Global liver transplant study finds recipient health and hospital volume outweigh donor factors

Liver transplantation has long been shaped by a stubborn assumption: the quality of the donated organ largely determines the fate of the person who receives it. A sweeping international study now challenges that view. The IMPROVEMENT collaborative, the largest balanced international analysis of liver transplantation to date, followed 5218 patients treated at 61 centres across Europe, Asia, Oceania, North America and South America, and found that donor factors had only a minor impact on survival once modern preservation and matching practices were accounted for. Instead, the health of the recipient, the speed of the transplant logistics and the volume of the transplanting hospital emerged as the dominant predictors of who survived.

The study, published in The Lancet Regional Health – Europe, was designed to capture a genuinely global picture of transplant practice. Centres were enrolled into a retrospective cohort of 3997 patients transplanted between January 2017 and December 2019, before the COVID-19 pandemic, and a prospective cohort of 1221 patients treated between July 2022 and December 2023, after the worst of the pandemic had passed. This split was deliberate: the researchers wanted to avoid the confounding effects of the pandemic, which had disrupted donor-recipient matching, transplant logistics and post-operative care. Each centre was required to contribute a fixed number of consecutive cases, a measure intended to reduce the centre-volume bias that arises when hospitals competitively enrol only their best-performing patients. The number of centres per region was set to mirror the distribution reported by the Global Transplant Observatory, and patient distribution was balanced between high-volume centres performing more than 65 transplants per year and intermediate-volume centres performing 25 to 65 per year.

The cohort reflected the full spectrum of modern transplant activity. Overall, 4668 patients, or 89.5 percent, received a graft from a deceased donor, either after brain death or after circulatory death, while 550 patients, or 10.5 percent, received a graft from a living donor. A total of 3365 patients, or 64.5 percent, were transplanted at high-volume centres and 1853, or 35.5 percent, at intermediate-volume centres. Patients had to be at least 18 years old and undergoing a first transplant; combined grafts, domino grafts, heterotopic grafts, double grafts, cholangiocarcinomas, liver metastases and advanced hepatocellular carcinomas were excluded. Data were collected through a dedicated REDCap platform and processed in compliance with the European Union’s General Data Protection Regulation, with approval from the coordinating centre’s institutional review board in Rome and from local boards.

To make sense of such a heterogeneous dataset, the researchers adopted an unusual four-step analytical process. First, they generated colourimetric world maps to visualise the parameters with the greatest heterogeneity and the strongest discriminatory potential for outcome variation. Second, they built radar graphs summarising sixteen parameters grouped into three domains: donor characteristics, recipient characteristics and risk-mitigation strategies. Third, they plotted countries on a Cartesian diagram of donor versus recipient risk-propensity, which they called aggressiveness, and then drew a histogram of country-specific mitigation strategy against that aggressiveness, a measure they termed coverage adequacy. Finally, they developed and validated prognostic models for mortality and complication rates using generalised linear mixed-effects models, with country as the cluster level, and cross-validated the models among regions.

The maps revealed striking geographic contrasts. Median MELD 3.0 scores, a measure of liver disease severity, ranged from 24 in the United States to 9 in Vietnam, New Zealand and the Netherlands. Hepatocellular carcinoma as the indication for transplant varied from 66 percent of cases in Vietnam to 10 percent in Poland. Metabolic dysfunction-associated steatotic liver disease, an increasingly common cause of cirrhosis, exceeded 40 percent of cases in Brazil and Argentina but fell below 5 percent in Russia, Poland, Vietnam and China. Living-donor transplantation accounted for more than 90 percent of transplants in Asia but was nearly absent in many European countries, China, Australia and South America, with low prevalence also in the United States at 8 percent and in individual European countries such as the Netherlands at 12 percent, Belgium at 5 percent and Germany at 4 percent. The median volume of packed red blood cells transfused during surgery, used as a surrogate for surgical complexity, exceeded 1200 millilitres in Japan, the United States, Germany, China and Australia but fell below 600 millilitres in France, Vietnam, Sweden and New Zealand.

The radar graphs added texture to these contrasts. In Europe, donor risk was mainly captured through extended criteria donors and donation after circulatory death, while recipient risk aligned with the global reference line and liver-specific risk remained below it, particularly in the 2022 to 2023 period. In Asia, risk was largely attributable to living-donor transplantation, with a strong emphasis on minimising cold ischaemia time, the interval during which the graft has no blood supply. North America exhibited the highest recipient risk, especially in the later cohort, with a high prevalence of steatotic liver disease, non-cancerous indications, ventilatory life support before surgery, high MELD 3.0 scores and large transfusion volumes. In Oceania, perceived donor and recipient risks were balanced, whereas in South America donor-risk exposure was low and recipient risk was sustained by high rates of steatotic liver disease, non-cancerous cases, ventilatory support, complex portal vein thrombosis and high MELD 3.0 scores.

