James L. Mills, Rebecca DerSimonian, Elizabeth Raymond, Jason D. Morrow, L. Jackson Roberts II, John D. Clemens, John C. Hauth, Patrick Catalano, Baha Sibai, L. B. Curet, Richard J. Levine. Prostacyclin and Thromboxane Changes Predating Clinical Onset of PreeclampsiaA Multicenter Prospective Study. JAMA. 1999;282(4):356–362. doi:10.1001/jama.282.4.356
Author Affiliations: Epidemiology Branch (Drs Mills, Clemens, and Levine) and Biometry and Statistics Branch (Dr DerSimonian), Division of Epidemiology, Statistics and Prevention Research, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Md; Family Health International, Clinical Trials Division, Durham, NC (Dr Raymond); Departments of Clinical Pharmacology and Medicine, Vanderbilt University, Nashville, Tenn (Drs Morrow and Roberts); and the Departments of Obstetrics and Gynecology, University of Alabama-Birmingham (Dr Hauth), MetroHealth Medical Center, Cleveland, Ohio (Dr Catalano), University of Tennessee-Memphis (Dr Sibai), and University of New Mexico, Albuquerque (Dr Curet).
Context An imbalance in vasodilating (prostacyclin [PGI2]) and vasoconstricting
(thromboxane A2 [TxA2]) eicosanoids may be important
in preeclampsia, but prospective data from large studies needed to resolve
this issue are lacking. Because most trials using aspirin to reduce TxA2 production have failed to prevent preeclampsia, it is critical to
determine whether eicosanoid changes occur before the onset of clinical disease
or are secondary to clinical manifestations of preeclampsia.
Objective To determine whether PGI2 or TxA2 changes occur
before onset of clinical signs of preeclampsia.
Design, Setting, and Participants Multicenter prospective study from 1992 to 1995 of subjects from the
placebo arm of the Calcium for Preeclampsia Prevention Trial. Women who developed
preeclampsia (n=134) were compared with matched normotensive control women
Main Outcome Measures Excretion of urinary metabolites of PGI2 (PGI-M) and TxA2 (Tx-M) as measured from timed urine collections obtained prospectively
before 22 weeks', between 26 and 29 weeks', and at 36 weeks' gestation.
Results Women who developed preeclampsia had significantly lower PGI-M levels
throughout pregnancy, even at 13 to 16 weeks' gestation (long before the onset
of clinical disease); their gestational age–adjusted levels were 17%
lower than those of controls (95% confidence interval [CI], 6%-27%; P=.005). The Tx-M levels of preeclamptic women were not
significantly higher overall (9% higher than those of controls; 95% CI, −3%
to 23%; P=.14). The ratio of Tx-M to PGI-M, used
to express relative vasoconstricting vs vasodilating effects, was 24% higher
(95% CI, 6%-45%) in preeclamptic women throughout pregnancy (P=.007).
Conclusions Our results show that reduced PGI2 production, but not increased
TxA2 production, occurs many months before clinical onset of preeclampsia.
Aspirin trials may have failed because an increase in thromboxane production
is not the initial anomaly. Future interventions should make correcting prostacyclin
deficiency a major part of the strategy to balance the abnormal vasoconstrictor-vasodilator
ratio present in preeclampsia.
The etiology of preeclampsia is one of the major unsolved mysteries
in obstetrics. Because vascular changes are a prominent feature of this condition,
numerous investigations during the past decade have studied the possibility
that preeclampsia is causally linked to an imbalance in the formation of prostacyclin
(PGI2), a vasodilator, and thromboxane A2 (TxA2), a vasoconstrictor.1- 8
The results of these studies, however, have been conflicting and confusing.
One problem is that assessment of eicosanoid formation was usually performed
after the onset of symptoms and, in some instances, after initiation of treatment.
Thus, it is difficult, if not impossible, to determine whether changes in
PGI2 or TxA2 are the cause or the result of the disease
(or even its therapy). Prospective studies have been carried out in an attempt
to circumvent the problem, but the number of women who developed preeclampsia
was small, limiting the power of these studies to detect differences.
