Abstract: SA-PO0018
Radiomics-Based Adipose Profiling Predicts Kidney Outcomes in Lupus Nephritis
Session Information
- Intelligent Imaging and Omics: Phenotyping and Risk Stratification
November 08, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Shi, Xiaolei, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
- Zhou, Yingting, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
- Huang, Yilong, Department of Radiology, Guangdong Provincial People’s Hospital, Guangzhou, Guangdong, China
- Tang, Rui Han, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
- Xia, Xi, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
- Liu, Zaiyi, Department of Radiology, Guangdong Provincial People’s Hospital, Guangzhou, Guangdong, China
- Chen, Wei, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, China
Background
Lupus nephritis (LN) is a serious complication of systemic lupus erythematosus (SLE), and reliable biomarkers for predicting renal prognosis remain limited. Emerging evidence suggests that adipose tissue plays a key role in immune-metabolic regulation; however, the prognostic significance of adipose tissue distribution in LN remains unclear.
Methods
We conducted a retrospective study involving 137 patients with biopsy-proven lupus nephritis (LN) (from January 2010 to December 2023) and 203 healthy controls matched for age, sex, and body mass index (BMI). Subcutaneous fat volume (SFV), visceral fat volume (VFV), and intermuscular fat volume (IMFV) were quantified from non-contrast-enhanced chest CT images using 3D U-Net segmentation via AutoPanoM, developed by Liu’s lab. Clinical, laboratory, and pathological data were collected. The primary outcome was defined as adverse renal events, including a ≥30% decline in estimated glomerular filtration rate (eGFR) or progression to end-stage renal disease (ESRD). Kaplan–Meier survival analysis and Cox proportional hazards models were used to identify prognostic factors.
Results
Among the 137 LN patients (median age: 32.6 years; 75.9% female), 21.9% experienced adverse renal events during a median follow-up of 38.5 months. Total fat volume (SFV, VFV, and IMFV) was significantly lower in LN patients compared to controls. Kaplan-Meier analysis revealed that patients with a lower intermuscular-to-subcutaneous fat volume ratio (IMSVR < 0.12) had a significantly higher risk of adverse renal events compared to those with higher IMSVR (p = 0.036). In multivariate analysis, a lower IMSVR remained independently associated with adverse renal events (HR = 2.81; 95% CI: 1.07–7.40; p = 0.036).
Conclusion
Quantitative CT-derived IMSVR may serve as a novel imaging biomarker for predicting long-term renal outcomes in LN. These findings underscore the potential of radiomics-based adipose profiling to enhance risk stratification in LN management.