Abstract: TH-PO0487
Validation of the Klinrisk Machine Learning Model in Patients with IgAN
Session Information
- Glomerular Diseases: Clinical, Outcomes, and Therapeutics Research - IgAN
October 22, 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
- Barr, Bryce, University of Manitoba Max Rady College of Medicine, Winnipeg, Manitoba, Canada
- Ferguson, Thomas W., University of Manitoba Max Rady College of Medicine, Winnipeg, Manitoba, Canada
- Tangri, Navdeep, University of Manitoba Max Rady College of Medicine, Winnipeg, Manitoba, Canada
Background
Management of IgA nephropathy (IgAN) has changed in the past five years, with greater focus on immunosuppressive therapies, and recognition of risk earlier in the disease course. The Klinrisk prediction model represents a potential tool for more proximal risk prediction using routinely collected data. We sought to validate the Klinrisk model in a population-based cohort of adults with IgAN in the Canadian province of Manitoba and in the CureGN cohort.
Methods
The Manitoba cohort was identified from the Manitoba Glomerular Diseases Registry, and included patients diagnosed with IgAN from 2006 to 2021, followed through 2023. The CureGN cohort comprised all eligible adult patients in the Cure Glomerulonephropathy consortium. Inclusion criteria were age 18 or older, biopsy-proven IgAN without IgA vasculitis, with an available basic metabolic panel and urine ACR or PCR within 90 days of biopsy. Discrimination was assessed using the area under the receiver operating characteristic curve (AUC). Calibration was assessed using Brier score, and calibration plots to compare predicted with observed risk.
Results
A total of 318 patients from Manitoba and 216 patients from CureGN with biopsy-proven IgA nephropathy were included, with 119 and 59 patients experiencing events in Manitoba and CureGN, respectively. Mean age at diagnosis was 43.0 years in Manitoba and 40.4 years in CureGN. Mean eGFR was 59 and 67 mL/min/1.73m2 in Manitoba and CureGN and median ACR was 156 in Manitoba and 164 mg/mmol in CureGN. Discrimination in Manitoba was very good, with an AUC of 0.84 (95% CI 0.79-0.89) at 2 years and 0.81 (95% CI 0.75-0.87) at 5 years, and excellent in CureGN, with an AUC of 0.91 (95% CI 0.86-0.96) at 2 years and 0.90 (95% CI 0.85-0.95) at 5 years. Brier scores were 0.139 and 0.165 at 2 and 5 years in Manitoba, and 0.088 and 0.136 at 2 and 5 years in CureGN.
Conclusion
In two independent IgAN cohorts, the Klinrisk model accurately predicted CKD progression from routinely collected labs alone, without requiring histologic scoring. These findings support its use for risk stratification at any time during the course of disease.
Acknowledgment
Funding for the CureGN consortium is provided by U24DK100845, U01DK100846, U01DK100876, U01DK100866, and U01DK100867 from the NIDDK. Patient Recruitment is supported by NephCure.
Funding
- Commercial Support – Klinrisk Inc.