{"paper_id":"b32af7df-abab-4fd4-ac0f-99cd9bba8421","body_text":"Adverse pregnancy outcomes may have antecedents in the pre- and\nperi-conception periods, and first trimester of pregnancy. This idea was supported\nby studies implicating dysregulation of endometrial maturation (decidualization)\nduring the secretory phase and early pregnancy in the genesis of\npreeclampsia 1 – 4 . The concept of “endometrium\nspectrum disorders” then emerged 3 , which was underpinned by the integration of multiple\nendometrial transcriptomic databases available in the public domain 5 . These bioinformatics analyses\nprovided evidence for dysregulation of molecular pathways in common among the\nclassic endometrial disorders—recurrent implantation failure, recurrent\nmiscarriage, endometriosis—and one of the great obstetrical or placental\nsyndromes, preeclampsia 5 , 6 . Conceivably, other adverse pregnancy\noutcomes that may arise from placental pathology including normotensive intrauterine\ngrowth restriction and preterm birth, also fall within the continuum of endometrial\nspectrum disorders affecting implantation, placentation or both depending upon the\nspecific molecular pathways disrupted and the severity of disruption 3 . Although the genesis of the great\nobstetrical syndromes including preeclampsia is likely to be multifactorial, in some\nwomen these disease entities may have antecedents in endometrial dysregulation\nduring early pregnancy or even before pregnancy.\nIn vitro  fertilization (IVF) is another setting in which\npre- and peri-conception, as well as early pregnancy factors may impact obstetrical\noutcome. In pregnancies conceived by IVF using artificial (programmed) cycles\ninvolving hypothalamic-pituitary suppression and development of the endometrium with\nestradiol and progesterone, a corpus luteum (CL) does not develop 7 . These IVF protocols were observed to perturb\nendometrial gene expression in the mid-secretory phase 8 , 9 , and\nto be associated with greater risk of post-term delivery, large for gestation age\ninfants and macrosomia, as well as placental accreta 10 , 11 .\nIn addition, artificial cycles were also linked to maternal hemodynamic\ndysregulation in the first trimester, and hypertensive disorders of pregnancy and\npreeclampsia 12 , 13 . Because the CL is a key regulator of\nendometrial function including decidualization in the secretory phase and early\npregnancy, one potential explanation for increased incidence of these adverse\nobstetrical outcomes is that, despite luteal support with exogenous estradiol and\nprogesterone, the absence of other crucial circulating CL factors(s) negatively\naffects endometrial maturation in artificial IVF cycles 7 , 10 , 12 . Another potential, albeit not\nmutually exclusive explanation is that the dosage and timing of estradiol and\nprogesterone administration for luteal support is suboptimal 8 , 9 .\nIn this review, the molecular evidence of impaired decidualization in\npreeclampsia, and the emerging concept of “endometrial spectrum\ndisorders”, in which dysregulated decidualization of preeclampsia, recurrent\nimplantation failure, recurrent miscarriage and endometriosis demonstrated\nsignificant overlap of molecular pathology will be presented. In addition, the\ndiscovery of dysregulated maternal hemodynamics during the first trimester of\nartificial (programmed) IVF cycles, as well as the association with increased risk\nfor hypertensive disorders of pregnancy and preeclampsia will be also be presented\nin the context of the corpus luteum, or more precisely, the lack thereof, in the\ncase of artificial cycles.\n\nOne widely held theory is that preeclampsia originates within the\nplacental bed during early gestation in many women. Normally, the fetal\nextravillous trophoblast emanating from the anchoring villous tips invade the\ngestational endometrium (decidua) and inner 1/3 of the myometrium, remodeling\nthe uterine spiral arteries from low caliber, high resistance to high caliber,\nlow resistance blood vessels. These physiological changes of the spiral arteries\nfacilitate increased maternal blood flow into the intervillous space. In\ncontrast, preeclampsia is often associated with impaired trophoblast invasion\nand spiral artery remodeling, thereby restricting blood flow into the\nintervillous space leading to placental ischemia. These placentation\ndeficiencies may not be unique to preeclampsia, as they have also been\ndescribed, albeit not universally so, in late sporadic miscarriage, normotensive\nfetal growth restriction, placental abruption, and preterm labor 6 .