Kidney biopsy in pregnancy: evidence for counselling? A systematic narrative review


Correspondence: Dr GB Piccoli, Struttura Semplice of Nephrology, Department of Clinical and Biological Sciences, University of San Luigi Gonzaga, Regione Gonzole 10, 10043 Orbassano, Torino, Italy. Email or



Kidney diseases, which have a prevalence of 3% in women of childbearing age, are increasingly encountered in pregnancy. Glomerulonephritis may develop or flare up in pregnancy, and a differential diagnosis with pre-eclampsia may be impossible on clinical grounds. Use of kidney biopsy is controversial, but a systematic review has not been carried out to date.


To review the literature on kidney biopsy in pregnancy, with a focus on indications, risks and timing.

Search strategy

Medline, Embase, CHINAL and the Cochrane Library were searched in September 2012, with ‘pregnancy’ and ‘kidney biopsy’ used as MESH and free terms, for the period 1980–2012. Results were filtered for ‘human’ if this option was available.

Selection criteria

Biopsies during pregnancy and within 2 months after delivery. Case reports (fewer than five cases) and kidney grafts were excluded. Paper selection was performed in duplicate.

Data collection and analysis

Data were extracted in duplicate. The high heterogeneity in study design necessitated that the review be narrative, except for data on adverse events, which were analysed with regard to the timing of kidney biopsy.

Main results

Of 949 references, 39 were selected, providing data on 243 biopsies in pregnancy and 1236 after delivery (timing was unclear in 106 women). The main aims of the studies were to define morphology in pre-eclampsia (23 studies), to carry out a risk–benefit analysis of kidney biopsy (11 studies), and to investigate pregnancy-related acute kidney injury (five studies). Four cases of major bleeding complications occurred at 23–26 weeks of gestation. Relevant complications were observed in 7% of women during pregnancy and 1% after delivery (P = 0.001). Kidney biopsy performed for the diagnosis of glomerulonephritis or pre-eclampsia led to therapeutic changes in 66% of cases.

Authors’ conclusions

The evidence on kidney biopsy in pregnancy is heterogeneous, but a significantly higher risk of complications (relative to postpartum biopsy) was found, with a possible peak at around 25 gestational weeks.


Many issues concerning whether, and in what circumstances, kidney biopsy should be performed in pregnant women remain unresolved[1-3]; these include indications for biopsy, for example sudden deterioration of kidney function before 32 weeks of gestation and differential diagnosis with pre-eclampsia,[4] and the effectiveness of empirical therapy, mainly with steroids.[5-12] The rapid healing of pre-eclamptic lesions led some authors to suggest postponing kidney biopsy until at least 2 months after delivery, whereas others have underlined the importance of early diagnosis and timely interventions in the presence of other glomerular diseases.[13, 14]

There are several good reasons for systematically reconsidering the pros and cons of kidney biopsy in pregnancy. Improvements in maternal and fetal care permit survival without major problems of babies born as early as 24 or 25 weeks of gestation, which changes the balance between the risks of early delivery and of invasive procedures.[15, 16] Conversely, the advantage of an early diagnosis of glomerular diseases is increasingly emphasised, but it is not clear whether this applies to pregnant women, in whom an additive role for hyperfiltration-induced damage has been suggested.[17-19]

The definitions of pre-eclampsia have changed in the last decade, and it is no longer synonymous with early delivery; the differential diagnosis between pre-eclampsia and underlying kidney disease remains an unmet challenge.[20-22]

Ethical issues are also changing; in a setting of shared decisions, kidney biopsy may be considered in specific situations, such as a systemic lupus erythematosus flare-up or rapid kidney function impairment in the presence of favourable fetal parameters, if it might help to identify therapies that will allow continuation of the pregnancy to a relatively safe stage in terms of fetal outcomes.[23-25]

Despite there being a great deal of theoretical interest in this issue, no systematic review has addressed kidney biopsy and pregnancy in the last few decades.

In this context, the aim of the study was to systematically review the literature on kidney biopsy in pregnancy, with the intention of clarifying the risks and benefits of this invasive procedure, according to PICOS criteria (explained in Data collection and analysis section). The analysis took into consideration the indications, risks and timing of the biopsy; biopsies performed in the 2 months following delivery served as controls for the incidence of complications.

Maternal and fetal adverse events recorded after a kidney biopsy in pregnancy were considered ‘risks’. The best outcome measure would be some measure of maternal and fetal morbidity; however, the small number of cases and the high heterogeneity in study design led us to choose changes in therapeutic strategies secondary to the biopsy results as a surrogate for ‘benefits’.

Search strategy

The study did not require a previous protocol. Medline (OVID), Embase, CHINAL and the Cochrane Library were searched. The first search was carried out in Medline (first week of November 2011), on kidney biopsy, Chronic Kidney Disease (CKD) and pregnancy. This was subsequently updated in September 2012 (second week), when searches in Embase, CHINAL and the Cochrane Library were also performed.

The search was deliberately broad to increase sensitivity, according to the guidelines of the Cochrane Collaboration. The search terms ‘pregnancy’ and, for kidney biopsy, ‘biopsy’, ‘morphology’ and ‘pathology’, combined with ‘renal’ or ‘kidney’, were employed as MESH and free terms. The limits of ‘human’ and publication after 1980 were employed when available.

The choice of 1980–2011 as the period of publication considered was motivated by the great advances in maternal and fetal medicine during this period and by the increased use of ultrasound-guided kidney biopsies since the early 1980s. Although the search was not limited to English, language barriers impaired the evaluation of eight papers in Japanese (all dealing with postpartum biopsies).

Selection criteria

The search was performed in duplicate, with the abstracts and titles being screened by GD and GBP (working independently and matching results). Reviews were screened for references. To reduce publication bias, we did not include case reports; only articles that reported on at least five women were selected. The final selection of the articles was agreed on and the data were extracted in duplicate. Women with kidney transplants were not considered.

Data collection and analysis

The data were analysed according to PICOS criteria, as follows. P (‘patients’): patients who underwent a kidney biopsy during pregnancy or within 2 months of delivery; I (‘interventions’): renal biopsy; C (‘control groups’): for the narrative review, as stated in the papers; for the comparison of the incidence of adverse effects, pregnancy versus postpartum; O (outcomes): morphological data for all women, changes of treatment and adverse effects; S (‘study design’): a narrative review of the main results, and a comparison (using the chi-square test) of adverse events in pregnancy versus postpartum.

