but different roles in the pathogenesis of pre-eclampsia. These include cytokines such as TNFα, IL-6, IL-1α, IL-1β, apoptotic factors such as Fas-ligand, oxidized lipid products, neurokinin B and asymmetric dimethylarginine and anti-angiogenic factors, such as soluble fms-like tyrosine kinase-1 (sFlt-1) and endoglin 118–123.
sFlt-1 is an alternative splice variant of VEGF-1 receptor that lacks the transmembrane and cytoplasmic portions of the protein 124. It is therefore able to act as an antagonist for both VEGF and PlGF and diminish their binding to the VEGF-1 and -2 receptors 124. Indeed, in cases of pre-eclampsia an increase in sFlt-1 is associated with a concordant decrease in both free VEGF and PlGF in the serum 125. sFlt-1 expression is regulated by oxygen, with increased expression in both in vitro and in vivo models of placental hypoxia 125–128. Endoglin is a co-receptor for transforming growth factor β1 and 3 (TGF-β1 and 3) and acts to regulate vascular tone via interactions with eNOS 129. The extracellular domain of endoglin can be shed as soluble endoglin (sEng), which is thought to impair TGFβ1 binding to cell surface receptors and decrease endothelial nitric oxide signalling, thus inhibiting angiogenesis and promoting vascular dysfunction 130–131. sFlt-1 is significantly increased in maternal serum approximately 5 weeks prior to the onset of pre-eclampsia, whereas sEng is increased slightly earlier, 2–3 months prior to the onset of pre-eclampsia 125, 132–136.
Other markers of pre-eclampsia have also been identified, including P-selectin, ADAM12, placental protein 13 (PP13), pentraxin 3 (Ptx3) and pregnancy-associated placental protein A (PAPP-A) (for review, see 137). The function of PP13 has not been fully discerned, but it is produced by the trophoblast in increasing amounts throughout gestation. PP13 is able to be detected from early in the first trimester, and has been reported to be lower in women who subsequently develop pre-eclampsia, in particular early-onset pre-eclampsia 138–140. Ptx3 is an inflammatory molecule produced by the placenta (among other tissues) that interacts with growth factors and extracellular matrix components, activates the complement system and facilitates the recognition of pathogens by phagocytes 141. Levels of Ptx3 are increased in all three trimesters in pregnancies that go on to develop pre-eclampsia or IUGR 142, 143. Whilst many of these markers provide hope for early clinical diagnosis, due to the heterogeneity of pre-eclampsia it is unlikely that any single measure in isolation will provide a test accurate enough to be used on a routine diagnostic basis, and it appears that the use of a ratio between factors will provide the most accurate predictive measure 144–146.
The final step in the chain of events that lead to pre-eclampsia is the ability of the maternal system to handle the deficits in placentation, and subsequent challenge to the maternal cardiovascular system. In addition to its potential role in spiral artery remodelling in the first trimester, the immune system plays an important role in the pathogenesis of pre-eclampsia in the third trimester, as systemic inflammatory stress plays a key role in endothelial cell activation. Systemic inflammation usually occurs in the vasculature as a response to injury, lipid peroxidation or infection 147. However, in pregnancy the presence of STMPs may also stimulate systemic inflammation. This results in the activation of leukocytes, which mediate the adherence of neutrophils to the endothelium. Following activation and binding, neutrophils transmigrate through the vascular wall and release cytokines that mediate vasoconstriction and vascular damage, including TNFα, IL-1 and IL-8 148. Women with pre-eclampsia demonstrate increased neutrophil activation, and increased neutrophil adhesion to endothelial cells 148, 149. The level of intracellular reactive oxygen species in leukocytes is also significantly increased during normal pregnancy, and is further increased in women with pre-eclampsia 150. Finally, maternal infections, such as urinary tract infections, Chlamydia pneumoniae or cytomegalovirus, may increase the total infectious burden and have been linked to an increased risk of pre-eclampsia 151, 152. Therefore, the underlying inflammatory state of the maternal immune and cardiovascular systems prior to the increase in inflammatory markers associated with pregnancy may be extremely important in the pathogenesis of pre-eclampsia.
Some genetic links to pre-eclampsia have been established 153, 154, but it is possible that the inherited susceptibilities may relate more to the ‘fitness’ of inherited maternal endothelial cell function. Indeed, pre-disposing risk factors all converge on widespread maternal endothelial dysfunction, with an increased risk of pre-eclampsia observed in women with pre-existing hypertension, renal disease, obesity and dyslipidaemia 155. Pre-eclampsia itself also appears to predispose women to other cardiovascular disorders later in life 156, 157.
Early-onset pre-eclampsia, late-onset pre-eclampsia and IUGR
Until this point, the focus of this review has been solely on pre-eclampsia. However, in up to one-third of cases, pre-eclampsia is associated with intrauterine growth restriction (IUGR)—the failure of the fetus to reach its optimal growth potential 158, 159. IUGR is also seen in 8–14% of normotensive pregnancies 159, 160. Pre-eclampsia and IUGR were for a long time considered to result from the same underlying pathophysiology. However, as our understanding of both of these conditions increases, it is becoming apparent that, while they may result from aberrations in many of the same pathways, there appear to be specific differences in the pathophysiology of the two conditions.
