Preconceptional and Periconceptional Pathways to Preeclampsia

Introduction

As a role for the corpus luteal (CL) product, relaxin (RLN), began to figure prominently in the maternal cardiovascular and renal changes during pregnancy in the gravid rat model, the obvious question arose about whether CL secretory product(s) might play a similar role in gravid women. This possibility was foreshadowed by the elegant work of Chapman and colleagues, Bernstein and coworkers, as well as other investigators, which demonstrated that the maternal hemodynamic changes in human pregnancy are anticipated, albeit to a lesser degree, in the luteal phase of the menstrual cycle ^, (and see citations in Conrad et al. ). Because maternal vasodilation of varying degrees beginning in early pregnancy (and even before pregnancy) underpins maternal cardiovascular and renal gestational changes, ^, a related question was whether CL secretory product(s) such as RLN might also reduce the risk of developing preeclampsia (PE), which at the peak of disease is a relative state of vasoconstriction or of inadequate vasodilation ( Chapter 11 ).

However, there were logistical challenges to addressing these two overarching questions. First, whereas a role for RLN in the maternal hemodynamic changes of rat pregnancy was uncovered by using RLN neutralizing antibodies or by removing the ovaries, the source of circulating RLN, ^, these manipulations could obviously not be implemented in women. Second, although women develop PE spontaneously, rats do not; therefore, to address the second question related to PE protection possibly conferred by CL product(s), only women could be investigated. Therefore, in order to address these two questions, investigation of women who planned to conceive by in vitro fertilization (IVF) seemed to be the only available option. Depending on the IVF protocol, women conceive without a CL, one CL, or multiple CL. In addition, women who conceive spontaneously with singleton pregnancies usually have one CL. Thus, gestational hemodynamic changes and PE risk could be prospectively investigated during pregnancy in the context of a “dose response” of CL number and hence of circulating CL factors including RLN.

Using this “experimental” approach afforded by IVF, an overarching hypothesis was investigated: CL factors such as RLN either mediate or contribute to the maternal hemodynamic changes of early human pregnancy (“developmental phase”), analogous to the gravid rat, while after the CL-placental shift, placental vasodilatory hormones such as placental growth factor (PlGF) supersede (“maintenance phase”; Fig. 4.1 ). Thus, according to this hypothesis, women who conceived using an IVF protocol that precluded the development of a CL might have a “hypodynamic” circulation due to the absence of circulating RLN and/or other CL factor(s), while women who conceived using an IVF protocol that led to the development of multiple CL might have a “hyperdynamic” circulation due to excess circulating RLN and/or other CL factor(s) (at least during the first trimester). As a corollary, in women who conceived without a CL, the maternal circulation should be restored to normal levels after the first trimester as a consequence of circulating placental hormones, e.g., PlGF, that intercede ( Fig. 4.1 ). One caveat to this overall experimental strategy, however, is that a role for any specific CL product(s) such as RLN cannot be discerned without the implementation of more invasive approaches (see Summary and Future Investigations , below). Of course, the CL factors estradiol and progesterone could not be responsible for observed differences in maternal hemodynamics between pregnancies with and without a CL, because they are supplemented in IVF protocols that preclude the development of a CL.

The second overarching hypothesis was that increased risk of PE (and perhaps of other pathologic pregnancy outcomes) would transpire in both types of IVF leading to 0 or >1 CL, at least in part due to the perturbed maternal circulation in early pregnancy (and/or of perturbed endometrial function imposed by the IVF protocols; see Endometrial Origins of Preeclampsia , below). That is, both absent and excessive circulating levels of RLN or other CL product(s) would also have a negative impact on pregnancy outcome.

Against this backdrop, the first objective of this section, which addresses the concept of corpus luteal origins of preeclampsia, is to highlight the emerging evidence demonstrating that the majority of the increased PE risk in IVF is specifically associated with IVF protocols that preclude the development of a CL. The evidence for cardiovascular dysregulation during early gestation in these pregnancies will also be presented.

