Abstract: FR-PO900
Analysis of Clinical Risk Factors of Prognosis in IgA Nephropathy Based on Machine-Learning Approaches
Session Information
- IgA Nephropathy: Clinical, Outcomes, and Therapeutics
October 25, 2024 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1402 Glomerular Diseases: Clinical, Outcomes, and Therapeutics
Author
- Zhu, Qin, Hangzhou Hospital of Traditional Chinese Medicine, Hangzhou, Zhejiang, China
Background
Immunoglobulin A nephropathy (IgAN) is one of the leading causes of end-stage kidney disease (ESKD). Many studies have shown the significance of pathological manifestations in predicting the outcome of patients with IgAN. Evaluating prognosis by using baseline clinical characteristics is beneficial for both clinicians and patients.
Methods
A dataset of 892 patients with IgAN is used to develop prediction model for estimating the prognosis of patients with IgAN based on the cox regression and random forest algorithm. The 5-year ESKD prediction models using clinical variables were developed to evaluate the probability of ESKD or doubled serum creatinine(Scr). Cross-validation were implemented to mitigate bias, while the area under the receiver operating characteristic curve (AUROC) was used for model evaluations.
Results
For the 5-year ESKD prediction model, the base model achieved an AUROC of 0.86 (95% CI: 0.75–0.97). Six clinical factors exhibited statistical significance (p < 0.05) in predicting IgAN prognosis: BMI (HR: 1.025, 95% CI: 1.003-1.047, p = 2.25×10−2), cholesterol (HR: 1.24, 95% CI: 1.097-1.398, p = 5.42×10−4), and T in Oxford classification (HR: 2.747, 95% CI: 1.573-4.797, p = 3.79×10−4) were identified as risk factors, while estimated glomerular filtration rate (eGFR) (HR: 0.964, 95% CI: 0.947-0.982, p = 1.16×10−4), use of immunosuppressive drugs (HR: 0.498, 95% CI: 0.264-0.941, p = 3.19×10−2), and renal pathology IgA (HR: 0.603, 95% CI: 0.386-0.940, p = 2.57×10−2) exhibited a protective effect against ESKD or doubled Scr.
Conclusion
A prediction model incorporating clinical characteristics was constructed to estimate prognosis in patients with IgAN. The findings suggest that BMI, cholesterol, and T score in Oxford classification increase the risk of ESKD or doubled Scr, while higher eGFR, immunosuppressive drug use, and renal pathology IgA confer a protective effect. Further validation through randomized controlled trials is warranted to corroborate these findings.