Hypertension in Pregnancy

F Broughton Pipkin1 and JM Roberts2 1School of Human Development, University of Nottingham, UK; 2University of Pittsburgh and Magee- Women’s Institute, USA

Hypertension arising during pregnancy remains one of fetal growth restriction. Prospective studies have sug- the two most frequently-cited causes of maternal death gested a hyperdynamic circulation in early pregnancy, in the UK. In some cases, pregnancy is unmasking with cardiac output only falling in established disease. underlying hypertension, which manifests itself in later Baroreﬂex sensitivity is decreased in normal pregnancy, life. Pregnant women who develop de novo proteinuric and still further decreased in established PE. Activation hypertension (pre-eclampsia, PE) can share many risk of endothelial cell function antedates the clinical diag- factors with patients with the metabolic syndrome, such nosis, and has features in common with atheroscler- as obesity, dyslipidaemia and insulin resistance. How- osis. Dyslipidaemia is common in PE and, via oxidation ever, more than half the women who develop PE remain of susceptible lipids, may contribute to endothelial acti- normotensive thereafter. There is a genetic component(s) vation. Oxidative ‘stress’ is increased in PE, perhaps to the disease, but it is most improbable that there is a through a variant of the hypoxia-reperfusion phenom- ‘PE gene’. Rather, there are factors such as genetically- enon in the developing intervillous spaces. Such early determined thrombophilias which are predisposers but changes might then lead to the clinically-evident not prerequisites. Impaired placentation is a feature, syndrome in susceptible women. PE is a protean, multi- with inadequate invasion of the spiral arteries by syncy- system, multifactorial disease, the causes of which are tiotrophoblast and poor remodelling. However, similiar only slightly less enigmatic than a decade ago. features are found in association with non-hypertensive Journal of Human Hypertension (2000) 14, 705–724

Keywords: pregnancy; hypertension; pre-eclampsia; gestational hypertension

Introduction collaboration with colleagues across numerous disciplines. It is a lamentable fact that, until relatively recently, However, pregnancy remains a higher-risk state there was little basic research in obstetrics. Obser- than non-pregnancy. In the most recent ‘Report on vation certainly, and consideration of management Confidential Enquiries into Maternal Deaths in the regimes, but investigations into the physiology and United Kingdom’,1 which is believed to have near- pathophysiology of pregnancy were the province of total ascertainment, pre-eclampsia/eclampsia a few dedicated individuals. Convulsions in preg- remain one of the two most frequently-cited causes nancy have been noted since at least the time of of maternal death, as they have for the last 30 years. Galen (4th century BC). As maternal death rates fell in the 20th century AD, at first slowly, then much In the United Kingdom, as in North America, the more rapidly after the introduction of antibiotics, absolute numbers of maternal deaths are relatively small1 (USA 9.2/100000 and UK 12.2/100000 and with improving antenatal and peri-partum care, 1,2 so the relative importance of the ‘toxaemias’ of preg- maternities; but in less developed countries, such nancy as causes of maternal death increased. They as Central and parts of South America and sub-Sah- have, however, always been one of the ‘Big Four’, aran Africa, rates, where available, can be 20–30 3 the others being infection, thrombosis and haemor- times higher (eg, Mungra et al, 1999; Etard et al, 4 rhage. A quick scan of Medline reveals that in 1969, 1999). a mere 22 articles with the keyword ‘pre-eclampsia’ As usual, the Greeks had a word for it. The word? or ‘preeclampsia’ were published. By 1979 this had Eclampsia, the thunderbolt, which strikes from the risen to 115 and to 285 in 1989. A scan in late 1999 blue, and can kill where it strikes. A prospective identified 468 such publications, and whereas 30 study which attempted to involve every consultant years ago, more than half of the papers were case obstetrician in the United Kingdom reported an inci- reports or related to management, those today are dence of eclampsia of 4.9/10000 maternities.5 far more likely to be concerned with basic science, Eleven percent had no recorded hypertension or utilising extremely sophisticated methodology and proteinuria at the antenatal clinic visit immediately preceding the onset of fitting, and a further 10% had proteinuria but no hypertension. Even in those who had been admitted to hospital, the highest recorded Correspondence: Professor F Broughton Pipkin, Department of diastolic pressure before the onset of convulsions Obstetrics and Gynaecology, University Hospital, Nottingham NG7 2UH, UK was 100 mm Hg or less in a third (34%). Diastolic Received and accepted 24 February 2000 pressures of 120 mm Hg or more before fitting were Hypertension in pregnancy F Broughton Pipkin et al 706 only recorded in 19% of patients. Thus it is unsafe supported by epidemiological observations that to consider eclampsia as being the end-point of sev- indicate a greater risk of recurrence and essential ere pre-eclampsia, although it can be associated. hypertension in later life for women with GH than Interestingly, a careful follow-up of 65 patients 6– those with PE. Additionally, many of the character- 42 months after eclampsia found that none of the istic markers of physiological abnormalities of PE women had neurological deficits after 6 months or such as increased cellular fibronectin concentrations repeated convulsions,6 emphasising that it is a preg- are not present in GH.12 If one imagines that GH may nancy-related condition. Of the 38 who sub- be a physiological mechanism to compensate for sequently had one or more pregnancies, none had a impaired utero-placental perfusion (see recurrence of fitting. ‘Conclusion’), and PE a pathological breakdown of The relatively low incidence of eclampsia, and its such a system, then the different outcomes become diversity of presentation, make systematic study of more comprehensible. It certainly seems improbable its pathogenesis very difficult indeed. This being so, that a condition should occur with such frequency the remainder of this paper will not consider it (GH ෂ10% in first pregnancy) at such a physiologi- further, and will focus on the several hypertensive cally important time of life if it were wholly malign. conditions found in pregnancy. Problems Definitions The fact that the defining symptoms of hypertension The question of the definition of the various forms and significant proteinuria arising de novo only of hypertension found in pregnancy has exercised become apparent late in the disease process, and are numerous groups, and serious attempts are being secondary, not primary, phenomena is a further made in 1999/2000 to arrive at an agreed definition obstacle to research. The seeds of PE are sown very for research purposes, which may be used world- early indeed in pregnancy. To compound the prob- wide. For purposes of this paper, gestational hyper- lem still further, the actual measurement of the tension (GH; also referred to as ‘pregnancy-induced blood pressure is notoriously unreliable, and hypertension’, PIH) is defined as the occurrence of inclusion criteria have differed from study to study. an increase of at least 30/15 mm Hg in the blood Worse still, the determination of the level of the pressure from pre- or early-pregnancy levels or an blood pressure even in normotension is itself multi- absolute value of at least 140/90 mm Hg or more on factorial. at least two occasions not less than 4–6 h apart in a woman known to have been normotensive before Heterogeneity of study groups 20 weeks of gestation and in whom the blood pressure has returned to normal by, at the latest, the Apparent PE occurring in a second or subsequent twelfth week post-partum.7 Korotkov sound V has pregnancy is associated with a considerably higher now been accepted as preferable for recording the incidence of hypertension in later life than if the diastolic pressure in pregnancy (eg, Shennan et al8). condition occurs only in the first pregnancy.13 Par- When this is accompanied by de novo significant ous women who develop PE or eclampsia are twice proteinuria (Ͼ300 mg/l or 500 mg/day in a 24-h as likely to die from ischaemic heart disease in later urine sample, or a minimum of two ‘pluses’ of pro- life than are women who develop PE or eclampsia tein on dipstick test), which also resolves after deliv- in their first pregnancy.14 Furthermore, renal biopsy ery, the condition is known as pre-eclampsia (PE). performed in primiparous and multiparous women Pre-eclampsia may also be accompanied by any/all diagnosed as having PE showed that while 84% of of haemolysis, elevated liver enzymes and a platelet primiparae did have diagnostic glomerular endo- count below 100 × 109 cells/l (thrombocytopaenia). theliosis, only 28% of multiparae did so.15 Thus When all are present, the condition is known as pathophysiological studies should logically concen- HELLP. Any of these conditions may be superim- trate on women who develop PE in their first preg- posed upon pre-existing essential hypertension. nancy. There is discussion as to whether GH and PE are Another problem is that the diagnosis undoubt- indeed part of the same spectrum of disease. This is edly encompasses more than one condition. For enhanced by the observation that GH at term is usu- example, in a prospective study of 212 primigravi- ally associated with normally-grown, or even dae, it was noted that 50 had had an isolated ‘spike’ slightly large, babies, particularly when centiles of systolic blood pressure of 140 mm Hg or more defined for the specific population are used during their first antenatal visit to their general prac- (reviewed in MacGillivray, 19839). Conversely, nul- titioner, but were normotensive by the time of their liparous women with PE have a four-fold greater first hospital antenatal clinic appointment. How- likelihood of delivering a baby small for its ges- ever, these women were twice as likely to be diag- tational age than do normotensive nullips10). Sau- nosed as having either non-proteinuric or pro- dan et al (1998)11 reported that 15–25% of women teinuric hypertension (PE) in pregnancy as those diagnosed initially with GH went on to develop PE, who did not show such a spike. Furthermore, other the likelihood being greater the earlier in pregnancy differences were noted between the groups (Table the hypertension developed. It may simply be that 1). later-onset GH does not have time to develop into An increased urinary calcium excretion is com- PE. Conversely the other 85% of women may have mon in essential hypertension (eg, Robinson, different, benign disease. This latter possibility is 198417), while renal vasoconstriction and reduction

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 707 Table 1 Values are shown as means with their s.e.m. or medians Speciﬁc study areas (interquartile range). All values differ signiﬁcantly over the two groups (P Ͻ 0.001 except urinary Ca/Cr for which P Ͻ 0.01) Genetics

No first SBP ‘spike’ Given all this, it is hardly surprising that research SBP ‘spike’ at first visit into the causes of PE has not proceeded faster. One area currently attracting increasing attention is that BMI (kg/m2) at first visit 23.1 ± 1.9 26.4 of the genetics of the condition. A pioneer of SBP at 20/52 (mm Hg) 120.5 ± 0.7 130.4 ± 1.4 research into PE, Professor Leon Chesley, showed in DBP at 20/52 (mm Hg) 66.6 ± 0.5 72.0 ± 1.0 his three-generation follow-up study that there is at Plasma AII at 28–32/52 44 (23, 75) 79 (46, 115) Urinary Ca/Cr at 28– 0.49 (0.29, 0.77) 0.79 (0.41, 1.08) least a genetic predisposition to the condition (see: 19 32/52 Lindheimer et al, 1999 ). All the requirements for scrupulous, standardised phenotyping come into Data from Broughton Pipkin et al, 1998.16 play here. This is no single-gene disease such as cys- tic fibrosis; we will not identify a ‘pre-eclampsia gene’. Rather, we are likely to be looking at the com- of renal blood flow are early companions of cardiac bined effects of several polymorphisms, probably in insufficiency, and will stimulate the renin-angioten- several different combinations, interacting with sin system. The angiotensin-converting enzyme their environment and eventually precipitating a inhibitors and, more recently, the angiotensin AT1 final common pathway of hypertension and pro- receptor blockers, are widely used in the treatment teinuria. of essential hypertension, and are highly effective Several recent reviews have considered the cur- (see Nicholls et al, 199818). It therefore seems likely rent state of knowledge of the genetics of PE (eg, that this form of ‘white coat hypertension’, under Morgan and Ward, 199920; Broughton Pipkin, the stressful circumstances of a first antenatal visit, 199921). In brief, there is some evidence for may be identifying a cohort of women who will be maternal-fetal sharing of a recessive gene(s), but a diagnosed as having GH or PE when in fact they polymorphism in a dominant gene with partial pen- have underlying chronic hypertension. etrance may also be implicated. Paternal genes The pathogenesis of GH and PE is almost certainly could contribute to the disease.22 Such studies are multifactorial. This being so, simplistic efforts to becoming possible with the increasing number of determine a single ‘cause’ are likely to be doomed women conceiving by more than one partner. to discouraging failure. Even when PE recurs in the Women who became pregnant by a partner who had same patient, it may do so in a rather different guise. fathered a pre-eclamptic pregnancy in another An example of this is given by a patient in Not- woman had nearly twice the predicted risk of them- tingham. Normotensive at the start of her first preg- selves having a pre-eclamptic pregnancy. There is nancy, she had by 24 weeks gestation suffered a rise also, of course, the possibility that the development of 35 mm Hg in both systolic and diastolic blood of PE is partly determined by a polymorphism in a pressure. By 3 days later, her blood pressure had paternally-imprinted, maternally-active gene which reached 200/120 mm Hg in spite of ␣-methyl dopa must be expressed by the fetus.23 Perhaps there is a therapy and her platelet count had dropped from threshold for either a necessary or a susceptibility 187 to 70 × 109 cells/l. Over the same period, her locus affected by fetal or other environmental influ- serum alanine transaminase (ALT) had risen from ences which may dictate who develops the disease. 22 to 140 U/l but the alkaline phosphatase, serum Several genetically-mediated thrombophilias have urate and albumin concentrations remained within been identified at higher incidence in women with the normal range. She was delivered by emergency obstetric complications, including PE.24 Two of the Caesarean section of a grossly growth-retarded major genes concerned are those for 5,10-methylene- infant who died shortly after delivery. Her blood tetrahydrofolate reductase (MTHFR) and Factor V. A pressure was 120/70 mm Hg by 2 weeks post-par- common mutation of the MTHFR gene (C677T) is tum. Thirty months later she conceived by another associated with increased plasma homocysteine partner, and was normotensive at 15 weeks ges- concentrations. Raised plasma homocysteine is tation. Again, she became severely hypertensive at associated with increased lipid peroxidation, 23 weeks gestation, with worsening hypertension in plasma triglycerides, and serum uric acid, as well as spite of therapy. Attempts were made to prolong the with blunted endothelial-dependent vasorelaxation pregnancy for the sake of fetal maturation, but at 28 and an altered phenotype of endothelium from anti- weeks, her blood pressure control was lost (180/115 coagulant to procoagulant. All these are features of mm Hg) and delivery was effected by Caesarean sec- PE. Three studies, of women of mixed parity, one tion. In this pregnancy, her platelet count over the Japanese, one Jewish and one Italian suggested an last few days ranged between 224 and 317 × 109 increased incidence of C677T in women with PE but cells/l and ALT between 16–25 U/l, with normal this was not confirmed in a study of white, nullipar- serum alkaline phosphatase and urate concen- ous women in Pittsburgh.25 In the Japanese study, trations. However, this time her serum albumin con- TT homozygosity was present in 30% of PE patients centration fell steadily as the disease worsened, to but only 11% of controls, but in the Pittsburgh study a nadir of 24 g/l in conjunction with urinary protein the proportions were 15% and 11% respectively. It output of 5.72 g/l. Thus although she clearly is worth noting that, as with many other mutations, developed pre-eclampsia in both pregnancies, the the mutant allele frequency varies widely by popu- associated system disorders were different. lation (see Table 2).

