In 2009, Schaubel et al. proposed a new deceased donor liver allocation model as an alternative to the current Model for End-Stage Liver Disease (MELD) and its pediatric equivalent [Pediatric End-Stage Liver Disease (PELD)]. The Organ Procurement and Transplantation Network (OPTN)/United Network for Organ Sharing (UNOS) Liver and Intestinal Organ Transplantation Committee has since considered this proposal and concluded that the current allocation system is not in need of modification. Rather, the committee has focused on addressing disparities in geographic distribution with the hope that these efforts will help to optimize the utilization of donor livers.[2, 3] In contrast, we will argue that the model proposed by Schaubel et al., with modifications, better balances a more comprehensive list of ethical concerns. It, therefore, represents a justifiable alternative to the current MELD/PELD system and complements current efforts to improve distribution. We will illustrate this need by analyzing the ways in which this survival benefit–based model better comprehends the ethical concerns shaping liver allocation.
The disparity between the demand for and supply of donor livers has continued to grow over the last 2 decades, and this has placed greater weight on the need for efficient and effective liver allocation. Although the use of extended criteria donors has shown great potential, it remains unregulated. A survival benefit–based model was recently proposed to answer calls to increase efficiency and reduce futile transplants. However, it was previously determined that the current allocation system was not in need of modification and that instead geographic disparities should be addressed. In contrast, we believe that there is a significant need to replace the current allocation system and complement efforts to improve donor liver distribution. We illustrate this need first by identifying major ethical concerns shaping liver allocation and then by using these concerns to identify strengths and shortcomings of the Model for End-Stage Liver Disease/Pediatric End-Stage Liver Disease system and a survival benefit–based model. The latter model is a promising means of improving liver allocation: it incorporates a greater number of ethical principles, uses a sophisticated statistical model to increase efficiency and reduce waste, minimizes bias, and parallels developments in the allocation of other organs. However, it remains limited in its posttransplant predictive accuracy and may raise potential issues regarding informed consent. In addition, the proposed model fails to include quality-of-life concerns and prioritize younger patients. We feel that it is time to take the next steps toward better liver allocation not only through reductions in geographic disparities but also through the adoption of a model better equipped to balance the many ethical concerns shaping organ allocation. Thus, we support the development of a similar model with suggested amendments. Liver Transpl 20:140-146, 2014. © 2013 AASLD.
extended criteria donation
living donor liver transplantation
Model for End-Stage Liver Disease
Organ Procurement and Transplantation Network
Pediatric End-Stage Liver Disease
United Network for Organ Sharing
CURRENT STATE OF LIVER TRANSPLANTATION
Liver transplantation remains the most effective and definitive means of treating patients with end-stage liver disease. However, improved outcomes have led to an ever-increasing demand for donor livers. In 2012, 5757 liver transplants were performed in the United States. There remains a great disparity, however, between the supply of donor livers and the demand, with 10,246 new candidates added to the wait list in 2012. This means that were 15,783 wait-list candidates as of February 2013.
In order to increase the supply of donor livers, the use of extended criteria donation (ECD) has become widespread. ECD refers to the use of donor livers that do not meet the standard criteria for transplantation. Many studies have suggested that the use of ECD allografts can safely reduce the time that potential recipients spend on the wait list without subjecting them to additional transplantation risks that would outweigh the benefits of a reduced wait-list time.[6-14] However, there is no clear consensus on 1 set of extended criteria. Extended age, for example, has been defined as an age > 59 years in one study and as an age > 80 years in another.[9, 13] That said, ECD has commonly included criteria such as a higher donor age, donor liver steatosis, an increased cold ischemia time, an increased warm ischemia time, a high donor body mass index, and a peak serum sodium level as well as the presence of infections or malignancies and procurement via donation after cardiac death.[15, 16] Although this practice has increased the number of transplantable livers, it remains unregulated by the current allocation system.
Another strategy for expanding the supply of donors has been living donor liver transplantation (LDLT). LDLT uses a partial liver allograft from a living donor. This has raised a number of ethical concerns, mostly because of the inherent risk to healthy donors. Supporters argue that as long as patient autonomy, a lack of coercion, and informed consent are emphasized, LDLT is justified by the growing demand for livers.[17, 18] Nevertheless, LDLT has failed to significantly expand the supply of donor livers since its implementation in the mid-1990s. Thus, new models for improving liver allocation have focused on the incorporation of ECD rather than LDLT.
