Pregnancy and Long-Term Kidney Function in CKD

Visual Abstract

Abstract

Key Points

• Outcomes for women with mild CKD are reassuring: Pregnancy does not impair long-term kidney function in mild CKD.

• Women with advanced CKD should be informed on risks of losing prepregnancy kidney function in pregnancy.

• Animal studies support clinical findings in mild CKD and offer insights in the potential underlying mechanistics.

Background

Pregnancy may accelerate kidney function decline in women with CKD, particularly in advanced stages. Some clinicians may therefore advise against pregnancy. Exact effect of pregnancy and subgroup risks (e.g., for patients with proteinuria or hypertension) are still uncertain. For patients with CKD who wish to conceive, establishing the effect of pregnancy and identifying those at risk for progression is crucial.

Methods

We conducted a systematic review and meta-analysis in human and animal studies separately, synthesizing all available evidence on long-term kidney function after pregnancy in women with preexisting CKD. Primary outcome was standardized mean difference (SMD) in kidney function before versus after pregnancy. For clinical implications, we calculated mean GFR differences. Secondary outcomes were incidences of kidney failure/KRT after delivery and kidney function deterioration. Subanalyses stratified studies as mild versus advanced CKD if studies had ≤25% versus >25% of participants CKD stage 3–5 (unless cohorts defined otherwise) and chronic hypertension as <25% versus >25% of participants having chronic hypertension.

Results

We analyzed 36 human studies including 2945 patients, 4623 pregnancies, and 12 animal studies. Pregnancy had no effect on long-term kidney function in mild CKD cohorts (SMD, −0.22 [−0.56 to 0.13]) over a mean follow-up of 4.4 (SD 0.7) years. However, kidney function was significantly lower after pregnancy in advanced CKD cohorts (SMD, −0.55 [−0.80 to −0.30]), pooled eGFR decline −8.96 ml/min (−17.4 to −0.48), mean follow-up 2.6 years. Chronic hypertension affected overall SMD (β=−0.01 [−0.02 to −0.001], P = 0.03). Pooled kidney failure/KRT incidence was 9%, with a mean follow-up 6.5 of years. Pregnancy did not affect kidney function after delivery in animal nephropathy models.

Conclusions

The results for patients with mild CKD are reassuring as pregnancy does not affect long-term kidney function. Animal studies support clinical findings in mild CKD and offer insights in the potential underlying mechanisms. Patients with advanced CKD should be informed on risking kidney function decline in pregnancy. Studies including eGFR-slopes prepregnancy and postpregnancy are scarce, limiting understanding of pregnancy's impact next to natural disease progression in advanced CKD. This highlights the need for future research including multiple eGFR measurements over time.

Introduction

CKD complicates 3% of pregnancies.¹ These pregnancies are considered high risk because of elevated risks of preeclampsia, fetal growth restriction, preterm birth, and fetal death.^2,3 There is still controversy regarding the effect of pregnancy on long-term kidney function.⁴

During normal pregnancy, hemodynamic changes increase kidney function by over 40% (hyperfiltration).^2,5,6 The current clinical paradigm is that in patients with CKD, gestational hyperfiltration leads to loss of kidney function possibly accelerating the need for KRT, especially in advanced stages. Therefore, some clinicians may advise a patient not to get pregnant.

Several studies in the mild CKD population have shown no effect of pregnancy on long-term kidney function outcomes.^2,7–9 A previous systematic review and meta-analysis concluded that pregnancy was not a risk factor for progression of kidney disease.³ However, this review focused on “kidney events” (i.e., kidney failure [KF], doubling of serum creatinine [SCr]) as primary outcome, without comprehensively analyzing the effect of pregnancy on long-term kidney function, and compared pregnant and nonpregnant individuals, potentially introducing selection bias in those who choose not to pursue pregnancy. In addition, risks for subgroups of proteinuria and hypertension remain uncertain. A recent study in women with pregnancy and advanced CKD showed a step-decline in eGFR of 5.3 ml/min per 1.73 m², equivalent to a loss of 2.5 years prepregnancy kidney disease.¹⁰ Previous studies in controlled animal models of CKD have not shown deleterious effects of pregnancy,^11–14 but no comprehensive overview of animal studies on the topic has been conducted.

It is of great importance to patients with CKD and a wish to conceive to establish what effect pregnancy has on long-term kidney function. By collating animal and human studies, we aim to investigate all available evidence on this effect and to provide guidance for prepregnancy counseling.

Methods

Study Protocol

This systematic review was conducted according to preregistered protocols on the International Prospective Register of Systematic Reviews (human protocol: CRD42023371856, animal protocol: CRD42023371863) and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.¹⁵ Deviations from the preregistered protocol including rationale are summarized in Supplemental Table 1. The review question was “What is the effect of pregnancy on long-term kidney function in animals and humans with prepregnancy CKD?.”

Literature Search

We searched PubMed and Embase to identify animal and human studies reporting on long-term kidney function after pregnancy complicated by CKD (animal studies up to April 3, 2024, human studies up to October 25, 2023). The search strings are presented in Supplemental Tables 2 and 3. No publication date restrictions were applied. References of included studies were screened for additional relevant articles (snowballing).

Eligibility Criteria

Two researchers (M.E. Gosselink and J.M.M. Sluijters) independently screened articles by title and abstract, followed by full-text review for final inclusion, data-extraction, and assessment of risk of bias and study quality. Researchers were blinded to each other's assessments, using Rayyan (https://rayyan.ai/cite). Discrepancies were resolved by discussion or by a third reviewer (R. Snoek and K.E. Wever).

Inclusion criteria for human studies were (1) patients older than 18 years with CKD before pregnancy, with (2) pregnancy more than 20 weeks duration, and (3) reported on long-term kidney function for a minimum follow-up time of 6 months after pregnancy. Studies conducted solely in patients with unsuitable comorbidities (i.e., diabetes and SLE) and on pregnancy after KRT (i.e., dialysis or kidney transplantation) were excluded.

