Effect of Pregnancy on Long-Term Kidney Function in Renal Transplant Recipients Treated with Cyclosporine and with Azathioprine



In order to investigate the effect of different immunosuppressive regimens and the time interval between transplantation and pregnancy on long-term outcome, we performed a case-control study in pregnant renal allograft recipients. Eighty-one pregnancies of kidney transplanted recipients were identified [cyclosporine (CYA): n = 40; azathioprine (AZA): n = 41]. Controls were matched with respect to important prognostic factors. Posttransplant follow-up was 91.3 ± 5 months. Graft and patient survival were similar in both groups and there was no apparent effect of immunosuppression. A total of 28 recipients (33%) delivered within 2 years and 6 (8%) subjects within 1 year after transplantation, but these short transplantation-to-pregnancy intervals had no apparent adverse effect on long-term outcome. In contrast to AZA-treated patients, CYA-treated patients experienced an increase in serum creatinine postpartum (1.15 ± 0.2 mg/dL vs. 1.61 ± 0.1 mg/dL; p < 0.05). Whole blood CYA levels decreased transiently during pregnancy from 115.9 ± 8 ng/mL to 80.7 ± 7 ng/mL leading to a gradual increase in drug dose from 240 ± 14 mg/day to 324 ± 21 mg/day (p < 0.05). Following delivery, there was an increase in CYA concentrations to 173 ± 5.4 ng/mL, requiring rapid dose tapering to baseline of 246 ± 15 mg/day. Pregnancies in renal recipients do not affect long-term patient and graft survival, independent of the immunosuppression. No detrimental effect of short transplantation-to-pregnancy intervals on long-term graft function was detected.


Successful renal transplantation usually restores reproductive function and leads to a reversal of the relative infertility that accompanies end-stage renal failure (1). In contrast to patients on dialysis, where pregnancies are rarely reported, about 1 in 50 women of childbearing age with a functioning graft becomes pregnant (2). Pregnancy in healthy gravidas is associated with an increase in glomerular filtration rate (GFR) of approximately 50% in early pregnancy, which may decrease in the third trimester (3). Increases in GFR have also been noted in women with mild renal dysfunction due to kidney disease, and/or renal transplantation, although increments are more modest (4), the magnitude of the increases usually related to the baseline GFR. The increment of renal hemodynamics in renal transplant recipients, however, is surprising since the increase of GFR occurs in a single kidney that would have been expected to have undergone compensatory hypertrophy with each nephron already hyperfiltrating to its limit (4).

The effect of pregnancy on long-term allograft survival and renal function is discussed controversely (5,6), but the majority of studies described a favorable maternal outcome (7–11). Good renal function and absence or a well-controlled hypertension are the most important prognostic factors for a favorable outcome (12,13).

Most women in the above described studies received azothioprine (AZA) as their major immunosuppressant, while more recently calcineurin inhibitors (CNIs) have been used. The overall long-term outcome of pregnancy in patients treated with CNIs seems to be similar (10–13), although women receiving CNIs may have worse renal function and a higher incidence of hypertension. It has been suggested, that these, among other factors may contribute to prematurity, fetal growth retardation and lower birthweights frequently observed in CNI-treated women (2,6,10–13). Furthermore, there are few reports (1,4,6) on the natural course of graft function during pregnancy, especially with respect to the different immunosuppressive regimens. The introduction of cyclosporine A (CYA) has improved graft and patient survival (14), however, nephrotoxicity limits its use. CYA-associated effects on kidney function can be divided into dose-related acute nephrotoxicity mainly due to vasocontriction and chronic changes leading to interstitial fibrosis with irreversible impairment of allograft function (15–17). As reported previously, CYA-treated subjects have a higher serum creatinine concentration at onset of pregnancy, as compared to AZA-treated patients (1). Thus, it is important to investigate the long-term consequences of pregnancy under different immunosuppressive regimens.

An area of confusion is how long to wait after transplantation before attempting conception. Many text recommend 2 years, but a recent consensus group at the American Transplant Society (18) has reduced this to 12 months, but data to support either policy are sparse.

