Wang M, Shui AM, Rodriguez Peralvarez M
… +10 more, Mauro E, Duarte-Rojo A, Rahimi R, King EA, Verna EC, Ganger D, Ladner DP, Pearson M, Kappus MR, Lai JC
The sodium-corrected Gender-Equity Model for Liver Allocation (GEMA-Na) was developed to mitigate gender disparities in liver allocation, which arise in part from lower muscle mass in women compared with men. Frailty, a...The sodium-corrected Gender-Equity Model for Liver Allocation (GEMA-Na) was developed to mitigate gender disparities in liver allocation, which arise in part from lower muscle mass in women compared with men. Frailty, a measure of muscle health more prevalent in women than in men, is a strong predictor of waitlist mortality (WLM) and can be feasibly assessed in cirrhosis patients using the Liver Frailty Index (LFI). We sought to determine whether LFI could improve WLM risk prediction beyond GEMA-Na. Ambulatory adults with cirrhosis awaiting liver transplant (LT) from the Functional Assessment in Liver Transplantation (FrAILT) study were included. The primary outcome was WLM (death or delisting due to being too sick). Performance of GEMA-Na versus GEMA-Na+LFI Cox proportional hazard models was assessed using Uno's concordance statistic, continuous net reclassification index (NRI), and integrated discrimination improvement (IDI) at 3, 6, and 12 months and in women. Among 1435 patients, 42% were women. Median (IQR) GEMA-Na was 19 (16-23), MELD-Na was 19 (16-23), MELD 3.0 was 19 (16-23), and LFI was 3.9 (3.4-4.3). Median follow-up was 11 months (5-25). WLM was 3.5%, 7.3%, and 13.0% at 3, 6, and 12 months. Performance characteristics demonstrate improved model prediction with GEMA-Na+LFI versus GEMA-Na. Incorporating LFI into GEMA-Na enhanced WLM prediction, including among women, a population historically disadvantaged in liver allocation. While GEMA-Na reduces gender disparities related to renal dysfunction, the added prognostic value of frailty highlights muscle health as an independent and actionable determinant of outcomes. These findings support routine frailty assessment to better stratify risk and to potentially guide timely interventions such as prehabilitation or expedited LT.
Frailty before liver transplantation (LT) is a critical determinant of outcomes after LT. While frailty often improves after LT, it persists-and even worsens-in some. The extent to which post-LT frailty impacts longer-te...Frailty before liver transplantation (LT) is a critical determinant of outcomes after LT. While frailty often improves after LT, it persists-and even worsens-in some. The extent to which post-LT frailty impacts longer-term outcomes has not been characterized. Adults with cirrhosis who underwent LT between 2013 and 2023 from 9 centers with an ambulatory frailty assessment [by Liver Frailty Index (LFI)] 12 months (±6 mo) post-LT were included. Kaplan-Meier methods were used to estimate post-LT survival beyond 1 year (1y) post-LT. Associations between 1y post-LT frailty and subsequent mortality were assessed using Cox regression. Linear regression was used to explore the association between 1y post-LT frailty and health-related quality of life (HrQoL) using the 36-Item Short Form Survey. Among 1097 LT recipients, 144 (13%) patients were frail 1y post-LT. Those who were frail 1y post-LT were more often women (47% vs. 33%), had MASLD (40% vs. 19%), diabetes (44% vs. 30%), stroke (3.5% vs. 0.6%), and pre-LT frailty (37% vs. 16%) (all p ≤0.001). At 3, 5, and 7 years after the 1y post-LT assessment, mortality was higher among those who were frail versus not frail 1y post-LT (13% vs. 6%, 24% vs. 10%, and 32% vs. 16%). After multivariate adjustment, 1y post-LT frailty was associated with a 2.0 times increased risk of post-LT mortality (95% CI 1.2-3.4, p =0.01) and worse physical HrQoL (estimate -9.99, 95% CI -12.5 to -7.5, p <0.001). Patients who were frail 1y after LT experienced >2-fold increased risk of long-term mortality and poorer physical HrQoL. Patients who transitioned from not frail pre-LT to frail 1y post-LT had a 2.5-fold increased risk of post-LT mortality (95% CI 1.3-4.8, p =0.005) compared with patients who remained not frail. These findings suggest the 1y period after LT may be a critical window for evaluating whether targeted frailty mitigation strategies may improve long-term survival among LT recipients.
