Introduction

It has been known for a long time that the occurrence of preeclampsia is somehow linked to the presence of a placenta [1]. During placentation the mother comes in close contact with semi-allogeneic fetal trophoblastic cells which play a key role in maternal–fetal physiological exchange. Because of the reduced number of layers separating the two circulations, the most intimate association between mother and fetus occurs in species with hemochorial placentation, in which fetal trophoblast is directly exposed to circulating maternal blood. In contrast to other placental types, hemochorial placentation is always associated with decidualization of the endometrium, which involves an ‘epithelioid’ transformation of the fibroblasts of the uterine mucosa [2], accompanied by extracellular matrix changes and infiltration by other cell types, notably uterine natural killer cells and macrophages. Multiple functions have been ascribed to the decidua, including the secretion of hormones and growth factors to allow embryo implantation and placental outgrowth in an orderly fashion, but at the same time also protecting the uterus against excessive damage [3]. Morphologically, various degrees of decidualization have been discerned in different species, notably in primates where a more elaborate decidua seems to be associated with deeper trophoblast invasion beyond the decidualized endometrium, as typically occurs in the human [4].

Decidualization not only involves the endometrial stroma, but also the spiral arteries which undergo a marked increase in length from the late luteal phase of the cycle onwards, leading to their spiral course [5]. After an early phase of ‘plugging’ by intraluminal trophoblast, spiral arteries undergo a retrograde (‘antidromic’) invasion by these cells, which are also largely responsible for the subsequent vascular remodeling (Chapter 11). Variations may occur in different species as to the involvement of endovascular and interstitial trophoblast, as well as to the depth of invasion which can either be restricted to the decidua or extended into the inner myometrium (Chapter 13). ‘Physiologically changed’ spiral arteries have undergone a loss of the vascular smooth muscle and the elastic lamina, resulting in a significant increase in vascular diameter to accommodate the increasing uteroplacental blood flow. Soon after the discovery of the characteristic spiral artery remodeling in the pregnant uterus, it was postulated that these changes resulted from a destructive action by invading trophoblast on the vessel wall [6] (see also Chapter 2). This initial emphasis on a primordial role of trophoblast in the process was subsequently toned down by the observation – in the human as well as in laboratory animals – that decidua-associated arterial changes, probably occurring under hormonal control, precede trophoblast invasion [7,8]. This seemed to imply that there exists an early preparatory phase which is essential for subsequent trophoblast-associated remodeling. Since in the human the trophoblast invades as deeply as the inner ‘junctional zone’ (JZ) myometrium, which is increasingly considered as a separate uterine compartment (see Chapter 9), the occurrence of a priming ‘decidua-associated’ vascular remodeling has to be considered also in this uterine compartment.

In 1972 Brosens and colleagues [9] showed for the first time that in preeclampsia the ‘physiological changes’ are restricted to the decidua. These observations were mainly based on studies of third trimester placental bed biopsies collected during cesarean sections, and since no data were available yet about an early ‘decidualization’ phase of vascular remodeling, it was logical to propose a failure of trophoblast invasion as a primary cause of the vascular defects. At that time little was known about the relationship between early invasive processes and vascular change during the first trimester. Although swelling of vascular smooth muscle in the decidual spiral arteries had already been described [10], an event obviously related to the development of perivascular decidual sheaths (‘Streeter’s columns’), nothing was known about the inner myometrium in early pregnancy, which was to be the site of the major vascular defects in preeclampsia. Subsequent histological studies of the junctional zone myometrial segments of spiral arteries in intact pregnant hysterectomy specimens of the first and early second trimester resulted in two major findings [11]. First, vascular changes including disorganization of the muscular wall are not exclusively due to the presence of trophoblast. Vascular smooth muscle becomes disorganized before the arrival of endovascular trophoblast and this process is enhanced in the presence of interstitial trophoblast. A second finding was the apparent occurrence of endovascular invasion in the inner myometrium as a second ‘wave’ occurring after a 4-week period of relative stability in the decidua. In later years this ‘two waves concept’ has been criticized, mainly following studies of large numbers of placental bed biopsies taken in early pregnancy [12,13]. Although the two waves concept is not definitely agreed upon, the point is relevant for considering possible mechanisms of failed invasion. Indeed, the occurrence of a distinct second wave might imply a specific mechanism for triggering deeper invasion, which might be disturbed in preeclampsia.

