Are We Ready To Match Donor and Recipient in Liver Transplantation?

On March 10, 2006, according to the United Network for Organ Sharing, there were 17,233 patients awaiting deceased donor liver transplantation (OLTx) in the United States. The same source indicates that 6,121 OLTx were performed in 2005 in this country.1 The average U.S. patient and graft survival after OLTx are 85.7% and 80.7%, respectively, at 1 year and 77.7% and 71.7%, respectively, at 3 years.2 In 2004, 2,136 patients died while on the liver waiting list, bringing the annual death rate per 1,000 waitlisted patients to 125.6.3 The discrepancy between donor supply and organ demand is the major limitation and an utter priority to face for all those working in liver transplantation. To date, a solution for this dramatic gap is still to be found, as we do not have enough organs to transplant all the patients currently placed on the waiting lists. On the one hand, this gap raises such technical issues as the selection of potential donors and recipients on the basis of biological and clinical parameters. On the other, it settles ethical questions: Who gets to be transplanted? Why one patient instead of another? These are arduous choices that transplant surgeons and physicians have faced and that they have debated for the longest time. OLTx, liver transplantation; MELD, model for end-stage liver disease. George Ioannou's study, published in this issue of Liver Transplantation,4 provides an excellent starting point for one of the most critical debates in the field, and it introduces the topic of prediction models applied to liver transplantation. A review of the literature available shows a considerable body of articles investigating the topic and confirming that in the past 10 years, scientific evidence, mathematical survival models and, I daresay, a stricter ethical awareness have supported organ allocation in the United States. Accuracy and equity have significantly improved. Nevertheless, the system still needs adjustments and could certainly benefit from the introduction of new tools that optimize the usage of the scarce donor organ pool.5-7 Several retrospective studies were conducted to test the organ allocation process based on the model for end-stage liver disease (MELD), adopted in the United States in 2002. When compared to mere clinical judgment, MELD has proved accurate and valuable. It appears to both predict and reduce waiting list mortality,8 when used correctly, without fueling geographic disparities,9 but its value in predicting post-OLTx outcomes still needs further analysis.10, 11 Opinions vary on the matter. In a 2004 paper,11 Desai et al. studied a cohort of 2,565 patients to assess whether the MELD score predicted graft and patient survival and length of post-OLTx hospitalization. The authors concluded that MELD was a poor predictor of post-OLTx outcome except for those patients with the highest scores. Also, a model with 4 variables was developed that showed a better ability than the MELD score to predict post-OLTx patient survival. In their conclusions, Desai et al. stressed that such statistical model could be used to rule out futile OLTx in cases with poor expected outcomes. An earlier study12 had been conducted on a larger population (25,772 patients), developing an 8-factor prognostic model that seemed to work well also in case of retransplantation and in patients with hepatitis C. Following previous studies on the effect of donor age and sex on the outcome of liver transplantation,13, 14 already in the late 1990s, with a retrospective multivariate analysis of 2,376 consecutive transplantations, the major risk factors associated with post-OLTx outcome had been identified, showing that donors and recipients could be stratified into differential categories at the time of surgery. In fact, using 3 donor and 7 recipient variables, this stratification proved useful when comparing results across different groups and in helping to choose the best donor-recipient combination based on the calculated probability of a favorable outcome.15, 16 George Ioannou's study examines United Network for Organ Sharing data derived from the Standard Transplant Analysis and Research file.4 The file is developed with the information collected by U.S. transplantation centers and by organ procurement organizations. The author excludes certain subgroups from his analysis, such as donors <10 and >75 years of age, non–heart beating donors, and donors with a serum sodium concentration >170 mmol/L, as well as recipients of multiple organs or patients who underwent previous liver transplantation. This is unfortunate, because information on outcomes of OLTx with donors >75 years of age and with serum sodium >170 mmol/L would be useful, as these are the riskier subgroups of deceased donors that raise the main concerns when it comes to crucial decisions of organ allocation. According to recent statistics, liver graft survival for donors >65 at 1 year is only 72%, dropping to 59% at 3 years.17 As for the influence of high donor serum sodium levels on postoperative graft function, we have long known that recipients of hepatic allografts from donors with uncorrected hypernatremia had a significantly greater incidence of graft loss compared with recipients of hepatic allografts from normonatremic donors.18, 19 Nevertheless, Ioannou's multivariate analysis is constructive when it identifies 4 donor variables (age, ischemia time, sex, and ethnicity) and 9 recipient variables (age, body mass index, MELD score, United Network for Organ Sharing priority status, sex, ethnicity, diabetes mellitus, liver disease and albumin), which emerge as statistically significant for predicting outcome after OLTx. This conclusion, however, is not different from that reported in many other previous, though smaller, analyses, including data from our group.13-16 Ioannou raises a critical point, stressing that “among patients without HCV, the 5-year survival of the average recipient with the average donor is 76%. However, when a high-risk recipient (MELD = 24, albumin <2.1) receives an average donor liver, the predicted 5-year survival is only 