Epidemiological assessment of misoprostol teratogenicity

Authors


Correspondence: Professor I. M. Orioli, Departamento de Genética, Universidade Federal do Rio de Janeiro, Caixa Postal 68011, 21944-970 Rio de Janeiro, Brazil.

Abstract

Objective To verify if any of the 15 congenital defects already reported in association with misoprostol can be found within an epidemiological registry of congenital defects.

Design Case-control study including case-sick and case-health controls.

Methods Comparison of misoprostol exposure for each specific defect, using the exposure for the rest of defects as a reference group.

Population Four thousand six hundred seventy-three consecutive newborn infants with malformations of unknown aetiology, in the Latin American Collaborative Study of Congenital Malformation.

Results There was no difference in exposure rate between the malformed (34/4673) and nonmalformed (23/4980) newborns. Four of the five more frequently cited defects in the literature were found to be in excess: constriction ring, terminal transverse-limb defects, hydrocephalus, and arthrogryposis. Equinovarus feet had a normal frequency in our study. Thirteen different defects not described in the literature were seen in our misoprostol exposed cases, but only holoprosencephaly and bladder exstrophy significantly exceeded the expected number.

Conclusions The confirmation from an epidemiological registry of an association for four of the five more commonly observed congenital defects among misoprostol exposed children described in the literature seems indicative of a real teratogenic effect. The defects are of vascular disruption type. However, additional attempts to achieve abortion could not be excluded as a concurrent contribution.

INTRODUCTION

Misoprostol, a synthetic prostaglandin E1 analogue, has been illegally used since, at least, 1988 to interrupt unwanted pregnancies in Brazil and Venezuela1,2. The report by Gonzalez et al.3 increased to 53 the number of malformed children attributable to prenatal misoprostol use. Considering the previous five case reports in the literature, there are now 15 specific congenital defects implicated from misoprostol exposure1,3–7.

Collins and Mahoney4 published the first case of congenital malformations, (hydrocephalus, finger reduction defect, and equinovarus feet) being attributable to prostaglandin 15-methyl F2α use at seven weeks after conception. Another fetus exposed to prostaglandin E2 and oxytocin at 15 weeks also had hydrocephalus and growth retardation5. A fronto-temporal defect with an asymmetric, well-circumscribed deficiency of the cranium and overlapping scalp was associated with prenatal exposure to misoprostol in a dubious number of cases by a single group of observers6,8,9. Gonzalez et al.1 presented seven cases with prenatal exposure to misoprostol, six of them with limb deficiency with or without Möbius sequence, and one with only facial nerve palsy, proposing a vascular disruption mechanism. Four of their cases with limb deficiencies had also other defects. Genest et al.7 described an abortus at 17 weeks after first trimester misoprostol exposure with limb defects, omphalocele, and findings diagnostic of early amniotic rupture.

Forty-two infants were described by Gonzalez et al.3, 17 of them with equinovarus feet and cranial nerve defects. Ten children had equinovarus as part of more extensive arthrogryposis. The most distinctive phenotypes were arthrogryposis confined to the legs and terminal transverse-limb defects with or without Möbius sequence.

Estimating that millions of pregnancies yearly are exposed to misoprostol, at least in Brazil, Brent10 recommended caution with case report publications, since almost every defect could be expected to occur in a misoprostol exposed pregnancy. In a prospective study of 17 cases exposed to misoprostol no defects were identified11. Extension of this cohort to 86 pregnancies still did not detect a teratogenic effect12. A case—control analysis with malformed and not malformed newborns infants suggested an association between misoprostol exposure early in pregnancy and vascular disruption defects2. A case—control study of 94 Möbius sequence compared with 120 neural tube defects showed a significant odds ratio if exposed pre-natally to misoprostol13.

Misoprostol is also legally used in ‘chemical’ abortion in many countries and, of course, as a drug for ulcer and related diseases. Thus, accidental prenatal exposures can occur, as well as the already cited misuse in Brazil and other South American countries, causing concern about the teratogenic status of misoprostol.

Since prohibition of misoprostol sales in Brazil under-registration of the drug has increased in recent years (1996–1998), and epidemiological data is not going to increase. We need to do the best with available data to investigate the potential hazards of this drug. We attempted to verify if any of the congenital defects reportedly associated with misoprostol exposure in the literature case reports was also associated with misoprostol exposure within a larger sample of an epidemiological registry of congenital defects.

