Pregnancy outcome after liver transplantation: A single-center experience of 71 pregnancies in 45 recipients†
See Editorial on Page 1037
Infertility is common in women with end-stage liver disease. Successful liver transplant (LT), however, can restore childbearing potential. Controversy exists regarding the most appropriate immunosuppressive regimen and timing of conception following LT. We report the outcomes of a review of all pregnancies occurring following LT at King's College Hospital, London, from 1988 to 2004. Seventy-one pregnancies were recorded in 45 women. Tacrolimus (60%) and cyclosporin A (38%) were the predominant primary immunosuppressive agents used. Median age at conception was 29 years (range, 19–42), with a median time from LT to conception of 40 months (range, 1–111). There were 50 live births, and no maternal or fetal deaths related to pregnancy. There were no graft losses. Median gestation was 37 weeks (range, 24–42) with a median birth weight of 2,690 g (range, 554–4,260). Caesarean section was performed in 40% of pregnancies. Complications included pregnancy-induced hypertension in 20%, preeclampsia in 13%, acute cellular rejection in 17%, and renal impairment in 11%. There was no statistically significant difference in complication rates observed between immunosuppressive groups. Pregnancies occurring within 1-year posttransplant had an increased incidence of prematurity, low birth weight, and acute cellular rejection compared to those occurring later than 1 year. In conclusion, this study confirms that favorable outcomes of pregnancy post-LT can be expected for the majority of patients. However, delaying pregnancy until after 1-year post-LT is advisable, since doing so maybe associated with a lower risk of prematurity. Liver Transpl 12:1138–1143, 2006. © 2006 AASLD.
Amenorrhea and infertility are common in women with end-stage liver disease, affecting up to 50% of patients.1 In premenopausal women, a return to normal reproductive function after liver transplant (LT) is expected. Indeed, over 80% of women will have a normal menstrual cycle within 8 months of transplantation and restoration of menstrual bleeding may occur as early as 2 months post transplant.1–4 This allows for the opportunity of pregnancy, with its associated potential risks to the patient and challenges to the attendant physicians.
The first successful pregnancy in a liver graft recipient was reported in 1978.5 Chief concerns at that time were that transplantation would have an unfavourable effect on the course of the pregnancy, that immunosuppressive agents might be teratogenic, and that pregnancy might precipitate deterioration in graft function. The outcome was a healthy male of normal weight delivered after a normal gestation.5
Early pregnancy series of liver transplant recipients utilized cyclosporin A (CyA) as the primary immunosuppressive agent, and despite generally favorable outcomes, of concern was the increased risk of fetal growth retardation, preterm delivery, hypertension, and preeclampsia.6–9 Recently, pregnancy outcomes with tacrolimus-based immunosuppression have noted lower incidences of hypertension and preeclampsia when compared to CyA-based therapy.10–13 The largest reported series to date is the United States National Transplantation Pregnancy Registry established in 1991 to monitor pregnancy outcomes in transplant recipients.14, 15 This registry has reported on 137 pregnancies from 41 centers and describes a live birth rate of over 70%, in addition to favorable fetal and maternal outcomes for the majority of recipients.14 Data from this registry have shown that CyA, in comparison to tacrolimus, was associated with a higher incidence of preeclampsia and hypertension. However, diabetes was more prevalent in the tacrolimus group.14
A further consideration in liver transplant recipients is the fact that there is no clear consensus regarding the appropriate time interval between liver transplantation and conception. This contrasts with the renal transplant literature, where clear guidance in relation to both the timing and management of pregnancy existed 30 years ago.16 Currently, most liver transplant centers recommend waiting for 1 year following successful transplant. In this report we have reviewed the complete experience of pregnancy among LT recipients at King's College Hospital in London. We report on maternal and fetal outcome and provide further data on the optimal timing for conception posttransplant.
We conducted a retrospective study of all self-reported pregnancies in patients transplanted as part of the liver transplant program at King's College Hospital. There were 1,829 transplants in 1,623 adult recipients performed between January 1988 and December 2003, of whom 823 liver transplants were performed in 717 women. A total of 341 women were aged less than 45 years and considered to be of potential childbearing age. The clinical case notes of patients who reported a pregnancy were reviewed. Antenatal care for post-LT pregnant patients is coordinated between a specialist maternal-fetal unit based at our institution and the transplant team. Standard follow-up consists of monthly antenatal review in the first trimester, review every two weeks in the second trimester and weekly review in the final trimester. Target trough levels of immunosuppression were 50–150 μg/L for CyA and 4–8 μg/L for tacrolimus.
