Correspondence: Dr A. Metin Giilmezoǧlu, UK Cochrane Centre, Summertown Pavilion, Middle Way, Oxford OX2 7LG, UK.
Objective To determine whether antioxidant therapy alters the disease process in severe early onset pre-eclampsia, in support of the hypothesis that increased lipid peroxides and reactive oxygen species production play an important role in the pathogenesis of the disease.
Setting Two tertiary care, referral hospitals in Johannesburg, South Africa.
Participants Women with severe pre-eclampsia diagnosed between 24 and 32 weeks of gestation.
Intervention Combined antioxidant treatment with vitamin E (800 IU/day), vitamin C (1000 mg/day), and allopurinol (200 mg/day).
Main outcome measures Primary outcomes: 1. prolongation of pregnancy and 2. biochemical assessment of lipid peroxides and antioxidants. Secondary outcomes: data on maternal complications, side effects of treatment, infant outcomes and regular assessment of haematologic and renal parameters.
Results The proportion of women delivered within 14 days in the antioxidant group was 52% (14/27) compared with 76% (22/29) in the placebo group (relative risk 0.68, 95% confidence interval 0.45–1.04). One woman in each group had eclampsia. Eleven women (42%) in the antioxidant and 16 (59%) in the placebo group required two antihypertensives for blood pressure control. Trial medications were well tolerated with few side effects. Lipid peroxide levels were not significantly altered in the antioxidant and placebo groups. Serum uric acid levels decreased and vitamin E levels increased significantly.
Conclusion The results of this explanatory randomised trial do not encourage the routine use of antioxidants against pre-eclampsia. However, further research with modified strategies such as earlier initiation of therapy or different combinations seem worthwhile.
Pre-eclampsia is a systemic disorder associated with the presence of placental tissue. The pathogenetic mechanisms involved in the initiation and progression of the disease process are not clear. Pre-eclampsia is associated with inadequate or shallow trophoblastic invasion of the decidual vessels, resulting in a high resistance-low flow uteroplacental circulation which causes placental ischaemia and hypoxia. The events behind this inadequate trophoblast invasion are unknown, but the trophoblast behaviour is probably modified by an immunologic factor1. Roberts et al2 and Hubel et al3 hypothesised that placental ischaemia could trigger lipid peroxidation causing endothelial damage, which would then initiate the self propagating course and the systemic manifestations of the disorder.
This study has been funded in part by South African Medical Research Council and South African Druggists. The trial medications (active and placebo) have been donated by South African Druggists and Her-Sol Laboratories, Johannesburg, South Africa.
Lipid peroxides are formed in two ways. The first is the free radical process in which free radicals form in excess, overcome the antioxidant defences and attack polyunsaturated fatty acids. These free radicals initiate chain reactions within the hydrophobic core of the cell membrane lipid bilayer, leading to membrane fragmentation. The second is an enzymatic process involving lipooxygenase and cyclooxygenase4. Free radicals and lipid peroxides can initiate reactions that damage cells and cell membranes; however, the latter mechanism is often a late component which accompanies cellular injury5.
The mechanism of uric acid elevation in pregnancy is controversial. Most obstetricians would agree that it is an indicator of the severity of the disease, whether the problem lies in the renal handling of uric acid or in the placenta6,7. It has been suggested that ischaemia of the placental villi may damage the nuclear rich syncytiotrophoblasts resulting in a high rate of purine catabolism, which would raise uric acid levels and possibly correlate with the severity of the disease8. The fetus can also generate uric acid at concentrations high enough to be detected in the maternal blood9.
An important consequence of ischaemia is inadequate cellular oxygenation. Oxygen is used in the cell to generate energy in the form of adenosine triphosphate (ATP). When there is inadequate oxygen supply to the cells a process called ATP degradation occurs. ATP degradation leads to the release of purine intermediate products and increased uric acid as the end product. This process is important for two reasons. Firstly, when purines are degraded to uric acid they cannot be re-used within the energy generating system of the cell. Secondly, under hypoxic conditions, the enzyme xanthine dehydrogenase is shifted to the oxidase form generating free oxygen radicals10. This mode of cellular injury has been studied extensively in adult respiratory distress syndrome11 and neonatal hypoxic-ischaemic-encephalopathy12,13. A family with mitochondrial dysfunction characterised by a deficiency in energy production within the cells and a high incidence of pre-eclampsia/eclampsia has been reported14. The finding of morphologic evidence of mitochondrial injury in pre-eclamptic women also supports this mechanism15.
