Unintended Pregnancy After Kidney Transplantation

In kidney transplant recipients, unintended pregnancy was not associated with acute peripartum kidney rejection but was associated with mycophenolate product exposure.

Abstract

OBJECTIVE:

To define risk factors, obstetric morbidity, and allograft outcomes associated with an unintended pregnancy after kidney transplant.

METHODS:

This is a retrospective cohort study of pregnancies in women after kidney transplantation enrolled in the Transplant Pregnancy Registry International with births between 1967 and 2019, with cohorts categorized as intended pregnancies and unintended pregnancies. The primary outcome was acute kidney rejection during pregnancy or by 6 weeks postpartum. Secondary outcomes included allograft loss, severe maternal morbidity, and neonatal composite morbidity. Multivariable logistic regression, Kaplan–Meier curves, and Cox proportional hazards regression models were performed, with adjustment for covariates pertinent to allograft function.

RESULTS:

Among 1,723 pregnancies of kidney transplant recipients, 1,081 (62.7%) were intended and 642 (37.3%) were unintended pregnancies. Risk factors for unintended pregnancy included younger age, Black race, nulliparity, chronic hypertension, and transplant from a deceased donor. Exposure to mycophenolate products (16.0% vs 5.7%) and termination (4.7% vs 0.4%) were more common in unintended pregnancies (P<.001). Unintended pregnancy was not associated with acute kidney rejection (2.3% vs 0.9%, adjusted odds ratio [AOR] 2.38, 95% CI, 0.91–6.30, P=.08). Unintended pregnancy was independently associated with allograft loss at 2 years from the end of pregnancy (8.1% vs 3.5%, AOR 2.27, 95% CI, 1.32–3.94, P=.003) but not allograft survival (adjusted hazard ratio 1.22, 95% CI, 1.00–1.49, P=.05). There were no differences in severe maternal morbidity (3.3% vs 3.6%) or neonatal composite morbidity (12.9% vs 14.3%) by pregnancy intention.

CONCLUSION:

Unintended pregnancy was not associated with acute kidney allograft rejection in the peripartum period, but it was associated with mycophenolate exposure and allograft loss at 2 years from pregnancy. The same social vulnerabilities that underlie difficulty in accessing reproductive care may be similarly important for transplant health. We recommend prepregnancy counseling and longitudinal follow-up of transplant recipients to reduce rates of unintended pregnancy and to optimize transplantation success over time.

Unintended pregnancy is defined as a pregnancy that is unwanted or mistimed. In the United States, approximately 45% of all pregnancies are unintended,¹ prompting the Department of Health and Human Services to identify reducing unintended pregnancies as a key goal in the Healthy People 2030 family planning initiative.² An unintended pregnancy can have significant medical and social repercussions for birthing people, including delays in prenatal care, substance use, depression and anxiety, relationship conflict, preterm birth, low birth weight, reduced educational attainment, and financial burden.^3,4 From a societal perspective, the medical costs of unintended pregnancies are estimated to be $4.6 billion annually.³

Pregnancies after kidney transplantation represent a high-risk obstetric population, with increased rates of preeclampsia, preterm birth, cesarean birth, and maternal and neonatal morbidity.^5–8 We hypothesize that an unintended pregnancy after kidney transplantation could have even more pronounced consequences than in the general population, with possible effects on maternal, neonatal, and graft outcomes. Therefore, the objective of this study was to understand risk factors, obstetric morbidity, and allograft consequences in the setting of unintended pregnancies in kidney transplant recipients.

