Abstract: TH-PO929
Biomarker Identification for Diabetic Kidney Disease at the Early Stage
Session Information
- Diabetic Kidney Disease: Biomarkers, Pathogenesis
November 07, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Diabetic Kidney Disease
- 602 Diabetic Kidney Disease: Clinical
Authors
- Fu, Haiyan, Nanfang Hospital, Southern Medical University, Guangzhou, Guangdong, China
- Liu, Shijia, the affiliated hospital of Nanjing University of Chinese Medicine, Nanjing, Jiangsu, China
- Zhu, Haili, Nanfang Hospital, Southern Medical University, Guangzhou, Guangdong, China
- Zhang, Lu, the affiliated hospital of Nanjing University of Chinese Medicine, Nanjing, China
- Zhou, Dong, University of Pittsburgh, Pittsburgh, Pennsylvania, United States
Background
Diabetic kidney disease (DKD) is the leading cause of end-stage renal disease. As the most common microvascular complication of diabetes, DKD is a knotty clinical problem in terms of its diagnosis and management. Currently, renal biopsy remains the most reliable approach to distinguishing a true DKD, non-DKD, or mixed form. Our present study focuses on early biomarker identification to monitor the onset of DKD in type2 diabetes (T2D) patients. We expect the identified biomarker could eventually be translated into the solid backup resources to avoid invasive biopsy in patients.
Methods
Based on our previous findings, combine with the data refinery in public repository centers, we identified 3 proteins which play critical roles in the development of DKD, including Serum Amyloid A-1 (SSA1), matrix metalloproteinase-7 (MMP-7), and Tenascin C (TNC). We designed a standard cohort study by dividing the trial into 3 phases: training, testing, and validation phase. In the training phase, we enrolled 202 candidates including DKD or T2D or healthy adults from a single medical center. In the testing phase, we enrolled 680 patients at different stages of DKD or healthy adults from the same medical center. Serum SAA1, MMP-7, and TNC, urinary MMP-7 and TNC were measured. Two machine learning models, linear discriminant analysis (LDA) and support-vector machines (SVM), were applied in this study. The external validation is now pending.
Results
Compared with the healthy adults, serum and urinary SAA1, MMP-7, and TNC were significantly increased at each stage of DKD in patients. However, if directly compared with the T2D patients without kidney disease, none of the above markers could individually serve as an early warning marker to alert the onset of kidney disease in the T2D patients. The diagnostic outcomes analyzed by SVM were shown via the receiver operating characteristic (ROC) for the direct comparison between the T2D and the DKD patients at early stage. In SVM model, combined serum SAA1, MMP-7, and TNC provided area under curve at 0.95 between the T2D and the DKD early stage patients. In addition, a combination of serum SAA1 and MMP-7 is able to more precisely distinguish the boundary between the early and advance stages of DKD.
Conclusion
Analysis using SVM model by combining SAA1, MMP-7, and TNC in serum is promising to predict the onset of kidney complication in T2D patients.