Risk of complications during pregnancy after Senning or Mustard (atrial) repair of complete transposition of the great arteries

Abstract

Aims To investigate magnitude and determinants of risks during pregnancy in women with Mustard or Senning repair for complete transposition of the great arteries (TGA).

Methods and results Using a nationwide registry (CONCOR), 70 women with Senning (23%) or Mustard (77%) repair for TGA were enrolled. A total of 28 patients had 69 pregnancies (two twins), including 17 spontaneous miscarriages and three elective abortions. During 39 of the 49 completed pregnancies, complications were observed. The most important cardiac complication was clinically significant arrhythmia (n=11, 22%), especially occurring in patients with a prior history of arrhythmia. Important general pregnancy complications were preeclampsia (n=5, 10.2%) and pregnancy-induced hypertension (n=4, 8.2%). Obstetric complications included premature rupture of membranes (n=7, 14.3%), premature labour (n=12, 24.4%), premature delivery (n=16, 31.4%), and thrombo-embolic complications (n=2, 4.1%). Mean (singleton) pregnancy duration was 36±5 weeks. Eleven of the 51 children (21.6%) were small for gestational age. Foetal and neonatal mortality combined was 11.8% (n=6). No recurrence of congenital heart disease in the offspring was documented.

Conclusion In this largest report on pregnancy in women with atrial-corrected TGA to date, a high incidence of obstetric complications and mortality in the offspring was observed.

Introduction

Physiological repair of complete transposition of the great arteries (TGA) started with the introduction of surgical procedures bearing the names of Senning (1959) and Mustard (1964). In these procedures, venous blood flow is redirected within the atrial compartment leaving the transposition itself untouched, which dramatically improved medium-term survival. Actuarial 30-year survival of early survivors of atrial repair, nowadays, equals almost 80%.^1,2 However, in addition to heart failure caused by the employment of the morphologically right ventricle as the systemic ventricle, rhythm disturbances are important long-term sequelae after Mustard/Senning procedures. Atrial correction was gradually abandoned after the introduction (late 1970s) of the arterial switch procedure hallmarked by the switch of the great arteries themselves, which avoids above-mentioned sequelae. Nevertheless, a substantial cohort of atrial-corrected women is now passing through the child bearing years of life.¹

Circulatory changes inherent to pregnancy, including the increased cardiac output and reduced systemic vascular resistance, potentially threaten the seemingly good health of both the women after Senning or Mustard repair and their offspring.^1–3 Current knowledge on the magnitude and determinants of complications in women with an atrial switch for complete TGA is limited.^4–6

The primary objective of this study was to identify the magnitude and determinants of pregnancy risks in women after Senning- or Mustard-corrected TGA and their offspring, including congenital heart disease recurrence risk. Secondary aims were to assess the prevalence of infertility and the reasons for being childless.

Patients and methods

Using the CONgenital CORvitia registry, a nationwide database that registers all CHD patients receiving care in all tertiary medical centres in the Netherlands which provide informed consent (consent rate 97%) funded by the Netherlands Heart Foundation, and the congenital thoracic surgery database registering all patients who received a Senning or Mustard repair in the participating tertiary Belgian Centre, we were able to identify, respectively, 70 and 6 female patients (aged 18–45 years) with atrial repair. A total of 70 (92%) patients provided written informed consent and were enrolled between January and November 2004. The main reason six patients did not want to cooperate was age-related; most (n=5) thought they were too young to address issues of pregnancy. The enrolled patients are all seen regularly on the outpatient clinics of the six participating tertiary medical centres. The institutional review board or ethics committee at each site approved the protocol.

Questionnaires were used only as a supplement for the available data from medical records. Baseline data included: prior surgical procedures, comorbidity, and medical history recorded using the European Paediatric Cardiac Coding; age at inclusion; medication; fertility (infertility: more than 2 years of pregnancy attempts and documented by gynaecologist); miscarriages (spontaneous foetal loss before 20 weeks of gestation), and/or elective abortions.