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 708 Table 2 Allele frequency for the C677T mutation in various population studied during, rather than after, preg- populations nancy, plasma angiotensinogen concentrations were actually lower in the TT homozygotes.43 The M235T Origin Allele Reference frequency mutation is in partial positive linkage disequilib- (control) rium with a dinucleotide repeat polymorphism at the 3Ј end of the gene. This is a very polymorphic Mexico 0.59 Mutchinick et al, 199926 site, which therefore allows tracking of allele trans- Ashkenazi Jewish 0.48 Rady et al, 199927 mission from mother to fetus. We have recently Japan 0.36 Sohda et al, 199728 reported a significantly increased maternal-fetal Greece 0.35 Antoniadi et al, 199929 USA ‘White’ 0.34 Powers et al, 199925 transmission of the A9 allele, which is associated Sweden 0.29 Brattstrom et al, 199830 with low plasma angiotensinogen concentrations, Italy 0.21 Cattaneo et al, 199931 in PE.43 A very recent study has shown that women with PE develop stimulatory autoantibodies against the angiotensin AT1 receptor.44 The only study to date Some 2–7% of the population carry the G1691A investigating possible linkage between AT1 receptor mutation (Leiden) of the Factor V gene, which genotype and the development of PE found no evi- results in the substitution of glutamine for arginine, dence that the three known polymorphisms in the making Factor V resistant to activated protein C. gene were associated with pre-eclampsia.45 How- Resistance to activated protein C is present in about ever, there is a dinucleotide repeat polymorphism 32 Ј one-fifth of women with pre-eclampsia and some in the 3 flanking region of the AT1 receptor gene, twice as many women with PE have the Factor V and a marked distortion of maternal-fetal trans- Leiden mutation as do normotensive women.33 mission of one of these alleles was shown in PE.45 Thus, polymorphisms associated with clotting dis- This allele is in partial linkage disequilibrium with orders may be predisposers for PE, but are evidently the C573T variant, which in normotensive preg- not pre-requisites. nancy is associated with higher levels of platelet ␣ ␣ 45 Tumour necrosis factor (TNF ) is a potent pro- AT1 receptors. An enhanced pressor response to inflammatory cytokine. TNF␣ mRNA expression is angiotensin II antedates the onset of clinically- significantly higher in leukocytes from pre-eclamp- detectable disease by many weeks.46 These obser- tic women.34 The high expression of TNF␣ may be vations may therefore be of functional significance. associated with the TNF1 allele, since the frequency It may perhaps also be relevant in this context, of this allele is markedly increased in pre-eclampsia. that one of the ‘best’ animal models of PE so far Furthermore, the common TNF␣ polymorphism at developed involves the renin-angiotensin system. −308 in the promoter region has been reported as Female transgenic mice carrying the human angiot- being associated both with raised TNF␣ mRNA and ensinogen gene were mated with males carrying the with PE.34 Concentrations of the soluble receptor for renin gene. The females developed hypertension TNF␣ (TNF␣ receptor 1; sTNFp55) are also from about two-thirds of the way through gestation, reportedly raised in pre-eclampsia before the clini- with much increased urinary protein excretion Sixty cal onset of the disease.35,36 However, the frequency percent of the mice died during pregnancy, but in of the TNF T2 mutation in the promoter region (the those which survived, the blood pressure had only polymorphism so far studied in this context) is returned to normal by 3 days after delivery. At auto- not increased in either pre-eclampsia or HELLP.37 psy, (paternally-derived) human renin was localised Another promising candidate gene appears to be to the chorionic trophoblasts. Other wild- the angiotensinogen gene. In the first published type/transgenic and transgenic/transgenic crosses study of pre-eclamptic patients,38 the variant M235T were not associated with the development of ‘pre- was found in significantly increased frequency in eclampsia’.47 women with PE, in association with a raised plasma The relationship of dyslipidaemias to pre-eclamp- angiotensinogen concentration. Similar findings sia is emphasised by the recent finding that poly- were reported in a case-control study of Japanese morphisms of the lipoprotein-lipase gene are five patients. However, this variant has also been times as common in women with well defined pre- reported in higher frequency in non-pregnant eclampsia as primiparas. Nearly 20% of these patients with essential hypertension (see: Caulfield women carried either the N291S or the D9N/-93G to et al, 199639) and coronary artery disease,40 again in T allele combination compared to 5% of controls.48 association with raised angiotensinogen concen- These polymorphisms are associated with increased trations. This must give rise to concern about the plasma triacylglycerol and decreased high-density possibility of underlying chronic hypertension in a lipoprotein cholesterol concentrations.49 proportion of these pregnant women (see: ‘Hetero- Mitochondrial DNA (mtDNA) encodes essential geneity of study groups’ above). Furthermore, recent subunits of the respiratory chain. It is critical for oxi- studies in Caucasian British, Caucasian Australian, dative phosphorylation and the production of cellu- Caucasian North American, African-American and lar energy stores in the form of ATP. Mitochondria Chinese populations have found no evidence for any are the major users of molecular oxygen within cells, increased frequency of the M235T variant in care- and so are also among the most important generators fully-characterised pre-eclamptic women compared of reactive oxygen species. Furthermore, mitochon- with normotensive pregnant controls (eg, Morgan et dria are enriched with polyunsaturated fatty acids, al, 199541; Guo et al, 199742). Furthermore, in a UK and are therefore very susceptible to damage by

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 709 lipid peroxidation. Over the last 10 years, a group clues to a unifying target in PE. Increased sensitivity of inborn errors of metabolism have been identified to pressors, platelet activation and loss of fluid from in which there is a failure in aerobic energy pro- the intravascular volume could all be explained by duction, considered to be due to mitochondrial dys- altered endothelial cell function. function. More than 50 mtDNA mutations and sev- Endothelial cells, once considered as merely for- eral nuclear gene mutations have been identified in ming a barrier to prevent the contact of vascular association with specific disease states. A common smooth muscle and collagen with intravascular con- 4977-bp deletion affects genes encoding seven poly- tents, are now recognised as subserving sophisti- peptide components of the mitochondrial respir- cated functions maintaining vascular homeostasis.57 atory chain, and five of the 22 tRNAs necessary for These functions are pertinent to pregnancy and PE. mitochondrial protein synthesis. Neither this nor Thus, normal endothelium produces factors that the 7436-bp deletion have been identified in buffer the contractile response of subjacent vascular samples of PE mitochondria,50,51 nor have any of the smooth muscle to circulating pressors. The endo- three most common mtDNA mutations been found51 thelial products, nitric oxide and prostacyclin, have (see also: Mitochondrial function below). such vasodilator effects but in addition inhibit plate- Candidate genes considered, and on available evi- let aggregation. Other molecules produced by endo- dence put aside, include those for renin, angioten- thelium are responsible for the generation of normal sin-converting enzyme, endothelial nitric oxide syn- anticoagulants. Additionally, vascular endothelium thetase and superoxide dismutase. Although the maintains normal intravascular volume by several human leukocyte antigen (HLA) system has been mechanisms in addition to the obvious barrier func- implicated, many of the findings reported have been tion. inconsistent or contradictory.52 In normal pregnancy vascular volume is expanded and cardiac output increased yet blood pressure is Future research in genetics: The search for candi- reduced. In addition, sensitivity to angiotensin is date genes has so far been conducted on small reduced.58 The differences in vascular response can samples with varying stringency in phenotyping. be demonstrated in controlled in vitro settings in The future has to lie in the accumulation of large small myometrial resistance vessels59 and omental samples, at the least from mother, father and new- vessels.60 They are largely eliminated by removal of born, which can be used for model-free analyses endothelium59 although little or no difference has such as transmission disequilibrium testing. This is been reported in other resistance arteries.61 This has made possible by the recent rapid advances in geno- led to the concept that at least a portion of the car- typing technology, which will allow genome-wide diovascular adaptations of normal pregnancy is sec- scanning and the identification of ‘hot spots’. That, ondary to modifications of endothelial function, in of course, is only one step, but a very important one. some vascular beds. The special importance of Such a study, sampling from mother, father, new- endothelium in normal pregnancy is further indi- born and mother’s parents, has recently been funded cated by the increased frequency of endothelial dis- by the British Heart Foundation as a 10-centre col- eases such as thrombotic thrombocytopaenic pur- laboration in the United Kingdom (the GOPEC pura and hemolytic uremic syndrome in association study); recruitment will begin in early 2000. with pregnancy.62 With injury or activation endothelial function is strikingly modified. Not only are vasodilator func- Endothelial dysfunction in pre-eclampsia tions and anticoagulant functions downregulated The woman with PE manifests a diverse array of but endothelial cells can also produce vasoconstric- pathophysiological changes.53 These include pro- tors and manifest procoagulant function. From this found abnormalities of hepatic, renal and cerebral description, increased sensitivity to pressors and function accompanied by activation of the coagu- activation of the coagulation cascade, the earliest lation cascade. Many of these changes can be pathophysiological changes of PE, are consistent explained by reduced organ perfusion, but the pro- with alteration of endothelial function. This has led foundly disordered physiology of the woman with to the hypothesis that endothelial dysfunction is a PE makes it impossible to separate cause and effect. central pathophysiological feature of PE. The A better insight into the sequence of pathophysiol- hypothesis also proposes that the poorly perfused ogical changes comes from attention to those placenta characteristic of PE generates factors that changes that antedate clinically evident PE. Acti- promote alteration of endothelial cell function.63 vation of the coagulation cascade, especially alter- Several lines of evidence support the involvement ations in platelet function, can be demonstrated of endothelial cells in the pathogenesis of PE.64 One weeks to months before clinically evident dis- of the most characteristic pathological findings in PE ease.54,55 Similarly, increased sensitivity to angioten- is the renal abnormality termed glomeruloendo- sin46 and reduced plasma volume56 have been found theliosis.65 This lesion consists of marked glomeru- in groups of women destined to develop PE. It is lar capillary swelling (often sufficient to occlude the important to point out that these changes are not capillary lumen), basement membrane inclusions universally present and thus can only be recognised and virtually no change in the renal epithelial podo- in groups of women destined to develop PE. Overlap cytes. This finding not only emphasises that PE is of the values of normal and pre-eclamptic women is not merely a variant of essential hypertension but such that the available markers are not valuable as supports the concept that endothelial cells are a tar- predictors. Nonetheless, these changes provide get in the disorder. Several markers of endothelial

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 710 Table 3 Evidence of endothelial activation in pre-eclampsia cated the presence of more than one factor responsible for these activities.66 Markers in blood Many of the risk factors that predispose to the Increased von Willebrand Factor (vWF)a74 functional endothelial alterations of atherosclerosis Reduced prostacyclin excretiona73 also come into play in PE. Elevated homocysteine,67 a75 Increased growth factor activity 68 222 elevated triglycerides with associated small dense Increased endothelin 69 Increased prevalence of antiendothelial antibodies80 low-density lipoprotein (LDL), and insulin resist- Increased thrombomodulin223 ance70 are all associated with increased risk for both Increased thromboxane224 disorders. Some of these changes such as elevated Increased cellular fibronection (cFN) concentrationa 225 homocysteine persist beyond pregnancy, yet the PE Increased plasminogen activation activity (PAI-1)226 Increased soluble VCAM227 syndrome and the attendant endothelial dysfunction Increased tissue plasminogen activator228 abate with delivery. This observation provides Increased ICAMa 229 further evidence supporting the unique sensitivity Increased E-selectin230 of endothelial cells during pregnancy. An important 231 Increased tissue factor recent observation may explain this unique sensi- Increased ELAM232 tivity. The Oxford group examined inflammatory Differences of in vitro vascular responses of vessels from markers on circulating blood cells of normal preg- pre-eclamptic women nant and non-pregnant women, women with PE and 71 Reduced endothelial dependent relaxation of subcutaneous non-pregnant women with sepsis. They found vessels233 striking increases in markers of inflammation in nor- Reduced relaxation of subcutaneous vessels to bradykinin234 mal pregnant compared to non-pregnant women that Reduced endothelial dependent relaxation of epigastric were increased only slightly more in women with arteries235 Impaired endothelial mediated relaxation in omental PE. The increased inflammatory response of normal vessels236 pregnancy may provide at least a partial explanation Reduced NO dependent relaxation of myometrial vessels237 for the increased sensitivity of vascular endothelium during pregnancy. aCan be demonstrated prior to clinically evident pre-eclampsia. What is the factor that alters endothelial function during pregnancy? The nature of endothelial dys- activation can be demonstrated in women with PE function during pregnancy may provide some (Table 3). Most importantly, several of these are insight. Endothelial dysfunction may be due to acti- manifest weeks to months before clinically evident vation of usual functions in an inappropriate setting disease, supporting the central role of altered endo- or, alternatively, physical injury of the cells. The thelial function. In vitro studies also provide evi- endothelial dysfunction of PE, especially that dence for products in the blood of pre-eclamptic present before the overt syndrome, is more consist- women that alter endothelial function. Endothelial ent with activation than with mechanical injury. functions can be modified by either exposure of Blood concentrations of thrombomodulin, an endo- endothelial cells in culture or isolated blood vessels thelial protein only released with actual injury, are only increased with clinically evident PE.72 By con- to plasma or serum from women with PE (Table 4). 73 With this approach several markers of endothelial trast, reduced prostacyclin production, increased circulating von Willebrand factor74 and growth fac- injury or activation (eg, cellular fibronectin (cFN), 75 platelet-derived growth factor (PDGF)) or endo- tor activity, features of endothelial activation are thelial functions (vasodilatation in vitro) are altered altered weeks to months before clinically evident more by exposure to serum/plasma from women PE. The relevance of in vitro effects of PE with PE than to serum/plasma from normal pregnant serum/plasma to the in vivo setting is open to ques- women or pre-eclamptic women after delivery. Pre- tion. With this proviso, in vitro studies do not sup- liminary characterisation of these factors has indi- port a generalised toxic effect of serum/plasma from women with PE. The effects are relatively selective, with some markers of activation being stimulated Table 4 In vitro effects of serum/plasma from pre-eclamptic while others are not. Also, in most settings cells women appear normal and maintain structural integrity.76 Thus, the nature of the dysfunction would suggest Effects upon cultured endothelial cells that the responsible factor is not a toxin but more Increased Cr51 release238 likely a usual factor present inappropriately or at 239 Increased lipid uptake higher than usual concentration. Increased expression of PDGF ␤-chain mRNA240 Increased prostacyclin release241 Several factors have been suggested. These Increased cFN release76 include cytokines, TNF␣ in particular,77 activated Increased nitric oxide (NO) release242 neutrophils, monocytes or platelets,78 stable pro- Increased nitric oxide synthase (NOS) expression243 ducts of lipid peroxidation,79 syncytiotrophoblast Reduced endothelial cell proliferation244 Increased VCAM on cell surface245 fragments and antibodies. The antibodies suggested 80 Increased permeability246 include autoantibodies to endothelial cells and more specifically autoantibodies to the angiotensin Effects on ex vivo vessels 44 (AT1) receptor that are capable of activating the Reduced endothelial dependent relaxation of myometrial receptor. With attention to the cautions above, pre- vessels247 liminary characterisation of the activities that alter endothelial in vitro indicates that one of these