\nIt should be noted that the classical view of the biological consequences\nof spiral artery remodeling or lack thereof, as presented above, has been\nrecently called into question. Revised computational modeling suggested that\nspiral artery remodeling is unlikely to contribute substantially to reducing\nuterine vascular resistance and increasing blood flow in normal pregnancy,\nrather the (proximal) radial artery is a more significant resistance\nsite 14 . Computational\nmodeling further revealed that spiral artery remodeling in normal pregnancy\nreduces the velocity of increased blood flow into the intervillous space,\nthereby protecting delicate villi from mechanical damage and increasing the\ntransit time of blood flow through the intervillous space allowing for adequate\nexchange of oxygen and nutrients across the syncytiotrophoblast layer 15 . According to this model,\nfailure of spiral artery remodeling in preeclampsia would lead to the opposite\nchain of events, i.e., mechanical damage of villi by high velocity blood flow\nand rapid transit time of blood flow through the intervillous space precluding\nadequate oxygen and nutrient exchange across the syncytiotrophoblast\nlayer 15 . Nevertheless,\nregardless of which model is apropos, each predicts that failure to remodel\nspiral arteries would impair placental function. In both scenarios,\nischemia-reperfusion injury would also occur as a consequence of spontaneous and\nhormone-induced constriction or relaxation of spiral arteries that were not\nremodeled and retained vascular smooth muscle.\nBecause uterine invasion and spiral artery remodeling by trophoblast can\nbe deficient in preeclampsia, this fetal cell has been intensively investigated.\nMoreover, the paternal genetic contribution to disease etiology could be\nmanifest, at least in part, through impairment of trophoblast invasion. The\nseminal work of Fisher and colleagues revealed the extensive molecular and\nfunctional aberrations of the extravillous trophoblast in early onset, severe\npreeclampsia as investigated at the end of pregnancy  in situ ,\nand after trophoblast isolation,  in vitro 16 . However, a potential caveat to this\nmethodological approach is that molecular pathology at the end of pregnancy may\nbe more related to the phenotypic expression of the disease, which typically\nemerges at that time or may even be a  consequence  of the\ndisease (e.g., sFLT1 could conceivably be injurious to endometrium, in addition\nto endothelium). Therefore, the molecular pathology of tissues procured at\ndelivery is likely to be unrelated to the molecular etiology which\n caused  the disease months before, when the physiological\nprocesses of uterine trophoblast invasion and spiral arterial remodeling\ntranspired. That is, the large temporal gap between the acquisition of placental\ntissues for molecular studies at delivery and the critical period of trophoblast\ninvasion and spiral artery remodeling occurring in early pregnancy, may preclude\nany insights into the molecular genesis of preeclampsia. One potential solution\nto this conundrum is  prospective  acquisition of early placental\ntissues (surplus chorionic villous samples or CVS) months before onset of\nclinical manifestations. Although the advent of noninvasive prenatal screening\n(NIPs) has markedly reduced the number of CVS procedures performed world-wide,\ncollaboration among large medical centers with the greatest volume of CVS cases\nannually, could lead to acquisition of sufficient sample numbers for molecular\nand functional investigation of preeclampsia etiology targeting the\ntrophoblast.\nAnother potentially relevant tissue that has received little attention in\nthe context of adverse pregnancy outcomes is the maternal decidua\n(“soil”), which extravillous trophoblast (“seed”)\ninvade ( vide supra ). Conceivably, insufficient or defective\nendometrial maturation (decidualization) that begins in the secretory phase and\ncontinues after implantation may impede trophoblast invasion and spiral artery\nremodeling, thereby contributing to the genesis of preeclampsia 1 , 17 . This alternative, but not mutually exclusive\nhypothesis is perhaps intuitive or self-evident, in light of the close\napposition of endometrial stromal, glandular epithelial and maternal immune\ncells with trophoblast and spiral arteries in the placental bed. Furthermore,\nthe maternal inheritance pattern of preeclampsia could be manifest, at least in\npart, through dysregulation of decidualization. Normally, massive molecular and\nfunctional changes occur in endometrial stromal and epithelial cells, spiral\narteries and immune cells during decidualization in the secretory phase and\nearly pregnancy. Implantation and placentation depend on the optimal and timely\nprogression of decidualization. Decidualization of the glandular epithelium is\nprerequisite to histiotrophic nutrition during