We, therefore, designed a case-control study seeking data from a Transplant Registry shared by five centers and addressing the following questions. What are the long-term effects of pregnancy on allograft function? Does it differ if AZA or CYA are used as the primary immunosuppressant, and do short intervals between transplantation and pregnancy adversely affect graft survival?

Patients and Methods


A registry serving five German transplant centers, including Erlangen-Nuremberg, Muenster, Berlin-Steglitz, Berlin-Charité and Munich-Grosshadern provided the data analyzed in this study. This registry was created to follow the course of all their renal transplant recipients, who became pregnant after successful renal transplantation. All pregnancies reported to this registry were enrolled into this study. Each subject gave informed consent to participate in the registry. All data were collected and centrally analyzed at the University of Erlangen/Nuremberg and the Technical University of Munich.

Female renal transplant recipients of childbearing age who did not become pregnant served as controls. Each pregnant women was matched to one of these control allograft recipients for age, age at transplantation (±2 years), diabetes mellitus and immunosuppressive therapy (cyclosporine vs. AZA). To avoid the effects of the era when the transplant occurred, as well as differences in center management protocols, the patients and controls were further matched to identical transplant centers with pregnant patients from the same transplant center who had received their respective allografts during the same year. Other criteria for matching included renal function (similar preconception serum creatinine levels ±0.3 mg/dL) transplantation to delivery interval was used to define a 'matching day' and to calculate a 'gestational period' in control subjects. For subsequent comparisons between both groups, the 'matching day' in controls was used instead of the delivery day in pregnant renal allograft recipients. For women with multiple pregnancies, only the initial posttransplant pregnancy was matched.


Data were collected from the patient's records. Serum creatinine concentration as an indicator for graft function had been measured every 4 weeks before and after pregnancy and weekly during pregnancy. None of the patients were taking drugs known to affect cyclosporine metabolism. Adjustments of cyclosporine dose were based on wholeblood trough concentrations, determined at each visit before, during and after pregnancy, by commercially available monoclonal antibody test systems, the same assay used for a given individual. To detect rejection episodes, a close clinical and laboratory monitoring was performed during pregnancy. During the postgestational time, clinical examination, blood pressure measurements and laboratory parameters were determined at least monthly. Graft rejection episodes were diagnosed on clinical grounds and confirmed by renal biopsy if possible. Preconceptional, gestational and follow-up data were available for all pregnant women and for the matched periods in controls.


Data are expressed as mean ± standard deviation (SD). The significance of differences between pairs of observations was assessed by Student's t-test in normally distributed values. Graft and patient survival were analyzed by Kaplan–Meier analysis. Log-rank tests were used to describe and compare the survival rates of grafts and patients. Values of p < 0.05 were considered as statistically significant.


Eighty-one successfully transplanted patients who subsequently conceived were identified and matched to a similar number of controls who had not become pregnant. All were Caucasian, their clinical characteristics are summarized in Table 1, where one notes that baseline characteristics were well matched (Table 1). Seventy-six (93.8%) had received their first transplant, nine (11.1%) of which were from a living donor. Eight (9.9%) recipients were diabetic.

Table 1. Patient characteristics
 Study group (n = 81)Control group (n = 81)
  1. 1Not significant versus study group.

Age at transplantation (years)25.9 ± 0.725.6 ± 0.61
Age at delivery/matching (years)29.0 ± 0.528.7 ± 0.71
Time between delivery/matching and transplantation (months)41.8 ± 3.241.8 ± 3.21
Follow-up after delivery/matching (months)49.2 ± 4.251.2 ± 3.01
Observation interval to transplant (months)91.3 ± 4.988.3 ± 4.51

Complete data were available in 81 pregnant transplant patients of whom 72 were primigravidae. Three women gave birth to twins. Forty-one of the pregnant subjects were treated with AZA (mean dose: 93 mg/day, range: 25–175 mg/day), and prednisolone (mean dose: 6.4 mg/day, range: 4–16 mg/day) with no significant changes of the AZA dose during the course of the pregnancies. Forty of the pregnant allograft recipients had been receiving CYA, their preconception regimens averaging 240 ± 14 mg/day.

