A predictive model of chronic kidney disease in patients with congenital anomalies of the kidney and...
Transcript of A predictive model of chronic kidney disease in patients with congenital anomalies of the kidney and...
1 23
Pediatric NephrologyJournal of the International PediatricNephrology Association ISSN 0931-041X Pediatr NephrolDOI 10.1007/s00467-014-2870-z
A predictive model of chronic kidneydisease in patients with congenitalanomalies of the kidney and urinary tract
Isabel G. Quirino, Cristiane S. Dias,Mariana A. Vasconcelos, IsabelV. Poggiali, Kerlane C. Gouvea,Alamanda K. Pereira, et al.
1 23
Your article is protected by copyright and all
rights are held exclusively by IPNA. This e-
offprint is for personal use only and shall not
be self-archived in electronic repositories. If
you wish to self-archive your article, please
use the accepted manuscript version for
posting on your own website. You may
further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.
ORIGINAL ARTICLE
A predictive model of chronic kidney disease in patientswith congenital anomalies of the kidney and urinary tract
Isabel G. Quirino & Cristiane S. Dias & Mariana A. Vasconcelos & Isabel V. Poggiali &Kerlane C. Gouvea & Alamanda K. Pereira & Gabriela P. Paulinelli & Amanda R. Moura &
Raquel S. Ferreira & Enrico A. Colosimo & Ana Cristina Simões e Silva &
Eduardo A. Oliveira
Received: 16 December 2013 /Revised: 22 May 2014 /Accepted: 28 May 2014# IPNA 2014
AbstractBackground The antenatal detection of congenital anomaliesof the kidney and urinary tract (CAKUT) has permitted earlymanagement of these conditions. The aim of this study was toidentify predictive factors associated with chronic kidneydisease (CKD) in CAKUT. We also propose a risk score ofCKD.Methods In this cohort study, 822 patients with prenatallydetected CAKUT were followed up for a median time of43 months. The primary outcome was CKD stage III or
higher. A predictive model was developed using the Coxproportional hazards model and evaluated by using cstatistics.Results Chronic kidney disease occurred in 49 of the 822(6 %) children with prenatally detected CAKUT. The mostaccurate model included bilateral hydronephrosis,oligohydramnios, estimated glomerular filtration rate andpostnatal diagnosis. The accuracy of the score was 0.95[95 % confidence interval (CI) 0.89–0.99] and 0.92 (95 %CI 0.86–0.95) after a follow-up of 2 and 10 years, respective-ly. Based on survival curves, we estimated that at 10 years ofage, the probability of survival without CKD stage III wasapproximately 98 and 58 % for the patients assigned to thelow-risk and high-risk groups, respectively (p<0.001).Conclusions Our predictive model of CKD may contribute toan early identification of a subgroup of patients at high risk forrenal impairment. It should be pointed out, however, that thismodel requires external validation in a different cohort.
Keywords Antenatal diagnosis . Congenital anomalies of thekidney and urinary tract . Chronic kidney disease . Clinicalpredictivemodel . Fetal hydronephrosis . Survival analysis
Introduction
The ability to impact nephrology conditions found in thedeveloping fetus and newborn is of critical importance [1].The antenatal detection of congenital anomalies of the kidneyand urinary tract (CAKUT) has enabled the early managementof these conditions [2–4]. However, although prenatal diag-nosis of CAKUT has been recognised as a prominent featureof such conditions for more than two decades, only limiteddata are as yet available on the risk factors related to the
This study was presented in part at the 16th Congress of InternationalPediatric Nephrology Association, August 30–September 3, 2013,Shanghai, China, and published in abstract form (Pediatr Nephrol 2013;28:1370).
I. G. Quirino : C. S. Dias :M. A. Vasconcelos : I. V. Poggiali :K. C. Gouvea :A. C. Simões e Silva : E. A. OliveiraPediatric Nephrology Unit, Department of Pediatrics, NationalInstitute of Science and Technology (INCT) of Molecular Medicine,Faculty ofMedicine, Federal University ofMinas Gerais, Av AlfredoBalena, 190, Belo Horizonte, MG 30130-100, Brazil
R. S. Ferreira : E. A. ColosimoStatistics Department, Federal University of Minas Gerais, BeloHorizonte, MG, Brazil
A. K. PereiraFetal Medicine Unit, Faculty of Medicine, Federal Universityof Minas Gerais, Belo Horizonte, MG, Brazil
G. P. Paulinelli :A. R. MouraDepartment of Pediatrics, Faculty of Medicine, Federal Universityof Minas Gerais, Belo Horizonte, MG, Brazil
E. A. Oliveira (*)Rua Engenheiro Amaro Lanari 389/501, Belo Horizonte,MG 30310-580, Brazile-mail: [email protected]
Pediatr NephrolDOI 10.1007/s00467-014-2870-z
Author's personal copy
occurrence of chronic kidney disease (CKD) in these patients[5]. Moreover, to establish a tailored medical management ofCAKUT, the clinician needs to be aware of the factors/diagnostic markers which may predict the occurrence ofCKD [6]. This issue, however, has not been clearly addressedin the current medical literature.
