Improving Patient Access to Liver Allografts Through a Mathematically Optimized Continuous Organ Distribution Model

Abstract

COMMENTARY Liver organ allocation policy in the United States has been at the forefront of controversy in the transplant community over the past decade. Arguments have intensified in the last year, following the Health Resources and Services Administration mandate to improve equity in organ distribution and the resulting litigations.1,2 As need continues to exceed organ availability, the disparities in distribution and access have become more evident. The common points of contention with current allocation practice is not only organ availability, but also center and regional variations in listing practices, application for model of end-stage liver disease (MELD) exceptions, recipient medical acuity at transplant, utilization of marginal donors, and waiting times.3–6 Combined, these factors profoundly affect a patient’s probability of transplant, and have prompted the search for a solution. In this issue of Transplantation, Bertsimas and colleagues present their article “Balancing efficiency and fairness in liver transplant access: tradeoff curves for the assessment of organ distribution policies.” Here, they compare the 3 distribution frameworks currently under evaluation by the OPTN/UNOS Ad Hoc Geography Committee: the approved, but contested, acuity circles (fixed radii around the donor hospital with patient stratification via model of end-stage liver disease sodium [MELD-Na] score); mathematically optimized geographic boundaries (optimized allocation districts); and continuous distribution (allocation by a score composed of factors such as medical urgency and proximity to donor without specific geographic limits).7 Using tradeoff curves and a standardized computer model for allocation, the group demonstrates that the continuous distribution system outperforms the former 2 by reducing recipient mortality and MELD variability between centers.8 The effect is maintained regardless of the disease acuity score used (MELD versus MELD-Na). Furthermore, the previously proposed novel medical need score, the optimized prediction of mortality, compounds the effects of continuous scoring and has the greatest effect on lives saved across all transport distances.2,8 This latter concept, while not the primary focus of the article, adds an additional layer of complexity to the mathematical algorithms that bears further evaluation. The results of the analysis of the continuous distribution framework are intriguing. They again highlight the need for additional deliberation, as the optimal allocation system is selected and implemented. There are several factors that require further clarification with the 3 described models. For example, what is the effect on actual travel time, which does not always correlate to linear distance, especially for land travel? This is important, as transit time directly increases liver cold ischemia time and contributes to organ viability and posttransplant function.9 The authors attempt to flush this out by extrapolating average time traveled compared with annual average deaths and SD in median MELD. The results are similar to those above, with the continuous distribution model again outperforming acuity circles and optimized districts. The substitution of optimized prediction of mortality for MELD or MELD-Na builds on this effect, significantly lowering average total patient deaths across all travel times.8 The results of the present analysis must be interpreted with caution, as the data analyzed arises from the pre-Share 35 era (2007–2011). Whether the current broader distribution policy (ie, Share 35) would alter the computer models is unclear and requires further evaluation. As geographic reach is expanded, other concerns, such as the greater costs associated with air transportation and requirements for increased logistical support, may modify the pertinent considerations for optimal distribution system design. Smaller transplants centers may be especially vulnerable to such changes. Additional analysis is necessary to determine which model will optimize these factors. Finally, any allocation scheme that favors wider distribution will ultimately impact utilization of “marginal” organs. The decrease in extended criteria donor organ use with greater distance is intuitive. The authors show a trend toward increased organ discard rates and decreased average graft survival for marginal grafts as transport distance increases8; however, the continuous distribution model may stabilize these effects. More research is clearly warranted, as we optimize both allocation frameworks and utilization of marginal grafts to maximize liver graft allocation and recipient benefit. Inclusion of a more comprehensive medical acuity score into the distribution policy may remove some of the inherent bias in the MELD/MELD-Na score to further improve equity. Evolving technologies, such as machine preservation techniques, have the potential to drastically alter the temporal limitations of organ distribution. Furthermore, as the landscape changes, constant reassessment will be necessary to ensure continued ethical practices. In the end, it all goes back to the final rule: the goal is to adopt an organ allocation system, which optimizes equity, patient access to this life saving therapy, is based on sound medical judgement, and promotes organ utilization and efficiency.10 This study showcases 1 of the 3 proposed allocation frameworks and demonstrates intriguing potential effects on future patient lives. However, while mathematical and statistical models are excellent in theory, the feasibility of their application and utilization remains to be seen.