The matched control group consisted of 40 patients with a CYA-based therapy (mean: 212 ± 11 mg/day), and 41 patients with AZA-based therapy (mean dose: 88 mg/day; range: 50–175 mg/day) and prednisolone (mean dose: 5.5 mg/day; range: 4–16 mg/day). No significant change of CYA and AZA dose was observed during the period used for comparison with the pregnant allograft recipients. Eight controls were also diabetic.

Patient survival

Five women died after a mean interval from delivery of 61.1 ± 22.6 months (range: 10–136) compared to 6 in the control group (matched follow-up 58.3 ± 17.3 months; range: 5–100). Kaplan–Meier analysis revealed no statistical differences between groups (Figure 1). Ten-year patient survival after delivery was 93.4%, and 88.7% for the controls. All patients died with functioning allografts, only two patients within 2 years after delivery, the overall patient survival was similar between CYA- and AZA-treated patients. A causal association between death and the pregnancy and/or delivery could not be identified in these patients: One patient died due to cancer of the bladder 10 months after delivery, while the other died due to sepsis 21 months postpartum. There were two early deaths in the control group as well (myocardial infarction, sepsis).

Figure 1.

Postpartum patient survival in study subjects (solid line) and controls (dashed line; n.s.).

Baseline characteristics of those patients with fatal outcomes were very similar in both groups, and there were no significant differences with respect to baseline characteristics between the 11 patients who died and the 151 survivors.

Graft survival

Graft loss (including death with functioning graft) occurred in 12 of 81 study subjects and in 10 of 81 controls (n.s.). Ten-year actual graft survival was 62.5% after delivery and 67.0% in the control group (Figure 2). When censoring for death with functioning graft, seven patients lost their graft after pregnancy compared to four in the control group (n.s.). Graft loss in both groups was due primarily to chronic allograft failure, no patient losing the graft because of acute rejection. Three graft losses were within 2 years of delivery, all in patients receiving CYA. In the control arm, two graft losses occurred within 2 years after matching.

Figure 2.

Postpartum graft survival in study subjects (solid line) and controls (dashed line; n.s.).

Mean age at transplantation in subjects who had been pregnant with graft loss was 23.4 ± 1.7 years (range: 17.6–29.7 years) versus 24 ± 0.9 years (range: 19.1–30.4 years) in control subjects (n.s.). Mean age at delivery was 25.9 ± 1.4 years. The grafts failed after 99.5 ± 18.5 months posttransplant in study subjects (range 27.8–137.5 months) versus 89.3 ± 14.3 months (range: 25.4–128. months) in control subjects, respectively (n.s.). Graft loss in study subjects was seen after 51.5 ± 14.7 months (range: 1–102 months) postpartum, four of the study patients having been treated with AZA and three with CYA, while two of the controls losing grafts were treated with AZA and two with CYA. There was no difference between CYA-and AZA-treated patients with respect to graft survival. In addition, there were no significant differences in baseline characteristics for mode of immunosuppression between patients with pregnancies and controls who lost grafts and those who did not.

Graft function related to pregnancy

Graft function was stable before pregnancy, indicated by a mean creatinine level of 1.30 ± 0.07 mg/dL at conception. During the first trimester, serum creatinine levels decreased to 1.08 ± 0.05 mg/dL (p < 0.01), but then rose during the last trimester (1.46 ± 0.09 mg/dL; p < 0.05 compared to baseline).

During puerperium, mean serum creatinine level decreased consecutively to a level of 1.35 ± 0.09 mg/dL by 20 weeks postpartum, similar to values before pregnancy (n.s.). Serum creatinine levels of controls were lower at baseline, and remained stable over the next 18 months (1.06 ± 0.04–1.13 ± 0.05 mg/dL; n.s.; Figure 3).

Figure 3.

Time course of serum creatinine before, during and after pregnancy of study subjects (n = 81; fat line; p < 0.05 gestational versus pre and postgestational data) and time course of serum creatinine of the matched controls (n = 81; thin line; n.s.).

