Long-term Graft Survival and Graft Function Following Pregnancy in Kidney Transplant Recipients: A Systematic Review and Meta-analysis.

BACKGROUND: The incidence of pregnancy in kidney transplantation (KT) recipients is increasing. Studies report that the incidence of graft loss (GL) during pregnancy is low, but less data are available on long-term effects of pregnancy on the graft.
METHODS: Therefore, we performed a meta-analysis and systematic review on GL and graft function, measured by serum creatinine (SCr), after pregnancy in KT recipients, stratified in years postpartum. Furthermore, we included studies of nulliparous KT recipients.
RESULTS: Our search yielded 38 studies on GL and 18 studies on SCr. The pooled incidence of GL was 9.4% within 2 years after pregnancy, 9.2% within 2-5 years, 22.3% within 5-10 years, and 38.5% >10 years postpartum. In addition, our data show that, in case of graft survival, SCr remains stable over the years. Only within 2 years postpartum, Δ SCr was marginally higher (0.18 mg/dL, 95%CI [0.05-0.32], P = 0.01). Furthermore, no differences in GL were observed in 10 studies comparing GL after pregnancy with nulliparous controls. Systematic review of the literature showed that mainly prepregnancy proteinuria, hypertension, and high SCr are risk factors for GL.
CONCLUSIONS: Overall, these data show that pregnancy after KT has no effect on long-term graft survival and only a possible effect on graft function within 2 years postpartum. This might be due to publication bias. No significant differences were observed between pre- and postpartum SCr at longer follow-up intervals.

INTRODUCTION

With increasing numbers of kidney transplantation (KT) performed worldwide and good short-term pregnancy as well as graft outcomes, there is an increasing incidence of pregnancy in KT patients. In 2011, over 11 000 births after KT have been reported worldwide.[1] The Transplant Pregnancy Registry International (TPR) reported, in 2018, a total of 1993 pregnancies in 1101 KT recipients in the United States.[2] Pregnancy in KT recipients is labeled as high risk with increased fetal and maternal risks for adverse pregnancy outcome. Reported live birth rates after KT are consistently between 72% and 80%.[1,3,4] Compared to the US general population, pregnancies after KT are associated with higher rates of cesarean sections (56.9% versus 31.9%), preterm (<37 wk of gestation), deliveries (45.6% versus 12.5%) and increased rates of small for gestational age and low birth weight (mean birth weight 2420 g versus 3298 g).[1,4] In addition, the pregnancies are reported to have high maternal complication rates of hypertension and proteinuria[5]: with an increased risk to develop pre-eclampsia 27% versus 3%. In previous meta-analysis, 4.2% of recipients experienced an episode of acute rejection during their pregnancy.[1]

Besides the pregnancy-related complications mentioned above, little is known on what effect pregnancy has on long-term graft survival and graft function. At the time of KT, the transplanted kidney develops compensatory renal hypertrophy, which results in hyperfiltration.[6] During pregnancy, physiological changes occur in the kidney and cardiovascular system, including vasodilatation and increase in glomerular filtration rate (GFR).[7] This increased pressure and/or plasma flow during pregnancy on top of the already existing hyperfiltration may cause progressive loss of graft function due to glomerular sclerosis.[6] It is unknown which effect this temporary extra demand has on the long-term graft survival and graft function. These insights would be helpful in preconceptional counseling of KT patients.

A meta-analysis of KT recipients published in 2011 analyzed graft loss (GL) incidence in a small number of retrospective studies, reporting 8% postpregnancy GL at 2 years, 7% at 5 years, and 19% at 10 years.[1] Limited studies reviewed a year later showed no significant increase in serum creatinine (SCr) at 3 months and GL at 2 years postpartum.[8] No reviews analyzed the effect on long-term consequence of pregnancy on graft function (SCr).

A limitation of previous meta-analysis and reviews is that they did not include a control group of nulliparous KT recipients.[1,8] Furthermore, they do not systematically report on predictive factors regarding long-term graft function after pregnancy.

Currently, optimal timings of pregnancy after KT are described as follows: an interval of >1 year between KT and pregnancy, and an interval of >1 year between the last episode of acute rejection. Furthermore, SCr levels should be below 1.5 mg/dL, no acute infections should be present, and stable maintenance of nonteratogenic immunosuppressive medication.[9,10] However, the aforementioned guidelines are based mainly on data from voluntary registries and expert opinions, focusing primarily on (predictors of) adverse pregnancy outcomes.