51%; when the same high-risk recipient receives a high-risk donor liver (age = 60 years, cold ischemia time = 14.3 hours), the predicted 5-year survival is down to 28%.”4 Now, a predicted 5-year survival as low as 23% is perfectly acceptable for surgical procedures such as pancreaticoduodenectomies,20 which, as opposed to OLTx, do not rely on the use of limited resources (Charles J. Yeo, personal communication). However, when we're faced with the responsibility of allocating priceless and limited supplies, such as livers from deceased donors, our perspective changes. What are we to do with the information provided by Ioannou, then? This model, as well as all others offered so far, has its limitations, but I believe that our duty is to focus on the general approach rather than on single aspects. To begin with, how should we use the information derived from the Score of Liver Donor? Ioannou shows that the higher the Score of Liver Donor, the lower the recipient survival, and this information is available prior to OLTx. Ethically, we cannot ignore this information, nor can we avoid sharing it with the patient. But how? The theory of informed consent tells us that a competent patient who is given all the necessary facts will make a truly informed, autonomous decision. But the reality is that as the facts increase in complexity, the patient's ability to make a truly informed decision decreases proportionally. Those working in the field know well that patients generally end up deciding whether to accept or refuse an organ by relying mainly on the judgment expressed by the physician who has followed them prior to OLTx; in liver transplantation, informed decisions are more the exception than the rule. That places a heavy burden on our shoulders. What is the correct policy: to try to perform an OLTx on every waitlisted patient, no matter what the chances for survival are, or to try to optimize outcomes, given the scarce supply of deceased liver donors? The outcome of OLTx is a complex phenomenon, resulting from the interaction of 2 different biological systems—the donor's and the recipient's. For this reason, when assessing risk factors before OLTx, we cannot limit ourselves to considering either only the donor or the recipient. For decades, deceased donor livers were allocated on the basis of accrued waiting time and of criteria that subjectively assessed the seriousness of patients' conditions. This system proves efficient and does not pose conflicts only when transplant candidates are young patients suffering from fulminant hepatitis, whose life expectancy is a matter of hours or of days. Deceased donor liver allocation for patients suffering from chronic liver diseases, on the other hand, raises priority issues that are not easy to solve. The allocation system currently in place stems from the feeling of unfairness shared among the transplant community and the general public that in 1998 triggered the intervention of the Department of Health and Human Services and the publication of the so-called “Final Rule.”21 This document inaugurated the process that in 2002 brought to the introduction of the MELD score system. Although this did bring about several advantages,22 as mentioned above crucial matters remain open. For example, it is evident that patients suffering from hepatocellular carcinoma used to be penalized in the pre-MELD era, when their average waiting time on the transplant list was 2.28 years, as opposed to today's 0.69 years.23 There is no doubt that hepatocellular carcinoma patients have benefited from the changes in the allocation system. Nevertheless, at the same time, it is very likely that the same system has been responsible for the higher mortality rates for waitlisted patients suffering from end-stage liver diseases different from hepatocellular carcinoma. Such patients presently need to be clinically much sicker then the others to compete in the allocation of a deceased donor liver. The crux of the matter remains the extent of objectivity and subjectivity required by any allocation system in order for it to result as fair as possible. I could not agree more with the concept, expressed by Robert Merion, that we have broadly accepted listing criteria, while delisting criteria have never been incorporated into a policy.24 As stated in his article, it is relatively easy to indicate when a patient is too well for an OLTx, while it is significantly more challenging to establish when a patient is too sick. We all know that a futile OLTx for an extremely sick patient will not help that liver candidate and will deprive another patient of a scarce resource: the deceased donor liver. It is hard to disagree with what Ioannou emphasizes in the discussion section of his paper4: MELD scores being equal, expected survival can significantly differ due to the donor's characteristics. Then, if candidates for OLTx can be stratified into different risk categories at the time of the actual organ allocation, the inevitable question arises: Should we try to match donors and recipients? In addition, should we try to implement rules to assess when a patient is too sick for OLTx and have uniform delisting criteria? I believe it is unfair to leave the decision up to single transplant centers. Criteria should be developed and applied uniformly, as we are all using the same scarce resource: deceased donor organs. However, the cutoff to assess when a patient is too sick to undergo OLTx cannot be set without the support of additional data. Such information can be derived from the application of criteria as those indicated by Ioannou. In other words, we might not be ready to match donors and recipients yet, but this procedure should be our ultimate goal. A large prospective stratification of OLTx candidates at the time of surgery should be implemented. Comparing results across different groups will then help optimize allocation criteria and define when a prospective donor should not be used for a prospective recipient. The author thanks Ms. Claudia Cirillo for her support with research and editorial work, and Dr. Howard R. Doyle for past collaboration and continuing discussion on this issue.