Pregnancy outcomes in women receiving eculizumab for the management of paroxysmal nocturnal haemoglobinuria

Background

Paroxysmal nocturnal haemoglobinuria (PNH) is a haematological disorder characterised by chronic complement-mediated intravascular haemolysis, thrombosis and varying degrees of bone marrow failure. ¹ Before the eculizumab era, the co-occurrence of PNH and pregnancy was associated with a disproportionately large degree of morbidity and mortality, to such an extent that pregnancy was discouraged in these women.^2,3

The physiological changes of gestation may exacerbate the PNH phenotype in a number of ways: early physiological haemodilution may worsen pre-existing anaemia, the pro-thrombotic state induced by pregnancy further elevates risk of thrombosis, and increased terminal complement activation in later gestations may exacerbate intravascular haemolysis.^4,5 Prior to the introduction of targeted therapies, prognoses for pregnant women with PNH were poor, with a maternal death rate of around 8%.^2,3 Fetal outcomes were similarly poor; one study reported fetal deaths at 5%, near ubiquitous low birth weights and prematurity in around one-third of cases. ³

Eculizumab is a terminal complement inhibitor, which hampers the formation of the membrane attack complex preventing complement-mediated intravascular haemolysis. ⁶ Large-scale RCTs have demonstrated a significant benefit from the use of eculizumab in non-pregnant women, reducing both symptom burden and complication rate.^7,8 However, data regarding benefits to pregnant women are limited.

To date, most studies investigating eculizumab use in pregnancy have leveraged small sample sizes, with only one previous large-scale study.^2,9–14 Therefore, the aim of the present work is to investigate maternal, fetal and neonatal outcomes in a larger cohort of pregnant women taking eculizumab in a national PNH referral centre.

Methods

Routinely collected medical records from 33 women with PNH spanning 70 pregnancies across UK centres were reviewed. Informed consent was not sought since this was a low-risk service evaluation. Pregnancies were excluded if PNH diagnosis post-dated pregnancy (n = 12), data were unavailable (n = 3) or if eculizumab was never received in pregnancy (n = 6), since women never requiring treatment with eculizumab had significantly milder disease, making comparisons between them and women receiving treatment problematic. This left 49 pregnancies from 21 women between October 2007 and October 2018 for analysis. Baseline was defined as the point when each woman first commenced eculizumab in pregnancy.

Eculizumab was commenced according to UK consensus guidelines, which included any pregnant woman with a PNH clone of over 20% in any lineage, using standard loading and maintenance doses. ¹⁵ Doses were increased where there was evidence of ‘breakthrough’ intravascular haemolysis – where complement activation increases to such an extent that it overwhelms the inhibitory effects of eculizumab. Where higher doses were required, these were reduced to pre-pregnancy baseline levels immediately after delivery. Breakthrough haemolysis was defined as an increase in LDH of 1.5–2 times the upper limit of normal and/or clinical assessment (fatigue, shortness of breath, dark urine). Complement haemolytic activity was not routinely measured since AH50/CH50 assays take weeks to be reported and are therefore not suitable in acute scenarios requiring immediate management.

Low molecular weight heparin (LMWH) was commenced in all women upon pregnancy diagnosis and was titrated according to anti-Xa assays, with a peak target anti-Xa level of 0.5–1.0 iu/mL.

Antepartum and postpartum haemorrhage was classified into minor and major, with major defined as bleeding that resulted in haemodynamic compromise and/or required emergency intervention. Gestational age was determined by ultrasonography, with prematurity defined as delivery before 37 weeks’ gestation. Prematurity was further classified as moderately preterm (32–36 weeks and six days of gestation), very preterm (28 to 31 weeks and six days) and extremely pre-term (less than 28 weeks), as per the World Health Organisation classification. Intrauterine growth restriction (IUGR) was defined as a fetus falling through growth chart centiles or a small for gestational age fetus with abnormal umbilical artery Dopplers.

Odds ratios were calculated to see if any maternal/fetal complications were more common depending on the point at which eculizumab was commenced.