We have recently published a retrospective cohort study inwhich we describe the clinical course of 822 children withprenatally detected nephropathies [7]. In the study reportedhere, we extended our analysis with the aim of identifyingpotential risk factors for the occurrence of CKD stage III orhigher in a long-term follow-up. We also propose a score tostratify the risk for CKD in patients with CAKUT.
Patients and methods
Patients
At the Fetal Medicine Unit, each fetus underwent a detailedanatomic study aimed at detecting CAKUT and other organabnormalities following procedures described previously [8].Briefly, specific ultrasound (US) findings, such as the pres-ence of renal, ureteral and bladder involvement, and volumeof amniotic fluid, were recorded, and the records of 914patients diagnosed with CAKUTwere then reviewed. Fetuseswith aneuploidy or multiple malformations were excludedfrom the analysis (n=82), as were ten patients lost to follow-up soon after birth. Therefore, 822 infants with prenatallydetected CAKUTwhowere admitted to the Pediatric Nephrol-ogy Unit of the University Hospital, Federal University ofMinas Gerais (UFMG, Brazil) between 1989 and 2009 wereincluded in this retrospective cohort study.
Clinical protocol
During the 20 years of this study, the systematic approach tothe postnatal investigations, treatment and follow-up of in-fants with prenatally detected CAKUT at our Unit haschanged. In the first decade of the study, infants were inves-tigated according to a comprehensive systematic protocoldescribed elsewhere [9]. After 2000, we developed a moretailored clinical protocol [10]. Briefly, laboratory tests, such asserum creatinine and urine analysis at baseline, were per-formed during the first month of life, and subsequent testswere scheduled at 4- to 6-month intervals or according to theclinical needs of the patient. In terms of imaging tests, a USscan was performed after the first week of postnatal life, andall infants underwent a voiding cystourethrogram (VCUG).Antibiotic prophylaxis was started on the first postnatal dayand maintained in accordance with the definitive diagnosis.US scans, clinical examination (including growth and bloodpressure measurements) and laboratory reviews (including
urine culture and serum creatinine) were scheduled at 6-month intervals. When the VCUG was normal but postnatalUS scans demonstrated renal pelvis dilatation (RPD) of≥10 mm, renal scintigraphy was performed after the firstmonth [11–14]. Since 2009, VCUG is only performed in aselected subgroup of patients [15].
Outcomes
The event of interest was the occurrence of CKD stage III orhigher. CKD stage III was defined as an estimated glomerularfiltration rate (eGFR) of ≤60 ml/min per 1.73 m2 in twoconsecutive examinations. CKD stages were classified ac-cording to the National Kidney Foundation practice guidelines[16]. GFR was estimated by the formula of Schwartz et al.[17].
Variables
The following candidate predictive variables were included inthe analysis: gender, RPD laterality (unilateral vs bilateral),fetal US findings (fetal hydronephrosis alone vshydronephrosis associated with lower urinary tract abnormal-ities), olygohydramnios, period of diagnosis (1989 to 1999 vs2000 to 2009), and eGFR at baseline.
Clinical definitions
Combined data obtained by VCUG, renal scintigraphy andsequential US were considered for the diagnosis of urinarytract anomalies. Data from post-mortem examinations werealso taken into consideration for the diagnosis when appropri-ate. The absence of any recognised uropathy was reason toclassify the condition as idiopathic hydronephrosis.Hydronephrosis alone was defined as the presence of RPD≥10 mm without any other alterations of the urinary tract.Associated hydronephrosis was defined as the presence ofRPD >10 mm combined with other alterations, such asmegaureter and megacystis. Oligohydramnios was deter-mined on the basis of the amniotic fluid index [18].
Statistical analysis and development of the risk predictionmodel
In this study all values are expressed as medians and inter-quartile range (IQR) or means and standard deviation (SD),where appropriate. The Mann–Whitney or Kruskal–Wallistest was used to compare medians of continuous variables.Univariate analyses of continuous and categorical prognosticfactors were performed using Cox regression. Missing valuesfor candidate predictive variables were replaced with thecohort mean. A Cox proportional hazards model was appliedto identify variables at baseline that were independently
Pediatr Nephrol
Author's personal copy
associated with the occurrence of CKD stage III or higher.Only those variables collected at admission that were associ-ated with the event of interest by univariate analysis (p<0.25)were included in the initial Cox regression model. Then, usinga backward elimination strategy, we included in the finalmodel those variables that retained a significant independentassociation. All reported P values are two-sided, and a P valueof <0.05 was considered to be statistically significant.
