Jacek B. Cywinski, M.D., Edward J. Mascha, Ph.D., Brian M. Parker, M.D.
麻醉外科，克利夫兰医院，克利夫兰，美国俄亥俄州。

简介：终末期肝病模型（MELD）评分标准已经被广泛用于等待肝移植的肝病患者，这种方法能为处于移植等待期且近期死亡风险最高的患者优先提供脏器。曾有人建议将血清钠水平纳入到MELD评分标准中，理由是患有低钠血症的终末期肝病（ESLD）患者的病情更为严重。本研究的目的是在预测同种异体原位肝移植患者的短期疗效和长期疗效时，评价和比较术前“钠-MELD”分数的预后值和传统MELD分数的预后值。

方法：回顾性收集的数据主要来自名为“统一移植数据库”的单一机构。我们筛选了2000年1月至2005年12月期间进行原位肝移植手术的250名患者并记录了移植前的血钠水平。血钠水平低于135 mEq/dl (n=106)的所有患者均参与了此项研究。随后根据Ruf等人提出的算法[钠-MELD = MELD + 1.59 * (135 –血清钠)]，计算出“钠-MELD”分数。移植术后肝功能（LFT）标准检测收集了天门冬氨酸转氨酶（AST）、丙氨酸转氨酶（ALT）和总胆红素水平，同时还有移植术后血清肌酸酐水平（SCR）、手术中血液制品需求量和受体的存活率等。
我们对每个MELD分数（MELD和钠-MELD）、移植后1至3天（POD 1-3）的LFT和SCR峰值、血液制品用量之间的皮尔森相关系数进行了分析，并采用标准方法比较了它们之间的相关性。对两种MELD分数POD 1至3天LFT的变化概率进行了对比，此外，还应用回归分析法c-statistic比较了二者对受体术后30天死亡率的预测能力。同样的，在Cox存活模型c-index的基础上比较了以上分数，用以预测长期死亡率。每种假说的显著性水平为0.05。

结果：MELD和钠-MELD二者都与ALT最高值（分别为P=.03 and .008）和SCR（P=.001, <.001）呈负相关，同时，与RBC（二者P值均小于.001）和FFP（P=.002, .04）呈正比关系。然而，两种MELD分数对任何一种检测结果的预测能力并无差别，包括LFT的变化概率（AST P=.54, ALT P=.42, BLB P=0.13）、LFT最高值、血液制品和SCR等。在30天死亡率（c-statistic=0.66 vs 0.60）或更长时间死亡率方面，二者的预测能力都不强。

结论：移植前钠-MELD 分数和传统MELD分数对原位肝移植患者术后疗效的预测能力都很差。原位肝移植患者的术后疗效很有可能取决于多种因素，这些因素并未全部包括在本项研究的评分系统中。

参考文献：
1. Ruf et al.,Liver Transpl. 2005 Mar;11(3):336-43.
2. Londono et al.,Gastroenterology. 2006 Apr;130(4):1135-43
3. Onaca NN, et al.,Liver Transpl. 2003 Feb;9(2):117-23.

Prediction of Outcome by Pretransplant Sodium-MELD and MELD Scores after Liver Transplantation
Jacek B. Cywinski, M.D., Edward J. Mascha, Ph.D., Brian M. Parker, M.D.
General Anesthesiology, Cleveland Clinic, Cleveland, Ohio

Introduction: The model end-stage liver disease (MELD) score has been widely used to prioritize patients with hepatic disease for liver transplantation. This approach was based on prioritizing organs for those patients with the highest short term mortality risk while awaiting transplant. It has been suggested that incorporating the serum sodium level into the MELD score would be appropriate, since hyponatremia in the patients with ESLD indicates more severe disease. The aim of this study was to evaluate and compare the prognostic value of the preoperative “sodium MELD” score compared to the traditional MELD score for both short and long term outcomes of allograft recipients after OLT.

Methods: Retrospective data collection was predominantly from the existing Unified Transplantation Database of a single institution. 250 patients that underwent OLT from January 2000 to December 2005 were screened for pretransplant serum sodium level. Any patient with a serum sodium level <135 mEq/dl (n=106) was included in the study. For each patient a “sodium MELD” score was then calculated according to Ruf et al1 [sodium MELD = MELD + 1.59 * (135 – serum sodium)]. Standard postoperative liver function tests (LFT) including aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total bilirubin were collected as well as peak postoperative serum creatinine level (SCR), intraoperative blood products requirements and recipient survival. Pearson's correlations between each MELD score (MELD and sodium MELD) and peak LFT and SCR on postoperative day (POD) 1-POD 3, as well as blood product usage was assessed, and the correlations compared using standard methods. Both MELD scores were also compared with respect to correlations with the rate of LFT change from POD 1-POD 3. Also, both scores were compared for their ability to predict recipient 30-day mortality using the c-statistic from logistic regression.Similarly, these scores were compared to predict long-term mortality based on the c-index from Cox survival modeling. The significance level for each hypothesis was 0.05.

Results: Both MELD and sodium MELD were correlated negatively with maximum ALT (P=.03 and .008, respectively) and SCR(P=.001, <.001) and positively with RBC (both P<.001) and FFP(P=.002, .04). However, the MELD scores did not differ in their ability to predict any of the measured outcomes, including rate of LFT change (AST P=.54, ALT P=.42, BLB P=0.13), maximum LFTs, blood products, and SCR. Neither MELD score was a strong predictor of 30-day (c-statistic=0.66 vs 0.60) or long-term mortality.
Conclusion：Both the pretransplant sodium MELD score and traditional MELD score were rather poor predictors of postoperative outcomes in patients undergoing OLT. Outcomes in OLT patients most likely depend on multiple factors not fully accounted for in the scoring systems evaluated in this study.

References:
1. Ruf et al.,Liver Transpl. 2005 Mar;11(3):336-43.
2. Londono et al.,Gastroenterology. 2006 Apr;130(4):1135-43
3. Onaca NN, et al.,Liver Transpl. 2003 Feb;9(2):117-23.

[责任编辑:刘 聪]