The following data were extracted: title, author(s), objective, year, journal, period of study, single centre or multicentre, country, type of study, number of cases, indications for kidney biopsy, control group, maternal age, type of disease, known/new diagnosis, subcategories, parity, hypertension, pre-eclampsia, proteinuria, drugs, additional care, gestational age, birthweight, indication for delivery, preterm delivery, complication of the kidney biopsy including bleeding (major or minor) and other maternal complications, stillbirth/neonatal death, small for gestational age, admission to intensive care unit, other neonatal complications (whenever reported), and maternal and fetal follow up. Timing of kidney biopsy and kidney biopsy modalities were gathered for stratification of the risks. All available data on the morphological diagnoses and on the consequent changes of therapeutic approaches after the kidney biopsy were also extracted.

The papers were divided into three major categories: those reporting on kidney biopsy during pregnancy; those reporting on kidney biopsy both during pregnancy and after delivery; and those reporting on kidney biopsy after delivery only. Only series in which all or most women underwent a biopsy within 2 months of delivery were considered. Biopsies performed at the time of delivery (caesarean section) were considered together with biopsies performed during pregnancy. The older term ‘gestosis’ was considered to be equivalent to pre-eclampsia; as definitions changed over time, definitions were also extracted. The decision to perform a narrative review or a meta-analysis was subordinated to the type and quality of evidence retrieved. Because we were expecting high heterogeneity, a descriptive narrative review was planned for the main results and outcomes, while pooling of the data was planned to assess the incidence of complications. Analyses were performed using spss version 18.0 (SPSS, Chicago, IL, USA).

Main results

Retrieving the evidence and summary data

In our search for articles related to kidney biopsy and pregnancy, 51 full-text articles were retrieved from 949 references (297 from Medline, with the limit ‘human’, 632 from Embase, 17 from CHINAL and three from the Cochrane Library). After the exclusion of four papers as duplicates, 39 fulfilled the selection criteria (Figure 1).[26-64] Of these, five studies dealt with kidney biopsy in pregnancy only, six included women who underwent a kidney biopsy in pregnancy and after delivery, and 28 included biopsies performed only after delivery (Table 1).

Table 1. Baseline data
First authorPeriodCountryDesignObjective, as stated in the studyBiopsies (n)Control
  1. FSGS, focal segmental glomerular sclerosis; HELLP, haemolysis, elevated liver enzymes and low platelet count; NR, not reported; Pro, prospective; Ret, retrospective.

  2. a

    Presumably including biopsies after delivery (not specified in the text).

  3. b

    Presumably at least partial overlap of a few cases between studies.

  4. c

    Controls: female transplant donors (nonpregnant)

  5. d

    Controls: neoplasia and kidney transplant donors (nonpregnant nephrotic women)

  6. e

    Other cases of acute kidney failure included in the paper, not in pregnancy

  7. f

    Controls: nonpregnant women with FSGS.

  8. g

    Cases selected among 170 postpartum biopsies

  9. h

    The number of biopsies performed in pregnancy was not reported

  10. i

    Selection from a large cohort of women with PE, reclassified after delivery.

Kidney biopsy in pregnancy only
Zineb 2012(years)MoroccoRet To assess utility of renal biopsy in pregnancy (abstract)90
Wide-Swensson, 20071999–2001SwedenProTo assess the value and adverse effects of ultrasound-guided renal biopsy in women with normal and pathological pregnancies3618
Chen, 20011990–1999TaiwanRetTo verify the role of renal biopsy in pregnancies complicated by renal dysfunction15
Weiner, 1990NRUSAPro To analyse the biopsy in early-onset PE and to correlate the histological diagnosis with antithrombin III activity12 (intrapartum)
Packham, 1987b1965–1985AustraliaRetTo review the experience with renal biopsy in pregnancy: indications, histopathological findings and complications111 (104 women)
Kidney biopsy during pregnancy and within 2 months of delivery
Day, 20081983–2004UKRetTo investigate the role of renal biopsy in the management of women with renal disease in pregnancy95 (20 in pregnancy)
Kuller, 20011989–2000USARetTo review the experience with renal biopsy in pregnancy: indications, findings, complications and neonatal outcomes18 (15 in pregnancy) 
Stettler, 19921970–1990USARetTo determine the significance of ‘asymptomatic’ proteinuria identified during pregnancy21 (1 in pregnancy)
Hill, 1988b1973–1987AustraliaRetTo investigate the morphological changes in the glomerulus and juxtaglomerular apparatus in women with PE and no renal disease 10 (7 intrapartum)
Packham, 1988bNRAustraliaRetTo perform a morphometric analysis of glomeruli with a clinical and histological diagnosis of PE16 (6 in pregnancy)
Sommers, 1982NRUSARetTo study glomerular lesions associated with PE106 (82 women)h
Kidney biopsy after delivery only (within 2 months of delivery)
Debasmita, 20102007–2010IndiaProTo analyse the spectrum of acute kidney injury-related nephropathies 9
Hassan, 2009a2007–2008PakistanNR To determine the aetiology and outcome of acute kidney injury in pregnancy31
Goplani, 20082004–2006IndiaProTo evaluate the contributing factors and frequency of cortical necrosis in pregnancy-related acute kidney injury11
Prakash 20071984–2005IndiaRetTo describe the changing trends in renal cortical necrosis in acute kidney injury in eastern India32
Pires do Rio, 20041990–1997BrazilRetTo investigate the glomerular alterations by electron microscopy in severe PE and their evolution 39
Nagai, 2001b1981–1998JapanRetTo clarify the relationship between changes in contractile proteins in renal vascular walls and prognosis of hypertension during pregnancy207
Khedun, 2000bNRSouth AfricaRetTo determine the diagnostic yield of postpartum renal biopsy and the nature and frequency of its complications 50
Murakami, 20001980–1999JapanRetTo identify women with PE or severe gestational proteinuria at risk of underlying renal disease86
Nishimoto, 1999b1983–1997JapanRetTo evaluate glomerular hypertrophy in PE associated with FSGS lesions2610
Lafayette, 1998NRUSARetTo explore the degree and nature of glomerular dysfunction in PE108c
Gartner, 1998NRGermanyRetTo investigate the role of the endothelial lesion in PE90
Naicher, 1997bNRSouth AfricaProTo assess the anionic charge and the pathology in early-onset PE 15
Nagai, 1997bNRJapanRetTo observe changes in endothelin-1 in PE kidney tissue1910,12d
Lee, 19951985–1992KoreaRetTo define the structural characteristics related to FSGS in atypical PE118f
Sandoval Rodriguez, 19931990–1992MexicoProTo assess the impact of postpartum reclassification of hypertensive disorders of pregnancy7i
Grcevska, 1992NRMacedoniaRetTo report on glomerular sclerosis in biopsies performed after pregnancy in women with PE (letter)21
Nagai, 1991b1977–1988JapanProTo study the association of FSGS with renal biopsy in ‘pure’ PE19
Shiiki, 1990b1983–1988JapanRetTo clarify if FSGS lesions in PE represent a primary glomerular disease or a variant of PE nephropathy15
Diaz, 19901966–1981South AmericaRetTo review the experience with acute kidney injury of obstetric origin17
Olmer, 1987aNRFranceRetTo assess the role of renal biopsy after a vasculo-renal accident in pregnancy, for the prediction of hypertension and future pregnancies201
Nochy,1986b1975–1983FranceRetTo study de novo glomerular sclerosis associated with PE11g
Kida, 19851981–1984Japan Ret To clarify the relationship between PE and focal-segmental glomerulosclerosis16
Beller, 1985Last 10 yearsGermanyRetTo compare clinical and morphological data in women with PE and HELLP syndrome12 (HELLP)118 (PE)
Heaton, 19851965–1985UKRetTo investigate evidence of persistent renal damage following PE15 (13 women)
Raidt,1984NRGermanyProTo test the prognostic value of electrophoretic urinary protein patterns in pregnancy-induced hypertension92
Beaufils, 1980NRFranceRetTo determine the renal lesions in PE, their relationship with clinical presentation and the prognostic value of renal biopsy 125
Nochy, 1980bNRFranceRetTo analyse the renal lesions in hypertensive syndromes of pregnancy in comparison to typical PE117 (114 women)
Oe, 1980NRNetherlandsProTo examine the resolution of the glomerular changes in edema-proteinuria-hypertension-PE22
Figure 1.