A predictive model was then built from these data bydividing each estimatedβ coefficient in the final multivariablemodel with significant predictors by the lowest estimated β.Theβ coefficients were used for factor weighting; points wereassigned to each independent prognostic factor, its coefficientbeing rounded to the nearest integer [19, 20]. Finally, a pre-diction score was calculated for each patient by summing upthe points. The prognostic scores derived were then groupedinto two categories: low risk and high risk. We assessed thepredictive accuracy of the derived model by looking at thecomponents of accuracy, i.e. discrimination and calibration[19, 20]. Discrimination was evaluated using the c statistic thatrepresents the area under the receiver operating characteristiccurve (for which larger values indicate better discrimination)[21]. To assess model calibration, or how closely the predictedprobabilities reflect actual risk, observed risk was calculatedon the basis of 2, 5, and 10 years of follow-up [21, 22]. Weused Kaplan–Meier univariate survival analysis by comparingobserved and predicted risk for each of the three categories ofrisk for the outcome (low risk, medium and high risk, respec-tively) [23, 24]. Calibration was also assessed graphically on aKaplan–Meier plot of survival without CKD for patients ofdifferent risk groups [25]. To adjust for overfitting and over-optimistic performance of the model, we performed an inter-nal validation of our model with a bootstrapping technique. Ineach bootstrap sample, the entire modeling process was re-peated, resulting in shrinkage of the regression coefficientswhen applicable [26–28].
Ethical aspects
The study was approved by the Ethics Committee of theUFMG.
Results
Baseline findings and outcome
The study cohort comprised 822 infants with prenatally de-tected nephropathies. Median follow-up time was 43 (IQR18.6–94.7) months for those patients who survived the neo-natal period, with 310 (37.7 %) patients followed up for>5 years and 126 (15 %) followed up for >10 years. Of the
822 patients, 35 (4%) developed CKD stage III or higher witha median GFR of 18.0 (IQR 9.2–29.0) ml/min per 1.73 m2.The median age of patients who reached the target outcomewas 14 (IQR 4–32) months. At the end of the follow-upperiod, nine patients had CKD stage III, two had CKD stageIV and 24 (2.8 %) had reached CKD stage V. Using Kaplan–Meier curves, we estimated that the probability of CKD stageIII or higher for patients with prenatally detected nephropa-thies was 5 % at age 10 years. The laboratory evaluation at thelast visit of the follow-up period showed an eGFR of 100 (IQR90.4–109.3) ml/min per 1.73 m2 for patients who did notprogress to CKD.
Univariate survival analysis
Univariate analyses suggested that associated hydronephrosis(p < 0.001) , b i la teral hydronephrosis (p < 0.001) ,oligohydramnios (p<0.001) and eGFR (p<0.001) were asso-ciated with the occurrence of CKD. The main baseline clinicalcharacteristics associated with progression to CKD stage III orhigher are summarised in Table 1.
Multivariate analysis
After adjustment by the Cox regression model, four variablesremained as independent predictors of CKD in children withprenatally detected nephropathies: bilateral hydronephrosis,oligohydramnios, eGFR (as a continuous variable) and asso-ciated hydronephrosis. The shrinkage factor obtained frombootstrap results was 0.0457 (Table 2).
A prognostic weighting score was derived for each variableby dividing each estimated β coefficient by the lowest one.The highest prognostic weighting score for a dichotomousvariable was bilateral hydronephrosis (7 points), with associ-ated hydronephrosis and olygohydramnios having prognosticweighting scores of 4 and 3 points, respectively. eGFR levelsat baseline were divided into six categories: 0 points (>95 ml/min/1.73 m2), 1 point (79–94 ml/min/1.73 m2), 2 points (63–78 ml/min/1.73 m2), 3 points (47–62 ml/min/1.73 m2), 4points (31–46 ml/min/1.73 m2) and 5 points (15–30 ml/min/1.73 m2). A total prognostic risk score was then calculated bysumming these weighting scores for the four variables. A riskscore was calculated for each patient by adding up thesepoints. The risk score ranged from 0 to 19 points (median 5points). The accuracy of the score applied to the sample washigh, with a mean c statistic of 0.953 [95 % confidenceinterval (CI) 0.932–0.969], 0.962 (95 % CI 0.936–0.9809),0.933 (95 % CI 0.881–0.967) and 0.899 (95 % CI 0.807–0.957) for the follow-up periods of 2, 5, 10 and 15 years,respectively. Figure 1 shows the capacity of differences in therisk score to predict CKD stage III or higher during up to 5 and10 years of follow-up.