The current MELD/PELD system for liver allocation is based primarily on medical need, which is objectively defined by the measurement of a patient's bilirubin level, prothrombin time (international normalized ratio), and creatinine level to provide an estimation of the patient's 3-month wait-list mortality risk. The MELD/PELD algorithm yields a numerical score from 6 to 40 that represents the patient's need for liver transplantation; patients with higher scores are given more priority. Schaubel et al. and others[19-21] have criticized this model primarily because of its inefficiency. Because patients with greater medical need tend to have worse posttransplant outcomes, the MELD/PELD system allocates livers to patients with less potential benefit than others. Thus, these authors have suggested that a model balancing medical need with potential benefit would be best because of the large disparity between the demand for and supply of donor livers.
The survival benefit–based model proposed by Schaubel et al. attempts to achieve this balance through the estimation of a patient's 5-year mean lifetime with and without transplantation and an examination of the difference between the two. In other words, a patient expected to live 1 year on the wait list and 5 years with transplantation would have a survival benefit of 4 years. To make these estimations, Schaubel et al. used separate Cox regression models to incorporate a wide array of donor and recipient characteristics that were selected through backwards elimination; they began with all relevant variables included in program-specific report models or previous Scientific Registry of Transplant Recipients analyses. Recipient characteristics included creatinine, bilirubin, international normalized ratio, albumin, sodium, age, body mass index, previous time on the waiting list, prior malignancy, diabetes, hepatitis C, and presence of hepatocellular carcinoma. Donor characteristics included age, cause of death, donation after cardiac death status, transplant factors (eg, cold ischemia time), and share type (regional or national). To assess the potential impact of implementing this model, they used a liver stimulated allocation model to estimate a potential benefit of 2223 additional life years saved per year. In our opinion, this model's potentially improved strategy for improving efficiency is further strengthened by its ability to better address a number of ethical concerns shaping liver allocation in comparison with the current MELD/PELD system. Thus, we will first identify the relevant ethical concerns and then suggest how each is strengthened by the survival benefit model.
ETHICAL CONCERNS SHAPING LIVER ALLOCATION
For this analysis, we first reviewed recent literature on organ allocation ethics. This review revealed a wide variety of ethical concerns considered to be important; they included, but were not limited to, equity, solidarity, fairness, efficiency, quality of life, maximum benefit, economical responsibility, informed consent, and minimum corruptibility.[19, 20, 22-24] Kerstein and Bognar offered 2 more general characteristics: they stated that an acceptable system should be based on secure moral foundations and offer practical guidance. We then grouped this more specific list of ethical concerns into 4 general principles: justice, utility, beneficence/nonmaleficence, and patient autonomy. The latter 2 characteristics will be addressed in our evaluation of the survival benefit model later in our discussion.
Most discussions concerning the allocation of scarce medical resources have emphasized the importance of justice. We use the term justice to encompass concerns of equity, medical need, and fairness. Equity calls for individuals under equal conditions to be treated equally or to have equal opportunities to live out their life plans: patient A with acute liver failure should receive the same treatment opportunities as patient B with acute liver failure.[19, 25] Medical need describes a commitment to providing injured individuals with a benefit proportional to their loss. This commitment to treat the sickest patients first is also supported by federal law. Fairness is a rather broad term. It has been used to describe the notion that patients who accept greater risks should receive some additional benefit or that priority should increase with time on the wait list. In the case of liver allocation, additional weight is placed on the need for fairness because transplantation is often the sole treatment option and the alternative is death. Overall, justice demands that allocation models be perceived as fair and uniformly applied.
Some authors have used concerns for justice to argue that social worth or responsibility should be factored into organ allocation policies.[26-29] Such a system would deprioritize candidates whose need for transplantation is the result of a poor choice such as alcohol or drug abuse. Because of the subjectivity in distinguishing truly autonomous decisions and the evolving understanding of alcohol and drug dependence as complex medical illnesses, using such criteria runs the risk of discriminating against patients who engage in activities viewed as moral failures by the societies in which they live.[29, 30] We and most other authors do not believe social worth criteria should be factored into organ allocation, and thus we have excluded social worth from our assessment.