Flow chart for the papers retrieved.

Overall, the review included 243 biopsies performed in pregnancy, of which 19 were performed at caesarean delivery and 1236 after delivery. In one paper (106 women) the timing was not clear.[35]

Most studies were carried out at single centres. Because case reports were not included, the number of observed women ranged from seven to 201. The largest series were the oldest ones, and all studies with more than 100 women were retrospective.

The studies had a wide range of origins: five were from North America, 12 from Europe, 13 from Asia (seven from Japan), three from Australia, three from South America and three from Africa. The studies were heterogeneous for duration (2–21 years) and period of study (1966–81 to 2007/08) (Table 1).

The aims and designs were also very heterogeneous. The most frequent aim was to define the renal lesion in pre-eclampsia and in pregnancy-induced hypertension (23 studies). The next most frequent aim was to assess the main risks of kidney biopsy or differential diagnosis with underlying kidney diseases (ten studies, plus one concerning reclassification of pregnancy-induced hypertension and pre-eclampsia), whereas five studies had the aim of analysing pregnancy-related acute kidney injury.

Seven studies included ‘normal controls’, but their definitions of ‘normal controls’ differed. Non-pregnant female kidney donors of childbearing age and women undergoing nephrectomies because of cancers were the most commonly used control groups. In one recent study, healthy pregnant women were used as controls.[27]

Indications for and timing of kidney biopsy

All studies reported the indications for the kidney biopsy, or, when retrospective analyses were performed, the selection criteria employed (Table 2). In accordance with the different study designs, the indications differed among studies. In pregnancy, the timing of the biopsy ranged from early pregnancy to 40 gestational weeks (Table 2). As a consequence of the broad inclusion criteria (all or most biopsies performed within 2 months of delivery, to account for the wide time ranges reported in most of the studies), the timing of the postpartum biopsy ranged from a few days to a few months after delivery, according to the indications (pre-eclampsia or gestational hypertension, or other kidney diseases).

Table 2. Indications for kidney biopsy and timing of the biopsy
First authorIndications for kidney biopsy—inclusion criteria Timing of the kidney biopsy
  1. AKI, acute kidney injury; FSGS, focal segmental glomerular sclerosis; HELLP, haemolysis, elevated liver enzymes and low platelet count; Ht, hypertension; PE, pre-eclampsia; PKD, polycystic kidney disease.

