Abstract Body (Do not enter title and authors here): Introduction: Cirrhosis, a late-stage liver disease, is associated with numerous systemic complications, making long-term mortality prediction challenging. Traditional models like Child-Pugh and MELD are primarily designed for short-term (3-month) mortality. This study evaluates the impact of incorporating echocardiographic (echo) variables in predicting long-term mortality in cirrhosis patients.
Methods: A retrospective cohort study was conducted using data from 60 patients with clinically diagnosed cirrhosis from Hippokration Hospital, Greece, spanning 2018 to 2022. Patients were followed for up to 52 months. Exclusion criteria included pre-existing cardiovascular diseases and significant co-morbidities. Patients were categorized into survived to end of study (n=31) and died during study (n=29) groups. Outcomes were analyzed using multivariate logistic regression and Cox proportional hazards models, and include parameters for age, blood pressure, lab-based biomarkers (potassium, pro-brain natriuretic peptide, hemoglobin, platelet count, aspartate transaminase, gamma-glutamyl transferase, uric acid), and echo/ecg measurements (QT interval, left ventricular (LV) ejection fraction, LV global longitudinal strain (GLS), left atrial GLS, annulus velocities, and E/A ratio).
Results: The multivariate logistic regression model (MELD+) that included echo variables showed a 10% improvement in area under the curve (AUC: 0.901) compared to the MELD model alone (AUC: 0.817). MELD+ improved risk classification accuracy by nearly 10% with a conservative threshold. The Cox proportional hazards model with MELD+ also demonstrated superior performance (AUC: 0.923; C-index: 0.875; Brier Score: 0.097) compared to the MELD model alone (AUC: 0.892; C-index: 0.831; Brier Score: 0.134). The MELD+ model showed less variation in hazard ratio confidence intervals, indicating more consistent risk estimation.
Conclusion: Incorporating echocardiographic measurements enhances the accuracy of long-term mortality predictions in cirrhosis patients. These findings suggest that cardiac imaging should be considered in risk assessments for cirrhosis patients, particularly for predicting survival beyond three months.