Recent literature on the ethics of alternative allocation models often argues for a greater emphasis on the effective and efficient use of donor organs.[23, 24, 31] In response to the growing demand for donor organs, the 1984 National Organ Transplant Act and the UNOS final rule have both called for the reduction of futile and wasteful transplants. To describe concerns of efficiency and economic responsibility, we will use the term utility, which is often used in transplant ethics to describe an effort to maximize the greatest benefit for the greatest number of patients. Utility is commonly used to support efforts to maximize life years gained: if patient A would live 10 years after he or she received a given treatment and patient B would live 8 years, patient A should receive the treatment before patient B. However, solely maximizing life years would discriminate against the very old and sick, who generally have worse outcomes. Utility is also difficult to measure because it relies on the soundness and accuracy of medical predictions. Nevertheless, the large demand for donor organs has placed a greater weight on demands for greater utility in organ donation.
Two important concepts in clinical ethics are beneficence and nonmaleficence. Beneficence refers to a commitment to choose treatments that benefit the patient, whereas nonmaleficence describes a commitment to prevent further injury or an increased risk of harm. We use these terms to also encompass concerns of quality and risk of harm that affect transplant outcomes. There are 2 primary quality judgments related to liver allocation: donor liver quality and recipient quality of life. Risk of harm, which is increased by any surgical operation, must be outweighed by the expected benefits to uphold nonmaleficence. This justification is supported by proportionality, which is the idea that a given treatment is ethical to the extent that the expected benefits outweigh the potential risks. Therefore, an ideal allocation model should support beneficence and nonmaleficence by maximizing patients' benefits while minimizing the risks of harm.
The remaining ethical concerns proposed by authors included in our review emphasize informed consent and system transparency with minimization of conflicts of interests.[19, 33] These concerns are all related to patient autonomy, which refers to respect for the moral right of an individual to follow and make free choices related to his or her life plan. In this respect, the principle of autonomy underpins informed consent. Patient autonomy is often left out of discussions of organ allocation, likely because of previous models' uniform application and objective criteria, which have minimized conflicts of interests. Furthermore, if no standard measure of organ quality exists, recipients likely view all organs that they may receive as more or less equal. If the limits of organ quality are pushed via ECD, patients should be fully informed of any additional risks. Overall concerns of autonomy demand that allocation models be transparent and that patients be fully informed of important information related to their potential liver transplantation.
MODEL FOR END-STAGE LIVER DISEASE/PEDIATRIC END-STAGE LIVER DISEASE VERSUS SURVIVAL BENEFIT–BASED MODEL
Using the ethical concerns discussed previously, we will first highlight the strengths and shortcomings of the MELD/PELD system before we summarize the ways in which the survival benefit–based model better comprehends the important ethical principles that we have just reviewed.
Model for End-Stage Liver Disease/Pediatric End-Stage Liver Disease
Since its implementation in 2002, the MELD system has reduced wait-list mortality and prioritized sicker patients by better predicting wait-list mortality. Past and current liver allocation models in the United States have heavily emphasized justice by scoring medical need on the basis of objective medical criteria and by applying the system uniformly across all US transplant centers. Before improved uniformity was achieved, allocation decisions were shown to be inconsistent and vulnerable to conflicts of interest. Furthermore, the PELD system prioritizes patients less than 12 years old and meets the common notion that children should be prioritized because they have not yet had an opportunity to live out their biographical lives. Inadequacies of the MELD algorithm, such as the scoring of patients with hepatocellular cancer (which is commonly felt to be underscored), have been addressed through the implementation of exception points. Overall, because of these strengths, the MELD algorithm strongly supports concerns of justice and protects patient autonomy.
Although the MELD system once existed as a better alternative to the prior Child-Turcotte-Pugh system, it may no longer be the best model for predicting medical need. For example, Myers et al. recently illustrated the importance of serum albumin in predicting liver wait-list mortality, which the MELD system does not take into account. The three measures used by the MELD score may also not be properly weighted to most accurately reflect other prognostically relevant medical factors.[1, 21] For example, a recent study showed that the incidence of end-stage renal disease in liver recipients after transplantation has significantly increased with the use of the MELD system. Even those authors who have agreed that the MELD score is well designed for predicting wait-list mortality have also criticized it for giving little priority to prognosis.[19, 20, 27, 38] Without the ability to predict potential benefit, it fails to answer calls for efficiency and reduction of probable futility.[22, 24, 31, 32]
The MELD system is also not well equipped to regulate the use of ECD for expanding the supply of donor livers. The allocation of high-risk livers only to patients who can achieve a significant medical benefit saves lower risk livers for other candidates. Without an objective measure of donor organ quality, ECD risks subjecting recipients to hidden risks associated with less healthy donor livers. Thus, the MELD algorithm also fails to allow for the wider use of ECD without undermining nonmaleficence and patient autonomy.