Pre-eclampsia and IUGR may result from specific differences in the same broad factors that contribute to the success of spiral artery remodelling and implantation in the first trimester. Cases of pre-eclampsia and IUGR both exhibit differences in the populations of immune cells present in the decidua, with pre-eclamptic placental bed biopsies demonstrating a reduction in all lymphocyte populations (CD3+ T lymphocytes, CD8+ T lymphocytes, CD14+ macrophages and CD56+ dNK cells), whereas biopsies from cases of IUGR showed only a reduction in CD56+ NK cells 32. This suggests that differences in the local cytokine balance may be important in the pathogenesis of each of these pregnancy disorders 32. In addition, differences in the differentiation of cytotrophoblast into syncytiotrophoblast and the secretion of human placental lactogen (hPL) and human chorionic gonadotropin (hCG) have been reported between cases of IUGR, pre-eclampsia with IUGR and pre-eclampsia alone 161. Finally, uterine artery Doppler sensitivity for small-for-gestational-age babies with pre-eclampsia is superior to that for IUGR alone, suggesting some differences in the underlying placental abnormalities in the first trimester 160, 162.
As discussed previously, increased levels of STMP shed into the maternal circulation are believed to be a major stimulus for the systemic inflammatory response and endothelial cell activation observed in pre-eclampsia. However, this increase in STMP shedding is not seen in cases of IUGR, and IUGR alone is also not associated with the major inflammatory and cardiovascular responses characteristic of pre-eclampsia 163, 164. In line with these findings, the anti-angiogenic factors sFlt-1, sEng and endostatin have been reported to be increased in pre-eclamptic, but not normotensive, IUGR pregnancies 135, 165, 167. However, findings in this field can be inconsistent with others finding an increase in sFlt-1 and endoglin in early severe IUGR 168–170, and more work is needed to clarify differences between pre-eclampsia and IUGR. These findings have important ramifications for the pathophysiology of IUGR, as they indicate that, while many of the aberrations in the first trimester of pregnancy appear to be broadly similar, in IUGR these factors do not converge at the level of hypoxia–reoxygenation injury and the resulting syncytiotrophoblast debris that appears to be the key trigger of pre-eclampsia (Figure 2).
If IUGR pregnancies do not share the characteristics of pre-eclampsia that activate the maternal endothelium, what alternative placental mechanisms may underlie this condition? It is possible that one key feature of IUGR alone may be an inability of the placenta to transport nutrients to the fetus successfully. Findings demonstrating compromised feto-placental angiogenesis, and subsequent villous development resulting in a reduced villous area as well as a reduction in system A amino acid transporter activity, resulting in reduced placental amino acid uptake, support this hypothesis 171–174. It is also interesting to speculate that, whilst incomplete conversion of the myometrial segments of the spiral arteries may result in hypoxia–reperfusion-type injuries and lead to pre-eclampsia, inadequacies in the response of the deeper radial and arcuate arteries may produce the more traditional scenario, where blood supply to the intervillous space is steady but insufficient, resulting in IUGR. These inadequacies are not necessarily mutually exclusive events, and could both be present in cases of pre-eclampsia associated with IUGR, and increased uterine artery resistance (as a result of inadequate dilation of the radial or arcuate arteries) would be expected to be observed in both cases. It is interesting that increased uterine artery resistance is associated with IUGR and early-onset pre-eclampsia, but not in late-onset pre-eclampsia, which is less often associated with fetal growth restriction 162.
As it becomes evident that there are differences in the pathophysiology of pre-eclampsia and IUGR, attention has also been drawn to the heterogeneity of pre-eclampsia itself, and the reasons for early (<34 weeks) or late presentation of the disease, often with significant consequences for its severity. Late-onset pre-eclampsia comprises approximately 80% of pre-eclampsia cases worldwide, which typically exhibit normal placental morphology and are not associated with growth restriction or altered umbilical artery Doppler profiles 175–177. In contrast, early-onset pre-eclampsia, while comprising a smaller number of cases, tends to be more severe. Early-onset pre-eclampsia is associated with abnormal placental villous and vascular morphology, including volume of the intervillous space, terminal villous volumes and surface areas of the terminal villi 176.
The differences in the aetiology and pathophysiology between early and late-onset pre-eclampsia remain unclear. Early-onset pre-eclampsia is often associated with inadequate and incomplete spiral artery remodelling, alterations in uterine Doppler artery profiles and clear signs of fetal growth restriction 175. There is also growing evidence of differences between early- and late-onset pre-eclampsia in placental gene expression and in the balance of factors involved in maternal endothelial function 122, 153, 178–181.
As we try to unravel the pathophysiology of pre-eclampsia, it is important not to just consider the various components that contribute to the disease state individually, but to be aware that this is a condition that involves numerous and constant interactions between the placental, immunological and cardiovascular systems, and as such can not be treated as a disorder of any one of these systems in isolation. The volume of literature and understanding of pre-eclampsia has exploded in the past 25 years, and considerable amounts of work have gone into understanding how the various elements of pre-eclampsia contribute to the disease. Hopefully, understanding how the pieces of pre-eclampsia fit together as a whole will enable us to further understand this syndrome, and continue to look for early markers of the disease and potential treatments.
This work was supported by grants from the New Zealand Foundation of Science and Technology (to JLJ).