METHODS

The Latin American Collaborative Study of Congenital Malformations is an ongoing study on congenital malformations in action since 1967. The aim of this study is to examine all births in collaborating hospitals across ten Latin-American countries in South America, recording detailed descriptions of defects and information about 50 risk factors14. The same information on each malformed baby is gathered on a control baby, defined as the next nonmalformed baby, of the same sex, that is born in the same hospital. Exposures during the pregnancy, encompassing maternal diseases, drug use and immunisation, were asked directly to the mother by a qualified paediatrician in the puerperium.

Misoprostol was available over-the-counter in all the Brazilian cities in the study period. Since its main use was to try to interrupt the pregnancy, an illegal action, the actual intake of misoprostol is expected to have been unreported. Furthermore, the birth of a malformed baby could possibly elicit more report of misoprostol use than the birth of a nonmalformed child. So, a case-sick control analysis approach was preferred. We calculated the odds ratios of being exposed to misoprostol for each specific defect (hydrocephalus, for example) as compared with the group of all other defects (all defects minus hydrocephalus) using Epi Info Software 6.015; the significance of the differences in misoprostol exposure for each defect was established by two-tailed Fisher exact tests with a critical level of 0.05. The specific defects tested were those congenital defects associated with misoprostol exposure in the literature (Table 1), as well as those that appeared only in our exposed sample. There is no overlapping between the published case series (Table 1) and the material reported in this present paper.

Table 1.  Congenital defects described previously with misoprostol prenatal exposure. Values are given as n/n total.
DefectsCollins & Mahoney 19834Wood et al. 19875Fonseca et al. 19916Gonzalez et al. 19931Genest et al. 19947Gonzalez et al. 19983Total of 53 patients
Terminal transverse limb defect1/10/10/16/71/15/4213/53
Hydrocephalus1/11/11/12/70/19/4214/53
Equinovarus feet1/10/10/11/70/122/4224/53
Intrauterine growth retardation0/11/10/10/70/101421/53
Scalp and skull defect0/10/11/10/70/101421/53
Cranial nerve palsy0/10/10/14170/121/4225/53
Limb constriction ring0/10/10/11/71/15/425/53
Dextrocardia0/10/10/11/70/101421/53
Hypospadias0/10/10/11/70/11/422153
Cleft palate/Bifid uvula0/10/10/11/70/11/422/53
Nail hypoplasia0/10/10/12/70/14/422/53
Microcephaly0/10/10/11/70/15/424/53
Syndactyly0/10/10/11/70/181427/53
Omphalocele0/10/10/10/71/11/422/53
Arthrogryposis0/10/10/10/70/110/4210/53
Braquidactyly0/10/10/10/70/13/423/53
Agenesis of abdominal muscles0/10/10/10/70/11/421/53

We investigated the role of maternal age, vaginal bleeding in the first trimester of pregnancy, as well as misoprostol dose, and timing of use (first, second, or third month of pregnancy) as confounding variables of the association of misoprostol exposure and congenital defects. These variables were considered in a logistic regression analysis where the dependent variable was to be malformed or nonmalformed newborn infant, all of them exposed to misoprostol. Maternal age was chosen as the only demographic confounder because it is associated to the occurrence of several congenital anomalies, and it covariates with many other characteristics as parity, socio-economical level, ethnicity, and reproductive behaviour.

Since the first registered misoprostol use in the Latin American Collaborative Study of Congenital Malformation was in 1989, and since 60 of 63 recorded exposures were born in Brazil, the present analysis was restricted to this country, and to the 1989–1995 period. From the original 4980 malformed newborn infants observed in this period we subtracted 307 with known genetic or ambiental syndromes, allowing 4673 malformed and 4980 controls.

RESULTS

There were 57 newborn infants exposed to misoprostol among 9653 records (4673 cases and 4980 controls). The recorded frequency of misoprostol use was 6 per 1000 records (95% CI4.4–7.7), which was homogenous among ten of the twelve reporting hospitals (χ2= 5.85; d.f. = 9; P > 0.05). One hospital, presented a higher frequency of 36.1 per 1000 (95% CI 14.6–72.2), and one hospital presented no cases among 780 records.