Strengths of the Survival Benefit–Based Model
The primary ethical strength of the survival benefit–based model lies in its attempt to balance the competing concerns of justice and utility.[1, 21] Justice-centered allocation tends to prioritize the sickest patients, who have worse outcomes, whereas utility-centered allocation would prioritize the healthiest patients. The model emphasizes justice through its use of better weighted MELD criteria and additional objective factors such as diagnoses that have previously required the assignment of exception points. By eliminating the need for exception points, the survival benefit–based model minimizes potential inequalities and bias toward some medical conditions over others (eg, cancer). Concerns for utility are addressed by the model's incorporation of posttransplant outcomes and donor liver quality. Rather than simply labeling livers as either healthy or not healthy, this provides a continuum of quality and thus will have the advantage of intrinsically being capable of regulating ECD. We believe, therefore, that this model may provide a fairer system that better addresses calls for efficiency.
Beyond balancing justice and utility, the survival benefit–based model addresses additional ethical concerns (and thus adds to its secure ethical foundations) and offers practical guidance through its powerful statistical design. The model is better equipped to assess the risks and benefits of a potential transplant, and this further supports concerns of beneficence and patient autonomy. For example, livers that are higher risk than those used today could provide a currently unknown benefit to certain patient populations; conversely, some ECD livers used today with an unfavorable risk/benefit ratio might no longer be used. The greater practical guidance offered by this model is the result of the model's use of Cox regression, which is a powerful statistical model used widely in biomedical literature and well equipped to balance the many recipient and donor factors considered. Thus, we believe that the survival benefit–based model has a more rigorous ethical foundation and offers more practical guidance than the current MELD system.
A final point in support of the survival benefit–based model is its close resemblance to current developments in lung and kidney allocation. Lungs are similar to livers in that transplantation is often the only treatment option for wait-list candidates. In order to improve lung allocation, the lung allocation score has recently been adopted; this is a normalized score that balances wait-list and posttransplant mortality predictions to maximize efficiency. A recent multicenter study showed that the lung allocation score was achieving its goals and significantly affecting lung allocation. Although kidney transplantation is not completely analogous to liver transplantation because of the availability of organ replacement therapy, the proposed 20/80 system also attempts to balance concerns of distributive justice and efficiency.[23, 40] The survival benefit–based model thus appears to mirror current trends in the development of organ allocation through the incorporation of concerns beyond wait-list mortality.
Shortcomings of the Survival Benefit–Based Model
One major issue with attempts to predict posttransplant outcomes is imprecision. Schaubel et al. acknowledged their model's less encouraging ability to accurately predict posttransplant outcomes, and they reported an index of concordance (C statistic) of 0.63 for 5-year posttransplant survival. The C statistic measures the percentage of time that a model correctly predicts which of 2 patients will die first, so flipping a coin would yield a C statistic of 0.50. In comparison, the MELD C statistic for 3-month wait-list mortality is 0.83. This implies that the survival benefit–based model may omit factors that significantly affect posttransplant survival.
Additionally, the measure of benefit used by Schaubel et al. is inadequate because it equates benefit with life years gained and omits quality-of-life concerns. Quality of life is relevant in assessing the medical benefit of a given treatment.[30, 38] A transplant candidate who is expected to gain three 25% quality years may be chosen over a candidate who could have gained two 90% quality years. Because of the subjective nature of determining quality of life, many like Schaubel et al. have instead focused on the more objective and easier to measure outcome of life years gained. Although we agree that simply gaining life years is a good thing, we feel that the absence of an incorporated and valid quality-of-life measure further limits the strength of the model's benefit prediction.
In addition to quality-of-life concerns, Schaubel et al. also acknowledge their model's lack of priority for younger patients. Furthermore, the survival benefit–based model uses a candidate's age as a factor for predicting transplant benefit, and this would deprioritize older patients. Without proper justification, we anticipate that this characteristic will limit the public acceptance of this model. Although one could employ an approach similar to that of the MELD/PELD system for younger candidates, we believe that questions of how to fairly allocate resources at any stage of life may have the same solution (see the suggestions presented later).