METHODS

We conducted a retrospective cohort study of data abstracted from the Transplant Pregnancy Registry International, which includes births from solid organ transplant recipients between 1967 and 2019. This study follows the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for cohort studies. The Transplant Pregnancy Registry International is an IRB- (Advarra Pro00008001) approved international registry enrolling from a diverse set of 289 transplantation centers since 1991. Information on pregnancies before 1991 was collected retrospectively instead of prospectively. Since 1991, recipients have been followed up longitudinally throughout pregnancy and every other year postpartum until the time of death or loss to follow-up. Demographic and outcome information is collected by trained transplantation coordinators through telephone interviews and medical record review, when available, to reduce recall bias. Participants are recruited through multiple channels to reduce selection bias, including transplantation center communication, social media, transplantation advocacy groups, and self-referral. Race was self-identified by participants and grouped into the predefined categories of Asian, White, and Black or African American. Additional races and ethnicities included American Indian, Alaska Native, and mixed race. If participants did not answer the question about race and ethnicity, their race was designated as unknown. Self-identified race was included in our analysis given the association of Black race with kidney allograft loss to determine whether unintended pregnancy was independently associated with graft outcomes.⁹ Each pregnancy is treated as a separate encounter for the purposes of this study because recipients may have had more than one pregnancy after their transplantation. For recipients who had more than one transplantation, characteristics of the most recent allograft transplanted before pregnancy were used in the analysis. All data used in this study were individually reviewed and validated by the primary author. Additional information on the registry can be found in the most recent Transplant Pregnancy Registry International report.¹⁰

Included in this study were pregnancies in adult (more than 18 years old) kidney transplant recipients, with all outcomes of live births, stillbirths, miscarriages, terminations, and ectopic pregnancies analyzed. Cohorts were categorized as intended pregnancy or unintended pregnancy. The participants were specifically asked the question, “Was this pregnancy planned?” We excluded pregnancies for which participants did not answer the prior question, as well as individuals with multiorgan transplantations. We did not restrict our study size and included all participants meeting eligibility criteria. We did not analyze contraceptive data for this study because this was beyond the scope of our study question.

The primary outcome of this study was acute kidney rejection, defined as rejection of the allograft during pregnancy or within 6 weeks postpartum. All acute rejection episodes were confirmed by kidney biopsy and, starting in the mid-1990s, classified by Banff histologic criteria.¹¹ Secondary outcomes included allograft loss, severe maternal morbidity, and neonatal composite morbidity. Allograft loss was defined as the need for dialysis or repeat transplantation. We studied both allograft loss within 2 years of end of pregnancy and allograft loss up to the last date of follow-up. Immunosuppressant change was defined as alteration in immunosuppressant regimen within 6 months of pregnancy or during the pregnancy. Severe maternal morbidity was defined as one or more of the 21 Centers for Disease Control and Prevention indicators occurring during labor or the postpartum period,¹² and neonatal composite morbidity was defined as one or more of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units adverse outcomes (Appendix 1, available online at http://links.lww.com/AOG/D873).^13,14

Statistical analysis was conducted with R Studio 2023.09. Univariate categorical variables were analyzed with χ² tests, and continuous variables were analyzed with the Wilcoxon rank-sum test, with reporting of medians and interquartile range. Missing data were removed from the analysis and not imputed. Any categories with missing data of more than 10% of either cohort are reported in the table footnotes, along with any difference in proportion of missingness. Multivariable logistic regression was used to evaluate independent risk factors for acute kidney rejection and allograft loss at 2 years from the end of pregnancy. We used an established stepwise selection process to arrive at the final model that has the least number of independent variables that best predicted the outcome of interest.¹⁵ Goodness of fit was evaluated by the Hosmer–Lemeshow test and by graphic evaluation of model residuals. Covariates chosen for adjustment were risk factors for kidney rejection and allograft loss,¹⁶ as well as year of pregnancy and mycophenolate product use as a proxy for medication change in pregnancy. Unadjusted death-censored and follow-up time–censored allograft survival was assessed with Kaplan–Meier analyses with the log-rank test. Cox proportional hazards regression models were then constructed for multivariable time-to-event analyses for allograft loss, with adjustment for covariates associated with allograft function as in the multivariable models.

We conducted sensitivity analyses for the prespecified subgroups of continuing pregnancies (live births, stillbirths, and miscarriages, excluding terminations and ectopic pregnancies) and contemporary pregnancies (year of birth 1997 and onward). The continuing pregnancy subgroup was chosen to collect information for the subgroup of patients who desired continuation of an unintended pregnancy. The contemporary pregnancies were chosen to see whether modern transplantation immunosuppression regimens and obstetric practices influence the results of the study.