With a median follow-up of 56 months, overall survival was 93 percent at 90 days and 90 percent at 365 days, declining to 80 percent at five years. Survival differed significantly between living-donor transplants at 86 percent and deceased-donor transplants at 80 percent over five years, and also varied according to MELD class, liver disease aetiology, cancer status, pre-transplant ventilatory support and the presence of complex portal vein thrombosis. At the univariate level, patient survival differed among the enroled centres, with centres from Oceania showing the highest survival, followed by North America, Asia, Europe and South America, and differences were stratified according to centre volume.

The multivariable models then distilled which factors independently predicted death. Nine parameters emerged. Only one was a donor parameter: cold ischaemia time of 360 minutes or longer, which predicted mortality at 90, 365 and 730 days, though the authors note it reflects logistical organisation as much as intrinsic organ quality. Seven were recipient parameters: recipient age, coronary artery disease score, pre-operative ventilatory life support, MELD 3.0 of 30 or higher, complex portal vein thrombosis, metabolic dysfunction-associated steatotic liver disease and the volume of packed red blood cells transfused. The ninth was logistical: transplantation at a high-volume centre, which predicted better survival at all three time points. The models achieved C-statistics of 0.78 at 90 days, 0.74 at 365 days and 0.73 at 730 days, with similar values in the cross-validation set. For complications, significant predictors were pre-operative ventilatory support, complex portal vein thrombosis, steatotic liver disease, pre-transplant TIPS, transfusion volume and prolonged cold ischaemia, while transplantation at a high-volume centre was protective.

The study’s central message is that donor quality is no longer the major limiting factor it was once thought to be. The authors attribute this in part to the spread of machine perfusion technologies, including ex-situ hypothermic oxygenated perfusion, normothermic regional perfusion, normothermic machine perfusion and ischaemia-free techniques, which have enabled the safe use of organs previously deemed unsuitable, including donors up to 100 years old. However, they were unable to establish a protective role for machine perfusion in their own country-adjusted analysis because the number of events was insufficient. They also caution that the absence of significant intrinsic donor effects may reflect confounding by centre practices, selection bias or underpowering, and that recipient and system-level factors appear to explain a larger proportion of outcome variability without excluding an important role for donor characteristics. Comparing the pre- and post-pandemic periods, mortality decreased while acceptance of extended-criteria donors and machine perfusion use increased, though the researchers cannot demonstrate a cause-and-effect relationship and prefer to describe the post-pandemic era as a period of new policies, dedicated pathways and better infection control.

The findings carry practical weight for clinicians and policymakers alike. Concentrating cases in high-volume centres, shortening cold ischaemia time, implementing machine perfusion and performing pre-transplant transjugular intrahepatic portosystemic shunts may mitigate the impact of high-risk clinical profiles, the authors suggest. They also stress that the aggressiveness and coverage adequacy indices remain exploratory and await external validation, and that geographic differences should be interpreted as qualitative and improvement-generating rather than as evidence of causal or performance-related differences across countries. The study was funded by the Italian Health Ministry, which played no role in the statistical analysis or writing of the manuscript. The researchers hope their framework will support benchmarking across transplant systems, identification of transferable organisational strategies, and more equitable allocation of health resources, particularly in countries where mitigation strategies are not yet fully implemented.

Subject of Research: A multicentre cohort study of regional and national variation in donor and recipient characteristics, transplant practice and outcomes after liver transplantation.

Article Title: Regional and national variation in donor and recipient characteristics, transplant practice, and outcomes after liver transplantation (the IMPROVEMENT collaborative): a multicentre cohort study

Article References: Avolio, A. W., Pascale, M. M., Lai, Q., Burra, P., Cillo, U., Croome, K. P., De Santibanes, M., Egawa, H., Gastaca, M. M., Guo, Z., Martins, P. N., Oniscu, G. C., Polak, W. G., Quintini, C., Rela, M., Sapisochin, G., Wiederkehr, J. C., Baccarani, U., Balci, D., … Agopian, V. G. (2026). Regional and national variation in donor and recipient characteristics, transplant practice, and outcomes after liver transplantation (the IMPROVEMENT collaborative): a multicentre cohort study. The Lancet Regional Health – Europe, 71, Article 101889. https://doi.org/10.1016/j.lanepe.2026.101889

Image Credits: AI Generated

DOI: 10.1016/j.lanepe.2026.101889

Keywords: liver transplantation, IMPROVEMENT study, machine perfusion, MELD 3.0, centre volume, cold ischaemia time, living donor transplant, donor risk index, prognostic modelling, geographic disparities, Regional, national