Pregnancy outcomes in patients with a fontan circulation and proposal for a risk-scoring system: single centre experience

Pregnancy in women with Fontan physiology poses a significant management challenge and is deemed high risk. The aim of this study is to describe short and long-term pregnancy outcomes in women who have undergone Fontan palliation and propose a novel risk-stratification model specific to women with a Fontan heart. We undertook a single-centre, retrospective cohort study of all female Fontan patients (n = 78) from 1991–2015. We recorded pregnancy outcome, maternal cardiovascular and obstetric complications and fetal outcome. We propose a risk stratification model to identify those women who might be at highest risk of adverse outcomes during pregnancy. Twenty-one women had 55 pregnancies, with 13 (24%) live births and 38 (69%) spontaneous miscarriages (p < 0.001). Eight (62%) out of 13 live birth pregnancies incurred maternal cardiovascular complications and six (46%) experienced maternal obstetric complications. Median gestational age at delivery was 32(27–39) weeks with 12 out of 13 (92%) pregnancies resulting in pre-term delivery. There were more pregnancies (OR 4.90, 95% CI 1.46–16.42, p ≤ 0.01) and a trend towards a higher proportion of live births (OR 7.60, 95% CI 1.81–31.97, p = 0.06), in the ‘lower risk’ compared to those women in the ‘very high risk’ group. There were no maternal deaths. We observed a high first trimester miscarriage rate, significant maternal cardiovascular and obstetric complication rates and a high rate of pre-term births in pregnant Fontan women. Our risk stratification model requires further investigation but may identify those women at particularly high risk of a poor outcome, and inform realistic pre-pregnancy counselling.


Background
The creation of a Fontan circulation has improved life expectancy for people born with functionally single ventricle congenital heart disease (CHD). However, the lack of a sub-pulmonary ventricle results in a chronic low cardiac output (CO) state with pulmonary blood flow dependent on adequate preload [1]. Patients often encounter complications in adulthood with a gradually declining functional capacity, atrial arrhythmias, myocardial dysfunction, thrombo-embolic events and hepatic dysfunction [2]. Reports suggest that the current UK single ventricle population is 1040 adults and 1700 children, with an expected increase in adult numbers by 60% in the next decade [3]. This improvement in survival to adulthood has resulted in an increased number of women with a Fontan circulation reaching child-bearing age.
Pregnancy is associated with considerable physiological stress. The normal heart can increase heart rate and stroke volume to adapt to changes in systemic vascular resistance and blood volume [4]. Patients with a Fontan circulation tolerate preload changes poorly, and their ability to increase CO is limited. Consequently, pregnancyrelated haemodynamic changes confer a considerable burden on the Fontan heart [5].
Published data regarding pregnancy in Fontan patients consists largely of small case series, as described in Table 1 [6][7][8][9][10][11][12][13]. Maternal cardiovascular morbidity, primarily due to atrial arrhythmias and deterioration of functional status has been observed. The rate of spontaneous miscarriages varied (27-65%) may have been underestimated in some of the series [6,10]. Moreover, a significant proportion of pregnancies (69-81%) resulted in premature births in the contemporary series [7,8,10]. Drenthen et al. and Cannobio et al. advised against pregnancy in these women, whereas more contemporary literature suggests that despite the comorbidities associated with the Fontan circulation, women can undergo pregnancy, albeit at the risk of encountering significant maternal morbidity [6][7][8][9].
We sought to determine the immediate and long-term outcomes of pregnancy within a large single-centre Fontan population. The main objective was to identify pregnancy outcomes, maternal cardiovascular morbidity and mortality, obstetric complications and fetal outcomes.
Current risk calculators such as CARPREG, ZAHARA and the modified WHO criteria either do not include women with a Fontan circulation, or simply classify them as high risk [14][15][16][17]. The CARPREG classification included women with both acquired and congenital heart disease and assigns a point for each predictor (prior cardiac event, New York Heart Association functional class > II or cyanosis, left heart obstruction, left ventricular ejection fraction < 40%). A score of 1 confers a 27% risk of maternal cardiovascular complications and a score > 1 confers a 75% risk [14]. The ZAHARA study also identified predictors of adverse maternal events, in patients with congenital heart disease [17]. The modified WHO criteria include specific cardiac lesions in addition to clinical cardiac status. It applies to women with acquired as well as congenital heart disease and the risk categories range from low risk (group I) to a very high risk (group IV) [15,16]. In this study, we also sought to identify factors that may confer increased maternal cardiovascular risk and propose a novel risk stratification model specific to women with a Fontan circulation to help identify those women at highest risk of suffering an adverse outcome during pregnancy.