Kidney biopsy in pregnancy only
Zineb Nephrotic syndrome in seven women; macroscopic haematuria and hypertension in one eachGestational age 13 ± 5 weeks
Wide- SwenssonI: Pregnancy-induced Ht; mild PE; severe PE (proteinuria > 3000 mg/l); II: nonproteinuric Ht; III: control: 18 healthy pregnancies Gestational age: I: 33.8; II: 34.7; III: 35.2 weeks
ChenUndiagnosed renal disease during pregnancy (proteinuria, haematuria, hypertension or azotaemia).Gestational age: 22.3 (20–25) weeks
WeinerSevere PE < 34 weeks in women planning future pregnancy, if delivery by caesarean section is necessary At caesarean delivery: 30 (24–36) weeks
Packham Impaired renal function, nephrotic syndrome, haematuria, proteinuria or HtGestational age: 15 (4–28) weeks
Kidney biopsy during pregnancy and/or within 2 months of delivery
DayI: In pregnancy: decrease of renal function, proteinuria, autoimmune serology, worsening proteinuria or Ht; II: postpartum: proteinuria, haematuria or AKII: Gestational age 20.5 weeks; II: AKI: postpartum; otherwise ≥ 6 months
KullerI: In pregnancy: to differentiate between PE and other renal diseases; II: postpartum: worsening kidney disease or HtI: Gestational age: 25 (11–30) weeks; II: postpartum: 5 (4–19) days
StettlerNondiabetic proteinuria ≥ 500 mg/day, no known renal disease, no reversible dysfunction or PEIn pregnancy: gestational age 16 weeks; postpartum: NR
Hill PE in the absence of other kidney diseaseIntrapartum: 28 (18–33) weeks; postpartum: 4–10 days
PackhamPE: hypertension and proteinuria in pregnancy, which resolved in the early postpartum period. I: in pregnancy; II: post partum I: Gestational age: 28 (20–34) weeks; II: postpartum: 4 (3–9) days
SommersClinical toxaemia of pregnancyMid-pregnancy, near term and postpartum
Kidney biopsy after delivery only (within 2 months of delivery)
DebasmitaPatients with pregnancy-related AKI, needing dialysis and not recovering within 2 weeksPresumably after 2 weeks of dialysis
Hassan Previously healthy, who developed AKI without improvement by 3 weeksNR; presumably after > 3 weeks
GoplaniOliguria or dialysis dependence after 3 weeksNR
PrakashAKI with oliguria lasting more than 4 weeksPresumably after 1 month
Pires do RioSevere Ht in pregnancy; selected cases with electron microscopy availableAfter delivery; timing NR
NagaiI: 7; Ht after delivery; II: 5; normotensive after delivery, later Ht; III: 8; normotensive after delivery; IV: 7; healthy controls (5 normotensive women after delivery; 2 from nephrectomy)I: 15 ± 1 days; II: 12.5 ± 2.4 days;III: 13.9 ± 1.8 days
KhedunEarly-onset PE and Ht and proteinuria not resolving 14 days postpartumEarly postpartum (presumably minimum 2 weeks)
MurakamiSevere PE or severe gestational proteinuria (exclusion: refusal, PKD or thrombocytopenia); I: with underlying renal disease; II: without renal disease1–2 weeks after delivery
NishimotoClinical diagnosis of pure PE (proteinuria and hypertension in the third trimester); controls: nonpregnant women with isolated haematuriaI-II: 14.5 (5–150) days after delivery
LafayettePE: clinical diagnosis: Ht and proteinuria in the second half of pregnancy (dipstick proteinuria ≥ 2); controls: 8 kidney transplantsWithin 48 hours after delivery
Gartner Diagnosis of PE: hypertension and proteinuria (3 g or more); eclampsia: the former plus convulsions1/3 at delivery or caesarean section; 2/3 at various intervals, up to 4 years
NaicherHt, proteinuria and/or oedema between 24 and 34 weeks of gestation; controls:nephrectomy from trauma2 weeks after delivery
NagaiPE (proteinuria, Ht and oedema in late pregnancy); nephrotic syndrome; I: underlying kidney disease; II: normal kidney from nephrectomiesI: 16.7 ± 1 days after delivery; II: 10.8 ± 2.9 days after delivery
Lee Clinical features of PE; controls: 8 nonpregnant FSGSWithin 30 days after delivery
Sandoval Rodriguez Differential diagnosis between PE and nephropathy (reclassification of previous diagnoses of PE)After delivery; timing NR
GrcevskaDifferential diagnosis between PE and nephropathy in the case of persistent clinical signs 6 weeks after delivery6–9 weeks after delivery
NagaiTypical PE (proteinuria, oedema and Ht) without underlying renal disease and hypertensive disorders (clinical and histological criteria)PE + FSGS: 18 ± 1.9 days after delivery PE: 14 ± 2 days after delivery
ShiikiClinical: PE with nephrotic syndrome; no histological evidence of other kidney disease (except for FSGS)14 (7–25) days after delivery
Diaz Oliguric AKINR
Olmer PE; pregnancy-induced Ht and proteinuria; eclampsiaNR, presumably after delivery
NochyFSGS on pregnancy-induced nephropathy, selected from 170 postpartum biopsies from 150 hypertensive women6 women: 8 days after delivery; 4 women: 1 month; 1 woman: 3 months
KidaProteinuria > 1 g/day 1 month after delivery. Focus on 4/16 cases with FSGSNR for the whole group; FSGS: 5–22 weeks after delivery
BellerHELLP syndrome, in comparison with PE (previously published data)Up to 14 days postpartum
HeatonPE (proteinuria, oedema and Ht)5 days to 10 months
RaidtSevere PE (diastolic blood pressure ≥110 mmHg and clinical signs of PE)0–6 weeks after delivery
BeaufilsSevere hypertension and/or proteinuria during pregnancyNR, presumably after delivery
NochyHigh blood pressure and/or proteinuria during pregnancy and/or after delivery Scheduled on the 8th day after delivery
OePE (oedema-proteinuria-Ht-gestosis)10–14 days after delivery

The modality of the kidney biopsy was reported in less than half of women, but was reported in all studies focused on indications for and complications of renal biopsy.

Main results and conclusions

The main results and conclusions of the studies reviewed were generally heterogeneous, mainly as a reflection of the heterogeneity of the designs and analyses. Some conclusions were in disagreement: kidney biopsy in pregnancy was considered safe in two studies[28, 30] but a choice to be avoided where possible in others[27, 31]; the most recent study, available as an abstract, reported ‘uneventful recovery’ and drew very cautious conclusions[26] (Table 3). On other issues, there was substantial agreement: no clinical or serological marker was found to be able to discriminate between pre-eclampsia and underlying kidney disease, and the histological changes of pre-eclampsia did not reflect the severity of the disease (the finding of glomerular endotheliosis also in healthy controls in the most recent study on kidney biopsy in pregnancy further supports this conclusion).[27, 43, 56, 61]

Table 3. The main results
First authorNumber of BiopsiesMain resultsMain conclusions
  1. AKI, acute kidney injury; ESRD, ESRF, end-stage renal disease/failure; FSGS, focal segmental glomerulosclerosis; GFR, glomerular filtration rate; HELLP, haemolysis, elevated liver enzymes and low platelet count; Ht, hypertension; IgAGN, immunoglobulin A nephropathy; PE, pre-eclampsia; PGBM, peripheral glomerular basement membrane; SLE, systemic lupus erythematosus.