early gestation prior to onset of\nmaternal blood flow; uterine [natural killer] NK cells become the major immune\ncell type in the placental bed and assume an immunomodulatory rather than\ncytotoxic phenotype, and they initiate spiral artery remodeling and stimulate\ntrophoblast invasion; uterine macrophages accumulate and they adopt an\n“M2” or alternatively active rather than pro-inflammatory\nphenotype; and T regulatory cells contribute to immune tolerance at the\nmaternal-fetal interface in the face of the fetoplacental\nsemi-allograft 3 , 18 . In essence, decidualization\nis preparation of the “soil” for the “seed”, i.e.,\nembryo implantation and subsequent placentation. Impairment of this process as\none possible etiology of preeclampsia would seem to be a reasonable hypothesis\nto explore.\nIn order to investigate relevant reproductive tissue temporally related\nto decidualization, trophoblast invasion and spiral artery remodeling, we\nprospectively obtained surplus CVS at ~11.5 gestational weeks in women\nwho developed preeclampsia with severe features (sPE) or who experienced normal\npregnancy (NP) outcome 5–6 months later 1 . These tissue samples were snap frozen in\nliquid nitrogen and ultimately analyzed by DNA microarray. Contrary to our\nhypothesis, we did not detect a molecular signature consistent with ischemia or\nischemia-reperfusion, rather many genes identified as biomarkers of\ndecidualization were downregulated in the CVS from women who developed sPE\nrelative to NP outcome including insulin-like growth factor binding protein-1\n(IGFBP-1), glycodelin or progesterone-associated endometrial protein (PAEP),\nprolactin (PRL) and IL-15. These initial observations prompted a wider text\nmining approach, which revealed many other dysregulated decidual genes that, in\nturn, provided the justification for a formal bioinformatics reanalysis of the\nraw data from our CVS microarray data 2 .\nThe bioinformatics reanalysis of the CVS microarray data revealed 396\ndifferentially expressed genes (DEGs) between CVS from sPE and NP-CVS, of which\n154 or 40% overlapped with DEGs changing during endometrial maturation either in\nthe secretory phase or early pregnancy (p=4.7 × 10 −14 ),\nthe latter DEGs obtained by reanalyzing publically available microarray datasets\nof normal decidualization. Moreover, approximately 73% of these 154 DEGs changed\nin the  opposite  direction compared to normal endometrial\nmaturation (p=0.01), and 75% overlapped significantly with DEGs between\nproliferative vs late secretory endometrium or DEGs between decidualized vs\nnondecidualized endometrium obtained from tubal ectopic pregnancies (p=4.4\n× 10 −9 ). Neither of these endometrial tissues contain\nextravillous trophoblast, thus suggesting a  primary  role for\ndysregulated decidualization. Moreover, 16 DEGs normally\n upregulated  in uterine compared to peripheral NK cells were\n downregulated  in sPE- compared to NP-CVS (p<0.0001).\nDEGs normally  upregulated  in uterine relative to peripheral\nmacrophages were  downregulated  in sPE- vs NP-CVS (p=9.5\n× 10 −3 ) and vice versa (p=1.1 ×\n10 −6 ) 3 .\nTaken together, these observations suggested deficient or defective endometrial\nmaturation including uterine NK cells and macrophages may precede the\ndevelopment of preeclampsia with severe features. The concept that dysregulated\ndecidualization is involved in the genesis of preeclampsia was supported by 6\nstudies published throughout the last 10 years or so, which demonstrated a\nreduction of circulating concentrations of IGFBP1 during early pregnancy in\nwomen who later developed PE (reviewed in 3 ).\nAnother notable finding from the CVS microarray study was that the\naverage mRNA expression of a cohort of 20 decidual genes  uniquely\nupregulated  in normal late secretory compared to proliferative\nendometrium were  downregulated  in sPE- vs NP-CVS by ~\n2-fold (p<0.0001) 2 . This\nobservation suggested that the dysregulation of endometrial maturation in the\nwomen who developed PE with severe features may have started\n before  pregnancy during the secretory phase. Indeed, the\nidea that endometrial pathology may reside in the secretory endometrium was\nstrongly reinforced by Garrido-Gomez and coworkers, who reported marked\nimpairment of  in vitro  decidualization of endometrial stromal\ncells isolated and then cultured from mid-secretory endometrial biopsies of\nwomen who experienced sPE during the previous 1–5 years 4 . In fact, there was significant\noverlap of DEGs that arose from sPE-CVS vs NP-CVS as reported by Rabaglino and\ncolleagues with the DEGs observed by Garrido-Gomez and coworkers in cultured\nendometrial stromal cells decidualized  in vitro  that were\nderived from women who experienced prior sPE vs normal pregnancy 5 .