Aplastic anemia (AA) is an acquired bone marrow failure (BMF) disorder characterized by hypocellular bone marrow and pancytopenia. The pathogenic mechanism is an immune-mediated attack against hematopoietic precursors, and the concomitant detection of a paroxysmal nocturnal hemoglobinuria (PNH) clone, or its emergence over time, is a frequent event. Pregnancy is one of the proposed risk factors for autoimmune disease development and cases of AA in pregnant women were described in numerous case series.^1-5 However, limited data are available on the management of BMFs during pregnancy or on feto-maternal outcomes.

We conducted a single-center retrospective cohort study including women with acquired AA and/or PNH, referred to a tertiary center in the United Kingdom over the last 25 years. The study focused on women who experienced either a relapse of known AA or a de novo occurrence of BMF during pregnancy. Demographic, clinical, and disease severity findings according to Camitta criteria⁶ were collected. Additionally, information regarding treatment administered before, during, and after each pregnancy, response to AA-treatment, according to RACE trial adopted criteria,⁷ as well as feto-maternal outcomes were systematically recorded.

Seventy pregnancies occurring in 52 women were included from a cohort of 266 women with BMF followed from 1989 until 2022; 17 patients developed AA de novo during pregnancy, while 28 had a diagnosis of AA predating pregnancy (Table S1). Additionally, seven patients had a hemolytic PNH (two with onset during pregnancy and five with a previous diagnosis).

Regarding 17 de novo AA, most presented during the first trimester, 57% versus 36% and 7% in the second and third trimester respectively, with severe presentation in 29% of cases. In seven patients a PNH clone was detected with a median granulocytes clone of 5.2% (1.7–19). During pregnancy, 16 (94%) required blood transfusions and eight (47%) immunosuppressive therapy (IST) with cyclosporin (CSA) (five patients, three responsive) or steroids (three subjects, one responsive). After delivery, spontaneous hematologic improvement occurred in seven subjects (41%), however, 10 (59%) required IST or transplant (5 ATG + CSA ± Eltrombopag; 4 CSA/Androgens; 1 HSCT) all with response. Maternal adverse events (AEs) were observed in 29% of pregnancies, including two peripartum bleedings, one preterm premature rupture of the membrane (PPROM), and a Cesarean scar infection. Three women needed an urgent Cesarean Section (CS) (one for PPROM, two unknown), and three received additional peripartum transfusion support. No spontaneous miscarriages occurred, but fetal AEs were reported in 24% of pregnancies, including three late preterm births (32–37 gestation weeks, according to WHO definition of prematurity⁸) and one fetal growth restriction (Table 1).

Considering the 38 pregnancies in 28 women with previous AA diagnosis, the median time between diagnosis and pregnancy was 10 years (0.5–26). PNH clone was detected in 16 cases (median granulocyte clone 13.6%, 0.08–41). Twelve non-severe patients had been managed with follow-up only, while 16 had received at least a previous therapy for AA (12 ATG-based regimen; four CSA/eltrombopag). At pregnancy onset, eight were in complete response, two were in partial response off-treatment, three were on CSA and one on eltrombopag, one was off-treatment not in response, and one was not known. Eltrombopag was interrupted and substituted with CSA after the positive pregnancy test. Relapse of AA occurred in nine pregnancies (9/38, 24%): five relapses in patients with previous IST (three off-treatment and two on CSA) and four untreated cases showed worsening of disease severity, with equal distribution among trimesters and non-severe presentation. During pregnancy, 14 (37%) required transfusions (including seven due to PNH-related hemolysis or for decrease of Hb/platelets counts without fulfilling relapse criteria). Relapse occurred in two of the three patients on CSA at pregnancy onset, while the one who switched from eltrombopag to CSA, even without an overt relapse, presented an increase in transfusion need. A total of seven patients received CSA during pregnancy, and three responded. Spontaneous improvement after delivery occurred in 5/9 relapsed patients, however, IST was necessary in four (3 CSA ± Eltrombopag; 1 ATG + CSA) with response. Four (11%) spontaneous miscarriages occurred, three in the first trimester and one in the second, all in women who evolved to hemolytic PNH on eculizumab. Maternal AEs occurred in 24% of pregnancies, including a Budd–Chiari syndrome after CS in an AA patient who evolved to overt PNH in the pre-eculizumab era, four infectious complications, two peripartum hemorrhages, and two cases of preeclampsia. Two women needed urgent CS for preeclampsia, and one received additional peripartum transfusion support. Fetal AEs were reported in 5% of pregnancies, including two extremely preterm births (<28 gestation weeks).