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 711 resides in the LDL fraction.66 The question as to the with atherosclerosis.89 Thus, the dyslipidaemia of identity of the responsible factor(s) is far from answ- PE acting through oxidative modification of suscep- ered but as will be discussed below, an interesting tible lipids may contribute to the altered endothelial theme present with the majority of these suggestions responses of the syndrome. If this concept is vali- is a relationship to oxidative stress. dated by future research, dyslipidaemia may serve as a target for preventive therapy either prior to or Future research on the endothelium: The in early pregnancy. increased evidence supporting the role of endothelial dysfunction in the genesis of PE has interest- Mitochondrial function ing implications. Could a marker of endothelial dysfunction be used as a more specific diagnostic In 1989, Torbergsen et al90 described carefully- marker of the disorder that currently used signs? characterised mitochodrial dysfunction in a family Would markers of endothelial dysfunction serve as with a high incidence of PE and eclampsia. The a useful indicator successful preventive therapies? recent upsurge of interest in the potential roˆle of What allows change in endothelial function in some lipid peroxidation and increased free radical con- but not all vascular beds? centration in the pathogenesis of PE (see section: Lipids in pre-eclampsia) has led to an examination of the cellular localisation in the placenta of pro- Lipids in pre-eclampsia teins specifically modified by lipid peroxidation. The alteration of circulating lipids is one of the more While not all placentae from PE pregnancies show striking metabolic changes of pregnancy. Normal evidence of peroxidation in the trophoblast, where pregnancy is associated with an approximate three- such modified proteins are identified, they can be fold increase in triglycerides and fatty acids, and a localised in the mitochondria.91 Malonyldialdehyde 50% increase in LDL cholesterol. High-density lipo- concentrations are a crude marker of lipid peroxi- protein (HDL) cholesterol is also increased.81 The dation; such concentrations were significantly triglyceride, fatty acid and LDL cholesterol changes higher in mitochondrial fractions from placentae are accentuated but HDL reduced in PE.68,82 Changes from PE than from normotensive pregnancies.92 in fatty acids and triglycerides antedate clinically Mitochondrial fractions from PE placentae were also evident PE.83 It is not known if they antedate preg- associated with greater stimulation of lipid peroxi- nancy. However, dyslipidaemia can be demon- dation by such agents as NADPH.92 strated to be more prevalent in later life (20 to 30 years later) in women with recurrent PE.84 Of the Maternal-fetal interactions: insulin resistance and lipid changes in pregnancy and PE perhaps the most the maternal syndrome surprising is a shift in the spectrum of LDL sub- classes toward proportional increases in smaller, Implicit to the understanding of PE is the concept denser LDL particles.69 The presence of the small that the disorder is a two-stage disease. The abnor- dense LDL variant is predictable from the elevated malities of placentation described below and the triglycerides associated with pregnancy but is sur- risk factors for PE that would be associated with prising from the perspective that this lipoprotein reduced placental perfusion (Table 5) implicate a moiety is considered most pathophysiologically rel- compromised blood supply to the placenta as an evant in atherosclerosis.85 Small dense LDL have integral component of the disorder. Further support increased access to the subendothelial space where for the concept comes from obstetrical risk factors they also reside longer than other LDL particles. In for pre-eclampsia. These include the large placental this setting, the particle is sequestered from the pro- mass associated with multiple gestations, hyda- tective effects of circulating antioxidants. In tidiform moles and hydropic placenta that are pro- addition, small dense LDL are inherently more eas- posed to result in a relative reduction of uterine per- ily oxidized. In atherosclerosis, this combination is fusion.93 proposed to lead to the formation of oxidized LDL However, it is evident that abnormal placental (ox-LDL) with subsequent alterations of endothelial perfusion (Stage 1) is not sufficient to explain the function.86 Additionally, increased VCAM maternal systemic syndrome (Stage 2) of PE. Many expression in response to ox-LDL recruits mono- growth restricted infants (IUGR) that must have cytes. Oxidized LDL is taken up by the monocytes received inadequate placental blood flow come from with subsequent formation of foam cells and the generation of the atheromatous plaque. It is tempt- ing to speculate that the same changes occur in PE Table 5 Risk factors for pre-eclampsia and account for altered endothelial function. The Hypertension morphological changes of atherosclerosis have not Diabetes been reported in the systemic circulation of women Collagen vascular disease with PE. However, decidual vessels do manifest a Thrombophilias morphological abnormality termed atherosis.87,88 Hyperhomocysteinemia This alteration includes fibrinoid occlusion of the Increased androgens Gestational diabetes vessels that are surrounded by foam cells, changes Black race reminiscent of atherosclerosis. Functionally, the alt- Insulin resistance ered endothelial responses of women with PE are ? Dyslipidaemias similar to the blunted endothelial responses present

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 712 pregnancies that are not complicated by PE. Even unknown. In a smaller study it was evident that more strikingly, the abnormal placentation charac- women who had pregnancies that were not compli- teristic of PE is also present in the implantation sites cated by PE had a lower risk of cardiovascular dis- of pregnancies complicated by IUGR and one-third ease than the unselected female population.13 This of those complicated by preterm birth.94,95 This has would suggest common risk factors for PE and car- stimulated the hypothesis that the abnormal per- diovascular disease. This is further supported by fusion must interact with maternal constitutional other risk factors for PE that are not part of the meta- factors to generate the maternal syndrome.96 What bolic syndrome. Elevated homocysteine67,110 and are these maternal predispositions? Table 5 lists risk black race111 are associated with increased risks of factors for PE. Note that whereas some of these fac- both conditions. tors could relate to PE by effects on placental perfusion, for many this relationship seems unlikely. Future research However, most of these are associated with insulin resistance and/or the ‘metabolic syndrome’.97 The number of genetic, environmental and behav- This syndrome has been described as associated ioural factors that predispose to the metabolic syn- with increased cardiovascular risk and includes drome could provide targets for prepregnancy ther- obesity, dyslipidaemia, hypertension, insulin resist- apy. This diversity of factors also emphasises the ance and, interestingly, elevated plasma uric acid heterogeneous nature of predisposing factors for concentrations. How this constellation of abnor- pre-eclampsia. mality relates to cardiovascular disease is not clearly established. There is evidence of reduced endo- Oxidative stress as the linkage for the two stages thelial mediated vasodilatation with the syn- of PE drome.98 The postulated effects of the dyslipidaemia are discussed above and the mechanisms of elevated Oxidative stress, an excess of free radicals and blood pressure to generate cardiovascular disease reactive oxygen species over antioxidant scavengers, are well established. However, direct contributions has been implicated in several disease states includ- of insulin resistance or the accompanying hyperin- ing atherosclerosis. The formation of these reactive sulinaemia are not clear. Also, where the circle species results in alterations of cellular function begins is also open to question. For example, insulin with endothelium particularly vulnerable. The simi- resistance will result in increased free fatty acids larities between atherosclerosis and PE, the sug- while free fatty acids increase insulin resistance.99 gested role of oxidative stress in the metabolic syn- Regardless of mechanisms, abnormalities that are drome and the involvement of oxidative stress in the components of the metabolic syndrome are associa- activities of many of the agents proposed to alter ted with and likely predispose to PE.100 Pre-preg- endothelial function in PE, stimulate the hypothesis nancy obesity has a dose response relationship to that oxidative stress may provide the linkage PE.101 As obesity increases so does the risk of PE. between the two stages of PE.112 Increased androgens are present in women with PE There is abundant evidence that oxidative stress and women with a history of PE.102,103 The dyslipi- is increased in PE. There are increased concen- daemia of PE is identical to that present in the syn- trations of stable metabolites of the reactive lipid drome. It seems likely that these changes in a more peroxides, malondialdehyde and isoprostanes in subtle form antedate pregnancy. Alterations of fatty blood and tissues.79 In addition, plasma from acids and triglycerides are present from early preg- women with PE oxidizes ascorbic acid in vitro. Not nancy83 and women with PE are five times as likely surprisingly, circulating concentrations of this to have function perturbing mutations of the lipo- lynchpin antioxidant are reduced in women with protein lipase gene.104 Women with PE have PE, likely indicating consumption in vivo.113 increased circulating insulin,105 and PE is more What is the source of free radicals? A prime mech- common in women with gestational diabetes.106 anism for the generation of free radicals is the Using ‘minimal model’ analysis of response to intra- hypoxia reperfusion phenomenon.114,115 In this venous glucose, insulin resistance was increased in scenario, oxygen delivery to a tissue is reduced such women with PE.107 The likelihood that this change that ATP can no longer be oxidized and is metab- may have antedated pregnancy was indicated by the olised to adenine. The bifunctional enzyme xan- persistence of increased insulin resistance for 12 thine oxidase dehydrogenase increases in response weeks post-partum. Similarly, women re-examined to hypoxia and is preferentially converted to its oxi- more than 10 years after a pre-eclamptic pregnancy dase form. In the dehydrogenase form the enzyme had higher fasting insulin concentration.108 catalyzes the conversion of purines including the The metabolic syndrome is associated with increased content of adenine to NADH and uric increased risk of cardiovascular disease. However, acid. By contrast, the products of the xanthine oxi- the classic studies of Chesley have been interpreted dase form of the enzyme are the superoxide radical, to indicate that with well defined PE there is no hydrogen peroxide, and uric acid. Superoxide and increased risk of cardiovascular disease in later hydrogen peroxide are buffered by available antioxi- life.109 It is important to point out that although the dants until they are consumed. The consequences of women in the control group of Chesley’s studies the generation of free radicals and antioxidant con- were matched for factors that might influence car- sumption is a second wave of tissue injury when diovascular disease and hypertension (age, race, oxygen is restored to the previously hypoxic tissue. socio-economic status), their pregnancy history was The intervillous space would seem to be particularly

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 713 vulnerable to this process. The blood flow of the practice, the effectiveness and even more uterus varies widely with daily activities and inter- importantly fetal safety, must be proven in larger villous blood flow is directly reduced by myometrial studies. contractions. Nonetheless, the variations of blood flow are usually well tolerated indicating the pres- The cardiovascular system ence of protective homeostatic mechanisms. For example, the concentration of xanthine oxidase It is increasingly apparent that many of the cardio- dehydrogenase is very low in the placenta.116 It is vascular adaptations to normal pregnancy are possible, however, that in a setting of reduced pla- initiated in the luteal phase of every menstrual cental perfusion, as is characteristic of PE, these cycle, and are merely amplified should conception mechanisms may be overcome and oxidative stress occur. For example, careful prospective studies of ensue. Support for this speculation is the demon- ovulatory cycles show significant falls in mean stration of increased xanthine oxidase mass and arterial pressure (MAP) and total peripheral resist- activity in the invasive cytotrophoblasts of women ance (TPR) in the mid-luteal phase, with a rise in with PE.117 Another potential source of free radical heart rate (HR), cardiac output (CO), renal plasma generation is elevated free iron which may be flow (RPF) and glomerular filtration rate (GFR) (eg, present in PE. Free iron can catalyse free radical pro- Davison and Noble, 1981123; Manhem and Jern, duction and lipid peroxidation.118 1994124; van Beek et al, 1996125; Chapman et al, Local formation of free radicals in the intervillous 1997126). The renin-angiotensin system (RAS) is also space or decidua could be transmitted to endo- activated (eg, Brown et al, 1964127; Kaulhausen et al, thelium by several of the suggested mediators of 1978128; de Hertogh et al, 1989129). These changes endothelial activation in PE. Stable products of lipid are all amplified should conception occur. They are peroxidation such as MDA are themselves capable thus proactive, not reactive. of oxidant activity. Also, neutrophils and other The TPR has been reported as being significantly blood elements activated by exposure to oxidative decreased by at least 6 weeks gestation age,130 by stress in the intervillous space could release reactive which time CO has also risen,130,131 through oxygen species on contact with endothelium. The increases in both stroke volume (SV) and HR. It may role of placental fragments may also relate to local be that the first response to pregnancy is the fall in oxidative stress. There is evidence that villous mem- vascular resistance. This in turn would cause a rela- branes are oxidatively modified in PE119 and frag- tive ‘underfilling’, associated with a fall in arterial ments of these modified tissue could induce distant blood pressure and concomittent increase in HR and endothelial oxidative modification. The altered SV. The raised circulating concentrations of plasma membrane fluidity of these membranes would fav- renin concentration, plasma renin activity, AII and our their shedding, as would apoptosis known to be aldosterone (ALD) in the luteal phase of the cycle induced by oxidative stress.120 Likewise, activation do not fall, but continue to rise should conception of the angiotensin II receptor by autoantibodies occur (Sundsfjord and Aakvaag, 1973132; Chapman could result in activation of the enzyme et al, 1998130). The RAS is thus one of the earliest NADH/NADPH-oxidase with the generation of hormone systems to ‘recognise’ pregnancy, and may reactive oxygen species.121 well be being activated as a response to the per- The consequences of the generation of free rad- ceived underfilling. icals will be influenced by maternal factors includ- If, as seems more and more likely, PE is initiated ing antioxidant capability, the presence of other pro- by a failure of adaptation to pregnancy, then one moters of oxidative stress (eg, homocysteine) and the might expect to see very early differences in women presence of susceptible substrate such as small who go on to develop PE. This, however, requires dense LDL. Thus, oxidative stress provides a plaus- very large-scale prospective studies throughout ible linkage between reduced placental perfusion pregnancy, beginning, preferably, before concep- and endothelial dysfunction that is influenced by tion. Such pre-conception studies have not been maternal factors. This hypothesis was directly tested done, but two prospective studies of nulliparae have in a recent trial of antioxidant therapy.122 The inves- been reported. Easterling et al (1990)133 reported a tigators proposed that treating women at risk for PE longitudinal study of 179 women, from whom serial with antioxidant dosages of vitamin E and C from measurements of blood pressure, CO (Doppler) and 22 weeks’ gestation would reduce evidence of endo- derived TPR were made from at least 22 weeks ges- thelial activation. They measured the ratio of plas- tation. Nine were described as having developed PE. minogen activator I (made by activated endothelial Unfortunately, the definition of PE used required cells) to plasminogen activator II (made by the only the presence of 1+ protein or more which is placenta) as a marker of endothelial activation. They now not considered to be sufficiently stringent (see: reported that antioxidant therapy was associated Definitions and Brown and Buddle, 1995134); only with a 21% reduction in the ratio, consistent with two of their patients would be classified today as reduced endothelial activation. Surprisingly, there having PE. With this caveat, the data are, however, was also a significant reduction in the incidence of interesting, in that they show a relatively raised CO PE in this small (140 in each group) study. These and MAP from late in the first trimester in women findings are quite interesting and support the role of who went on to develop a raised blood pressure oxidative stress in the genesis of endothelial acti- with non-significant proteinuria. The TPR in these vation in PE. However, before the exciting possi- women was slightly but not significantly lower than bility that this therapy prevents PE moves to clinical in those who remained normotensive. A prospective