For animal studies, we included studies on (1) mammals with prepregnancy nephropathy and (2) pregnancy when (3) reporting on kidney function before and after pregnancy. Among the included nephropathy models were KF models, such as five/sixth nephrectomy, drug-induced nephropathy, and nephropathy induced through immunization (i.e., Fx1A antigen). Studies conducted in animal models with unsuitable comorbidities (i.e., diabetes, SLE, and spontaneous hypertension without nephropathy) were excluded.

Data Extraction

For human studies, baseline characteristics (i.e., prepregnancy kidney function, chronic hypertension, and prepregnancy proteinuria) were extracted. Extracted primary outcomes were kidney function measurements (eGFR, SCr, plasma creatinine, or creatinine clearance) before (or in early pregnancy if no prepregnancy values were available) and after pregnancy. Extracted secondary outcomes were incidence of kidney function deterioration after pregnancy, KF, or need for KRT after pregnancy. Where possible, separate incidences for kidney function deterioration and KRT were extracted, assigning patients to one group exclusively to avoid overlap.

For animal studies, information on study characteristics, i.e., methods of KF induction, were extracted. Extracted primary outcomes were kidney function measurements (eGFR, SCr, plasma creatinine, creatinine clearance, and inulin clearance) before (or in virgin controls) and after pregnancy. Although this study primarily focuses on long-term kidney function after pregnancy, we also included kidney function values at delivery due to limited long-term follow-up in animal studies. Extracted secondary outcomes were other kidney function measurements if available, i.e., glomerular capillary pressure (mm Hg), preglomerular and efferent arteriole resistance (Ra, Re, dyn-s-cm×10¹⁰), Qa (glomerular plasma flow, nl/min), Kf (glomerular capillary Kf, Nl/s per mm Hg), BP (mm Hg), and urinary protein volume (mg/24 hours). Per outcome, means, SD, and number of subjects per group were collected. If necessary, SD was calculated from SEM.

Assessment of Risk of Bias and Study Quality

For human studies, we used the Newcastle-Ottawa Scale for cohort studies.¹⁶ For animal studies, we used the Systematic Review Center for Laboratory animal Experimentation risk of bias.¹⁷ For adaptations to these tools, see Supplemental Table 4. Selective outcome reporting in animal studies was assessed by comparing methods and results sections because of missing prespecified protocols. No studies were excluded based on quality.

Data Synthesis

MA of Studies in Humans

Next to the primary outcome of standardized mean difference (SMD) in kidney function before versus after pregnancy, for clinical implications, the GFR and creatinine differences were analyzed separately using the mean difference (MD) before versus after pregnancy. For cohorts that reported data on the secondary outcome for a nonpregnant CKD control group, risk ratio was calculated. MA were performed for outcomes reported in a minimum of two studies by pooling effect sizes using a random effects model, based on restricted estimates of maximum likelihood. We used a random effects model to account for anticipated between-study heterogeneity because of differences in the studies' populations, settings, and quality. If studies reported subgroup results separately, data were pooled and included in the analysis. All data were expressed as effect size and corresponding 95% confidence intervals (CIs). We calculated weighted mean follow-up time after delivery based on the cohorts' sample sizes with its variability expressed as weighted SD.

Meta-regression analyses were performed to study the effect of CKD stage prepregnancy (linear, percent patients with advanced CKD in the cohort), chronic hypertension prepregnancy (linear, percent chronic hypertensive patients in the cohort), proteinuria prepregnancy (linear), kidney disease etiology (stratified), follow-up time after delivery (linear), and publication year (linear). A minimum of five studies were required for linear and stratified variables (for at least two strata). For prepregnancy kidney function subgroup analyses, studies were stratified by CKD severity, with advanced CKD defined as >25% of the cohort having advanced CKD (CKD-stage 3–5, eGFR <60 ml/min, SCr >1.4 mg/dl or as stated otherwise by cohorts, see Tables 1 and 2).

Table 1.

