Improving Kidney Exchange Program Effectiveness and Equity by Using Allele and Eplet HLA Compatibility Technologies
Author(s)
Valentina Peralta Clarke, MSc1, Hans De Ferrante, PhD2, Francisco Perez-Galarce, PhD3, Joris Van de Klundert, PhD4.
1Facultad Ingeneria y Ciencias, universidad adolfo ibanez, Santiago, Chile, 2Mathematics, TU Eindhoven, Eindhoven, Netherlands, 3Faculdad de Negocias y Ingeneria, Universidad de Las Americas, Santiago, Chile, 4Business School, Universidad Adolfo Ibanez, Chile, Vina del Mar, Chile.
1Facultad Ingeneria y Ciencias, universidad adolfo ibanez, Santiago, Chile, 2Mathematics, TU Eindhoven, Eindhoven, Netherlands, 3Faculdad de Negocias y Ingeneria, Universidad de Las Americas, Santiago, Chile, 4Business School, Universidad Adolfo Ibanez, Chile, Vina del Mar, Chile.
OBJECTIVES: Kidney Exchange Programs (KEPs) provide access to the most (cost-)effective treatment of living donor transplantation for patients with a willing yet incompatible donor. The HLA compatibility between the recipient and potential donors is traditionally determined using antigen level data. theoretical and technological advancements presently enable to base HLA compatibility on allele and eplet level data. These technologies can improve transplant compatibility assessment and effectiveness yet may reduce equity of access and outcomes. The very scarce evidence on the use of allele and eplet level data in KEPs solely considers effectiveness and is from small scale case studies. This study aims to strengthen evidence on the impact on effectiveness and equity of long run KEP performance achieved when adopting allele and eplet level technologies and to propose KEP optimization methods that jointly maximize effectiveness and equity.
METHODS: Long run KEP performance is determined using discrete event simulation on a constructed OPTN/SRTR population of 991 KEP participants. HLA level conversions are based on Eurotransplant tables and HLAmatchmaker. KEP runs are optimized using Integer Linear Programming. Equity weights that jointly optimize effectiveness and equity are set using Bayesian optimization (in Optuna).
RESULTS: Use of antigen level data in KEP optimization resulted in significantly more favorable access and compatibility for (non-Hispanic) Whites. Use of the allele and eplet level data improved long run KEP performance but especially allele level data enlarged the inequities, with transplant access for Asians particularly diminished. For each of antigen, allele, and eplet level data, different equity weights can optimize effectiveness while virtually eliminating inequities.
CONCLUSIONS: Use of allele and eplet level data in KEPs can improve effectiveness yet negatively impact equity. Using appropriate optimization models and methods, these novel HLA compatibility technologies the negative impact on equity can be avoided.
METHODS: Long run KEP performance is determined using discrete event simulation on a constructed OPTN/SRTR population of 991 KEP participants. HLA level conversions are based on Eurotransplant tables and HLAmatchmaker. KEP runs are optimized using Integer Linear Programming. Equity weights that jointly optimize effectiveness and equity are set using Bayesian optimization (in Optuna).
RESULTS: Use of antigen level data in KEP optimization resulted in significantly more favorable access and compatibility for (non-Hispanic) Whites. Use of the allele and eplet level data improved long run KEP performance but especially allele level data enlarged the inequities, with transplant access for Asians particularly diminished. For each of antigen, allele, and eplet level data, different equity weights can optimize effectiveness while virtually eliminating inequities.
CONCLUSIONS: Use of allele and eplet level data in KEPs can improve effectiveness yet negatively impact equity. Using appropriate optimization models and methods, these novel HLA compatibility technologies the negative impact on equity can be avoided.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR131
Topic
Clinical Outcomes, Health Policy & Regulatory, Methodological & Statistical Research
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics
Disease
Surgery, Urinary/Kidney Disorders