Pediatr Nephrol
Author's personal copy
Finally, the prognostic risk score for progression to CKDwas divided into three categories: low-risk (0–5 points),medium-risk (6–13 points) and high-risk groups (14–19points). Of the 822 patients, 449 (54.6 %) were assigned tothe low risk group, 322 (39.2 %) to the medium-risk group and51 (6.2 %) to the high-risk group. The model demonstratedbetter calibration for 2 and 5 years of follow-up for all riskcategories, while plots for 10 and 15 years of follow-up hadlarger deviations between the observed and predicted proba-bilities for the high-risk group. Figure 2 illustrates the calibra-tion plots for the model of risk prediction during 2 and 10 yearsof follow-up, respectively. Model calibration was also assessedby survival analysis (Fig. 3). We estimated that, at 10 years ofage, the probability of survival without CKD stage III wasabout 98, 88 and 58 % for the patients assigned to the low-risk,medium-risk and high-risk groups, respectively (p<0.001).
Discussion
In this retrospective cohort study, we investigated possiblepredictive factors associated with the occurrence of CKD in
a large series of pediatric patients with CAKUT. We alsodeveloped a risk prediction model for CKD stage III or higherin this population. Our model uses simple laboratory andimaging data that are routinely obtained for infants withantenatal hydronephrosis or other suspected CAKUT. Of par-ticular interest, in a previous study we estimated the overallprobability of CKD stage III or higher to be 6 % at 10 years ofage in our cohort of prenatally detected CAKUT [7]. Never-theless, in the present study, we found that the probability of amore severe renal impairment increased to about 42 % forthose patients assigned to the high-risk group.
The clinical spectrum of CAKUT includes a variety ofmalformations of the kidney and urinary tract [29, 30]. Theprevalence of CAKUT, based on prenatal US, has been re-ported to range between 3 and 6 per 1,000 births [30–32], withCAKUT being the most frequent congenital organmalformations in humans, comprising about 23 % of birthdefects [30, 31, 33–39]. Understanding the rate of renal func-tion impairment and predicting the long-term outcome arecritical for an accurate management of these patients [40, 41].
Our model relies on fetal US findings and serum creatinineduring the first month of life to predict the risk of renalfunction deterioration. In our analysis, four variables remained
Table 1 Baseline findings for822 infants with prenatally de-tected nephropathies according tothe occurrence of chronic kidneydisease (stages III–V)
CKD, Chronic kidney disease;US, ultrasonography
Data are presented as the number(of infants) with the percentage inparenthesis
Baseline findings No CKD (n=787) Progression to CKD (n=35) p
Gender
Male 530 (95.2) 27 (4.8) 0.18
Female 257 (97.0) 8 (3.0)
Laterality (fetal US)
Unilateral 478 (99.6) 2 (0.4) <0.001
Bilateral 309 (90.4) 33 (9.6)
Period of admission
1989–1999 252 (92.3) 21 (7.7) 0.001
2000–2009 535 (97.4) 14 (2.6)
Fetal US findings
Isolated hydronephrosis 692 (98.6) 10 (1.4) <0.001
Associated hydronephrosis 95 (79.2) 25 (20.8)
Oligohydramnios
Absent 779 (96.9) 25 (3.1) <0.001
Present 8 (44.4) 10 (55.6)
Table 2 Factors associated with CKD after adjustment by the Cox regression model (n=822)
Variables Coefficient Coefficienta Hazard ratio 95 % Confidence interval p value
Basal eGFR (ml/min) −0.066 −0.020 0.94 0.91–0.96 <0.001
Bilateral hydronephrosis 2.241 2.195 9.39 2.15–40.9 0.003
Oligohydramnios 0.872 0.826 2.39 1.04–5.49 0.04
Associated hydronephrosis 1.171 1.125 3.22 1.35–7.69 0.008
eGFR Estimated glomerular rate, CKD chronic kidney diseasea Shrunk coefficients using the shrinkage estimator obtained from the bootstrap results
Pediatr Nephrol
Author's personal copy
in the final model as independent predictors of CKD:oligohydramnios, bilateral US findings, eGFR and associatedhydronephrosis. Bilateral renal disease with oligohydramniosindicates significant global fetal renal dysfunction and is also arisk factor for the development of pulmonary hypoplasia[42–44]. Our findings are similar to those reported in the studyby Klaassen et al. [45] in showing a higher proportion of CKDamong children with a history of oligohydramnios. Theseauthors also reported that all patients with renal
