In the early 1980s, organ allocation in the United States was initially based on anecdotal experience, self-interest, and single-center opinions with little in the way of scientific evidence, mathematical survival modeling, or validation to support these patterns of organ allocation. As liver transplantation (LT) became more successful, the disparity between the number of patients on the waiting list and available donor organs became an issue as a result of increasing wait-list deaths, and thus, a more justifiable donor organ allocation scheme became necessary. An initial attempt at establishing organ allocation policy came in 1987 when the U.S. government established the Organ Procurement and Transplantation Network as part of the Transplantation Act. The Organ Procurement Transplantation Network operates under federal contract with the United Network for Organ Sharing (UNOS). Since the formation of the Organ Procurement Transplantation Network, attempts to improve and standardize organ allocation have been ongoing and evolving.
Liver allocation policy was initially based on patients' location of care. Patients requiring continuous intensive care, including patients with acute esophageal variceal bleeding not responding to endoscopic therapy; patients who developed hepatorenal syndrome; and patients with intractable ascites or intubated with stage 4 portosystemic encephalopathy received first priority. Organ allocation was prioritized next to patients requiring continuous hospitalization, and finally to patients who were cared for at home.1 However, as the waiting list continued to grow, waiting time became a major factor in determining who received a donor organ.
Before February 2002, liver allocation was prioritized according to 4 UNOS-defined categories: status 1, 2A, 2B, and 3. These categories were based on whether the patient required admission to an intensive care unit and on the patient's Child-Turcotte-Pugh score.2 One problem with this scheme was that there were no established criteria for defining which patients truly required intensive care unit admission, thus allowing less ill patients to remain in the intensive care unit and gain an advantage for liver organ allocation. Furthermore, there were large numbers of patients in each of the 4 UNOS listing strata, and deaths on the waiting list continued to increase. Consequently, waiting time became the tiebreaker and the ultimate major determinant of organ allocation in the United States. However, as demonstrated by Freeman and others,3, 4 waiting list mortality did not correlate with waiting time. As a result of these disparities, the Department of Health and Human Services issued its Final Rule mandate in 1998 stating that donor livers should be allocated according to medical urgency and that a more continuous system should be used as mentioned by the Institute of Medicine.5, 6 In response to the Department of Health and Human Services mandate, UNOS formed the Liver Allocation Committee, and in February 2002, the Model for End-Stage Liver Disease (MELD) was adapted for the allocation of donor livers in the United States.
The MELD score was initially developed to predict mortality in patients receiving transjugular intrahepatic portosystemic shunts to treat variceal bleeding or resistant ascites. The initial model, called the Mayo Model for End-Stage Liver Disease, consisted of 3 objective variables: serum creatinine, serum total bilirubin, international normalized ratio, and a fourth variable based on liver disease etiology.7 During further assessment, the name was changed to the MELD scoring system, and investigators found that etiology contributed minimally to predicting short-term survival. Thus, the final MELD model used for liver allocation policy is based on serum creatinine, serum total bilirubin, and international normalized ratio. The MELD score has been retrospectively and prospectively shown to be highly predictive of short-term mortality in patients with all causes of end-stage liver disease who are awaiting LT.8 The model has been validated for prediction of 3-month and 1-year mortality (and inversely survival) in a broad spectrum of patients with chronic liver disease.9
The advantages of the MELD system for organ allocation are that all variables are objective and statistically weighted, and the model has a continuous scale with no ceiling or floor effects, thereby reducing a large number of ties (multiple patients with equal waiting times) and virtually eliminating dependence on waiting time for ranking candidates except as a tiebreaker for patients with equal MELD scores.
Since the implementation of MELD system, assessment of the UNOS liver allocation policy has revealed marked changes in the dynamics of organ allocation.10 The mean MELD score at transplantation increased from 17 in the pre-MELD era to 22 in the post-MELD era. Despite a shift to sicker patients receiving transplants, there have been no differences in 1-year patient and graft survival rates since implementation of the MELD system. And there has been a reduction in median waiting time from 656 days to 416 days. The major reason for the reduction in waiting time has been the removal of waiting as a criterion for liver offer. Thus, the most ill patients receive an offer first, regardless of how long they have waited. This also removed the incentive to “list patients early” to gain waiting time at a time when their liver disease was not very severe. Perhaps the most important indicator of the superiority of the MELD system over the previous allocation system was a reduction in waiting list mortality by 3.5% after its implementation. These changes clearly met the requirements of the Department of Health and Human Services, which state that organs should be allocated on the basis of medical urgency rather than waiting time.
