The MELD system and liver transplant waiting-list mortality in developing countries : lessons learned from São Paulo , Brazil

The interpretation and reporting of these data are the sole responsibility of the authors and should in no way be seen as representing official policy of or interpretation by the São Paulo State Government. RESUMO Objetivo: Este estudo foi desenhado para avaliar os resultados da nova política de alocação em relação à mortalidade na lista de espera. Métodos: O banco de dados de transplante hepático do Estado de São Paulo foi revisado de forma retrospectiva, de julho de 2003 até julho de 2009. Os pacientes foram divididos naqueles transplantados antes (Grupo Pré-MELD) e depois (Grupo Pós-MELD) da implementação do sistema MELD (Model for End-stage Liver Disease). Foram incluídos apenas os candidatos adultos para transplante de fígado. O desfecho primário foi a mortalidade na lista de espera. Resultados: A taxa não ajustada de óbitos na lista de espera diminuiu significativamente após a implementação do sistema MELD (de 91,2 para 33,5/1.000 pacientes por ano; p<0,0001). A análise multivariada mostrou uma queda significativa no risco de morte na lista de espera para pacientes após o MELD (HR de 0,34; p<0,0001). Atualmente, 48% dos pacientes são transplantados no primeiro ano na lista (versus 23% na era pré-MELD; p<0,0001). A sobrevida dos pacientes e do enxerto não mudou com a implementação do MELD. Conclusão: Houve redução no tempo de espera e na mortalidade na lista após implementação do sistema MELD em São Paulo. Os pacientes na lista no período pós-MELD apresentaram uma redução significativa no risco de mortalidade na lista de espera. Não houve mudanças nos resultados após o transplante. O MELD pode ser utilizado com sucesso para alocação para transplante fígado em países em desenvolvimento.


INTRODUCTION
Brazil is the largest country in Latin American, with a population surpassing 190 million, and its economy currently ranks among the top ten in the world (1,2) .The State of São Paulo has more than 40 million habitants and accounts for one third of the country Gross Domestic Product (GDP) (1) .In 1997, the Brazilian government launched the Transplant National System (SNT, acronym in Portuguese), which divided the country into different regions and set allocation rules.Initially, allocation was patient-based and priority was given by time on the waiting list (3) .Single center reports found an initial mortality on the waiting list for liver transplantation ranging between 20 to 38% (4,5) .
When faced with high mortality of liver transplant candidates on the list, Americans adopted the MELD (Model for End-stage Liver Disease) system to proper allocate grafts for transplantation.This change of liver allocation policy was linked to decreased time to transplant and reduction in waiting list mortality, with no concomitant changes in post-transplant outcomes (6) .Some European groups also adopted the MELD system for liver allocation, such as the North Italian Transplant, in 2003; the Eurotransplant, in 2006; the Etablissement Français des Greffes, and the Swiss Transplant, in 2007.A recent European report described similar positive waiting list findings with the utilization of the MELD system (7) .
Despite significant advances in the Brazilian transplant arena, concerns regarding post-transplant outcomes were evident prior to MELD implementation.Local specialists were divided among those who were worried about high mortality on the waiting list versus those who did not believe that the healthcare system would support the perioperative care of the sickest patients (4,(8)(9)(10)(11) .
To better understand the local landscape, it is important to first appreciate that a national public healthcare system provides free access to unrestricted services for the entire transplant population.Public healthcare is also the only coverage for a large segment of the population (approximately 130 million habitants).Second, there is an enormous regional variation in GDP per capita, with multiple areas experiencing high poverty level and substandard quality of living.Third, the donor pool is extremely limited (ranging from 6.3 donors per million, in 2005, to 8.7 donors per million, in 2009) (12) .Finally, there are substantial inequities in access to and quality of healthcare services, with a quarter of the country not having a liver transplant center in close proximity (8) .Therefore, a liver transplant allocation system driven by MELD score had not yet been attempted under those circumstances.After an intense and long debate, the Brazilian liver allocation was modified.Allocation is now patient based and priority was shifted to a MELD/Pediatric End-Stage Liver Disease (PELD) based system on July 16, 2006.A landmark publication has highlighted the successful implementation of the PELD system (9) .However, the overall impact of the MELD allocation in the waiting list mortality and in post-transplant outcomes of the adult population has not been yet fully exploited and could be potentially different than early reports from other developed countries (7,13,14) .

