Predicting the clinical outcome of congenital unilateral ureteropelvic junction obstruction in newborn by urinary proteome analysis Stephane Decramer 1–3 , Stefan Wittke 4 , Harald Mischak 4 , Petra Zu ¨rbig 4 , Michael Walden 4 , Franc ¸ois Bouissou 1–3 , Jean-Loup Bascands 1,2 & Joost P Schanstra 1,2 We analyzed urinary polypeptides from individuals with neonatal ureteropelvic junction (UPJ) obstruction to predict which individuals with this condition will evolve toward obstruction that needs surgical correction. We identified polypeptides that enabled diagnosis of the severity of obstruction and validated these biomarkers in urine collected in a prospective blinded study. Using these noninvasive biomarkers, we were able to predict, several months in advance and with 94% precision, the clinical evolution of neonates with UPJ obstruction. Proteome analysis is increasingly investigated for its use in identifying biomarkers that can assist in diagnosis and disease staging 1–3 . The use of this approach to predict clinical evolution of disease, however, has not been fully developed. Noninvasive proteomics-based prognosis of clinical progression would be of interest, especially in newborns and children. In kidney-related diseases, urine is a potential source for such biomarkers 4 . Although detected by echography before birth, congenital unilateral neonatal UPJ obstruction is a common clinical problem after birth 5 . One issue in UPJ obstruction is determining whether infants should undergo surgery to correct this condition 6 . Currently, this requires medical surveillance, including repetitive invasive diuretic renograms relying on arbitrary threshold values. Urinary markers for UPJ obstruction have been identified, but their prognostic value in terms of informing the decision to operate has yet to be shown 6 . There is thus a lack of nonarbitrary, noninvasive biomarkers to determine, rapidly after birth, the necessity for relief surgery in UPJ obstruction. Furthermore, during the often prolonged medical surveillance, tem- poral obstruction might initiate processes that reduce nephron num- bers often observed in renal biopsies of infants with UPJ obstruction 7 , which in turn might lead to hypertension and proteinuria later in life. We hypothesized that these processes should be reflected even at an early stage of the condition by typical changes in urinary polypeptides. To evaluate this hypothesis, we used an on-line combination of capillary electrophoresis and mass spectrometry (CE-MS 8 ) to obtain data on urinary polypeptides of 103 neonates with UPJ obstruction. First, we compared the data obtained from healthy newborns with that of normal adults (Fig. 1a and Supplementary Methods online). The distribution of the newborn urinary polypeptides was markedly less scattered than the adult pattern, resulting in much higher consistency of the data. The origin of this difference is unclear, but it might reflect the frequency of physical exercise, the homogenous feeding pattern of newborns and the absence of acquired kidney lesions. This consistency of urinary polypeptides in urine from newborns allows a clearer definition of pathological urinary polypeptides. We next classified newborns at birth according to their clinical criteria into three groups (Supplementary Table 1 online): nonoper- ated individuals with UPJ obstruction (No_OP), individuals who might possibly undergo operation (OP_Poss) and individuals with severe UPJ obstruction who were scheduled to undergo surgery rapidly after birth (OP). For the OP group, urine samples were taken before surgery. OP_Poss individuals were the neonates who needed repetitive and invasive medical surveillance to determine the necessity of relief surgery 6 . Typical CE-MS reference data of healthy newborns, No_OP and OP groups are shown in Figure 1a. Compar- ison of these three different data sets enabled the identification of several putative biomarkers. For example, 19 biomarkers discrimi- nated between the healthy newborn + No_OP group and the OP group (Fig. 1b and Supplementary Tables 2–4 online). For four of these markers, the frequency of appearance in each group is indicated. Exclusive polypeptide expression in one of the three groups was not observed and thus a single polypeptide does not allow classifying an individual sample with high confidence. The analysis of 51 distinct polypeptides (Supplementary Table 2 online), however, allowed us to discriminate between the three groups: randomly chosen individuals of these three groups could be classified with 98% sensitivity and 98% specificity. Blinded validation on an independent group of OP and No_OP individuals (n ¼ 16) yielded a classification with 94% sensitivity, and 80% specificity for No_OP and 100% specificity for OP. As described above, the main problem in newborns with moderate UPJ obstruction (OP_Poss) is deciding between relief surgery or continued invasive and repetitive medical surveillance based on surrogate endpoints, risking unnecessary prolonged ureteral obstruc- tion. Using a hierarchical disease model 9 based on the polypeptides used to discriminate between healthy newborns, No_OP and OP groups (Supplementary Table 2 online), we predicted the clinical outcome of individuals in the OP_Poss group. Using this approach in a prospective blinded study of 36 OP_Poss individuals, 25 were predicted to evolve toward the OP group, and 11 were predicted to evolve toward the No_OP group (Fig. 2a). Nine months later, which is in general the time necessary to determine whether individuals need surgery or they have spontaneously resolved the © 2006 Nature Publishing Group http://www.nature.com/naturemedicine Received 10 January; accepted 23 February; published online 19 March 2006; doi:10.1038/nm1384 1 Inserm, U388, 1 Avenue Jean Poulhes, F-31432 Toulouse, France. 2 Universite ´ Toulouse III Paul Sabatier, IFR31, Institut Louis Bugnard, F-31432 Toulouse, France. 3 CHU de Toulouse, Nephropediatric Unit, Hopital des Enfants, TSA 70034, 31059 Toulouse Cedex 9, France. 4 Mosaiques Diagnostics and Therapeutics AG, Mellendorfer Strasse 7-9, 30625 Hannover, Germany. Correspondence should be addressed to J.P.S. (schans@toulouse.inserm.fr). NATURE MEDICINE ADVANCE ONLINE PUBLICATION 1 BRIEF COMMUNICATIONS