Baseline characteristics of human studies

Author (Year)	Study Design	Patients, N	No. of Pregnancies, N	Cause of CKD	Prepregnancy Chronic Hypertension	Prepregnancy Proteinuriaa
						N	Mean±SD, g/d	Yes/Total
Studd and Blainey (1969)29	Retrospective	19	29	Nephrotic syndrome	5/14 (36%)	NR	NR	NR
Bear et al. (1976)30	Retrospective	37	44	Heterogeneous	12/34 (35%)	NR	NR	NR
Katz et al. (1980)9	Retrospective and prospective combined	89	121	Heterogeneous	20/74 (27%)	67	0.6±1.95	16/67 (24%)
Gregory and Mansell (1983)31	Retrospective	22	40	Cystinuria	NR	NR	NR	NR
Hou et al. (1985)32	Retrospective	19	21	Heterogeneous	14/17 (82%)	NR	NR	NR
Barceló et al. (1986)33	Retrospective	48	66	Primary glomerular disease	8/48 (17%)b	48	1.49±0.53	NR
Becker et al. (1986)34	Prospective	6	6	Reflux nephropathy	NR	1	1.2	NR
Jungers et al., primary GN (1986)35	Retrospective	122	240	Primary chronic glomerulonephritides	23/240 (10%)	NR	NR	NR
Jungers et al., PKD (1986)35	Retrospective	64	172	Polycystic kidney disease	18/64 (28%)	NR	NR	NR
Packham et al. (1989)36	Retrospective	238	395	Primary GN	48/376 (13%)	NR	NR	NR
Cunningham et al. (1990)37	Prospective	37	37	Heterogeneous	25/37 (68%)	NR	NR	NR
Abe (1991)7	Retrospective	118	168	IgA nephropathy	15/168 (9%)	32	0.6±0.4	NR
Chapman et al. (1994)38	Prospective	235	605	ADPKD	13/170 (8%) b	NR	NR	NR
Hemmelder et al. (1995)39	Retrospective	19	30	Primary glomerular disease	4/19 (21%) b	19	3c (IQR, 1.3–4.7)	17/19 (89%)
Alexopoulos et al. (1996)40	Retrospective	17	24	Glomerular disease	3/17 (18%) b	NR	NR	NR
Jones and Hayslett (1996)41	Retrospective	67	82	Heterogeneous	29/67 (43%)	NR	NR	NR
Jungers et al. (1996)8	Retrospective	158	375	Reflux nephropathy	34/375 (9%)	NR	NR	NR
Bar et al. (2000)42	Retrospective	38	46	Heterogeneous	12/46 (26%)	NR	NR	NR
North et al. (2000)43	Prospective	46	54	Reflux nephropathy	19/54 (35%)	NR	NR	10/52 (19%)
Malik et al. (2002)44	Retrospective	9	30	Primary membranous GN	NR	NR	NR	NR
Köhler et al. (2003)45	Retrospective	89	242	Reflux nephropathy	NR	NR	NR	NR
Imbasciati et al. (2007)46	Retrospective	49	49	Heterogeneous	25/49 (51%) b	49	1.2±1.2	NR
Limardo et al. (2010)47	Retrospective	136	229	IgA nephropathy	27/136 (20%) b	136	1.0c range 0–6	NR
Shimizu et al. (2015)48	Prospective	64	64	IgA nephropathy	NR	64	0.49±0.38	NR
Davidson et al. (2015)49	Retrospective	55	55	Heterogeneous	34/55 (62%)	NR	NR	39/55 (71%)d
Singh et al. (2015)50	Retrospective	51	51	Heterogeneous	NR (continuous)	51	1.03±0.96	NR
Fukasawa et al. (2016)51	Retrospective	139	156	Heterogeneous	30/156 (19%)	NR	NR	22/156 (14%)
Wu et al., ADPKD (2016)52	Retrospective	54	146 (92>20 wk)	ADPKD	NR	NR	NR	8/54 (15%)
Wu et al., simple cyst (2016)52	Retrospective	92	265 (153>20 wk)	Simple cyst	NR	NR	NR	NR
O'Shaugnessy et al. (2017)53	Retrospective	43	48	Heterogeneous	17/42 (40%)	33	1.6c (IQR, 0.6–4.1)	NR
Su et al. (2017)54	Prospective	104	110	IgA nephropathy	15/104 (14%) b	104	1.04c (range, 0.03–7.25)	50/104 (48%)
Li et al. (2018)55	Retrospective	29	29	Primary chronic glomerular nephritis	2/29 (7%)	29	0.5±0.3	NR
Kaul et al. (2018)56	Retrospective	172	NR	Heterogeneous	NR	NR	NR	NR
He et al. (2018)57	Retrospective	293	300	Heterogeneous	49/300 (16%)	300	1.06±1.44	NR
Liu et al. (2020)58	Retrospective	25	27	Primary membranous nephropathy	2/27 (7%)	11	0.6±0.6	NR
Wiles et al. (2021)10	Retrospective	159e	178	Heterogeneous	96/121 (79%)	140	63c,f (IQR, 18–215)	NR
Marinaki et al. (2022)59	Retrospective	22	29	Glomerular disease	5/22 (23%) b	29	0.19c (range, 0.05–1.59)	6/29 (21%)
He et al. (2022)60	Retrospective	53	54	Heterogeneous	22/54 (41%)	43	0.97c (range, 0.01–5)	27/54 (50%)

IQR, interquartile range; NR, not reported.

^aProteinuria in g/d.

^bPatient-level BP measurement.

^cMedian.

^dPCR spot.

^eIncluded 43 kidney transplant recipients.

^fPCR, mg/mmol.

Table 2.