Study design and patient population
A total of 98 women out of 224 Fontan patients were identified from our database of Adult Congenital Heart Disease (ACHD) patients at this Regional Tertiary ACHD Centre. Three women were subsequently followed-up elsewhere, two patients underwent cardiac transplantation prior to pregnancy, three patients were post-menopausal on referral to our unit and 12 patients died due to non-pregnancy related causes. A total of 78 women were included in the analysis.
A retrospective case review was undertaken of all the women with a Fontan circulation who were managed under the care of the combined Cardiology/Obstetric service from January 1991 to December 2015. Medical records from both the ACHD unit at the Queen Elizabeth Hospital Birmingham and the obstetric records from the Birmingham Women's Hospital were examined.
Baseline data collection included underlying cardiac anatomy, details of Fontan repair, New York Heart Association (NYHA) functional status defined as class I -IV [18], resting oxygen saturations, ventricular function defined as good, mild, moderate or severe impairment [19], and Fontan-related complicationssustained cardiac arrhythmias, hepatic dysfunction, thromboembolism, protein-losing enteropathy and plastic bronchitis. Systemic ventricular function was determined by echocardiography using a series of measurements including qualitative assessment, ejection fraction (EF) using single-plane endocardial tracings, M-mode, spectral Doppler and tissue Doppler derived indices for assessment of systemic ventricular function, as described elsewhere [19]. Where ejection fraction was calculated, the British Society of Echocardiography grading was used to determine the degree of impairment-normal (EF ≥55%), mild impairment (EF 45-54%), moderate impairment (EF 36-44%) and severe impairment (EF ≤ 35%) [20]. For those women who became pregnant, complications prior to the first pregnancy were taken as baseline data. The neverpregnant Fontan patients were sub-divided into two age groups -< 25 years and ≥ 25 years. This age cutoff was chosen as the likelihood of women under the age of 25 years not having a pregnancy may have been more likely to be the result of social and personal factors rather than ill health.
We proposed a novel risk-stratification model specific to women with a Fontan circulation based on their preexisting cardiovascular morbidity at the time of prepregnancy counselling. Women were divided into three groups, based on their pre-pregnancy cardiovascular status (Table 2). We used predictors of poor outcomes in Fontan patients and data from the modified WHO criteria, which is considered to be the best available pregnancy risk calculator in women with CHD [15,16,[20][21][22]:-(1) Lower risk -women with good functional class (NYHA I) with no previous Fontan-related complications (as described above), resting oxygen saturation ≥ 94%, normal or mildly impaired exercise capacity defined by peak myocardial oxygen consumption (MVO 2 ) on exercise ≥ 60% predicted, good ventricular function and no/mild atrio-ventricular valve regurgitation (AVVR). (2) Intermediate riskwomen with at least one of NYHA class II symptoms, treated Fontan-related complication (previous single, sustained arrhythmic event and/or thrombo-embolic complication with no recurrence at time of pregnancy), resting oxygen saturation 90-93%, at least moderately impaired exercise capacity defined by MVO 2 on exercise 50-60% predicted, mildly impaired ventricular function or moderate AVVR.