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 714 study of 400 nulliparae, using the more stringent was either unchanged144 or somewhat definition, has also recently reported a raised CO increased.145,149 This suggests that the abnormalities and unchanged TPR prior to the onset of clinical of tone found in established disease represent PE,135 with a progressive fall in CO and rise in TPR homeostatic, rather than primary pathological, as the disease became manifest. Interestingly, those mechanisms. Women with severe PE have reduced who developed GH only retained a relatively raised venous distensibility antepartum, but are indis- CO until delivery. In established PE, CO has been tinguishable after delivery from women who were measured by Swan–Ganz thermodilution immedi- normotensive during pregnancy.147 Their forearm ately prior to the initiation of therapy and shown vascular resistance and tone are higher than those to be significantly lower, and TPR higher, than in recorded in women with GH.150 normotensive controls136. Thus, in the first trimes- What factors might be driving these changes? ter, a proportion of nulliparae show a hyperdynamic Ambulatory blood pressure monitoring shows that circulatory response (future GH), with a later switch in normal pregnant women, as in the non-pregnant in a small proportion of patients (future PE) in state, the blood pressure falls at night, but in women response to an as yet unknown stimulus. The total with PE, this fall is blunted,151 and even reversed in blood and plasma volumes are both reduced in ෂ20%.152 The HR variability is decreased in normal established PE, with a smaller fall in both in pregnancy153,154 in the face of a normal increase of GH;137,138 the timing of this volume contraction has 10–15 beats/min. The baroreflex sensitivity, the not been established. Interestingly, in PE the total linkage between changing arterial pressure and HR, extracellular fluid volume (ECFV) is unchanged, so has not been intensively studied in pregnancy, and that the ratio of plasma volume: ECFV is signifi- data available are conflicting. An increased barore- cantly reduced compared with normal pregnancy,137 flex sensitivity during normal pregnancy was presumably in association with a ‘leaky’ circu- reported in earlier studies,155,156 the inference being lation.139 that the lower blood pressure of pregnancy is ‘per- There is a small fall in systolic, and a greater fall ceived’ as normal, and that the baroreceptors have in diastolic blood pressure during the first half of been reset. However, in subsequent studies using pregnancy in both normal and future hypertensive cross-spectral analysis and non-invasive blood women (eg, Christianson, 1976140; Halligan et al, pressure monitoring, a decrease in baroreflex sensi- 1993141). It is important that the blood pressure be tivity has been reported from the second trimester recorded at the first antenatal clinic visit to give the of normal pregnancy,154 and in the third trimester base against which subsequent change can be by comparison with non-pregnant women157 or with assessed (see: Definitions). The blood pressure rises the same subjects post-partum.158 The changing car- steadily thereafter, in both normal and future hyper- diovascular response to head-up tilt as pregnancy tensive women, but the rate and amplitude of the progresses also suggests a fall in baroreflex sensi- rise is accelerated in women who develop GH or PE. tivity.159 A further decrease in baroreflex sensitivity Longitudinal studies in primigravid pregnancy have has been reported in PE.157,160 Indeed, Wasserstrum shown that women who go on to develop GH or PE et al (1989)160 reported that: ‘In patients with higher have relatively raised blood pressures at the first baseline blood pressures, the severe impairment of antenatal clinic visit, but since these are still well baroreflex function eliminated the normal circulat- within the normotensive range, and the overlap is ory buffer against vasodilator-induced hypotension’, considerable, this has a low specificity in predicting presumably accounting for the profound falls in women at risk (eg, Masse´ et al, 1993142,143). blood pressure which can be seen in such patients Forearm blood flow is increased by up to 30% in during the administration of hydrallazine. normal pregnancy (eg, Martin et al, 1997144; On the assumption that the fall in blood pressure Anumba et al, 1999145). Normotensive pregnancy is initially, and the subsequent rise, accelerated in GH associated with a progressive venodilatation.146 and PE represented changing sympathetic tone, There are changes in venous distensibility and Assali et al (1952)161 examined the effect of ganglion capacitance, to accommodate the increased flow on blockade with TEAC. To their surprise, the adminis- the venous side, both distensibility147 and capaci- tration of TEAC was associated with greater falls in tance being higher in the normal pregnant than in BP in pregnant than non-pregnant women. This the non-pregnant woman. In non-pregnant women, enhanced response returned to non-pregnant levels venous capacitance is greater in the leg than in the during the third trimester. This suggested activation forearm, but in pregnancy, this difference is lost, of the sympathetic nervous system, presumably in presumably because of the haemodynamic effect of response to some (hormonal?) vasodilator stimulus. the gravid uterus.144 Vasodilator hormone synthesis and excretion is cer- Women with GH have significantly-raised forearm tainly markedly enhanced from early in normal venous tone by comparison with normal pregnant pregnancy, presumably to facilitate expansion of the women,148 but interestingly, a prospective study plasma and blood volumes. For example, the uri- showed this to be preceded by a period of increased nary excretion of prostaglandin 6-keto F1␣ is signifi- venodilatation.146 When the central blood volume in cantly increased by the end of the first trimester73 such women was increased by a period of passive while the excretion of kallikrein is increased from leg elevation, the women with GH showed venodil- at least 16 weeks gestation.162 The local generation atation, unlike normal pregnant women or non-preg- of nitric oxide (NO) may also be involved; there is nant women, who showed no change.148 Further- increased NO activity in the veins of the dorsum of more, forearm blood flow in women with GH or PE the hand in pregnancy.163 Blood flow itself is a major

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 715 regulator of endothelial NO synthesis, and Cockell nancy have an enhanced pressor response to AII (eg, et al (1997)164 have shown in small arteries from fat Talledo et al, 1968);179 this is foreshadowed by a biopsies in healthy pregnant women that NO more rapid rise in pressor response to AII from at responses to flow are enhanced by comparison with least the second half of pregnancy in such women46 those from non-pregnant subjects. Furthermore, the and paralleled by increased platelet AII bind- forearm vascular response to NO synthase inhibition ing.180,181 However, in women with established GH is enhanced in normal pregnancy by comparison or PE, the pressor response to other agonists is also with non-pregnant women, also suggesting a roˆle for enhanced, implying a general underlying dysfunc- NO in the vasodilatation of pregnancy.145 tion. A probable early mechanism for this has been Might pressor responsiveness to vasoconstrictors described above (see: Endothelial dysfunction in be reduced? Normal pregnancy is indeed associated pre-eclampsia). However, there is also a later con- with a decrease in the pressor response to AII, first tributory factor, in the shape of increased smooth described nearly 40 years ago.165 However, this is muscle intracellular free calcium concentrations, specific to AII, and is not due to a generalised which is seen in established PE but not in GH.182 decrease in vascular responsiveness, as the response The mechanism for this has not been elucidated, but to noradrenaline has been repeatedly described as there are known abnormalities in renal calcium being unchanged.166–169 The pressor responsiveness handling in PE (see below: Renal function). to AII returns to non-pregnant levels in late ges- tation46 in parallel with the rise in blood pressure. Future research in the cardiovascular system: The The pressor response to AII is mediated via the type available observations suggest that at least some of AT1 receptors. When platelets have been used as the cardiovascular changes of PE are initiated during models of vascular smooth muscle, decreased AII the first months of pregnancy. A main aim of future binding has been shown from early in the first trim- research must therefore be the study of very early ester,170 which may result in the decrease in pressor pregnancy, preferably including pre-conception response. Again, this change begins to return to non- data. Such data are very time-consuming to collect, pregnant levels in the second half of gestation. The but are invaluable as a resource. Non-invasive fall in platelet AII binding has also been observed methods of measuring cardiac output, blood press- in the second half of the menstrual cycle,171 and is ure, blood flow and baroreflex sensitivity are now therefore another proactive change, preparing for available, and may be used in sequential studies. pregnancy. The reported changes in baroreflex sensitivity in Assali et al (1952)161 also studied the effect of normal and hypertensive pregnancy have not been TEAC in women with ‘PE’, and showed it to have a considered in the light of mechanisms. In such stud- lesser effect, suggesting that there was withdrawal, ies attention should be directed to the variance of not activation, of sympathetic tone in these women. the data. It is quite likely that not all women, for Differing results have, however, been reported example, will manifest increased cardiac output or recently, using power spectral analysis. This altered baroreceptor response and that any cardio- showed a pronounced increase in power in the very vascular change could be a predisposing factor to low-frequency range of heart rate variability, sug- other more universal pathogenetic mechanisms. gesting a marked increase in sympathetic tone.172 There is increasing awareness of similarities The urinary excretion of both prostaglandin 6-keto between insulin resistance and some features of F1␣ and kallikrein is reduced long before clinical established PE (see above: Maternal fetal interac- signs of the disease are apparent (eg, Fitzgerald et tions: insulin resistance and the maternal al, 198773; Millar et al, 1996162; Mills et al, 1999173), syndrome).These features may persist for at least 3 perhaps again reflecting early endothelial damage. months after pregnancy,70,107 with potentially detri- There is considerable interest in the possibility of mental effects on future health. This is an emerging decreased endothelial NO synthesis contributing to area of research with potentially wide-reaching the vasoconstriction of PE. Adequately diet-con- consequences. Indeed, from the point of view of the trolled studies of (nitrite+nitrate) excretion have yet well-being of the individual patient, long-term fol- to be performed in women with PE, although a study low-up studies should also be initiated, not only of of plasma concentrations after one day of dietary the mother, but also of her baby. control reported higher levels than in normotensive 174 145 pregnant women. Anumba et al (1999) reported Renal function no difference in response to inhibition of NO synthase between normal and pre-eclamptic pregnant ‘Glomerular endotheliosis’ is unique to PE, and women. The surrogate measurement of cGMP will resolves post-partum.65 There is glomerular also reflect plasma concentrations of atrial natriur- hypertrophy, with swelling of the glomerular endo- etic peptide,175 which have also been reported to be thelium and increased glomerular wall thickness. higher in PE (eg, Thomsen et al, 1987176; Irons et al, The glomerular tufts appear obstructed, with nar- 1997177). Indeed, one report suggests that a rise in rowing of the glomerular capillary loops (see: Con- plasma ANP antedates the clinical diagnosis of rad and Lindheimer, 1998183). Lipid droplets may be PE.178 Thus two major vasodilatatory systems seem present in the cytoplasm (‘foam cells’) and there to have enhanced activity in hypertensive preg- may be fibrin deposition. nancy, presumably as a secondary response to the On the basis of available evidence, it appears raised pressure. likely that the undoubted changes in renal function Women who have become hypertensive in preg- in PE are secondary, rather than primary. Thus,