There was no difference (χ2= 2–47; d.f. = 1; P > 0–05) in the frequency of misoprostol exposure between malformed (34/4673) and nonmalformed (23/4980) babies. Only two of the 34 malformed cases presented multiple anomalies, probably belonging to the oromandibularlimb hypogenesis spectrum plus hydrocephalus/holoprosencephaly. They came from the same hospital and in one of them the information of misoprostol exposure was not obtained in the postpartum maternal interview, it being reported to the registry later. All the others are isolated cases or sequences: holoprosencephaly (1), hydrocephalus (1), cephalocele (1), constriction ring/skin scar (3), one of them with absence of phalanges, gastroschisis (1), bladder exstrophy (1), scalp defect (1), inferior limbs arthrogryposis (2), hypospadias (2), hydronephrosis (2), cleft lip (2), postaxial polydactyly (2), oropharingeal teratoma, and 12 cases with minor defects.

Table 2 presents the probability of each one of the 15 specific congenital anomalies reported as case reports to be present among the misoprostol exposed cases as compared with the not exposed cases. Constriction ring, terminal transverse-limb defects, hydrocephalus, and arthrogryposis were found to be in excess. Equinovarus feet, the defect occurring in 24/53 literature cases (not as part of arthrogryposis), had a normal frequency in our exposed cases. Thirteen defects not described in the literature (Table 2) were seen in our misoprostol exposed cases, but only holoprosencephaly and bladder exstrophy exceeded the expected number.

Table 2.  Different types of congenital anomalies among children born to mothers exposed to misoprotol. Values are given ass n(%), unless otherwise indicated.
 Misoprostol exposed (n= 34)Misoprostol unexposed (n= 4639)P*OR (95% CI)
  1. *.Fisher's exact test, two-tailed probability value.

  2. Decreasing order of the literature frequency.

Defects described earlier in case reports    
Equinovarus feet2 (5.88)268 (5.77)0.591.02 (0,24428)
Arthrogryposis2 (5.88)34 (0.73)0.038.47 (1.95–36.74)
Hydrocephalus4 (11.76)141 (3.04)0.024.23 (1.47–12.15)
Terminal transverse limb reduction3 (8.82)31 (0.67)0.00312.04 (3.52–41.12)
Limb constriction ring or skin scars3 (8.82)11 (0.24)0.000140.72 (10.83–153.12)
Syndactyly1 (2.94)73 (1.57)0.421.90 (0.26–14.05)
Microcephaly0 (0.0)29 (0.62)  
Braquidactyly0 (0.0)10 (0.21)  
Hypospadias2 (5.88)174 (3.75)0.371.60 (0.38–6.75)
Cleft palataifid uvula2 (5.88)50 (1.08)0.055.74 (1.34–24.59)
Nail hypoplasia1 (2.94)12 (0.26)0.0911.68 (1.48–92.46)
Omphalocele0 (0.0)38 (0.82)  
Scalp and skull defect1 (2.94)8 (0.17)0.0617.54 (2.13–144.24)
Dextrocardia0 (0.0)9 (0.19)  
Agenesis of abdominal muscles0 (0.0)9 (0.19)  
Defects not described earlier in case reports    
Extra nipple4 (11.76)202 (4.35)0.062.93 (1.02–8.39)
Cleft lip * palate3 (8.82)131 (2.82)0.073.33 (1.0.1–11. 03)
Post-axial polydactyly2 (5.88)316 (6.81)0.590.86 (0.20–3.58)
Holoprosencephaly2 (5.88)16 (0.34)0.00718.06 (3.99–81.79)
Hydronephrosis2 (5.88)77 (1.66)0.113.70 (0.87–15.73)
Pre-auricular tag2 (5.88)363 (7.82)0.500.74 (0.18–3.08)
Pigmented nevus2 (5.88)651 (14.03)0.130.38 (0.09–1.60)
Interatrial septal defect1 (2.94)38 (0.82)0.253.67 (0.49–27.52)
Ear defect1 (2.94)35 (0.75)0.233.99 (0.53–29.96)
Cephalocele1 (2.94)26 (0.56)0.185.38 (0.71–40.79)
Bladder exstrophy1 (2.94)3 (0.06)0.0346.83 (4.75–461.95)
Gastroschisis1 (2.94)22 (0.47)0.156.36 (6.8348.57)
Malpositioned toes1 (2.94)50 (1.08)0.3I2.78 (0.37–20.73)

Terminal transverse-limb defects were part of an oromandibular-limb hypogenesis spectrum in two multiple malformed cases, and they were combined with constriction ring in one case. Hydrocephalus also participated of the two oromandibular-limb hypogenesis spectra, besides being present in one holoprosencephaly case. The other defects in excess had no associated defects.