Ge WH, Bishara A, Braun H
… +2 more, Simonetto DA, Ge J
Liver Transpl
· 2026 May · PMID 42190268
·
Full text
Artificial intelligence (AI) has emerged as a transformative force in liver transplantation (LT), spanning patient selection, donor-recipient matching, intraoperative management, and post-transplant care. Historically, a...Artificial intelligence (AI) has emerged as a transformative force in liver transplantation (LT), spanning patient selection, donor-recipient matching, intraoperative management, and post-transplant care. Historically, applications have relied on traditional machine learning (ML) and deep learning (DL) techniques trained on structured clinical datasets to improve risk stratification and outcome prediction. Since 2023, however, advances in generative AI (GenAI) and large language models (LLMs) have expanded the scope of AI in LT, enabling multimodal reasoning and the extraction of clinically meaningful information from unstructured data. This review synthesizes recent AI-driven developments across the LT patient journey, from advanced chronic liver disease management to long-term post-transplant outcomes. Representative applications include ML-based models to optimize treatment selection and waitlist prioritization for hepatocellular carcinoma, random forest and gradient-boosting approaches to improve donor utilization and predict graft survival, and DL-based imaging models for graft volumetry and steatosis assessment. Post-transplant innovations encompass neural network models for graft injury classification, recurrent architectures for immunosuppression dosing and fibrosis prediction, multi-task frameworks for complication risk stratification, and integrative omics-based approaches to detect rejection and dysfunction. Emerging GenAI applications include LLM-enabled digital phenotyping of social determinants of health and multi-agent simulations of transplant selection committees. Despite rapid progress, translation into routine practice remains limited by challenges in local validation, workflow integration, and infrastructure requirements. Future opportunities lie in incorporating underutilized data domains, such as psychosocial factors, -omics, and intraoperative physiologic streams, and leveraging GenAI to enhance clinical decision support, research scalability, and patient-clinician communication. AI is poised not only to refine prediction but to reshape the conceptual and operational landscape of liver transplantation.
Patients with cirrhosis awaiting transplantation have well-documented reductions in aerobic capacity, yet the physiological responses during functional assessments that underlie these limitations are not fully understood...Patients with cirrhosis awaiting transplantation have well-documented reductions in aerobic capacity, yet the physiological responses during functional assessments that underlie these limitations are not fully understood. This study compared the physiological responses during a six-minute walk test (6MWT) in patients with cirrhosis awaiting transplantation to age-matched controls. Thirty participants with cirrhosis awaiting transplantation (11 female, 58±7 y, 77% Child-Pugh B/C) and 20 controls (5 female; 62±11 y) were enrolled. Heart rate (HR), oxygen consumption (VO 2 ; portable device), and calf muscle oxygenation (SmO 2 ; near-infrared spectroscopy) were continuously measured before (5-min), during, and after (5-min) the 6MWT. HR and VO 2 recovery kinetics were analyzed using non-linear exponential decay modeling (OriginLab, 2023b) to calculate the mean response time (MRT), while calf muscle oxygenation (SmO 2 ) was assessed using the half-time recovery metric ( t50 ). At rest, the cirrhosis group had a higher HR ( p =0.020) and lower calf SmO 2 ( p =0.036) compared with controls, while VO 2 was not significantly different between groups. The cirrhosis group walked a shorter distance than controls (414±131 m vs. 624±72 m; p <0.001), and had a significantly lower HR, VO 2 , and tissue deoxygenation during exercise. Despite these exercise-related impairments, recovery kinetics, measured by HR MRT and SmO 2t50 , were comparable between groups ( p >0.999 and p =0.151, respectively). In contrast, VO 2 MRT was significantly slower in the cirrhosis group (85±23 s vs. 61±18 s; p <0.001). Individuals with cirrhosis awaiting transplantation have reduced aerobic endurance and delayed VO 2 recovery, reinforcing the rationale for prehabilitation and suggesting that extended recovery periods during exercise training may be beneficial.