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 716 there is not a major research effort being devoted to in serum albumin is enhanced in PE, and concen- the topic at present. There is an excellent recent trations may reach values as low as 20 g/l. The over- review of renal function in PE,183 to which the inter- all pattern of change in serum proteins resembles ested reader is referred. In summary, in established that found in the nephrotic syndrome,190 and is a PE the GFR is reduced by one-third, with a smaller type of acute phase response. An increase in intra- percentage fall in effective RPF; the filtration frac- glomerular pressure would result in increased fil- tion thus falls. These changes can completely wipe tration of intermediate-sized molecules, such as has out the normal increases in renal function during been reported in PE; such an effect can be mimicked pregnancy, as shown by studies of the same women by the infusion of AII.191 It is of interest that the uri- before and after delivery. The reduced RPF is still nary albumin excretion rate remains high up to 3–5 demonstrable 4 months post-partum, as is a contrac- years after delivery following PE.192,193 It is not ted plasma volume.184 The renal clearance of uric known whether this is a long-term consequence of acid is reduced in established PE, resulting in hyp- the pregnancy disease or whether a degree of vascu- eruricaemia, the degree of which varies directly with lar disease was present before conception. severity of disease. There appears to be an increased Urinary calcium excretion is two- to three-fold net reabsorbtion of uric acid which, together with higher in normal pregnancy than in the non-preg- the lower GFR, will result in lower renal clearance. nant state, even though there is a need to conserve The most likely explanation for the altered renal calcium for both maternal and fetal development. handling lies in the volume contraction of PE (see There is however, an increase in the concentrations above). of 1,25-dihydroxy vitamin D and a reciprocal fall in The normal pregnant woman retains some 950 intact parathyroid hormone levels. The increase in mmol sodium in the course of pregnancy. AII is anti- serum 1,25-dihydroxy vitamin D values is probably natriuretic in physiological concentrations, but from a key factor in providing for these increased calcium as early as the first trimester, this effect is blunted,185 requirements in pregnancy.194 Changes in calcium in parallel with the decreased pressor response to metabolism are well-recognised in essential hyper- the hormone. Interestingly however, the adrenal tension. Hypocalciuria is common in established PE responsiveness to AII is maintained. A prospective (eg, Taufield et al, 1987195; Sanchez-Ramos et al, study reported that, prior to 32 weeks gestation, no 1991196). Furthermore, the mean urinary calcium:- difference could be demonstrated in the association creatinine ratio is decreased from the first trimester between urinary sodium excretion and arterial in women who go on to develop PE,197 although the pressure in women who remained normotensive and overlap with women who remain normotensive is those who became hypertensive.186 The excretion of considerable, and measurement of the ratio is not a sodium load does not differ significantly between helpful as a predictor of outcome.198,199 The mech- non-pregnant and normotensive pregnant women in anism for the hypocalciuria of PE has not been fully the second trimester, but in the third trimester, there worked-out. Women with established PE have is a tendency to sodium retention and expansion of decreased plasma 1,25-dihydroxy vitamin D concen- plasma volume under these circumstances.187 These trations and increased plasma parathyroid hormone authors reported that in a prospective study of salt concentrations.200,201 The lower 1,25-dihydroxy vit- loading in second trimester primigravidae, those amin D concentrations could be due to diminished who went on to develop GH did not handle a renal or placental hormone synthesis, and would be sodium load differently from those who remained expected to decrease intestinal calcium absorption. normotensive. However, once PE was established, Serum ionised calcium is lower in established they retained sodium avidly, with a reduction in fil- PE,200,201 but intracellular free calcium concen- tered sodium.187 Normal pregnant women have a trations are increased, for example in the plate- raised ALD:PRC ratio by comparison with non-preg- lets.182 Given the impact of intracellular calcium nant women, and in established GH, the ratio is still concentrations on vascular reactivity, it is perhaps further raised,188 suggesting an altered adrenal surprising that more research has not been carried responsiveness to the RAS. An enhanced natriuretic out in this area. response to ANP has been described in PE,177 which, in conjunction with the increased circulating con- Placentation centrations, might contribute to the volume contraction. Human placentation involves an intensely invasive One of the pathognomic features of PE is the process, in the course of which there is major development of significant proteinuria (see: remodelling of the resistance vessels of the uterine Definition). Serum albumin concentrations fall in circulation. In a normal pregnancy, the cytotrophob- normal pregnancy, as a consequence of plasma vol- lasts (epithelial stem cells from future placental ume expansion outstripping albumin synthesis. Uri- tissue) invade the spiral arterioles and progressively nary albumin excretion is significantly increased in erode their walls until they are structurally floppy, normal women by 16 weeks gestation, and takes up thin-walled conduits (See: Robertson et al, 1984202), to 3 months to return to pre-conception values.189 with the capacity to support a greatly enhanced Masse´ et al (1993)142 studied urinary microalbumin blood flow. This erosion continues beyond the deci- excretion measured prospectively in more than 1300 duo-myometrial junction, and the arteriolar struc- nulliparas and reported that it was already signifi- ture is never regained after pregnancy. Indeed, it is cantly higher between 15–24 weeks gestation in never regained once a pregnancy has progressed women who went on to develop GH or PE. The fall beyond 14–16 weeks of gestation, which has been

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 717 linked to an apparently protective effect of a mid- the ␣V␤3 integrin characteristic of angiogenic endo- trimester pregnancy loss in relation to PE. The thelium, and enhanced in cytotrophoblasts which maternal placental circulation is established surpris- have invaded the spiral arteries, is weak.214 In con- ingly late, at 13–14 weeks203,204 so that at 8–10 trast, expression of ␣V␤6 is maintained; this is tran- weeks gestation, the oxygen pressure within the siently expressed in remodelling endothelium and intervillous space is only about 18 mm Hg.205 Once is confined to sites of column formation in normal the placental circulation is established, the blood placentae.214 It thus appears that the invasive cyto- perfusing the cytotrophoblasts is at maternal oxygen trophoblasts from PE placentae have failed to trans- tensions. It has thus been suggested that the cytotro- form their phenotype to mimic that of vascular phoblast invasion may be partly regulated by the endothelium. Furthermore, TGF-␤3 is overex- prevailing oxygen tension. This has been tested by pressed in placentae from PE pregnancies, leading Genbachev et al (1997),206 using in vitro method- to the suggestion that a failure to downregulate such ology, who showed that hypoxia stimulated cytotro- expression in early pregnancy might contribute to phoblast proliferation and differentially altered the the abnormal placentation.207 The localization of expression of some stage-specific antigens. Another VEGF in placental biopsy samples at delivery has proposed stimulatory mechanism concerns pla- been reported not to differ qualitatively between cental TGF-␤3 expression, which is high in early normal and PE placentae although the intensity of pregnancy but begins to fall at ෂ9 weeks’ gestation. VEGF immunostaining in syncytiotrophoblast was These changes are reciprocally correlated with tro- significantly reduced in PE.209 phoblast outgrowth and fibronectin synthesis, A ‘deportation’ of trophoblast to the maternal cir- which are markers of trophoblast differentiation culation has been reported in established PE.215 Tro- toward an invasive phenotype.207 A fascinating phoblast cells were found in the uterine venous observation has recently been reported,208 which blood (but rarely in peripheral venous blood) of nor- suggests that the invasive cytotrophoblasts may mal pregnant women with higher levels in pre- switch their repertoire of adhesion molecules to eclampsia. There was no correlation between tro- mimic the adhesion phenotype of endothelial cells, phoblast numbers and indices of severity of the dis- with enhanced molility and invasiveness. Vascular ease, and the authors concluded that increased tro- endothelial growth factor (VEGF) has been localised phoblast deportation in PE is secondary to the immunohistochemically in normal term villous pla- changes occurring in the placenta. The medium centa in the syncytiotrophoblast with less intense from co-cultures of syncytiotrophoblast microvill- staining in stromal cells209 and is assumed to be con- ous membranes (STBM) and endothelial cells cerned with placental angiogenesis. caused significant activation of granulocytes and One of the major pathological features of PE is the monocytes derived from male subjects.216 Such an relative failure of trophoblast invasion210,211, which effect might account for the observed activation of does not proceed beyond the deciduo-myometrial peripheral blood leukocytes in PE. Furthermore, junction. This is not unique to PE, but is also found vesicles prepared from STBM reduced the relaxation in association with intra-uterine growth restriction response of preconstricted subcutaneous fat arteries occurring in the absence of hypertension.212 Since from normotensive pregnant women to acetylcho- flow is proportional to the fourth power of the line, presumably due to decreased endothelial radius, it is not surprising that a reduced spiral dependent relaxation.164 Thus some of the endo- artery diameter is associated with fetal growth thelial dysfunction seen in PE might be secondary to restriction. There may thus be a differential respon- the consequences of increased trophoblast invasion. siveness to impaired foeto-placental blood flow in the two groups of women, but formal studies of fac- Future research in placentation tors mediating this differential response have yet to be performed. It has to be accepted that any studies of function on Histologically, the cytotrophoblasts do not appear tissue derived from PE pregnancies may be mislead- to be adequately ‘anchored’ to the vascular endo- ing in relation to the initiation of the disease pro- thelium, and do not invade, or invade inadequately. cess, which is likely to occur very early in gestation. A major difficulty in studying mechanisms for the By the time the disease is clinically-manifest, the impairment of placentation is that the diagnosis of placenta, like other tissues, has been exposed to PE is made late in pregnancy, and data gathered deteriorating conditions for months. Thus, from histological and in vitro studies of placental attempting to unravel cause from effect is extremely tissues gathered at the end, even a premature end, difficult. Furthermore, the placenta has a built-in of gestation, may bear little relationship to what was obsolescence even in normal pregnancy, and is pro- happening at the initiation of placentation. All that grammed to function under conditions very differ- can be done is to match control and PE tissues for ent from those pertaining in culture. Its enzyme sys- gestation age. Lower trophoblast attachment on fib- tems on the fetal side function under conditions of ronectin and vitronectin-coated slides was observed low oxygen (ෂ25–30 mm Hg) and high carbon diox- in samples from PE than normotensive pregnancies, ide (ෂ50–55 mm Hg) tension; the effect of the sud- possibly reflecting differences in expression of den perfusion with well-oxygenated blood which matrix receptors.213 There was also a decreased occurs after the first breath, before placental separ- multinuclear cell formation, indicating lesser syncy- ation, has not been studied, but is likely to be large. tialization of trophoblast in this group. It has also Intrinsic placental coagulation systems are also rap- been shown that the cytotrophoblastic expression of idly activated at placental separation.

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 718 Our knowledge of the processes of normal placen- through clinical observation rather than expensive tation is still only partial, and hindered by the fact laboratory measurement makes this a potentially that when samples are obtained in early human very valuable prophylactic treatment for the pregnancy, the majority of such pregnancies will not developing world, where PE kills numerous women proceed to term. Nevertheless, further studies of each year. normal placentation are likely also to be of help in unravelling the pathophysiology of PE. Research into the expression and function of the ever-increas- Conclusion ing family of growth factors in relation to placen- The upsurge in research into the pathophysiology of tation is likely to burgeon. Data at present are not PE over the last 20 years has still further emphasised entirely consistent, and it may well be that, in a pro- the protean presentation and multifactorial nature of cess as vital as placentation, there are a number of the disease. In some women, pregnancy seems to be ‘safety nets’, such that more than one cytokine can unmasking underlying hypertensive disease which exert a particular effect. will become manifest in later life. In others, the blood pressure can reach life-threatening levels in a For the future? terrifyingly short space of time, only to subside nearly as fast after delivery, and never to recur. In We have to accept that, unlike many disease states, between is all the spectrum familiar to obstetricians. there is still no adequate animal model for PE. Given It is possible to speculate that mild, late-onset GH the apparently central roˆle of placentation in the dis- which usually occurs without detriment to fetal ease, and the enormously variable nature of placen- well-being, and in some studies in association with tation in the Mammalia, this should not cause sur- increased birthweight is in fact a physiological prise. Certainly, hypertension can be induced in response to an inadequate utero-placental blood pregnant animals, and the consequences of that flow. This might ‘account for’ its high prevalence, hypertension examined, but that is not the same as and would imply that, although careful monitoring having a naturally-occurring animal model. Sheep is still needed in case of progression, antihyperten- ‘toxaemia’ is a disorder of carbohydrate metabolism. sive therapy is not needed. PE might then represent Although there is developing interest in insulin a breakdown of this homeostatic mechanism, in any resistance as a predisposing factor for PE, one feels one of a number of systems. that human PE is unlikely to respond to the oral It is a condition or cluster of conditions with administration of calcium borogluconate, which, as aspects which can interest those with research shepherds are well aware, is the remedy in sheep. expertise in a wide variety of fields, and it will be The long-running searches for a predictor or pre- through multidisciplinary research that the patho- dictive index for PE, and, in parallel, for an adequate physiology is unravelled. The techniques of molecu- prophylactic regime, are likely to continue until we lar biology must be put into the context of integrated identify the causes of PE and can therefore apply physiology and clues from the new genetics be inter- more logic to the quest. The literature is littered with preted in the light of environmental factors. And attempts to derive predictive indices, some on finally, we should never lose sight of the fact that in small, very tightly-controlled, data sets, other on the developing world the hypertensive diseases of much larger sets, genuinely attempting to identify pregnancy remain major killers of young women. predictors in the frenetic activity of an antenatal 217 Their needs are all too often overlooked, but we clinic. These have recently been reviewed. As for should try to put results from sophisticated research prophylaxis, very large-scale trials have shown that into their context, as well as our own. low dose aspirin does have a place in women with a previous history of severe, early-onset disease (eg, CLASP, 1994)218 and in those with thrombo-embolic References disease, but not in the general population. Nor does calcium supplementation;219 in fact, as the recent 1 Department of Health, W.O., Scottish, Office Depart- title of a paper suggests: ‘Prevention of pre-eclamp- ment of Health, Department of Health and Social Secur- sia is a big disappointment’.220 That said, a recent ity Northern Ireland, Why mothers die, 1999, Report on = confidential enquiries into maternal deaths in the placebo-controlled pilot study (n 283) of dietary United Kingdom, pp 1994–1996. supplementation with vitamins C and E reported a 2 Koonin LM et al. Pregnancy-related mortality surveil- decreased incidence of PE and an improvement in lance—United States, 1987–1990. (MMWR CDC) Sur- plasma markers of endothelial activation and pla- veillance Summary 1997; 46, pp 17–36. cental dysfunction.122 Funding is being sought to 3 Mungra A et al. Nationwide maternal mortality in Suri- conduct a large-scale trial, but this is likely to be nam. Br J Obstet Gynaecol 1999; 106: 55–59. blurred by women self-medicating following the 4 Etard J-F, Kodio B, Traore´ S. Assessment of maternal widespread publicity surrounding publication of the mortality and late maternal mortality among a cohort of trial. The convincing demonstration that magnesium pregnant women in Bamako, Mali. Br J Obstet Gynaecol sulphate is the treatment of choice for the preven- 1999; 106: 60–65. 221 5 Douglas KA, Redman CWG. Eclampsia in the United tion of recurrent eclamptic fits has led to the Kingdom. BMJ 1994; 309: 1395–1400. ongoing MAGPIE trial, investigating its utility in 6 Sibai B et al. Eclampsia IV. Neurological findings and preventing fitting in women with PE. The relative future outcome. Am J Obstet Gynaecol 1985; 152: ease of use and cheapness of magnesium sulphate, 184–192. and the possibility of monitoring for toxic levels 7 National High Blood Pressure Education Program.