Baseline characteristics human studies, kidney function at baseline

Author (Year)	Definition Advanced CKD	Advanced CKD	Kidney Function, Continuous
			Unit of Measurement	N, Pre	Prepregnancy	N, after	After Delivery	Follow-Up after Delivery, yr
Studd and Blainey (1969)29	Clearance <60 ml/min	0/8 (0%)	Creatinine clearance, ml/min	8	92.38±28.28	15	99.67±37.49	7.3±5.3
Bear (1976)30	SCr >1.6 mg/dl	8/37 (22%)	NR	NR	NR	NR	NR	NR
Katz et al. (1980)9	SCr >1.4 mg/dl	0/55	SCr, mg/dl	55	0.95±0.16	62	0.99±0.23	5.8±5.2
Gregory and Mansell (1983)31	NR	NR	Plasma creatinine, mmol/l	8	0.08±0.023	21	0.08±0.023	Median: 10 yr (range, 8 mo–48 yr)
Hou et al. (1985)32	SCr >1.4 mg/dl	14/15 (93%)	SCr, mg/dl	15a	1.83±0.33	12	3.65±3.04	2.08±2.41
Barceló et al. (1986)33	SCr >1.3 mg/dl	5/48 (10%)	SCr, mg/dl	48	0.83±0.36	NR	NR	NA
Becker et al. (1986)34	SCr >1.4 mg/dl	2/2 (100%)	Plasma creatinine, mmol/l	2	0.24±0.05	6	1.05±0.43	1.18±0.45
Jungers et al., primary GN (1986)35	CKD: plasma creatinine >135 μmol/L (=1.53 mg/dl) or blood urea >8.3 mmol/L (=23.24 mg/dl)	7/122 (6%)	NR	NR	NR	NR	NR	NR
Jungers et al., PKD (1986)35	CKD: plasma creatinine >135 μmol/L (=1.53 mg/dl) or blood urea >8.3 mmol/L (=23.24 mg/dl)	4/64 (6%)	NR	NR	NR	NR	NR	NR
Packham et al. (1989)36	Plasma creatinine >0.11 mmol/L (=1.24 mg/dl) or plasma urea >8.3 mmol/L (=23.24 mg/dl) in first trimester	4/394 (1%)	NR	NR	NR	NR	NR	NR
Cunningham et al. (1990)37	SCr >1.4 mg/dl	37/37 (100%)	NR	NR	NR	NR	NR	NR
Abe (1991)7	GFR <60 ml/min	0/118 (0%)	GFR, ml/min	32	81±13	26	69.0±26.0	3±NR
Chapman et al. (1994)38	SCr >1.2 mg/dl	5/428 (1%)	SCr, mg/dl	12	1.1±0.1	12	1.1±0.1	0.5±NR
Hemmelder et al. (1995)39	SCr >1.2 mg/dl	2/19 (11%)	SCr, mg/dl	19	0.9b median IQR, 0.8–1.0	19	NA	NR
Alexopoulos et al. (1996)40	SCr >1.3 mg/dl or increase ≥50% prepregnancy kidney function	2/17 (12%)	SCr, mg/dl	17	0.9±0.3	16	1.2±0.6	4.67±3.75
Jones and Hayslett (1996)41	SCr >1.4 mg/dl	82/82 (100%)	SCr, mg/dl	82a	1.9±0.8	62	2.59±1.03	1±NR
Jungers et al. (1996)8	SCr >0.11 mmol/L (=1.24 mg/dl)	30/375 (8%)	NR	NR	NR	NR	NR	NR
Bar et al. (2000)42	SCr >1.4 mg/dl	4/38 (11%)	SCr, mg/dl	38c	0.95±0.62	NR	NR	NR
North et al. (2000)43	SCr >0.13 mmol/L (>1.4 mg/dl)	5/51 (10%)	SCr, mmol/l	NR	NR	NR	NR	NR
Malik et al. (2002)44	NR	NR	SCr, μmol/l	9	68±17	9	80±48	NR
Köhler et al. (2003)45	NR	NR	NR	NR	NR	NR	NR	NR
Imbasciati et al. (2007)46	SCr >1.4 mg/dl or GFR <60 ml/min recorded within 3 mo before conception or within first month of pregnancy	49/49 (100%)	SCr, mg/dl	49	2.1±1.0	NR	NR	NR
Limardo et al. (2010)47	SCr >1.2 mg/dl	0/136 (0%)	Creatinine clearance, ml/min	136	92±17	NR	NR	NA
Shimizu et al. (2015)48	eGFR <60 ml/min	16/64 (25%)	eGFR, ml/min	64	75.84±7.25	64	71.23±8.24	5±NR
Davidson et al. (2015)49	eGFR <60 ml/min	28/55 (51%)	eGFR, ml/min per 1.73 m2	55	65.93±NR	55	58.81±NR	1±NR
Singh et al. (2015)50	eGFR <60 ml/min	19/51 (37%)	eGFR, ml/min per 1.73 m2	51	71.7±27.6	51	50.7±34.7	1±NR
Fukasawa et al. (2016)51	eGFR <60 ml/min	58/156 (37%)	eGFR, ml/min per 1.73 m2	156a	71.94±11.21	156	67.67±22.34	3.8±NR
Wu et al., ADPKD (2016)52	SCr ≥1.2 mg/dl	1/54 (2%)	NR	NR	NR	NR	NR	NR
Wu et al., simple cyst (2016)52	SCr ≥1.2 mg/dl	0 (0%)	NR	NR	NR	NR	NR	NR
O'Shaugnessy et al. (2017)53	eGFR <60 ml/min	8/35 (23%)	eGFR, ml/min per 1.73 m2	35a	90b (IQR, 61–126)	NR	NR	NR
Su et al. (2017)54	eGFR <60 ml/min	11/104 (11%)	NR	104	102.6±23.9	NR	NR	NA
Li et al. (2018)55	NR	NR	SCr, μmol/l	29	68.2±8.0	29	70.2±7.9	0.5±NR
Kaul et al. (2018)56	GFR <60 ml/min	76/172 (44%)	NR	NR	NR	NR	NR	NR
He et al. (2018)57	eGFR <60 ml/min	30/300 (10%)	NR	300a	106.22±26.11	NR	NR	NR
Liu et al. (2020)58	eGFR <60 ml/min	0/25 (0%)	eGFR, ml/min per 1.73 m2	11a	108.3±11.4	24	117±11.2	1±NR
Wiles et al. (2021)10	eGFR <60 ml/min	178/178 (100%)	Creatinine, μmol/l	167a	140b (IQR, 123–167)	NR	NR	NA
Marinaki et al. (2022)59	SCr ≥1.2 mg/dl	5/22 (23%)	eGFR, ml/min per 1.73 m2	22	115b (range 30–120)	NR	NR	NR
He et al. (2022)60	eGFR <60 ml/min	53/53 (100%)	eGFR, ml/min per 1.73 m2	43a,d	44.9±13.5	53	38.4±16.2	0.5±NR

Kidney function and follow-up time after delivery are reported in mean6SD unless stated otherwise.

IQR, interquartile range; NR, not reported; SCr, serum creatinine.

^aMeasurement on pregnancy level.

^bMedian.

^cMeasurement at first admission during pregnancy, first or second trimester.

^dMeasurement in early pregnancy.

MA of Animal Studies

MA was performed for outcomes reported in a minimum of eight studies. Descriptive synthesis was performed for outcomes reported in less than eight studies. For each outcome, data were expressed as SMD (95% CI) and pooled CIs using a random effects model, based on restricted estimates of maximum likelihood. Meta-regression analyses were performed to assess the effect of type of KF model (stratified), species (stratified), follow-up time after delivery (linear), and proteinuria (linear) on the outcome.

Statistical analyses were performed using R software (R.4.3.3, Metafor and Metareg package). A two-sided P value of <0.05 was considered significant. For outcomes reported in at least 20 studies, we assessed publication bias by visual inspection of funnel plots. Since the Egger test is not validated for incidence proportions, it was not performed. Heterogeneity between included studies was assessed by I² statistics.

Results

Study Selection

The search identified 17,810 records (Figure 1, A and B). In this systematic review, 48 studies were included: 36 human and 12 animal studies. Owing to considerable heterogeneity, the results were analyzed separately for humans (Tables 1 and 2, and Supplemental Table 5) and animals (Table 3 and Supplemental Table 6).

Figure 1.

Flowcharts of study selection. (A) Human studies. (B) Animal studies.

Table 3.

