Using ethics, epidemiology and high-quality data to optimize the allocation of livers for transplantation

7.项目总结/摘要 在美国，对适用于肝移植（LT）的肝脏的需求超过了供应。2017年，虽然 7，500名患者接受了LT，超过2，500人在死亡或变得“病重”后从等待名单中删除 移植”。这些数字并不能解释这样一个事实，即6名终末期肝病患者中有5名 符合LT标准的ESLD永远不会被列入等待名单。因此，患者不可避免地在等待名单上死亡，或死亡 从未被列入候补名单一个最佳的移植系统应该通过实现三个价值来分配器官：1） 平等-所有人都能公平地获得移植; 2）优先考虑情况最差的人; 3）效用-最大限度地实现“预期” 患者群体的“总体良好净量”。利用终末期肝病模型 （MELD）评分来优先考虑患者，目前的系统主要侧重于紧急情况， 目前病情最严重的是“最差的情况”。“这将最大限度地减少等待名单的死亡率。然而，这并没有考虑公平准入 对于不同的患者亚组（即，传统上处于不利地位的群体），或公用事业（例如，最大 生存或LT的净效益）。因此，许多LT被分配给预期寿命有限的患者， LT，而等待名单上的其他人同样强烈的道德要求LT死亡。改善LT分配需要 新的数据和创新的范式，寻求优化使用稀缺资源，可以延长生命， >15年。这需要的不仅仅是一个修订的终末期肝病模型（MELD）评分（ 用于等待列表优先化的分数）。首先，为了估计没有LT的生存率， 考虑到平均等待时间为<180 days and >50%， 患者在上市后一年内接受LT。其次，必须建立长期的LT后风险模型， 考虑到对长期生存具有高度歧视性的因素（例如，年龄，急性与慢性肾脏 失败）。第三，肝细胞癌（HCC）患者的分配政策必须考虑相对 考虑到优先考虑HCC的非预期后果，LT相对于其他HCC治疗的长期获益 失代偿期肝硬化患者的比例。这项工作将是未来设计的基础， 改善分配系统，大幅提高ESLD患者的生存率。我们将利用 来自美国最大的肝脏护理提供商退伍军人健康管理局的强大而详细的数据， 联合器官共享网络（UNOS），以实现这些目标：1）制定及时更新的，长期的预 LT生存模型解释急性和慢性肾脏疾病; 2）开发一个时间更新，长期 使用HCC患者的肿瘤、实验室数据和临床变量的LT前生存模型; 3）开发一个 长期LT后风险模型，包括人口统计学，临床和实验室变量; 4）建立分配 模拟，量化在几种不同的分配下获得的生命年差异的大小 计划在一系列的时间范围内，重点是NIH指定的健康差距人群。

7. Project Summary/Abstract The demand for livers suitable for a liver transplant (LT) in the US outnumbers the supply. In 2017, although 7,500 patients received a LT, more than 2,500 were removed from the waitlist after dying or becoming “too sick to transplant.” These numbers don't account for the fact that 5 out of 6 patients with end-stage liver disease (ESLD) meeting LT criteria are never waitlisted. Consequently, patients inevitably die on a waiting list, or die never having been waitlisted. An optimal transplant system should allocate organs by realizing three values: 1) equality—fair access to transplant for all; 2) priority to the worst off, and 3) utility—maximizing the `expected net amount of overall good' for the population of patients. By using the Model for End Stage Liver Disease (MELD) score to prioritize patients, the current system focuses predominantly on urgency, favoring the currently sickest as the `worst off.' This minimizes waitlist mortality. However, this does not consider fair access (equality) for different patient subgroups (i.e., traditionally disadvantaged groups), or utility (e.g., maximizing survival or net benefit of LT). As a result, many LTs are allocated to patients with limited life expectancy after LT, while others on the waitlist with equally strong ethical claims to LT die. Improving LT allocation requires new data and an innovative paradigm that seeks to optimize use of a scarce resource that can extend life by >15 years. This requires more than simply a revised Model for End-Stage Liver Disease (MELD) score (the score used for waitlist prioritization). First, to estimate survival without a LT, robust longitudinal data are needed that extend beyond waitlist registry data, given that median waitlist time is <180 days and >50% patients receive a LT within one year of listing. Second, long-term post-LT risk models must be developed, to account for factors that are highly discriminatory for long-term survival (e.g., age, acute vs chronic kidney failure). Third, allocation policies for patients with hepatocellular carcinoma (HCC) must consider the relative long-term benefit of LT vs. other HCC treatments given the unintended consequences of prioritizing HCC patients over those with decompensated cirrhosis. This work will be fundamental to the future design of an improved allocation system that drives major increases in survival for patients with ESLD. We will leverage powerful and detailed data from the Veterans Health Administration, the largest US provider of liver care, and United Network for Organ Sharing (UNOS), to address these aims: 1) to develop time-updated, long-term pre- LT survival models accounting for acute and chronic kidney disease; 2) to develop a time-updated, long-term pre-LT survival model using tumor, laboratory data, and clinical variables for patients with HCC; 3) to develop a long-term post-LT risk model incorporating demographic, clinical, and laboratory variables; 4) to build allocation simulations that quantify the magnitude of difference in life-years gained under several different allocation schemes across a range of time horizons, with a focus on NIH-designated health disparity populations.