Background: Diagnostic criteria for Balkan endemic nephropathy (BEN) have not been precisely established. In the present study the predictive value of variables previously proposed as diagnostic criteria for BEN was examined.
Methods: The study involved 182 patients: 98 patients with BEN, 57 patients with other kidney diseases (20 with glomerulonephritis, 17 with tubulointerstitial diseases and 20 with hypertensive nephrosclerosis) and 27 healthy subjects. The BEN group comprised patients who fulfilled criteria for BEN and suspected BEN, together with patients with proteinuria and at least two tubular abnormalities or one tubular abnormality and a history of urothelial tumour. Demographic, clinical, laboratory and ultrasound variables of examined groups were combined in univariate/multivariate logistic regression analysis.
Results: Out of 28 analysed variables only urine alpha1-microglobulin (MG) and kidney length were selected as significant predictors in differentiating BEN from other kidney diseases and healthy controls. Using ROC curves the cutoff values of these variables and proteinuria and kidney volume, variables collinear with them, were found. Moderate sensitivity and specificity characterized all these cutoff values except for proteinuria, which provided high sensitivity and specificity in combination of BEN and healthy persons. The predictive value of different combinations of selected variables was not significantly different from the predictive value of each variable individually.
Conclusions: Proteinuria, urine alpha1-MG, kidney length and volume were selected as significant predictors of BEN. Variables related to kidney failure as well as several tubular disorders (urine specific gravity, FENa and TRP) had an insignificant predictive value and could not be used for differential diagnosis of BEN.