Study outcome and follow-up assessment
For each pregnancy, pregnancy outcome, maternal death, maternal and/or fetal complications during pregnancy and gestation length were recorded. Further detailed information regarding all on-going pregnancies was collected and included, hospital stay, indication for delivery, mode of delivery and anti-coagulation regime. Urgency/indication of caesarean section (CS) was classified as emergency, semielective and elective based on the Royal College of Obstetrics and Gynaecology classification [21].
Complications were divided into maternal death, maternal cardiovascular complications, maternal obstetric complications and fetal/neonatal complications. Cardiac complications included deterioration in functional status by one class, sustained arrhythmia requiring medical intervention/hospitalisation and heart failure requiring medical treatment. Obstetric complications included significant (≥500ml) intra-uterine/vaginal bleeding either during pregnancy or immediately post-partum or requiring a blood transfusion, thromboembolic events, preeclampsia (hypertension beyond 20 weeks gestation with blood pressure ≥ 140mmHg systolic and ≥ 90mmHg diastolic and proteinuria), pregnancy-related hypertension (blood pressure as defined for pre-eclampsia without proteinuria) and HELLP syndrome (haemolysis, raised liver enzymes, low platelets). Prolonged maternal hospital stay was recorded and was defined as in-hospital stay ≥ 7days.
Post-partum and long-term follow-up data were recorded for all women with live births including NYHA functional status, resting arterial oxygen saturations, ventricular function and cardiac events including sustained arrhythmias, heart failure, thromboembolic complications, conduction disease and Fontan-related complications (protein-losing enteropathy, plastic bronchitis and liver cirrhosis).

Statistical analysis
Statistical analysis was performed using XLSTAT (Microsoft Excel 2011). Categorical and continuous variables are expressed as numbers and percentages and as median (interquartile range), respectively. The Fisher Exact test was used to compare categorical variables, the Kruskal-Wallis test to compare ordinal variables, and the Mann-Whitney U test to compare continuous variables. The association between risk stratification and pregnancy outcomes was assessed by logistic regression analysis. P-value < 0.05 (two-tailed) was considered statistically significant.

Results
Seventy eight Fontan women of reproductive age were identified and their data are outlined in Table 3. All women were offered pre-pregnancy counselling with 82% of women undergoing pre-pregnancy counselling. There were fewer pregnancies in women < 25 years of age. Nine out of ten women who completed a pregnancy were in NYHA class I-II. Median baseline arterial oxygen saturation was 93 (92-94)% in the ten women with live births. Eleven women suffered 32 miscarriages with none of these women achieving live births. The median oxygen saturation in this group was 92 (90-95)%. In the 'never-pregnant' group the median arterial oxygen saturation was 93 (90-96)% and 94 (92-95)% in those ≥ 25 years and those < 25 years of age, respectively. There was no statistically significant difference in the baseline arterial oxygen saturations between the different groups (Table 3).

Pregnancy outcomes
There were 55 singleton pregnancies in 21 women. Maternal baseline (pre-pregnancy) characteristics are outlined in Table 4. Of these pregnancies there were 13 (24%) live births in ten patients, 38 (69%) spontaneous miscarriages (p < 0.05 compared to live births), three (5%) terminations and one (2%) ectopic pregnancy. Eleven women suffered a total of 38 miscarriages (36 in the 1st trimester, two in the 2nd trimester). There were no intra-uterine fetal deaths after 24 weeks gestation. There were no maternal deaths.
44% (24 out of 55) of pregnancies occurred in women in the proposed lower risk group, with 69% (nine) of the total livebirths. 29% [16] of pregnancies occurred in the intermediate risk group with 23% (three) live births and 27% [15] of the pregnancies occurred in the very high risk group with only 8% (one) live birth (Fig. 1). There were more pregnancies (OR 4.90, 95% CI 1.46-16.42, p < 0.01) and a trend towards a higher proportion of live births per total pregnancies for women in the group proposed to be at lower risk of pregnancy-related complications (OR 7.60, 95% CI 1.81-31.97, p = 0.06).
There was a trend towards the highest rate of miscarriage (including one termination and one ectopic pregnancy) in the proposed very high risk group (93.3%, 14 out of 15 pregnancies) vs 81.2% (13 out of 16 pregnancies, intermediate group) vs 62.5% (15 out of 24 pregnancies, lower risk group, p = 0.07 compared to the high risk group).