Oxidative stress can also cause rises in intracellular free calcium ions by interfering with normal calcium sequestering mechanisms5. In the presence of intracellular energy deficiency, calcium ions that enter the cell cannot be removed by the calcium-pump. Calcium supplementation reduces parathyroid hormone release and intracellular calcium, which decreases smooth muscle reactivity. Calcium supplementation is currently a promising intervention for the prevention of hypertensive disorders of pregnancy16. Calcium channel blockers may also have a beneficial effect by preventing calcium influx into the cell17.
In a preliminary case-control study lipid peroxide levels were significantly highest in women with eclampsia followed by severe pre-eclampsia and healthy pregnant controls, suggesting an increase in lipid peroxides which followed the disease severity18. Our hypothesis for this study was that the pathology that follows placental ischaemia is a local cellular energy deficiency which initiates excessive free radical production and triggers the multiple chain reactions which result in widespread endothelial damage. Another hypothesis was that by giving antioxidants which act at different sites we could slow down the disease process or interrupt the self-propagatory course of the disease. This effect would be manifest in prolonga- tion of pregnancy.
An exploratory randomised trial was designed in which allopurinol, vitamin E and vitamin C were administered as antioxidants to a group of women with severe early onset pre-eclampsia who were to be managed conservatively. The main clinical outcome measure was prolongation of pregnancy. Serial biochemical assessments were also undertaken to investigate the effect of this therapeutic regimen on lipid peroxides (as measured by the malondialdehyde-thiobarbituric acid reactive assay[MDA-TBAR]) as well as antioxidant levels. The secondary outcomes were, maternal morbidity, side effects, fetal and neonatal outcome and changes in haematological and renal function.
The trial was conducted at two tertiary care academic hospitals which serve a low income urban population with a mixed racial composition. The study group comprised women who were admitted to the antenatal wards with a diagnosis of severe pre-eclampsia as defined by Odendaal et al19. Briefly, these women had at least 2+proteinuria on urine dipstix testing (in at least two consecutive tests 4 to 6 hours apart), with a blood pressure of 160/110 mmHg, or 3+ proteinuria with blood pressure 150/100 mmHg. With these criteria proteinuric hypertensive women were selected regardless of the presence or absence of underlying hypertensive or renal disease. Gestational age limits for inclusion were between 24 and 32 weeks. In these two hospitals women with severe pre-eclampsia before 24 weeks are advised termination of pregnancy while after 32 weeks, delivery can usually be undertaken. Only women with a live single fetus and who did not have any systemic disorder, such as diabetes or systemic lupus erythematosus or allergy to the study medications, were eligible. Conservative management of early onset severe pre-eclampsia is the standard treatment in these hospitals once the woman's condition is stabilised in the ward and biochemical indices are acceptable. The latter are absence of significant renal impairment, the syndrome of haemolysis, elevated liver enzymes and low platelets (HELLP) syndrome, or thrombocytopenia alone. These women are advised to stay in hospital until delivery with weekly betamethazone injections up to 32 to 34 weeks and frequent fetal and maternal monitoring. Uric acid levels are not used for decision making.
The sample size calculation was based on reports indicating that 80% of such women needed delivery within 14 days of admission20. We estimated that the number of women delivering within 14 days would be halved to 40%, which yielded a sample size of 54 women (α= 0.05; l–β= 80%).
Conduct of the trial
Once conservative management was initiated the women were approached and asked to participate in the trial. Of the 59 women approached, 56 gave written informed consent and comprised the study population. Logistical problems, such as distance, or difficulties with communication, with the second hospital resulted in a high proportion of missed data with regard to blood and placental samples, especially for lipid peroxide assays which constituted the laboratory part of the trial. For this reason enrolment in this hospital was stopped after eight women; the women in this hospital who gave consent for participation in the trial were transferred to the co-ordinating hospital.
The study medications were vitamin E (400 IU twice daily), vitamin C (500 mg tabs twice daily) and allo-purinol (100 mg twice daily) or matching placebos, administered twice daily. For allopurinol and vitamin C, placebos were identical. Vitamin C placebos were used as placebos for vitamin E as well because it was not possible to obtain two separate sets of placebos from the supplier. Vitamin E tablets were similar in appearance but slightly thicker and cream coloured compared with the lighter colour of their placebos. To preserve double blinding all medications were placed in dark brown coloured bottles and 20-day supply of all three medications (active or placebo) were placed in consecutively numbered sealed opaque paper bags. None of the authors was involved in the day to day management of the women. One of the authors (GJH) had access to the code. This was necessary for those who remained undelivered for more than 20 days (only a 20 day treatment could fit into the bottles), where a separately coded, ‘second’ treatment pack could be used. The treatment packs were randomised by computer generated random numbers in blocks of ten. The randomisation was done by an independent researcher who was not involved in the study. As all of the women were kept in hospital, these medications were left with them and the ward sisters were asked to observe whether the women were taking their medications or not. After delivery the bottles were taken back and any remaining tablets counted at the end of the trial to check for compliance.