Graft function related to immunosuppressive therapy

The preconception baseline creatinine level was significantly lower in AZA-treated women, as compared to CYA-treated subjects (p < 0.05; Figure 4). Those receiving immunosuppression with AZA demonstrated a decrease from a pregestational level of 1.10 ± 0.1 to 0.84 ± 0.05 mg/dL (p < 0.001) at 20th week's gestation. During the last trimester, serum creatinine levels then increased to baseline levels (1.15 ± 0.17 mg/dL (Figure 4; p < 0.05 vs. baseline level). The mean AZA dosage of 93 mg/day remained unchanged during pregnancy as did prednisolone doses (mean: 6.4 mg/day). Similar to AZA-treated patients, all recipients who received CYA-based immunosuppression experienced a decrease in serum creatinine from a preconception level of 1.46 ± 0.06 to 1.21 ± 0.06 mg/dL (p < 0.05) at 20th week's gestation, followed by a return to baseline values at the end of the pregnancy. Serum creatinine concentrations in the immediate puerperium rose to a maximum of 1.61 ± 0.11 mg/dL 1 week postpartum (p < 0.05 vs. baseline values, and were decreasing in the following 6 months back to baseline values (1.48 ± 0.1 mg/dL; n.s. compared to baseline; Figure 4).

Figure 4.

Time course of serum creatinine in the study subjects treated with CYA (n = 41, fat line) and AZA (n = 40; thin line) before (−60 to −40 weeks), during (−40 to 0 weeks) and after (0–30 weeks) delivery.

Graft function after early onset of pregnancy

A total of 28 subjects (32.6%) were delivered within 2 years after transplantation and 6 women (7.9%) within 1 year after transplantation, respectively. Age at transplantation was significantly higher in subjects who became pregnant within 2 years after transplantation as compared to others (27.8 ± 0.75 years vs. 24.16 ± 0.72 years; p < 0.05). We did not observe any adverse effect of early pregnancy (<2 years posttransplant) on long-term outcome in our patients. Patient and graft survival as well as renal function were similar to those women with late pregnancy (>2 years posttransplant). Furthermore, changes in serum creatinine during gestation were similar between women with early versus late delivery after transplantation. Even the six women who delivered during the first year after transplantation had similar renal function and long-term outcome. Acute rejection episodes, deterioration of graft function and serious complications were not observed in these subjects.

Interaction between pregnancy and cyclosporine

Mean administered CYA dose increased gradually from 240 ± 14 mg/day before pregnancy to 324 ± 21 mg/day (p < 0.001) at the time of delivery (Figure 5). The increase in dose was most prominent after the 20th gestational week, but despite this mean CYA levels decreased from 115.9 ± 8 ng/mL before conception to a minimum of 80.7 ± 6.9 ng/mL at 20th week's gestation. With increased efforts to reach the therapeutic window, the trough level then slowly increased in late pregnancy reaching a level of 148 ± 5.1 ng/mL at delivery, and immediately postpartum there was a profound increase of CYA trough concentrations reaching 173 ± 5.4 ng/mL 1 week after delivery (p < 0.001; Figure 6). Over the following 10 weeks, CYA trough levels rapidly declined reaching baseline values again after 3 months. Subsequently, the CYA dose was tapered rapidly over the first 8 weeks postpartum and then gradually declined to the preconception level of 246 ± 15 mg/day (32 weeks postpartum; p < 0.05; Figure 5). Prednisolone dose was not altered in CYA-treated patients (6.4 ± 3.1 mg/day).

Figure 5.

Time course of CYA dose (mg/day) before (−60 to −40 weeks), during (−40 to 0 weeks) and after delivery in 41 study subjects treated with CYA (0–30 weeks; p < 0.05 gestational versus pre and postgestational CYA dose).

Figure 6.

Time course of CYA whole blood drug concentration (ng/mL) before (−60 to −40 weeks), during (−40 to 0 weeks) and after delivery in 41 study subjects treated with CYA (0–30 weeks; p < 0.05 gestational versus pre and postgestational drug concentration).