Despite the lack of definitive evidence that thromboxane (Tx) levels
are elevated in preeclampsia, aspirin, which reduces levels of Tx,9 has been used in preeclampsia prevention trials. The
results generally have been disappointing.
To determine whether alterations in production of PGI2 and
TxA2 are important in the pathophysiology of preeclampsia, we conducted
a prospective study using the 2294 women enrolled in the placebo group of
the National Institute of Child Health and Human Development Calcium for Preeclampsia
Prevention Trial (CPEP). This cohort provided a sufficiently large number
of women who later developed preeclampsia to detect abnormalities in TxA2 or PGI2 production long before onset of clinical symptoms
CPEP methods have been published in detail elsewhere.10,11
Between April 1992 and March 1995, nulliparous women between 11 and 21 gestational
weeks were screened, and follow-up continued until September 30, 1995. Those
who were diagnosed as having conditions that could markedly influence the
study end points, affect the absorption or metabolism of calcium, make the
use of calcium inappropriate, or impede compliance were dropped. Eligible
women then underwent a single-blind test of compliance with a placebo. Those
who took at least 75% of the tablets were enrolled if they: (1) had blood
pressures consistently below 135/85 mm Hg; (2) had a dipstick test for urinary
protein with negative or trace results; and (3) were between 13 and 21 weeks
pregnant determined by ultrasonography.
Subjects were randomly assigned to receive either calcium supplements
or placebo tablets. For this analysis, only women who received the placebo
were included to eliminate any possible effect of calcium on eicosanoid production.
Other than prenatal vitamins, acetaminophen, and a non–calcium-containing
antacid, which were provided to participants, women were instructed not to
take vitamin supplements, analgesics, or antacids. The study protocol was
approved by the institutional review boards at the 5 participating medical
centers, and all women gave written informed consent.
All women participating in the CPEP trial were asked to provide 24-hour
urine specimens at enrollment (13 to 21 weeks), between 26 and 29 weeks',
and at 36 weeks' gestation. When a woman was found to have developed hypertension
or proteinuria, another specimen was collected if possible. All samples were
frozen, sent to the central CPEP repository, and stored at −70°C.
Specimens were rejected for the study if they contained visible blood (n=3),
if the total volume was less than 400 mL or the collection period was less
than 10 hours (n=22), or if rupture of membranes occurred before collection
was complete, unless the specimen was collected by catheter (n=11) (see "Diagnosis
of Preeclampsia" section). In all, 11% of specimens from preeclamptic women
and 6% of specimens from controls were rejected.
Women were evaluated for preeclampsia by a trained staff member every
4 weeks through the 29th week of gestation, every 2 weeks through the 35th
week, and weekly thereafter. Medical records of all outpatient visits and
hospitalizations were reviewed for blood pressure and urine protein. Women
were not asked to perform any self-diagnosis. A preliminary diagnosis of preeclampsia
was made by the principal investigator after reviewing the medical data. The
final diagnosis was determined by 3 researchers working independently, assisted
by a computer algorithm. All women with a preliminary diagnosis were considered
potential cases; those in whom the diagnosis was not confirmed were dropped.
Preeclampsia was defined as pregnancy-associated hypertension (diastolic
blood pressure ≥90 mm Hg on 2 occasions 4 to 168 hours apart) and pregnancy-associated
proteinuria (≥300 mg of protein in a 24-hour urine specimen, or 2 urine
specimens 4 to 168 hours apart containing ≥1+ protein by dipstick measurement,
or a single urine specimen with a protein-creatinine ratio of ≥0.35 or
containing ≥2+ protein by dipstick). Collection of samples via catheter
was mandatory after rupture of membranes or in the presence of vaginitis.