RESULTS

Between 1967 and 2019, there were 2,030 pregnancies after kidney transplantation in the Transplant Pregnancy Registry International. After the exclusion of 307 pregnancies (15.1%) with unknown intention, 1,723 pregnancies were included in this study; 1,081 (62.7%) were intended pregnancies and 642 (37.3%) were unintended. There were significant demographic differences based on pregnancy intention (Table 1). Unintended pregnancies were characterized by younger age (median age 29.3 years vs 31.3 years, P<.001), nulliparity (57.6% vs 51.1%, P=.01), self-reported Black race (11.5% vs 4.0%, P<.001), and chronic hypertension (48.5% vs 41.9%, P=.01) Similar significant differences were seen in subgroup analysis (Appendix 2, available online at http://links.lww.com/AOG/D873). Rates of pregnancy termination were higher in unintended (4.7%) compared with intended (0.4%) pregnancies (P<.001).

Table 1.

Demographics, Transplant Characteristics, and Transplant Outcomes in Kidney Transplant Recipients by Pregnancy Intention^*

Variable	Intended Pregnancy (n=1,081)	Unintended Pregnancy (n=642)	P
Age (y)	31.3 (6.6)	29.3 (7.8)	<.001
BMI (kg/m2)†	24 (7)	24 (7)	.56
Year	2003 (17)	1997 (15)	<.001
Nulliparity	552 (51.1)	370 (57.6)	.01
Gestation			.27
Singleton	1,038 (96.0)	627 (97.7)
Twin	37 (3.4)	14 (2.2)
Triplet	5 (0.5)	1 (0.2)
Quadruplet	1 (0.1)	0 (0.0)
Race			<.001
Asian	59 (5.5)	39 (6.1)
Black or African American	43 (4.0)	74 (11.5)
White	869 (80.4)	433 (67.4)
Additional races and ethnicities	76 (7.0)	73 (11.4)
Unknown	34 (3.1)	23 (3.6)
Diabetes			.33
Gestational, diet controlled	23 (2.1)	7 (1.1)
Gestational, insulin controlled	39 (3.6)	23 (3.6)
Pregestational	46 (4.3)	34 (5.4)
Chronic hypertension	450 (41.9)	309 (48.5)	.01
Prenatal care‡	783 (89.2)	445 (88.5)	.69
Birth outcome			<.001
Live birth	858 (79.4)	483 (75.2)
Stillbirth	10 (0.9)	11 (1.7)
Miscarriage	200 (18.5)	114 (17.8)
Termination	4 (0.4)	30 (4.7)
Ectopic	9 (0.8)	4 (0.6)
Transplantation–pregnancy interval (y)	5 (5.3)	4 (5.1)	<.001
Transplantation no.			.003
1	930 (86.0)	529 (82.4)
2	145 (13.4)	96 (15.0)
3	6 (0.6)	15 (2.3)
4	0 (0.0)	2 (0.3)
Donor type			<.001
Deceased	346 (32.0)	280 (44.0)
Living, related	581 (53.8)	319 (50.1)
Living, unrelated	153 (14.2)	38 (6.0)
History of rejection	275 (26.6)	186 (30.7)	.08
Pretransplantation pregnancy	292 (27.1)	199 (31.1)	.07
Creatinine level (mg/dL)
Before pregnancy§	1.2 (0.4)	1.2 (0.5)	.77
During pregnancy‖	1.2 (0.5)	1.2 (0.5)	.84
After pregnancy¶	1.2 (0.5)	1.2 (0.6)	.22
Immunosuppression in pregnancy
Cyclosporine	453 (41.9)	293 (45.6)	.14
Tacrolimus	435 (40.2)	213 (33.2)	.003
Azathioprine	800 (74.6)	431 (67.2)	<.001
Prednisone	910 (84.9)	573 (89.5)	.01
Mycophenolate	61 (5.7)	102 (16.0)	<.001
Sirolimus	6 (0.6)	18 (2.8)	<.001
Acute rejection	10 (0.9)	15 (2.3)	.02
Immunosuppressant change	7/10 (70.0)	12/15 (80.0)	.65
Allograft loss 2 y from the end of pregnancy	37 (3.5)	51 (8.1)	<.001
Immunosuppressant change	11/37 (29.7)	23/51 (45.1)	.18
Allograft loss, all	319 (29.7)	283 (44.4)	<.001
Adult follow-up (y)	12.3 (15.6)	15.4 (15.6)	<.001

BMI, body mass index.