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 719 Report on high blood pressure in pregnancy. Am J a Greek population of blood donors. Am J Hematol Obstet Gynecol 1990; 163: 1691–1712. 1999; 61: 265–267. 8 Shennan A et al. Lack of reproducibility in pregnancy 30 Brattstrom L et al. A common methylenetetrahydrofol- of Korotkoff phase IV as measured by mercury sphyg- ate reductase gene mutation and longevity. Athero- momanometry. Lancet 1996; 347: 139–142. sclerosis 1998; 141: 315–319. 9 MacGillivray I. Pre-eclampsia: the Hypertensive Dis- 31 Cattaneo M et al. The G20210A mutation of the pro- ease of Pregnancy. WB Saunders, Eastbourne, West thrombin gene in patients with previous first episodes Sussex, 1983. of deep-vein thrombosis: prevalence and association 10 Eskenazi B et al. Fetal growth retardation in infants of with factor V G1691A, methylenetetrahydrofolate multiparous and nulliparous women with preeclamp- reductase C677T and plasma prothrombin levels. sia. Am J Obstet Gynecol 1993; 169: 1112–1118. Thromb Res 1999; 93: 1–8. 11 Saudan P et al. Does gestational hypertension become 32 Lindoff C et al. Preeclampsia is associated with a pre-eclampsia? Br J Obstet Gynaecol 1998; 105: reduced response to activated protein C. Am J Obstet 1177–1184. Gynecol 1997; 176: 457–460. 12 Taylor RN et al. High plasma cellular fibronectin levels 33 Dizon-Townson DS et al. The factor V Leiden mutation correlate with biochemical and clinical features of pre- may dispose women to severe pre-eclampsia. Am J eclampsia but cannot be attributed to hypertension Obstet Gynecol 1996; 175: 902–905. alone. Am J Obstet Gynecol 1991; 165: 895–901. 34 Chen G et al. Tumour necrosis factor-alpha (TNF- 13 Chesley LC, Annitto JE, Cosgrove RA. The remote alpha) gene polymorphism and expression in pre- prognosis of eclamptic women. Am J Obstet Gynecol eclampsia. Clin Exp Immunol 1996; 104: 154–159. 1976; 124: 446–459. 35 Stark M et al. Tumor necrosis factor. Lancet 1995; 345: 14 Jonsdottir LS et al. Death rates from ischaemic heart 648–649. disease in women with a history of hypertension in 36 Williams MA et al. Maternal second trimester serum pregnancy. Acta Obstetricia et Gynecologica Scandi- tumor necrosis factor-alpha-soluble recepto p55 navica 1995; 74: 772–776. (sTNp55) and subsequent risk of preeclampsia. Am J 15 Fisher KA et al. Hypertension in pregnancy: Clinical- Epidemiol 1999; 149: 323–329. pathological correlations and remote prognosis. Medi- 37 Dizon-Townson DS, Major H and Ward K. A promoter cine 1981; 60: 267–276. mutation in the tumor necrosis factor alpha gene is not 16 Broughton Pipkin F, Lal S, Sharif J. Predicting high associated with preeclampsia. Reproduct Immunol blood pressure in pregnancy—a multivariate 1998; 38: 55–61. approach. J Hypertens 1998; 16: 221–119. 38 Ward K et al. A molecular variant of angiotensinogen 17 Robinson BF. Altered calcium handling as a cause of associated with preeclampsia. Nat Genet 1993; 4: 59– primary hypertension. J Hypertens 1984; 2: 453–460. 61. 18 Nicholls MG, Richards AM, Agarwal M. The impor- 39 Caulfield M et al. Angiotensinogen in hyman essential tance of the renin-angiotensin system in cardiovascu- hypertension. Hypertension 1996; 28: 1123–1125. lar disease. J Hum Hypertens 1998; 12: 295–299. 40 Winkelmann BR et al. Angiotensinogen M235T poly- 19 Lindhemier MD, Roberts JM, Cunningham FG (eds). morphism is associated with plams angiotensinogen Hypertensive Disorders in Pregnancy. 2nd edn, ed. and cardiovascular disease. Am Heart J 1999; 137: Chesley. Appleton & Lange: Stamford, Connecticut, 698–705. USA, 1999. 41 Morgan L et al. Pre-eclampsia and the angiotensinogen 20 Morgan T, Ward K. New insights into the genetics of gene. Br J Obstet Gynaecol 1995; 102: 489–4900. preeclampsia. Seminars in Perinatol 1999; 23: 14–23. 42 Guo G et al. Angiotensinogen gene variation in a popu- 21 Broughton Pipkin F. The genetics of pre-eclampsia— lation case-control study of preeclampsia/eclampsia in ideas at the turn of the Millennium. Curr Obstet Gyn- Australians and Chinese. Electrophoresis 1997; 18: aecol 1999; 9: 178–182. 1646–1649. 22 Lie RT et al. Fetal and maternal contributions to risk 43 Morgan L et al. Maternal and fetal angiotensinogen of pre-eclampsia: population based study. BMJ 1998; gene allele sharing in pre-eclampsia. Br J Obstet Gyn- 316: 1343–1347. aecol 1999; 106: 244–251. 23 Graves JAM. Genomic imprintin, development and 44 Wallukat G. et al. Patients with preeclampsia develop disease—is pre-eclampsia caused by a maternally- agonistic autoantibodies against the angiotensin AT1 imprinted gene? Reproduct Fertil Develop 1998; 10: receptor. J Clin Invest 1999; 103: 945–952. 23–29. 45 Morgan L et al. Functional and genetic studies of the 24 Kupferminc MJ et al. Increased frequency of genetic angiotensin II type 1 receptor in pre-eclamptic and thrombophilia in women with complications of preg- normotensive pregnant women. J Hypertens 1997; 15: nancy. N Engl J Med 1999; 340: 9–13. 1389–1396. 25 Powers RW et al. Methylenetetrahydrofolate reductase 46 Grant NF et al. A study of angiotensin II pressor polymorphism, folate, and susceptibility to preec- response throughout primigravid pregnancy. J Clin lampsia. J Soc Gynecol Invest 1999; 6: 74–79. Invest 1973; 52: 2682–2689. 26 Mutchinick OM et al. High prevalence of the thermo- 47 Takimoto E et al. Hypertension induced in pregnant labile methylenetetrahydrofolate reductase variant in mice by placental renin and maternal angiotensinogen. Mexico: a country with a very high prevalence of neu- Science 1996; 274: 995–998. ral tube defects. Mol Genet Metab 1999; 68: 461–467. 48 Hubel CA, Roberts JM, Ferrell RE. Association of pre- 27 Rady PL et al. Methylenetetrahydrofolate reductase eclampsia with common coding sequence variations in (MTHFR): the incidence of mutations C677T and the lipoprotein lipase gene. Clin Genet 1999; 56: A1298C in the Ashkenazi Jewish population. Am J 289–296. Med Genet 1999; 86: 380–384. 49 Fisher RM, Humphries SE, Talmud PJ. Common vari- 28 Sohda S et al. Methylenetetrahydrofolate reductase ation in the lipoprotein lipase gene: effects on plasma polymorphism and pre-eclampsia. J Med Genet 1997; lipids and risk of atherosclerosis. Atherosclerosis 34: 525–526. 1997; 135: 145–149. 29 Antoniadi T et al. Prevalence of factor V Leiden, pro- 50 Furui T et al. Decrease in cytochrome c oxidase and thrombin G20210A, and MTHFR C677T mutations in cytochrome oxidase subunit I messenger RNA levels

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 720 in preeclamptic pregnancies. Obstet Gynecol 1994; 84: blood leukocytes akin to those of sepsis. Am J Obstet 283–288. Gynecol 1998; 179: 80–86. 51 Vuorinen K et al. Placental mitochondrial DNA and 72 Hsu CD et al. Elevated circulating thrombomodulin in respiratory chain enzymes in the etiology of preec- severe preeclampsia. Am J Obstet Gynecol 1993; 169: lampsia. Obstet Gynecol 1998; 91: 950–955. 148–149. 52 Kilpatrick DC. Influence of human leukocyte antigen 73 Fitzgerald DJ et al. Decreased prostacylin biosynthesis and tumour necrosis factor genes on the development preceding the clinical manifestation of pregnancy- of pre-eclampsia. Hum Repro Updata 1999; 5: 94–102. induced hypertension. Circulation 1987; 75: 956–963. 53 Roberts JM. Pregnancy related hypertension. In: Creasy 74 Redman CW et al. Factor-VIII consumption in pre- RK, Resnick R (eds). Maternal Fetal Medicine.WB eclampsia. Lancet 1977; 2: 1249–1252. Saunders: Philadelphia, 1998. 75 Taylor RN, Heilbron DC, Roberts JM. Growth factor 54 Hutt R et al. Increased platelet volume and aggregation activity in the blood of women in whom preeclampsia precede the onset of preeclampsia. Obstet Gynecol develops is elevated from early pregnancy. Am J 1994; 83: 146–149. Obstet Gynecol 1990; 163: 1839–1844. 55 Janes SL et al. Flow cytometry detection of activated 76 Roberts JM et al. Sera from preeclamptic women platelets in pregnant women prior to the development specifically activate human umbilical vein endothelial of pre-eclampsia. Thromb Haemostasis 1995; 74: cells in vitro: morphological and biochemical evi- 1059–1063. dence. Am J Reproduct Immunol 1992; 27: 102–111. 56 Gallery EDM et al. Relation between plasma volume 77 Conrad, KP, Miles TM, Benyo DF. Circulating levels of and uric acid in the development of hypertension in immunoreactive cytokines in women with preeclamp- pregnancy. In: Bonnar M, MacGillivray I, Symonds MS sia. Am J Reproduct Immunol 1998; 40: 102–111. (eds). Pregnancy Hypertension. University Park Press: 78 Clark P et al. The neutrophil and preeclampsia. Sem- Baltimore, 1980. inars Reproduct Endocrinol 1998; 16: 57–64. 57 Rubanyi GM. The role of endothelium in cardiovascu- 79 Hubel CA. Oxidative stress and preeclampsia. Fetal lar homeostasis and diseases. J Cardiovasc Pharmacol Matern Med Rev 1997; 9: 73–101. 1993; 22 (Suppl 4): S1–S14. 80 Rappaport VJ et al. Anti-vascular endothelial cell anti- 58 Talledo OR. Renin-angiotensin system in normal and bodies in severe preeclampsia. Am J Obstet Gynecol toxemic pregnancies. I. Angiotensin infusion test. Am 1990; 162: 138–146. J Obstet Gynecol 1966; 96: 141–143. 81 Roy AC et al. Interrelationships of serum paraoxonase, 59 Cockell AP, Poston L. Flow-mediated vasodilation is serum lipids and apolipoproteins in normal preg- enhanced in normal pregnancy but reduced in preec- nancy. A longitudinal study. Gynecol Obstet Investig lampsia. Hypertension 1997; 30: 247–251. 1999; 38: 10–13. 60 Pascoal IF, Umans G. Effect of pregnancy on mech- 82 Tripathy C et al. Serum insulin and lipid profile in anisms of relaxation in human omental microvessels. normal pregnant and pregnancy-induced hypertensive Hypertension 1996; 28: 183–187. women from North India. Aust NZ J Obstet Gynaecol 61 van de Voorde J, Depypere H, Vanheel B. The influ- 1999; 39: 321–323. ence of pregnancy on endothelium-derived nitric 83 Lorentzen B et al. Fasting serum free fatty acids and oxide mediated relaxations in isolated human resist- triglycerides are increased before 20 weeks of gestation ance vessels. Fundament Clin Pharmacol 1997; 11: in women who later develop preeclampsia. Hypertens 371–377. Pregnancy 1994; 13: 103–109. 62 Egerman RS et al. Thrombotic thrombocytopenic pur- 84 Hubel C et al. Women with a history of eclampsia pura and hemolytic uremic syndrome in pregnancy: manifest dyslipidemia during later life. J Soc Gynecol Review of 11 cases. Am J Obstet Gynecol 1996; 175: Investig 1998; 5 (Suppl): 40A–41A. 950–956. 85 O’Keeefe JHJ, Lavie CJJ, McCallister BD. Insights into 63 Roberts JM et al. Preeclampsia: An endothelial cell dis- the pathogenesis and prevention of coronary artery order. Am J Obstet Gynecol 1989; 161: 1200–1204. disease [Review]. Mayo Clin Proc 1995; 70: 69–79. 64 Roberts JM. Endothelial dysfunction in preeclampsia 86 Witztum J. The oxidation hypothesis of atheroscler- (Review). Seminars Reproduct Endocrinol 1998; 16: osis. Lancet 1994; 344: 793–795. 5–15. 87 Assali PM, Assali NS. Vascular changes in the decidua 65 McCartney CP. Pathological anatomy of acute hyper- associated with eclamptogenic toxemia. Am J Clin tension of pregnancy. Circulation 1964 (Suppl II): 30. Pathol 1950; 20: 1099. 66 Davidge ST et al. Distinct factors in plasma of pre- 88 Staff AC et al. Increased contents of phospholipids, eclamptic women increase endothelial nitric oxide or cholesterol, and lipid peroxides in decidua basalis prostacyclin. Hypertension 1996; 28: 758–764. inwomen with preeclampsia. Am J Obstet Gynecol 67 Powers RW et al. Plasma homocysteine concentration 1999; 180: 587–592. is increased in preeclampsia and is associated with 89 Clarkson P et al. Endothelium-dependent dilatation is evidence of endothelial activation. Am J Obstet Gyn- impaired in young health subjects with a family his- ecol 1998; 179: 1605–1611. tory of premature coronary disease. Circulation 1997; 68 Hubel CA et al. Fasting serum triglycerides, free fatty 96: 3378–3383. acids and malondialdehyde are increased in preec- 90 Torbergsen T et al. Pre-eclampsia—a mitochondrial lampsia, are positively correlated, and decrease within disease? Acta Obstet Gynecol Scand 1989; 68: 145– 48 hours post partum. Am J Obstet Gynecol 1996; 174: 148. 975–982. 91 Morikawa S et al. Increased mitochondrial damage by 69 Hubel CA et al. Small low-density lipoproteins and lipid peroxidation in trophoblast cells of preeclamptic vascular cell adhesion molecule-1 are increased in placentas. Biochem Mol Biol Int 1997; 41: 767–775. association with hyperlipidemia in preeclampsia. 92 Wang Y, Walsh SW. Placental mitochondria as a Metabolism: Clin Experiment 1998; 47: 1281–1288. source of oxidative stress in pre-eclampsia. Placenta 70 Kaaja R. Insulin resistance syndrome in preeclampsia. 1998; 19: 581–586. Seminars Reproduct Endocrinol 1998; 16: 41–46. 93 Page EW. The relation between hydatid moles, relative 71 Sacks GP et al. Normal pregnancy and preeclampsia ischemia of the gravid uterus, and the placental origin both produce inflammatory changes in peripheral of eclampsia. Am J Obstet Gynecol 1939; 37: 291–293.