Study characteristics animal studies

Author	Species (Strain)	Nephropathy Model	N	Pregnancies per Animal, N	BP Controlsa				Proteinuria Controlsa
					Unit	Time after Disease Induction	N	BP, mm Hg	Time after Disease Induction	N	Proteinuria, mg/24 h
Baylis and Wilson (1989)14	Rat (MW)	Uninephrectomy	6	5	Arterial	36	7	95±5	36	7	21±9
Packham et al. (1991)61	Rabbit (NZW)	Uninephrectomy	5	3	MAP	4.5	5	87±3	4.5	5	0.07±0.02
Leaker_p-inf (1992)19	Rat (HE)	Five/sixth kidney ablation by kidney infarction	10	2.33 (range, 1–3)	SBP	20	9	139±7	20	10	110±40
Leaker et al._p-neph (1992)19	Rat (HE)	Five/sixth kidney ablation by nephrectomy	10	2.66 (range, 1–3)	SBP	20	10	151±9	20	7	210±20
Podjarny et al. (1992)20	Rat (W)	Adriamycin nephropathy	14	1	MAP	5.5	6	111.7±3	5.5	NR	NR
Baylis et al. (1995)13	Rat (SD)	Heymanns nephritis (Fx1A antibody injection)	8	1	Arterial	7	7	110±6	7	7	518±115
Deng and Baylis (1995)22	Rat (SD)	Five/sixth kidney ablation/infarction	8	1	Arterial	4	8	140±4	4	8	161±43
Podjarny et al. (1995)21	Rat (W)	Adriamycin nephropathy	6	1	MAP	5.5	6	111.7±3	5.5	6	265±45
Pomeranz et al. (1995)62	Rat (W)	Adriamycin nephropathy	15	2	MAP	16	5	110±1	16	5	115±26
Salas et al. (2003)63	Rat (SD)	Five/sixth nephrectomy	11	1	SBP	7	6	149±7.9	7	6	144b±26
Pedrycz et al. (2005)64	Rat (W)	Adriamycin nephropathy	8	1	Arterial	NR	NR	NR	4	8	84.4 SD 5.4
Han et al. (2022)65	Rat (SD)	Five/sixth nephrectomy	3	1	SBP/DBP	1	3	174/113.8 SD 5.7/21.1	1	3	2c SD 0

Author	Species (Strain)	Nephropathy Model	Unit	Kidney Function, Mean±SD
				Baseline, Time after Disease Induction	N	Controls	Control Group	Follow-Up Measurement Time Relative to Delivery	N	After Delivery
Baylis and Wilson (1989)14	Rat (MW)	Uninephrectomy	GFR, ml/min	36	7	1.27±0.16	Virgin neph	4–6 wk	6	1.17±0.22
Packham et al. (1991)61	Rabbit (NZW)	Uninephrectomy	Creatinine clearance, ml/min	4.5	5	6.0±0.89	Pregnant neph	1–4 wk	5	6.6±1.12
Leaker et al._p-inf (1992)19	Rat (HE)	Five/sixth kidney ablation by kidney infarction	SCr, mmol/l	20	10	0.1±0.06	Virgin neph	1–48 h	10	0.11±0.09
Leaker et al._p-neph (1992)19	Rat (HE)	Five/sixth kidney ablation by nephrectomy	SCr, mmol/l	20	8	0.17±0.17	Virgin neph	1–48 h	10	0.09±0.03
Podjarny et al. (1992)20	Rat (W)	Adriamycin nephropathy	GFR (inulin clearance), ml/min	5.5	6	1.94±1.08	Virgin neph	Gestational day 22	6	1.22±0.42
Baylis et al. (1995)13	Rat (SD)	Heymanns nephritis (Fx1A antibody injection)	GFR, ml/min	7	7	1.0±0.26	Virgin neph	Gestational day 17–19	8	0.8±0.28
Deng and Baylis (1995)22	Rat (SD)	Five/sixth kidney ablation/infarction	GFR, ml/min	4	8	0.63±0.14	Virgin neph	Gestational day 18–19	8	0.98±0.25
Podjarny et al. (1995)21	Rat (W)	Adriamycin nephropathy	GFR (inulin clearance), ml/min	5.5	6	1.93±0.10	Virgin neph	Gestational day 22	6	1.22±0.42
Pomeranz et al. (1995)62	Rat (W)	Adriamycin nephropathy	Inulin clearance, ml/min	15.5	5	0.79±0.2	Virgin neph	35 d	5	0.84±0.20
Salas et al. (2003)63	Rat (SD)	Five/sixth nephrectomy	Creatinine clearance, ml/min	7	6	0.70±0.24	Virgin neph	Gestational day 21	11	0.8±0.33
Pedrycz et al. (2005)64	Rat (W)	Adriamycin nephropathy	SCr, mg/dl	4	8	0.65±0.08	Virgin neph	4 wk	8	0.68±0.12
Han et al. (2022)65	Rat (SD)	Five/sixth nephrectomy	Creatinine, mg/dl	1	3	0.56±0.1	Pregnant neph	Gestational week 3 (end of pregnancy)	3	0.81±0.07

Mean SEM unless stated otherwise. Time after disease induction is presented in weeks.

DBP, diastolic BP; HE, Hooded-Evans; MAP, mean arterial pressure; MW, Munich Wistar; NR, not reported; NZW, New Zealand White; Pregnant neph, pregnant nephropathy controls; SBP, systolic BP; SCr, serum creatinine; SD, Sprague-Dawley; Virgin neph, virgin nephropathy controls, W, Wistar.

^aNonpregnant controls with nephropathy, either virgin controls or prepregnancy measurements from pregnant rats.

^bmg/16 hours.

^cMean protein score.

Human Studies

Study Characteristics

All studies were retrospective or prospective cohort studies. Definition of CKD was heterogeneous: 15/36 cohorts included various CKD etiologies. Ten of the 36 cohorts included more than 25% patients with advanced CKD. Thirteen of 36 cohorts consisted of more than 25% of patients or pregnancies with chronic hypertension. Most cohorts did not report prepregnancy proteinuria values (21/36 cohorts missing).

