Eg J Anaesth  2005  ,21 :191-9

Abstract: 
The philosophy underlying organ allocation in liver transplantation is to prioritize recipients with a substantial risk of death. This requires a disease severity index that grades liver transplantation urgency for patients on the waiting list.  We with discuss common disease severity indices including the Child-Pugh scoring system, the United Network of Organ or Online sharing (UNOS) allocation scheme, and the models for end-stage liver disease in adults MELD) and pediatrics (PELD).  We will draw attention to a suggested new model for end-stage liver disease (NMELD) inviting future research to elucidate its formula.

Introduction
The philosophy underlying organ allocation in liver transplantation is to prioritize recipients with a substantial risk of death.  This requires a disease severity index that grades liver transplantation urgency for patients on the waiting list. (1)
The Child-Pugh classification is the widely used model to determine the prognosis of surgical patients with liver disease (table 1) (2). Although its original purpose was to assess operative risk in patients undergoing postosystemic shunts, it has been also used to stratify patients on the waiting list for liver transplantation.

Table (1): Grading of the Child-Pugh scoring system

Grades: mild: 5-6 points, moderate: 7-9 points, severe: 10-15 points.

As a disease severity index, the Child-Pugh system, has a number of limitations:

• Limited discriminatory ability:

It has only 8 points of difference between the least (8 points) and most (15 points) sick transplant candidates.  All patients with markedly abnormal scoring points (10-15), are C-classed equally (ceiling effect).  This creates many ties (i.e. transplant candidates with the same score) and necessitates an emphasis on waiting time as a tie-breaker. 

• Variability in measuring objective parameters:

Bilirubin, albumin and PT are objective parameters that may have variable evaluation by different laboratories.

• Variability in assessing subjective parameters:

Encephalopathy and ascites are subjective parameters that are difficult to standardize.  Although the degree of ascites may be based on ultrasound examination, no standard definition has been given to refractory encephalopathy.
Before 1997, the initial allocation scheme of the United Network of Organ or Online Sharing (UNOS) stratified liver transplantation urgency according to hospital status and waiting time.  In patients with chronic liver disease, highest status was awarded to the patient in an intensive care unit (ICU), next priority was given to the hospitalized non-ICU patient, and the ambulatory (at-home) patient was given lowest priority. This system of broad categorization resulted in large numbers of patients in each group, all waiting for donated livers. In this situation, time on the waiting list served as a tie-breaker for allocation of the organs.  So, early listing of patients with well-compensated liver disease, gave them an advantage and became a common practice (3).  It has been recently shown that recipients who were outpatients before transplantation have longer survival than those transplanted from the hospital or ICU. (4)
The UNOS, then in 1997, adopted a minimal Child-Pugh score equal to or greater than 7 (floor effect) as essential for patient listing where accumulated waiting time, not the disease severity, continued to determine organ allocation and was used also, as a tie-breaker between competing candidates with the same UNOS status (5).  It was, then, obvious that the Child Pugh system, which categorized patients with cirrhosis into three groups (UNOS 2A, 2B and 3) was not an optimal system for ranking patients at risk of dying without liver transplantation.
In 1999, in response to the directions of the Department of Health and Human Services in USA, UNOS proposed that a better scheme for donating livers should be developed by the transplantation community.