oligohydramnios who survived the neonatal period developedCKD [45]. Nevertheless, somewhat unexpectedly, Sanna-Cherchi et al. [6] have recently shown that the risk for dialysiswas significantly higher for patients with a solitary kidney orwith renal hypodysplasia associated with posterior urethralvalves (hazard ratios 2.43 and 5.1, respectively) than forpatients with renal hypodysplasia or multicystic kidney andwas independent of other prognostic factors. Although all ofthese laboratory and imaging markers have been previouslyassociated with progression to CKD, our study integrated allof them into a single predictive model and used only simplevariables that are standardly available at baseline. Our scoreshowed excellent discrimination and was consistently highthrough 2 to 10 years of follow-up (c statistic always>0.900) and also showed good calibration, as reflected bythe calibration plots for the model of risk prediction andsurvival curves for the risk groups. Nevertheless, it shouldbe pointed out that the calibration plots showed a decline inthe prediction of renal survival for patients who had ≥10 yearsof follow-up. There was a persistent underestimation of renalfunction impairment for the medium- and high-risk groups,which clearly draws attention to the need for accurate bio-markers to predict renal function on a long-term basis and alsoto the need for prolonged follow-up for patients assigned tothe medium- and high-risk groups. Therefore, we believe thatthis model seems to be clinically useful and may improve thedecision-making process. Consequently, the model can bepotentially applied in several clinical settings. For example,the prognostic information provided by our model is useful toinform/advise patients and parents about the likely clinicaloutcome. Moreover, knowledge of the prognosis for sub-groups of patients can be used as a guide for the selection oftherapeutic options.
There are a number of limitations to our study, and severalmethodological considerations should be taken into accountwhen evaluating our findings. First, we did not validate thisrisk prediction instrument in an independent cohort [46, 47].Second, our study sample consisted of a referred CAKUTpopulation, of which all infants were born and followed upat a tertiary center. Therefore, our conclusions should not begeneralised to a non-referred CAKUT population. In addition,we were not able to systematically analyse clinical and demo-graphic variables such as sonographic renal measurements,birth weight, preterm birth and ethnicity that could contributeto the prediction of renal outcome for infants with CAKUT[48–50]. Furthermore, it should be pointed out that the per-formance of our clinical predictive model must be interpretedwith caution. The assessment of GFR in neonates is a verycomplex task, and it is difficult to ascertain accurately the truedegree of renal impairment. Therefore, we are aware that theinclusion of a group of individuals who are not possibly freeof the outcome of interest almost certainly resulted in anoverestimation of the performance of the clinical predictive
Fig. 1 Receiver operating characteristic curves estimated in order toevaluate the capacity of the risk score to predict chronic kidney diseasestage III or higher: a up to a 5-year follow-up ( n=333 infants; n = 27events7), b up to a 10-year follow-up (n=157 infants, n = 29 events).Dotted lines 95 % confidence intervals
Pediatr Nephrol
Author's personal copy
Fig. 2 Calibration plots for riskcategories comparing predictedand observed renal survivalaccording to the duration of thefollow-up period. a 5-yearfollow-up (n=333, ; n = 27events, b 10-year follow-up(n=157, n = 29 events)
Fig. 3 Kaplan–Meier estimatesfor overall renal survival by riskcategory (n=822). Shaded areas95 % confidence intervals.Number of patients at risk in eachcategory are shown below theX-axis
Pediatr Nephrol
Author's personal copy
model derived in our study. Nevertheless, some features of ourstudy may increase the strength of our findings, such as thelarge dataset collected over many years and the well-established protocol during both the prenatal and postnatalperiods. In addition, we are currently investigating the effectof including of other variables in our clinical prediction mod-el, i.e. imaging data such as RPD and clinical data such asbirth weight.
In conclusion, we have developed an accurate predictivemodel for the occurrence of CKD in children with prenatallydetected CAKUT. Our model uses fetal US findings andlaboratory data routinely available at baseline and can predictthe long-term risk for renal impairment with high accuracy. Itshould be pointed out, however, that external validation indiverse CAKUT prospective cohorts is clearly needed.
Acknowledgements This study was partially supported by CNPq (Bra-zilian National Research Council, Grants 401949/2010-9, 478742/2011-8and 481649/2013-1), FAPEMIG (Fundação de Amparo à Pesquisa doEstado deMinas Gerais, Grants PPM-00345-11 and PPM-00273-13) andthe INCT-MMGrant (FAPEMIG: CBB-APQ-00075-09 / CNPq 573646/2008-2). GPP was the recipient of a CNPq fellowship. Dr. EA Oliveira,Dr. AC Simões e Silva and Enrico A Colosimo received a research grantfrom CNPq.