However, at the time the MELD liver allocation system was implemented, policy makers recognized that not all LT candidates benefited from LT because they faced a high risk of dying from their intrinsic liver disease. These patients would not have their need for LT accurately characterized by their calculated MELD or Pediatric End-Stage Liver Disease (PELD) score.11
Hepatocellular carcinoma (HCC) is one such example. Rule makers defined the need for LT for these candidates as the risk of progressing beyond tumors meeting the so-called Milan criteria, a stage at which excellent posttransplant results could be achieved.12 At the initial time of MELD and PELD policy implementation, there were no good data quantifying the risk of HCC progression, so estimates of probability were arbitrarily assigned and equated to the MELD-defined probability of death, resulting in awarding of additional MELD points to the calculated MELD scores.11 Thus, with additional points, the initial assessment of the MELD allocation system demonstrated that there was a reduction from 24 to 7% in patients with HCC falling off the waiting list.13 Subsequently, many studies have examined the progression of HCC in waiting LT candidates,14 and the HCC prioritization policy has been revised several times.
There are patients with other conditions for which the need for LT is not accurately defined by the MELD score because their prognoses depend on factors other than liver disease mortality risk.15 As such, these patients have potential for being underserved (wait list removals or death) if they are ranked for allocation of deceased donor livers solely on the basis of a calculated MELD score derived from their pretransplant laboratory values. Some of these conditions, in addition to HCC, such as the progressive pulmonary compromise seen in hepatopulmonary syndrome, familial amyloid polyneuropathy, metabolic liver diseases such as urea cycle defects or hereditary oxaluria, have been identified and termed the so called “exceptional diagnoses.”16 In addition, because the MELD/PELD score accurately predicts death from liver disease in approximately 80 to 85% of patients, as many as 15 to 20% of patients with chronic liver disease may not be accurately prioritized by their MELD/PELD score.
It was recognized that it is unlikely that any scoring system would serve all potential LT candidates equally well, and for these reasons, the original MELD allocation policy included a mechanism by which centers could request increased priority for any patient for whom the MELD/PELD score was thought to inaccurately estimate their need for LT. The policy stipulated that these requests would be reviewed by a regional peer review system to determine the appropriateness of the requested increase in priority on the basis of medical evidence from the literature or expert opinion. The goals behind allowing certain disease entities to receive increased priority on the waiting list as MELD exceptions are severalfold: (1) decrease patient risk of death on the waiting list; (2) increase timeliness of LT before the underlying liver disease progresses to the point that the patient can no longer be considered a LT candidate; (3) decrease the risk of disease of recurrence after LT; (4) prevent disease progression that is dependent on a metabolic or genetic disease that may preclude LT, and (5) improve survival after LT. With the exception of HCC, however, MELD-equated priority guidelines were not identified for these other exceptional diseases.
After the first year of use of the MELD allocation policy, Rodriguez-Luna and collagues17 assessed Regional Review Board practices. They found that large variations existed throughout the country in how exceptional cases were handled. Because these differences exist and there are no guidelines available for the consistent assessment of exceptional case applications other than for HCC, UNOS convened a study group, MELD Exceptional Case Study Group (MESSAGE), a subcommittee of the UNOS Liver and Intestinal Committee, to develop a consensus and to advance the field and provide written recommendations to UNOS, the Liver and Intestinal Committee within UNOS, and guidelines to the Regional Review Boards to aid in the assignment or denial of assignment of MELD score upgrades. Thus, the goals of MESSAGE and the conference were to make evidence-based recommendations when possible and create a uniform approach to a number of conditions that were believed to be underserved when the calculated MELD score was used.
The MESSAGE group, with the assistance of the Scientific Registry of Transplant recipients and UNOS, identified 17 diagnoses or groups of diagnoses for which MELD exceptions were requested for Regional Review Board deliberation. The members of this committee reviewed the literature for evidence or expert opinion to support or refute the validity of waiting list death, waiting list removals for progressive disease, waiting list removals as too ill, and evidence of nonhepatic end-organ injury that would obviate LT for patients with these diagnoses and determine if these patients with specific diagnoses should receive increased waiting-list priority. These findings were presented at a national conference held in Chicago, March 1 to 2, 2006, where final recommendations were reached. The MESSAGE committee members and the members of the Liver and Intestinal Committee who were involved in this project emphasize that the Regional Review Boards across the country have held widely divergent opinions on which conditions should receive additional MELD points, if any, and on how much priority should be given for exceptional case requests. The purpose of the MESSAGE group's work was to provide a consistent, evidence-based approach for listing patients with additional MELD points across the liver transplant programs and regions within the UNOS. This series of articles summarize MESSAGE's deliberations and conclusions reached at this conference.