Detailed information concerning each completed (>20 weeks of gestation) pregnancy was recorded (when applicable pre-, peri-, and postpartum data): mode of delivery; parity status; use of cigarettes, alcohol and/or drugs; use of medication; New York Heart Association (NYHA) functional class; physical examination (including blood pressure and heart rate); 12-lead electrocardiogram, and/or 24-h electrocardiogram (Holter) registrations. Although transthoracic echocardiograms were available for all patients, merely those recorded <2 years prior to pregnancy and technically adequate to assess systemic (right) ventricular systolic function (qualitative wall motion scoring)/end diastolic diameter (apical four chamber view), intra-atrial baffle, and AV valvular function (qualitative assessment using colour Doppler imaging) were used.

Documented complications were grouped into cardiac, general, obstetric, and neonatal events. Cardiac complications: symptomatic documented arrhythmia or heart failure (according to attending cardiologist) requiring treatment, myocardial infarction, and/or endocarditis. General complications: pregnancy-induced hypertension (PIH, new onset hypertension after ≥20 weeks of gestation, >140 mmHg systolic, or 90 mmHg diastolic without proteinuria); preeclampsia (gestational hypertension with >0.3 g of proteinuria/24-h urine sample); eclampsia (preeclampsia with grand mal seizures); Haemolysis Elevated Liver enzymes Low Platelets (HELLP) syndrome; thrombo-embolic complications; gestational diabetes, and/or stroke. Obstetric complications: assisted (forceps/vacuum/caesarean) delivery; premature rupture of membranes (PROMs, membrane rupture before the onset of uterine contractions); prolongation of second stage of delivery (nullipara >2 h, multipara >1 h); premature labour (spontaneous onset of labour <37 weeks gestation); postpartum haemorrhage (vaginal delivery >500 ml, caesarean section (CS) >1000 ml, documented by gynaecologist and requiring transfusion). Neonatal complications: premature birth (delivery <37 weeks of gestation); small-for-gestational-age birth weight (<10th percentile);^7,8 foetal demise (intra-uterine death ≥20 weeks of gestation); perinatal death (within the first month after birth), and/or recurrence of congenital heart disease.

A Clintrial data entry program was used to record information and was converted to SPSS (version 11.0) for statistical analysis. Descriptive statistics for nominal data, depending on context per pregnancy or patient, were expressed in absolute numbers and percentages. After checking for normality, mean values and standard deviations were calculated for normally distributed continuous variables, and non-normal distributed continuous variables are presented as median with quartiles. Comparison of baseline characteristics (including history of arrhythmia, type of atrial repair, prior rashkind/blalock procedures, NYHA class, use of cigarettes/alcohol/cardiac medication, and parity) and above-mentioned cardiac, general pregnancy, obstetric, and neonatal complications were preformed using Chi-square test or Fisher's exact test. Outcome rates were analysed by logistic regression with random effects with Gaussian distribution (STATA, version 8.0), considering that patients represent clusters of pregnancies and the dependence of pregnancies with respect to outcome within the same patient. All test were two-tailed and a P-value <0.05 was considered as statistically significant.

Clinical outcome data were compared with known incidences of normal pregnancies in the western world and available literature.⁹

Results

Baseline characteristics of the enrolled 70 patients with Mustard and Senning repair for complete TGA are summarized in Table 1. Forty-two women (60%) were childless, of whom 35 (83%) wished to bear children in the future. The most important reasons for being childless were: age (n=21), assumed TGA-related increased risk of pregnancy (n=6), socio-economic reasons (n=6), currently trying to get pregnant (n=5), comorbidity (n=3), and negative pregnancy advice (n=1).

The remaining 28 patients had 69 pregnancies (between 1990 and 2004): 49 completed pregnancies (two twin gestations), 17 spontaneous miscarriages (miscarriage rate 24.6%, 10 patients), and three elective abortions (three patients). Elective abortions were performed for socio-economic reasons in two patients and development of systemic AV valvular regurgitation with systemic ventricular dilatation in the third patient. None of these patients remained childless.

Four patients were referred to a gynaecologist based on suspected sub-/infertility; however, spontaneous pregnancy occurred in three of them during follow-up. The fourth patient is trying intra-uterine insemination to achieve pregnancy.

Miscarriage pregnancies and those who are electively terminated are difficult to investigate, and therefore we focus on the completed pregnancies (Table 2). On the basis of predefined criteria, useful echocardiographic data were available in 34 (69%) of the 49 completed pregnancies. Qualitative wall motion scoring showed that the anatomic systemic ventricular function was reduced in 33% of these pregnancies. The mean systemic ventricle end-diastolic diameter (apical four chamber view) was 51 mm (range 41–58). Systemic AV valvular regurgitation was absent in 10%, minor in 70%, moderate in 20%, and severe in none of the pregnancies. No clinical significant baffle leakage or obstruction was documented.

