The increasing number of patients dying while awaiting liver transplantation (1) has forced a critical examination of our allocation system for cadaveric liver donors. The previously established system, in which waiting time on the liver list had a strong influence on allocation, was studied by both the Institute of Medicine (2) (at the request of the Human Resources Service Administration (HRSA)) and researchers within the transplant community (3). All reached a similar and intuitively not surprising conclusion: death awaiting liver transplantation correlated with severity of illness, not with waiting time. Given this information, and in accordance with the goal of both HRSA and the transplant community to reduce deaths while waiting, impetus was given to develop a new system which allocated livers to patients ranked by severity of illness, with less emphasis on waiting time.
To meet this challenge the liver transplant community has spent the past two years developing and scientifically validating scores that quantitate end-stage liver disease. There was general agreement that the components of the score must be a few objective parameters proven to be predictive of death with chronic liver disease. From the outset it was also agreed that the most important predictors of death for children on the liver transplant waiting list would be different from those for adults. Although the pediatric score would be developed separately, the general principles and methods of statistical validation would be similar for the two scores. The results of these efforts, described in detail elsewhere, are the Pediatric End-stage Liver Disease (PELD) score and Medical End-stage Liver Disease (MELD) score (4,5). Each equation uses regression co-efficients derived from multi-variate analyses as multiplicative factors for the variables included. There are two important differences between MELD and PELD:
1. The components of MELD are bilirubin, INR, and creatinine, whereas for PELD they are bilirubin, INR (International Normalized Ratio), albumin, age under 1 year, and growth failure (defined as less than 2 standard deviations below the mean for height or weight).
2. The populations of patients with end-stage liver disease that were used for the multivariate analyses, and from which the scores were derived and tested, were different. MELD was developed originally in a population of adult patients after a TIPS procedure (6). In contrast, PELD was developed and tested in a large population of children listed for transplantation in the Studies of Pediatric Liver Transplantation (SPLIT) database (7) and was representative of a broad cross section of all children awaiting liver transplantation. MELD has subsequently been extensively tested in other populations more representative of the adult transplant waiting list, including the UNOS waiting list, and has been found to be a highly accurate predictor of pre-transplant death. PELD has also been validated both in the UNOS transplant waiting list and a large database of children awaiting liver transplantation at the University of Pittsburgh (8).
The open debate accompanying the intensive efforts to develop PELD and MELD, coupled with peer-reviewed publications, has engendered cautious confidence in the transplant community that MELD and PELD can quantify end-stage liver disease, and correctly rank patients by severity of illness on the liver transplant waiting list. Subsequently, as of February 27th 2002, MELD and PELD have been implemented as the new liver allocation system. Although the development and implementation of PELD and MELD are important achievements, now is not the time for complacency. Critical questions remain before us—if anything, our scrutiny and oversight of the effect of MELD and PELD on liver allocation practices and outcomes must increase.
The most critical question is what will be the effect on post-transplant outcome of a policy which prioritizes the sickest patients first for transplant? Will post-transplant survival decrease? Will re-transplant rates increase? Will we risk squandering a vital resource or alternatively exacerbate the donor shortage? How do we understand and study the strongest predictors of survival for patients with liver disease from the time of eligibility for transplant? This question requires analyzing and understanding the relative effects of pre-transplant factors, peri-transplant factors, donor factors, and post-transplant events. How do we balance the conundrum of justice vs. utility? Fairness would dictate that transplanting the sickest person first is morally and medically just, but the utilitarian objective would be to transplant those with the best chance of survival. Have we in fact pushed the pendulum too far toward the side of justice? None of these questions could be answered by statistical models before MELD and PELD were implemented. We have now set in motion, with the best of intentions, a system whose effects are as yet unknown. As we stand and watch the results, will PELD and MELD be the light at the end of the liver allocation tunnel, or is there a train approaching?
Of particular concern for the pediatric liver transplant community is how the application of PELD and MELD to an integrated waiting list of children and adults will affect the access of children to liver transplantation. After considerable debate, it was decided to use the actual calculated PELD and MELD scores to rank all patients irrespective of age on the list, even though the scores are derived from different equations. The important question then becomes: does the same PELD and MELD score predict the same probability of death on the waiting list? From analyses of PELD and MELD scores in different populations of patients with end-stage liver disease, the probability of death on the waiting list for any given MELD or PELD score can be calculated. As can be seen from Figure 1, for the same PELD or MELD score, the probability of death waiting is less for children in the SPLIT database, as compared with adults in either the TIPS population (Malinchoc data (6)), or a subset of the adult UNOS liver waiting list. Because the probability of death is calculated from the known baseline survival in a given population, these differences can be explained by the different baseline survivals in the populations compared. However, what we do not know is if the probability of transplant is the same for the same PELD or MELD score. Given that adults outnumber children on the waiting list by about 15 to 1, for any given MELD or PELD score, adults will always outnumber children. As well, waiting time will be the tiebreaker for patients eligible for the same organ with the same score. Many of the sicker children requiring liver transplantation have not been alive long enough to accumulate substantial waiting time. There is also the influence of an increasing number of adults with hepatocellular carcinoma (HCC). These patients will be assigned MELD scores, equivalent to a 20 to 40% probability of death within 3 months on the waiting list, with the added provision that the score will be upgraded every 3 months if the patient remains eligible. Before long, substantial numbers of adults with HCC may well rise to the upper quartiles of the MELD and PELD scoring systems and preferentially attract donors over other patients.
