Abstract: PO2349
Pediatric-Specific Models Improve Prediction of Kidney Transplant Survival for Children Under 5 Years Old
Session Information
- Pediatric Nephrology: Glomerular Disease and Transplantation
October 22, 2020 | Location: On-Demand
Abstract Time: 10:00 AM - 12:00 PM
Category: Pediatric Nephrology
- 1700 Pediatric Nephrology
Authors
- Manca Barayre, Florian, Emory University, Atlanta, Georgia, United States
- Greenbaum, Larry A., Emory University, Atlanta, Georgia, United States
- Garro, Rouba, Emory University, Atlanta, Georgia, United States
- Winterberg, Pamela D., Emory University, Atlanta, Georgia, United States
- Patzer, Rachel E., Emory University, Atlanta, Georgia, United States
- Hogan, Julien, Emory University, Atlanta, Georgia, United States
Group or Team Name
- Emory Transplant Center
Background
The kidney allocation system directs high-quality kidneys to pediatric recipients, but the kidney donor profile index (KDPI) used to quantify donor quality may not accurately predict graft survival in small pediatric recipients. We aimed to determine if a pediatric-specific KDPI could improve the prediction of graft longevity for young recipients.
Methods
We evaluated first-time kidney transplantations in pediatric recipients between 1/2005 and 8/2018 in the U.S. (SRTR) and used a Cox model to assess KDPI accuracy for combined primary outcome of death or graft loss in young recipients. We developed an adapted KDPI score for recipients <5 years old of deceased (SP-KDPI) or living (SP-LDKPI) adult donor transplants using multivariate cox regression and scaled these models to allow comparison between living and deceased donors. Models’ accuracies were validated internally by cross-validation.
Results
KDPI C-statistic was 0.52 (95% CI = 0.50 – 0.54) in recipients less than 5 years old. Ethnicity, age, body surface area, gender, cold ischemia time and number of HLA-B mismatches, were significant predictors for deceased donors (SP-KDPI model) while race, age, HLA-B mismatch and donor/recipient body surface area ratio were used in the living donor model (SP-LKDPI). C-statistics were 0.64 (95% CI = 0.57 – 0.70) for SP-KDPI and 0.65 (95% CI = 0.58 – 0.73) for SP-LKDPI. Figure 1 shows allograft survival by donor type and SP-(L)KDPI stratum. The SP-LKDPI model identified 16.8% of living donors with predicted graft survival superior to any deceased donor (denoted as SP-LKDPI <0).
Conclusion
Our adaptation of the KDPI demonstrated a higher accuracy to predict graft loss in young recipients of deceased donors. Furthermore, our SP-LKDPI model may allow direct comparison of living versus deceased donors offered to the youngest recipients.
Observed graft survivals by type of donor and SP-(L)KDPI