Effect of Pregnancy on Kidney Function

SMDs and Subanalyses

Sixteen human studies reported on the primary outcome of SMD of kidney function before versus after pregnancy. Meta-analysis showed a significantly better kidney function before pregnancy versus after pregnancy: SMD, −0.38 (95% CI, −0.55 to −0.22; P < 0.001) over a mean follow-up of 2.9 years (SD 0.6; Figure 2). Decline in kidney function was more pronounced in cohorts with more advanced CKD and prepregnancy chronic hypertensive patients (meta-regression β=−0.005 [95% CI, −0.01 to −0.0006; P = 0.03] and β=−0.01 [95% CI, −0.02 to −0.001; P = 0.03], respectively). There was no significant difference in kidney function before versus after pregnancy in mild CKD cohorts (SMD, −0.22 [95% CI, −0.56 to 0.13; P = 0.22]) over mean follow-up of 4.4 (SD 0.7) years (Supplemental Figure 1A). In advanced CKD cohorts, kidney function declined significantly (SMD, −0.55 [95% CI, −0.80 to −0.30; P < 0.001]) over mean follow-up of 2.2 (SD 0.8) years (Supplemental Figure 1B).

Figure 2.

Meta-analysis of human studies, SMD kidney function (both GFR and creatinine values) before versus after delivery. CI, confidence interval; KFnc pre, kidney function prepregnancy; KFnc pp, kidney function after pregnancy; SMD, standardized mean difference.

Cohorts including >25% patients with chronic hypertension showed a significant decline in kidney function (SMD, −0.45 [95% CI, −0.75 to −0.15; P = 0.004]; Supplemental Figure 2, A and B). Prepregnancy proteinuria did not affect study results (β=−1.10, P = 0.21) nor did follow-up time after delivery (β=0.03, P = 0.54), kidney disease etiology (β=−0.0007, P = 0.99), or publication year (β=0.0006, P = 0.92).

Pooled MDs

To translate SMD findings into clinical measures, we calculated pooled mean GFR difference before and after pregnancy. Pooled mean GFR difference was −5.2 ml/min (95% CI, −11.7 to 1.2; P = 0.11) and creatinine difference was 0.17 mg/dl (95% CI, −0.0047 to 0.35; P = 0.06) over a mean follow-up of 3.1 years (SD 0.6) and 2.9 years (SD 1.1), respectively. Subgroup analyses (Figure 3, A and B) showed a GFR change of −8.96 ml/min (95% CI, −17.4 to −0.5; P = 0.04) in advanced CKD and −1.5 ml/min (95% CI, −11.4 to 8.4; P = 0.76) in mild CKD cohorts, over a mean follow-up of 2.6 (SD 0.9) years and 4.1 (SD 0.8) years, respectively.

Figure 3.

Subanalysis of human studies with mild CKD and advanced CKD: MD of eGFR before versus after delivery. (A) Mild CKD (<25% patients with advanced CKD in cohort). (B) Advanced CKD (>25% advanced CKD in cohort). N pp, number of experimental animals (pregnant); N pre, number of controls; % adv CKD, percentage of patients with advanced CKD in the cohort; MD, mean difference.

Two mild CKD cohorts reported mean yearly decline in kidney function in pregnant patients with CKD versus nonpregnant CKD patients. No pregnancy effect was observed in these studies (SMD, −0.14 [95% CI, −0.31 to 0.04]).

KF/KRT in Patients with CKD after Pregnancy

Meta-analysis of 36 human studies showed a pooled incidence of KF/KRT after pregnancy of 9% ([95% CI, 6 to 13]; P < 0.001; I²=79%) over a mean follow-up of 6.5 years after delivery (SD 2.2, Supplemental Figure 3). The percentage of patients with advanced CKD in the cohort was a significant moderator (β=0.02, P < 0.001). Pooled KF/KRT incidence after pregnancy was 7% (95% CI, 4 to 10) in mild CKD and 17% (95% CI, 9 to 29) in advanced CKD cohorts, over mean follow-up of 8.9 years (SD 2.9) and 1.8 years (SD 0.6) years, respectively (Supplemental Figure 4, A and B). Chronic hypertension was a significant effect modifier (β=0.03; 95% CI, 0.01 to 0.04; P < 0.001), follow-up time after delivery (β=0.03; 95% CI, −0.03 to 0.08; P = 0.36), prepregnancy proteinuria (β=2.0; 95% CI, −0.25 to 4.25; P = 0.08), CKD etiology (β=−0.17; 95% CI, −0.45 to 0.10; P = 0.23), and publication year (β=−0.02; 95% CI, −0.05 to 0.005; P = 0.11) were not.

In studies controlled with nonpregnant CKD patients, pregnancy did not increase KF/KRT risk in patients with CKD (risk ratio, 1.07 [95% CI, 0.62 to 1.84]). Notably, only two of six studies met our advanced CKD definition.

Effect of Pregnancy on Kidney Function Deterioration

Supplemental Table 5 presents cohorts' definitions of kidney function deterioration. Pooled incidence of kidney function deterioration was 13% ([95% CI, 9 to 18], I²=84%) over a mean follow-up of 3.2 years after delivery (SD 0.9). The percentage of patients with advanced CKD was a significant moderator (β=0.01, P = 0.02). Pooled incidence of kidney function deterioration after pregnancy was 9% (95% CI, 5 to 15) in mild CKD cohorts versus 25% (95% CI, 19 to 33) in advanced CKD cohorts, over mean follow-up of 4.6 (SD 1.6) years and 1.6 (SD 0.6) years after delivery, respectively. Chronic hypertension was a significant modifier of the pooled estimate (β=0.02; 95% CI, 0.004 to 0.04; P = 0.02), whereas follow-up time after delivery (β=−0.04; 95% CI, −0.16 to 0.08; P = 0.47), prepregnancy proteinuria (β=1.13; 95% CI, −0.12 to 2.38; P = 0.08), and CKD etiology (β=−0.08; 95% CI, −0.44 to 0.28; P = 0.67) were not.

Animal Studies

Study Characteristics

In total, 12 animal studies were performed, 11 were performed in rats, and 1 in rabbits. Five studies (42%) induced CKD using five/sixth nephrectomy, four using adriamycin nephropathy (33%), two using uninephrectomy (17%), and one using Heymanns nephritis (8%). Most animal models showed hypertension and proteinuria, and ten studies (83%) reported decreased kidney function.¹⁸ Two studies (17%) used the pregnant rats as their own controls and ten studies (83%) used virgin controls.