From the previous it follows that a failure of spiral artery remodeling might find its origin either in the fetal trophoblast or in the maternal environment (the decidualized endometrium and JZ myometrium, including the spiral arteries), or maybe even in both. In addition to these local uterine phenomena wider pathophysiological disturbances should also be considered, which may have an impact upon these local vascular defects.

Arguments for trophoblast defects

Focusing on spiral artery invasion, there is little direct evidence for the occurrence of intrinsic trophoblastic defects in early pregnancy. The only relevant observation so far might be the absence of endovascular trophoblast in myometrial spiral arteries in one out of seven post-15 weeks specimens in a study of early pregnant uteri [11]. While this finding is obviously very intriguing, it is of course pointless to guess whether or not preeclampsia would have occurred if the pregnancy had been allowed to continue.

Arguments for intrinsic defects in acquiring an ‘invasive phenotype’ during extravillous trophoblast differentiation were put forward during the high days of integrin research. A key observation was the discovery of a spatial gradient of shifting integrin expression within the cytotrophoblastic cell columns of anchoring villi, as revealed by immunohistochemistry on tissue samples of early abortions [14,15]. This integrin shift must result in an altered binding capacity to different extracellular matrix components, thought to be necessary for allowing trophoblast migration. This integrin shift could still be observed in late second trimester normal pregnancies, but was absent in preeclamptic women [16], thus indicating a possible cause for failed invasion. Although the idea was very attractive, no differences in integrin expression were observed in extravillous trophoblasts in the basal plate and associated decidua in third trimester placentae of normal and complicated pregnancies [17]. This observation does not refute the postulated role of a failed integrin shift in preeclampsia, however, since at this late stage of pregnancy all invasive activity may have stopped at the basal plate and associated decidua. Another problem with the defective integrin shift hypothesis is that mainly interstitial trophoblasts were evaluated in the quoted studies. While a defective endovascular invasion was implied by the restricted spiral artery remodeling, the question as to whether defective interstitial invasion might also be involved in preeclampsia has never been convincingly answered [18,19]. Although admittedly anecdotal, placental bed biopsies of even severe preeclamptic women may sometimes show intense interstitial invasion of the inner myometrium. This should not be surprising, since interstitial trophoblast numbers vary considerably throughout the whole extent of the placental bed, not only in preeclamptic women but also in normal pregnancies [18]. Recent evidence for defects in interstitial trophoblast in hypertensive pregnancies is the finding of an inadequate fusion of invading cytotrophoblast into multinuclear giant cells in both gestational hypertension and preeclampsia, associated with a maintenance of E-cadherin expression [20]. This finding is in agreement with a report of a failed downregulation of E-cadherin in preeclampsia [16], but contradicts previous claims of increased fusion into multinuclear giant cells in this condition [21]. Whether alterations in E-cadherin expression may result from an intrinsic trophoblast defect, or rather are induced by maternal factors is not known.

Impaired trophoblast invasion in spiral arteries may not only be due to an intrinsic defect in invasive properties, but may also be induced by maternal cells. Inflammatory cells are inevitably present within invaded areas of the placental bed, and an aggravated maternal response might well lead to an ‘overkill’ of trophoblasts. Such events might be related to the concept that normal – and a fortiori preeclamptic – pregnancies represent a hyperinflammatory state [22]. Although not regularly seen in near-term placental bed biopsies, physiologically remodeled spiral arteries occasionally show extensive leukocytic infiltrations. It is possible, however, that invasion-related acute maternal inflammatory responses mainly occur in earlier stages of pregnancy when biopsies are not routinely taken, and this may account for the rarity of such observations.

Arguments for maternal defects

For a long time trophoblast invasion was thought to be controlled by a restrictive action of the decidua [3, 23]. One proposed mechanism, emerging from rodent studies, was the presence of a mechanical barrier, possibly effected by the tight intercellular junctions joining decidual stromal cells [2]. Such restriction would in the first place apply to interstitial trophoblasts which are directly in contact with the decidualized endometrial stroma. In the human, however, it is obvious that the decidua does not really act as a barrier but rather as a passage-way for trophoblasts to colonize the junctional zone myometrium, as witnessed by the tremendous numbers of interstitial trophoblasts appearing in this compartment during the first trimester [24]. Indeed, more recent evidence indicated that the human decidua may actually stimulate the invasive behavior of the trophoblasts by inducing their synthesis of matrix metalloproteinases (MMPs) [25]. This idea was already implicit in the comparative study by Ramsey and colleagues [4] who tried to find an explanation for the obviously higher degree of decidualization in the human as compared to other primate species with less deep trophoblast invasion. It is therefore conceivable that defective decidual function may be a possible reason for impaired trophoblast invasion in complicated pregnancies, although, as far as the interstitial invasion is concerned, the available (contradictory) evidence needs to be substantiated [18,19].