References
1. McKenna PH (2008) Prenatally detected hydronephrosis: impact onpostnatal outcomes. J Urol 179:18–19
2. Braga LH, Mijovic H, Farrokhyar F, Pemberton J, Demaria J,Lorenzo AJ (2013) Antibiotic prophylaxis for urinary tract infectionsin antenatal hydronephrosis. Pediatrics 131:e251–e261
3. Ismaili K, Hall M, Piepsz A, Alexander M, Schulman C, Avni FE(2005) Insights into the pathogenesis and natural history of fetuseswith renal pelvis dilatation. Eur Urol 48:207–214
4. Yamacake KG, Nguyen HT (2013) Current management of antenatalhydronephrosis. Pediatr Nephrol 28:237–243
5. Nguyen HT, Herndon CD, Cooper C, Gatti J, Kirsch A, KokorowskiP, Lee R, Perez-Brayfield M, Metcalfe P, Yerkes E, Cendron M,Campbell JB (2010) The Society for Fetal Urology consensus state-ment on the evaluation andmanagement of antenatal hydronephrosis.J Pediatr Urol 6:212–231
6. Sanna-Cherchi S, Ravani P, Corbani V, Parodi S, Haupt R, Piaggio G,Innocenti ML, Somenzi D, Trivelli A, Caridi G, Izzi C, Scolari F,Mattioli G, Allegri L, Ghiggeri GM (2009) Renal outcome in patientswith congenital anomalies of the kidney and urinary tract. Kidney Int76:528–533
7. Quirino IG, Diniz JS, Bouzada MC, Pereira AK, Lopes TJ, PaixaoGM, Barros NN, Figueiredo LC, Cabral AC, Simoes ESAC, OliveiraEA (2012) Clinical course of 822 children with prenatally detectednephrouropathies. Clin J Am Soc Nephrol 7:444–451
8. Melo BF, Aguiar MB, Bouzada MC, Aguiar RL, Pereira AK, PaixaoGM, Linhares MC, Valerio FC, Simoes ESAC, Oliveira EA (2012)Early risk factors for neonatal mortality in CAKUT: analysis of 524affected newborns. Pediatr Nephrol 27:965–972
9. Oliveira EA, Diniz JS, Cabral AC, Leite HV, Colosimo EA, OliveiraRB, Vilasboas AS (1999) Prognostic factors in fetal hydronephrosis:a multivariate analysis. Pediatr Nephrol 13:859–864
10. Bouzada MC, Oliveira EA, Pereira AK, Leite HV, Rodrigues AM,Fagundes LA, Goncalves RP, Parreiras RL (2004) Diagnostic accuracyof fetal renal pelvis anteroposterior diameter as a predictor of uropathy:a prospective study. Ultrasound Obstet Gynecol 24:745–749
11. Bouzada MC, Oliveira EA, Pereira AK, Leite HV, Rodrigues AM,Fagundes LA, Goncalves RP, Parreiras R (2004) Diagnostic accuracyof postnatal renal pelvic diameter as a predictor of uropathy: aprospective study. Pediatr Radiol 34:798–804
12. Coelho GM, Bouzada MC, Lemos GS, Pereira AK, Lima BP,Oliveira EA (2008) Risk factors for urinary tract infection in childrenwith prenatal renal pelvic dilatation. J Urol 179:284–289
13. Coelho GM, Bouzada MC, Pereira AK, Figueiredo BF, Leite MR,Oliveira DS, Oliveira EA (2007) Outcome of isolated antenatalhydronephrosis: a prospective cohort study. Pediatr Nephrol 22:1727–1734
14. Dias CS, Silva JM, Pereira AK, Marino VS, Silva LA, Coelho AM,Costa FP, Quirino IG, Simoes e Silva AC, Oliveira EA (2013)Diagnostic accuracy of renal pelvic dilatation for detecting surgicallymanaged ureteropelvic junction obstruction. J Urol 190:661–666
15. Dias CS, Bouzada MC, Pereira AK, Barros PS, Chaves AC, AmaroAP, Oliveira EA (2009) Predictive factors for vesicoureteral refluxand prenatally diagnosed renal pelvic dilatation. J Urol 182:2440–2445
16. Levey AS, Coresh J, Balk E, Kausz AT, Levin A, Steffes MW, HoggRJ, Perrone RD, Lau J, Eknoyan G (2003) National KidneyFoundation practice guidelines for chronic kidney disease: evalua-tion, classification, and stratification. Ann Intern Med 139:137–147
17. Schwartz GJ, Brion LP, Spitzer A (1987) The use of plasma creati-nine concentration for estimating glomerular filtration rate in infants,children, and adolescents. Pediatr Clin North Am 34:571–590
