Model for end-stage liver disease (MELD) exception for ascites


Ascites is a common clinical finding in liver transplant (LT) candidates, accounting for 5% of all petitions and 20% of nonstandard petitions for Model for End-Stage Liver Disease (MELD) exception.1 Refractory ascites, as defined by the International Ascites Club,2 occurs in 5 to 10% of patients with portal hypertension and has a 1-year mortality rate of approximately 50%.3 In the transition from Child-Turcotte-Pugh score–based liver allocation to MELD-based liver allocation, ascites, which is a well-recognized predictor of mortality in liver disease,4 was removed as a priority measure for LT.5 In contrast to the initial MELD validation studies that reported little or no additional risk of ascites after controlling for MELD,6, 7 more recent studies have reported that the MELD score may not adequately predict the risk of waiting list mortality as a result of persistent ascites.8, 9 Heuman and colleagues8 found that the excess waiting list mortality risk associated with persistent ascites was equivalent to 4.5 MELD score points and was most evident in, but not limited to, patients with MELD scores <21.8

The current goal is to reliably identify a subgroup of patients with severe or complicated ascites that should be compensated with additional MELD score points. Under the Child-Turcotte-Pugh–based liver allocation system, subjective measurements of the severity of ascites were frequently misused by listing centers. When considering MELD exceptions for ascites, every effort must be made to use objective and reliable evidence-based data. One of the primary strengths of the MELD-based allocation system is the sole reliance on objective criteria.

Standardized definitions for grading the severity of uncomplicated and refractory ascites has been agreed on by the International Ascites Club.2 Currently however, little objective, quality information based on these definitions exists for predicting which patients with ascites will experience higher waiting list mortality. The International Ascites Club defines refractory ascites as the requirement for intensive diuretic therapy (spironolactone 400 mg/day and furosemide 160 mg/day) for at least 1 week) while salt restricted (serum sodium <90 mM/day), with an inadequate response (<0.8-kg weight loss or positive sodium balance), recurrence of ascites (grade 2 or 3) within 4 weeks of initial mobilization, and/or the occurrence of severe diuretic-induced complications (i.e., encephalopathy, hyponatremia, hypokalemia, hyperkalemia, or renal insufficiency).2 In addition, complications of ascites such as spontaneous bacterial peritonitis10, 11 and hepatic hydrothorax12 may be used to identify a subgroup of patients with a higher mortality risk, but whether these complications represent a mortality risk beyond that measured by MELD is unclear.6, 7 Although the transjugular interahepatic portosystemic shunt (TIPS) is an effective intervention for the management of refractory ascites in selected patients,13, 14 it provides no survival advantage over repeated paracenteses.13, 14 However, it seems intuitive that patients with refractory ascites occurring despite a functioning TIPS represent a higher-risk group and may warrant consideration for a MELD exception. Although any or all of these parameters could be considered potential MELD exceptions, all are subjective in nature, and there is potential for misuse.

The serum sodium concentration shows promise as an objective surrogate marker for increased mortality risk in patients with ascites on the LT waiting list. Hyponatremia has a well-recognized association with portal hypertension,15, 16 ascites,10, 17, 18 and mortality in the patient with cirrhosis.4, 10, 19 Three retrospective single-center studies8, 20, 21 and a prospective multicenter study22 have shown hyponatremia to be a strong predictor of waiting list mortality, with a 5 to 7% increase in waiting list mortality for each 1 mEq/L decrease in serum sodium concentration, after controlling for the MELD score.20, 21 The association between hyponatremia and waiting list mortality observed in patients with and without ascites8, 20–22 may be stronger if it occurs in the absence of precipitating events such as spontaneous bacterial peritonitis or variceal bleeding,17 and it may be weaker or negligible at higher MELD scores (i.e., MELD score >21).8, 22 A recent multicenter study developed, but has not validated, a mortality risk model, MELD-Na, that incorporates the serum sodium concentration into the MELD score.22 The formula is: MELD-Na score = [MELD score + 1.59 × (135-Na)], where the allowed range of the serum sodium concentration is 120 to 135 mEq/L.

The United Network for Organ Sharing has required documentation of serum sodium levels for new registrants on the LT waiting list since November 2004 and models, such as the MELD-Na, should be validated by using this data. If validated, such objective models could be considered as guides for the amount of priority, in terms of exceptional MELD, to assign candidates who meet criteria for refractory or complicated ascites, or more broadly to candidates with and without ascites.


There is limited evidence that candidates with persistent ascites have a higher waiting list mortality than that predicted by the MELD score alone. Candidates meeting the proposed criteria for refractory or complicated ascites (Table 1) may be at even higher risk of waiting list mortality, although no studies have investigated this relationship. There is evidence that a decreasing serum sodium level predicts waiting list mortality even after controlling for the MELD score in candidates with and without ascites. The incremental improvement in predicting waiting list mortality as a result of the addition of either persistent ascites or serum sodium to MELD may be greater in candidates with low MELD scores.

Table 1. Refractory or Complicated Ascites
• International Ascites Club grade 3 ascites (massive abdominal distension)2 with current imaging documenting the presence and severity of ascites and ≥2 of the following additional criteria:
 ○ ≥3 therapeutic paracenteses (≥2 L each) in last 60 days.
 ○ ≥2 episodes of spontaneous bacterial peritonitis with supporting documentation (ascitic polymorphonuclear granulocyte count ≥250 or positive culture).
 ○ Previous transjugular interahepatic portosystemic shunt.
 ○ Ascites unresponsive to maximum doses of diuretics (i.e., spironolactone 400 mg/day and furosemide 160 mg or equivalent).
 ○ ≥2 therapeutic thoracenteses.
 ○ Serum sodium ≤125 mEq/L.

LT, liver transplantation; MELD, Model for End-Stage Liver Disease; TIPS, transjugular interahepatic portosystemic shunt.


Additional MELD points should not be automatically given to LT candidates with ascites. Candidates with refractory or complicated ascites, as defined in Table 1, could be considered for additional priority for LT on a case-by-case basis once a national review board is established. Any additional priority should be based on the median MELD score for a given blood type in the given Organ Procurement Organization–designated service area. The serum sodium concentration shows promise as an objective surrogate marker for the excess waiting list mortality associated with portal hypertensive ascites, and further investigation is encouraged.

Documentation submitted for case review and collected by United Network for Organ Sharing should include the following:

  • Clinical and imaging evidence of ascites after sodium restriction.

  • Number and volume of therapeutic paracenteses in the last 3 months.

  • Number of episodes of spontaneous bacterial peritonitis and any available supporting diagnostic evidence (polymorphonuclear granulocyte count, culture report).

  • Any evidence of contraindication to TIPS, or ascites with a patent TIPS.

  • Current and maximum (in the last 3 months) doses of diuretics.

  • Number of days in hospital for ascites management in the last 2 months.

  • Evidence of hepatic hydrothorax (imaging studies, number and volume of thoracenteses).

  • Most recent serum sodium concentration and MELD score.