Abstract
Objective:
In Hungary, vitamin E is frequently used to prevent repeated or threatened abortion. A previous study showed a higher risk of cardiovascular malformations in the children of pregnant women who had a high vitamin E intake either in their diet or by taking supplements. The objective was to examine this association.
Methods:
The Hungarian Case–Control Surveillance System of Congenital Abnormalities, 1980–1996, is a large, population-based dataset including 22,843 cases with congenital abnormalities, 38,151 healthy controls matched to the cases, and 834 patient controls with Down syndrome. Vitamin E treatment was compared in the mothers of these children.
Results:
The mothers of 1418 cases with congenital abnormalities (6.2%), 2267 controls (6.0%) and 43 patient controls (5.2%) had vitamin E treatment during pregnancy. A preliminary comparison of cases and controls showed a higher risk for four congenital abnormality groups, including cardiovascular malformations. However, if only prospectively and medically recorded vitamin E treatments in the prenatal maternity logbook were evaluated during the critical period of different congenital abnormalities, the higher risk for these congenital abnormalities was not found.
Conclusions:
The results of this study were based on relatively high-dose vitamin E intake in pregnant women and were not able to confirm the previously reported teratogenic effect of vitamin E.
Keywords
Introduction
Tocopherols or tocotrienols comprise eight essential naturally occurring fat-soluble nutrients, and among these isomers, alpha-tocopherol is the highest biologically active form [Lubin and Machlin, 1982; Machlin, 1980]. Alpha-tocopherol is more commonly known as vitamin E and this vitamin is essential for human health. Vitamin E is frequently used in pregnant women [Gagné et al. 2009; Rumbold and Crowther, 2005], mainly with vitamin C for the prevention of pre-eclampsia. However, recent studies have not confirmed the preventive effect of these vitamins for pre-eclampsia [Xu et al. 2010; Villar et al. 2009]. Moreover, in a Dutch case–control study, Smedts and colleagues [Smedts et al. 2009] showed a ninefold increased risk of cardiovascular malformations in children born to women who had taken vitamin E doses above 14.9 mg/day as periconception supplements or dietary intake.
In Hungary, pregnant women with sterility, repeated abortion, threatened abortion and preterm delivery are frequently treated with vitamin E. Therefore the objective of this study was to check the effect of vitamin E treatment (VET) in pregnant women on different structural birth defects, that is, congenital abnormalities (CAs), including cardiovascular CAs, using a large population-based dataset, the Hungarian Case–Control Surveillance of Congenital Abnormalities (HCCSCA) [Czeizel et al. 2001].
Materials and methods
The HCCSCA is based on the comparison of CA cases against matched controls (i.e. healthy newborns).
Cases and controls
CA cases were selected from the dataset of the Hungarian Congenital Abnormality Registry (HCAR) [Czeizel, 1997] for the HCCSCA. In Hungary it is mandatory for physicians to report CA cases to the HCAR from birth until the end of the first postnatal year. Most CA cases are reported by obstetricians and paediatricians. Practically all deliveries take place in inpatient obstetric clinics and the birth attendants are obstetricians. All infants with CAs are treated in the neonatal units of inpatient obstetric clinics, or in various general and specialist (e.g. cardiological) inpatient and outpatient paediatric clinics. During the study period (1980–1996), autopsy was mandatory for all infant deaths and was common (80%) in stillborn fetuses. Pathologists sent a copy of the autopsy report to the HCAR if defects were identified. Since 1984, fetal defects identified in prenatal diagnostic centres after elective or nonelective termination of pregnancy have also been included in the HCAR.
Two main categories of CA cases were differentiated: isolated (only one organ is affected) and multiple (two or more CAs in the same infant affecting at least two different organ systems) CAs.
The recorded total (birth + fetal) prevalence of cases with CA diagnosed from the second trimester of pregnancy to the age of 1 year was 35.0 per 1000 informative offspring (live-born infants, stillborn fetuses and electively terminated malformed fetuses) in the HCAR, 1980–1996 [Czeizel, 1997], but about 90% of major CAs were recorded during the 17 years of the study period [Czeizel et al. 1993].
There were three exclusion criteria for the HCCSCA dataset: cases with CA reported after 3 months of birth or pregnancy termination (77% of cases were reported during the first 3 months, and in most cases the remainder had mild CA); three mild CAs (congenital dysplasia of hip, inguinal hernia, large haemangioma); and CA syndromes caused by major gene mutations or chromosomal aberrations with preconceptional origin.
