Pediatr Radiol (2004) 34: 798–804 DOI 10.1007/s00247-004-1283-8
Maria Caˆndida Ferrarez Bouzada Eduardo Araujo Oliveira Alamanda Kfoury Pereira Henrique Vitor Leite Andre´ia Moura Rodrigues Lı´ via Alvarenga Fagundes Ricardo Peixoto Gonc¸alves Ricardo Parreiras
Received: 13 April 2004 Revised: 17 May 2004 Accepted: 22 June 2004 Published online: 5 August 2004 Ó Springer-Verlag 2004 M. C. F. Bouzada Neonatal Unit, Department of Pediatrics, Hospital das Clı´ nicas, Federal University of Minas Gerais, Avenida Professor Alfredo Balena 190, Belo Horizonte, 30130-100, Brazil E. A. Oliveira Pediatric Nephrourology Unit, Department of Pediatrics, Hospital das Clı´ nicas, Federal University of Minas Gerais, Avenida Professor Alfredo Balena 190, Belo Horizonte, 30130-100, Brazil E. A. Oliveira Ultrasound Unit, CNPq Researcher, Hospital das Clı´ nicas, Federal University of Minas Gerais, Avenida Professor Alfredo Balena 190, Belo Horizonte, 30130-100, Brazil A. K. Pereira Æ H. V. Leite Fetal Medicine Unit, Department of Gynecology and Obstetrics, Hospital das Clı´ nicas, Federal University of Minas Gerais, Avenida Professor Alfredo Balena 190, Belo Horizonte, 30130-100, Brazil A. M. Rodrigues Æ L. A. Fagundes R. P. Gonc¸alves Æ R. Parreiras FAPEMIG and CNPq studentship, Hospital das Clı´ nicas, Federal University of Minas Gerais, Avenida Professor Alfredo Balena 190, Belo Horizonte, 30130-100, Brazil E. A. Oliveira (&) Rua Patagonia 515 / 701, 30320-080 Belo Horizonte, Minas Gerais, Brazil E-mail:
[email protected] Fax: +55-31-32223584
ORIGINAL ARTICLE
Diagnostic accuracy of postnatal renal pelvic diameter as a predictor of uropathy: a prospective study
Abstract Background: There is a necessity to standardize postnatal ultrasonographic markers predictive of significant uropathies. The purpose of this study was to assess the accuracy of anteroposterior renal pelvic diameter (APD) on postnatal ultrasound to discriminate between significant uropathy and idiopathic renal pelvic dilatation. Materials and methods: Neonates (n=110) who were found to have isolated fetal renal pelvic dilatation underwent systematic investigation and were prospectively followed up. A US scan was performed after the 1st week of life, and all infants underwent voiding cystourethrography. Neonates with an APD larger than 10 mm were examined with renal scintigraphy. Receiver-operating characteristic (ROC) plots were constructed to determine the best cutoffs for APD to identify renal units with significant uropathy as well as those requiring surgical intervention. Results: The area under the curve (AUC) estimated by the ROC curve was 0.921 (95% CI=0.869–0.957), indicating excellent discriminant power between nonsignificant pelvis dilatation and significant uropathy. A cutoff of 10 mm for APD showed the best accuracy for identifying infants with a significant uropathy (sensitivity, 90.4%; specificity, 91%). To identify infants who required surgical inter-
vention, the calculated AUC was 0.976 (95% CI=0.940–0.993). A cutoff of 15 mm showed the best diagnostic performance for identifying the renal units requiring pyeloplasty (sensitivity, 100%; specificity, 92.5%). Conclusion: Our results suggest that postnatal APD is an excellent test for identifying fetuses with significant uropathy, as well as those requiring postnatal intervention.
Keywords Pyelectasis Æ Accuracy Æ ROC curve Æ Fetus Æ Children
799
Introduction Fetal US screening has become widely established and allows a new understanding of certain congenital urinary tract diseases from their very beginning to after birth [1]. The postnatal investigation of severe hydroureteronephrosis is well established and usually includes US scan, voiding cystourethrogram (VCUG) and diuretic renography [2, 3]. However, it has become apparent that many cases of antenatally detected hydronephrosis will spontaneously improve with no surgical intervention [4]. There is a consensus that postnatal US must be the initial study for the evaluation of fetal hydronephrosis [1, 5, 6]. The major limitation of the identification of renal dilatation is that hydronephrosis does not necessarily imply obstruction, and US gives no information about the function of the affected kidney [6]. However, a neonatal US scan is potentially able to confirm the prenatal findings and to suggest a differential diagnosis [7]. In addition, it is possible to evaluate the degree of urinary tract dilatation according to the US appearance of the renal parenchyma and pelvicalyceal system [8] or according to various measurements of the kidney [6]. The anteroposterior renal pelvic diameter (APD) has been a useful US parameter indicating those fetuses with a possible significant renal abnormality [9, 10]. However, in the postnatal setting there have been few studies designed to evaluate the accuracy of APD in neonates with fetal hydronephrosis. The purpose of this study was to assess the accuracy of APD on postnatal US in discriminating between significant uropathy and benign transitory renal pelvic dilatation. The data are based on a series of infants submitted to a systematic protocol of investigation and prospectively followed up.