Kidney biopsy in pregnancy only
Zineb9The renal biopsy led to treatment definition in all cases (abstract)Renal biopsy is a morbid procedure to be considered only if it offers the opportunity to make a diagnosis other than severe PE in a woman remote from term (abstract)
Wide-Swensson54With or without proteinuria, all women with hypertension showed morphological changes typical for PE. Similar changes seen in 11/18 controls Glomerular endotheliosis is not pathognomonic for PE. It is not advisable to perform antepartum biopsies in cases of rapidly deteriorating function and swollen kidneys
Chen 15Different GN discovered, the most common being SLE nephropathy; therapy modified according to biopsy in 11 women. After 2 years: 7 women were in complete remission, 3 died and 2 had ESRFPregnancy does not increase the risk associated with renal biopsy. Biopsy should be performed if there is a likelihood that information will significantly change the therapy
Weiner 12The diagnosis of PE was confirmed in 9 women; 3/9 had underlying renal disease; no evidence of PE in 2 women. Antithrombin III predicted the histological changes of PE in 82% of casesHigh frequency of underlying disease in early-onset PE; normal antithrombin III activity is reassuring and more consistent with a non-PE renal lesion
Packham 104If women with previous PE and isolated microscopic haematuria were excluded, diagnosis of glomerulonephritis was made in 95% of cases (total: 80%)The bioptic procedure is safe and yields a definitive histological diagnosis in most cases, which may affect treatment and modify the subsequent clinical course
Kidney biopsy during pregnancy and/or within 2 months of delivery
Day 95Biopsy in pregnancy revealed a glomerular disorder in 95% of women. Postpartum glomerular abnormality was found in 64%. At last follow up (47 women), 20.7% had significant proteinuria, 42.6% had GFR < 60 ml/min and 12.7% had ESRDIn selected cases (unexplained nephrotic proteinuria, progressive CKD or renal disease in systemic disease), renal biopsy in pregnancy can allow progression to fetal viability. After 28 weeks, delivery should be induced and biopsy performed postpartum
Kuller 18Glomerular endotheliosis was found in 5 women only. Kidney biopsy complications occurred in 7 women. Fetal outcome was poor, apparently unrelated to the biopsy Renal biopsy in pregnancy is a morbid procedure, to be considered only for diagnosis other than PE in a woman remote from term (not be performed after 28–30 gestational weeks)
Stettler21Biopsy revealed underlying diseases: 7 membranous GN, 6 FSGS, 3 chronic GN, 2 chronic pyelonephritis, 1 minimal change GN, 1 IgAGN and 1 PKD. ESRD: 20%‘Asymptomatic’ proteinuria is associated with adverse pregnancy outcomes and long-term morbidity. Postpartum renal evaluation is of paramount importance
Hill10Juxtaglomerular regions not prominent and poorly granulated; ultrastructurally: myoepithelioid cells with sparse renin and scant endoplasmic reticulum and GolgiIn women with clinical and renal biopsy evidence of PE, there is no significant stimulation of the renin-angiotensin system
Packham16Biopsies during pregnancy showed electron-dense subendothelial deposits in capillary loops; these were rare or absent in postpartum biopsies In pregnancy, large subendothelial deposits, endothelial swelling and luminal encroachment. Postpartum disappearance of deposits but persistence of endotheliosis
Sommers 106Pure toxaemic changes (glomerular arteriolar, tubular and juxtaglomerular alterations) were present in 40% of women. Arteriolar nephrosclerosis and glomerulonephritis were also foundMore than half of the women had pre-existing lesions; their presence alters the ability of the kidney to compensate for the large increase in blood volume during pregnancy
Kidney biopsy after delivery only (within 2 months of delivery)
Debasmita9Renal biopsy performed in women not recovering after 2 weeks; 3 acute tubular necrosis, 2 cortical necrosis and 2 interstitial nephropathies; 6 recovered, 3 developed CKD and 1 developed ESRDPregnancy-related AKI is associated with high maternal mortality and poor fetal outcome
Hassan3121 women had tubular necrosis and 10 renal cortical necrosis. Of the 36 surviving women, 18 had complete recovery of renal function, 12 had partial recovery and 6 required dialysisPeripartum haemorrhage is the most common aetiological factor. Duration of anuria and renal replacement therapy are the predictors of long-term outcome
Goplani 11Renal biopsy revealed that 10 women had cortical necrosis and 1 had glomerular endotheliosisThe incidence of biopsy-proven renal cortical necrosis is high in postabortion sepsis
Prakash 32Incidence of cortical necrosis was higher in the postabortion groupThe incidence of post-abortal renal cortical necrosis is decreasing over time
Pires do Rio39The most frequent alterations: subendothelial fibrinoid deposits and podocyte fusion. Endothelial oedema in 84% of group I and in 92.2% of group II. No association between degree of Ht and severity of endothelial oedema; podocyte fusion associated with proteinuriaSubendothelial fibrinoid deposits, podocyte fusion, mesangial interposition and endothelial edema are a spectrum of complex and dynamic lesions characteristics of PE
Nagai 20Group I showed marked decreases in actin, SM-1 and SM-2, which were less frequent in group II. The small arteries (interlobar) were most affected in hypertensive women. Group I showed changes in afferent arterioles, which were rare in group IIPhenotypic changes in smooth muscle cells in renal vascular walls reflect the stage of Ht and progress from large to minimum-sized arterioles with the progression of hypertension. SM-2 is a sensitive marker of these changes
Khedun50Ultrastructural changes in 16 women with normal renal function. Glomerular sclerosis suggested underlying GNRenal biopsy is safe, and ultrastructural and histological findings obtained from postpartum renal biopsies are more informative than the routine renal function tests
Murakami86Underlying kidney disease in 22.1% of women. IgA nephropathy in 12/19. Women with underlying kidney disease had a significantly earlier onset of proteinuria (< 0.001)Women with gestational proteinuria or PE before the 30th gestational week are more likely to have underlying renal disease
Nishimoto26The mean glomerular tuft area, whole area and extracellular matrix area were negatively correlated with day at biopsy in PE with FSGS. Mesangial cells increased in PE with FSGS Glomerular hypertrophy that develops during severe PE plays an important role in the pathogenesis of FSGS lesions and is reversible about 40 days after delivery
Lafayette10Reduction in endothelial fenestrae; subendothelial fibrinoid deposits, lower glomerular permeability in PE. Mesangial cell interposition reduced filtration surfaceHypofiltration in PE is a consequence of structural changes impairing the intrinsic glomerular ultrafiltration capacity
Gartner 90The endothelial lesions display a close correlation between clinical and morphological data; FSGS presents as a hyperperfusion lesion, developing only facultatively in PEThe substrate is an endothelial lesion resulting in disturbance of permeability and leading to an imbalance in the mediator systems, with dominance of vasoconstrictive actions
Naicher15The mean number of anionic sites was significantly lower than that of the control group; significant correlation of proteinuria with number of anionic sites The loss of glomerular charge induces structural alterations of the glomerular filtration barrier and may be responsible for proteinuria in early-onset PE
Nagai 19Anti-ET-1 antibody showed negative staining in the glomerular capillary walls in both PE women and normal pregnant women with underlying glomerular diseaseThe production of ET-1 in glomeruli decreases after delivery, probably during pregnancy, and the condition may be caused by pregnancy itself
Lee 11Mesangial volume fraction increased in PE-FSGS, as compared with primary FSGS. PGBM surface area per glomerulus in PE-FSGS correlated with GFRDecreased glomerular filtration function in atypical PE is linked to decreased PGBM surface area resulting from mesangial expansion and from mesangial interposition
Sandoval Rodriguez 7158 women reclassified; the initial PE was confirmed in 33%. The kidney biopsy (7 cases) defined the presence of an underlying kidney disease not suspected in pregnancyThe postpartum reclassification is essential to avoid errors in diagnosis and to establish adequate maternal and perinatal outcomes in subsequent pregnancies
Grcevska21Different forms of glomerulonephritis found in 16 women; in 5 there was partial or complete glomerular sclerosis without signs of glomerulonephritis. These recovered during follow upGFS-like lesions can occur not infrequently in women with pure PE
Nagai 19Positive correlation between glomeruli with FSGS, proteinuria at delivery and weeks during which proteinuria ≥3 g/day continued after delivery FSGS-like lesions occur during pregnancy in ‘pure’ PE; lesions may disappear after delivery. Patients with FSGS-like lesions show more severe proteinuria than those without
Shiiki 15FSGS in 13 women (2–34% glomeruli). Variable extent of FSGS; early lesions predominated and may be an expression of PE. All women were nonproteinuric within 1–30 monthsPE associated with nephrotic syndrome frequently induces FSGS; prognosis is favourable. Discrimination with primary FSGS should be based on clinical and morphological criteria
Diaz 17The biopsy revealed acute tubular necrosis in 16 women and cortical necrosis in 1. Two died, and 9 underwent different surgical interventions. 12 women recovered kidney function Acute tubular necrosis is the most frequent lesion in obstetric oliguric AKI; functional recovery is possible
Olmer 201Clinical and biological alterations did not allow definition of the kidney lesions. Permanent Ht is frequent in women with predominant vascular lesionsRenal biopsy is the only means to determine the vascular state. Predominant vascular lesions suggest that Ht will recur in later pregnancy or will become permanent
Nochy11FSGS affected 10–50% of glomeruli: lesions mainly localised to a single area, affecting a small number of capillaries. 9/11 women normoproteinuric and normotensive within 3 monthsFSGS with pregnancy-induced nephropathy is frequent. FSGS may first appear in PE with the same mechanism of experimental models of FSGS with hyperfiltration
Kida164/16 women displayed hyalinosis, sclerosis, and swollen and foamy endocapillary cellsFSGS may develop during PE, and may follow a nonprogressive course after delivery
Beller1212 women displayed HELLP syndrome, out of 130 women with PEHELLP and ‘very severe’ PE are difficult to differentiate; severity is better expressed as liver involvement
Heaton13Endotheliosis could persist for weeks after delivery. FSGS, interstitial scarring and arteriolosclerosis suggested that permanent damage may follow PEPE is capable of producing persistent renal damage in a small but significant percentage of cases
Raidt92No close correlation between the results of urine analysis and kidney histology. A clear predominance of reversible glomerular protein patterns in pregnancy-induced HtElectrophoretic protein analysis may predict the risk of developing pregnancy-induced Ht and identify underlying glomerular diseases
Beaufils125The histological pattern could not be inferred from clinical criteria. 70% in groups I–II had metabolic disorders or a family history of Ht. Ht recurred in 70% of women in groups I–II; Ht was permanent in 47% of women in group II and 8% in group IHt in pregnancy affects a genetically prone population; renal biopsy is of little value in predicting the prognosis in future pregnancies. Renal biopsy should be performed in women with persistent proteinuria 3 months after delivery
Nochy114Proteinuric women showed pathological changes of PE. 42/62 other women showed PE. The changes were endothelial swelling, double contours and glomerular depositsHt and persistent hyperuricaemia in pregnancy are indicative of glomerular lesions of ‘pregnancy-induced nephropathy’
Oe22A diffuse increase in mesangial cells and matrix with varying intensity, compatible with mesangium activity more than endothelial proliferationIt is possible that the resolution of the characteristic renal lesions in toxaemia is promoted by mesangium activity