A stimulatory role of decidua for trophoblast invasion may also apply to the endovascular invasion of the spiral arteries. First, it is not inconceivable that the ‘decidualized’ vascular smooth muscle may also induce MMP production by the migrating endovascular trophoblasts and thus stimulate their invasion or incorporation into the vessel wall. Second, since the decidualization process of the spiral arteries implies a loss in the coherence of the vascular smooth muscle, this early vascular change might allow easier intramural penetration by the trophoblast. This loss in coherency of the smooth muscle must be related to alterations in the extracellular matrix, and such matrix changes have been reported for both the decidual stroma and the decidual segments of spiral arteries [26]. Decidua-associated vascular remodeling not only occurs in the decidua but also in the ‘junctional zone’ myometrial compartment [27]. Is there any evidence for a disturbed decidua-associated remodeling, either in the decidua or in the JZ myometrium, in complicated pregnancies? Unfortunately not, mainly because of the impossibility to routinely collect placental bed samples in the early stages of an ongoing pregnancy. There was certainly no morphological evidence of failed disorganization – i.e. maintenance of a tight vascular smooth muscle coherence – in myometrial spiral arteries of the one non-invaded post-15 weeks specimen in the previously quoted study [11].

Because of the disorganization of the vascular smooth muscle and subsequent vasodilatation, decidua-associated vascular remodeling may enhance blood flow to the implantation site. Doppler studies at the time of embryo replacement after IVF revealed an increased vascularity of the endometrium in conception cycles [28], which may result from angiogenic processes associated with early decidua-associated remodeling. It is not yet known to what degree defects in early endometrial and junctional zone vascularity are responsible for pregnancy complications. Disturbances in uterine arterial blood flow have been demonstrated as early as the twelfth week [29]. A marked increase in uteroplacental oxygenation has been observed after 12 weeks in normal pregnancies [30], i.e. before the onset of the alleged second wave of endovascular trophoblast invasion into the inner (junctional zone) myometrium. The question as to what comes first, increased blood flow or trophoblast invasion, has not yet been fully resolved. The most favored scenario still is that by their invasive action, trophoblasts open up the vessels, destroy the vascular smooth muscle, and transform the arteries into permanently dilated tubes, thus increasing maternal blood flow to the placenta. Taking a different point of view, one might envisage that, besides the invasion of the spiral arteries, the steroid-controlled rise in blood supply to the pregnant uterus also has to be taken into account [31]. Hemodynamics may indeed be a real, albeit imperfectly understood, factor directing trophoblast migration [32]. It is not unlikely that an inadequate rise in uteroplacental blood flow, which may result from various disturbances, is the real cause of failed trophoblast invasion and spiral artery remodeling in preeclampsia (Chapter 11).

Preeclampsia as a failure in the maternal–fetal dialogue

In the past the placental bed was frequently compared with a battle-field where invading trophoblasts are countered by a defense line of maternal decidua [33] (Chapter 16). This possible scenario appealed to many investigators who were looking for possible causes of the uniquely human disease preeclampsia, fueling the idea that the mother fights back against the threat of deeply invading trophoblast. It was silently assumed that deep trophoblast invasion was an exclusive feature of human pregnancy, and that there was no need to consider a maternal ‘rejection’ in other primate species which undergo only shallow invasion. However, a recent study of specimens from historical tissue collections revealed that deep invasion does occur in chimpanzees, and also in gorillas (Chapter 12). Interestingly, for both species case reports of suspected (pre-)eclampsia have been published [34,35,36]. Of course most pregnancies do not become preeclamptic, so that as a rule deep trophoblast invasion is well tolerated, without any ensuing ‘battle’.

