Detailed Abstract
[Liver Oral Presentation 2 - Liver (Transplantation)]
[LV OP 2-S8] Advancing Graft Failure Prediction Post-Liver Transplantation: External Validation of a Model Using Aspartate Aminotransferase, Total Bilirubin, And Coagulation Factors
Sunghyo AN 1, Jinsoo RHU 1, Jong Man KIM 1, Gyu-Seong CHOI 1, Jae-Won JOH 2
1 Department of Surgery, Samsung Medical Center, REPUBLIC OF KOREA, 2 Department of Surgery, Samsung Changwon Hospital, REPUBLIC OF KOREA
Background : Developed from single-institution data, our model predicting early graft failure post-liver transplantation showed high internal validity, warranting external validation for broader applicability.
Methods : Our study externally validated a Cox regression model for early graft failure post-liver transplantation, using post-transplant aspartate aminotransferase, total bilirubin, and international normalized ratio. We analyzed data from SMC (Living: 342, Deceased: 114; 2019-2021), SNUH (Living: 716, Deceased: 157; 2015-2021), and Severance Hospital (Living: 967, Deceased: 428; 2005-2021). The model's efficacy was compared with MEAF and EAD benchmarks through C-index and time-dependent AUC.
Results : The C-index of the model for SMC living donor (0.96,CI=0.91–1.00) was significantly higher compared to those of both MEAF (0.9, P=0.003) and EAD (0.84, P=0.04) while C-index for SMC deceased donor (0.96, CI=0.94–0.99) was only significantly higher compared to C-index of EAD. (0.7, CI=0.58-0.81, P<0.001) The C-index of the model for SNUH living donor (0.81,CI=0.7–0.93) was significantly higher compared to C-index of EAD (0.65, P=0.006) while C-index for SNUH deceased donor (0.89, CI=0.77–1) was significantly higher compared to C-index of EAD. (0.7, P=0.009) The C-index of the model for Severance hospital living donor (0.89,CI=0.8–0.96) was significantly higher compared to those of both MEAF (0.84,P=0.03) and EAD (0.77,P=0.03) while C-index for Severance hospital deceased donor (0.88, CI=0.82–0.95) was significantly higher compared to those of both MEAF (0.75,P=0.007) and EAD (0.67,P<0.001)
Conclusions : The external validation of our Cox regression model demonstrated its superiority over MEAF and EAD benchmarks in predicting early graft failure post-liver transplantation, indicating its potential for widespread clinical application.
Methods : Our study externally validated a Cox regression model for early graft failure post-liver transplantation, using post-transplant aspartate aminotransferase, total bilirubin, and international normalized ratio. We analyzed data from SMC (Living: 342, Deceased: 114; 2019-2021), SNUH (Living: 716, Deceased: 157; 2015-2021), and Severance Hospital (Living: 967, Deceased: 428; 2005-2021). The model's efficacy was compared with MEAF and EAD benchmarks through C-index and time-dependent AUC.
Results : The C-index of the model for SMC living donor (0.96,CI=0.91–1.00) was significantly higher compared to those of both MEAF (0.9, P=0.003) and EAD (0.84, P=0.04) while C-index for SMC deceased donor (0.96, CI=0.94–0.99) was only significantly higher compared to C-index of EAD. (0.7, CI=0.58-0.81, P<0.001) The C-index of the model for SNUH living donor (0.81,CI=0.7–0.93) was significantly higher compared to C-index of EAD (0.65, P=0.006) while C-index for SNUH deceased donor (0.89, CI=0.77–1) was significantly higher compared to C-index of EAD. (0.7, P=0.009) The C-index of the model for Severance hospital living donor (0.89,CI=0.8–0.96) was significantly higher compared to those of both MEAF (0.84,P=0.03) and EAD (0.77,P=0.03) while C-index for Severance hospital deceased donor (0.88, CI=0.82–0.95) was significantly higher compared to those of both MEAF (0.75,P=0.007) and EAD (0.67,P<0.001)
Conclusions : The external validation of our Cox regression model demonstrated its superiority over MEAF and EAD benchmarks in predicting early graft failure post-liver transplantation, indicating its potential for widespread clinical application.
SESSION
Liver Oral Presentation 2
Room F 3/22/2024 4:30 PM - 5:30 PM