Association of Hypertensive Disorders of Pregnancy and Preeclampsia with Frozen Autologous Embryo Transfer Cycles

The increased risk of adverse obstetrical outcomes including hypertensive disorders of pregnancy and preeclampsia in the context of IVF has been known for some time, but without much consideration given to the specific types of IVF protocols with the notable exception of donor oocytes. ^, Recent investigations of PE incidence according to specific IVF protocol compared frozen autologous (oocyte derived from the same woman undergoing IVF) vs. fresh embryo transfer cycles, and/or spontaneously conceived pregnancies. It was clear that the majority of PE risk is associated with frozen embryo transfer cycles (FET). This relationship was also observed for the more generic classification “hypertensive disorders of pregnancy”. The process of cryopreservation that is specific to FET was proposed as a possible explanation for this association. ^,

However, another potentially decisive factor to consider is the CL status that also distinguishes fresh from frozen ET. (As discussed above, the concept that the CL may be a determining factor arose from earlier work on the CL factor, RLN, in relation to maternal hemodynamic changes of pregnancy in the gravid rat model). FET conducted in an artificial or programmed cycle does not produce a CL as a consequence of hypothalamic-pituitary suppression (FET-AC). In FET-AC, the endometrium is developed with estradiol administration and progesterone is subsequently provided for luteal support. This hormonal support is typically continued through 10–12 gestational weeks. Then, there is FET conducted in the setting of a natural or modified natural cycle with one CL (FET-NC), which most resembles spontaneous conceptions. FET can also be performed in a cycle with mild ovarian stimulation, in which one or at most a few CL will develop (FET-SC). Finally, in fresh ET protocols, multiple CL develop as a consequence of controlled ovarian stimulation in which higher degrees of stimulation lead to the formation of more CL up to as many as 25.

Table 4.1 specifically includes those studies in which the type of FET protocol was specified (with the exception of Sazonova et al. that was nevertheless presented in Table 4.1 , because of the rare inclusion of a spontaneous conception cohort for comparison). When PE risk is examined in the context of the specific IVF protocol employed and the corresponding number of CL that developed, i.e., 0, 1, or >1 CL, it is evident that most of the PE risk is associated with FET-AC protocols that lack a CL, and consequently, circulating CL products such as RLN. In this regard, an early 1997 study by Wennerholm et al. is consistent, because PE occurrence was similar in the FET cohort that was mostly FET-NC (82% FET-NC, 18% FET-stimulated cycle), 7.2%; fresh ET, 7.7%; and spontaneously conceived pregnancies, 6.2% ( N = 208 per group). Numerous publications have recently emerged supporting the concept that the majority of PE risk in IVF involving autologous oocytes is associated with artificial cycles (see citations in Conrad et al. ).

Maternal Cardiovascular Dysregulation during Early Pregnancy in Women Who Conceived Using Artificial Cycles

While investigating PE risk in the context of CL status using IVF as the “experimental model,” maternal systemic hemodynamics were also examined in a separate IVF population. ^, As discussed above, when compared to spontaneous pregnancy, the occurrence of “hypodynamic” and “hyperdynamic” circulations was hypothesized for FET-AC (0 CL) and fresh ET (>1 CL), respectively, during the first trimester, when CL influence typically dominates before the emergence of placental factors ( Fig. 4.1 ). Indeed, gestational changes in several cardiovascular parameters were considerably subdued in the first trimester of women who conceived using FET-AC with absent CL; notably, there was an attenuated gestational rise in cardiac output and global arterial compliance, as well as a subdued decline in systemic vascular resistance and carotid-femoral pulse wave velocity during the first trimester signifying a failure of the arterial circulation to properly dilate and increase compliance ^, (pulse wave velocity depicted in Fig. 4.2 ). Consistent with the overall hypothesis, after the first trimester (and CL-placental shift), cardiovascular parameters in the 0 CL cohort were more or less restored to levels observed in the control population of spontaneous pregnancies. ^, However, not all elements of the hypothesis were supported, insofar as the women who conceived using fresh-ET with multiple CL did not demonstrate a “hyperdynamic” circulation in the first trimester, rather their cardiovascular function was comparable to women who conceived spontaneously. ^, Perhaps not coincidentally, women who conceived by fresh ET with multiple CL also demonstrated a PE risk comparable to the spontaneous conception cohort ^, ( Table 4.1 ). Additional publications have emerged, which demonstrated dysregulation of other cardiovascular parameters and of endocrine function particularly in women who conceived using FET-AC without a CL and circulating CL factors.

Why Is FET-AC Specifically Associated with First Trimester Cardiovascular Dysregulation and Increased PE Risk?