OBJECTIVE
To evaluate the impact of the MELD system on the waiting list mortality in the State of São Paulo, Brazil.

Protocol, design, data sources, inclusion criteria and groups
A retrospective cohort study was initially conducted including data from all liver transplant candidates waitlisted from July 15, 2003 through July 16, 2009 in the State of São Paulo.Data were drawn from the liver transplant database of research files from the State Health Secretariat of São Paulo.There is no national database for liver transplant in Brazil.
In the present study, inclusion was restricted to adult patients (≥18 years) who were candidates for a liver transplantation.Patients with liver-kidney transplants were included while all other multiple organ transplants were excluded.
Candidates were then divided into a Pre-MELD Group for those listed from July 15, 2003 to July 15, 2006 and into a Post-MELD Group, for those listed from July 16, 2006 to July 16, 2009.For the analysis of post-transplant outcomes, the recipients were divided into the same groups based on the date of transplant (instead of date of listing).
Patients with hepatocellular carcinoma (HCC) according to the Milan criteria were granted additional MELD points.On the first year after implementation of MELD these patients received arbitrarily a MELD of 29.This was subsequently changed to a MELD of 24 on the second year after implementation.Salvalaggio P, Afonso RC, Pereira LA, Ferraz-Neto BH

MELD and donor risk index
The biological MELD at time of the transplant (or the last score available) was calculated as previously described (15) .Donor risk index (DRI) was computed according to the formula defined by Feng et al. (16) .Donation after cardiac death (DCD) is not present in this series.Race of donors is not reported in the database since there is much variety in the country (8) .To calculate the DRI, DCD scores were set to zero and input race scores to 0.15 (the average between minimum and maximum allowed scores).

Clinical outcomes and covariate definitions
The main endpoint was death or removal on the waiting list.Post-transplantation survival was a secondary endpoint.Laboratory data and MELD score were not collected in the pre-MELD era.For those who receive a transplant after implementation of the MELD system, the MELD score at transplantation was used as a marker of liver disease severity.
Covariates included gender, age, race, ethnicity, blood type, cause of liver failure (viral hepatitis, HCC, and other causes), region, retransplantation, split grafts, deceased donor recipient, kidney co-transplantation, DRI and body mass index (BMI).The definitions of allograft loss and patient death used were the same found in the Organ Procurement and Transplant Network (OPTN) registry.

Analysis of risk factors for waiting list death
Demographic data of patients and recipients and transplant information were submitted to univariate analysis to identify risk factors for waiting list death or removal from the list.For the risk factors identified with p<0.2 in the univariate analysis, a Cox analysis was performed to test whether different grades contributed to waiting list death or removal from the list.

Statistical analysis
Rates of removals for death/too sick and transplantation were calculated based on the number of patients and their length of time on the list and expressed as cases per 1,000 patient-years.Comparisons between rates for demographic, clinical, and geographic strata for the two eras were performed using the normal approximation to the binomial distribution.A χ 2 analysis was used to examine qualitative variables and t tests to study quantitative variables.Kaplan-Meier curves were drawn depicting the post-transplant patient and graft survival differences of patients for each group.The log-rank test was used to determine if there was a significant difference in the curves.Missing data (<5% of all variables) on the characteristics examined was categorized as "other" or "unknown" or excluded from analysis (in most circumstances), depending on the frequency of missing data for a given characteristic.No imputation technique was used.An alpha level of 0.05 was used for all significance tests.Analyses were performed using SAS v.9.2 (SAS Institute, Cary, NC).

Clinical characteristics of the study cohort
The demographics of transplant recipients are depicted on table 1.Except for a large rate of listing for HCC in the post-MELD era, there were no major differences regarding the demographics of the patients over time.Although MELD scores and laboratory tests were not recorded prior to implementation of the MELD system, in 2007, it is evident that the average MELD score has changed over time.The demographics of waitlisted transplant candidates were also assessed and the findings were similar to data presented on table 1. Considering the demographics of donors, there was a significant shift towards an older population that is dying of stroke.Moreover, a surge of split and liverkidney transplants was observed, as well as a reduction in cold ischemic time in the post-MELD era.