Overall, in 80% of the completed pregnancies (in 26 of the 28 women) complications ranging from only mild (nine patients, e.g. vacuum or forceps) to complex and extremely serious (e.g. offspring mortality) were observed (Tables 3 and 4).

Cardiac complications (Tables 3 and 4)

Clinically significant arrhythmias occurred in 11 pregnancies (22.4%), in eight patients (28.6%). A medical history of arrhythmia (P=0.023) was associated with occurrence of arrhythmia during pregnancy. Previous arrhythmia was present in five women (Patients A–E): paroxysmal atrial fibrillation (PAF, n=4) and recurrent non-sustained ventricular tachycardia (n=1). PAF re-emerged during six gestations (Patients A–D) in spite of medication. Changes in medication prior to pregnancy may have contributed to the recurrence of arrhythmia in three patients (Patients B, C and E).

De novo development of clinically significant arrhythmia occurred in three women (Patients F–H), e.g. symptomatic ventricular tachycardia (VT, max. 153BPM, 8 beats) combined with supraventricular tachycardia (most probably atrial flutter), atrial flutter (mean 145BPM), and an exercise-dependent multifocal VT.

Heart failure occurred during two pregnancies in two patients: both episodes were successfully treated with diuretics. In addition, deterioration of NYHA class was found during 17 (34.7%) pregnancies (53.5% of patients). In four patients (8.2%), the observed NYHA class deterioration persisted (on average for more than 1 year) postpartum. No myocardial infarction, endocarditis, or cardiac death was documented. No relation of baseline (i.e. before pregnancy) NYHA class, systemic ventricular function, and presence/severity of systemic AV valvular regurgitation with any of the events could be established.

General pregnancy complications

Hypertensive disorders occurred during a total of nine (18.4%) pregnancies (in 28.6% of the patients), which included preeclampsia (n=5, 10.2%, including one case complicated by the development of HELLP syndrome) as well as isolated pregnancy-induced hypertension (n=4, 8.2%). Four patients reported infections requiring antibiotic therapy (all genito-urinary tract infections). Thrombo-embolic complications were diagnosed during two pregnancies: pulmonary embolism (Patient A) and a deep venous thrombosis in combination with symptoms of pulmonary embolism (Patient E). No eclampsia, pregnancy-induced diabetes, or stroke was reported.

Obstetric complications

Forceps and vacuum-assisted deliveries were performed in three (6.1%, three different patients) and seven (14.3%, seven different patients) pregnancies. Four primary Caesarean sections (CSs, 8.2%, four patients) were performed on maternal congenital cardiac indication. Breech presentation of both children in a twin pregnancy was the reason for the fifth primary CS. Two secondary CSs could also be attributed to a breech presentation. Epidural analgesia was only used during six deliveries, mainly to reduce maternal pain and only on one occasion to reduce sympathic stress to dampen haemodynamic shifts. Other important observed obstetric complications were PROMs (n=7, 14.3%, in 21.4% of patients), premature labour (n=12, 24.4%, in 28.6% of patients), prolongation of second stage of delivery (n=5, 10.2%, in 17.9% of patients), and postpartum haemorrhage (n=7, 14.3%, in 25% of patients).

Neonatal complications

Mean pregnancy duration for all singleton gestations were 36.4±4.5 weeks (primiparous 37.1±3.5 weeks and multiparous 35.4±5.6 weeks). Forty-seven of the 51 children (57% male) were born alive. Four children (7.8%, including one half of twin) died in utero after a pregnancy duration ranging from 23 to 39 weeks. Post-mortem examinations were performed in two children, the pathologist suggested that a maternal Parvo-B19 infection could have played a role in the death of one child, however no congenital cardiac or other defects were detected. Perinatal mortality was documented in two cases (3.9%, including the other half of the twin that died in utero) both subsequent to pre-/immaturity. Three of the six children that died (including the twin) were offspring of the same mother (Patient J). Her pregnancies were generally uneventful and her cardiac condition based on echography and exercise performance (NYHA class I) was reasonable. Sixteen children (31.3%) were born prematurely to 12 different mothers (42.9%) after on average 31.4±4.7 weeks of gestation. Eleven children (21.6%) of eight women (28.6%) were small for gestational age, including four prematurely born children. No recurrence of congenital heart disease was reported, prenatal ultrasound evaluations were performed in all pregnancies, and specific foetal echocardiograms were performed in 31 (63%) of all completed pregnancies.

