Validation of a predictive model of idiopathic membranous nephropathy: Its clinical and research implications

Abstract

Although a number of factors have consistently correlated with progression to chronic renal insufficiency (CRI) in idiopathic membranous glomerulonephropathy (IMGN), they appear late, are not quantitative in nature and have not been validated. We have determined that the highest sustained six-month period of proteinuria is an important predictor of progression. Using multiple logistic modelling, the only additional prognostic variables of importance in 184 Canadian patients were the initial creatinine clearance and the rate of change in function over this six-month internal. Independent data from Italy (101 patients) and Finland (78 patients) were obtained for comparison. Sensitivity, specificity, negative and positive predictive values and overall accuracy, as well as Pearson's goodness-of-fit and Harrell's 'C' statistic were used to assess the fits of the model. Accuracy of prediction was ≤85% in all three countries. Pearson's Chi-square goodness-of-fit showed good agreement across the spectrum and Harrell's 'C' statistic was ≤90%. Therefore, a predictive, semiquantitative algorithm in IMGN has been validated. Its relevance in patient management and in clinical trials is illustrated.