The MELD scoring system
In 2001, the article of Kamath et al (6) in Hepatology, established the value of a model for end-stage liver disease (MELD) in predicting survival in four independent adult populations with advanced liver disease including patients hospitalized for decompensated liver disease, ambulatory patients with non-cholestatic cirrhosis, ambulatory patients with primary cirrhosis, and historical unselected cirrhotic patients from the 1980’s.  The authors found that the model was useful for grading the risk of death at three time intervals (one week, three months and one year) to the decision to undertake liver transplantation. They reported that the addition of parameters associated with decompensation, such as ascites, encephalopathy, variceal bleeding and spontaneous bacterial peritonitis, did not substantially improve the accuracy of the model.  This indicated that complications are not independent predictors of survival.
In February 27, 2002, the MELD as a score was instituted to replace UNOS status 2A, 2B, and 3 for organ allocation in adult patients.  UNOS status 1 patients with acute liver failure and life expectancy less than seven days remain in place and are not affected by MELD (6).  The MELD score denotes how urgently a patient needs a liver transplant within the next three months.  It has 34 points of difference between the least (6 points) and the most (40 points) sick transplant candidates.  Such increased number of categories (34 versus 4 for UNOS) of severity of illness, lacking floor and ceiling effects, was intended to reduce the impact of waiting time on organ allocation.
The MELD system included bilirubin as a measure of the ability of the liver to break-down haemoglobin and to effectively excrete bile, creatinine as a measure of kidney function, PT calculated as the international normalized ratio (INR) to measure the blood ability to clot, and the etiology of liver disease (6).
The MELD score was originally designed to predict 3-month mortality after transjuglar intrahepatic portosystemic shunt surgery (TIPPS), done to control an episode of variceal bleeding (7).  It was considered a TIPSS-MELD score (TMS), and for ease of use, the score was multiplied by 10 and the rounded to the nearest integer.  The original MELD formula was: 3.8 log (bilirubin mg/dl) + 9.6 log (creatinine mg/dl) + 11.2 log (INR) + 6.4 etiology of liver disease (0 if cholestatic or alcoholic, 1 if otherwise).  It was then slightly modified to act as a general predictor of survival in hepatic patient predictor of survival in hepatic patients by excluding the etiology of liver disease as a parameter (6).  For adults who have had dialysis twice within the last week, the creatinine value will be automatically set to 4 mg/dl.
Merion et al (8) have shown that MELD was superior to Child-Pugh classification as well as the UNOS scheme, and they considered the change from Child-Pugh score to MELD score a step in the right direction.  Moreover, they showed that a patients MELD score may go up or down over-time depending on the status of his/ her liver disease, while being on the waiting list.  Using a change in MELD score rather than a single MELD score is a further step to maximize benefits of liver transplantation.  For a patient with a stable MELD score for 30 days (MELD = 0), the mortality was lower than for a patient with increasing MELD score (positive MELD), but higher than for a patient with decreasing score (negative MELD).
An increase in MELD score may be due to reversible factors as dehydration or sepsis, while a decrease way follow correction of such factors.  A rise in creatinine component of MELD may be of pre-renal origin, simply in the form of dehydration.  A rise in both creatinine and bilirubin components may take place in a septic state.  In addition, the PT representing the INR component may be prolonged in an ICU patient who is on enteral nutrition and is on systemic antibiotics, simply causing vitamin k deficiency. So long as transient and reversible factors are ruled-out or corrected, a persistencly increasing MELD score over-time is related to worsening of liver function.  A rapidly increasing MELD score is likely to be associated with increasing mortality (4, 8).  If a patient’s MELD score increases over-time, only the waiting time at the higher level will count.  An increased rate of death was observed in recipients of the organ with a MELD score higher than 36 and more urgent UNOS status 2A (9).
When the MELD system was originally developed, scores were taken in haemodynamicaly stable patients after assuring that all reversible factors were corrected. Thus, creatinine levels were recorded after the patients were clinically isovoelemic.  Patients with intrinsic renal disease or hepatic carcinoma were excluded. In patients with sepsis, scores were recorded after infection had cleared out and PT, creatinine level, and bilirubin level were stable for a week. Therefore, it became clear that MELD score may not be an accurate indicator of mortality in patients with cirrhosis who have intrinsic renal disease, are dehydrated, or who have liver carcinoma.  Merion et al (8) hold the opinion that the increase in MELD may be a result of multiorgan failure. In other words, an increase in MELD may be, in fact, a part of the event rather than a predictor of death, which mean that one should be certain that reversible factors have been ruled out and the increase in MELD is not a preterminal event.

MELD versus Child-Pugh scoring systems
Two variables that correlate with severity of liver disease; bilirubin and PT; are key components in the calculation of both the MELD and Child-Pugh scoring systems. Historically, these two variables composed the modified discriminant function (DF) score to predict mortality in patients with hepatic dysfunction (10).
DF = 4.6 (patients PT-Control PT) + total bilirubin (mg/ dl)
Based on DF, a study on patients with severe alcoholic hepatitis showed a 30 day mortality greater than 50% (10).  This study validated the use of the MELD score as an alternative to DF score in predicting short-term mortality in patients with alcoholic hepatitis. (11)
If encephlapoathy and ascites, as subjective parameters, are eliminated from the Child-Pugh score, the remaining difference between MELD and Child-Pugh scores is creatinine in MELD versus albumin in Child-Pugh (table 2).  So, the magnitude of the superiority of MELD score over Child-Pugh score is limited to populations at high risk of renal failure (12).  
Table 2: Model for end-stage liver disease
(MELD) versus Child-Pugh parameters