Results

Study population

Around one-third of women (n = 8) became pregnant prior to ever receiving eculizumab therapy and then went on to have subsequent pregnancies in which eculizumab was used. This enabled us to report on 12 pregnancies in which eculizumab was not used. In total there were 37 pregnancies in which eculizumab was used and it is these pregnancies that the majority of our data focusses on (Table 1). Mean age of eculizumab commencement in pregnancy was 29 years.

OPEN IN VIEWER

Table 1.

Characteristics of the participants and their pregnancies.

Baseline patient characteristics (n = 21)
Age –mean (SD)	29 (5.4)
History of aplastic anaemia –n (%)	14 (66.7)
History of thrombosis –n (%)	1 (4.8)
Pregnancies not on eculizumab (n = 12)
Pregnancy outcomes –n (%)
Live birth	7 (58.3)
Miscarriagea	2 (16.7)
Termination	3 (25.0)
Pregnancies on eculizumab (n = 37)
Number of pregnancies per woman – median (range)	2 (1–4)
Eculizumab commencement –n (%)
Prior to conception	22 (59.5)
First trimester	4 (10.8)
Second trimester	8 (21.6)
Third trimester	3 (8.1)
Pregnancy outcomes – n (%)
Live birth	31 (83.8)
Miscarriagea	3 (8.1)
Intrauterine deathb	2 (5.4)
Ectopic pregnancy	1 (2.7)

^aMiscarriage was defined as fetal loss before 24 weeks’ gestation.

^bIntrauterine death was defined as fetal loss beyond 24 weeks.

Maternal outcomes

There were no reported cases of maternal death. Eculizumab was commenced prior to conception in over half of cases and all women continued eculizumab through the post-partum period (Table 1). One woman had twins and the remainder had singleton pregnancies. There was one reported case of a tubal ectopic pregnancy requiring salpingectomy. The most common maternal complication was breakthrough intravascular haemolysis requiring an increase in the dose or frequency of eculizumab infusions, occurring in 54% of the pregnancies (Figure 1). Increases were primarily required at later gestations. Whilst the majority of women required only one increase, several required two or three to control haemolysis (Figure 1). Women who commenced eculizumab prior to pregnancy were more likely to experience breakthrough haemolysis than those who started it during their pregnancy (OR of breakthrough in pre-pregnancy vs. during pregnancy = 9.35, 95% CI 2.05–42.7, p = 0.004).

OPEN IN VIEWER

Figure 1.

Breakthrough haemolysis requiring one or more eculizumab dose increase was seen in 54% of pregnancies.

All women were anticoagulated with LMWH throughout pregnancy and the postpartum period. In one case, a woman anti-coagulated with warfarin prior to pregnancy had the diagnosis of pregnancy missed. Sadly, the fetus displayed signs of warfarin embryopathy and an intrauterine death was confirmed at 16 weeks.

There were no reported cases of thrombosis during pregnancy or in the three-month postpartum period. Bleeding, however, was a relatively common occurrence. Minor bleeding occurred in 18.9% of pregnancies. Major bleeding necessitating urgent intervention was reported in 10.8% of cases (Table 2). Antepartum causes of bleeding included two cases of placental abruption, leading to fetal loss in one. Minor bleeding included four cases of recurrent epistaxis, and one case each of minor vaginal bleeding and gum bleeding. Postpartum, one woman had a severe primary postpartum haemorrhage, and another had a secondary postpartum haemorrhage due to retained products of conception requiring surgical evacuation. A further woman developed a small haematoma following LSCS.

OPEN IN VIEWER

Table 2.

Fetal and maternal outcomes in eculizumab pregnancies.