Waiting list dynamics and number of transplants
Since time on the list was no longer a requirement to receive a graft and a policy was established to remove patients with MELD <15 from the list, a clear reduction in the number of patients added to the list after 2007 was noted (from an average of 325 per quarter in the pre-MELD era to an average of 125 per quarter in the post MELD era; p<0.05).At the same time, the number of deceased donor liver transplants has more than doubled from 50 per month in the pre-MELD era to approximately 85 per month in the post-MELD period (p=0.02).The number of split and liver-kidney transplants also followed the same trend (Table 2).Deceased donor liver transplantation grew from 60% of the donor pool, in 2005, to 88.5%, in 2009.It also became apparent that the number of HCC patients who were transplanted increased more than 8-fold (from 2% in the pre-MELD era to 17% in the post-MELD period, p<0.0001).
As a result of implementing the MELD system and of the rise in donors, mortality in the waiting listed plunged (Table 3).This benefit was observed for all genders, races, blood types, age groups and was not related to diagnosis or waiting list time.Furthermore, we observed a reduction in time to be removed from the list was found (in the post-MELD era, 48% of patients were removed of the list in less than 1 year, Figure 1), a decrease in time to transplant (from 887 to 434 days;

DISCUSSION
Earlier reports attested the efficiency of MELD-driven algorithms for liver transplant allocation in the United States and Europe (6,7) .Since liver transplantation grows around the globe, it might be also important to test efficiency of a MELD-driven allocation system in developing countries.Herein, we investigated the impact of a MELD allocation system on wait-listed liver transplant candidates in Brazil.
The main finding of our study was validation of efficiency of the MELD allocation system in São Paulo, Brazil.We experienced a reduction in mortality on the waiting list, associated with a shorter time to receive a transplant after the introduction of the MELD system.Importantly, the Cox analysis pointed to switch in allocation system as the main driver of waiting list death or removal, attesting the efficiency of the MELD system.The major "motto" of the MELD system is reduction in mortality in the waiting list.This was fulfilled for the entire population, including the pediatric population and minorities (9) .
For the Brazilian liver transplant community, it is extremely important to have an analysis of the MELD p=0.003), and a drop in the number of patients removed from the list due to death or for being too sick to transplant (from 91.2 to 33.51 deaths/1,000 patients per year in the waiting list; p<0.05).
A univariate analysis showed that listing in the pre-MELD era and having a diagnosis of HCC were significant risk factors for waiting list death.However, the multivariate analysis showed that only the listing period posed a significant risk factor for mortality on the liver transplant waiting list (Table 4).In the pre-MELD period, 1-year and 3-year patient survival rates were 70.1 and 64.8%, respectively.In the post-MELD phase, 1-year and 3-year patient survival rates were 68.4 and 64.4%, respectively (Figure 2A; p>0.1).