Severe preeclampsia was defined as preeclampsia with a diastolic blood pressure
of at least 110 mm Hg or severe proteinuria (≥3.5 g per 24 hours or ≥3+
by dipstick on 2 occasions). Eclampsia and HELLP syndrome (hemolysis, elevated
liver enzymes, and low platelet count) were also counted as severe preeclampsia.11
Controls were randomly selected from women with no preliminary diagnosis
of preeclampsia or any related disorder (gestational hypertension or proteinuria).
Also, controls were frequency-matched to cases (using random numbers) by study
center and date of last menstrual period within 1 calendar year. Subjects
were matched on study center because the preeclampsia rates differed significantly
between centers, perhaps due to differences in the pathophysiology of the
disease. In addition, the centers may have used slightly different procedures
for collecting, preparing, and storing specimens. The rationale for matching
by last menstrual period was to ensure that the duration of storage of specimens
would be comparable in cases and controls. One control was selected per case
by random number.
Because women did not always provide specimens on schedule, specimens
collected before 154 days were classified as "early (baseline)," between 154
and 237 days as "middle," and after 237 days as "late." To avoid potential
problems with deterioration of urine specimens over time, samples were selected
for assay based on initial clinical reports of preeclampsia and were assayed
over the course of the study. When the final diagnosis was established, subjects
were reclassified as necessary for data analysis.
Endogenous production of TxA2 and PGI2 was assessed
by measuring excretion of their major urinary metabolites, 11-dehydro-TxB2 (Tx-M), and 2,3 dinor-6-keto PGF1α (PGI-M), respectively.
Data are expressed as picograms per milligram of creatinine to correct for
intersubject differences in creatinine clearance. Both compounds were quantified
by stable isotope dilution methods using gas chromatography and negative ion
chemical ionization mass spectrometry.12,13
The accuracy of both assays is greater than 95% and precision less than ±10%.
Interday and intraday variability were less than 12% for both assays. Laboratory
personnel were blinded to case-control status. The Tx-M/PGI-M ratio is a commonly
used index of the net vasoconstricting and vasodilating effects of TxA2 and PGI2, respectively.
Differences in prostaglandin metabolite production between cases and
control subjects were evaluated by a repeated-measures analysis using the
PROC MIXED procedure in SAS,14 which adjusts
for the correlation of observations for the same woman.15
For comparisons that required a Gaussian assumption, the data were transformed
to the logarithmic scale. Differences in the baseline characteristics of the
case and control groups were evaluated by the Wilcoxon 2-sample test (continuous
variables) or the Fisher exact test (categorical variables). All analyses
were performed using SAS version 6.12. P<.05 was
Of the 154 subjects with a preliminary diagnosis of preeclampsia, 15
had a final diagnosis of pregnancy-associated hypertension or pregnancy-associated
proteinuria only and were excluded. One had no abnormalities and was used
for the control pool. Of the remaining 138 cases, 4 were dropped because they
had used nonsteroidal anti-inflammatory drugs. The final case group consisted
of 89 mild and 45 severe cases (total 134). Of the 154 potential controls
selected, 9 were lost to follow-up, 3 had pregnancy-associated hypertension
or pregnancy-associated proteinuria, and 4 used nonsteroidal anti-inflammatory
drugs. One normal subject who met the matching criteria was transferred from
the case group, bringing the total to 139.
and controls did not differ significantly in age, height, gestational week
at entry, racial distribution, smoking, education, prior induced abortions,
or marital status (Table 1). As
reported previously,16 preeclamptic women had
significantly greater weights, body mass indexes, upper arm circumferences,
and higher systolic and diastolic blood pressures (P<.001
for all). The median gestational age at diagnosis of preeclampsia was 274
days due to a large number of mild cases diagnosed near term (mean, 39 weeks)
and a relatively small number of severe cases diagnosed earlier (mean, 36.9
weeks). Baseline specimens were collected, on the average, 150 days before
Figure 1 shows the distribution and mean levels of PGI-M , Tx-M
, and Tx-M/PGI-M ratio values in preeclampsia cases and controls in the baseline,
middle, and late time periods. Levels of PGI-M were significantly lower in
the preeclampsia group in the baseline period. Levels of Tx-M were significantly
higher in the preeclampsia group in the late period. The vasoconstrictor-vasodilator
ratio (Tx-M/PGI-M) was significantly higher in the preeclampsia group at both
the baseline and late periods (P=.04 and P=.01, respectively) and of borderline significance at the middle period
It is important to note that it is not possible to take into
account several important factors in this crude analysis, namely, gestational
age strongly influences both PGI-M and Tx-M levels, and crude data do not
reflect changes over time in individual women. Therefore, the mixed model
was used to account for the effect of gestational age and longitudinal changes
in values from the same subject.