Extended-criteria donation after circulatory death (ECD-DCD) livers require functional assessment to ensure safe utilization. Sequential (dual) hypothermic oxygenated machine perfusion ((D)HOPE) with controlled oxygenate...Extended-criteria donation after circulatory death (ECD-DCD) livers require functional assessment to ensure safe utilization. Sequential (dual) hypothermic oxygenated machine perfusion ((D)HOPE) with controlled oxygenated rewarming (COR) followed by normothermic machine perfusion (NMP) enables viability assessment of ECD-DCD grafts while promoting mitochondrial recovery. Previous work has investigated the added value of flavin mononucleotide (FMN) in risk stratification during DHOPE. This diagnostic accuracy study aimed to compare the predictive value of FMN release during DHOPE with bile quality assessment during subsequent COR-NMP in consecutively accepted ECD-DCD livers. FMN concentrations were measured fluorometrically in perfusate samples collected during DHOPE and not used to guide clinical decisions on transplantability. Viability assessment after 2.5 hours of NMP served as the reference standard. Associations between FMN levels and graft utilization were evaluated using receiver operating characteristic analyses. A total of 103 ECD-DCD livers were included, of which 67% were transplanted. FMN demonstrated discriminative ability for graft nonutilization based on bile quality assessment, with AUROC of 0.74 (95% CI: 0.60-0.86) at 30 minutes, 0.75 (95% CI: 0.63-0.85) at 60 minutes, and 0.70 (95% CI: 0.56-0.83) at 120 minutes of DHOPE. Exploratory FMN thresholds showed increasing specificity and positive predictive value over time, reaching 0.92 (95% CI: 0.83-0.98) and 0.75 (95% CI: 0.53-0.94), respectively, at 120 minutes. Transplanted livers had significantly lower FMN concentrations than declined livers at all time points, with Youden-optimal cutoffs of 9.32, 20.97, and 45.53 ng/mL at 30, 60, and 120 minutes, respectively. FMN release during DHOPE is associated with subsequent NMP viability outcomes in ECD-DCD livers. FMN concentrations may help identify viable and nonviable grafts early during DHOPE, potentially avoiding resource-intensive comprehensive viability assessment with COR-NMP.
Predicting short-term liver dysfunction after locoregional therapy (LRT) for hepatocellular carcinoma (HCC) remains challenging, and the relative prognostic value of liver function, tumor diameter, and patient characteri...Predicting short-term liver dysfunction after locoregional therapy (LRT) for hepatocellular carcinoma (HCC) remains challenging, and the relative prognostic value of liver function, tumor diameter, and patient characteristics is not well defined. Improved risk stratification may inform patient selection and peri-procedural decision-making. We performed a retrospective cohort study of patients with HCC undergoing LRT in the Veterans Health Administration. Clinical, laboratory, and tumor-related variables were evaluated for their ability to predict 30-day and 90-day outcomes of liver dysfunction. Model discrimination was assessed in the test dataset using the area under the receiver operating characteristic curve (AUC) with 95% confidence intervals, and calibration was evaluated using calibration plots and Brier scores. Among 1183 patients with early- to intermediate-stage HCC, liver function-based scores consistently showed the strongest predictive performance. For 30-day outcomes, the Model for End-stage Liver Disease (MELD) score demonstrated the highest discrimination, followed by MELD-Na and serum bilirubin. Similar patterns were observed for 90-day outcomes. In contrast, total tumor diameter measures and demographic variables exhibited limited discriminatory ability. Overall model calibration was acceptable across risk strata. In patients undergoing LRT for HCC, measures of liver function outperform tumor-related and demographic variables in predicting liver failure. These findings underscore the central role of hepatic reserve in peri-procedural risk assessment and support prioritizing liver severity metrics when evaluating candidates for LRT.
