ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: SA-PO0017

Towards Precision Prognostication in IgAN: Artificial Intelligence (AI)-Infused Pathomic Approach

Session Information

Category: Artificial Intelligence, Digital Health, and Data Science

  • 300 Artificial Intelligence, Digital Health, and Data Science

Authors

  • Cortez Ferreira, Beatriz, Universidade de Coimbra, Coimbra, Coimbra District, Portugal
  • Pardinhas, Clara, Universidade de Coimbra, Coimbra, Coimbra District, Portugal
  • Paul, Anindya S., University of Florida, Gainesville, Florida, United States
  • Rodrigues, Luis, Universidade de Coimbra, Coimbra, Coimbra District, Portugal
  • Abdelazim, Haitham M, University of Florida, Gainesville, Florida, United States
  • Sa, Helena Oliveira, Universidade de Coimbra, Coimbra, Coimbra District, Portugal
  • Eadon, Michael T., Indiana University, Bloomington, Indiana, United States
  • Sarder, Pinaki, University of Florida, Gainesville, Florida, United States
Background

Predicting long-term renal outcomes in IgA nephropathy (IgAN) remains challenging. Established predictors include histologic features (e.g., mesangial/endocapillary hypercellularity, segmental sclerosis, tubular atrophy, crescents) and clinical parameters like eGFR. Tools like MEST-C and the IgAN Prediction Tool (IgAN-PT) aim to improve risk stratification but may miss morphometric data now accessible via AI-powered computational pathology. This study compared the predictive performance of conventional models with an AI-enhanced pathomic approach.

Methods

We retrospectively analyzed biopsy-proven IgAN cases (1998–2023) at ULS Coimbra with ≥5 years of follow-up. Slides were digitized and analyzed using the Computational Renal Pathology Suite (ComPRePS) [https://compreps.wiki/home], extracting 78 glomerular pathomic features via a human-AI loop. Expert renal pathologists provided feedback to refine AI outputs. Clinical data and MEST-C features were also assessed. The primary outcome was ≥50% decline in eGFR or initiation of renal replacement therapy within 5 years. We used forward stepwise logistic regression and AUC-ROC analysis for performance evaluation.

Results

Among 58 patients (median follow-up: 104.7 months), a composite model integrating clinical and pathomic features identified four key predictors: age and eGFR at biopsy, plus two pathomic variables—maximum distance transform by object area luminal space and mean nuclear area, quantifying endocapillary and mesangial hypercellularity. This model achieved an AUC of 0.89, outperforming IgAN-PT (AUC=0.73) and MEST-C (AUC=0.57).

Conclusion

While clinical and histologic scores remain useful, integrating AI-derived pathomic data via ComPRePS significantly improved predictive accuracy. This approach demonstrates the promise of expert-guided computational pathology for enhanced IgAN prognostication.

Funding

  • NIDDK Support

Digital Object Identifier (DOI)