18. Phelan JP, Smith CV, Broussard P, Small M (1987) Amniotic fluidvolume assessment with the four-quadrant technique at 36–42 weeks’gestation. J Reprod Med 32:540–542
19. Steyerberg E (2008) Clinical prediction models: A practical approachto development, validation, and updating. Springer SBM, New York
20. Sullivan LM, Massaro JM, D'Agostino RB Sr (2004) Presentation ofmultivariate data for clinical use: The Framingham Study risk scorefunctions. Stat Med 23:1631–1660
21. Harrell FE Jr, Lee KL, Mark DB (1996) Multivariable prognosticmodels: issues in developing models, evaluating assumptions andadequacy, and measuring and reducing errors. Stat Med 15:361–387
22. Pencina MJ, D'Agostino RB (2004) Overall C as a measure ofdiscrimination in survival analysis: model specific population valueand confidence interval estimation. Stat Med 23:2109–2123
23. Cook NR, Buring JE, Ridker PM (2006) The effect of including C-reactive protein in cardiovascular risk prediction models for women.Ann Intern Med 145:21–29
24. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M,Obuchowski N, Pencina MJ, Kattan MW (2010) Assessing theperformance of prediction models: a framework for traditional andnovel measures. Epidemiology 21:128–138
25. Mallett S, Royston P, Waters R, Dutton S, Altman DG (2010)Reporting performance of prognostic models in cancer: a review.BMC Med 8:21
26. Moons KG, Kengne AP, Grobbee DE, Royston P, Vergouwe Y,Altman DG, Woodward M (2012) Risk prediction models: II.External validation, model updating, and impact assessment. Heart98:691–698
27. Moons KG, Kengne AP, Woodward M, Royston P, Vergouwe Y,Altman DG, Grobbee DE (2012) Risk prediction models: I.Development, internal validation, and assessing the incremental val-ue of a new (bio)marker. Heart 98:683–690
28. Steyerberg EW, Harrell FE Jr, Borsboom GJ, Eijkemans MJ,Vergouwe Y, Habbema JD (2001) Internal validation of predictivemodels: efficiency of some procedures for logistic regression analy-sis. J Clin Epidemiol 54:774–781
Pediatr Nephrol
Author's personal copy
29. Adalat S, Bockenhauer D, Ledermann SE, Hennekam RC,Woolf AS(2010) Renal malformations associated with mutations of develop-mental genes: messages from the clinic. Pediatr Nephrol 25:2247–2255
30. Estrada CR Jr (2008) Prenatal hydronephrosis: early evaluation. CurrOpin Urol 18:401–403
31. Bulum B, Ozcakar ZB, Ustuner E, Dusunceli E, Kavaz A, Duman D,Walz K, Fitoz S, Tekin M, Yalcinkaya F (2013) High frequency ofkidney and urinary tract anomalies in asymptomatic first-degreerelatives of patients with CAKUT. Pediatr Nephrol 28:2143–2147
32. Wiesel A, Queisser-Luft A, Clementi M, Bianca S, Stoll C (2005)Prenatal detection of congenital renal malformations by fetal ultraso-nographic examination: an analysis of 709,030 births in 12 Europeancountries. Eur J Med Genet 48:131–144
33. Harambat J, van Stralen KJ, Kim JJ, Tizard EJ (2012) Epidemiologyof chronic kidney disease in children. Pediatr Nephrol 27:363–373
34. LoaneM, Dolk H, Kelly A, Teljeur C, Greenlees R, Densem J (2011)Paper 4: EUROCAT statistical monitoring: identification and inves-tigation of ten year trends of congenital anomalies in Europe. BirthDefects Res A Clin Mol Teratol 91[Suppl 1]:S31–S43
35. Sanna-Cherchi S, Kiryluk K, Burgess KE, Bodria M, Sampson MG,Hadley D, Nees SN, Verbitsky M, Perry BJ, Sterken R, LozanovskiVJ, Materna-Kiryluk A, Barlassina C, Kini A, Corbani V, Carrea A,Somenzi D, Murtas C, Ristoska-Bojkovska N, Izzi C, Bianco B,Zaniew M, Flogelova H, Weng PL, Kacak N, Giberti S, Gigante M,Arapovic A, Drnasin K, Caridi G, Curioni S, Allegri F, Ammenti A,Ferretti S, Goj V, Bernardo L, Jobanputra V, Chung WK, Lifton RP,Sanders S, State M, Clark LN, Saraga M, Padmanabhan S,Dominiczak AF, Foroud T, Gesualdo L, Gucev Z, Allegri L, Latos-Bielenska A, Cusi D, Scolari F, Tasic V, Hakonarson H, Ghiggeri GM,Gharavi AG (2012) Copy-number disorders are a common cause ofcongenital kidney malformations. Am J Hum Genet 91:987–997