Cardiovascular complications are a leading cause of morbidity and mortality after liver transplantation. Major Adverse Cardiovascular Events (MACE) are increasingly used as composite endpoints in both clinical trials and...Cardiovascular complications are a leading cause of morbidity and mortality after liver transplantation. Major Adverse Cardiovascular Events (MACE) are increasingly used as composite endpoints in both clinical trials and observational studies; however, definitions are highly variable, limiting comparability. Liver transplant recipients present unique pathophysiological features and perioperative risks that are not adequately captured by standard cardiovascular frameworks. We conducted a systematic review of 49 studies reporting MACE after liver transplantation, including randomized trials, prospective and retrospective cohort studies, and registry-based analyses. Data on study design, follow-up, MACE definitions, and the components included in composite endpoints were extracted and analyzed descriptively. None of the studies applied the classical three-point MACE definition (non-fatal myocardial infarction, non-fatal stroke, and cardiovascular death). Components included in composite endpoints varied widely: myocardial infarction and heart failure were reported in 78% of studies, stroke in 69%, arrhythmias in 47%, and cardiac death in 35%. Pulmonary embolism was inconsistently included. Heart failure and arrhythmias accounted for a substantial proportion of events, whereas ischemic events represented a minority. Based on event frequency, pathophysiology, and prognostic relevance, we propose that post-transplant MACE include coronary artery disease (defined as myocardial infarction, unstable angina requiring hospitalization, or coronary revascularization, including elective procedures), stroke, heart failure requiring hospitalization, clinically significant arrhythmias (including perioperative events within 30 days of transplantation), cardiac arrest, and cardiovascular death. All non-fatal components should require cardiovascular hospitalization. Pulmonary embolism should be excluded. This definition represents the authors' proposal and requires further consensus and formal validation. A structured international process, ideally using Delphi methodology and a position paper endorsed by transplant societies, could provide a reference framework for future studies, improving comparability, risk stratification, and clinical management of liver transplant recipients.
Waitlist mortality is a key quality metric in liver transplantation, reflecting the interplay of disease severity, organ availability, and clinical management at both system and center levels. A clear understanding of ho...Waitlist mortality is a key quality metric in liver transplantation, reflecting the interplay of disease severity, organ availability, and clinical management at both system and center levels. A clear understanding of how waitlist mortality is defined, measured, and interpreted is essential. In this perspective, we review the current definitions of waitlist mortality and highlight important trends and predictors influencing waitlist mortality among liver transplant candidates in the United States. We discuss the implications of using waitlist mortality in center performance monitoring and review potential system-level and center-level strategies to mitigate waitlist mortality, including allocation policy refinements, candidate selection practices, donor utilization, and clinical management while awaiting transplant. Finally, we explore future directions, including improved risk prediction, integration of waitlist mortality into continuous distribution, and the importance of equity-focused policy.
Building on models developed for recipients of deceased donor liver transplant, a modified early allograft dysfunction score of bilirubin>10 or INR>1.6 at postoperative day 7 has been applied to patients undergoing livin...Building on models developed for recipients of deceased donor liver transplant, a modified early allograft dysfunction score of bilirubin>10 or INR>1.6 at postoperative day 7 has been applied to patients undergoing living donor liver transplant. However, this score has never been validated in the living donor liver transplant population or separately analyzed in patients receiving left lobe (LL) grafts. A multicenter cohort of patients (Adult-to-Adult Living Donor Liver Transplantation Cohort Study [A2ALL]) and a single-center cohort University of California, San Francisco (UCSF) were combined. Kaplan-Meier estimates were computed for graft loss (GL), and Cox proportional hazards models were used to evaluate lab values from 1-week post-transplant. Separate multivariable risk scores for GL in recipients of right lobe (RL) and LL were created and bootstrap-validated, and their risk associations compared to modified early allograft dysfunction. In total, 724 and 156 patients with RL and LL grafts, respectively, were included in this analysis. GL occurred in 7.5% of patients with RL and 7.8% patients with LL. Recipients of RL who scored above the RL score cut-point and recipients of LL who scored above the LL score cut-point had a significantly higher risk of GL (RL: HR 6.00, 95% CI: 3.15-11.4, p <0.001; LL: HR 12.6, 95% CI: 3.80-42.0, p <0.001). The modified early allograft dysfunction score was only significantly associated with GL in recipients of RL (HR: 4.04, 95% CI: 2.40-6.80, p <0.001) but not in recipients of LL (HR: 1.98, 95% CI: 0.63-6.23, p =0.24). We developed and validated separate risk scores for recipients of RL and LL, which may allow for earlier and more accurate identification of patients at high risk of GL and early intervention to protect the graft.