Data are n (%) or median (interquartile range).

^*Small deviations in total number or percentages are attributable to missingness of less than 10% in each cohort except for the following: denominator: intended pregnancy n=1,081, unintended pregnancy n=642.

^†BMI: intended pregnancy (n=245, 22.7%), unintended pregnancy (n=232, 36.1%); no difference in proportion by missingness.

^‡Prenatal care: intended pregnancy (n=203, 18.8%), unintended pregnancy (n=139, 21.7%); no difference in proportion by missingness.

^§Creatinine before pregnancy: intended pregnancy (n=161, 14.9%), unintended pregnancy (n=119, 18.5%); no difference in proportion by missingness.

^‖Creatinine during pregnancy: intended pregnancy (n=175, 16.2%), unintended pregnancy (n=118, 18.4%); no difference in proportion by missingness.

^¶Creatinine after pregnancy: intended pregnancy (n=213, 19.7%), unintended pregnancy (n=154, 24.0%), with more missing data in the unintended pregnancy group (P=.04).

Transplantation characteristics and outcomes are also detailed in Table 1. Unintended pregnancies had a shorter transplantation-to-pregnancy interval (4 years vs 5 years, P<.001), were more commonly repeat transplantations (17.6% vs 14.0%, P=.003), and were more commonly transplanted from deceased compared with living donors (44.0% vs 32.0%, P<.001). Unintended pregnancies had lower use of tacrolimus (33.2% vs 40.2%, P=.003) and azathioprine (67.2% vs 74.6%, P<.001) in pregnancy and, conversely, higher rates of prednisone use (89.5% vs 84.9%, P=.01). Mycophenolate product use (includes both mycophenolic acid and mycophenolate mofetil) (16.0% vs 5.7%, P<.001) and sirolimus use (2.8% vs 0.6%, P<.001) were significantly more common in unintended pregnancies. Similar significant differences were seen in subgroup analysis (Appendix 3, available online at http://links.lww.com/AOG/D873). The trend in mycophenolate product exposure by pregnancy intention is depicted in Figure 1.

Fig. 1. Mycophenolate product exposure in pregnancy based on pregnancy intention from time of mycophenolate U.S. Food and Drug Administration (FDA) approval in the United States (1995).

Yin. Unintended Pregnancy After Kidney Transplantation. O&G Open 2024.

The primary outcome of biopsy-proven acute kidney rejection occurred more frequently after unintended pregnancies compared with intended pregnancies (2.3% vs 0.9%, P=.02) (Table 1). After adjustment for birthing age, self-reported race, chronic hypertension, transplantation-to-pregnancy interval, type of organ donor, history of rejection, transplantation number, creatinine level (milligrams per deciliter) before pregnancy, pretransplantation pregnancy, mycophenolate product use, and year of pregnancy, unintended pregnancy was not associated with acute kidney rejection (adjusted odds ratio [AOR] 2.38, 95% CI, 0.91–6.30, P=.08); the most important factor associated with acute rejection was creatinine level before pregnancy (Fig. 2A). Immunosuppressant changes in pregnancy were common to both cohorts experiencing acute kidney rejection (80.0% unintended vs 70.0% intended, P=.65). There was a lower rate of optimized allograft before an unintended pregnancy, with optimized graft defined as creatinine level less than 1.5 mg/dL, minimal proteinuria, no cytomegalovirus, and no mycophenolate product use (37.9% unintended vs 50.5% intended, P<.001).

Fig. 2. Associations between unintended pregnancy and graft outcomes. Acute peripartum kidney rejection (A) and graft loss at 2 years from the end of pregnancy (B). *P<.05, **P<.01, ***P<.001. Associations between unintended pregnancy and graft outcomes. Long-term graft survival (C). Additional races and ethnicities include American Indian, Alaska Native, and mixed race. cHTN, chronic hypertension.