Characteristics of live births (n = 31)
Mode of delivery –n (%)
Vaginal	15 (51.7)
Induced – 12 (80)
Spontaneous – 3 (20)
Instrumental	1 (3.4)
LSCS	13 (44.8)
Emergency – 7 (53.8)
Elective – 6 (46.2)
Gestation in weeks –median(range)	37 (28–40)
Maternal complications
Antepartum haemorrhage –n (%)
Minor	6 (16.2)
Major	2 (5.4)
Postpartum haemorrhage –n (%)
Minor	1 (2.7)
Major	2 (5.4)
Thrombosis –n (%)	0 (0)
Infection requiring antibiotics –n (%)	6 (16.2)
Pre-Eclampsia –n (%)	5 (13.5)
Fetal complications
IUGR –n (%)	4 (10.8)
Premature birth –n (%)	8 (25.8)

IUGR: Intrauterine growth restriction; LSCS: lower segment caesarean section.

Six pregnancies were complicated by infection (Table 2). The majority of these were minor antenatal infections requiring courses of oral antibiotics: two urinary tract infections, one upper respiratory tract infection (URTI) and two lower respiratory tract infections. Three women had more severe postpartum infections requiring admission and treatment with intravenous antibiotics – one with mastitis and sepsis, another with pyrexia of undetermined origin and the third with endometritis secondary to retained products. One woman had a minor URTI six weeks postpartum treated with oral antibiotics. There were no cases of meningococcal infection.

Pre-eclampsia was diagnosed in five pregnancies from four women. In one case, this was complicated by haemolysis, elevated liver enzymes and low platelet (HELLP) syndrome leading to a rapidly worsening thrombocytopenia that necessitated emergency delivery.

Fetal outcomes

Overall, there were 31 live births, three miscarriages and two intrauterine deaths (IUD) (Table 1). One IUD occurred at 30 weeks following antepartum haemorrhage due to placental abruption, and the other at 26 weeks following severe growth restriction. Placental histopathology in the latter case demonstrated an immature placenta with patchy chorioamnionitis and established infarcts with retroplacental haematoma consistent with IUGR. Notably, this woman was not started on eculizumab until around the time IUD was diagnosed. Delivery was induced and she had an uncomplicated postpartum course.

Of the live births, around half delivered vaginally with the majority of those induced at 37 weeks and only 20% going into spontaneous labour. One woman required instrumental delivery by vacuum extraction. Just under half delivered by lower segment caesarean section (LSCS) with an approximately equal split between elective and emergency procedures (Table 2).

Prematurity was the most common fetal complication, seen in 25.8% of live births (Table 2). Seven infants were moderately preterm (34–36 weeks’ gestation) and one was very preterm (28 weeks). The only other documented fetal complication was IUGR, seen in four cases (Table 2).

Neonatal outcomes

Data for neonatal outcomes was sparse with no data available on newborn vital status or developmental progress. There were no documented cases of neonatal death. Mean birth weight was 2.83 kg (Table 3). Three infants required a prolonged stay in hospital – one for transient jaundice and hypoglycaemia, another with respiratory distress and microcolon requiring temporary stoma formation and the third due to excessive primary weight loss (11%). Over half were bottle fed. Eculizumab breast milk assay results were available for three neonates, with no drug detected in any. Cord blood samples were available in four cases. In two cases, no drug was detected and in the remaining two eculizumab was detected, but in levels below therapeutic range (Table 3).

OPEN IN VIEWER

Table 3.

Neonatal outcomes.

Birth weight – mean (range)	2.83 (2.21–3.36)
Feeding status – n (%)
Breast	9 (60)
Bottle	6 (40)
Eculizumab assays – n (% detected)
Cord blood	4 (50)a
Breast milk	3 (0)

^aWhere detected, levels below therapeutic range.

Discussion

This study aimed to complement the existing literature around the safety and efficacy of eculizumab during pregnancy, which previously only one large-scale study has investigated. ¹⁰ To this end, the present work has demonstrated that eculizumab may largely be considered safe, as evidenced by a low rate of maternal, fetal and neonatal complications.

There were no reported cases of maternal death, compared to a rate of around 8% in the pre-eculizumab era. ³ Fifty-four per cent of women experienced breakthrough intravascular haemolysis despite eculizumab therapy, a figure almost identical to that reported in previous work. ¹⁰ Haemolysis was most common in the late second and third trimesters, which is consistent with physiological increases in terminal complement activation during late gestation.^5,16 Our data showed that women who start eculizumab before pregnancy were more likely to experience breakthrough haemolysis, almost certainly because those who required earlier complement blockade had more severe disease. Future work should seek to identify if there are factors that can predict the development of breakthrough haemolysis, which may provide a rationale for prophylactically increasing eculizumab doses at earlier gestations.

