Kidney Week

Abstract: TH-PO803

Discovery and Validation of a Kidney Rejection Prediction Model Using Urinary CXCL9 and CXCL10

Session Information

• Transplantation: Clinical - 2
  October 24, 2024 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Transplantation

• 2102 Transplantation: Clinical

Authors

• Chen, Kenny W., Northwestern University Feinberg School of Medicine, Chicago, Illinois, United States

Kenny W. Chen, MS, MSc

• Zhao, Lihui, Northwestern University Feinberg School of Medicine, Chicago, Illinois, United States

Lihui Zhao

• Park, Sookhyeon, Northwestern University Feinberg School of Medicine, Chicago, Illinois, United States

Sookhyeon Park, MD, MS

• Paudel, Sujay Dutta, Northwestern University Feinberg School of Medicine, Chicago, Illinois, United States

Sujay Dutta Paudel, MD

• Lacombe, Ronnie, Eurofins Viracor BioPharma Services, Lenexa, Kansas, United States

Ronnie Lacombe, PhD

• Kleiboeker, Steven, Eurofins Viracor BioPharma Services, Lenexa, Kansas, United States

Steven Kleiboeker, PhD, DVM

• Friedewald, John J., Northwestern University Feinberg School of Medicine, Chicago, Illinois, United States

John J. Friedewald, MD, FASN

Background

Urinary chemokines (CXCL9 + CXCL10) have shown to be non-invasive biomarkers in diagnosing kidney rejection. We aim to discover and validate a rejection prediction model combining the two by assessing the diagnostic performance in two independent cohorts.

Methods

We analyzed 638 biopsy paired urine samples paired from the CTOT-08 trial for discovery and then samples from an internal study cohort as a validation. We fit univariate logistic regression models using urine CXCL9 and CXCL10 normalized to urine creatinine and a logistic regression model with both to predict kidney rejection. AUC was used to evaluate the model's diagnostic performance.

Results

The univariate logistic model with CXCL9 had an AUC of 0.694 while the univariate model with CXCL10 had an AUC of 0.634. In the logistic model with both CXCL9 and CXCL10, the AUC was 0.678. In the validation cohort, the univariate model with CXCL9 had an AUC of 0.674, the univariate model with CXCL10 had an AUC of 0.677 and the model with both had an AUC of 0.678.

Conclusion

Incorporating urinary biomarkers CXCL9 and CXCL10 into prediction models to monitor kidney rejection resulted in moderately high diagnostic value (AUC: 0.63 – 0.68). This performance held in the external validation cohort. The combination of both biomarkers did not significantly improve model performance.

Funding

• Other NIH Support – Eurofins - Viracor