Gestational period, mode of delivery, and birth weight were determined. Renal function, serum electrolytes, hepatic biochemistry, and urinary protein analysis were recorded at baseline and measured serially throughout the course of pregnancy.
For quantitative data, analysis was performed using Mann-Whitney tests for comparisons of 2 groups. Differences in proportions was performed by Fisher's exact test when the number of subjects was <5, and chi-square test for 2 × 2 tables when the number of subjects in all cells was >5. Data are expressed as means ± standard deviations or median and range when appropriate. Analysis was performed using StatsDirect statistical software (CamCode, Ashwell, Hertfordshire, England).
A total of 71 pregnancies were reported in 45 women. These pregnancies form the basis of this analysis. The indications for LT are summarized in Table 1. The primary immunosuppressive agent at the time of conception was CyA in 27 (38%) of pregnancies and tacrolimus in 43 (60%) of pregnancies. Mycophenolate mofetil was used in conjunction with tacrolimus in 2 cases, whereas 1 patient received prednisolone and azathioprine alone without a calcineurin inhibitor.
Table 1. Indications for Liver Transplantation Among 45 Patients of King's College Hospital Who Became Pregnant Following Liver Transplant
|Acetaminophen toxicity||11 (24%)|
|Autoimmune hepatitis||9 (20%)|
|Seronegative hepatitis||4 (8.9%)|
|Budd-Chiari syndrome||4 (8.9%)|
|Wilson's disease||4 (8.9%)|
|Secondary biliary cirrhosis||3 (6.7%)|
|Primary biliary cirrhosis||2 (4.4%)|
For all pregnancies, median age at conception was 29 years (range, 19–42) with a median time between transplantation and conception of 40 months (range, 1–111). For women who delivered live infants, the median age at conception was 29 years (range, 19–42) with a median gestation period of 37 weeks (range, 24–42). These data are summarized in Table 2. There was no significant difference noted between age at conception, time between LT and conception, median gestation, or birth weight irrespective of the primary immunosuppression used.
Table 2. Patient Details at Time of Conception and Timing of Conception in Relation to Immunosuppression Regimen for All Patients, and for Those Patients Who Had a Live Birth
|Age at conception, years (range)||29 (19–42)||28 (23–29)||29 (19–42)||NS|
|LT to conception, months (range)||40 (1–111)||48 (5–111)||35 (1–109)||NS|
|Live Births|| || || || |
|Age at conception, years (range)||29 (19–42)||28 (23–39)||29 (19–42)||NS|
|LT to conception, months (range)||31.5 (1–111)||45 (6–111)||25 (1–108)||NS|
|Median gestation, weeks (range)||37 (24–42)||37 (24–42)||37.5 (26–40)||NS|
|Birth weight, g (range)||2,688 (554–4,256)||2,951 (554–3,800)||2,660 (732–4,256)||NS|
Three patients who had been pregnant died during follow-up with all deaths occurring more than 1-year postpartum. The causes of death related to posttransplant lymphoproliferative disease 18 months following delivery for autoimmune hepatitis in 1 patient, and septic shock in a second recipient transplanted for autoimmune hepatitis. Finally, 1 patient died from graft failure due to recurrent Budd-Chiari syndrome, 15 years after original transplant. No death was considered to be pregnancy related. Two patients required admission to the liver intensive care unit either during or immediately after delivery. One patient with established chronic ductopenic rejection awaiting retransplantation decompensated due to spontaneous bacterial peritonitis, whereas the second patient developed gram-negative sepsis following receipt of intravenous methylprednisolone as treatment of acute cellular rejection in the immediate postpartum period.
