Risk score distribution and one-year MACE prediction by CAD-OLT and CAD-LT in liver transplant candidates
C Huerta, W Giesing, M Mogg, A Abdulmanan, C Delgado, B Barrientos, H Cheema, A Khetani, N KhanAbstract
Background
With an ageing population and rising metabolic comorbidities among liver transplant candidates, perioperative cardiovascular risk assessment has become increasingly important. Pre-transplant cardiac evaluation strategies before liver transplantation vary widely across centres. Risk stratification tools include the Cardiovascular Risk in Orthotopic Liver Transplantation (CAR-OLT) score (predicting perioperative major cardiovascular events) and Coronary Artery Disease in Liver Transplantation (CAD-LT) score (predicting significant coronary artery disease). These were developed to estimate perioperative cardiovascular risk, yet their real-world distribution across evaluation strategies and performance for predicting post-transplant events remain incompletely characterised.
Purpose
To compare CAR-OLT and CAD-LT score distributions between invasive (left heart catheterisation [LHC]-based) and non-invasive strategies and assess discrimination for one-year major adverse cardiovascular events (MACE).
Methods
We conducted a retrospective cohort study of 637 adults undergoing orthotopic liver transplantation between 1 January 2017 and 31 December 2024. Cardiac evaluation was categorised as LHC-based (invasive) or non-invasive, primarily stress testing. CAR-OLT and CAD-LT scores were calculated for all patients. One-year MACE was defined as myocardial infarction, new or decompensated heart failure, new or worsening pulmonary hypertension, cardiogenic shock, cardiac arrest, arrhythmia, cardiovascular death, or cerebrovascular accident. Score distributions were compared between strategies using non-parametric testing. Discrimination for one-year MACE was assessed using the area under the receiver operating characteristic curve (AUC).
Results
Patients undergoing LHC-based evaluation had higher baseline predicted risk. Median CAR-OLT score was 32.0 (IQR 25.0–39.2) in the invasive group versus 23.0 (IQR 16.0–31.0) in the non-invasive group (p<0.001). Median CAD-LT score was 12.0 (IQR 10.0–15.0) versus 8.0 (IQR 6.0–10.0), respectively (p<0.001). When evaluated for the prediction of one-year MACE, discrimination was modest. The AUC was 0.651 for CAR-OLT and 0.565 for CAD-LT, indicating limited ability of either model to differentiate patients who experienced events from those who did not.
Conclusion
CAR-OLT and CAD-LT scores were higher among patients selected for invasive pre-transplant cardiac evaluation, supporting the presence of risk-based selection. However, both models show only modest discrimination for one-year post-transplant MACE, suggesting a disconnect between preoperative risk prediction and post-transplant cardiovascular outcomes. These findings highlight the need for more accurate, transplant-specific risk stratification approaches to better guide pre-transplant evaluations and optimise outcomes.Figure 1:A/B/C/DFigure 2:ROC Curve