Post-transplant outcomes
In the pre-MELD era, 1-year and 3-year graft survival rates were 65 and 60%, respectively; whereas in the post-MELD period, 1-year and 3-year graft survival rates were 71,9% and 60%, respectively (Figure 2B; p>0.1).The MELD system and liver transplant waiting-list einstein.2012;10(3):278-85 impact and this manuscript tried to fulfill this gap.Moreover, we would like to set a model, to encourage other countries in Latin America, Africa, Middle East and Asia, which are trying to organize and regulate liver transplantation, and to select an appropriate allocation method for their particular healthcare system.As previously mentioned, poverty, lack of resources, difficult access to healthcare and technology and low donation rates are common features of other developing countries.Future reports from other regions of the globe will be important contributions for shaping transplantation and liver transplant allocation across the world.
There are other important contributions of the MELD implementation in our State that could be beneficial in the developing world.First, we have switched the registration and data report of patients listed and transplanted to an on-line system, which facilitated tremendously the communication among members of the transplant community and added significant transparency to the system.Second, the Review Board that was initially available at local level has now been expanded to a national level.The Boards are responsible not only for judging exception cases, but also for refining the rules of allocation (such as how many points to grant to HCC cases).Third, appropriate management of listed patients is now much more feasible for all transplant teams.Now there is a combination that enables listing only patients with MELD >15 with automatic removal of those with laboratory tests older than 90 days.A shorter list, associated with an expansion of the donor pool and an increase in the number of transplants contributed to improve future outcomes (17) .
A systematic advantage in the current MELD system for HCC recipients has been extensively reported in the literature.In a recent manuscript, Washburn et al. examined how much the MELD system benefited HCC patients (18) .In our study HCC recipients were significantly assisted by the MELD system.There was a 8-fold increase in the number of transplants in HCC patients in the post-MELD era.Although all subpopulations were found to experience drop in the mortality list, our analysis was unable to specifically verify whether the benefit for HCC recipients was deleterious to non-HCC patients.Granting additional points to HCC recipients evolved over time, in Brazil.First a maximum MELD score of 29 was granted to these recipients.As the average MELD score went down, more evidence was found in the international literature about "the MELD inflation" and HCC granting points are now capped at 24 (starting at 20) (19) .Future studies should continue to address this important point in order to be fair to the system with no increment in dropout rates of the waiting list in HCC patients (20) .
Multiple studies also pointed to increased liverkidney transplantation in the MELD era and the association of renal failure with increased resource utilization and poor outcomes (21)(22)(23)(24) .Therefore, it was not surprising to find a growing number of liver-kidney recipients in the post-MELD period in our study.We were pleased to also discover a substantial growth in the split-livers that were probably related to the aggressive PELD policy in place (9) .
The Brazilian experience mirrors in some aspects the North American reports, with a significant reduction in waiting list mortality, without changes in overall post-transplant survival (6,7) .Nonetheless, in the post-MELD era, it seems that severity of liver disease at transplantation (as measured by MELD score) has a direct impact on post-transplant survival (data not shown).Thus, this early analysis (not adjusted) has probably indicated some similarities with German and other transplant center reports that experienced a worsening in post-transplant outcomes according to the MELD score of recipients (25)(26)(27) .
It is also paramount for the Brazilian liver transplant community to move forward by addressing critical issues that have not yet been revised, starting by a focus on quality.Post-transplant survival rates similar to those from American and European transplant centers are already achieved in few national transplant centers, but this reality is not yet widespread in the whole country.Better training of teams, efforts to better select recipients and donors, associated with major investment in hospital infrastructure can certainly make a difference.Second, transparency in published data is relevant.Previous experience in multiple medical fields proved that when data are accessible to the public and are transparent, the overall quality of medical care and its outcomes improve (27)(28)(29) .Third, it is necessary to spread the model employed in the State of São Paulo (and in few other Brazilian States) to the entire country.
The present study has several limitations.First, those inherent to retrospective studies.Second, our method of comparing two different allocation systems in different time points might be biased.Third, we have arbitrarily selected the size of the groups, the time frame and follow-up of this study according to Salvalaggio P, Afonso RC, Pereira LA, Ferraz-Neto BH estimated statistical power.With adjustments for exceptions, changes in the population demographics and modifications in the characteristics of transplant centers, it might be important to continue to monitor efficiency of the MELD system in the future.Fourth, we could not acquire transplant center identifiers and correct the analysis for clustering.Finally, further assessment of post-transplant survival rates corrected for co-morbidities, disease severity and other factors that can change outcomes.

CONCLUSION
There was a decrease in the waiting time and in mortality list after implementing the MELD system in Brazil.Patients listed in the post-MELD era had a significant reduction in death risk in the waiting list.Despite changes in donor demographics, there were no modifications in post-transplant outcomes.MELD can be successfully utilized for liver transplant allocation in developing countries.

Figure 1 .
Figure 1.Average time to be removed from the list

Figure 2 .
Figure 2. Patient (A) and Graft (B) survival of liver transplant recipients in São Paulo, Brazil

Figure 2
Figure 2 depicts post-transplant outcomes during the study period.Graft and patient survival did not change after implementing the new allocation system.In the pre-MELD period, 1-year and 3-year patient survival rates were 70.1 and 64.8%, respectively.In the post-MELD phase, 1-year and 3-year patient survival rates were 68.4 and 64.4%, respectively (Figure2A; p>0.1).In the pre-MELD era, 1-year and 3-year graft survival rates were 65 and 60%, respectively; whereas in the post-MELD period, 1-year and 3-year graft survival rates were 71,9% and 60%, respectively (Figure2B; p>0.1).

Table 2 .
Donor demographics and transplant characteristics of liver transplants performed in São Paulo, Brazil DRI: donor risk index.

Table 3 .
Unadjusted death rate/1.000patients per year in the waiting list

Table 4 .
Univariate (A) and Cox analysis (B) of the likelihood of death or becoming too sick to transplant on the waiting list RR: 95%CI: 95%confidence interval.