To increase insight in the effect of pregnancy on long-term graft survival and function as guidance for preconceptional counseling: the aim of this study was to perform an updated meta-analysis on graft survival with comparison with nonpregnant KT recipients and for the first time long-term (up to 10 y) graft function (SCr) after pregnancy. We included new studies since 2010 and studies with nulliparous KT recipient control groups. In addition, systematic review was performed to give an overview of predictors for adverse long-term graft outcomes after pregnancy.

MATERIALS AND METHODS

Search Strategy and Study Selection

A systematic search of literature was performed in Pubmed, Embase, and Cochrane library to identify all studies on SCr and GL after pregnancy in KT recipients until September 2018 (Appendix 1, SDC, http://links.lww.com/TP/B836). Two reviewers (A.S. and N.P.) independently screened the abstracts of all eligible studies. Studies reported in English, focusing on SCr or GL following pregnancy in KT recipients were eligible. Furthermore, we conducted snowballing strategy to include eligible reports. Case studies, reviews, and studies that reported <6 months postpregnancy follow-up were excluded.

Data Extraction

Two independent reviewers (A.S. and M.B.) extracted data from all eligible studies. The following data were extracted from each study: study outcomes on prepregnancy SCr, postpregnancy SCr, and GL or graft survival. For all the included studies, data on pre- and postpregnancy SCr were extracted or calculated in mean ± SD (in mg/dL). When median with range were reported, the mean ± SD were calculated by the method of Hozo et al.[11] SCr levels that were reported in µmol/L were converted into mg/dL. If graft survival was reported, this was converted to GL. In 11 studies, there were missing or incomplete data, this was requested from the authors and in 3 cases we gained enough information to include them in our meta-analysis. Using the observational cohort studies with a control group, it was examined whether pregnancy affects GL or SCr, versus nulliparous KT recipients. In addition, all included studies were reviewed for different predictors of adverse graft outcomes (eg, hypertension, proteinuria, SCr before pregnancy, and transplant to conception interval).

Pooled Estimates

To pool data on GL and postpregnancy SCr, subcategories were created based on the number of years postpartum. Articles on GL were divided into 4 categories based on timing since pregnancy: GL within 2 year postpregnancy, 2–5 year, 5–10 year postpregnancy, and >10 year postpregnancy. Data on SCr postpregnancy were divided into 3 subcategories: within 2 year postpregnancy, 2–5 years, and 5–10 years postpregnancy. The difference between postpregnancy SCr and prepregnancy SCr (Δ SCr) was calculated. For binary outcomes (GL), pooled estimates and 95% confidence intervals were calculated using Excel.[12] For continuous outcomes (pre- and postpregnancy SCr), pooled estimates and 95% confidence intervals were calculated using mean difference and random effect size, conducted by Review Manager 5.3.[13]

Quality Assessment and Assessment of Publication Bias

Two reviewers (A.S. and M.B.) screened the studies for full text and performed a critical appraisal on applicability and validity (Appendix 2, SDC, http://links.lww.com/TP/B836). Every study was scored for design, size, domain, determinant, outcome, missing data, lost to follow up, standardization of outcome, analysis, confounding factors, and the possibility to extract data. To test for publication bias, we performed funnel plot analysis for every subtopic within GL and SCr.

RESULTS

As a result of the search from 3 electronic databases, 1416 studies qualified for abstract screening. Among these, 43 individual publications were selected for inclusion of which 38 studies reported on GL and 18 articles on SCr postpregnancy (Figure 1). One study by Levidiotis[14] divided graft survival in different periods of time that is why we could only use the data of the subanalysis of the matched cohort. Ten of these were observational cohort studies with a control group,[3,14-22] Table 1 presents the study characteristics and reported graft outcomes for all of the included studies.

FIGURE 1.

Study selection for studies reporting postpregnancy graft function and/or postpregnancy graft loss.

TABLE 1.

Study characteristics and outcomes

Pooled Incidence of GL After Pregnancy in KT Recipients

A total of 38 studies reported on GL in 2453 recipients. Median follow-up time was very heterogenic among the studies and varied from 6 months until 15 years after pregnancy GL occurred in 321 (13%) patients following pregnancy. The risk on GL is increasing in time with pooled incidences of 9.4%, 9.2%, 22.3%, and 38.5% for <2 year, 2–5 year, 5–10 year, and >10 years after pregnancy, respectively (Figure 2A–D).

FIGURE 2.