Maternal cardiovascular complications during pregnancy
Of the 13 live-birth pregnancies in ten women, the majority of the women (eight out of 13, 62%) described an increase in symptoms e.g. breathless, fatigue and/or limited exercise tolerance (Table 4). In four (31%) pregnancies in three women, there was a significant deterioration in functional class, necessitating treatment with diuretics and hospital admissionone with heart failure and sustained arrhythmia requiring intervention/hospitalisation and later an implantable cardiac defibrillator. Two further pregnancies (in two women) were complicated by atrial arrhythmias, requiring treatment with electrical cardioversion and beta-blockers.
In the pregnant women who did not experience a live birth (42/55 pregnancies with 41 pregnancies ending in the 1st trimester), there were no maternal cardiovascular complications.

Maternal obstetric complications during pregnancy
In four out of 13 (31%) live birth pregnancies, there was vaginal (PV) bleeding requiring hospital admission. In all of these cases, the baby was delivered by CS, with two women having pre-term CS as a direct result of the bleeding.
Three further women (whose pregnancies did not result in a live birth) suffered spontaneous miscarriages associated with PV bleedingat 12, 16, and 23 weeks gestation. One woman required a blood transfusion.
There were no cases of thrombo-embolic complications during pregnancy. One woman suffered a minor embolic cerebrovascular event three weeks post-partum despite therapeutic anticoagulation. She made a full recovery.

Anti-coagulation
Twenty out of 21 women received anticoagulation with warfarin prior to pregnancy. One woman was not on anticoagulation due to poor compliance. She suffered three 1 st trimester miscarriages whilst not taking warfarin, and had no live births.
In 52 out of 55 pregnancies, warfarin was converted to low-molecular weight heparin (LMWH, enoxaparin 1mg/kg twice daily) on confirmation of pregnancy, and continued for the duration of the pregnancy. Dosing regimen was as per Royal College of Obstetrics and Gynaecology guidelines and supervised by a haematologist with expertise of managing anticoagulation during pregnancy [13]. The remaining four women suffered early 1 st trimester spontaneous miscarriages, prior to conversion to LMWH. CVS cardiovascular, CS caesarean section, SGA small for gestational agebirth weight < 10 th centile. a Continuous data is presented as median (inter-quartile range). *p < 0.05 compared to live births. b A semi-elective CS was classified as those women requiring early delivery (< 37 weeks gestation) due to maternal or fetal reasons, but without an immediate risk to life

Neonatal complications during pregnancy
Twelve out of 13 (92%) of the live births resulted in pre-term delivery (< 37 weeks gestation). Four (31%) of the deliveries occurred between 33-36 weeks gestation, seven (53%) between 28-32 weeks gestation and one (8%) of the live births occurred at 28 weeks gestation (Table 4). There was one 2 nd trimester intrauterine death, complicated by placental bleeding requiring a blood transfusion. Fetal post-mortem revealed small-sized organs and no congenital or chromosomal abnormality.
Two women underwent CS at 31 and 34 weeks gestation for intra-uterine fetal growth restriction (small for gestational, age estimated fetal weight < 10 th customised centile)one of these women was significantly cyanosed with pre-pregnancy resting oxygen saturations of 88%. Neither of these women had been treated with beta-blockers during their pregnancy.
Mode of delivery is shown in Table 4. One woman had a vaginal delivery at 35 weeks gestation, three women had an emergency CS at 30, 31 and 39 weeks, and nine women had semi-elective CS (requiring early delivery but no immediate maternal or fetal compromise).
Five out of 13 neonates experienced an adverse outcome including cardiac congenital lesion (n = 2), significant bradycardia secondary to maternal beta blocker use (n = 1), SGA (n = 3) and developmental delay (n = 2). Two infants had congenital heart defects and neither required immediate cardiac intervention. Two neonates had cerebral palsyone born at 30 weeks gestation and the other at 34 weeks. There were no neonatal deaths.