\nA priori, decidual tissue at delivery is likely to be markedly\ndissimilar from decidual tissue in the secretory phase or early pregnancy\n( vide supra ). This important point was highlighted by\nadditional bioinformatics analysis of differential gene expression in these\ntemporally disconnected decidual tissues from women who experienced sPE vs NP,\ninsofar as there was little or no overlap 5 . Thus, designing strategies to address the molecular\ngenesis of PE which resides in the secretory endometrium and/or placental bed of\nearly pregnancy based upon the molecular pathology of delivered tissue may be\nmisleading, and unlikely to lead to preventative or early corrective\nmeasures.\nIn summary, emerging evidence supports the concept that preeclampsia may\narise at least in some women from dysregulated decidualization including\naberrant endometrial immune cell number and/or function in the secretory phase\nand during early pregnancy 1 , 2 , 4  ( Fig. 1 ). In\ndelivered placentas, decidual function is also perturbed, which may contribute\nto or arise from deleterious circulating placental factors like sFLT1\n(e.g. 19 , and reviewed\nin 3 ). But, as discussed\nabove, the transcriptomics of delivered decidua are distinct from those of early\npregnancy or the secretory phase in women who developed preeclampsia, and as\nsuch, may not be relevant to disease etiology 3 , 5 . Perhaps not\ntotally unexpected in light of this potential link between aberrant\ndecidualization and preeclampsia, an elegant recently published study provided\nevidence that intrauterine growth restriction, another disease entity classified\nunder the great obstetrical or placental syndromes, may also have origins in\nimpaired decidualization 20 .\nBecause dysregulated decidualization was associated with preeclampsia,\nwe asked the question whether there might be molecular overlap with other\nendometrial disorders 5 . To this\nend, we reanalyzed 8 microarray databases in the public domain from normal and\npathologic endometrium or decidua. A significant proportion of the DEGs up- or\ndownregulated in CVS from women who experienced PE with severe features compared\nto NP, or in cultured endometrial stromal cells decidualized  in\nvitro  derived from mid-secretory biopsies of women who experienced\nsevere PE relative to NP ( vide supra ), demonstrated overlap\nwith, and the same directional change as DEGs in recurrent implantation failure\n(RIF), recurrent miscarriage (RM) and endometriosis (OSIS) compared to their\nrespective control tissues 5 .\nIn order to further explore this idea, a functional analysis and\npathway-driven approach was taken 5 . The cytokine-cytokine receptor interaction pathway (264\ngenes) was one of the most prominent and significant molecular pathways in\ncommon among normal and pathological endometrium. Principal component analysis\n(PCA) was employed to compare gene expression in this pathway among the\ndifferent normal and pathological endometrial tissues represented by 8\nmicroarray databases ( Fig. 2 ). CVS and\n in vitro  decidualized endometrial stromal cells derived\nfrom mid-secretory phase biopsies of women who suffered sPE segregated with the\nthree endometrial disorders. In contrast, decidua procured at delivery from\nwomen affected by sPE clustered with normal endometrium, indicating that the\nexpression pattern of the genes of these tissues at least in the\ncytokine-cytokine receptor pathway more resembled the normal than pathological\nendometrium. Of course, this does imply that other molecular pathways in the\ndecidua obtained at delivery from women who suffered sPE may not be abnormal.\nOverall, however, the differentially expressed genes affected in delivered\ntissues were not overlapping with those found in the CVS or  in\nvitro  decidualized endometrial stromal cells from mid-secretory\nphase biopsies of women who suffered sPE. In the same vein, proliferative\nendometrium and non-decidualized early pregnancy endometrium as histologically\nassessed, clustered with pathological endometrium in the context of the\ncytokine-cytokine receptor interaction pathway.\nTaken together, integration of multiple microarray datasets derived from\nnormal and pathologic endometrium suggested that, at least in some women,\npreeclampsia may be part of a continuum of endometrial disorders involving\nvarying degrees of molecular dysregulation affecting implantation, placentation\nor both. Indeed, other disease entities classified as placental syndromes may\nalso fall along this continuum ( Fig. 3 ).\nThat PE has, in common with the classical endometrial disorders, many\ndifferentially expressed genes and gene pathways strengthens the concept that\nthe genesis of the disease may reside in the decidua at least for some women.