The number of transplant recipients who have had successful pregnancies and even some repeating such successes (19) has increased substantially during the past two decades, primarily due to the improved success rates after organ transplantation (2). However, the effect of pregnancy on long-term maternal and graft prognosis is still controversial as is the effect of different immunosuppressive regimens on renal transplant function in these patients. This reflects the paucity of data, especially long-term outcomes from large groups of patients making it difficult to perform adequate assessments of how pregnancy affects patient survival. Our data, a moderately large German case-control study strongly suggests that pregnancies in renal recipients do not negatively affect long-term patient and graft survival, independent of the immunosuppressive regimen. We also found no detrimental effect of transplant to pregnancy intervals on long-term graft function. We did not investigate additional maternal and fetal complications, such as preeclampsia, fetal growth retardation and prematurity, as these factors were investigated in several studies before (10–13). Close clinical monitoring enabled us to show the natural course of graft function under different immunosuppressive treatment regimens, including AZA and CYA. Furthermore, our data reveal important effects of the interaction of pregnancy and cyclosporine metabolism, which should improve the management of these patients.

Studies in renal allograft recipients on the long-term consequences of pregnancy on renal function and graft survival come to different conclusions (5,6,10–13,20). While most studies report no deleterious effect of pregnancy on maternal outcome, a single study suggested an adverse effect of pregnancy on graft function (5). But as the control group had an outstanding good 10-year graft survival of 100% in this center, it is questionable whether the data are generalizable to other centers (6). The National Transplantation Pregnancy Registry (NTPR) has noted a 7.5% graft loss rate at 2 years in almost 200 cyclosporine-treated who had pregnancies (21), data that also suggest that graft function is not adversely affected by pregnancy. Similarly the European Dialysis and Transplant Association registry published a case-controlled study in AZA-treated patients, who did not show any negative impact of pregnancy on long-term outcome (22). In another study out of the NTPR registry, CYA treatment was not associated with graft failure in multivariate analysis, instead graft dysfunction and rejection were significant predictors of graft failure (23). Unfortunately, the NTPR registry does not compare the outcome to patients without pregnancies, thus analysis from this large database is limited. In order to specifically assess the effect of pregnancy on long-term outcome, it is important to compare pregnant renal allograft recipients with a well-matched control group, because renal transplant patients becoming pregnant will tend to be those with a good physical condition accompanied with a good graft function. Therefore, we initiated a case-control study with a well-matched control group and could not detect statistical differences between different immunosuppressive regimens with and without CYA.

This observation extends the results obtained from two recent large series on the consequences of pregnancy on graft function (12,13). An accurately cataloged series of 48 pregnancies in 24 women (12) did not observe major differences of immunosuppressive regimens (with and without CNI) on long-term maternal outcome. More importantly, the authors identified levels of preconception renal dysfunction, which may be predictive of permanent loss of function postdelivery. Another large study from the UK, although not yet completely published, analyzed 193 pregnancies in 176 women (13). Using logistic regression, the authors confirmed previous reports (2,6,12,21) on the importance of preexisting hypertension and renal dysfunction for preterm birth. Additionally, a matched case-control set was employed for the analysis of pregnancy on maternal outcome. Preliminary results from this study suggest that pregnancy has no significant effect on long-term graft function (13). While most authors address the effect of CNIs on outcome (10–13,21), the effects of AZA on outcome are less clear. AZA has less immunosuppressive strength, however no obvious effects on renal function and hypertension. Because most authors observe similar outcomes (6,10,13), an independent effect of AZA on outcome seems unlikely, and due to fear of rejection most physicians do not consider a conversion of immunosuppressive regimen. In summary, the majority of studies concluded that graft function is usually not adversely affected by pregnancy in women with good renal function treated either with AZA or CNI. Based on the results of the large NPTR database, it has been recommended that recipients who want to become pregnant should wait for 2 years after transplantation, as shorter transplant intervals had less favorable neonatal outcomes (4,24,25). However, these differences were only marginal, and no further reports on the long-term maternal outcome with respect to the time interval between pregnancy and gestation exist. Due to the high incidence of pregnancies within the first 2 years after renal transplantation (33%), we performed a subgroup analysis in our population. The present data demonstrate that graft and patient survival were not adversely affected in these women. Thus, based on our results a reflection of the recommendations may appropriate, although our study did not focus on maternal or fetal complications. Most patients achieve stable renal function within 6 months after transplantation and immunosuppressive therapy is tapered to long-term maintenance therapy after 6–12 months in most centers, as rejection rates are very low beyond this timepoint. As preconception counseling should always be cautious, based on our data a period of 1 year after transplantation without evidence of rejection seems to be sufficient. However, good renal function and a well-controlled blood pressure have to be taken into account during preconception counseling, because they are the most important prognostic factors (12,13).