A–D, Pooled incidence of postpregnancy graft loss. A, Graft loss within 2 y postpregnancy: 9.4%, n = 1347 (range 10–1100), total graft loss n = 126 (range 0–111). B, Graft loss 2–5 y postpregnancy: 9.2%, n = 600 (range 8–139), total graft loss n = 55 (range 1–8). C, Graft loss 5–10 y postpregnancy: 22.3%, n = 395 (range 12–81), total graft loss n = 88 (range 0–18). D, Graft loss >10 y postpregnancy: 38.5%, 234 (range 18–118), total graft loss n = 90 (range 1–51).

Among the 10 studies with a nulliparous control groups, matching criteria differed as shown in Table 2. The median follow-up time of these studies was 100 months (range 45–168) after pregnancy.

TABLE 2.

Characteristics matched cohort studies on graft loss

Pooled Incidence of SCr After Pregnancy in KT Recipients

The postpregnancy data of 18 individual studies on SCr within women were pooled in 3 postpartum time intervals. Fourteen studies reported on 1-year postpregnancy SCr in KT recipients. A pooled increase in SCr is seen of 0.18 mg/dL, 95% CI [0.05-0.32], P = 0.01 in the group comparing prepregnancy SCr within 2-year postpregnancy SCr. (Figure 3A) Four studies reported on SCr 2–5 years following pregnancy, and only 2 studies on long-term (5–10 y) postpregnancy SCr and no significant differences were found when comparing pre- versus postpregnancy SCr (Figure 3B and C).

FIGURE 3.

A–C, Pooled difference (mean difference [95% CI] in prepregnancy SCr and postpregnancy SCR (Δ SCr pre- and postpregnancy). A, Delta SCr within 2 y postpregnancy: SCr 0.18 mg/dL [0.05, 0.32], P = 0.007, n = 441. B, Δ SCr 2–5 y postpregnancy: SCr 0.17 mg/dL [–0.03, 0.37], P = 0.09, n = 175. C, Δ SCr 5–10 y postpregnancy: SCr 0.10 mg/dL [–0.12, 0.32], P = 0.38, n = 101. CI, confidence interval; SCr, serum creatinine; SD, standard deviation.

Predictors of Adverse Outcomes on Graft Function and Risk of GL

Among the included studies,[3,4,16,18,19,22,24,26,31,37,41-43,47,48,50,54] different predictors of adverse outcomes on graft function were described, including hypertension before pregnancy, presence of proteinuria before pregnancy, pre-eclampsia, SCr before pregnancy, and transplant to conception interval (TCI). An overview of the literature on these risk factors is given and described in more detail below (Table 3). In addition to these most reported risk factors, some incidental risk factors were reported. Type of delivery or type of donor was no significant risk factor for GL.[23] High panel reactive antibody (PRA) levels and donor-specific anti-HLA antibodies (DSA) have a high risk of antibody-mediated rejection (AMR) and have more pre-eclampsia.[25] The type of immunosuppressive regime had no effect on graft survival.[38,48]

TABLE 3.

Predictors of graft loss or renal function deterioration after pregnancy

Preconceptional Hypertension as a Risk Factor for Accelerated GL

Four studies reported an effect of hypertension, before, or at the beginning of pregnancy, in relation to long-term graft function.[4,23,42,48] Hypertension was defined as bloodpressure >140/90 mm Hg. These 4 authors concluded that (drug treated) hypertension before pregnancy is associated with worse graft function or is a risk factor for graft function decline and/or chronic rejection. In one of these studies, postpregnancy graft function (SCr) was compared between patients with hypertension before pregnancy (n = 5) and no hypertension before pregnancy (n = 15). The SCr was significantly worse (P = 0.03) in patients with hypertension before pregnancy.[23] Another study showed that hypertensive patients (n = 28) compared to normotensive patients (n = 23) had worse graft function (SCr) before pregnancy (1.39 mg/dL versus 1.10 mg/dL), P ≤ 0.01.[47] Two recent studies of which one was a matched cohort study did not see a relation between graft failure and chronic hypertension.[22]

Preconceptional Proteinuria and Preconceptional SCr as Risk Factors for Accelerated GL

Proteinuria before or during pregnancy, especially proteinuria of >1 g/day, is associated with worse graft survival.[48,55] Two studies, which compared high levels of proteinuria (>0.3 or 0.5 g/d), found no deleterious effect on SCr or GL.[50,54]