\nViewing PE in this light may also partly explain why women with endometriosis\nwho become pregnant experience increased PE risk as reported by some, but not\nall investigators. Similarly, recurrent miscarriage was also associated with\nincreased PE risk (see 5  for\ncitations).\n\nAn association between IVF and hypertensive disorders of pregnancy or\npreeclampsia has been thoroughly documented ( Tables 1 – 3 ). Several\ngroups of researchers reported increased frequency of hypertensive disorders of\npregnancy or preeclampsia in frozen embryo transfer (FET) vs fresh ET. However,\nthe FET protocol(s) were not delineated, and whether donor gametes were included\nor not was only specified in one of the studies 21 – 23 . In the investigation by Opdahl and colleagues,\nrelative risk (RR) for hypertensive disorders of pregnancy was 7.0% and 4.7%,\nrespectively, for FET and spontaneous pregnancy, aOR 1.41, 95%CI\n1.27–1.56 (adjusted for maternal age, parity, birth year, infant sex and\ncountry). The same authors also noted higher risk in siblings conceived by FET\nvs fresh ET, aOR 2.39, 95%CI 1.48–3.86 23 . More recently, the\nrisk of preeclampsia was also found to be increased for autologous FET in\nartificial cycles (AC) vs fresh ET 24 – 26 . In\none of these studies, patients with polycystic ovary syndrome (PCOS) were\nrandomized to FET-AC or fresh ET cycles 24 . Finally, in another investigation, autologous FET-NC\nand FET-stimulated cycles were employed, and the authors observed no significant\ndifferences in the rate of hypertensive disorders among women conceiving by\nFET-NC, FET-stimulated cycle, fresh ET or spontaneous conception 27 . Taken together, these studies\nsuggested that FET, and in particular FET-AC protocols may be associated with\nincreased rates of hypertensive disorders of pregnancy and preeclampsia as\ncompiled from  Table 1  and summarized in\n Tables 2  and  3 , respectively.\nIn a recently published prospective study, we recruited women during\nearly pregnancy with singleton intrauterine pregnancies who conceived using\nautologous oocytes and delivered live born infants (n=878) 12 . No participants had an infertility\ndiagnosis of premature ovarian failure or were recipients of donor oocytes or\nembryos. After adjustment for several preeclampsia risk factors (i.e., maternal\nage, nulliparity, history of hypertension, BMI, PCOS, pre-gestational and\ngestational diabetes), women conceiving by FET in artificial cycles, in which a\nCL did not develop, had increased risk for preeclampsia (aOR 2.73, 95%CI\n1.14–6.49) and preeclampsia with severe features (aOR 6.45, 95%CI\n1.94–25.09) compared to sub-fertile women with one CL. In a sub-analysis\nof FET in artificial cycles compared to FET in modified natural cycles with one\nCL, the adjusted odds ratios were 3.55, 95%CI 1.20–11.94 for developing\npreeclampsia, and 15.05, 95%CI 2.59–286.27 for preeclampsia with severe\nfeatures. Importantly, women conceiving by fresh ET in ovarian stimulation\ncycles who had multiple CL did not show increased preeclampsia risk. This study\nwas the first to evaluate preeclampsia risk in IVF from the standpoint of CL\nstatus. The findings implicated absence of the CL as a possible contributor to\nthe development of preeclampsia ( Tables 1 \nand  3 ).\nIn a parallel study, we serially evaluated cardiovascular function in\nwomen before, during and after pregnancies, who conceived after controlled\novarian stimulation (COS) (>1 CL), autologous FET or fresh donor\noocyte-derived embryos transferred in artificial cycles (0 CL), or spontaneous\nconceptions (1 CL) 12 , 13 . We observed significant attenuation of\nthe gestational changes in numerous cardiovascular parameters during the first\ntrimester in women who conceived by IVF without a CL, which mostly recovered\nduring the second trimester. These findings were consistent with the hypothesis\nthat circulating CL factor(s) mediate cardiovascular adaptations to pregnancy\nduring the first trimester in spontaneous pregnancy, and placental factors\nsupersede after the corpus luteal-placental shift 7 . The cardiovascular adaptations to\npregnancy in the IVF participants with multiple CL were comparable to those\nobserved in spontaneous pregnancies. Although we established an association\nbetween absent CL, dysregulated cardiovascular adaptations in the first\ntrimester, and increased preeclampsia risk, whether these factors were causally\nlinked remains to be proven.