Effect of Pregnancy on eGFR and SCr in Animals

Meta-analysis of 12 animal studies, including 104 animals and 183 pregnancies, showed no difference in kidney function after pregnancy compared with virgin controls (SMD, −0.18; 95% CI, −0.72 to 0.37; I²=64%, Figure 4). Type of nephropathy model (β=0.20; 95% CI, −0.17 to 0.56; P = 0.29), follow-up time after pregnancy (β=0.006; 95% CI, −0.03 to 0.04; P = 0.76), species (β=−0.78; 95% CI, −2.87 to 1.31; P = 0.47), and proteinuria (β=−0.0009; 95% CI, −0.005 to 0.003; P = 0.67) did not affect the outcome. MA was not performed for secondary outcomes (less than eight studies per outcome). See Supplemental Table 6 for details. In most studies, secondary outcomes showed no differences between pregnant animals and virgin controls (e.g., kidney histology^{13,14,19–22}: six of eight studies) or were even reduced in the pregnancy groups (e.g., glomerular capillary pressure was lower in pregnant animals in all three studies reporting it^13,14,22).

Figure 4.

Meta-analysis of animal studies, SMD kidney function after delivery versus virgin controls.

Publication Bias in Human and Animal Evidence

A funnel plot was created for the pooled KF/KRT incidence (>20 studies per outcome). Visual inspection did not show asymmetric results (Supplemental Figure 6).

Risk of Bias in Human and Animal Studies

Assessment of human studies showed overall moderate study quality. Domains describing the representativeness of exposed cohort, comparability, and attrition bias showed higher risks of bias (Supplemental Figure 5 and Supplemental Table 7). Risk of bias was considered unclear for most animal studies (Supplemental Figure 5 and Supplemental Table 8).

Discussion

This systematic review and meta-analysis of 36 human studies including 2945 patients with 4623 pregnancies and 12 animal studies with 104 animals and 183 pregnancies investigated the effect of pregnancy on kidney function after delivery in CKD. This study has four key findings. First, pregnancy has no effect on long-term kidney function after delivery in mild CKD cohorts (SMD, −0.22 [95% CI, −0.56 to 0.13]) over mean follow-up of 4.4 years. However, in advanced CKD cohorts, kidney function after pregnancy declined (SMD, −0.55 [95% CI, −0.80 to −0.30]) over mean follow-up of 2.2 years. For advanced CKD cohorts, pooled GFR drop after delivery was −8.96 (95% CI, −17.4 to −0.48) ml/min over a mean follow-up of 2.6 years, albeit with borderline significance (P = 0.04) and high heterogeneity (80%). Second, next to CKD stage, chronic hypertension is a moderator of SMD in kidney function after delivery (β=−0.01, P = 0.03). Proteinuria was not a significant moderator in analyses but was missing in 21 of 36 cohorts. Third, over a mean follow-up of 6.5 years (SD 2.2), the pooled incidence of KF/KRT after delivery is 9%. Finally, albeit after a short follow-up time, pregnancy does not affect kidney function after delivery in animal models likely mimicking mild CKD. Overall, human studies showed moderate study quality. Risk of bias was considered unclear in most animal studies. Future studies should enhance quality of reporting to facilitate risk of bias evaluations.

Our findings for mild CKD align with previous papers^2,4 and a review by Zhang et al.³ that reported no effect of pregnancy on CKD disease progression. In that review, incidences of KF and doubling of SCr levels in women with CKD and pregnancy were not increased compared with those without pregnancy (pooled incidence nonpregnant cohorts 18%). Relatively short follow-up after delivery in several studies, particularly in advanced CKD cohorts, limits conclusions on long-term kidney event incidence. In addition, the lacking validation of eGFR formulas for eGFR >90 ml/min per 1.73 m² hampers true interpretation of MDs in women with still normal kidney function based on eGFR.²³

Although not previously addressed in a systematic review, the kidney function decline after pregnancy observed in advanced CKD is consistent with the current clinical paradigm and with previous summary reports.^2,4 Isolating pregnancy's effect on kidney function from natural disease progression is challenging. Ideally, prepregnancy and postpregnancy eGFR slopes including multiple eGFR measurements per person should be compared. Only one study in women with advanced CKD by Wiles et al. has used this design, highlighting a significant gap in the literature.¹⁰ They reported an eGFR decline during pregnancy of −5.3 ml/min−equivalent to 2.5 “years of prepregnancy disease” progression. After this drop, eGFR slope was the same as before pregnancy: −2.1 ml/min per year.¹⁰ To compare with Wiles' findings, subtracting their −5.3 ml/min pregnancy drop from our overall decline of −8.96 ml/min gives a pooled eGFR loss of −3.66 ml/min outside pregnancy, corresponding to −1.98 ml/min per year. This comparable but slightly better eGFR slope reflects our advanced CKD definition (more than 25% of patients with advanced CKD per cohort, including mild cases) and should be interpreted cautiously due to limited studies (n=3 advanced CKD reporting on eGFR) and heterogeneity. The study by Wiles et al. provides a translation for prepregnancy counseling: an estimated loss of 2.5 years of kidney function for patients with advanced CKD.

The influence of moderators such as chronic hypertension on kidney outcomes was expected.^2,4,24 However, the absence of an effect of proteinuria in our analyses may be attributed to missing prepregnancy proteinuria values in 21 of 36 cohorts. More data are needed, as prepregnancy proteinuria is known to affect pregnancy outcomes.²⁵

As there is no official cutoff to define CKD severity in animal models, it remains unclear how closely these models align with human CKD stages. The lack of pregnancy-related effects—similar to mild CKD in humans—suggests that these models represent mild CKD.^26–28 Animal studies had short follow-ups. Inclusion of end-of-pregnancy eGFR measurements, potentially overestimating kidney function due to pregnancy-related changes, did not affect results in subgroup analyses for timing of measurement and follow-up duration. Some animal studies included only three animals per group, limiting reliability. Still, micropuncture experiments conducted in animal models^13,14,22 demonstrated that glomerular pressure is not elevated during pregnancy. No histologic damage to the glomeruli was seen, suggesting no negative effect on kidney structures during pregnancy. This aligns with human studies that found no difference in long-term kidney function after pregnancy in mild CKD.

