Abstract: SA-PO0846
Plasma Proteomic Endotypes Refine Diagnosis and Risk Assessment Across Primary Glomerular Diseases
Session Information
- Glomerular Diseases: Management, Evolving Strategies, and Practice-Changing Advances
October 24, 2026 | Location: Exhibit Hall A, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1402 Glomerular Diseases: Clinical, Outcomes, and Therapeutics
Authors
- Hong, Shaun G., Purdue University Weldon School of Biomedical Engineering, West Lafayette, Indiana, United States
- Park, Sehoon, Seoul National University Hospital Department of Internal Medicine, Jongno-gu, Seoul, Korea (the Republic of)
- Kim, Dong Ki, Seoul National University Hospital Department of Internal Medicine, Jongno-gu, Seoul, Korea (the Republic of)
Background
Biopsy-based diagnosis does not fully capture biologic heterogeneity in primary glomerular disease. We tested whether plasma proteomics could define reproducible cross-disease molecular endotypes beyond histology.
Methods
We measured 2,832 plasma proteins by Olink Explore HT in discovery (n=155), external validation (n=79), remission follow-up (n=28), and independent IgAN prognostic validation (n=151) cohorts. Factor analysis identified co-varying modules, and k-means clustering of selected module scores defined endotypes in discovery only. Frozen loadings and centroids were then applied to all other cohorts.
Results
Twelve proteomic modules were identified, including immune-inflammatory, podocyte-nephrotic, and kidney-function-decline programs. Continuous module scores improved eGFR modeling beyond histology alone in both discovery and validation (adjusted R2 0.17 to 0.43 and 0.17 to 0.51). Three endotypes defined a biologic gradient across diagnoses. In IgAN, the immune-inflammatory module independently predicted 50% eGFR decline (HR 2.73, 95% CI 1.34-5.53), and event rates rose from 2.6% in the lowest quartile to 64.9% in the highest. In MCD, module profiles differentiated clinical course (AUC 0.77) and remained abnormal despite remission.
Conclusion
Circulating proteomics complements biopsy classification by identifying cross-disease endotypes, improving clinical modeling, and revealing prognostic heterogeneity in IgAN and persistent molecular abnormality in MCD.
Acknowledgment
We are grateful to the participating patients, research coordinators, and clinical teams for their contributions to sample collection and clinical phenotyping. UK Biobank resources were used under approved application 53799.
| Domain | Metric | Value |
| Clinical modeling | Discovery eGFR adjusted R2 | 0.17 to 0.43 |
| Clinical modeling | Discovery eGFR adjusted R2 | 0.17 to 0.51 |
| Clinical modeling | Discovery log-proteinuria adjusted R2 | 0.44 to 0.50 |
| Clinical modeling | Validation log-proteinuria adjusted R2 | 0.58 to 0.69 |
| IgAN prognosis | Module 2 HR for 50% eGFR decline | 2.73 (1.34-5.53) |
| IgAN prognosis | Module 2 event rate, Q1 to Q4 | 2.6% to 64.9% |
| MCD course | Module-only LOOCV AUC | 0.77 |
Funding
- Government Support – Non-U.S.