Discussion

This study describes 69 pregnancies in 28 women with Senning or Mustard repair for TGA, which renders it the largest series thus far. Cardiac complications during pregnancy in these patients were frequent but manageable. However, a high incidence of obstetric complications (65%, multiple or serious obstetric complications were found in 49% of pregnancies, obstetric complication(s) occurred during one or more pregnancies in 85% of patients) and mortality in the offspring (11.7%, normally limited to less than 1%) were observed. Therefore, in contrast to what is generally assumed, pregnancy is not always well tolerated in TGA patients corrected with Mustard or Senning procedures. Although arterial switch is nowadays the mainstay surgical procedure, a substantial cohort of TGA patients with Mustard or Senning repair will enter their childbearing years. Roughly estimated, approximately 16/1 000 000 inhabitants are women with Senning or Mustard procedures who will reach or have reached childbearing age [USA±4500 patients (see the following table)].

With fertility apparently uncompromised and most (still) childless patients wishing to have children in the future, a substantial health problem is born.

No severe cardiac complications occurred in this study. In accordance with the available literature, the observed incidence of arrhythmias was approximately 20% (28.6% of patients) with a predilection for supraventricular tachycardias.^10–13 In addition to the atrial scar tissue formation after repair for TGA that acts as a substrate for re-entry circuits, the circulatory burden of pregnancy might trigger the development of arrhythmia. Change in pharmacological therapy prior or during pregnancy may also have contributed. We found that in patients with a history of arrhythmia the incidence was higher, although all episodes were accessible for pharmacological intervention.

Although new onset heart failure during pregnancy has been reported in atrial-corrected TGA patients,^6,12–17 the incidence was low in this study, as in only two cases diuretic therapy was needed. In agreement with Guedes et al.,¹⁷ NYHA class deterioration was found in approximately one-third of the gestations. However, it must be noted that also most healthy women experience a decline of exercise tolerance during pregnancy. Almost all patients recovered shortly after pregnancy, but persistence (>1-year postpartum) of the NYHA class deterioration was observed in three patients. The risk score introduced by Siu et al. applied to our population showed that during 28 pregnancies with zero predictors (estimated risk of 5%) three cardiac events (10.7%) occurred and in the remaining 21 pregnancies with one predictor (all prior cardiac events or arrhythmias, estimated risk of 27%) nine cardiac events (42.9%) were documented. In our population, the risk score developed by Siu underestimated the risk for cardiac complications.

The observed rate of spontaneous miscarriages (n=17, 24.6%) is higher than expected in the general population in the Netherlands (10–15%).⁹ Literature on miscarriages after Senning or Mustard repair for TGA is limited.^4,5,12 It needs to be taken into account, however, that analysis of miscarriage pregnancies is difficult. Future prospective research is necessary.

Hypertension-related disorders (particularly preeclampsia) were observed in 18.4% of the pregnancies (28.6% of patients), which is higher than the 8% reported in the general population.¹⁸ In particular, the high incidence of preeclampsia is noteworthy. A known risk factor for the development of preeclampsia is nulliparity, which was applicable to four of the five preeclampsia pregnancies. These women had neither a medical history of diseases known to be associated with preeclampsia (including diabetes, hypertension, renal failure, and thrombophilia) nor were they older than 35 years (range 22–29) at pregnancy. Review of previous publications reveals five cases of hypertension-related disorders in 23 pregnancies (21.7%).^4,10,12 In combination with our current data, the incidence of hypertension-related disorders, especially preeclampsia, appears to be elevated in women with atrial-corrected TGA. Several mechanisms, either solitary or combined, may be hypothesized. First, the reduced capacity of the systemic ventricle to increase cardiac output in atrial-corrected TGA patients possibly diminishes placental perfusion, which plays an essential role in preeclampsia.^19,20 Second, activation of neurohormonal pathways in patients with congenital heart disease may alter vascular remodelling associated with preeclampsia.²¹ Third, endothelial dysfunction and oxidative stress present in some congenital heart disease patients may interfere with the preeclampsia pathophysiology.²²