The absence of the CL is a logical explanation. However, the differences in PE risk and cardiovascular function between autologous FET-AC and fresh ET could be due not only to CL status, but as mentioned previously, the presence or absence of embryo cryopreservation, respectively. However, this potential mechanism was addressed in the more recent studies that compared FET-AC with FET-NC, in which cryopreservation was implemented in both protocols, and within each investigation, the method of cryopreservation whether slow freezing or vitrification was presumably the same in the two FET protocols. These investigations revealed first-trimester cardiovascular dysregulation and increased PE risk in the FET-AC cohort, thus specifically implicating the absence of a CL ^, ( Table 4.1 ). (As alluded to earlier, however, another possibility is that the dose and timing of estradiol and progesterone administration for endometrial support may be suboptimal, thereby contributing to suboptimal endometrial maturation and increased PE risk. See Endometrial Origins of Preeclampsia , below). Further consistent with the CL hypothesis was the finding that deficient maternal cardiovascular function observed in the first trimester of women who conceived using FET-AC was restored thereafter, when the placenta was sufficiently developed to secrete factors that may have come to the rescue, e.g., PlGF (vide supra). However, whether impaired cardiovascular function during early pregnancy contributed to the increased PE risk remains unproven.

Which CL product(s) might regulate maternal cardiovascular changes in early pregnancy and confer some degree of protection against developing PE? RLN would be a logical candidate based on what we already know about the cardiovascular actions of the hormone. ^, An additional piece of supporting evidence is that low circulating concentrations of RLN in the first trimester were associated with increased risk of developing PE. But, of course, to the extent that low RLN concentrations in that study arose from suboptimal CL function, other circulating CL factor(s) that were not measured could also be deficient. This raises the important point that, with the exception of few substances, e.g., RLN, estradiol, progesterone, and their metabolites, possibly VEGF and a few others, our knowledge of what the CL secretes into the maternal circulation during the luteal phase and early pregnancy is lacking. Perhaps in the not-too-distant future, high sensitivity -omics might reveal multiple proteins, lipids, and derivatives circulating during the luteal phase and early pregnancy in women who conceive spontaneously or by FET-NC, but which do not circulate in women who conceive using FET-AC, thereby supporting their CL origin. In addition, one might anticipate increased plasma concentrations of CL products (or products arising elsewhere as a consequence of being stimulated by CL factors) in women who conceive by fresh ET with multiple CL, analogous to RLN. Identification of the complete suite of CL secreted products may not only further our understanding of normal pregnancy and pregnancy complications in a profound way, but could also suggest potential novel therapeutics both within and outside of pregnancy. ^,

Summary

After autologous frozen embryo transfer using artificial (or programmed) cycles in which CL development was precluded, maternal cardiovascular changes during early pregnancy were significantly attenuated, and preeclampsia risk elevated relative to spontaneous conceptions and pregnancies conceived by autologous frozen embryo transfer in a natural cycle with one CL (or after mild stimulation/ovulation induction at least in the case of preeclampsia risk). In contrast, after controlled ovarian stimulation and fresh embryo transfer maternal cardiovascular changes during early pregnancy and preeclampsia risk were comparable to spontaneous pregnancies or pregnancies conceived by autologous frozen embryo transfer in a natural cycle. The increased PE risk was observed in women who conceived using FET in artificial cycles that resulted in singleton live births, and after adjusting for maternal age, nulliparity, history of hypertension, BMI, PCOS, and pregestational or gestational diabetes mellitus. In addition, subsequent analysis of the data from one study revealed that none of the women with singleton live births who developed PE or pregnancy-induced hypertension had vanishing twins (von Versen-Hoynck, F., Baker, V.L., personal communication). Because of the detrimental impact on maternal cardiovascular adaptations to pregnancy and PE risk in the setting of IVF specifically when FET-AC protocols were used, the question arises about whether suboptimal CL function outside of IVF might also compromise maternal pregnancy physiology and outcome.