Infants with Down syndrome were used as patient controls.
Controls were identified from the National Birth Registry of the Central Statistical Office for the HCCSCA. Controls were defined as newborn infants without CAs. In general, two controls were matched to every case according to sex, birth week in the year, and district of parents' residence.
Collection and evaluation of exposures and confounders
Prospective medically recorded data
A letter was sent to mothers immediately after the selection of cases and controls to explain the purpose of the HCCSCA. Mothers were asked to send us their prenatal maternity logbook. At the time of the study, prenatal care was mandatory for pregnant women in Hungary (in order for them to receive a maternity grant and leave), thus nearly 100% of pregnant women visited prenatal care clinics an average of seven times during pregnancy. The first visit was between the sixth and 12th gestational week. Obstetricians recorded all pregnancy complications, maternal diseases and related drug prescriptions in the prenatal maternity logbook. Mothers were also asked to send us all other medical records, particularly if their infant had a CA. These medical documents were sent back after 3 weeks.
Retrospective self-reported maternal information
A structured questionnaire and an informed consent form were also sent to the mothers. The questionnaire requested information on pregnancy complications and maternal diseases, on medicinal products (e.g. vitamin E) taken during pregnancy according to gestational months, and on family history of CAs. To standardize the answers, mothers were asked to read an enclosed list of medicinal products and diseases as a memory aid before they filled in the questionnaire. Mothers were also asked to sign an informed consent form so that their name and address could be recorded in the HCCSCA.
The time (mean ± standard deviation) between the birth or pregnancy termination and the return of the ‘information package’ (logbook, discharge summary, questionnaire and informed consent form) in our prepaid envelope was 3.5 ± 1.2, 5.2 ± 2.9 and 3.8 ± 2.0 months in the case, control and patient control groups respectively.
Supplementary data collection
Regional nurses were asked to visit all nonrespondent case and patient control mothers at home. The nurses helped mothers to fill in the questionnaire, evaluated the available medical records, obtained data about smoking and drinking habits by interviewing mothers and their close relatives, and a so-called ‘family consensus’ was recorded. Smoking was evaluated using the number of cigarettes smoked per day while drinking was assessed using three categories: abstinent or occasional drinkers (less than one drink per week); regular drinkers (from one drink per week to one drink daily); hard drinkers (more than one drink per day). Regional nurses visited only 200 nonrespondent and 600 respondent control mothers in two validation studies [Czeizel and Vargha, 2004; Czeizel et al. 2003] because the ethics committee considered this follow up to be disturbing to the parents of all healthy children. Regional nurses used the same method for control mothers and for nonrespondent case mothers.
Overall, the necessary information was available for 96.3% of cases (84.4% replied by mail, 11.9% responded during the nurse visit), 83.0% of controls (81.3% by mail, 1.7% from visit) and 96.0% of patient controls (86.0 by mail, 10.0% from visit) of patient controls. The informed consent form was signed by 98% of mothers; the names and addresses of the remainder were deleted.
The data collection procedure for the HCCSCA was changed in 1997. Since then, regional nurses visit and question all cases and controls. However these data had not been validated at the time of this analysis and so only data for the 17 years between 1980 and 1996 are evaluated here.
The evaluation of vitamin E treatment
In Hungary during the study period, vitamin E (Richter, Budapest, Hungary) was used as a tablet/capsule containing 100 mg of tocopherol aceticum for oral treatment and 30 mg of tocopherol aceticum in a 1 ml phial for parenteral treatment. The recommended oral dose of vitamin E in pregnant women with threatened abortion or preterm delivery is one to three tablets once or twice daily. Vitamin E is not available over the counter in Hungary.
Although it was planned to differentiate between oral and parenteral administration, preliminary evaluation of the data showed that parenteral VET did not occur in the HCCSCA. In this study, therefore, VET means oral intake of tablets. Two types of treatments were recorded: VET alone and VET plus other drugs.
The duration of VET is important from two aspects. First, how many days/weeks/months it was received. Second, as a special consideration in pregnant women, the time and duration of treatment according to gestational age. The gestational age was calculated from the first day of the last menstrual period. Three time intervals were considered. The first was the first month of gestation because it is before organogenesis. The first 2 weeks are before conception while the third and fourth weeks comprise the preimplantation and implantation period of zygotes and blastocysts, including omnipotent stem cells. Thus CAs cannot be induced by short-term environmental agents in the first month of gestation and it explains the ‘all-or-nothing effect’ rule, that is, total loss or normal further development [Czeizel, 2001]. The second time interval was the second and third months of gestation. This is the sensitive ‘critical period’ for most major CAs [Czeizel et al. 2008]. The third time interval was the fourth to ninth months of gestation, that is, pregnancy after the organ-forming period.
