Refining the Liver Donor Risk Index With Machine Perfusion: A Bayesian Approach

Area of Science:

• Transplantation research
• Organ procurement analytics
• Statistical modeling in medicine

Background:

• The Donor Risk Index (DRI) is a standard tool for assessing liver transplant allograft risk.
• Current DRI models do not incorporate machine perfusion (MP), a growing practice in organ preservation.

Purpose of the Study:

• To integrate machine perfusion (MP) data into the Donor Risk Index (DRI) using Bayesian updating.
• To develop an updated risk model (DRI-MP) that reflects current liver transplant practices.

Main Methods:

• Bayesian updating was employed to incorporate MP into the DRI framework, creating the DRI-MP model.
• A Bayesian proportional hazards model utilized Organ Procurement and Transplantation Network data (Jan 2022-June 2024).
• Model performance was evaluated using Harrell Concordance-statistic, calibration plots, and Brier scores.

Main Results:

• The DRI-MP model, defined as DRI × 0.7 for MP cases, demonstrated improved 90-day graft survival discrimination (Harrell Concordance-statistic: 0.546 vs 0.535, P = 0.040).
• The updated model maintained robust calibration, indicating reliable risk predictions.

Conclusions:

• The Bayesian-updated DRI-MP offers a modest improvement in donor risk discrimination for liver transplantation.
• This updated model reflects contemporary transplant practices and provides a practical tool with continuity from the original DRI.