36. Sanna-Cherchi S, Sampogna RV, Papeta N, Burgess KE, Nees SN,Perry BJ, Choi M, Bodria M, Liu Y, Weng PL, Lozanovski VJ,Verbitsky M, Lugani F, Sterken R, Paragas N, Caridi G, Carrea A,Dagnino M, Materna-Kiryluk A, Santamaria G, Murtas C, Ristoska-Bojkovska N, Izzi C, Kacak N, Bianco B, Giberti S, Gigante M,Piaggio G, Gesualdo L, Kosuljandic Vukic D, Vukojevic K, Saraga-Babic M, Saraga M, Gucev Z, Allegri L, Latos-Bielenska A, Casu D,State M, Scolari F, Ravazzolo R, Kiryluk K, Al-Awqati Q, D'AgatiVD, Drummond IA, Tasic V, Lifton RP, Ghiggeri GM, Gharavi AG(2013) Mutations in DSTYK and dominant urinary tractmalformations. N Engl J Med 369:621–629
37. Soares CM, Diniz JS, Lima EM, Oliveira GR, Canhestro MR,Colosimo EA, Simoes e Silva AC, Oliveira EA (2009) Predictive
factors of progression to chronic kidney disease stage 5 in apredialysis interdisciplinary programme. Nephrol Dial Transplant24:848–855
38. Soares CM, Diniz JS, Lima EM, Silva JM, Oliveira GR, CanhestroMR, Colosimo EA, Simoes e Silva AC, Oliveira EA (2008) Clinicaloutcome of children with chronic kidney disease in a pre-dialysisinterdisciplinary program. Pediatr Nephrol 23:2039–2046
39. Song R, Yosypiv IV (2011) Genetics of congenital anomalies of thekidney and urinary tract. Pediatr Nephrol 26:353–364
40. Lee RS, Cendron M, Kinnamon DD, Nguyen HT (2006) Antenatalhydronephrosis as a predictor of postnatal outcome: a meta-analysis.Pediatrics 118:586–593
41. Woodhouse CR, Neild GH, Yu RN, Bauer S (2012) Adult care ofchildren from pediatric urology. J Urol 187:1164–1171
42. Hogan J, DourtheME, Blondiaux E, Jouannic JM, Garel C, Ulinski T(2012) Renal outcome in children with antenatal diagnosis of severeCAKUT. Pediatr Nephrol 27:497–502
43. Kemper MJ, Mueller-Wiefel DE (2007) Prognosis of antenatallydiagnosed oligohydramnios of renal origin. Eur J Pediatr 166:393–398
44. Mehler K, Beck BB, Kaul I, Rahimi G, Hoppe B, Kribs A (2011)Respiratory and general outcome in neonates with renaloligohydramnios–a single-centre experience. Nephrol DialTransplant 26:3514–3522
45. Klaassen I, Neuhaus TJ, Mueller-Wiefel DE, Kemper MJ (2007)Antenatal oligohydramnios of renal origin: long-term outcome.Nephrol Dial Transplant 22:432–439
46. Bleeker SE, Moll HA, Steyerberg EW, Donders AR, Derksen-Lubsen G, Grobbee DE, Moons KG (2003) External validation isnecessary in prediction research: a clinical example. J Clin Epidemiol56:826–832
47. Steyerberg EW, Bleeker SE, Moll HA, Grobbee DE, Moons KG(2003) Internal and external validation of predictive models: a sim-ulation study of bias and precision in small samples. J Clin Epidemiol56:441–447
48. Kandasamy Y, Smith R, Wright IM, Lumbers ER (2013) Extra-uterine renal growth in preterm infants: oligonephropathy and pre-maturity. Pediatr Nephrol 28:1791–1796
49. Silva JM, Diniz JS, Silva AC, Azevedo MV, Pimenta MR, OliveiraEA (2006) Predictive factors of chronic kidney disease in severevesicoureteral reflux. Pediatr Nephrol 21:1285–1292
50. Silverwood RJ, Pierce M, Hardy R, Sattar N, Whincup P, Ferro C,Savage C, Kuh D, Nitsch D (2013) Low birth weight, later renalfunction, and the roles of adulthood blood pressure, diabetes, andobesity in a British birth cohort. Kidney Int 84:1262–1270
Pediatr Nephrol
Author's personal copy