Further work is needed to bolster the proposal that the CL, and specifically the CL product RLN, plays a critical role in human pregnancy. First, the prospective and retrospective cohort studies presented in Table 4.1 are sufficiently compelling to initiate randomized controlled trials (RCTs) that compare PE risk in FET-AC vs. FET-NC protocols. Indeed, at least two such RCTs were initiated at the end of 2019 to determine whether FET-NC reduces PE risk relative to FET-AC (clinicaltrials.gov: NCT04551807 and NCT04092829 ). Second, perhaps the only approach available to ascertain whether it is the absence of circulating RLN in FET-AC protocols that contributes to the attenuated maternal hemodynamic adaptations during early pregnancy in this IVF population is to investigate the short-term infusion of either recombinant or synthetic human RLN, in order to restore circulating RLN to physiological levels. The study objective would be to determine whether maternal hemodynamics are restored (or at least partly so) by short-term RLN administration. RLN has a good safety record both within and outside of pregnancy. Even if FET-NC or FET-SC protocols are shown to improve PE risk in the RCTs, some infertile women cannot conceive using FET-NC or FET-SC protocols. Therefore, another approach is needed to reduce PE risk in this group of infertile women. To this end, long-term RLN replacement starting in the late luteal phase and continuing through at least the first trimester, again to restore physiological levels of circulating hormone, might be tested in order to determine whether the increased PE risk observed in pregnancies conceived using FET-AC protocols would be reduced. This investigation is obviously considerably more complicated and challenging, but if short-term RLN infusion improved maternal hemodynamics, thus supporting an important role for RLN in human pregnancy, then perhaps long-term RLN infusion to ascertain whether PE risk might be reduced would become more compelling and worthy of undertaking despite the considerable logistical challenges.

Perhaps more in reach now is a definitive assessment of renal function in women who conceive by IVF. Because as much as 25% of the cardiac output is distributed to the renal circulation, one would anticipate impaired gestational increases of renal plasma flow (RPF) and glomerular filtration rate (GFR) during the first trimester in women who conceived using FET-AC protocols with absent CL. That is, one would expect that the attenuated decline in systemic vascular resistance previously noted in this IVF cohort would be partly a consequence of attenuated decline in renal vascular resistance. Indeed, one pilot study showed subdued gestational increase of 24-h endogenous creatinine clearance, while another study demonstrated that plasma creatinine concentration, a reflection of GFR, was significantly higher during the first trimester in women who conceived using FET-AC protocols. However, the gold standard for measuring RPF and GFR is the renal clearances of para-aminohippurate and inulin or inutest, respectively, which have a long track record of use in human pregnancy, e.g., . Unfortunately, these compounds are no longer available for human use in the United States. Therefore, this critical study would need to be conducted elsewhere, where they are still available.

Finally, as already discussed, identification of the entire repertoire of products that the CL secretes during early human pregnancy is needed, which would likely pave the way for new discoveries both within and outside of pregnancy.

Introduction

Preeclampsia (PE) is widely believed to begin within the first trimester as a result of defective placentation ( Chapter 5 ). However, since PE is a heterogeneous syndrome ( Chapter 1 ), it should be acknowledged that this etiology may not apply to or be a major factor in all women who develop the disease. Nevertheless, a fundamental problem with investigating the early origins of PE in relation to placentation is that, at least to date, the critical tissues, chorion and decidua, have been unattainable except for surplus chorionic villous samples (CVS) (described below). As a consequence, there are many reports that elegantly describe the molecular pathology of chorionic and decidual tissues obtained at delivery from women who suffered PE, because these delivered tissues are readily obtainable (and see citations in Conrad et al. ). It is possible, however, that the molecular pathology of tissues acquired at delivery may not be informative of the early origins of PE, which transpired months before in the first trimester, i.e., there is a gaping temporal disconnect between the (presumed) origins of preeclampsia in early pregnancy and choriodecidual tissues that are acquired at delivery for investigation. The molecular pathology of chorion and decidua obtained at the end of pregnancy could as much result from the disease than reflect its origins. For example, after disease onset, there are a multitude of circulating factors such as sFLT1, TNF-α, syncytial extracellular vesicles, etc., that could be just as injurious to decidual and trophoblast cells as endothelial cells, leading to disruption of molecular pathways. In summary, the molecular pathology of placental tissues obtained at delivery might relate more to the clinical manifestations than the genesis of preeclampsia that occurred in the first trimester (or even earlier). In support of this idea, we found little or no molecular overlap of CVS with decidua basalis or isolated villous trophoblasts obtained from women who experienced PE ^, (see Impairment of Endometrial Maturation—(Pre)Decidualization , below).

To address this overarching conundrum, surplus CVS was prospectively collected from women who later developed PE. Unexpectedly, after subjecting the CVS tissues to global gene expression analysis, a molecular signature of dysregulated (pre)decidualization emerged rather than perturbation of oxygen or oxidative stress related gene pathways, the latter predicted by the findings in delivered placenta. Since this initial investigation, the concept that the antecedents of PE may reside in the endometrium during early pregnancy and even before pregnancy (again, at least in some women) has received strong support. The major objective of this section, which addresses the evidence for endometrial origins of preeclampsia, is to summarize the literature to date surrounding the potential decidual origins of preeclampsia and in the context of other endometrial disorders.