The fundamental importance of trying to address the alarming attrition of patients, both pediatric and adult, on the living waiting list is not in question. However, pediatric advocates in particular may question if it is appropriate to allocate organs based on a system which seeks to equalize the chances that a child will die awaiting transplant, with that of an adult. Is it an acceptable societal value that livers should be allocated based on the concept that a child's chance of dying waiting should be equivalent to that of an adult? Do we still believe (women and) children get the first places in the life-boat? Simply put, should the chance that a 60-year-old with hepatitis C will die before transplant, be the same as 2-year-old with biliary atresia? This example, although it may seem extreme, is actually representative of the problem. Adults, of whom 2,092 were transplanted at more than 50 years of age in the UNOS 2000 Annual Report (1), are predominantly transplanted for diseases with a higher probability of recurrence. Not only is their natural life span after transplant shorter, so may their graft survival be impacted by the recurrence of disease. In contrast, the great majority of children transplanted at less than 5 years of age, have diseases with no risk of recurrence. Should the 2-year-old, in the midst of the critical stages of growth and neuropsychologic development, which effect their life-long potential, wait as long as the equally sick 60-year-old in the degenerative phases of his or her life cycle?
Some advantages for the pediatric liver transplant candidates exist in the new system. Unlike adults, children with chronic liver disease with life-threatening complications requiring intensive care unit care, can still be listed at Status 1. There also remains some preference for pediatric donors to be allocated to pediatric patients: for all patients with a PELD or MELD score predicting a greater than 50% mortality after 3 months on the waiting list, a pediatric donor will first be allocated to the pediatric candidate and then to the adult candidate within this category. However as the number of adults with HCC moving their score up every 3 months increases, adults are likely to outnumber children in this high-mortality bracket. Do we want pediatric donors preferentially allocated to adults with HCC?
There are important groups of patients for whom neither MELD nor PELD represents the urgency of their need for transplant; for example those with hepatopulmonary syndrome, pulmonary hypertension, or renal impairment out of proportion with liver dysfunction. Particularly relevant for children are those with metabolic diseases for whom a delay in transplantation may result in profound central nervous system damage. As with adults, children with liver tumors will also have a low calculated PELD score and will require special consideration. The Regional Review Boards will have an increasing role in adjudicating the appropriate MELD and PELD assignation to try and achieve timely transplant. There may be regional differences in how the same PELD or MELD score affects the chances of transplantation, as well in differences in the judgments of the Regional Review Board as to the appropriate score to be given. It will be important to make efforts to smooth out some of the regional differences to remain fair to all adults and children nation wide.
As with any major change, the anxiety of unpredictable consequences still permeates the liver transplant community after the implementation of MELD and PELD. Within the pediatric liver transplant community, we will need to continue to actively partake in the ongoing performance analyses of MELD and PELD including the effects on waiting time to transplant, the number of children transplanted, pre-transplant mortality and post transplant outcome.
Although many of us who care for children with life-threatening but potentially curable diseases are tempted to seek special advantages for children in acquiring liver transplants, we must still try to walk in the shoes of the adult hepatologists and surgeons concerned with the well-being of the larger number of adults with serious disease. Valuing one human life over another is dangerous ethical territory. The stark reality is quite simple; if there were an adequate donor supply we would not have to devise rationing systems with their inherent injustices. However, given the current discrepancy between organ supply and demand, we must move forward with implementation of new ideas developed and validated in a scientific manner, open to ongoing analysis, and tempered by compassion and reasoned discussion between ourselves and our patients.
1. Annual Report the U.S. Scientific Registry of Transplant Recipients and the Organ Procurement and Transplantation Network Transplant Data 1990–1999 U.S. Department of Health and Human Services;2000.
2. Organ Procurement and Transplantation: Assessing current policies and potential impact of the DHHS final rule. Committee on Organ Procurement and Transplantation Policy, Division of Health Sciences Policy, Institute of Medicine. Washington, D.C. National Academy Press;1999:75.
3. Freeman RB, Edwards EB. Liver transplant waiting time does not correlate with waiting list mortality: Implications for liver allocation policy. Liver Transplantation. 2000; 6:543–52.
4. Wiesner RH, McDiarmid SV, Kamath PS, et al. Meld and Peld: Application of Survival Models to Liver Allocation. Liver Transplantation 2001; 7:567–80.
5. McDiarmid SV, Anand R, Lindblad A. Development of a pediatric end stage liver disease score to predict poor outcome awaiting liver transplantation. Transplantation 2002;in press.
6. Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, Ter Borg PC. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology 2000; 31:864–71.
7. SPLIT Research Group. Studies of Pediatric Liver Transplantation (SPLIT):Year 2000 Outcomes. Transplantation
8. Mazariegos GV, Anand R, McDiarmid SV. Validation of PELD Severity Score in a Pediatric Transplant Candidate Database. Am J Trans 2002; 2:251–251.[Abstract]