Fifty of the 71 (70%) pregnancies resulted in a live birth. There were 19 (27%) abortions, of which 13 were spontaneous and 6 were planned terminations. Planned terminations were performed as a consequence of deterioration in previously stable graft function (n = 1), warfarin treatment within the first trimester (n = 3), or uncontrolled psychiatric illness (n = 1). There was 1 molar pregnancy and 1 intrauterine death. Of the live births, 23 were male and 27 were female. The mode of delivery was vaginal in 25 (50%), caesarean section in 20 (40%), and assistance with either ventouse suction or forceps in 5 (10%). This compares with background rates of cesarean section of 23% (of which greater than 50% are considered emergency) in England for all pregnancies.17 For all live births, median birth weight was 2,688 g (range, 554–4,256). Median birth weights stratified according to the primary immunosuppression were 2,951 g (range, 554–3,800 g) for patients who received CyA, and 2,660g (range, 732–4,256 g) for patients who received tacrolimus (P = not significant). Scrutiny of birth weights demonstrated that 35 infants (70%) were of normal birth weight (>2,500 g), 10 (20%) were of low birth weight (1,500–2,500 g), and 5 (10%) were of very low birth weight (<1,500 g). This compares with a prevalence of low birth weight and very low birth weight of 5% and 1%, respectively, for all births in England reported for the years 2003–2004.17 The 5 very low birth weight children were all premature, with 2 born between 24 and 28 weeks' and 3 born between 28 and 35 weeks' gestation. All the very low birth weight children were admitted to the pediatric intensive care unit and all survived, with only 1 demonstrating delay in developmental milestones. There were no congenital abnormalities reported in our series, and only the 1 child born at 24 weeks who required prolonged pediatric intensive care unit stay had significant cognitive impairment. Fetal outcomes are summarized in Table 3.
Table 3. Outcome for 70 Self-Reported Pregnancies at King's College Hospital Following Immunosuppression With Either CyA or Tacrolimus
|All Patients||70||27||43|| |
|Abortions||19||7 (26)||12 (28)||1.0|
|Spontaneous abortion||13||6 (22)||7 (16)||1.0|
|Termination||6||1 (4)||5 (13)||0.475|
|Intrauterine death||1||1 (4)||0 (0)||0.813|
|Molar pregnancy||1||0 (0)||1 (2)||1.0|
|Live Births||50||21||29|| |
|Vaginal delivery||25||8 (38)||17 (59)||0.251|
|Assisted delivery||5||4 (19)||1 (3)||0.181|
|Cesarean section||20||9 (43)||11 (38)||0.842|
|Normal birth weight (>2,500 g)||35||15 (71)||20 (69)||1.0|
|Low birth weight (1,500–2,500 g)||10||3 (14)||7 (24)||0.616|
|Very low birth weight (<1,500 g)||5||3 (14)||2 (7)||0.702|
|Pediatric ICU admission*||5||2 (10)||3 (26)||1.0|
Graft Dysfunction and Pregnancy
With regard to graft function, there were 12 episodes of biopsy-proven acute cellular rejection during pregnancy. All patients underwent ultrasound guided liver biopsy without complication. Eight of 12 episodes represented only mild degrees of rejection and were managed by augmentation of the baseline immunosuppression. Four cases had rejection of a severity requiring addition of pulsed intravenous methylprednisolone (1 g/day for 3 days followed by an oral steroid taper, prednisolone 20 mg/day for 2 weeks, reduced to 5 mg/day thereafter until completion of pregnancy). In the immediate postpartum period, there were 2 further rejection episodes, 1 of which responded to an adjustment in the baseline immunosuppression (in a patient who had rejection proven on biopsy during pregnancy), and the aforementioned episode requiring pulsed intravenous steroids and subsequent intensive care unit admission.
There were no graft losses during pregnancy, although 7 patients subsequently underwent retransplantation. Regarding these 7 patients, all were more than 1-year postpartum, and graft loss was deemed to be unrelated to pregnancy. Five of these 7 patients had undergone an episode of rejection during pregnancy. Their indications for retransplant included: recurrent Budd-Chiari syndrome 11 years following original grafting and 8 years following pregnancy; chronic rejection 10 years following pregnancy in a patient transplanted for secondary biliary cirrhosis; hepatitis C virus cirrhosis (likely acquired around the time of original transplant) in a patient transplanted for autoimmune hepatitis, 8 years following pregnancy; and chronic rejection in a patient transplanted for Wilson's disease with recurrent cytomegalovirus related hepatitis infection 2 years following pregnancy. Finally, the last patient received 4 grafts as a consequence of recurrent autoimmune hepatitis over a 5-year period. Of those patients who did not exhibit rejection, the indications for retransplantation were late hepatic artery thrombosis in a patient transplanted for Budd-Chiari syndrome 7 years earlier and pregnant 5 years prior, and chronic rejection 8 years following transplant for familial idiopathic ductopenia and 6 years following pregnancy.