Cardiovascular status at follow-uppost-partum and long-term
Thirteen post-partum clinic reviews were undertaken eight (5-11) weeks after delivery ( Table 5). The majority of women reported return of symptoms back to baseline, with 11 out of 13 reports of NYHA class I -II symptoms. Despite therapeutic anticoagulation, one woman suffered a confirmed embolic cerebrovascular accident three weeks post-partum. One woman (who underwent two pregnancies) had a significant and permanent deterioration of ventricular function following the first pregnancy.
Late follow-up data with a median of 4.8 (3.3-10) years were available for nine out of ten women. One woman had recently given birth and therefore no long-term postpregnancy data were available. There had been a deterioration in functional class from NYHA I to III in two women who had completed pregnancies. Both of these women had deterioration of ventricular function during pregnancy. Pre-pregnancy ventricular function was normal in both of these women and was moderately impaired in one woman and severely impaired in the second woman.
Five out of nine (56%) women with no prior arrhythmia had developed arrhythmia at the latest follow-up. Two women had developed a 'failing Fontan' state. One of these women had been in the proposed intermediate risk group and the other in the very high risk group, at time of pre-pregnancy counselling. Thromboembolic complications (n = 1), and significant conduction disease (n = 2) were also seen.

Discussion
With an improvement in survival following Fontan palliation, we may expect to see an increase in the number of Fontan women wanting to become pregnant. In our cohort of 78 women with a Fontan circulation, those with a shorter duration since Fontan repair were more likely to achieve a pregnancy compared to the women (≥25 years) who had a longer duration since Fontan repair. As might be expected, this latter group of women had a greater number of Fontan-related complications. In the lower age group (< 25 years) other factors such as relationship status and youth may also have been more likely to influence the decision to not pursue pregnancy.
We observed a high rate of spontaneous miscarriages (69%), cardiovascular morbidity (62%) and preterm delivery (92%). These pregnancies were predominantly in a group of 'well-functioning' Fontan patients. The high miscarriage rate in women with a Fontan heart is well described [7]. The cause of the high spontaneous miscarriage is unclear. Previous reports suggest an association between hypoxia and spontaneous miscarriages; however this was not borne out by our data [22].
Antepartum and postpartum haemorrhage was the main cause of obstetric complications in our series, and was similar to the 50% described by Cauldwell et al. [7]. In both studies, women received therapeutic or prophylactic low molecular weight heparin that was paused at delivery [13]. In contrast, in the series from the Mayo clinic [8], the majority of women received no anticoagulation or aspirin during pregnancy and suffered no significant haemorrhagic or thromboembolic complications. These important differences suggest that approaches to anticoagulation in pregnant women with a Fontan circulation require further study and may need to be reconsidered.
In contrast to published series, we observed a high rate of pre-term delivery (92%) and a higher rate of CS delivery [7,8,10]. The latter may be attributed to a higher rate of maternal and fetal morbidity seen in our cohort necessitating delivery rather than allowing spontaneous labour to take place. CS was the mode of delivery chosen for these women for preterm delivery to avoid risk of an emergency CS following failed induction, the risks of which would have been particularly high in this patient group [23].
Scoring systems to estimate maternal cardiovascular risk for women with heart disease attribute a high risk of adverse outcome for women with the Fontan circulation [14,16,17,24]. This risk is further increased with poor pre-pregnancy cardiovascular status (poor functional class and hypoxia) and in the presence of other complications such as thromboembolic disease or arrhythmias [14,15,22]. Our data show for the first time a trend towards livebirth rate being affected by maternal cardiovascular statuswomen with the worst cardiovascular status had a higher rate of miscarriage (93.3%) than those with good cardiovascular status (62.5%). Similarly, there was a trend towards a higher live birth rate in the lower risk group, with only one live birth in the veryhigh risk group. These data require validation in larger studies as they may provide a useful framework in informing pre-pregnancy counselling to be further refined to counsel women appropriately, i.e. women with the highest risk of complications during pregnancy may also be at the highest risk of miscarriage and have the lowest chance of having a pregnancy resulting in a live birth.
Our long-term follow-up data (median 4.8 years) demonstrated that Fontan complications were common in the years after pregnancy, and included atrial arrhythmia and heart failure. In this study, it is not possible to determine whether these complications were precipitated by pregnancy or occurred as part of the natural history of a Fontan circulation. Although it is difficult to draw firm conclusions from this dataset, our experience in a small number of individual women suggests that any deterioration in cardiac function that occurs during pregnancy may not recover. NYHA New York Heart Association classification, PPM permanent pacemaker; a One patient had no late follow-up data due to very recent pregnancy