There were no exclusions after enrolment. All women remained in their allocated group for analysis (intention-to-treat) regardless of whether they continued their allocated treatment or not. The duration of pregnancy from the time of randomisation had a nonnormal distribution and therefore was analysed by the nonparametric Mann-Whitney test. Wilcoxon paired ranks test was used in the analysis of umbilical artery Doppler flow measurements. Categorical data were analysed by the χ2 test with Fisher's exact test where appropriate. Relative risks (RR) and 95% confidence intervals (CI) were calculated for dichotomous outcomes using Epi Info version 6.02.
Biochemical assessment methods
For the purposes of this study the full blood counts, liver function tests, urea and electrolytes, as well as lipid peroxide and vitamin E assays, were performed on day 0, 3, 7 and weekly afterwards until delivery. An attempt was made to repeat the tests close to the time of delivery if they had not been performed within the preceding 24 to 48 hours. All blood samples were centrifuged and the supernatant snap frozen at −70°C within 30 minutes of collection. For lipid peroxides, malondialdehyde-thiobarbituric acid reactive (MDA-TBAR) assay was performed with the standard curve generated using tetramethoxypropane as working stock21. Sera for vitamin E measurements were handled as above and protected from light until analysis. Measurements were made by high performance liquid chromatography22 and the assay validated by the United States National Institute of Standards and Technology Standard Reference Material 968 a.
Fetal monitoring consisted of daily nonstress tests, weekly Doppler umbilical artery resistance index measurements and ultrasound scanning every two weeks for fetal growth. Renal function was monitored by urea, creatinine and weekly creatinine clearance measurements. More frequent testing or other relevant tests (such as lupus anticoagulant or coagulation tests) were decided on by the obstetric staff.
All laboratory tests were performed without any knowledge of trial group allocations.
The trial was approved by the Committee for Research on Human Subjects of the University of the Witwatersrand.
Of the 56 women enrolled, 27 were in the antioxidant and 29 were in the placebo group. The groups were comparable with regard to entry characteristics (Table 1). The number of women who were delivered within 14 days (the main outcome measure) was 14 (52%) in the antioxidant group and 22 (76%) in the placebo group (RR 0.68, 95% CI 0.45–1.04). Figure 1 is a Kaplan-Meier survival curve, ‘survival’ being the time from enrollment to delivery. Although the logrank test for survival (number still pregnant) was not statistically significant (P= 0.24), there was a trend towards women in the antioxidant group delivering later. There were no maternal deaths but serious complications occurred in both groups. In the antioxidant group, one woman had mild postpartum pulmonary oedema and one woman had eclampsia with the HELLP syndrome. In the placebo group, two women had mild postpartum pulmonary oedema, one had eclampsia and one had disseminated intravascular coagulation with shock and renal failure which required dialysis. All three cases of pulmonary oedema occurred within 24 hours after delivery. Three women in the antioxidant and six in the placebo group received magnesium sulphate. One woman in each group received this treatment because of eclampsia. Placental abruption was diagnosed clinically in two of the antioxidant women (7%) and six controls (21%). Sixteen women (59%) in the antioxidant group and 11 (38%) in the placebo group were delivered by pre-labour caesarean section. The reasons for delivery are listed in Table 2.
Table 1. Baseline data in the antioxidant and placebo groups expressed as median [range] or proportions (%).
Antioxidant group (n-27)
Placebo group (n= 29)
Gestational age (weeks)
Mean arterial pressure
Impaired fetal growth at enrolment
Proteinuria (g/24 h)
Serum uric acid (mmol/L)
Table 2. Reasons for delivery in the two groups. Values are given as n (%)
Antioxidant group (n= 27)
Placebo (n = 28*)
*Missing data for 1 woman. †Seven women in each group delivered because of intrauterine fetal death.
Spontaneous labour and delivery
Obstetric (rupture of membranes)
All but one woman in each group were put on anti-hypertensive treatment. Alpha methyldopa was the first choice and nifedipine second in the majority of the patients. Eleven women (41%) in the antioxidant and 16 (57%) in the placebo group required two antihypertensives for blood pressure control (RR 0.74, 95% CI 0.43–1.28).
The umbilical artery resistance index measurements were similar in the two groups at enrolment and one week after enrolment. The paired measurements within groups also did not reveal any significant changes.