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 721 94 Arias F et al. Maternal placental vasculopathy and 116 Many A et al. Xanthine oxidase/dehydrogenase is infection: Two distinct subgroups among patients present in human placenta. Placenta 1996; 17: 361– with preterm labor and preterm ruptured membranes. 365. Am J Obstet Gynecol 1993; 168: 585–591. 117 Many A et al. Invasive cytotrophoblast manifest evi- 95 Khong TY, De Wolf F, Robertson WB. Inadequate dence of oxidative stress in preeclampsia. Am J maternal vascular response to placentation in preg- Pathol 2000; 156: 321–331. nancies complicated by pre-eclampsia and by small- 118 Entman SS et al. Elevated serum iron in toxemia of for-gestational age infants. Br J Obstet Gynaecol 1986; pregnancy. Am J Obstet Gynecol 1982; 143: 398–404. 93: 1049–1059. 119 Cester N et al. Pregnancy induced hypertension: a 96 Ness RB, Roberts JM. Heterogeneous causes constitut- role for peroxidation in microvillus plasma mem- ing the single syndrome of preeclampsia: a hypoth- branes. Mol Cell Biochem 1994; 131: 151–155. esis and its implications [Review]. Am J Obstet Gyn- 120 Rosenfeld ME. Inflammation, lipis, and free radicals: ecol 1996; 175: 1365–1370. Lessons learned from the atherogenic process 97 Reaven GM. Syndrome X: 6 years later. J Int Med [Review]. Seminars Reproduct Endocrinol 1998; 16: 1994; 236: 13–22. 249–261. 98 Bellamy MF et al. Syndrome X and endothelial dys- 121 Romero J, Reckelhoff J. Role of angiotensin and oxi- function. Cardiovasc Res 1998; 40: 410–417. dative stress in essential hypertension. Hypertension 99 Belfiore F, Iannello S. Insulin resistance in obesity: 1999; 34: 943–949. metabolic mechanisms and measurement methods. 122 Chappell LC et al. Effect of antioxidants on the occur- Mol Genet Metab 1998; 65: 121–128. rence of pre-eclampsia in women at increased risk: a 100 Barden AE et al. Does a predisposition to the meta- randomised trial. Lancet 1999; 354: 810–816. bolicsyndrome sensitize women to develop pre- 123 Davison JM, Nobel MCB. Serial changes in 24 hour eclampsia? J Hypertens 1999; 17: 1307–1315. creatinine clearance during normal menstrual cycles 101 Sibai BM et al. Risk factors for preeclampsia in heal- and the first trimester of pregnancy. Br J Obstet Gyn- thy nulliparous women: a prospective multicenter aecol 1981; 88: 10–17. study. The National Institute of Child Health and 124 Manhem K, Jern S. Influence of daily-life activation Human Development Network of Maternal-Fetal on pulse rate and blood pressure changes during the Medicine Units. Am J Obstet Gynecol 1995; 172: menstrual cycle. J Hum Hypertens 1994; 8: 851–856. 642–648. 125 van Beek E et al. Peripheral haemodynamics and 102 Acromite MT et al. Androgens in preeclampsia. Am renal function in relation to the menstrual cycle. Clin J Obstet Gynecol 1999; 180: 60–63. Sci 1996; 91: 163–168. 103 Laivouri H et al. Evidence of high circulating testos- 126 Chapman AB et al. Systemic and renal hemodynamic terone in women with prior preeclampsia. J Clin changes in the luteal phase of the menstrual cycle Endocrinol Metab 1998; 83: 344–347. mimic early pregnancy. Am J Physiol 1997; 237: 104 Hubel C, Roberts J, Ferrell R. Association of preec- F777–782. lampsia with common coding sequence variations in 127 Brown JJ et al. Variations in plasma renin during the the lipoprotein lipase gene. Clin Genet 1999; 56: menstrual cycle. Br Med J 1964; 2: 1114–1115. 289–296. 128 Kaulhausen H et al. The relationship of the renin- 105 Kaaja R et al. Serum lipoproteins, insulin and urinary angiotensin-aldosterone system on plasma gonado- prostanoid metabolites in normal and hypertensive tropin, prolactin and ovarian steroid patterns during pregnant women. Obstet Gynecol 1995; 85: 353–356. the menstrual cycle. Archiv fur Gyna¨ kologie 1978; 106 Joffe GM et al. The relationship between abnormal 225: 179–200. glucose tolerance and hypertensive disorders of preg- 129 De Hertogh R et al. Immunoradiometric assays of nancy in healthy nulliparous women. Am J Obstet total renin and gonadotropins in serum during the Gynecol 1998; 179: 1032–1037. menstrual cycle. Fertility and Sterility 1989; 52: 107 Kaaja R et al. Evidence of a state of increased insulin 209–215. resistance in preeclampsia. Metab: Clin Experiment 130 Chapman AB et al. Temporal relationships between 1999; 48: 892–896. hormonal and hemodynamic changes in early human 108 Laivouri H, Tikkanen MJ, Ylikorkala O. Hyperinsuli- pregnancy. Kidney Int 1998; 54: 2056–2063. nemia 17 years after preeclamptic first pregnancy. J 131 Robson SC et al. Serial study of factors influencing Clin Endocrinol Metab 1996; 81: 2908–2911. changes in cardiac output during human pregnancy. 109 Chesley LC. Eclampsia: the remote prognosis. Sem- Am J Physiol 1989; 256: H1060–H1065. inars Perinatol 1978; 2: 99–111. 132 Sundsfjord JA, Aakvaag A. Variations in plasma aldo- 110 Mayer EL, Jacobson DW, Robinson K. Homocysteine sterone and plasma renin activity throughout the and coronary atherosclerosis. J Am Coll Cardiol 1966; menstrual cycle, with special reference to the pre- 27: 517–527. ovulatory period. Acta Endocrinol 1973; 73: 499–508. 111 Eskenazi B, Fenster L, Sidney S. A multivariate 133 Easterling TR et al. Maternal hemodynamics in nor- analysis of risk factors for preeclampsia. J Am Med mal and preeclamptic pregnancies: a longitudinal Assoc 1991; 266: 237–241. study. Obstet Gynecol 1990; 76: 1061–1069. 112 Roberts JM, Hubel CA. Is oxidative stress the link in 134 Brown MA, Buddle ML. Inadequacy of dipstick pro- the two-stage model of pre-eclampsia? Lancet 1999; teinuria in hypertensive pregnancy. Aust NZ J Obstet 354: 788–789. Gynaecol 1995; 35: 366–369. 113 Hubel CA et al. Increased ascorbate radical formation 135 Bosio PM et al. Maternal central hemodynamics in and ascorbate depletion in plasma from women with hypertensive disorders of pregnancy. Obstet Gynecol preeclampsia—implications for oxidative stress. Free 1999; 94: 978–984. Radical Biol Med 1997; 23: 597–609. 136 Visser W, Wallenburg HCS. Central hemodynamic 114 Bulkley GM. Reactive oxygen metabolites and reper- observations in untreated preeclamptic patients. fusion injury: aberrant triggering of reticuloendothel- Hypertension 1991; 17: 1072–1077. ial function. Lancet 1998; 344: 934–936. 137 Brown MA, Zammit VC, Mitar DM. Extracellular 115 Hogg N. Free radicals in disease. Seminars Reproduct fluid volumes in pregnancy-induced hypertension. J Endocrinol 1998; 16: 241–248. Hypertens 1992; 10: 61–68.

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 722 138 Silver HM, Seebeck M, Carlson R. Comparison of impaired in pre-eclampsia. J Hum Hypertens 1999; total blood volume in normal, preeclamptic, and non- 13: 179–183. proteinuric gestation hypertensive pregnancy by sim- 158 Silver HM et al. A longitudinal study of maternal ultaneous measurement of red blood cell and plasma hemodynamic and baroreflex function in the puer- volumes. Am J Obstet Gynecol 1998; 179: 87–93. perium. When is a stable baseline established? J Soc 139 Brown MA, Zammit VC, Lowe SA. Capillary per- Gynecolog Investig 1998; 5 (Suppl): 51A. meability and extracellular fluid volumes in preg- 159 Valdes G et al. The neurogenic response to 55-degree nancy-induced hypertension. Clin Sci 1989; 77: head-up tilt in normal pregnant women. Hyperten- 599–604. sion 1986; 8 (Suppl I): I.161–I.164. 140 Christianson RE. Studies on blood pressure during 160 Wasserstrum, N et al. Implications of sino-aortic pregnancy. I. Influence of parity and age. Am J Obstet baroreceptor reflex dysfunction in severe preeclamp- Gynecol 1976; 125: 509–513. sia. Obstet Gynecol 1989; 74: 34–39. 141 Halligan A et al. Twenty-four-hour ambulatory blood 161 Assali NS et al. Studies on autonomic blockade VI: pressure measurement in a primigravid population. J the mechanisms regulating the hemodynamic Hypertens 1993; 11: 869–873. changes in the pregnant woman and their relation to 142 Masse´ J et al. A prospective study of several potential the hypertension of toxaemia of pregnancy. Am J biologic markers for early prediction of the develop- Obstet Gynecol 1952; 63: 978–988. ment of preeclampsia. Am J Obstet Gynecol 1993; 162 Millar JG et al. Early prediction of pre-eclampsia by 169: 501–508. measurement of kallikrein and creatinine on a ran- 143 Sibai BM et al. Risk factors for preeclampsia in heal- dom urine sample. Br J Obstet Gynaecol 1996; 103: thy nulliparous women: a prospective multicenter 421–426. study. The National Institute of Child Health and 163 Williams DJ et al. Nitric oxide-mediated vasodilation Human Development Network of Maternal-Fetal in human pregnancy. Am J Physiol 1997; 272: Medicine Units. Am J Obstet Gynecol 1995; 172: H748–H752. 642–648. 164 Cockell AP et al. Human placental syncytiotrophoblast microvillous membranes impair maternal vas- 144 Martin A et al. Measuring venous capacitance and cular endothelial function. Br J Obstet Gynaecol blood flow in pregnancy. Aust NZ J Obstet Gynaecol 1997; 104: 235–240. 1997; 37: 335–339. 165 Abdul-Karim R, Assali NS. Pressor response to angi- 145 Anumba DOC et al. Nitric oxide activity in the per- otonin in pregnant and non-pregnant women. Am J ipheral vasculature during normotensive and pre- Obstet Gynecol 1961; 82: 246–251. eclamptic pregnancy. Am J Physiol 1999; 277: 166 Chesley LC et al. Vascular reactivity to angiotensin H848–H854. II and norepinephrine in pregnant and non-pregnant 146 Pickles CJ et al. Changes in peripheral venous tone women. Am J Obstet Gynecol 1965; 91: 837–842. before the onset of hypertension in women with ges- 167 Lumbers ER. Peripheral vascular reactivity to angiot- tational hypertension. Am J Obstet Gynecol 1989; ensin and noradrenaline in pregnant and non-preg- 160: 678–680. nant women. Austr J Experiment Biol Med Sci 1970; 147 Sakai K et al. Venous distensibility during preg- 49: 493–500. nancy. Comparisons between normal pregnancy and 168 Nisell H, Hjemdahl P, Linde B. Cardiovascular preeclampsia. Hypertension 1994; 24: 461–466. responses to circulating catecholamines in normal 148 Stainer K, Pickles C, Cowley AJ. The effects of pos- pregnancy and in pregnancy-induced hypertension. ture on abnormalities of the forearm venous tone in Clin Physiol 1985; 8: 479–493. women with pregnancy-induced hypertension. Clin 169 Ramsay M, Broughton Pipkin F, Rubin PC. Compara- Sci 1987; 73: 267–269. tive study of pressor and heartrate responses to angi- 149 Macpherson MBA. A study of vascular adaptation to otensin II and noradrenaline in pregnant and non- human pregnancy. In: Obstetrics and Gynaeology. pregnant women. Clin Sci 1992; 82: 157–162. Nottingham: Nottingham, 1988. 170 Baker PN, Broughton Pipkin F, Symonds EM. Longi- 150 Marlettini MG et al. Plasma concentrations of atrial tudinal study of platelet angiotensin II binding in natriuretic factor and hemodynamics in pregnancy- human pregnancy. Clin Sci 1992; 82: 377–381. induced hypertension. Clin Experiment Hypertens 171 Baker PN, Broughton Pipkin F, Symonds EM. [A]. 1991; 13: 1305–1327. Changes in platelet angiotensin II binding sites asso- 151 Halligan A et al. Diurnal blood pressure difference in ciated with the human menstrual cycle. Am J Obset the assessment of preeclampsia. Obstet Gynecol 1996; Gynecol 1990; 163: 175–176. 87: 205–208. 172 Lewinsky RM, Riskin-Mashiah S. Autonomic inbal- 152 Olofsson P. Characteristics of a reversed circadian ance in preeclampsia: evidence for increased sym- blood pressure rhythm in pregnant women with pathetic tone in response to the supine pressor test. hypertension. J Hum Hypertens 1995; 9: 565–570. Obset Gynecol 1998; 91: 935–939. 153 Ekholm EM, Hartiala J, Huikuri HV. Circadian 173 Mills JL et al. Prostacyclin and thromboxane changes rhythm of frequency-domain measures of heart rate predating clinical onset of preeclampsia: a multicen- variability in pregnancy. Br J Obstet Gynaecol 1997; ter prospective study. J Am Med Assoc 1999; 282: 104: 825–828. 356–362. 154 Blake MJ et al. Blood pressure variability and cardiac 174 Ranta V et al. Nitric oxide production with preec- baroreceptor sensitivity (BRS) in normotensive preg- lampsia. Obstet Gynecol 1999; 93: 422–445. nancy and the puerperium. J Soc Gynecol Investig 175 Schneider F et al. Plasma cyclic GMP concentrations 1998; 5 (Suppl): 135A. and their relationship with changes of blood pressure 155 Seligman SA. Baroreceptor reflex function in pre- levels in preeclampsia. Acta Obstetricia et Gynecol eclampsia. J Clin Endocrinol Metab 1971; 78: 413– Scand 1996; 75: 40–44. 416. 176 Thomsen JK et al. Atrial natriuretic peptide concen- 156 Leduc L et al. Baroreflex function in normal preg- trations in pre-eclampsia. Br Med J 1987; 294: nancy. Am J Obstet Gynecol 1991; 65: 886–890. 1508–1510. 157 Molino P et al. Baroreflex control of heart rate is 177 Irons DW et al. Atrial natriuretic peptide in preec-