Normal pregnancy is accompanied by marked increases in GFR that may reach levels 50% above preconception measurements (4). As in healthy pregnant women, GFR increased in the pregnant renal recipients, associated with the potential for glomerular damage due to glomerular hyperfiltration (26). During the third trimester, serum creatinine as a marker of GFR returned to pregestational levels in pregnant renal allograft recipients. Compared to healthy pregnant subjects, this maternal adaptation in GFR was similar.

Similar to AZA-treated patients and healthy gravidas, CYA-treated pregnant women exhibit a temporal hyperfiltration until the 20th gestational week followed by slow normalization of serum creatinine until delivery. In contrast to normal gravidas and AZA-treated patients, a significant proportion of CYA-treated renal transplant recipients (63%) developed a significant deterioration of their renal function immediately after delivery. Fortunately, serum creatinine of our study subjects returned back to pregestational levels in the puerperium, suggesting acute but reversible effects on kidney function. As we observed a simultaneous increase of serum creatinine and CYA trough levels, it is suggestive that the deteriorated renal function is a direct consequence of toxic CYA concentrations. Other factors such as hypovolemia, surgery and antibiotics might further contribute to the decreased renal function in these patients, as it is known that CYA treatment renders the allograft more susceptible to such injuries. As a consequence, close clinical monitoring and more rapid CYA dose adjustments are needed in the postgestational care of CYA-treated renal allograft recipients.

A reduction of CYA whole blood drug concentration was seen in all study subjects, starting during the first trimester with a nadir in the 16th week's gestation. This phenomenon was previously described for small study cohorts only (23,27), others did not observe any clear impact on CYA dose and trough levels, probably because of less thorough monitoring and low patient numbers (28–30). Similar effects were seen for tacrolimus (31), another CNI, for which favorable pregnancy outcomes have been described (32). Several effects might contribute to the decrease in CYA trough concentrations (33). Pregnancy is associated with changes in total blood volume and alterations in gut mobility. Both might affect drug absorption, disposition and drug concentrations in the blood. Additionally, enhanced drug metabolism, caused by hormonal changes during gestation (34), and CYA metabolism of the fetus (35,36) might be responsible for the decrease of CYA concentrations, especially during the last trimester. As a consequence of lower trough levels, CYA dose had to be increased in all patients to maintain therapeutic drug levels. A recent review of the large American database revealed worse outcome, when CYA was decreased or discontinued during pregnancy, providing evidence to maintain adequate CYA dosages (37). We did not observe any rejections during gestation, suggesting that the achieved drug concentrations in our patients were sufficient. Because the increased CYA dose caused toxic drug concentrations immediately after delivery, we recommend more rapid and more profound dose adjustments in the early postgestational period. Absorption monitoring using CYA drug concentrations 2 h after dosing (38) might further improve the monitoring of these critical patients and lead to a lower frequency of nephrotoxic side effects in CYA-treated allograft recipients.

In summary, our data endorse the best of earlier studies that pregnancy does not appear to compromise long-term maternal and renal prognosis. Both immunosuppressive regimens yielded good results during and after pregnancy. The encouraging results of early pregnancy after renal transplantation may lead to favor shorter intervals in preconception counseling, if other more important prognostic factors such as renal function and hypertension are taken into account. Our study shows the need for close drug monitoring during pregnancy and most importantly during the early postgestational period.


We appreciate the support and contribution provided by the following: R. Bollmann, Department of Gynecology and Obstetrics, University Berlin/Charite, Germany; G. Hillebrandt, Department of Nephrology, University of Munich/Grosshadern, Germany; W. Holzgreve, Department of Gynecology and Obstetrics, University of Basel, Switzerland; H. Hopp, Department of Gynecology and Obstetrics, University Berlin/Steglitz, Germany; H. J. Voigt, Department Gynecology and Obstetrics, Hospital Kaiserslautern, Germany. Special thanks to Professor Marshall D. Lindheimer, University of Chicago, for the review of this paper.