Compliance and adverse events
A compliance index was calculated for each trial medication. This was the ratio of the number of tablets thought to have been taken (according to the count) to the expected number if compliance was perfect. The median compliance indexes were 89% (antioxidant) compared with 100% (placebo) for vitamin C, 93% compared with 86% for allopurinol, and 84% compared with 75% for vitamin E. In the antioxidant group one woman complained of an increase in acne, two of transient weakness at the beginning of treatment, and one woman had a skin rash after delivery on the day the medications were stopped. One woman who complained of weakness stopped the trial medications after taking four doses altogether. In the second case it was found that she had been given a higher than prescribed dose of antihypertensives at night. No side effects were reported in the placebo group.
Three women in each group received vitamin C (100 mg/day), together with iron supplementation; one woman in each group received low dose aspirin as part of their management.
The perinatal outcome is shown in Table 3. Poor perinatal outcome was defined as having either stillbirth or neonatal death. The reasons that were thought to cause the poor outcome are shown in Table 4.
Table 3. Perinatal outcomes. Values are given as n (%) or median [range].
* Missing data
Apgar 1 min < 7
Apgar 5 min < 7
Umbilical artery pH < 7-2
Admission to intensive care unit
Table 4. Reasons for poor perinatal outcome. Values are given as n (%).
Antioxidant group (n= 12)
Placebo group (n= 10)
Abruption (diagnosed clinically)
Death before delivery with severely impaired fetal growth
Severe hyaline membrane disease
Grade III intraventricular haemorrhage + sepsis
No specific cause found
Lipid peroxide and vitamin E levels
Baseline MDA-TBAR concentrations were higher in the control group compared with the antioxidant group (P= 0.02). As the women were enroled according to clinical criteria and all other enrolment characteristics were comparable (Table 1), we believe this difference must be a chance effect. MDA levels did not show any consistent changes in the active or placebo groups over the course of the study. There were also no significant differences between groups with respect to changes in MDA-TBAR measurements between baseline and day 3. After day 14, the number of patients in each group (six in active and four in placebo) became too small for meaningful comparison.
Serum vitamin E levels greatly increased in the antioxidant group, whose treatment included vitamin E. Of interest was a statistically significant increase in vitamin E levels in the placebo group (P= 0.02). The change in vitamin E concentrations was evident both in measurements before delivery and paired analyses of the difference between baseline and day 7 measurements. The change in vitamin E in the antioxidant group was greater than that in the placebo group (P= 0.003).
Haematological and renal function parameters
The greatest change was in serum uric acid levels which decreased significantly within the first few days in the antioxidant group and increased slightly in the placebo group. In the antioxidant group the platelet count (median difference between day 7 and baseline, 12 × 109/L [range −59 to 215 × 109]) and white cell counts (median difference 0.6 × 109/L [range −8 to 4.8 × 109]) tended to remain static compared with a trend toward a decrease in platelet counts (median difference 0 [range −190 to 93 × 109]), and a more prominent increase in white cell counts (median difference 1.65 × 109/L [range −2.6 to 11.7 × 109]) in the placebo group, but these changes were not statistically significant. There were no discernible changes in renal function between the two groups.
There were no statistically significant differences between placental MDA-TBAR and glutathione levels in the antioxidant and placebo groups. Placental lipid peroxide and glutathione results have been discussed in more detail separately23.
Control of severe early onset pre-eclampsia remains a problem for those in charge of the care of affected women. The reason for this is that there is no way to treat effectively the underlying disease. Decisions are dictated by gestational age, assessments of fetal well-being and neonatal facilities. Currently, conservative management is widely practised, although it increases the risk of maternal morbidity24,25. The high maternal morbidity rate in both groups reflects the risks of expectant management in severe pre-eclampsia. The perinatal mortality rate in this trial, although higher than developed countries, is comparable to previous reports from South Africa26,27. Neonatal outcome is influenced most by the capacity of neonatal intensive care units, which admit only newborn infants whose birthweight is > 1000 g or who are 28 weeks of gestational age in the two hospitals in which this trial was undertaken. It is not uncommon for a baby to have no intensive unit care or mechanical ventilation, despite being above these limits in these hospitals.