Yin. Unintended Pregnancy After Kidney Transplantation. O&G Open 2024.

For secondary outcomes, unintended pregnancies had a higher incidence of allograft loss at 2 years from the end of pregnancy (8.1% vs 3.5%, P<.001) (Table 1), which remained significant after adjustment (AOR 2.27, 95% CI, 1.32–3.94, P=.003) (Fig. 2B). Unintended pregnancy was associated with decreased survival on unadjusted analysis (Fig. 3), but after adjustment for allograft risk factors, unintended pregnancy was not associated with decreased allograft survival (adjusted hazard ratio 1.22, 95% CI, 1.00–1.49, P=.05) (Fig. 2C).

Fig. 3. Kidney transplant allograft survival from the end of pregnancy based on pregnancy intention.

Yin. Unintended Pregnancy After Kidney Transplantation. O&G Open 2024.

In the sensitivity analysis, the subgroup of continuing pregnancies showed results similar to those of the entire study. In contrast, the subgroup of contemporary pregnancies from 1997 onward did not show an association between unintended pregnancy and acute rejection (AOR 1.68, 95% CI, 0.57–4.79, P=.3), graft loss at 2 years (AOR 1.50, 95% CI, 0.76–2.92, P=.20), or graft survival (adjusted hazard ratio 0.93, 95% CI, 0.68–1.27, P=.64). Although chronic hypertension was not associated with graft survival in the larger study (Fig. 2), it was associated in contemporary pregnancies (Appendix 4, available online at http://links.lww.com/AOG/D873).

Obstetric and neonatal outcomes are detailed in Table 2, which include the subgroup of 1,341 live births and 1,395 neonates. There was no difference in preeclampsia, cesarean birth rate, severe maternal morbidity, gestational age at birth, congenital anomalies, or neonatal composite morbidity based on intention status of the pregnancy. These results were similar in the contemporary subgroup (Appendix 5, available online at http://links.lww.com/AOG/D873).

Table 2.

Live Birth Obstetric and Neonatal Outcomes by Pregnancy Intention in Kidney Transplant Recipients^*

Variable	Intended	Unintended	P
Obstetric outcome	n=858	n=483
Preeclampsia	268 (31.3)	140 (29.0)	.38
Cesarean birth	441 (51.8)	254 (52.8)	.71
Severe maternal morbidity	31 (3.6)	16 (3.3)	.77
Neonatal outcome	n=897	n=498
Fetal sex			.93
Female	429 (47.8)	237 (47.6)
Male	468 (52.2)	261 (52.4)
Gestational age at birth (wk)	37 (3.3)	37 (4)	.87
Birth weight (g)	2,721.8 (978)	2,693 (970.8)	.23
Congenital anomalies	44 (5.0)	31 (6.2)	.32
NICU admission	207 (23.2)	92 (18.6)	.05
Neonatal composite morbidity	128 (14.3)	64 (12.9)	.46
Neonatal death	12 (1.3)	3 (0.6)	.20

Data are median (interquartile range) or n (%) unless otherwise specified.

^*Small deviations in total number or percentages are attributable to missingness of less than 10% in each cohort.

DISCUSSION

Our large registry study of 1,723 pregnancies after kidney transplantation found an unintended pregnancy rate of approximately 40%. Prior studies report similar rates of unintended pregnancy ranging from 29% to 30%.^17–19 Rates of contraceptive counseling after transplantation range from only 5%²⁰ to 50% at the highest,²¹ and among those with an unintended pregnancy, 70–92% report not using any kind of contraception.^18–20 These data point to gaps in our ability to facilitate adequate reproductive planning in the transplantation population, despite these patients' engagement in the medical system during their pretransplantation evaluation, periodic medical assessments while on the waiting list, and the perioperative period. We identified several key risk factors for unintended pregnancy, including younger age, nulliparity, Black race, hypertension, deceased donor organ, and repeat transplantation. Patients with these risk factors may be particularly vulnerable and merit targeted intervention related to family planning and their transplant health. The median transplantation-to-pregnancy interval of 4 years in our study also emphasizes the need for long-term reproductive health follow-up, much beyond the initial 1–2 years after transplantation when recipients receive the most medical attention. Although analysis of contraceptive use was beyond the scope of this analysis, we aim to explore this data in future work.