\nA recent comprehensive publication from Sweden based on a retrospective\nregistry study of singleton pregnancies after autologous FET reported a\nfrequency of 8.2% for preeclampsia in artificial cycles (0 CL; n=1446) compared\nto 4.4% in natural cycles (1 CL; n=6297)—aOR 1.78, 95%CI 1.43–2.21\n(adjusted for maternal age, BMI, parity, year of birth of infant, maternal\nsmoking, chronic hypertension, child’s sex, level of maternal education,\nand years of involuntary childlessness) 10 . The women conceiving by fresh ET with multiple CL\n(n=24,365) showed a lower rate of preeclampsia closer to that of spontaneous\nconceptions (n=1,127,566)—3.7% and 2.8%, respectively. Similar trends\nwere observed for hypertensive disorders of pregnancy. 10  Additional published studies\ndemonstrated that women conceiving by autologous FET in artificial cycles had\nincreased risk for hypertensive disease of pregnancy or preeclampsia compared to\nautologous FET in natural cycles, or fresh ET in ovarian stimulation cycles.\nHowever, a potential etiologic role for absent CL in the elevated risk of\nhypertensive disorders of pregnancy or preeclampsia in artificial cycles was not\nhypothesized in these reports (e.g., 24 – 26 , 28 ;  Tables 1 – 3 ).\nIn summary, although not yet confirmed by a rigorous randomized\ncontrolled clinical trial comparing autologous FET-AC and FET-NC or modified NC,\nthe emerging data suggest that use of IVF protocols which lead to suppression of\nCL formation may increase preeclampsia risk. These data are concerning due to\nthe immediate- and long-term detrimental consequences of preeclampsia for both\nmother and child. Thus, in addition to pre-pregnancy maternal characteristics in\nmany IVF patients such as older maternal age and subfertility, absence of the CL\nas an etiological factor in the impaired maternal cardiovascular adaptations\nduring early pregnancy and increased preeclampsia risk should also be\nconsidered. The absence of critical circulating CL factor(s) is perhaps the most\nlikely explanation for the dysregulation of maternal cardiovascular function\nobserved during early pregnancy in women conceiving by IVF without a CL, in part\nbecause either full or partial recovery subsequently transpired after the\n“corpus luteal-placental shift” coincident with secretion of\nplacental factors 12 , 13 . But, whether the absence of CL\nfactor(s) and of their vasodilatory and pro-decidualizing attributes, or the\npossibility of suboptimal luteal support with estrogen and progesterone for\nendometrial preparation in artificial cycles (dose and/or\ntiming —vide supra ) 5 , 8 ,\nor both underlie increased preeclampsia risk is less clear. Ultimately, if\nreplacement of the missing CL factor(s) (e.g., relaxin) restores maternal\ncardiovascular function in early pregnancy and reduces preeclampsia risk, then\nthis approach might be an alternative preventative strategy to autologous FET in\na natural cycle for some women, and perhaps the only approach available for\nwomen who have ovarian failure requiring donor oocytes or embryos to conceive.\nMild ovarian stimulation, which would permit CL development in a FET cycle,\nmight be used is women who do not ovulate on a regular basis.\nThe absence of a CL and circulating CL product(s) likely contribute to\nthe increased risk of preeclampsia in autologous FET-AC vs FET-NC. However,\nwhether or not cryopreservation, in addition to absence of a CL, may confer\nadded risk of preeclampsia in FET-AC compared to fresh ET is difficult to test.\nClose examination of the study by Sites et al. in the context of CL status may\nshed some light on this question 26 . Autologous fresh embryo transfer (>1 CL) and\nautologous frozen embryo transfer in an artificial cycle (0 CL) yielded rates of\nPE of 4.29 and 7.51%, respectively ( Tables\n1  and  3 ). The difference could\nhave been a consequence of embryo state (fresh vs frozen) and/or CL number\n(>1 vs 0 CL). Donor fresh and frozen embryo transfer in artificial cycles\n(0 CL) yielded rates of preeclampsia of 12.13 and 10.78%, respectively 26  (use of artificial cycles for\ndonor frozen embryo transfers was standard of care according to Dr. Sites,\npersonal communication). These preeclampsia rates were not significantly\ndifferent, which suggested that the freeze/thaw manipulation of embryos did not\nconfer increased risk for PE (although a ceiling effect cannot be excluded).\nComparing autologous (4.29%) and donor fresh (12.13%) ET revealed that the\ndifference, 7.17%, was PE risk attributable to “donor” (vs\nautologous) and “CL” (>1 vs 0 CL) effects. Comparing\nautologous (7.51%) and donor frozen (10.78%) ET both using artificial cycles (0\nCL) revealed that the difference, 4.62%, was the contribution to PE attributable\nto “donor” (vs autologous) effect, alone. Thus, the difference\nbetween the PE rates attributable to “donor” and\n“CL” effects (7.17%) and “donor” (4.62%) effect,\n2.55%, must be due to the “CL” effect, alone. Although any\nconclusion based on these rough estimates must be regarded cautiously, the\nartificial cycle (0 CL), in addition to a donor embryo source appeared to\naccount for the considerably higher rates of preeclampsia in women who were\nrecipient of donor-oocyte derived embryos.