The definitions of pre-eclamptic lesions were not fully concordant. In all studies supplying a detailed description, they were defined as glomerular endotheliosis with endothelial swelling and subendothelial, fibrin-like deposits. Additional elements of the definitions, used in some studies but not others, were: an increase in the number of mesangial cells, mesangial swelling, mesangial deposits, double contours, glomerular hypercellularity and immunoglobulin M deposits.

The prevalences of the different types of underlying kidney disease reflected the indications for biopsy and the populations studied, a notable exception being focal segmental glomerular sclerosis, whose relationship with pre-eclampsia was not evident in the older papers. Focal segmental glomerular sclerosis-like lesions, almost uniformly found in the first stages of the disease, were later considered part of the pre-eclamptic syndrome in a series of studies, mainly by Japanese authors, but were not mentioned in several recent analyses[45, 50, 53, 57] (Table 3).


Categorising studies according to the timing of the biopsies, complications of the biopsy were reported in four of the five studies of kidney biopsies performed during pregnancy, in two of the six studies of biopsies performed both during and after pregnancy, and in only four of the 28 studies of biopsies performed after pregnancy, reflecting in part the high number of retrospective studies in the last subset (Tables 1 and 4).

Table 4. Main clinical problems and their timing (only papers reporting on them are included)
First authorNo. of biopsiesBleedingOtherRelationship between complication and timing of the biopsyObstetric outcomeEffect on the therapeutic approach
  1. IUGR, intrauterine growth restriction; NR, not reported; SGA, small for gestational age.