There were three sources of information about VET: only from the prenatal maternity logbooks and/or other medical records; only from the questionnaire; and concordant data from both medical records and the questionnaire.
Among confounders, maternal age, birth and pregnancy order, marital and employment status (as an indicator of socioeconomic status) [Puhó et al. 2005], maternal diseases, related drug treatments and other pregnancy supplements were evaluated. The use of pregnancy supplements may indicate the level of pregnancy care, and the motivation of mothers to prepare and/or to achieve a healthy baby. In addition, it is necessary to consider folic acid and folic acid-containing multivitamins in the evaluation of preventable CAs [Czeizel, 2009].
Statistical analysis
The software package SAS version 8.02 (SAS Institute, Cary, North Carolina, USA) was used. First, frequency tables were developed for the main maternal variables in order to describe the groups of mothers who had VET and those who did not have VET as a reference. Student’s t-test was used for quantitative testing, while the chi-square test was used for categorical variables. Second, the prevalence of acute and chronic maternal diseases, related drug treatments and other pregnancy supplements used during the study pregnancy were compared between case and control mothers who had VET, and crude odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Third, the prevalence of VET was evaluated in different CA groups during any time of pregnancy and in the second and/or third gestational months. This prevalence was compared with the frequency of VET in matched controls, and adjusted ORs with 95% CIs were evaluated in a conditional logistic regression model. The ORs were adjusted for maternal age (<20 years versus 20–29 years versus 30+ years), birth order (first delivery versus one or more previous deliveries), maternal employment status (professional–managerial–skilled worker versus semiskilled worker–unskilled worker–housewife versus others), other frequently used drugs (yes/no) and folic acid/multivitamin supplementation in the first trimester (as a dichotomous variable).
Results
The case group consisted of 22,843 newborns or fetuses (‘informative offspring’) with CAs, of whom 1418 (6.21%) had mothers who received VET. The total number of births in Hungary was 2,146,574 during the study period between 1980 and 1996. Thus the 38,151 controls without CAs represented 1.8% of all Hungarian births. Among the controls, 2287 (5.99%) were born to mothers who had VET. The patient control group comprised 834 newborn infants or fetuses with Down syndrome and 43 (5.16%) had mothers who received VET. Only two case mothers, 12 control mothers and no patient control mothers had VET alone, therefore VET and other drugs were evaluated together.
VET was medically recorded in the prenatal maternity logbook for 631 of 1418 case mothers (44.5%), 1438 of 2287 control mothers (63.0%) and 23 of 43 patient control mothers (53.5%). Most medically recorded VET was prescribed by obstetricians in prenatal care clinics for pregnant women with threatened abortion and/or preterm delivery.
The dose of VET was not documented for about one-third of pregnant women; these women belonged to the group for whom only maternal information was available. In about half of the pregnant women, the daily dose was 600 mg (two tablets three times daily); the remainder used two tablets once or twice daily. However, there was a wide variation in the number of days per week when VET was received, but these estimates were possible only in half of the women because of available data. VET was recorded each day, every second day and less frequently per week in about 15%, 60%, and 35% of pregnant women respectively. There was no significant difference in the dosages of VET among the study groups. It is important to stress that the dosages of VET were higher than recommended; this is because vitamin E was not used as a supplement but as a treatment for pregnancy complications and infertility.
Onset and duration of vitamin E treatment according to gestational month in the study groups.
SD, standard deviation.
Characteristics of case and control mothers receiving or not receiving vitamin E treatment, the latter as a reference.
SD, standard deviation; VET, vitamin E treatment.
Among pregnancy supplements, folic acid use was more frequent in mothers who had VET, mainly from the second and third gestational month. However, increased use of all other pregnancy supplements (including vitamin E containing multivitamins) was characteristic for pregnant women receiving VET. In particular, the very high use of vitamin D is worth noting.
Smoking and drinking status were evaluated in women visited and questioned at home. The rate of smokers was lower in pregnant women who had VET compared with pregnant women who did not have VET (case mothers: 14.7% versus 22.1%; control mothers: 14.6% versus 19.2%). The proportion of regular or hard drinkers was low (about 1.0%) in the case and control mothers.
