Abstract: FR-PO813
The Application of Artificial Intelligence in IgA Nephropathy After Kidney Transplantation
Session Information
- Transplantation: Clinical - Outcomes
November 04, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
Abstract Time: 10:00 AM - 12:00 PM
Category: Transplantation
- 2002 Transplantation: Clinical
Authors
- Liu, Xumeng, Nanjing University, Nanjing, Jiangsu, China
- Lei, Qunjuan, Southern Medical University, Guangzhou, Guangdong, China
- Hou, Xiaoshuai, Ping An Healthcare Technology, Shanghai, Shanghai, China
- Liang, Dongmei, Nanjing University, Nanjing, Jiangsu, China
- Xu, Feng, Nanjing General Hospital of Nanjing Military Command Research Institute of Nephrology, Nanjing, Jiangsu, China
- Liang, Shaoshan, Nanjing General Hospital of Nanjing Military Command Research Institute of Nephrology, Nanjing, Jiangsu, China
- Liang, Dandan, Nanjing General Hospital of Nanjing Military Command Research Institute of Nephrology, Nanjing, Jiangsu, China
- Yang, Fan, Nanjing General Hospital of Nanjing Military Command Research Institute of Nephrology, Nanjing, Jiangsu, China
- Ni, Yuan, Ping An Healthcare Technology, Shanghai, Shanghai, China
- Xie, Guotong, Ping An Healthcare Technology, Shanghai, Shanghai, China
- Zeng, Cai-hong, Nanjing General Hospital of Nanjing Military Command Research Institute of Nephrology, Nanjing, Jiangsu, China
Background
A convolutional neural network named Analytic Renal Pathology System (ARPS) was trained to identify glomerular lesions in immunoglobulin A nephropathy (IgAN). However, whether ARPS can be applied to renal graft biopsy is still unknown. This study aimed to analyze the application of ARPS in transplant patients with IgAN.
Methods
From January 2016 to April 2020, patients diagnosed as IgAN by renal graft biopsy in our center were collected. The performance of ARPS in transplant patients with IgAN was evaluated. The correlation of clinicopathologic data were further analyzed.
Results
A total of 57 patients were enrolled. The median (interquartile range: IQR) age at renal biopsy was 32 (28, 40) years. ARPS could identify the types of glomerular lesions and intrinsic cells in transplant patients with IgAN, achieving F1-scores for different lesions ranged between 72.40% and 96.05%. The ratio of mesangial cells (M), endothelial cells (E), and podocytes (P) was 0.37:0.39:0.25. Compared with autologous IgAN patients (0.41:0.36:0.23), the percentage of E was higher in transplant patients with IgAN. Urine protein level was negatively correlated with the number of P (p<0.05), but positively correlated with mesangial area (p<0.05). Serum creatinine level was negatively correlated with the number of E (p<0.05). The course of disease after transplantation was positively correlated with the glomerular and mesangial area (p<0.05). According to the results of receiver operating characteristic curve, the lower number of E or P, and higher percent segmental sclerosis (SS) or glomerular sclerosis (GS) could indicate poorer prognosis, with all the area under curve of them greater than 0.7 (p<0.05).
Conclusion
ARPS can automatically identify and quantify the types of glomerular lesions and intrinsic cells in transplant patients with IgAN. The ARPS quantified glomerular lesions and intrinsic cells correlated well with key clinical indicators and the prognosis of transplant patients with IgAN.
Funding
- Government Support – Non-U.S.