The results did not show any effect of the antioxidant treatments on MDA-TBAR levels. This could be due to several factors. One possibility is that, although all samples were handled similarly and snap frozen within a reasonable time (< 30 min), it may be that the amount of artefactual MDA produced between thawing and measuring, concealed any real differences that might have existed. Secondly, the enrolment to delivery times in this study were relatively short (median 11 and 9 days in active and placebo groups, respectively) and the biochemical effects may take longer to appear. Duthie et al.28 used 1000 mg α-tocopherol/day for 14 days in smokers and did not detect any significant changes in MDA-TBAR despite significant changes in the levels of conjugated dienes. In a more recent study29 the same researchers used 280 mg dl-α tocopherol acetate for 10 weeks and could show a statistically significant suppression of both conjugated dienes and MDA-TBAR.
As expected, vitamin E supplementation increased serum concentrations significantly in the antioxidant group (Table 5). This increase is probably indicative of good compliance. There could be several reasons for the increase in the placebo group. Vitamin E is known to be mobilised from the tissues to sites of oxidative stress29. This could indicate increasing oxidative stress with continuation of pregnancy in severe pre-eclampsia. The other possibility is the effect of increasing gestational age. Wang et al.30 determined vitamin E levels longitudinally throughout normal pregnancy and observed a progressive increase in vitamin E concentrations which was most prominent between 24 and 32 weeks. A subgroup analysis of baseline vitamin E levels in early (< 28 weeks) and late ( 28 weeks) groups showed a trend towards higher levels in the latter group (P= 0.06).
Table 5. Biochemical data, lipid peroxides (MDA-TBAR) and vitamin E levels at enrolment, before delivery and postpartum day 3 (the differences in sample sizes are due to missing data. Urine protein and creatinine clearance measured in 24 hour collection and were not repeated after delivery)
Vitamin E (umol/L)
White cell count (109/L)
Platelet count (109/L)
Serum uric acid (mmol/L)
Creatinine clearance (mL/min)
Urine protein (g/day)
Vitamin E (umol/L)
White cell count (109/L)
Platelet count (109/L)
Serum uric acid (mmol/L)
Creatinine clearance (mL/min)
Urine protein (g/day)
Postpartum day 3
Vitamin E (umol/L)
White cell count (109/L)
Platelet count (109/L)
Serum uric acid (mmol/L)
Serum uric acid concentrations were suppressed within the first few days of treatment. However, this was not associated with any change in the condition of the mother or the fetus. Although creatinine clearance before delivery was higher in the antioxidant group, a suggestion of improvement in renal function would be highly speculative as neither the change within one week (baseline to day 7; data not shown) nor other renal function tests supported this finding. The trends in platelet and white cell count may be important. Leucocyte activation is a well documented feature of pre-eclampsia31 and vitamin E's modulatory effect on adhesion molecule expression32 could play a role in restricting leucocyte activation as well as its antioxidant effects.
Although the results are negative with regard to the primary study objective, there was a tendency towards prolongation of pregnancy in the antioxidant group (Fig. 1). The trial was not designed, in terms of sample size, to detect differences in perinatal outcome. There may be several reasons for the lack of significant results:
1The trial was too small. If the true effect of antioxidants was to decrease the number of women who delivered after 14 days by 25% the number of women necessary to show this difference would be 182 (α= 0.05; 1 - β= 0.80).
2The treatment was initiated when these women were diagnosed as having severe pre-eclampsia. As the pre-eclamptic process begins around the time of tro-phoblast invasion of decidual vessels it may be argued that the treatment was initiated too late to have a significant effect.
3There is no known ideal antioxidant dose or combination, although the doses used in this trial were similar to those used in other reports of antioxidant interventions and probably adequate for an antioxidant effect.
4It may be argued that women with severe early onset pre-eclampsia are heterogeneous and are not truly representative of ‘pure’ pre-eclampsia. However, we believe that once the pathophysiological process of pre-eclampsia is initiated the factors causing the disorder and its self-propagatory course are likely to be similar whether or not pre-eclampsia is superimposed on renal or chronic hypertension or there is any other underlying factor.
To our knowledge, this is the first randomised controlled trial of antioxidants in pre-eclampsia. There is considerable circumstantial epidemiological, laboratory and clinical evidence to suggest a significant role for free radical activation, and deficient or over-utilised antioxidant defences in the pathogenesis of pre-eclampsia. The results of this explanatory randomised trial do not encourage the routine use of antioxidants against pre-eclampsia. However, further explanatory randomised trials with modified strategies such as earlier initiation of therapy or with different combinations and more sensitive laboratory assays are worthwhile.
The authors would like to thank Her-Sol Laboratories and South African Druggists for supplying the trial medications, Professor C. J. Van Gelderen, Drs J. Paiker, R. Kulier and A. da Ponte, Srs T. Nkonyane, C. Nikodem and M. de Jager and to JG Strijdom Hospital Antenatal Ward staff for assistance.