In this series, 7 pregnancies occurred in mothers who were hypertensive at confirmation of conception. Subsequently, 14 patients developed de novo hypertension during pregnancy. In all cases, a single antihypertensive agent was sufficient to control blood pressure. Preeclampsia was observed in 9 (12.6%) patients, and there was 1 episode of eclampsia. Interestingly, there was no statistical difference in the incidence of these complications between the CyA and tacrolimus groups.
There were 3 patients with diabetes in this series; 2 mothers had known insulin dependant diabetes prior to pregnancy, whereas only 1 patient developed gestational diabetes. All patients with diabetes were immunosuppressed with tacrolimus as their primary immunosuppressive agent. Eight patients experienced infectious complications, 5 (7%) with bacterial infection, and 3 with viral infections (4.2%) including 1 episode of cytomegalovirus reactivation requiring intravenous gancyclovir in the CyA group, and 2 episodes of parvovirus infection (one each in tacrolimus and CyA group). Maternal complications are summarized in Table 4.
Table 4. Maternal Complications During Pregnancy and in the Postpartum Period in Relation to the Primary Immunosuppressive Regimen
|Prepregnancy hypertension||7 (10)||3 (11)||4 (9)||1.0|
|Pregnancy-induced hypertension||14 (20)||3 (11)||10 (23)||0.339|
|Preeclampsia||9 (13)||4 (15)||4 (9)||0.749|
|Eclampsia||1 (1)||1 (4)||0 (0)||0.813|
|De novo renal impairment||8 (11)||2 (7)||6 (14)||0.651|
|Diabetes||2 (3)||0 (0)||2 (5)||0.689|
|Gestational diabetes||1 (1)||0 (0)||1 (2)||1.0|
|Sepsis||5 (7)||1 (4)||4 (9)||0.682|
|Viral infections||3 (4)||2 (7)||1 (2)||0.677|
|Rejection during pregnancy||12 (17)||7 (26)||5 (12)||0.222|
|Rejection postpartum||2 (3)||2 (7)||0 (0)||0.289|
|Maternal ICU admission||2 (3)||1 (4)||1 (2)||1.0|
|Maternal death||3 (4)||1 (4)||2 (5)||1.0|
Timing of Pregnancy
Since many transplant centers recommend that pregnancy be delayed for at least 1-year posttransplant, we explored the validity of this advice comparing the outcomes for women who conceived within 1 year of transplantation (early group) to those who conceived after 1 year (late group). In total, 12 (17%) pregnancies occurred within the first year posttransplantation compared to 59 (83%) after 1 year. There was no statistically significant difference between abortion rate (spontaneous or planned) and live birth rates between the 2 groups. Of note was the finding of a higher caesarean section rate of 50% in the early group vs. 24% in the later group. Interestingly, these data demonstrated no statistical difference between the incidence of hypertension, preeclampsia, or renal dysfunction in the 2 groups. However, rates of acute cellular rejection were significantly higher in the early, compared to the late pregnancy group (33% and 14%, respectively). The effect of time of conception post-LT in relation to fetal outcome and maternal complication rates are summarized in Table 5.
Table 5. Comparison of the Rates of Maternal Complications and Overall Fetal Outcomes in 71 Pregnancies Classified According to Conception Within 12 Months of Transplant or Conception Beyond 12 Months From Transplant
|All pregnancies||12||59|| |
|Abortions spontaneous/elective||1 (8)||18 (30)||0.155|
|Live births||11 (91)||39 (66)||0.152|
|Cesarean section||6 (50)||14 (24)||0.135|
|Hypertension||4 (33)||16 (27)||0.932|
|Preeclampsia and eclampsia||3 (25)||6 (10)||0.351|
|Renal dysfunction||3 (25)||5 (8)||0.250|
|Rejection||4 (33)||8 (14)||0.213|
|Premature||7 (58)||17 (29)||0.101|
|Pediatric HDU/ICU admission||2 (17)||4 (7)||0.582|
|Maternal ICU admission||2 (17)||0 (0)||0.021|
|Maternal death||0 (0)||3 (5)||0.991|
In this large single-center experience, we have demonstrated that successful pregnancy is a reasonable expectation in female patients of childbearing age following LT. Moreover, these data reinforce the concept that for the majority of patients, pregnancy is safe for both mother and offspring. Although 3 maternal deaths are reported, all three occurred at a time point that precludes a direct relationship with pregnancy per se.