The preeclampsia and control
groups' PGI-M levels are shown by 4-week intervals in Figure 2, A. The mixed model demonstrates that the preeclampsia
group had lower PGI-M levels at virtually all intervals. The preeclampsia
group had an overall decrease of 17% (95% confidence interval [CI], 6%-27%)
in PGI-M levels (P=.005) compared with the control
group. In particular, the preeclampsia group had a significantly lower level
(P=.03) at weeks 13 to 16, long before the onset
of clinical disease.
Levels of Tx-M are shown in Figure 2, B. Contrary to what has been reported in the literature,
the preeclampsia group had significantly lower levels (P=.01) of Tx-M in the earliest time period (13-16 weeks). This may
have been a chance event, because the preeclampsia group had higher Tx-M levels
in each subsequent time interval. Over the entire study period, the preeclampsia
group had a median Tx-M level 9% higher (95% CI,−3 to 23%) than controls,
but the difference was not significant (P=.14). If
one wishes to argue that the early levels were a statistical accident, or
that changes in Tx-M occur later in pregnancy, it can be said that beyond
the baseline period (gestational age greater than 21 weeks) the preeclampsia
group had a 15% higher median Tx-M (95% CI, 1%-31%) than controls. The difference
was significant (P=.04). As shown in Figure 2, B, Tx-M levels clearly increased more rapidly in the preeclampsia
group. The preeclampsia group had a consistently higher Tx-M/PGI-M ratio after
16 weeks, indicating a predominant vasoconstrictor effect (Figure 2, C). The preeclampsia group showed an overall increase
of 24% (95% CI, 6-45%) compared with the control group (P=.007).
Next we compared women with severe preeclampsia
with the other women in the preeclampsia group. Levels of PGI-M and Tx-M and
the Tx-M/PGI-M ratio were not significantly different in the baseline, middle,
or late time periods or overall using the mixed procedure.
Many studies have reported eicosanoid values obtained after onset of
clinical preeclampsia. To determine whether such values reflect the preclinical
state, we also examined Tx-M and PGI-M levels in 51 specimens obtained after
onset of clinical preeclampsia. Levels of TxM were significantly higher (P<.001) in specimens collected after rupture of membranes
(n=28) (age-adjusted mean, 2208 pg/mg of creatinine) than in those collected
before (n=23) (age-adjusted mean, 735 pg/mg of creatinine). Women treated
with magnesium sulfate (n=20) also had significantly higher levels (P<.001) than women who were untreated at the time of
collection (n=31) (1998 vs 1097 pg/mg of creatinine, respectively). The group
that had rupture of membranes had a significantly increased Tx-M/PGI-M ratio
(1.1) compared with those who had intact membranes (0.7) (P=.018). Those who were treated with magnesium sulfate had a significantly
higher ratio (1.7) than those who were untreated (0.6) (P=.002). Results for PGI-M were inconsistent; magnesium sulfate treatment
was associated with lower PGI-M levels if membranes were ruptured and higher
levels if membranes were intact.
It should be emphasized that there could have been major differences
in the clinical condition of the treated and untreated groups (eg, clinicians
are more likely to treat severe cases and those in whom delivery is not imminent)
and that samples were available for only a small proportion of study subjects.
These findings illustrate the difficulty of using samples obtained after onset
of clinical disease.