Importantly, no cases of venous thromboembolism (VTE) were reported during pregnancy or the three-month postpartum period in a condition known to be pro-thrombotic. Before eculizumab it has been reported that 14% of pregnancies were complicated by thrombosis, suggesting that complement inhibition confers a protective effect against pregnancy-associated VTE in these women. ² However, there were two documented cases of thrombosis occurring at five months and seven months postpartum, respectively. In both cases, this appeared to coincide with the cessation of eculizumab, which cautions against prematurely withdrawing complement blockade in women who have recently given birth. Indeed, Kelly et al. suggest that an ‘extended’ post-partum period of six months should be considered in these women. ¹⁰

Balancing the risks of VTE with haemorrhage presents a challenge in pregnancy complicated by PNH. All women in this population were anticoagulated with LMWH throughout pregnancy, and 18.9% experienced bleeding, which represents a similar risk to that reported in the pre-eculizumab era. ² VTE remains one of the leading causes of maternal mortality and is especially pertinent in PNH women. It is essential that anticoagulation is closely monitored and accordingly titrated to prevent severe haemorrhagic events. This may be achieved by regular anti-Xa monitoring, which should form part of standard antenatal care for these women.

Few women went into spontaneous labour, with delivery usually planned for at 37 weeks by induction of labour or LSCS. This practice allows for birth to occur in a highly controlled environment with sufficient planning of intrapartum care such as ensuring availability of blood products, or further doses of eculizumab.

There were no cases of fetal malformation, suggesting that eculizumab is not teratogenic. The percentage of pregnancies that ended in live birth was 83.8, an improvement on figures from the pre-eculizumab era where fetal survival lay at 70%. ² In total, there were 12 pregnancies in which eculizumab was not used, since these women became pregnant at a time prior to when they first received the drug. Data from these 12 pregnancies demonstrate only 58.3% live births, with the remainder of pregnancies ending in miscarriage or elective termination. The most common fetal complication was pre-term delivery, affecting around one in four births, a figure notably higher than the average UK prematurity rate of 7.3%. ¹⁷ Premature birth was almost always iatrogenic by emergency LSCS, necessitated due to maternal medical complications (placental abruption, pre-eclampsia, thrombocytopenia) or fetal distress. Prior to eculizumab, studies report a prematurity rate of around one-third, suggesting eculizumab does not significantly modify prematurity risk. ³

Data on neonatal outcomes were limited which makes drawing conclusions difficult. Future work should aim to collect more extensive data on the neonates to enable more certainty of results described. Eculizumab was not detected in the three breast milk samples reported. This suggests breastfeeding is safe when taking eculizumab and that women should be encouraged to do so if they wish. Cord blood samples suggest that eculizumab is able to cross the placenta, but at levels well below the therapeutic range, and previous findings suggest that eculizumab does not affect complement activity in the newborn. ¹⁴

This study has several limitations. The retrospective design based on routinely collected medical records risks bias as some data may be missing or incomplete. However, it would not be ethical to conduct randomised controlled trials on these women since the risks of pregnancy without pharmacological complement blockade are so significant.^2,3 Furthermore, since PNH is exceedingly rare, the sample size is relatively small, which makes detecting complications with a low rate of occurrence difficult. This said, the study was conducted in a national PNH referral centre which undoubtedly allowed identification of a greater number of women than would be possible elsewhere.

Despite these limitations, this study provides further evidence to suggest that eculizumab is safe and effective as a treatment for PNH during pregnancy. This is evidenced by a low rate of maternal, fetal and neonatal complications and is supported by previous work in this area. ¹⁰ Avenues for future work include investigating optimal anticoagulation strategies to maximise thromboprophylaxis whilst limiting major bleeding events, as well as further studying the causes of prematurity in this population.