Twenty-two women (seven PNH and 15 AA) experiencing 34 pregnancies had a granulocyte PNH clone >10% at gestation onset (median 40.5%, 13.6–99). Seven patients were on eculizumab at the time of the first pregnancy, while 15 initiated it during the first or following pregnancies. During gestation, 21 of the 34 pregnancies (62%) required transfusions and 33 (97%) eculizumab (13 preemptive due to clone >20% and modest hemolysis, LDH <2 xULN, and 20 due to overt hemolysis). Three women received both CSA and eculizumab. Eculizumab dose varied from 900 to 1500 mg every 2 weeks, and progressive dose increases were needed in 17 (52%) due to worsening of hemoglobin and hemolytic markers. Complement inhibitor was continued beyond puerperium in 26 pregnancies. Thromboprophylaxis was given during 31 pregnancies, mainly with low molecular weight heparin. Five (15%) spontaneous miscarriages were reported: four and one in the first and second trimester, respectively, all in women already on eculizumab. Maternal AEs occurred in 29% of pregnancies, including the Budd–Chiari syndrome described above, four infective events (only one was related to neutropenia and never requiring intensive care), two preeclampsia, two preterm premature membrane rupture. Five patients required urgent CS or preterm delivery (two for PPROM, two for preeclampsia, one unknown), and four women received additional peripartum transfusion support. Fetal AEs were reported in 8% of pregnancies, including two extremely preterm and one very preterm births (28–32 gestation weeks).

Eight patients (seven with a previous AA diagnosis and one de novo) had multiple pregnancies (six patients had two pregnancies, and two patients had three pregnancies). No relapses were observed during the following pregnancies, and in particular, three women with AA onset/relapse during the first pregnancy had a successful second pregnancy without relapse.

This is one of the largest case series on the management and outcomes of pregnancy in women with either de novo or previously identified BMFs. In our cohort, AA onset during pregnancy was a rare event (17/266; 6%) and occurred mainly during the first trimester with a severe presentation in 29% of cases. It required transfusion support in 94% and IST in 71%, either during or after pregnancy, and feto-maternal AEs were reported in 24% and 29% of pregnancies, respectively, mainly in the perinatal period, due to bleeding secondary to thrombocytopenia, but with no maternal deaths. Along with previous literature, these data confirm that this is a high-risk condition, warranting close monitoring and awareness of potential complications.^{2, 5} Concerning AA preceding pregnancy, relapses during pregnancy occurred in 24% of cases, slightly higher than previously reported.^{2, 3} In particular, comparing our results with those reported by Tichelli et al.,³ despite more relapses and similar transfusion dependency, we observed fewer feto-maternal AEs, mainly related to bleeding, minor infective complications, and no maternal deaths. Relapses occurred both in women in treatment-free remission and on CSA, suggesting that CSA is not completely protective against relapse. We were unable to identify any predictors of AA relapse and/or spontaneous improvement, however, the mainly non-severe presentation, the frequent spontaneous improvement during postpartum period, and the possibility of having a subsequent uneventful pregnancy appear reassuring. Safety of CSA was confirmed in our cohort, with no treatment-related adverse events, as shown in renal transplant women on cyclosporin during pregnancy.⁹ CSA induced a response in about 50% of de novo/relapsed cases, and even higher response after delivery, suggesting a role of the hormonal/immunological alteration in pregnancy and the postpartum period.^2-4 Given the retrospective nature of this study it is difficult to make recommendations about the management of AA relapse during pregnancy. In general, given the safety concern for the fetus, the most possible conservative approach is adopted with optimization of supportive care with therapeutic levels of CSA, but no safety data is available for the use of ATG during pregnancy. The high possibility of spontaneous recovery postpartum and improvement with CSA with transfusion support during the intrapartum period, also adds further evidence that intensification of treatment with ATG during this period is not required. Finally, the use of eltrombopag during pregnancy in AA is limited to a single case report¹⁰ and its routine use should be avoided. Comparing to immune thrombocytopenia (ITP), eltrombopag in AA is used at higher dose and this could expose the pregnant and the fetus to side effects not described in ITP. However, its use may be acceptable in refractory cases.

In the pre-eculizumab era, pregnancy in PNH patients was often discouraged due to the high maternal/fetal morbidity and mortality rate secondary to thromboembolism.¹¹ The advent of eculizumab improved pregnancy outcomes and our data are very similar to those previously reported.¹¹ However, the detrimental impact of PNH on pregnancy has not been completely abolished by eculizumab, as reported by Alashkar et al.¹² and our study confirmed that miscarriages and main feto-maternal AEs almost all occurred in PNH women already on eculizumab. To date, eculizumab is the only complement inhibitor which can be safely administered during pregnancy, as no safety data is available for other anti-complement therapies in this setting. A clear cutoff for initiating eculizumab during pregnancy is not established and is generally advised for clone size >20%. During pregnancy, close monitoring of hemolytic markers is needed to intercept breakthrough hemolysis and to evaluate drug dose increases. Eculizumab is generally continued until at least 3 months postpartum, and withdrawal after this period should be carefully evaluated.

In conclusion, pregnancy remains an immunomodulatory state, associated with a risk of manifestation/relapse of AA, a quintessential autoimmune disorder. Complications associated with AA, such as PNH, previously considered a relative contraindication to pregnancy, can nowadays be successfully managed with anti-complement therapies. Equally, AA during pregnancy can be managed with a judicious transfusion program, careful antenatal monitoring, and planning for delivery with close liaison between BMF specialists and obstetricians.

AUTHOR CONTRIBUTIONS

MB and AK designed the study. AK, RT, SS, and JL contributed to data acquisition. MB performed the research, analyzed the data, and wrote the first draft of the paper. AK, BF, WB, RT, EV, SG, and JJ critically reviewed the manuscript, and all authors approved the final version.

CONFLICT OF INTEREST STATEMENT

No conflict of interest to declare.