Kidney biopsy in pregnancy only
Zineb 9Not clear; generic statement of no major problemsNot specifiedNot clear; generic statement of no major problems2 therapeutic terminationsTherapy and termination decided on biopsy in 9/9
Wide-Swensson 361 large retroperitoneal haematoma needing embolisation, followed by placental abruption; 1 small perirenal haematomaLoin pain, 3 womenA serious problem occurred in a twin pregnancy referred at the 25 weeks of gestationRecovery in the mother and in the twins after the severe complicationNR
Chen 151 macroscopic haematuria (8 hours' duration); 15 microscopic haematuria NoneNR (biopsy before 25 weeks of gestation)14/15 live births (7 term and 7 preterm; 5 SGA); 1 stillborn11 women received steroid therapy (pulse and oral)
Weiner 12NRNRAll biopsies at delivery50% low birthweightNR
Packham1111 macroscopic haematuria; 1 perirenal haematoma 4 days after biopsy concomitant with antepartum haemorrhage (multiple transfusions)Mild loin pain, 3 women; in 3, insufficient tissueA severe complication at 25 weeks of gestation; macrohaematuria at 10 weeks1 neonatal death, after preterm birth, in the woman with perirenal haematoma Stated as potentially relevant in the Discussion
Kidney biopsy in pregnancy and within 2 months of delivery
Day 95 (20)1 macroscopic haematuriaNone In pregnancy, range 17–39 weeks; no complication in biopsies after delivery17/20 live infants; 2 abortions; 1 neonatal death9 women in pregnancy: immediate change of therapy
Kuller 18 (15)2 large perirenal haematomas needing transfusion, 1 with Mallory–Weiss tear; 5 haematomas without need of transfusion (3 × 1 to 4 × 6 cm)None Large haematomas: 23 and 26 weeks; smaller haematomas: 14–41 weeks. No complication in biopsies after delivery4 intrauterine fetal deaths; 6 neonatal deaths; 3 severe IUGR10 women received a histological diagnosis that allowed pregnancy to be prolonged
Hill 10 (7)NRNRNR6 live; 4 still birthsNR
Packham 16 (6)NRNRNR2 neonatal deaths; 4 fetal deaths; 2 still birthsNR
Kidney biopsy after delivery only (within 2 months of delivery)
Pires do Rio 392 cases (1 perirenal haematoma; 1 macrohaematuria)NRNRNRNR
Khedun 50All microscopic haematuriaLoin pain, 12 womenNRNRNR
Murakami861 macroscopic haematuriaNoneNRSGA: group I, 21.1%; II, 50.7% NR
Gartner 90NRNRNRStillbirth 35%NR
Naicher15 + ?NoneNoneNRFetal mortality rate 67%NR
Shiiki15NRNRNR16 infants: 12 SGA; 1 died of pulmonary haemorrhageNR
Beller12NRNRNR2 neonatal deaths; 2 intrauterine deathsNR
Olmer201NRNRNR75 fetal deathsNR

Fetal outcome was reported in 14 of the 39 studies. The range was wide: from no fetal deaths in the study by Wide-Swensson et al.,[27] published in 2007, to fetal death occurring in 37.5% of women in the study by Olmer et al. in 1987,[56] who analysed data on 201 biopsies performed after delivery. The period of the study presumably played a major role in fetal outcome, with improving results over time; differences in the prevalences of the different types of kidney disease may also have been important, but stratification was not possible in the retrieved papers (Table 4).

Interestingly, over the long period analysed, there is no clear trend of a reduction of risk of biopsy, suggesting that technical advances during this time had a minimal effect on risk.

Four cases of major complications of kidney biopsy were observed during pregnancy (representing about 2% of the 197 kidney biopsies performed during gestation that were included in papers also reporting on complications). In all four, there was major bleeding, with large perirenal haematomas, requiring blood transfusion. In a twin pregnancy, there was an association between kidney biopsy, placental abruption and preterm delivery; and in another a relationship between kidney biopsy and preterm delivery and fetal death could not be excluded.[27, 30, 32] The major complications occurred at a median of 25 weeks of gestation (range 23–26 weeks) (P = 0.015 for the comparison with biopsies performed after pregnancy) (Table 4).

Excluding microhaematuria, minor but relevant complications (smaller haematomas that did not require a transfusion and macrohaematuria of several hours' duration with strong loin pain) were observed in a further 5% of cases. The timing of these events was not mentioned in most studies, and could not be determined accurately from the information provided (Table 4).

In the few studies providing relevant information on biopsies performed after delivery (a total of 268 biopsies), only three relatively minor problems were reported: one macrohaematuria and two haematomas (i.e. complications were reported in about 1.3% of biopsies). Loin pain was cited in 12 women in one study; however, as the Abstract and Discussion of this paper reported that ‘no complication’ had occurred, this loin pain was assumed to have been mild and was not considered as a complication in our analysis.[43] The figure of 1.3% for the rate of complications is significantly lower than that found during pregnancy (= 0.008).

Five studies provided information on the therapeutic changes made on the basis of the renal biopsy results during pregnancy.[26, 28, 30-32] The aim of these studies was to investigate differential diagnosis with pre-eclampsia and they included 59 kidney biopsies; a therapeutic change (mainly in steroid treatment) was made in 39 of these women (66.1%). No study reported this outcome after pregnancy, but the relationship between diagnosis and therapy was presumably taken for granted in this context.


Main findings

Our systematic search, encompassing the last three decades, considered articles reporting data on at least five kidney biopsies in pregnancy or within 2 months of delivery. Most of the retrieved data relate to the postpartum period, and this in itself provides indirect evidence of the reluctance of nephrologists to perform kidney biopsies in pregnancy (Tables 1–4). The high heterogeneity of cases, study designs and biopsy indications allowed pooling of data only for the analysis of the risks associated with kidney biopsy (Tables 1–4).

The largest body of evidence concerns the morphology of pre-eclampsia (22 of 39 papers). A pre-eclampsia-related lesion was defined as glomerular endotheliosis with endothelial swelling and subendothelial, fibrin-like deposits; additional elements of this definition, used in some studies but not others, were an increase in the number of mesangial cells, mesangial swelling and deposits, double contours and IgM deposits. Focal segmental glomerular sclerosis-like lesions (usually mild and often found when specifically searched for) are hypothesised to reflect hyperfiltration-induced kidney injury. Their role must still be defined, also in the development of long-term kidney damage, and these lesions were not reported in several recent studies (Table 3).[27, 28, 31-33]

There was substantial agreement among the studies on a negative observation: histological changes do not reflect the severity of pre-eclampsia; after delivery, the timing of the kidney biopsy may be a confounding factor, as most pre-eclamptic changes resolve within a few weeks (Table 3).

The finding of glomerular endotheliosis in 11 of 18 healthy controls further supports the lack of correlation between histological changes and clinical presentation.[8, 27] The study in which that finding was obtained, which was a duplicate publication, required women without risk factors to undergo a kidney biopsy in the course of a normal pregnancy, and is therefore noteworthy in terms of both theoretical interest and ethics. The study design, approved by a local ethics committee, was felt to be unethical by the obstetrics and nephrology community, and provoked an intense ethical debate.[5, 65-70]

The second largest set of papers dealt with the risks and benefits of a kidney biopsy in pregnancy or shortly after delivery (Tables 1, 2 and 4).