\n\nThe emerging evidence suggests that perhaps not all IVF protocols are\ncreated equal with respect to increased risk for hypertensive disorders of\npregnancy and preeclampsia. Although IVF protocols were frequently not presented\nin sufficient detail in many of the publications, after close inspection of\nthose in which they were delineated, the balance of evidence implicated the\nartificial (or programmed) cycle protocol. That is, elevated risk for\nhypertensive disorders of pregnancy and preeclampsia primarily resulted from\nFET-AC, not FET-NC or FET-stimulated, or fresh ET cycles ( Table 1 ). Perhaps not coincidentally, the maternal\nhemodynamic adaptations to pregnancy were perturbed in AC, but not COS cycle\nprotocols 12 , 13 . Close inspection of the grand averages\nof the rates for hypertensive disorders of pregnancy and preeclampsia listed in\n Table 1  further highlight that\nincreased risk is associated with the artificial cycle ( Tables 2  and  3 ).\nIn most of the studies that reported increased risk for preeclampsia in\nautologous FET-AC protocols, the gestational age of preeclampsia onset and the\nseverity of disease were not specified ( Table\n1 ). However, a few did provide these details. Chen and coworkers\nobserved increased risk for term, but not preterm preeclampsia or preeclampsia\nwith severe features 24 ;\nincreased frequency of term preeclampsia and preeclampsia with severe features,\nbut not preterm preeclampsia were noted by both von Versen-Hoynck et\nal. 12  and Barsky and\ncolleagues 25 ; and\nSites and coworkers reported increased incidence of both preterm and term\npreeclampsia, and preeclampsia with severe features in autologous\nFET-AC 26 . Although the\nnumber of studies are too few to draw any definite conclusions, with the\nexception of Sites and coworkers, term preeclampsia both with and without severe\nfeatures was associated with autologous FET-AC protocols. A recent theory for\nthe pathogenesis of term preeclampsia proposes that it arises from villous\novercrowding, which leads to compression of intervillous spaces that, in turn,\nimpedes blood flow causing placental ischemia. That is, villous growth outstrips\nuterine capacity 29 \n( vide supra ).\nInterestingly, women with low circulating relaxin concentration in early\npregnancy were observed to be at increased risk of developing late onset\npreeclampsia (≥34 weeks) 30 . Possibly, the vasodilatory attributes of relaxin are\nimportant in some women to mitigate the physiological rise in circulating\nvasoconstrictors such as sFLT1, thereby restraining the normal restoration of\nthe maternal circulation to the non-pregnant state of relative vasoconstriction\ntowards the end of pregnancy 13 , 31 , 32 . In fact, circulating sFLT1 and the sFLT1/PLGF ratio\nwere significantly higher at the end of pregnancy in women conceiving by IVF\nespecially for AC (0 CL) protocols 33 , perhaps reflecting villous overcrowding and placental\nischemia. Whereas circulating relaxin is absent in artificial cycles,\nconcentrations are either comparable to spontaneous pregnancy or markedly higher\nin controlled ovarian stimulation cycles, the latter possibly explaining the\nequivalent rates of preeclampsia in COS and spontaneous pregnancies as noted\nabove ( Tables 1  and  3 ).\nThe finding by Sites et al. of increased preterm, in addition to term\npreeclampsia after autologous FET-AC protocol should not be ignored\n( vide supra ) 26 . Indeed, this investigation may have identified increased\nrisk for both term and preterm PE due to larger cohort sizes, and hence,\nincreased study power. However, on the surface, it is difficult to reconcile\npreterm and term PE based on a common decidual etiology. Preterm preeclampsia is\nwidely believed to be associated with impaired trophoblast invasion and spiral\nartery remodeling, while recent theory suggests that term preeclampsia does not\ninvolve deficient placentation, but rather villous overcrowding ( vide\nsupra ). Conceivably, villous overcrowding might be exacerbated by\npost-term delivery and larger placentas associated with large for gestational\nage or macrosomic infants—adverse pregnancy outcomes also associated with\nartificial (programmed) IVF cycles (e.g., 10 , 34 ). Indeed,\npost-term delivery itself has been associated with increased preeclampsia and\neclampsia risk 35  presumably as\na consequence of the mechanisms outlined above being exacerbated by prolonged\ntime for placental growth 29 .