We had originally hypothesized that unintended pregnancy might be associated with acute peripartum kidney rejection, with the mechanism attributable to acute medication change at time of pregnancy discovery; however, those with acute rejection across both the intended and unintended cohorts had equally high rates of medication changes, and we ultimately did not find that unintended pregnancy or mycophenolate product use was associated with acute kidney rejection. Rather, factors related to suboptimal allograft function at the start of pregnancy such as short transplantation pregnancy interval and elevated baseline creatinine level were the most significant factors associated with peripartum kidney rejection. Therefore, recipients should be counseled that immediate transplantation outcomes in pregnancy are associated with prepregnancy graft health.

The secondary finding that unintended pregnancy was associated with allograft loss at 2 years is not fully explained by the demographic and biological factors adjusted for in our analysis. We hypothesize that the relationship is driven by unmeasured social determinants of health. Unintended pregnancy can be thought of as a marker of social vulnerability, occurring in the background of risk factors that also predispose to allograft loss such as insurance status, socioeconomic status, and health literacy.^22,23 We could not measure these covariates given the limitations of the registry, but our study identifies key parameters to collect and understand in future transplantation pregnancy work, especially because our registry and others mainly track biological covariates. These findings were no longer significant in the contemporary cohort, possibly indicating improved support structures in more recent pregnancies and increased importance of the measured risk factors for graft loss.

We did find, however, that unintended pregnancy was associated with mycophenolate product use, which is teratogenic²⁴ and is known to cause miscarriage. In addition, there are potential adverse effects on the allograft of abrupt discontinuation of mycophenolate products in the context of unintended pregnancy; the independent association of mycophenolate products and graft survival in our study suggests that this may be the case. To reduce mycophenolate exposure, optimize prepregnancy allograft health, and decrease unintended pregnancy rates, multiple health care members, including transplantation nurses, nephrologists, primary care physicians, obstetrician–gynecologists, and transplantation surgeons involved in transplantation care, can educate patients about the risks of pregnancy after transplantation and decrease barriers to contraceptive access. Beyond contraceptive access, abortion access for kidney transplant recipients expands their reproductive choice. Since the Dobbs v. Jackson Women's Health Supreme Court decision in 2022 overturned constitutional protections for abortion, 26 states²⁵ have severely limited or banned abortion. The repercussions of Dobbs have the potential to worsen outcomes for transplant recipients who have a pregnancy that is unintended or unexpectedly complicated; abortion access and the ways that this modifies risk on the basis of pregnancy intentionality represent an important future direction for research.²⁶

Strengths of this study include adequate power to detect differences in allograft outcomes, biopsy reports to support acute rejection, longitudinal follow-up across a diversity of transplantation and obstetric centers, detailed maternal and neonatal outcomes, and availability of medications used throughout pregnancy. In terms of limitations, reporting bias may skew our results; those with the worst pregnancy outcomes may be less likely to participate with the registry or continue follow-up over time; conversely, those with teratogen exposure or complicated pregnancies may be more likely to register. Participants may have not reported pregnancies that ended in induced abortion; bias and stigma often lead to underreporting. Although there were missing data for body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), creatinine levels after pregnancy, and prenatal care, none of these were variables included in our multivariable or survival analyses. The generalizability of our study is limited by the concentration of recipients from the United States and the high proportion of participants who self-identify as White race. The study should be interpreted in the context of the primary outcome, and nonsignificant associations for obstetric and neonatal morbidity cannot necessarily be generalized.

Unintended pregnancy was not associated with acute kidney allograft rejection in the peripartum period, but it was associated with mycophenolate exposure and allograft loss at 2 years. The same social vulnerabilities that underlie difficulty in accessing reproductive care may be similarly important for transplant health. Pregnancy counseling and longitudinal follow-up of transplant recipients are important steps to reduce rates of unintended pregnancy and optimize transplantation success over time.