There was no difference in the incidence of acute maternal diseases and the prevalence of chronic maternal diseases among pregnant women who had/did not have VET.
The use of drugs for the treatment or prevention of threatened abortion (allylestrenol, diazepam, drotaverine, promethazine) and threatened preterm delivery (magnesium salts, terbutaline, aminophylline) did not differ among case, control and patient control mothers who had VET.
Estimation of risk for different congenital abnormalities by comparing vitamin E treatment as exposure in the mothers of cases and matched controls during any time of pregnancy and in the second and/or third gestational month using a conditional logistic regression model. Bold numbers show significant associations.
CA, congenital abnormality; CI, confidence interval; OR, odds ratio, adjusted for maternal age and employment status, birth order, other drugs and use of folic acid.
The HCCSCA data were not appropriate for the differential diagnosis of different subtypes of oesophageal and rectal/anal atresia/stenosis, but the subgroups of cardiovascular CAs and multiple CAs were evaluated.
The number (and observed/expected rates) of isolated cardiovascular CAs in 306 cases (all were live-born babies) were: ventricular septal defects: 125 (40.8%/34.9%); atrial septal defects type II: 27 (8.8%/10.3%); tetralogy of Fallot: 18 (5.9%/1.9%); transposition of great vessels: 15 (4.9%/3.4%); coarctation of aorta 11: (3.6%/2.6%); patent ductus arteriosus: 10 (3.3%/3.9%); however, 50 cases had unspecified cardiovascular CA (16.3%/19.9%), while the rest included other cardiovascular CA groups with less than 10 cases. There was no significant cluster of any specific cardiovascular CAs.
The distribution of component CAs within 97 multimalformed cases did not show any characteristic pattern.
In the next step, VET was analyzed only during the second and/or third gestational months, the critical period of most major CAs including oesophageal and rectal/anal atresia/stenosis, cardiovascular CAs and multiple CAs. A significant association between maternal VET and higher risk of oesophageal atresia/stenosis, cardiovascular CAs and multiple CAs was not found. There was only a higher risk for rectal/anal atresia/stenosis.
We repeated this analysis based only on medically recorded VET and the adjusted OR did not indicate any association of VET with higher risk of total CA or any specified CA group. Of 101 mothers who had infants with cardiovascular CA, 70 had appropriate medical records to evaluate the daily doses and the number of days of VET per week. Thus, it was possible to identify 38 pregnant women with an estimated daily mean dose of VET of 322.2 mg, and 32 pregnant women with an estimated daily mean dose of VET of 84.4 mg. If these two groups were compared, there was no higher risk for cardiovascular CA in the group receiving the higher doses than in the group receiving the lower doses (adjusted OR 1.2; 95% CI 0.5 to 1.8). The duration of treatment was similar in these two groups.
Finally, the frequency of VET was compared between case mothers with different CAs and patient control mothers, and a higher risk was not found in any CA group, including cardiovascular CAs. (These data are not shown here because of the limited number (43) of patient control mothers with VET, particularly with medically recorded VET (23).)
Discussion
The birth of an infant with a CA is a traumatic event for mothers, who will try to find a causal explanation, such as illness or drug use during pregnancy. Of course, this does not happen after the birth of a healthy newborn infant. The objective of this study was to evaluate the association between VET in pregnant women and CAs in their infants. Our study did not show a higher risk for any CA, including cardiovascular CAs.
There were obvious differences between pregnant women who received VET compared with those who did not receive VET. On the one hand, threatened abortion and/or preterm delivery, and infertility (i.e. the indication of VET and related drug treatments) occurred more frequently in pregnant women who had VET. On the other hand, pregnant women who had VET had a healthier lifestyle than pregnant women who did not have VET; for example, smoking was less frequent and the use of pregnancy supplements was higher among pregnant women who had VET. These confounding factors are important when comparing the study groups.
The placental transfer of vitamin E is by passive diffusion; its passage to the fetus is dependent on plasma lipid concentrations [Gagné et al. 2009; Haga et al. 1982; Martinez et al. 1981; Hill and Longo, 1980]. Vitamin E levels rise throughout pregnancy in women [Leonard et al. 1972], thus the concentrations in pregnant women at terms are approximately four to five times higher than those of newborns [Gagné et al. 2009; Leonard et al. 1972].