Ten different studies analyzed the influence of prepregnancy SCr on graft outcomes (SCr postpregnancy and GL or graft survival). Multiple different cut-off values were described among these studies, ranging from SCr > 1.47 mg/dL to 2.26 mg/dL and >1.0 mg/dL to 2.1 mg/dL.[18,24,26,42,43,47,48,50,54] Ten studies found a negative effect on graft function in patients with high SCr before pregnancy. Eight of them used SCr > 1.47 mg/dL as cut off point,[18,24,26,31,43,47,48,50] 1 study defined worse graft function as >1.24 mg/dL, 2 of them used no cutoff point where one described a negative effect of worse graft function (OR 1.71, 95% CI [1.15-3.45], P = 0.04)[22] and one found no relationship between prepregnancy SCr and GL (OR –0.11, 95% CI [–0.44 to 0.23], P = 0.52). Two other studies used cut off points <2.26 mg/dL and <1.3 mg/dL also found no negative effect on postpregnancy graft function in women with high SCr before pregnancy[41,54] (Table 3).

Pre-eclampsia as a Risk Factor for Accelerated GL

The development of pre-eclampsia during pregnancy was mentioned by 1 study as factor for graft dysfunction during pregnancy.[24] Pre-eclampsia was defined as hypertension and proteinuria >0.30 g/24 h. One study showed that pre-eclampsia was a “borderline” risk factor for GL (OR 1.09; 95% CI [0.92-1.34], P = 0.09).[22] The latest matched cohort study did not see a relation between pre-eclampsia and GL.[51]

Transplant to Conception Interval as a Risk Factor for Accelerated GL

The relationship between transplant to conception interval (TCI) and graft function is reported by 5 individual studies, which report on different outcomes of TCI (in general, TCI < 1 y, TCI < 2 y, TCI > 5 y).[3,16,19,26,37] Stoumpos et al found no negative relationship between graft function and TCI.[3] One study found more GL in patients with TCI < 1 year,[26] whereas another study found no significant impact on graft outcome.[19] In another study, there was no adversely effect on graft survival in patients with TCI < 2 year, compared to other subgroups.[16] A TCI of >5 years has acceptable outcomes on postpregnancy graft function and rejection during pregnancy and up to 3 months postpartum.[37]

Assessment of Quality and Publication Bias

We assessed study quality with the use of critical appraisal on applicability and validity (Appendix 2, SDC, http://links.lww.com/TP/B836). Seventeen studies had a sample size of <20 women. Missing data were not well described in 21 of the studies. Ten studies did not describe their statistical analysis precisely. In 6 studies, possible confounding factors were not mentioned in the article. Publication bias for studies on GL is unlikely as GL funnel-plot shows symmetry (Appendix 3, SDC, http://links.lww.com/TP/B836). There is a funnel-plot asymmetry in the subgroup of Δ SCr <2 years after pregnancy indicating publication bias toward the publication of small studies with positive Δ SCr values (Appendix 4A, SDC, http://links.lww.com/TP/B836).

DISCUSSION

In this meta-analysis and systematic review, we aimed to investigate the effect of pregnancy on long-term graft survival and function as guidance for preconceptional counseling using data derived from 42 studies. This meta-analysis gives an update on GL after pregnancy after KT. It includes cohort studies with nulliparous control groups and pooled data on graft function after pregnancy after KT. We are the first to analyze pooled data on long-term SCr after pregnancy in KT recipients. GL and SCr after pregnancy in KT recipients are reassuring with no difference in GL when compared to nulliparous KT recipients and stable SCr up to 10 years postpartum. We only found a slight significant rise in SCr in the period within 2 years after delivery of 0.18 mg/dL of which it can be discussed if such a small increase is clinically relevant, especially since Δ SCr was not increased at later time points after pregnancy.

The present meta-analysis added >500 women from 23 additional studies to the literature since the last meta-analysis from 2011 on the subject.[1] We report slightly higher outcomes on GL within 2 years (9.4% versus 8%) and higher numbers of GL of 22.3% versus 19% after 5–10 years postpregnancy, this was mainly caused by the TPR report.[2] Desphande reported 12.5-year postpregnancy GL of 11%, based on one study of Gorgulu.[39] We could not include this study in our meta-analysis because they only reported on GL after KT and not on GL after pregnancy. Our outcome of GL of 38.5% >10-year postpartum is based on a pooled incidence of 5 new studies.[14,20-22,43]

We added 10 studies that compared the result of GL after KT with a nulliparous KT control group. The absence of a difference in GL between parous and nulliparous is reassuring. We ascertained that the control groups used were heterogenic among the studies: almost all studies were age and SCr before conception matched. The question remains whether the used control groups are really comparable because the reason they did not conceive might be the result of other underlying conditions, which also can influence SCr and GL.

