Identifying Factors Predicting Kidney Graft Survival in Chile Using Elastic-Net-Regularized Cox's Regression

Area of Science:

• Nephrology
• Transplantation immunology
• Biostatistics

Background:

• Kidney graft survival is crucial for patients with end-stage renal disease.
• Predictive models can optimize donor-recipient matching and improve transplant outcomes.
• Chilean kidney transplant data is limited, necessitating robust statistical approaches.

Purpose of the Study:

• To develop a predictive statistical model for kidney graft survival in the Chilean population.
• To identify donor, recipient, and transplant characteristics associated with graft failure.
• To provide objective criteria for improving deceased kidney allocation in Chile.

Main Methods:

• A retrospective multicenter study analyzed data from 822 adult kidney transplant recipients (1998-2018).
• Cross-validated regularized Cox regression with Elastic Net penalty was used for variable selection.
• Multivariate Cox regression estimated hazard ratios for predictors of death-censored graft failure.

Main Results:

• Seven variables were significantly associated with death-censored graft failure.
• Donor factors included age, male sex, history of hypertension, and history of diabetes.
• Recipient factors included years on dialysis and previous solid organ transplant history.
• Transplant factor included number of HLA mismatches.

Conclusions:

• A robust risk model for kidney graft failure was developed and validated for the Chilean population.
• The model identified objective criteria beyond HLA mismatches for predicting graft failure.
• These findings can inform and improve the efficiency of deceased kidney allocation strategies in Chile.