The periconception supplementation of vitamin E, in addition to a high dietary vitamin E intake (above 14.9 mg/day), was based on self-reported maternal information at 16 months after the index pregnancy in the study of Smedts and colleagues [Smedts et al. 2009]. The preliminary analysis of CAs in this study based partly on retrospective maternal information of VET resulted in a higher risk for some CAs, including cardiovascular CAs, in the newborns of pregnant women who had VET. Thus, our findings seemed to confirm those of Smedts and colleagues. However, a detailed analysis of our data did not confirm this association. When we evaluated only the medically recorded VET of pregnant women in the critical period for cardiovascular CA, a higher risk was not found. When we differentiated between a very high dose (about 322 mg/day) and a lower dose of VET, we did not find a dose–effect relation in the risk for cardiovascular CAs. However, the mean dose of VET was higher in our low-dose group than the high-dose vitamin E intake in the Smedt study.
In our opinion, the discrepancies between our results and those of Smedt and colleagues [Smedts et al. 2009] can be explained by recall bias. In the study of Smedts and colleagues, VET was based on retrospective maternal information. In our study, 55% of case mothers and 37% of control mothers only had retrospective maternal information, although there was no significant difference in the proportion of medically recorded VET between the two study groups. However, the higher proportion of maternal reports resulted in a relatively lower proportion of medically recorded VET in the group of case mothers. This recall bias might inflate an increased risk for CAs and may cause a spurious association between drugs and CAs, with a biased OR up to a factor of 1.9 [Rockenbauer et al. 2001].
Our study was designed to limit recall bias as follows:
We focused our analysis on the critical period for CAs because we expected an underreporting of VET in both the critical and noncritical periods of CAs in the control group. We evaluated separately only medically recorded VET as a gold standard. We use a patient control group with a similar recall bias as case mothers.
Thus, the comparison of prospectively and medically recorded VET in case mothers and control mothers during the critical period of CAs did not indicate any teratogenic effect of VET despite the very high doses received (the estimated mean dose was about 322 mg/day in a subgroup receiving higher doses). Although the National Academy of Sciences recommended dietary allowance (RDA) for vitamin E is 10 mg/day [American Hospital Formulary Service, 1997], our pregnant women had a special indication for VET.
Vitamin E deficiency was found to have teratogenic effects in animal experiments [Kalter, 1968]. The teratogenic effect of high doses of VET, namely a higher rate of exencephaly, hydrocephalus and cleft palate, was found in the offspring of pregnant rats and mice [Steele et al. 1974; Momose et al. 1972; Cheng et al. 1960; Cheng and Thomas, 1953]. However, other animal investigations did not find teratogenicity with VET [Hurley et al. 1983; Krasavage and Terhaar, 1977; Hook et al. 1974; Sato, 1973], although Kappus and Diplock showed that doses used in these animal investigations were hundreds to thousands of times higher than the human RDA [Kappus and Diplock 1992]. Previously, a causal association of VET with a higher risk for CAs has not been published in human studies [Briggs et al. 2005; Shepard and Lemire, 2004; Friedman and Polifka, 1996], although vitamin E was frequently used in pregnant women in the past decades. Thus, the results of our study are in agreement with these experiences.
The strengths of HCCSCA are that it is a large population-based dataset of ethnically homogeneous (white) Hungarian people. It includes 3705 pregnant women who received VET. Additional strengths include the matching of cases to controls without CA and available data for potential confounders. The diagnosis of medically reported CAs was checked by the HCAR [Czeizel, 1997] and the HCCSCA later modified, if necessary, on the basis of recent medical examinations [Czeizel et al. 2001]. Thus, of 360 cases with cardiovascular CAs, 310 (86.1%) were specified according to different types based on medical records.
However, this dataset also has limitations. For example, most of the VET data were based on self-reported maternal information, although medically recorded prospective data could be evaluated separately. In general, multivitamins used by pregnant women include vitamin E, but at relatively low doses, and less than 10% were supplemented with these products. We had no data on dietary vitamin E intake but such complementary low doses could not significantly modify the very large doses of VET considered in our study.
In conclusion, high doses of VET in pregnant women were not associated with a higher risk of CAs, including cardiovascular CAs, in their infants.
Footnotes
Acknowledgements
The authors thank Erzsébet H. Puho, PhD, for her help in the statistical analysis of data and Mrs Judit Erdei for her word processing and administrative assistance.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The available data of the HCCSCA were evaluated.
Conflict of interest statement
The authors declare no conflict of interest in preparing this manuscript.