Abstract: FR-PO459
Urine Metabolites Predict Kidney Function Decline in Type I Diabetic Subjects
Session Information
- Diabetic Kidney Disease: Clinical - I
October 26, 2018 | Location: Exhibit Hall, San Diego Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Diabetic Kidney Disease
- 602 Diabetic Kidney Disease: Clinical
Authors
- Darshi, Manjula, University of Texas Health San Antonio, San Antonio, Texas, United States
- Valo, Erkka A., Folkhälsan Institute of Genetics, Folkhälsan Research Center,Helsinki , Helsinki, Finland
- Zhang, Jing, University of California San Diego, La Jolla, California, United States
- Kim, Jiwan John, University of Texas Health San Antonio, San Antonio, Texas, United States
- Saito, Rintaro, Institute for Advanced Biosciences, Keio University, Tsuruoka, Yamagata, Japan
- Montemayor, Daniel, University of Texas Health San Antonio, San Antonio, Texas, United States
- Ahluwalia, Tarunveer S., Steno Diabetes Center Copenhagen, Gentofte, Denmark
- Snell-Bergeon, Janet, University of Colorado Denver , Aurora, Colorado, United States
- Rossing, Peter, Steno Diabetes Center Copenhagen, Gentofte, Denmark
- Groop, Per-Henrik, Folkhälsan Institute of Genetics, Folkhälsan Research Center,Helsinki , Helsinki, Finland
- Natarajan, Loki, University of California San Diego, La Jolla, California, United States
- Sharma, Kumar, University of Texas Health San Antonio, San Antonio, Texas, United States
Background
Diabetes is the most common cause of chronic kidney disease (CKD) and end-stage renal failure. Albuminuria and eGFR are widely approved biomarkers to identify kidney disease progression. However, due to considerable heterogeneity not all subjects progress at the same rate. In the current study, we evaluated a set of urinary metabolites toward prediction of rapid progression of CKD in patients with type I diabetes.
Methods
We used a nested case-control study in four diverse cohorts (CACTI, EDC, FinnDiane, and Steno) with long-standing type I diabetes and normal kidney function. Subjects were classified into slow decliners/controls with eGFR decline ≤1ml/min/1.73m2/yr or rapid progressors/cases with eGFR decline of ≥3ml/min/1.73m2/yr. Thirty four urine metabolites were measured by gas chromatography-mass spectrometry. Logistic regression and Random Forest models were used to predict rapid eGFR decline. Area under the curve (AUC) were used to assess model performance.
Results
Baseline mean eGFR and median ACR in controls (n=340) and cases (n=212) were 91.97 (sd 18.68) and 9.44 (IQR 30), and 98.37 (sd 25.44) and 33.49 (IQR 283.01), respectively. Analysis with clinical variables revealed age, baseline ACR, and baseline eGFR to be significant predictors of rapid decline. Among the metabolites, three were univariately associated with rapid decline (FDR p=0.015, 0.025, 0.025). A model to predict eGFR decline using clinical variables had an AUC of 0.71 (95% CI 0.63-0.79); no significant improvements in AUC were observed with added metabolites. In a stratified analysis of eGFR≥60 ml/min/1.73m2 and combined micro- and macroalbuminuria (MA+) group, metabolites significantly improved the AUC from 0.69 with clinical variables (0.45-0.84) to 0.76 (0.61-0.89) when combined with metabolites. Random Forest selected 6 metabolites as top 7 variables and the metabolites significantly improved the AUC from 0.61 (0.44-0.76) to 0.75 (0.61-0.89), 95% CI.
Conclusion
In subjects with albuminuria and normal eGFR (≥60 3ml/min/1.73m2) clinical variables are not optimal to predict kidney function decline and urine metabolites may be useful as prognostic biomarkers for loss of renal function.
Funding
- Other NIH Support