Objectives: Numerous biomarkers have been shown to be associated with albuminuria. However, few of them are valuable separate predictors of albuminuria development. This study aimed to develop a model for predicting the short-term risk of new-onset albuminuria in normoalbuminuric patients with type 2 diabetes (T2D).
Methods: 213 patients with T2D who were normoalbuminuric at the baseline were enrolled in this study. Basal levels of clinical characteristics and renal biomarkers including urinary orosomucoid (alpha-1-acid-glycoprotein, UORM), neutrophil gelatinase-associated lipocalin, retinol-binding protein, alpha-1-microglobulin, transferrin, and albumin-to-creatinine ratio (ACR) were utilized to analyze the association with the short-term risk of new-onset albuminuria.
Results: 19.72% of normoalbuminuric subjects at baseline progressed to albuminuria over the 2-year follow-up period. Except for NGAL, the basal levels of the other five renal biomarkers were significantly associated with new-onset albuminuria risk in the univariate analysis. In the multivariate logistic regression analysis using Forward: LR method, a model incorporating UORM/Cr, ACR, and HbA1c was established. Comparatively, this model had a higher potential to predict new-onset albuminuria risk compared with the single use of renal markers. In the validation of this model performed by 5-fold cross-validation method, the accuracy of this model was 0.818 ± 0.008 in the training sets, 0.827 ± 0.062 in the test sets, indicating a good capability for assessing albuminuria risk. Finally, a nomogram based on this model was constructed to facilitate its use in clinical practice.
Conclusion: The combined analysis of UORM/Cr, ACR and HbA1c may be of potential value for predicting the short-term risk of new-onset albuminuria in such patients.
Keywords: K-fold cross-validation; New-onset albuminuria; Nomogram; Prediction; Type 2 diabetes.
Copyright © 2021 Elsevier Inc. All rights reserved.