Only 22/34 malformed and 7/23 nonmalformed cases had information about used dose of misoprostol. In all but two controls and one case the drug was used only in the first trimester of pregnancy. Vaginal bleeding that could also be related to misoprostol use occurred in approximately one third of exposed cases from both groups. The misoprostol administration route was poorly specified in case and controls. The above cited variables and maternal age had no significant odds ratio in the logistic regression analysis and seemed not to be confounding factors between misoprostol exposure and congenital defects (data not shown). Nevertheless, the use of misoprostol in the second month of pregnancy could be related to the malformed status (OR 25.04 [0.98–636.54]).

DISCUSSION

No excess of malformed infants was found when compared with controls. This finding suggests the lack of a relevant case/control ascertainment bias for misoprostol in this material.

The epidemiological confirmation of the association of four of the five more observed congenital defects among misoprostol exposed children1,3–7 seems indicative of a real teratogenic effect. As a matter of fact, cranial nerve palsies were the second most cited defect (25/53) in the literature, but this defect was not expected to be obvious at birth.

With the exception of constriction rings, arthrogryposis, and bladder exstrophy, the defects found in excess in our material were concentrated in two cases with the oromandibular-limb hypogenesis spectrum. Less severe examples of this spectrum, constriction ring, and inferior-limbs arthrogryposis are the common phenotypes seen in the literature1,3–7. Since our sample was observed at birth, while the literature cases were observed at different ages, more severe defects, as holoprosencephaly or bladder exstrophy may have been undetected previously due to their low survival rate. Nevertheless, since those two defects were not included in the working hypothesis based on published reports, care must be taken in the cause-effect interpretation of these associations. However, age of diagnosis cannot be the explanation for the lack of association in our material between misoprostol and pes equinovarus1,3–7. If talipes equino-varus cases caused by misoprostol were a small fraction of all cases with this common defect, such association could be overlooked due to sample size limitations.

The intense vasoconstriction and uterine contraction produced by prostaglandin were proposed to result in distal ischemia in the fetus, leading to the digital anomalies observed4. The mechanism to explain the hydrocephalus was also through vasoconstriction and hypoxia, with or without associated haemorrhage, leading to foramina stenosis at a stage of development where the Munro foramina were occupied by exuberant choroid plexi4. Almost all of the defects described in the published cases exposed to misoprostol1,3–7 (Table 1) could be included in a theoretical list of defects caused by vascular disruptions.

Anterior-posterior embryonic compression was also proposed to occur when misoprostol is used around the 45-day, explaining the preferential lesions of cranial nerve nuclei in Möbius cases16. Accordingly, our finding of a higher exposure rate to misoprostol in the second than in the first or third month of pregnancy of malformed, as compared with nonmalformed children, also pointed towards an increased susceptibility to defects at this gestational age.

Other causes than uterine cramps could occur associated with the prenatal use of misoprostol to explain the defects. The most important one could be the expected concurrent instrumental abortion attempts, expected after the failure of misoprostol as an abortificient. The effects of inducing abortion by instrumental means were reviewed by Hall17. Puncture of the amniotic sac18, or early chorionic villus sampling procedures19, also seem to produce vascular disruption defects, including Möbius anomalies, but ring constrictions were described in only one report20. One of our cases with Potter facies and eritematous aplasia cutis vertex, plus oligohydramnios due to chronic leakage of amniotic fluid, could be an example of amniotic rupture caused by the effect of misoprostol or other abortive instrument. The likelihood of additional attempts to induce abortion is based on the fact that all these cases reported here and elsewhere are actually abortion failures producing a newborn infant.

In conclusion, there is an association of prenatal use of misoprostol as abortifacient and congenital defects of vascular disruption type. Under-reporting of misoprostol use due the abortion prohibition law, and concurrent abortion attempts by other chemicals or instrumental means are the main confounders of this association. Since misoprostol sales are now strictly controlled in several Brazilian cities, the used sample is hardly going to be large enough to clear these dubious aspects. A better material would be that of prenatal accidental exposures of misoprostol used for treatment of gastro-duodenal ulcer. However, the problem here could be that of gathering a sufficient number of exposed pregnancies within a reasonable time period.

Acknowledgements

This work has been partially supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Fundação Universitária José Bonifácio (FUJB), Fundação de Apoio à Pesquisa do Estado do Rio de Janeiro (FAPERJ), Projeto de Apoio à Pesquisa Estratégica em Saúde (PAPES), Brazil; and Consejo Nacional de Investigaciones Cientifica y Técnica (CONICET), Argentina.

Ancillary