\nEnhanced frequency of LGA and macrosomia in autologous FET during artificial\ncycles are also consistent with the increased risk of term preeclampsia, insofar\nas it is not infrequently accompanied by a large for gestational age (LGA)\nfetus 36 , 37  and large placenta 38 . Whether term preeclampsia may be\nassociated with excessive trophoblast invasion, albeit to lesser degree than\naccreta spectrum disorders that also occur more frequently in artificial IVF\ncycles (e.g., 11 ), is not\nknown.\nOn the one hand,  excessive  trophoblast invasion is\nobserved in tubal pregnancy and accreta spectrum disorders, in which decidua is\ndeficient and/or dysregulated 39 – 41 . On\nthe other, dysregulated decidualization is associated with preterm preeclampsia\nwith severe features, in which trophoblast invasion is\n deficient 2 – 4 \n( vide supra ). These apparently disparate actions of the\ndecidua on trophoblast invasion are difficult to reconcile mechanistically,\ni.e., how can decidual pathology lead to both excessive and deficient\ntrophoblast invasion? One potential explanation is that activation of different\nmolecular pathways account for these divergent actions of the decidua on\ntrophoblast behavior that may be regulated, at least in part, by factors derived\nfrom the corpus luteum, or lack thereof. A priori, it seems logical to presume\nthat decidual pathology would not be restricted to one phenotypic expression of\nexcessive trophoblast invasion as in some cases of placental accreta disorders,\nbut rather different molecular pathology could also arise, which leads to\nimpaired trophoblast invasion frequently observed in preterm preeclampsia.\n\nIn light of the association between dysregulated decidualization and\npreeclampsia, the underlying molecular mechanism(s) of the pathologic decidua now\nneed to be identified, in order to design prophylactic or corrective interventions.\nEventually, efforts to improve decidualization before and during early pregnancy\nmight be indicated in those women at increased risk for the disease (e.g., by\nadministration of hormones known to promote decidualization). Finally, circulating\nor urinary biomarkers or a panel of biomarkers reflecting endometrial dysfunction\nmight be helpful in identifying women at increased risk (e.g., low circulating\nIGFBP-1 or glycodelin before and/or during early pregnancy) 3 .\nGiven the perturbed maternal physiology and increased risk of several\nadverse pregnancy outcomes in IVF cycles involving autologous frozen embryo transfer\nin artificial (programmed) cycles, what can be done to intervene? Careful inspection\nof the data revealed that the increased risk for hypertensive disorders of pregnancy\nand preeclampsia was not observed in frozen embryo transfer using natural or\nstimulated cycles, or controlled ovarian stimulation cycles. Based on this\nrevelation, it is reasonable to propose that a large multi-site, randomized clinical\ntrial be conducted comparing pregnancy outcomes between autologous FET-AC and\nFET-NC, FET-modified NC, or FET-stimulated cycles 12 . In a subgroup of patients, maternal\nphysiology could be intensively investigated, in order to determine whether it would\nbe normal after FET-NC, FET-modified NC or FET-stimulated cycles in contrast to\nFET-AC as predicted 7 , 12 , 13 , 42 . If a RCT confirms the hypothesis\nthat maternal physiology and pregnancy outcome will be improved, then FET-NC,\nFET-modified NC or FET-stimulated cycles might be preferred protocols in many women.\nA common denominator is the absence of a corpus luteum in artificial IVF cycles,\nwhereas at least one CL develops in FET-NC, FET-modified NC and FET-stimulated\ncycles 7 , 12 , 13 .\nAll CL product(s) are missing in FET-AC (except for E2 and P4 administered for\nluteal support), and therefore, the absence of any one or several of them could\nunderlie the dysregulated maternal cardiovascular adaptations to pregnancy and\nincreased risk for adverse pregnancy outcomes. Indeed, both the cardiovascular\nsystem and endometrium are known targets of at least one CL factor that is not\nreplaced in AC protocols, relaxin ( vide supra ; 42 , 43 ).\nTherefore, including the missing CL factor(s) like relaxin with E2 and P4 for luteal\nsupport in artificial cycles might be investigated, in order to determine whether\nthe addition of CL factor(s) like relaxin to the IVF medical regimen would correct\nthe dysregulated maternal cardiovascular physiology and reduce the risk for adverse\npregnancy outcomes. For women with ovarian failure for which natural IVF cycles are\nunattainable, replacing the missing CL factor(s) may be the only option.","source_license":"public-domain-us","license_restricted":false}