Thrombo-embolic complications are well-known complications of pregnancy; however, incidence is limited to 1 per 1000–2000 pregnancies.²³ The presence of arrhythmia and immobility due to deterioration of NYHA class may have contributed to the development of the two thrombo-embolic complications in our study. Thus far, only one stroke during pregnancy has been described in literature.⁶

PROMs complicated seven pregnancies (one twin, four patients, 14.3%). All ruptures occurred prior to 37 weeks of gestation, also known as preterm PROMs, normally seen in only 1.25% of gestations.²⁴ All four patients had a medical history of miscarriages, in two of them cervical incompetence was suspected to be the underlying cause. Maternal smoking may have contributed in one pregnancy. All deliveries were premature (range 23 (twin)–36 weeks) and both children of the twin pregnancy died. The only earlier article mentioning PROM describes two episodes during five pregnancies.¹⁰

Premature delivery occurred in 16 (31.4%) pregnancies (including two twins), which is high in the Western world, where the preterm birth rate equals 10–12%.^25,26 Preterm PROM was associated with premature delivery in five cases (35.7%); this is in agreement with the 40% documented in literature.²⁴ Moreover, preterm delivery is the leading cause of infant morbidity and mortality in developed countries and was associated with four of the six deaths in the offspring.²⁷ In three other reports, similar increased percentages of premature labour and delivery have been reported.^10,16,28

In our study, 11 children were small for gestational age (21.6%). Five of these children were born after more than 37 weeks of gestation and no health problems were documented in these children. In the premature born and small for gestational age group, three out of the six children died.

In concordance with earlier series, no recurrence of congenital heart disease in the offspring was found.^4,5 Not all miscarriages and deaths, however, were thoroughly examined. Furthermore, research by Digilio et al.²⁹ showed the existence of rare monogenetic or oligogenetic inheritance lines in certain kindreds.

Limitations

First, the retrospective design necessitated a review of patient's medical records consequently leading to missing values. Nevertheless, all mentioned complications had to be documented by medically qualified personnel in the records according to the pre-set definitions before data entry. Additional medical records were examined when cardiology records did not provide sufficient information. In particular, the obtained echographical data should be interpreted with caution, because of bias introduced by the ample missing data. In contrast, reporting echographical data that are over 2 years old would also be clinically incorrect, as changes over time are expected. Second, the availability of parameters of ventricular/valvular functioning was limited to the level of qualitative assumptions. Third, selection bias could be introduced by excluding miscarriage/abortion pregnancies. However, these pregnancies are difficult to investigate in a retrospective setting especially due to the lack of post-mortem examinations. In addition, the increased health awareness of CHD patients makes it difficult to draw the line between a miscarriage and a delayed menstruation, which might lead to an over-presentation of miscarriages. Furthermore, these pregnancies cannot develop complications associated with the second and third trimesters (e.g. preeclampsia, also cardiac complications are unlikely in this early stage of pregnancy with its limited volume load); therefore, including these pregnancies would underestimate these risks. Additional selection bias is introduced by the fact that some patients remain childless because of anticipated/expected risks during pregnancy, which might lead to an underestimation of risks. Together with the given sample size and the possible effects of multitesting, e.g. inflation of type I error, all conclusions of the present study must be drawn with caution.

Conclusion

In conclusion, in addition to frequent though manageable cardiac complications, a high incidence of (serious) obstetric complications (65%) and mortality in the offspring (11.7%, normally limited to 1%) were observed during pregnancy in atrial-corrected TGA patients. However, with the exception of the finding that patients with a history of arrhythmia are at greater risk for complications, we could not identify additional (preconceptional cardiac) risk markers. Therefore, we recommend that all TGA patients, even those of apparent good health, should be monitored closely during their pregnancy, because early detection and intervention is necessary to prevent complications in the future.

Acknowledgements

We would like to thank Dr. J.H. Hillege for his statistical input in the preparation of the manuscript. P.G.P. was supported by the Netherlands Heart Foundation Grant 2002 B125 and the Interuniversity Cardiology Institute the Netherlands (ICIN). Prof. Dr. van Veldhuisen is an established investigator for the Netherlands Heart Foundation (Grant D97.017).

Conflict of interest: none declared.

References