Currently, allocation and distribution of organs are the most controversial aspects of liver transplantation. The scarcity of deceased donor organs has forced transplant centers and patients listed for transplant to compete with other local programs and locally listed patients, respectively, for liver grafts. The requirements for an optimal organ allocation system are (1) evidence-based outcome measures, (2) justice and equity, (3) maximal utility and (4) transparency. Balancing justice and equity for all candidates and individual vs. societal utility for a scarce resource is one of the largest challenges. The change in the allocation system from a waiting time-based system to a risk-based system using the Model for End stage Liver Disease (MELD) score has lowered pretransplant mortality, without adversely impacting posttransplant survival despite increased severity of illness at the time of transplantation (1,2). This is a model of how data-driven policy changes can improve outcomes.
In this issue of the American Journal of Transplantation, Merion et al. present a detailed analysis of the scientific registry of transplant recipients to determine the transplant benefit for various MELD scores using time-dependent Cox regression models (3). Transplant benefit was defined as the difference in the likelihood of 1-year survival if one undergoes a transplant at a given MELD score vs. remaining on the waiting list. There is an overall survival benefit of transplantation in the first year; however, the benefit varied markedly by MELD score. A significant survival benefit with transplantation was seen only for patients with high MELD scores (i.e. >18); the survival benefit continuing to increase as MELD score increased, with greatest survival benefit being achieved in those at the highest MELD scores. Conversely, there was increased mortality in the first year posttransplant compared to remaining on the waiting list for patients with lower MELD scores (<15), particularly those with MELD scores less than 12.
These models were adjusted for other factors that might independently influence outcome, including age, sex, race, diagnosis, education, insurance coverage and the changes in MELD over time (delta MELD). The overall covariate adjusted mortality risk was 79% lower for patients transplanted compared to those remaining on the waiting list. However, the risk of dying in the first year posttransplant was almost four-fold higher than remaining on the waiting list for patients with MELD scores 6–11 and two-fold higher for those with MELD 12–14. A significant transplant benefit was seen for all candidates with MELD greater than 18, increasing exponentially at higher MELD scores with up to a 96% reduction in mortality risk for patients with MELD scores of greater than 40.
The reason for this large difference is because pretransplant mortality risk increases exponentially rather than linearly for higher MELD scores. Though the risk of mortality posttransplant for high MELD scores is increased, it does not parallel the reduction in pretransplant mortality, resulting in increased survival benefit with each increase in MELD score. Over the range of MELD from 6 to 40, the impact on posttransplant survival was modest, 1.5-fold, compared to the over 300-fold change in the pretransplant mortality risk.
These data need to be interpreted carefully. First, it analyzes the patient's risk at each MELD score at a given point in time. When patients transit from one MELD score to another score, they are no longer counted in the initial MELD score risk group. Thus, the transplant benefit is for a composite group of patients at any given MELD score. Second, the patients remaining on the waiting list at a given MELD may be fundamentally different than those transplanted; however, it is unlikely that a selection bias can account for the magnitude of the difference seen. At an individual level, transplant professionals want to advise their patients of all the likely outcomes, including the risk of disease progression. Patients may transit to higher MELD scores; does this analysis reflect their risk or will the transplant benefit change in the short-run? Fortunately, it appears individual short-run risk is accurately measured. The MELD scores remain stable in low MELD candidates; over 75% of the patients with scores between 6 and 11 were in that range 1 year later. Additionally, delta MELD was not a significant covariate in the regression models. The largest limitation of this study is that it measures survival benefit only for the first year. Over time, the impact of perioperative mortality is minimized, and the benefits of transplant increase. Though long-term benefit is our goal, the relative benefit will likely remain the same as the impact of MELD on posttransplant survival is principally in the early posttransplant period. Other variables, e.g. age, center outcome, will also affect transplant benefit and require further study. Finally, these analyses only address survival and not the impact of transplantation on quality-of-life.
These concerns need to be tempered by the fact that we are allocating a scarce resource. With unlimited organs, one might apply transplantation to a broader range of patients. Since differences in transplant benefit will likely remain in the long term between high and low MELD candidates, these data should be used to guide prioritization of the patients with the greatest benefit and to inform potential recipients of the relative risks of early transplantation vs. waiting. It also may be used as a guide for the timing of live donor liver transplantation.
The implications of this analysis for organ allocation policy are apparent. Currently, 24% of all deceased donor liver transplants are performed in patients with MELD scores less than 14, and 10% with MELD scores less than 11. This number has significant OPO-to-OPO and regional variation from 0% to over 25% for the very low MELD (<11) scores. Increased transplant benefit can be obtained if the UNOS allocation algorithm shifts the distribution of transplants to patients with MELD scores greater than 15. Since the vast majority of patients with MELD scores less than 14 can be observed for up to a year without significant increases in MELD, this change should not result in increased mortality. Setting a minimum listing transplant score and wider sharing areas for distribution of organs are under active consideration; the OPTN board has already approved broader sharing agreements before local use of grafts in patients with MELD scores less than 15.
The concern with broader sharing and transplantation of high MELD candidates is an increase in futile transplants. No absolute MELD score predicted >50% mortality or transplant futility. Thus, it does appear that a meaningful absolute transplant benefit is achieved, not merely a large relative benefit. Centers have a higher rate of waiting list removal for the reason of “death/too sick” at high MELD scores, thus avoiding futile transplants by careful patient selection. Broader sharing arrangements are unlikely to change this practice in any significant way.
The implementation of MELD for organ allocation has already resulted in improved outcomes in transplantation. It has also broadened transplant research from a sole focus on posttransplant survival. As a therapy for end stage liver disease, the goal of transplantation is reduction in overall mortality for the entire cohort, i.e. all those eligible for transplant, not only those undergoing the procedure. The data in this report elucidate which patients derive the most benefit from transplantation. Thus, another shift is required where patients who have a greater benefit of transplant are allowed more timely access to organs and patients who have low and stable MELD score wait until they can derive a significant survival benefit from transplant. More rapid transplantation of high MELD candidates may even improve posttransplant outcomes to a point where this benefit is not accompanied by any increase in posttransplant mortality. Most importantly, as a transplant community, we should remember our goals in allocation policy are to save lives, not liver transplant programs.