Materials and methods In total, 110 infants consecutively born at Hospital das Clı´ nicas-UFMG (Belo Horizonte, Brazil) from May 1999 to July 2003 who were found to have isolated renal pelvic dilatation—defined as the presence of APD equal to or greater than 5 mm by prenatal US after 28 weeks’ gestation—underwent systematic investigation for uropathies and were prospectively followed up. Fetuses with hydronephrosis associated with ureteral dilatation or bladder distension or with oligohydramnios were excluded from the analysis. Fetuses with aneuploidy or multiple malformations were also excluded. Postnatally, neonates were investigated according to a systematic protocol described elsewhere [11]. All infants underwent a US scan after the 1st week of life. All renal pelves with a fetal APD equal to or greater
than 5 mm were included in the analysis. In order to evaluate the diagnostic accuracy of renal pelvic dilatation, patients with ureteral or bladder abnormalities on postnatal sonography were excluded. Therefore, patients with ureterocele (n=4), primary megaureter (n=3), posterior urethral valves (n=1), vesicoureteral reflux (VUR) associated with duplex system (n=2), and hypoplastic kidney (n=5) were excluded from the analysis because US findings other than renal pelvic dilatation often indicated the presence of significant uropathy. The first postnatal US scan was performed within a median of 14 days of life (7–40 days), with 90% of them being performed during the 1st month of life. The postnatal US scans were performed by the same trained examiner using a Siemens machine (Sonoline Prima SLC, Erlangen, Germany, 5-MHz probe), with the patient in the prone and supine positions. Sonograms were usually performed after breastfeeding, on a well-hydrated infant. First, the full bladder and retrovesical ureter were examined; then, after micturition, APD was measured at the point where it exits the kidney. A VCUG was acquired for all infants within 3 months of life (87% between 18 days and 2 months of life). If both the postnatal scan and VCUG were normal, antibiotic prophylaxis was discontinued and the infant was followed up to 12 months of age. If the postnatal scan showed mild pelvis dilatation (APD40%) as ascertained by DMSA
800
scan, whatever the pattern on DTPA ( intermediate or obstructed). Renal units with an uptake of less than 40% were managed surgically by Anderson-Hynes dismembered pyeloplasty. Combined data from VCUG and renal scintigraphy were regarded as the reference standard. Significant uropathy was defined as the presence of well-established urinary tract abnormality or the presence of an abnormal DMSA scan and a diuretic renogram. Otherwise, renal pelvic dilatation was labeled as idiopathic dilatation and regarded as a nonsignificant uropathy in the analysis. Extrarenal pelves were considered as a normal variant and were included in the group of nonsignificant uropathy. Those renal units with fetal renal pelvic dilatation not confirmed on postnatal US and a normal VCUG were labeled as transitory dilatation and were also included in the group of nonsignificant uropathy. Therefore, after a minimum of 6 months of follow-up, renal units were classified into three groups for accuracy analysis: (1) nonsignificant uropathy, (2) significant uropathy amenable to clinical management, and (3) significant uropathy requiring surgical management. The guidelines from the Standard for Reporting of Diagnostic Accuracy (STARD) were used in the present study [13]. The receiver-operating characteristic (ROC) plots were analyzed to determine the optimal APD needed to define renal units with significant uropathy and also to define those units at risk of significant postnatal obstruction, which would require pyeloplasty or other urologic surgical intervention. The prevalence of disease (pre-test probability) was regarded as 28%, according to the findings of Ismaili et al. [14]. The index test (APD) and the reference standard tests were performed by examiners who were blind to the results of the other tests. The goals were: (1) to choose the threshold with the highest sensitivity for predicting uropathy and to minimize the chance of missing a clinically significant urinary tract abnormality, and (2) to choose the threshold with the best accuracy for predicting the need for postnatal surgery. Informed consent was obtained from the parents of each patient, and the study was approved by the Ethics Committee of UFMG.
Table 1 Postnatal diagnosis of urinary tract anomalies and surgical interventions
Results A total of 110 infants (24 females) were included in the analysis. Fifty-six infants presented fetal bilateral renal pelvic dilatation, corresponding to 166 renal units for study. Of the 166 dilated renal units, 45 (27%) presented with significant uropathy and 121 (73%) with nonsignificant uropathy (Table 1). There was a predominance of ureteropelvic junction obstruction (n=38) among renal units with significant uropathy. Nineteen renal units (11.5%) required surgical intervention for relief of the obstruction. The median follow-up time was 30 months (6–58). Sixty-six (60%) patients were followed up more than 2 years, and 97 (88%) more than 1 year after birth. Various postnatal renal pelvic APD thresholds were analyzed and correlated with the presence of significant uropathy. The area under the curve estimated by the ROC curve was 0.921 (95% CI=0.869–0.957) (Fig. 1). A cutoff of 5 mm for postnatal APD detected most neonates with a significant uropathy. The sensitivity was 95.3% (84.2–99.3), the specificity was 28.0% (20.3–37.0), and the positive and negative likelihood ratios were 1.33 and 0.17, respectively. The relative risk for significant uropathy was 5 (95% CI=2–20, P