A key consideration in the management of pregnancy in transplant recipients is the teratogenic potential of immunosuppressive drugs. In the early years of our transplant experience, immunosuppression was reduced in response to these fears in 2 patients. However, this permissive reduction of immunosuppression precipitated acute cellular rejection in both patients. Subsequent pregnancies were managed without specific alterations to trough immunosuppression levels, and indeed, that early experience emphasized the importance of both monitoring and maintaining therapeutic drug levels during pregnancy. No congenital abnormalities were observed in this series, and in that regard, these data are consistent with other reports of pregnancy outcomes in immunosuppressed patients.14
In this series, the frequency of cesarean section was high relative to the reported rate of cesarean section in the United Kingdom of 22.7%.17 Regardless of this rate, our reported rate of 40% is in keeping with most published series of LT recipients,9, 18 and it predominantly reflects the increased incidence of premature delivery among recipients. Interestingly, as our experience of managing these patients has grown, it is notable that the need for cesarean section has fallen in tandem with the reduction in prematurity. This reduction is in contrast to the trend toward increasing rates of cesarean section occurring in the United Kingdom, where rates have increased from 12% in 1991 to 22.7% in 2003–2004.17
Previous studies have reported that pregnant women post-LT have higher rates of renal impairment, hypertension, and preeclampsia.8, 9, 14, 19, 20 This is predictable, given the vasoconstrictive effects of both CyA and tacrolimus.21 Our experience reiterates these previous findings, albeit at lower incidences than those reported in other series.14 One possible explanation for these differences may relate to variability in target immunosuppression levels achieved at individual transplant centers. In this series, we did not identify any significant difference in complication rates between patients who received CyA compared to those who received tacrolimus-based primary immunosuppression. These discrepancies, however, may be a reflection of the small numbers involved in our study.
In general, most transplant centers recommend a delay of pregnancy until after 1 year following transplant. This recommendation is supported by National Transplantation Pregnancy Registry data that suggests a greater likelihood of a live birth and less graft rejection with delayed pregnancies.14 There is sound clinical reasoning underpinning this advice. First, in the early posttransplant period, higher levels of immunosuppression may be required, rejection episodes are typically more frequent, and infections such as cytomegalovirus are more common.22 To clarify further the optimal timing of pregnancy among recipients, we compared outcome from pregnancies occurring within 1-year posttransplant to those occurring greater than 1-year posttransplant. Although no statistically significant differences in outcome was identified, other than for increased risk of intensive care admission for the mother, when patients were stratified in this fashion, interpretation of data is confounded by the small number of pregnancies occurring in the group of patients who conceived within 12 months of transplant.
Regarding graft loss, we experienced no losses during pregnancy, yet despite excellent graft survival our acute cellular rejection rate of 17% was over twice that reported by the National Transplantation Pregnancy Registry.14 This finding almost certainly represents the acceptance of lower levels of immunosuppression in this center. However, most patients responded to relatively minor adjustments to their immunosuppression regimen. Taken in this context, the higher acute rejection rates reported in this series may represent the price of lower levels of pregnancy-related complications. A further important observation from this study was the safety of ultrasound-guided liver biopsy in assessing graft dysfunction in pregnancy. In most circumstances, histological specimens can be obtained to differentiate different causes of acute graft dysfunction with confidence.
Of note, 7 patients required retransplantation, and although 5/7 recipients had undergone a previous rejection episode, the indications for retransplantation were diverse and attributable to disease recurrence in 2 patients, acquired hepatitis C virus infection in a third and chronic rejection in the remaining 2 patients. For these latter 2 patients, chronic rejection was as much related to recurrent viral infection in the form of cytomegalovirus and parvovirus infection as it was to adequacy of immunosuppression. Moreover, viral infection is a well-recognized cofactor in the development of chronic rejection among liver transplant recipients.22
In conclusion, this large single-center study reiterates the findings of favorable outcomes for pregnancy posttransplantation. Given the rapid return of libido and fertility after transplantation, family planning advice is essential. Although it may be prudent to delay pregnancy for a period of at least 1 year posttransplant, if patients conceive prior to that time point, then a tone of cautious optimism may be adopted once rigorous hepatological and obstetric follow-up is undertaken.