Finally, the survival benefit–based model fails to define the upper limit of donor liver risk. For example, a very poor quality liver may become available that would add an additional life year to the ideal recipient. Who is to say that this expected benefit outweighs his or her somewhat subjective opportunity costs? What if the recipient waits another year and, although his or her condition worsens, a liver becomes available that would extend his or her life 2 years? Without the clear limits of the current system in which all healthy livers are assumed to produce similar outcomes, a new complexity arises that places additional weight on the role of informed consent and requires further discussion.
Suggested Improvements to the Survival Benefit–Based Model
In order to improve the accuracy and practicality of the survival benefit–based model's transplant benefit prediction, future models should focus on significant factors available at procurement, such as donor age, graft steatosis, and the cause of death, as suggested by Freeman et al. Other factors shown to significantly affect posttransplant outcomes, such as the warm and cold ischemia times, are not always known at the time of procurement; therefore, a less accurate posttransplant prediction may be inevitable.[19, 20] The success of the model may also be limited by its use of uncorrected national data, which show greater benefit for patients with higher MELD scores. Liver transplant patients with higher creatinine levels due to kidney dysfunction have been shown to have significantly reduced posttransplant survival. Because these patients tend to have higher MELD scores, the corrected data show the opposite result: a greater transplant benefit for patients with lower MELD scores. Although the use of national data has many advantages, it may also be easier to overlook such factors that profoundly affect results.
Future liver allocation models should also incorporate objective measures of the quality of life years gained. This may be achieved with a tool from health economics called quality-adjusted life years. Quality-adjusted life years are used to compare the expected benefit of a given treatment to the patient's opportunity costs with benefit adjustments for inflation and quality of life. However, it should be noted that estimations of a given patient's opportunity costs and quality-of-life judgments are both rather subjective. Further research will be required to investigate how this tool could be incorporated into a liver allocation model. What is clear is that the quality-adjusted life year model has valuable potential for enhancing the ethical foundations of future allocation models.
To better address the prioritization of different stages of life, we recommend incorporating an approach similar to that used by the Equal Opportunity Supplemented by Fair Innings model. Under the Equal Opportunity Supplemented by Fair Innings model, all ages have an equal chance of receiving a donor organ, but higher quality organs are given to younger wait-list candidates. Younger wait-list candidates could receive an additional weighted benefit for livers above a certain quality, and this would create a continuum of weighted benefits rather than a particular cutoff age. Although some may argue that such a model discriminates against older patients, age is different from characteristics such as race and sex because it is experienced by all individuals. It is, therefore, an ethically neutral characteristic. If different stages of life are treated differently but everyone ages, then everyone is treated equally over his or her lifetime. Furthermore, because of the scarcity of donor livers, this unequal treatment may be to the benefit of all individuals over their lifetimes.
In order to set limits on the use of ECD, one must consider economic concerns and those of patient autonomy. A study by Axelrod et al. suggested that the use of high-risk livers increased both transplant costs and posttransplant days in the hospital. In our opinion, such economic concerns can be ethically addressed only at the level of the society at large. Freeman suggested allowing candidates to indicate the highest risk that they are willing to take. This more rigorous informed consent process may empower patients to set personal limits on the use of ECD and choose the economic burden that they are willing to bear. Thus, we suggest that policies should eventually be developed along with allocation models incorporating ECD to allow candidates to have an active role in determining the quality of the organ that they may receive.
SUMMARY AND CONCLUSION
The demand for donor livers has continued to grow over the last 2 decades, and this has placed greater weight on the need for efficient and effective means of increasing the supply. ECD has expanded the supply of donor livers, but it remains unregulated. Schaubel et al. recently proposed the survival benefit–based model to balance wait-list survival and potential transplantation benefits in the era of ECD. However, the OPTN/UNOS Liver and Intestinal Organ Transplantation Committee feels that the current MELD/PELD allocation system does not require modification and has instead focused on reducing geographic disparities. Although distribution issues are also important, we believe that there is a significant need to adopt the survival benefit–based model (or a similar model), which better balances the most important ethical principles.
To improve this model even further, we suggest incorporating other factors used to predict posttransplant outcomes (including the quality of life with a quality-adjusted life year model), weighting age differently for different stages of life, and developing a means for candidates to have an active role in weighing risks and economic concerns.
It is time to take the next steps toward better liver allocation through not only a reduction of geographic disparities but also the adoption of a model better equipped to balance the many ethical principles underpinning organ allocation. Thus, we urge the OPTN/UNOS Liver and Intestinal Organ Transplantation Committee and the transplant community to strongly consider this evaluation.
The authors thank Dr. A. Joseph Tector III for his feedback and support.