Complications were mentioned in only about half of the studies that included biopsies during pregnancy and in only four of the 23 studies dealing with biopsies after delivery, which was possibly a reflection of the high number of retrospective studies in the latter subset (Tables 1 and 4). Information on timing was available for severe adverse events only. Within these limits, biopsy-related adverse events were stratified according to the stage of pregnancy at which they occurred: ‘early’ pregnancy, from 0 to 21 weeks (i.e. before fetal viability, set at 22 weeks); the ‘grey phase’, from 23 to 28 weeks (in which resuscitation of the newborn is attempted but the prognosis is poor); ‘late’ pregnancy, from 28 weeks to term (when the probability of survival of the newborn without long-term sequelae progressively increases); and after term.[71, 72]

In ‘early’ pregnancy, no severe complications were recorded; however, macrohaematuria was reported in one woman at 10 weeks of gestation, and at least one haematoma not requiring a transfusion occurred in this period.[27, 29]

The ‘grey phase’ was the period in which all the severe biopsy-related complications were recorded (recorded in the period 23–26 weeks of gestation). There were four cases of major bleeding (2% of the kidney biopsies performed in pregnancy) with the need for a transfusion; one of the women with major bleeding required embolisation and one had a Mallory–Weiss strain. The bleeding was associated with severe obstetric complications, early preterm deliveries and in one case a presumably related fetal death.[25, 28, 30] Five further relevant haematomas not requiring a transfusion (3 × 1 to 4 × 6 cm) were reported (median timing of the kidney biopsy: 25 weeks).[30]

In ‘late’ pregnancy, six cases of loin pain and five perirenal haematomas were reported (Table 4).[25, 28-30]

After delivery, the overall prevalence of complications, all of which were mild, was about 1% (two haematomas and one macrohaematuria in 268 biopsies). The incidence was significantly lower than the incidence of 7% observed in biopsies performed in pregnancy (= 0.008) (Table 4).

Comparison with the literature for nonpregnant women is difficult, partly because recent series included elderly and high-risk women and because definitions of the ‘severity’ of complications vary. However, in recent large series, life-threatening complications were reported in about 0.1% of cases, suggesting the presence of pregnancy-related risks, possibly resulting from the increased blood flow through the kidneys.[73, 74]

It was not possible to investigate the influence on therapy in the different periods of pregnancy previously identified, as only four papers reported on this issue, all of which had the aim of examining differential diagnosis with pre-eclampsia during pregnancy, and which included 59 biopsies. Therapy mainly consisted of steroids, and was initiated or modified as a result of the biopsy results in 66.1% of cases (Table 4). No study reported on this outcome after pregnancy, but the relationship between diagnosis and therapy was presumably taken for granted in this context.

Strengths and limitations

The strengths of our study are that it took a novel, systematic approach to the old problem of the pros and cons of kidney biopsy in pregnancy and that it has enabled a working hypothesis to be proposed, to be tested in future work, namely that the risk–benefit ratio changes according to gestational age.

Systematic reviews reflect the biases and limits of the retrieved evidence, and in our study the high heterogeneity of the original studies is reflected in the tentative nature of the conclusions that can be drawn regarding the risk–benefit ratio of a kidney biopsy in pregnancy. The limits of the present evidence may demonstrate the need for more detailed reporting of the timing of kidney biopsies, their adverse effects and therapeutic changes in future studies.


This review supports the conclusions of previous nonsystematic studies in that it suggests that kidney biopsy in pregnancy is a morbid procedure, with a high risk of severe complications, that should be limited to women in whom a diagnosis is needed for urgent therapy[4] (Tables 1–4).

Our analysis also suggests that the ‘grey area’ of mid-pregnancy may be the period in which there is the highest risk of severe maternal and fetal adverse events, as most of the severe adverse events occurred in this period, although in terms of average timing kidney biopsies were distributed more or less evenly throughout pregnancy (Tables 2 and 4). However, no stage of pregnancy seemed to be free of considerable morbidity, the incidence of which appeared to be much higher during pregnancy than after delivery. Although no severe adverse events were recorded after delivery, the occurrence of minor but relevant adverse events was relatively high (1% of women), suggesting a gradual decrease of the risks, possibly as a consequence of the adjustment of the kidney function after pregnancy.[15]

In pregnancy, the risks of kidney biopsy are partly counterbalanced by an approximately 66% probability of therapeutic changes. Novel diagnostic approaches, such as the possibility of testing for anti-PLA2R (M-type phospholipase A2 receptor) to detect membranous nephropathy[75] or of analysing the patterns of proteinuria, and therapeutic approaches that were not available when the largest studies were carried out may further change the indications for kidney biopsy in pregnancy.[76] The discussion of these options, in guiding empirical therapy for glomerular diseases in pregnancy, is far beyond the scope of this review, which, however, suggests that pregnant women should be among those tested for novel serum or urine markers of glomerular diseases or pre-eclampsia.[77-79]


The evidence regarding the pros and cons of kidney biopsy in pregnancy was found to be heterogeneous; however, the risks of complications were higher in pregnancy than in the postpartum period (7% versus 1%, respectively).

Our review suggests that the procedure should be limited to women with a suspicion of glomerular disease severe enough to warrant immediate treatment. The risks and advantages of empirical therapeutic approaches have also to be considered in each case. As evidence is still scant, we should also inform women about the limits of the present knowledge, emphasising that, although the first kidney biopsies in pregnancy were carried out in the late 1950s, the treatment of a pregnant woman with proteinuria suspected to have a cause other than PE is still based on sporadic and, to some extent, experimental experiences.

Disclosure of interests

The authors declare that they have no conflict of interest.

Contribution to authorship

GBP and RA designed the study; GBP drafted the manuscript. CN, DR and TT participated in the writing of the final version of the manuscript. GD and GBP selected the papers, extracted the data and drafted the tables. NC retrieved the evidence and supervised the searches. SP, FF, MF and MCD participated in the bibliographical searches, the drafting of the tables and the writing of the manuscript.


No financial support was received for this study.


The authors thank Dr Peter Christie for help with English language, and the staff of the library of the ASOU san Luigi for their invaluable assistance in retrieving the selected papers.