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 723 lampsia: metabolic clearance, sodium excretion and development of preeclampsia. Metabolism 1997; 46: renal hemodynamics. Am J Physiol 1977; 273: 1107–1108. F493–487. 198 Raniolo E, Phillipou G. Prediction of pregnancy- 178 Malee MP et al. Increases in plasma atrial natriuretic induced hypertension by means of the urinary cal- peptide concentration antedate clinical evidence of cium: creatinine ratio. Med J Australia 1993; 158: preeclampsia. J Clin Endocrinol Metab 1992; 74: 98–100. 1095–1100. 199 Baker PN, Hackett GA. The use of urinary albumin- 179 Talledo OE, Chesley LC, Zuspan FP. Renin, angioten- creatinine ratios and calcium-creatinine ratios as sin system in normal and toxaemic pregnancies. III screening tests for pregnancy-induced hypertension. differential sensitivity to angiotensin II and norepi- Obstet Gynecol 1994; 83: 745–749. nephrine in toxaemia of pregnancy. Am J Obstet Gyn- 200 August P et al. Abnormal 1,25-dihydroxyvitamin D ecol 1968; 100: 218–221. metabolism in preeclampsia. Am J Obstet Gynecol 180 Baker PN, Broughton Pipkin F, Symonds EM. Platelet 1992; 166: 1295–1299. angiotensin II binding sites in normotensive and 201 Seely EW et al. Lower serum ionized calcium and hypertensive women. Br J Obstet Gynaecol 1991; 98: abnormal calciotropic hormone levels in preeclamp- 436–440. sia. J Clin Endocrinol Metab 1992; 74: 1436–1440. 181 Graves SW, Moore TJ, Seely EW. Increased platelet 202 Robertson WB et al. The making of the placental bed. angiotensin II receptor number in pregnancy-induced Eur J Obstet Gynecol Reproduct Biol 1984; 18: 255– hypertension. Hypertension 1992; 20: 627–632. 266. 182 Kilby MD, Broughton Pipkin F, Symonds EM. Cal- 203 Foidart JM et al. The human placenta becomes cium and platelets in normotensive and hypertensive haemochorial at the 13th week of pregnancy. Int J Dev human pregnancy. J Hypertens 1992; 10: 997–1003. Biol 1992; 36: 451–453. 183 Conrad KP, Lindheimer MD. Renal and cardiovascu- 204 Jauniaux, E et al. Doppler ultrasonographic features lar alterations. In: Lindheimer MD, Roberts JM, Cun- of the developing placental circulation: correlation ningham FG (eds). Chesley’s Hypertensive Disorders with anatomic ﬁndings. Am J Obstet Gynecol 1992; 166: 585–587. in Pregnancy. Appleton & Lange: Stamford, Con- 205 Rodesch F et al. Oxygen measurements in endometr- necticut, USA, 1998. ial and trophoblastic tissues during early pregnancy. 184 van Beek E et al. Persistent abnormalities in plasma Obstet Gynecol 1992; 80: 283–285. volume and renal hemodynamics in patients with a 206 Genbachev O et al. Regulation of human placental history of preeclampsia. Am J Obstet Gynecol 1998; development by oxygen tension. Science 1997; 277: 179; 690–696. 1669–1672. 185 Brown MA, Broughton Pipkin F, Symonds EM. The 207 Caniggia I et al. Inhibition of TGF-beta 3 restores the effects of intravenous angiotensin II upon blood invasive capability of extravillous trophoblasts in pressure and sodium and urate excretion in human preeclamptic pregnancies. J Clin Investig 1999; 103: pregnancy. J Hypertens 1988; 6: 457–464. 1641–1650. 186 Franx A et al. Sodium-blood pressure interrelation- 208 Zhou Y et al. Human cytotrophoblasts adopt a vascu- ship in pregnancy. J Hum Hypertens 1999; 13: 159– lar phenotype as they differentiate. A strategy for suc- 166. cessful endovascular invasion? J Clin Investig 1997; 187 Brown MA et al. Sodium excretion in normal and 99: 2139–2151. hypertensive pregnancy: a prospective study. Am J 209 Lyall F et al. Placental expression of vascular endo- Obstet Gynecol 1988; 159: 297–307. thelial growth factor in placentae from pregnancies 188 Brown MA et al. Renin-aldosterone relationships in complicated by pre-eclampsia and intrauterine pregnancy-induced hypertension. Am J Hypertens growth restriction does not support placental hyposia 1992; 5: 366–371. at delivery. Placenta 1997; 18: 269–276. 189 Taylor AA, Davison JM. Albumin excretion in normal 210 Robertson WB, Brosens I, Dixon HG. The pathological pregnancy. Am J Obstet Gynecol 1997; 165; 895–901. response of the vessels of the placental bed to hyper- 190 Studd JWW. Immunoglobulins in normal pregnancy, tensive pregnancy. J Patholog Bacteriol 1967; 93: pre-eclampsia and pregnancy complicted by the 581–592. nephrotic syndrome. J Obstet Gynaecol Br Common- 211 Meekins JW et al. A study of placental bed spiral wealth 1971; 78: 786–790. arteries and trophoblast invasion in normal and sev- 191 Wood SM, Burnett D, Studd J. Selectivity of pro- ere pre-eclamptic pregnancies. Br J Obstet Gynaecol teinuria during pregnancy assessed by different 1994; 101: 669–674. methods. Perspect Nephrol Hypertens 1976; 5: 75–83. 212 Brosens I, Dixon HG, Robertson WB. Fetal growth 192 Nisell H et al. Blood pressure and renal function retardation and the arteries of the placental bed. Br J seven years after pregnancy complicated by hyper- Obstet Gynaecol 1977; 94: 656–663. tension. Br J Obstet Gynaecol 1995; 102: 876–881. 213 Pijnenborg R et al. Attachment and differentiation in 193 Bar J et al. Microalbuminuria after pregnancy compli- vitro of trophoblast from normal and preeclamptic cated by pre-eclampsia. Nephrol Dialysis Transplant human placentas. Am J Obstet Gynecol 1996; 175; 1999; 14: 1129–1132. 30–36. 194 Seely EW et al. A prospective study of calciotrophic 214 Zhou Y, Damsky CH, Fisher SJ. Preeclampsia is asso- hormones in pregnancy and post partum: reciprocal ciated with failure of human cytotrophoblasts to changes in serum intact parathyroid hormone and mimic a vascular adhesion phenotype. One cause of 1,25-dihydroxyvitamin D. Am J Obstet Gynecol 1997; defective endovascular invasion in this syndrome? J 176: 214–217. Clin Investig 1997; 99: 2152–2164. 195 Tauﬁeld PA et al. Hypocalciuria in preeclampsia. N 215 Johansen M et al. Trophoblast deportation in human Engl J Med 1987; 316: 715–718. pregnancy—its relevance for pre-eclampsia. Placenta 196 Sanchez-Ramos L et al. Calcium excretion in preec- 1999; 20: 531–539. lampsia. Obstet Gynecol 1991; 77: 510–513. 216 von Dadelszen P, Hurst G, Redman CW. Supernatants 197 Izumi A et al. Calcium-to-creatine ratio in spot urine from co-cultured endothelial cells and syncytiotro- samples in early pregnancy and its relation to the phoblast microvillous membranes activate peripheral

Journal of Human Hypertension Hypertension in pregnancy F Broughton Pipkin et al 724 blood leukocytes in vitro. Hum Reproduct 1999; 14: 234 Knock GA, Poston L. Bradykinin-mediated relaxation 919–924. of isolated maternal resistance arteries in normal 217 Myatt L, Miodovnik M. Prediction of preeclampsia. pregnancy and preeclampsia. Am J Obstet Gynecol Seminars Perinatol 1999; 23: 45–57. 1996; 175: 1668–1674. 218 Clasp CG. A randomised trial of low-dose aspirin for 235 Oguogho A, Aloamaka CP, Ebeigbe AB. Depressed the prevention and treatment of pre-eclampsia among endothelium-dependent relaxation responses to ace- 9364 pregnant women. Lancet 1994; 343: 619–629. tylcholine and histamine in isolated human epigas- 219 Levine RJ et al. Trial of calcium to prevent preec- tric arteries from pre-eclamptic women. Clin Auto- lampsia. N Engl J Med 1997; 337: 69–76. nomic Res 1996; 6: 153–155. 220 Sibai BM. Prevention of preeclampsia: a big disap- 236 Pascoal I et al. Preeclampsia selectively impairs pointment. Am J Obstet Gynecol 1998; 179: 1275– endothelium-dependent relaxation and leads to oscil- 1278. latory activity in small omental arteries. J Clin 221 Duley L. Magnesium sulphate in eclampsia. Lancet Investig 1998; 101: 464–470. 1998; 352: 67–68. 237 Ashworth JR et al. Mechanisms of endothelium- 222 Taylor RN et al. Women with preeclampsia have dependent relaxation in myometrial resistance ves- higher plasma endothelin levels than women with sels and their alteration in preeclampsia. Hypertens normal pregnancies. J Clin Endocrinol Metab 1990; Pregnancy 1999; 18: 57–71. 71: 1675–1677. 238 Rodgers GM, Taylor RN, Roberts JM. Preeclampsia is 223 Hsu CD et al. Elevated circulating thrombomodulin associated with a serum factor cytotoxic to human in severe preeclampsia [published erratum appears in endothelial cells. Am J Obstet Gynecol 1988; 159: Am J Obstet Gynecol 1994; 171: 1165]. Am J Obstet 908–914. Gynecol 1993; 169: 148–149. 239 Lorentzen B et al. Sera from preeclamptic women 224 Kraayenbrink AA et al. Endothelial vasoactive increase the content of triglycerides and reduce the mediators in preeclampsia. Am J Obstet Gynecol release of prostacyclin in cultured endothelial cells. 1993; 169: 160–165. Thromb Res 1991; 63: 363–372. 225 Friedman SA et al. Plasma cellular fibronectin as a 240 Taylor RN et al. Preeclamptic sera stimulate measure of endothelial involvement in preeclampsia increased platelet-derived growth factor mRNA and and intrauterine growth retardation. Am J Obstet protein expression by cultured human endothelial Gynecol 1994; 170: 838–841. cells. Am J Reproduct Immunol 1991; 25: 105–108. 226 Halligan A et al. Haemostatic, fibrinolytic and endo- 241 Branch DW et al. Sera from preeclamptic patients thelial variables in normal pregnancies and pre- contain factor(s) that stimulate prostacyclin pro- eclampsia. Br J Obstet Gynaecol 1994; 101: 488–492. duction by human endothelial cells. Prostaglandins 227 Lyall F et al. The cell adhesion molecule, VCAM-1, Leukotrienes and Essential Fatty Acids 1992; 45: is selectively elevated in serum in pre-eclampsia: 191–195. does this indicate the mechanism of leukocyte acti- 242 Baker PN, Davidge ST, Roberts JM. Plasma from vation? Br J Obstet Gynaecol 1994; 101: 485–487. women with preeclampsia increases endothelial cell 228 Friedman SA et al. Biochemical corroboration of nitric oxide production. Hypertension 1995; 26: endothelial involvement in severe preeclampsia. Am 244–248. J Obstet Gynaecol 1995; 177: 443–449. 243 DAvidge ST, Baker PN, Roberts JM. NOS expression 229 Krauss T et al. Circulating endothelial cell adhesion is increased in endothelial cells exposed to plasma molecules as diagnostic markers for the early identi- from women with preeclampsia. Am J Obstet Gynecol fication of pregnant women at risk for development 1995; 269: H1106–H1112. of preeclampsia. Am J Obstet Gynecol 1997; 177: 244 Smarason AK, Sargent IL, Redman CWG. Endothelial 443–449. cell proliferation is suppressed by plasma but not 230 Daniel Y et al. Plasma soluble endothelial selectin is serum from women with preeclampsia. Am J Obstet elevated in women with pre-eclampsia. Hum Repro- Gynecol 1996; 1174: 787–793. duct 1998; 13: 3537–3541. 245 Endresen MJR et al. Serum from preeclamptic women 231 Bellart J et al. Tissue factor levels and high ratio of induces vascular cell adhesion molecule-1 fibrino peptide A: D-dimer as a measure of endo- expression on human endothelial cells in vitro—a thelial procoagulant disorder in pre-eclampsia. Br J possible role of increased circulating levels of free Obstet Gynaecol 1999; 106: 594–597. fatty acids. Am J Obstet Gynecol 1998; 179: 665–670. 232 Tempfer C et al. Serum levels of ELAM-1, but not 246 Haller H et al. Endothelial-cell permeability and pro- CD44, predict the clinical outcome of patients with tein kinase C in pre-eclampsia. Lancet 1998; 351: preeclampsia. Hypertens Pregn 1999; 18: 45–55. 945–949. 233 McCarthy AL et al. Abnormal endothelial cell func- 247 Ashworth JR et al. Plasma from pre-eclamptic women tion of resistance arteries from women with preec- and functional change in myometrial resistance lampsia. Am J Obstet Gynecol 1993; 168: 1323–1330. arteries. Br J Obstet Gynaecol 1998; 105: 459–461.

Journal of Human Hypertension