This study provides us insight into incidence of GL per years postpartum. It would have been informative to perform the same analyses per years posttransplant. Unfortunately, posttransplant years were rarely reported, which restricted us from performing this analysis. Therefore, it is hard to compare our results with the GL numbers from the registries. When comparing GL results after pregnancy to the age group of 16–34 years (men and women) of the Eurotransplant region, the number of GL after pregnancy are lower than the number of GL after KT in the general KT population. Ten years GL after KT (living and postmortal donors) is 46% and 15 years GL is 60% for this age group in the Eurotransplant region.[56] This finding of relatively good graft survival in women with pregnancy after KT is reassuring. Although the argument that KT recipients with worse renal and physical condition are less likely to get pregnant also counts for this comparison.

In addition to previous meta-analysis,[1] we examined long-term graft function after pregnancy in KT recipients. A small significant rise in SCr within 2 years after delivery as described in 87 KT recipients derived from 3 studies.[30,43,44] Possibly, this might be caused by physiological changes after pregnancy or restart of medication such as ACE inhibitors. On the contrary, this could be the result of high rate of risk factors in the study population (65.9% hypertension, 36.5% SCr >1.5 mg/dL before pregnancy[43]), which makes these women more prone for deterioration of graft function or even GL. Most importantly, we do not find an increase in SCr during the period 5 years after pregnancy. However, women with a malfunctioning graft or lost to follow-up are not present in subgroups longer time after pregnancy possibly inducing bias. This is in line with the recent systematic review on the effect of pregnancy in chronic kidney disease, which reported no shift in CKD stage after pregnancy.[57]

Risk factors for GL after pregnancy in KT were hypertension, proteinuria, transplant to conception/delivery interval, and preconception graft function. However, only a few of the studies reporting on these risk factors performed a multivariate analysis, influenced by power. It is difficult to establish cause-relationship effects of risk factors. These risk factors are mentioned in the European and American guidelines, aiming at improving outcome in KT recipients.[9],[10] The TCI is a point of discussion as it was stated by the European guidelines for 2 years after KT. The American guidelines changed their advice to postpone pregnancy at least until 1 year after pregnancy. Studies such as Fischer and Pour Reza Gholi showed reassuring results of pregnancies after 1 year after pregnancy.[16,19] Pregnancy within 1 year after KT is associated with an increased risk on GL, which Rose et al showed in their recent study.[58] Data on the association of pre-eclampsia with GL show confliction results.[23,24,51]

The strength of our meta-analysis is that we pooled data on GL including studies with a nulliparous control group and for the first time examined pooled incidences of graft function after pregnancy. One of the limitations of this study is that the quality of some studies was poor with small sample size. The funnel plot analysis in the subgroup of Δ SCr 24 months after pregnancy showed an asymmetry, possibly publication bias is present (Appendix 4A, SDC, http://links.lww.com/TP/B836). In addition, this is an unadjusted meta-analysis in which we could not account for factors such as differences in healthcare systems or socio-economic status or difference in SCr measurements because of lack of such information.

Unfortunately, we were not able to perform a meta-analysis on estimated glomerular filtration rate (eGFR). Most studies only report SCr without age, calculation of eGFR was not possible.[59] We assumed that preconceptional creatinine was really preconceptional as it was stated in the text. It could be possible that the preconceptional SCr that was used for the included studies were not completely preconceptional and that the SCr was already physiologically increased. Ultimately, evaluation of individual slope of eGFR pre- and postpregnancy would be performed by means of a multilevel analysis to answer the question whether pregnancy has effect on longer-term GFR. Additionally, it would be possible to identify the most important predictors for worse graft outcomes after pregnancy after KT in relation to eGFR slope change.

In conclusion, this systematic review and meta-analysis showed a possible association with short-term SCr decline postpartum, but no association at longer periods of time after delivery. The incidence of GL up to 10 years postpregnancy is limited but data analyzed show reassuring data on GL with pregnancy after KT compared to nulliparous controls and age-matched and SCr-matched controls. This should be taken into consideration during preconceptional counseling. Based on the risk factors for GL, it could be concluded that if prepregnancy KT function is good, it remains good after pregnancy. Systematic review of the literature showed that mainly prepregnancy proteinuria, hypertension, and high SCr are risk factors for GL.

ACKNOWLEDGMENTS

We thank all members of the Dutch PARTOUT (Pregnancy After Renal Transplantation Outcomes) network for their support.