We have previously argued that a completely different concept of interaction may better reflect the reality, namely a concept of a mutual maternal–fetal support or ‘dialogue’ between uterus and trophoblast, as a result of an intricate coevolutionary process [37]. Although coevolution usually occurs at the level of interspecies interactions, the process is obviously also applicable to interactions between males and females, or between mothers and their offspring [38]. This could be considered as a ‘Red Queen’ scenario, in which both runners (mother and fetus) have to move as fast as possible in order to keep themselves ‘at the same place’. Such stepwise coevolution between increasing trophoblast invasion and uterine adaptive (decidual) changes may have resulted in a progressively deeper invasion in the course of our evolution, thus setting a compromise in the inherent conflict between fetal nutrient requirements and maternal self-protection. Both partners then reap the benefit of optimizing their reproductive chances [39] (Chapter 16).

At this point we have to ask which benefits may be gained from the deeper trophoblast invasion and the associated deep vascular remodeling. At first sight there is no reason to consider the placenta of a baboon, which shows limited trophoblast invasion, to be less efficient than the human placenta. One possibility is that deep invasion may compensate for possible vascular disturbances due to our upright position [40]. Indeed, it has been reasoned that bipedalism may carry a risk of compressing the vena cava which may compromise uteroplacental blood flow. Since chimpanzees, which are basically knuckle-walkers, also show deep trophoblast invasion (Chapter 12), the upright position of the human has probably not been a major factor in the evolution of deep placentation. Another popular idea is that deep placentation provides a better support for a more extensive fetal brain development, and for that reason deep placentation used to be considered as being unique to humans. Also chimpanzee fetuses show considerable brain development [41] and this might also be related to their deep placentation. A possible link between fetal brain development, deep trophoblast invasion, and preeclampsia has been suggested [42]. Indeed, the extended reciprocal exposure of maternal and fetal cells must carry an increased risk for pregnancy complications such as preeclampsia due to the inevitable discordancies – genetic or other – between mother and fetus. It has even been proposed that accelerated fetal brain development in Neanderthalers, who had larger brains than the present Homo sapiens, must have been associated with more extensive trophoblast invasion and an associated higher risk of preeclampsia, contributing to their extinction [43]. More data are required, however, to support the proposed link between deep trophoblast invasion and increased fetal brain development, but also between deep invasion and placental efficiency. Following all these considerations, babies born after preeclamptic pregnancies must have been deprived of the benefits associated with deep placentation, and especially the long-term consequences of the disease should be a matter of concern.

Conclusion: defective remodeling as a disturbed partnership

This chapter provides a general overview of the two major components of pregnancy-associated spiral artery remodeling, the maternal (decidual) and the fetal (trophoblastic). A full understanding of the vascular remodeling process necessitates detailed knowledge of the fate of each arterial component. Based upon observations in early pregnancy specimens, we distinguished four major steps in arterial remodeling: (1) decidua-associated remodeling; (2) the intraluminal phase of trophoblast invasion; (3) intramural trophoblast incorporation (trophoblast-associated remodeling); and (4) a partial maternal repair [44].

Defects in each step may interfere with vascular remodeling and be associated with the whole spectrum of pregnancy complications. So far, most attention has been paid to the enigmatic disease preeclampsia, which is no longer considered as one specific disease, but rather as a syndrome which may have several underlying causes. The fact that different interfering steps may lead to failed spiral artery remodeling can be an additional reason for this multicausality. There is now general agreement that maternal constitutional factors play a role in the development of the preeclamptic syndrome [45]. Although such factors are usually considered in the light of the symptomatic ‘second stage’ of the disease, they might also have an impact on early remodeling steps in spiral arteries. It is known for instance that decidualization is impaired in diabetic rats [46] and diabetic NOD mice show underdevelopment of decidual spiral arteries [47]. It is not yet known whether in diabetic women the decidua-associated remodeling may be disturbed, but diabetes is a well-known risk factor for preeclampsia [48]. Other maternal physiological maladaptations to pregnancy such as a failure in plasma volume expansion or a failure in the redirection of maternal blood flow to the expanding uterus might have a direct negative effect on intraluminal trophoblast migration. Similar considerations may apply to other health issues, including fertility problems which may be due to a suboptimal uterine preparation for implantation and placental development, and which may contribute to poor pregnancy outcome after assisted reproduction (Chapter 20).

In order to unravel the possible causes of defective spiral artery remodeling it will be necessary to focus on early pregnancy, where the key of failed invasion and/or associated or preceding vascular defects has to be found. The big problem remains the impossibility of acquiring relevant study material of early placentation stages in ongoing pregnancies, and this problem is not likely to be solved in the foreseeable future.