To determine whether the results could be due to underlying differences
in the 2 groups, particularly differences in initial blood pressure, we adjusted
for factors in which the groups differ significantly (Table 1). Adjusting for baseline systolic and diastolic blood pressure,
weight, body mass index, and arm circumference, either individually or together,
did not change the results. Levels of PGI-M remained significantly lower in
preeclamptic women for all adjustments (P<.02
for all). Similarly, the Tx-M/PGI-M ratio remained significantly higher in
the preeclamptic women (P<.03 for all). No adjustments
produced a significant association between preeclampsia and Tx-M levels. Thus,
the association between preeclampsia and PGI-M was not explained by underlying
differences in the study groups.
Preeclampsia remains a major cause of morbidity and mortality in pregnancy.
Preventive efforts using aspirin or calcium supplements have produced mixed,
but generally disappointing, results.11,17- 21
Although interest in the role of eicosanoids is high, research has been hampered
by the lack of prospectively collected data for large numbers of affected
This is by far the largest prospective study of preeclampsia
and eicosanoids performed to date. Because of the large number of specimens
available over the course of pregnancy, we were able to document changes in
PGI2, Tx, and Tx/PGI2 ratios many months before the
onset of signs in women who would eventually develop preeclampsia.
Women who developed preeclampsia had significantly lower PGI2
production and higher TxA2/PGI2 ratios, indicative of
vasoconstriction beginning in the second trimester of pregnancy and persisting
throughout. Women who developed preeclampsia did not have significantly higher
Tx levels overall, yet they did have significantly higher Tx levels if only
the latter part of pregnancy (after 21 weeks) is considered, suggesting that
the rise in Tx levels could be a secondary event.
suggest that a deficiency in PGI2 production is an early and important
event in women who develop preeclampsia. This is associated with an early
increase in the vasoconstrictor-vasodilator (TxA2/PGI2)
ratio. While it is not yet clear whether PGI2 deficiency is the
primary cause of preeclampsia, our results indicate that endothelial dysfunction
may be a critical factor because of the role of the endothelium in PGI2 production.
Our findings may explain the failure of trials
of low-dose aspirin to prevent preeclampsia. These trials assumed that platelet
hyperactivity, mediated by excess Tx, produced preeclampsia. Low-dose aspirin
is known to reduce Tx production by selective inhibition of platelet eicosanoid
production with sparing of PGI2 production by the vascular endothelium.
Our data show that decreased PGI2, rather than increased Tx, is
the earlier, probably more important, eicosanoid disruption of preeclampsia,
and Tx elevation is a later, perhaps secondary event. Thus, it is not surprising
that reducing Tx did not prevent preeclampsia.
of prostacyclin and thromboxane metabolites in preeclampsia have produced
inconsistent results. While several studies reported significantly lower PGI2 levels in preeclamptic subjects,1- 5
the only long-term prospective study6 found
no difference. For TxA2, some studies found no difference between
patients with and without preeclampsia,1,3- 6
while others have found significantly higher levels in preeclamptic patients.7,8 Reasons for these discordant results
are easy to find. The problems with collecting data after onset of symptoms
have been noted; indeed, our own analysis of samples collected after rupture
of membranes and treatment produced uninformative results. The problem with
low power in some prospective studies has also been noted previously and could
account for the difference between their findings and ours.
summary, this large, prospective study demonstrated that, in women who develop
preeclampsia, PGI2 production is significantly decreased and the
vasoconstrictor-vasodilator ratio is significantly increased many months before
signs appear. Clinically, these findings are important because they suggest
that the primary therapeutic strategy for preventing the complications of
preeclampsia should be restoring PGI2 homeostasis; reducing Tx
production might be a secondary goal to assist in reducing the elevated vasoconstrictor-vasodilator
ratio. Long-term infusion of PGI2 has proved useful in treating
endothelial dysfunction and vasoconstriction in primary pulmonary hypertension
and pulmonary complications of connective tissue disorders.22,23
Future trials should explore the use of PGI2 or PGI2
analogs in the prevention and treatment of preeclampsia.