This systematic review is the first to collate human and animal studies on effect of pregnancy on CKD progression. It offers additional mechanistic insights into clinical findings in mild CKD. In addition, it is the first to comprehensively investigate kidney function after pregnancy, using patients as their own controls instead of nonpregnant controls groups. Furthermore, it updates the 2015 systematic review by including recent studies on advanced CKD.^3,10 However, this review also had a few limitations. First, because studies with eGFR-slope data are scarce (only one in advanced CKD), we could not truly establish the effect of pregnancy on top of natural disease progression in advanced CKD. The same goes for the effect of chronic hypertension and proteinuria. Second, we analyzed eGFR and creatinine separately for clinical relevance, but these analyses contain a subset of studies included in the primary outcome analysis. Further subgroup analyses got small, reducing precision and reliability.

In addition, although the percentage of patients with advanced CKD per cohort was analyzed linearly in meta-regression analyses, for subgroup analyses, the results should be interpreted in the context of an arbitrary definition of advanced CKD (more than 25% of patients with advanced CKD per cohort). Third, while inclusion of studies without publication date restrictions is a strength of this review, it resulted in data spanning from the 1960s to 2020s. Substantial changes in health care practices, including CKD definitions and diagnosis, should be considered when interpreting the findings, although publication year did not affect the analyses. Owing to limited studies included in the meta-analysis of continuous outcomes, risk of publication bias could not be assessed.

Recommendations and Future Research

This review and meta-analysis can be used for prepregnancy counseling. It shows the reassuring result that pregnancy does not negatively affect long-term kidney function in patients with mild CKD. Therefore, patients with mild CKD should not be discouraged from pregnancy for concerns about kidney function decline. Patients with advanced CKD should be informed on the risks of losing kidney function in pregnancy. Animal models have provided additional mechanistic insights into clinical findings in mild CKD, but future research should prioritize human studies to ensure clinical relevance and applicability in daily practice. These studies should include multiple eGFR measurements over time, with longer follow-up periods, to accurately assess the effect of pregnancy on the eGFR slope compared with normal disease progression. Establishing larger prospective cohorts could help investigate the relationship between pregnancy, hypertension, proteinuria, and kidney function, enabling a more comprehensive assessment of who is truly at risk of kidney function decline while overcoming limitations of missing data and small sample sizes.

Supplementary Material

cjasn-20-1190-s001.pdf ^{(1.4MB, pdf)}

cjasn-20-1190-s002.pdf ^{(268.7KB, pdf)}

cjasn-20-1190-s003.pdf ^{(2.8MB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/CJN/C334.

Author Contributions

Conceptualization: Margriet E. Gosselink, A. Titia Lely, Rozemarijn Snoek, Kimberley E. Wever, Albertien M. van Eerde.

Data curation: Margriet E. Gosselink, Jolijn M.M. Sluijters, Rozemarijn Snoek.

Formal analysis: Margriet E. Gosselink, A. Titia Lely, Jolijn M.M. Sluijters, Rozemarijn Snoek, Kimberley E. Wever.

Investigation: Margriet E. Gosselink, A. Titia Lely, Jolijn M.M. Sluijters, Rozemarijn Snoek, Kimberley E. Wever, Albertien M. van Eerde.

Methodology: Margriet E. Gosselink, A. Titia Lely, Jolijn M.M. Sluijters, Rozemarijn Snoek, Kimberley E. Wever, Albertien M. van Eerde.

Project administration: Margriet E. Gosselink.

Supervision: A. Titia Lely, Kimberley E. Wever, Albertien M. van Eerde.

Visualization: Margriet E. Gosselink.

Writing – original draft: Margriet E. Gosselink.

Writing – review & editing: Margriet E. Gosselink, A. Titia Lely, Jolijn M.M. Sluijters, Rozemarijn Snoek, Kimberley E. Wever, Albertien M. van Eerde.

Funding

This work was supported by the ZonMW MKMD Synthesis of Evidence (114024176) and by the Dutch Kidney Foundation (Nierstichting, 24OSR032).

Data Availability Statements

All original data, including deidentified patient-level data or individual laboratory data measurements, are included in the manuscript and/or supplemental material.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/C335.

Supplemental Table 1. Methods deviations from prespecified study protocols in International Prospective Register of Systematic Reviews, including rationale of adjustments.

Supplemental Table 2. Search string human studies for PubMed and Embase.

Supplemental Table 3. Search string animal studies for PubMed and Embase.

Supplemental Table 4. Quality assessment criteria, adjustments to the Newcastle-Ottawa Scale for cohort studies.

Supplemental Table 5. Extended characteristics and data extraction table human studies.

Supplemental Table 6. Extended characteristics and data extraction table animal studies.

Supplemental Table 7. Quality assessment of included human studies.

Supplemental Table 8. Quality assessment of included animal studies.

Supplemental Figure 1, A and B. Subanalyses of SMD between kidney function before versus after pregnancy in cohorts with <25% (A) and >25% (B) patients with advanced CKD.

Supplemental Figure 2, A and B. Subanalyses of human studies, SMD between kidney function before versus after pregnancy in cohorts with <25% and >25% chronic hypertensive patients.

Supplemental Figure 3. Meta-analysis of human studies, pooled incidence KF/KRT after pregnancy, n=36 studies.

Supplemental Figure 4A. Subanalysis of pooled incidences for KF/KRT after pregnancy, mild CKD cohorts.

Supplemental Figure 4B. Subanalysis of pooled incidences for KF/KRT after pregnancy, advanced CKD cohorts.

Supplemental Figure 5. Overall summary figure of quality assessment human and animal studies.

Supplemental Figure 6. Funnel plot for secondary outcome KF/KRT.