Renal failure complicating chronic liver disease may be caused by hypovolaemia, hepato-renal syndrome or intrinsic renal disease.  Changes in nutrition, volume status, and the use of diuretics; all significantly influence creatinine level.  Volume loading against prerenal volume depletion, frequently treats transient increases in creatinine. In such group of patients, the accuracy of MELD and Child-Pugh scores to predict outcome should be similar with equivalent accuracy in the absence of renal failure (6).  In a study, allowing entry of renal failure patients, the predictive accuracy of MELD proved to be superior to Child-Pugh (13).  Thus, the main advantage of MELD over Child-Pugh score may be in waiting list patients who have the probability to develop renal failure.  While day-to-day fluctuations in bilirubin and PT are less dramatic, the dependence of MELD score on the more dramatic day-to-day fluctuating creatinine levels means that the score also fluctuates.
The UNOS has mandated updates of the MELD score to reduce the risk of the effect of transient fluctuations in creatinine in determining organ allocation (12). The MELD score is recalculated every 7 days for patients with a score of 25 or more, every 30 days for those of 19-24, every 90 days for those of 11-18, and every year for those equal to or less than 10 scoring points.
It should be remembered that the MELD and Child-Pugh scores have 8 and 34 points range between the least and most sick transplant candidates respectively.  The MELD score increases as its three constituent objective parameters deteriorate, whereas individual scoring elements in the Child-Pugh score remain fixed (ceiling effect) once a defined threshold has been reached, i.e. all patients with 10-15 scoring points are C-classed equally.
It has been shown that the MELD score is superior to the Child-Pugh score in estimating pre-orthotopic liver transplant (OLT) disease severity In UNOS status 2A listed patients, and thus, may help to risk stratify status 2A or decompensated status 2B OLT candidates and to optimize the timing of OLT (14). 

MELD versus modified Pugh scoring systems
A modified-Pugh scoring system has been recently introduced by Moemen et al (15) to categorize surgical patients with chronic liver disease (table 3).  It has been validated in a prospective study including a relatively large number of patients and proved to be more accurate and specific than the original Child-Pugh scoring system (16).

Table (3) Grading of the Modified-Pugh Scoring System

Grades: mild: 9-10 points, moderate: 11-14 points, severe: 15-27 points.

If encephalopathy and ascites (as a subjective parameters) and albumin (a parameter excluded in MELD) are eliminated from Moemen scoring system, a suggested New Model for End-Stage Liver Disease (NMELD), as a disease severity index includes 3 more parameters (serum sodium, white blood count and an oxygenation parameter) compared to MELD (table 4).  The serum sodium parameter identifies the contribution of hyponatraemia in hepatic encephalopathy and the diagnosis of the hepatoremanl syndrome.  The white blood count parameter identifies sepsis complicating hepatic disease.  Arterial/ alveolar oxygen tension ratio [P(a/A) O2], as an oxygenation parameter, explores the respiratory-ventilatory component of hepatic failure.  A planned future research may elucidate the NMELD formula for prognostic outcome in non-transplant patients as well as for organ allocation in liver transplantation.

Table 4: NMELD versus Modified-Pugh parameters

NMELD: New model for end-stage liver disease
PT:              Prothrombin time
INR:            international normalized ratio
P(a/A) O2:      arterial/ alveolar oxygen tension ratio

HCC-adjusted MELD
Hepato cellular carcinoma (HCC) is the most common primary liver malignancy, and 50-80% of patients with HCC have associated cirrhosis(17, 18).  About 10-20% of cirrhotic patients of any cause will develop HCC (18).  The incidence of HCC as a primary disease or as a complication of cirrhosis continues to rise, and may need either hepatic resection or liver transplantation.
For patients with HCC, prolonged waiting time often leads to tumour growth exceeding acceptable criteria for OLT, in terms of the size and number of tumour nodules (dropout), often equivalent to death within 6 months. (19)
Since patients with HCC may have preserved hepatic function, and consequently a MELD score predicting a low risk of death from the liver disease alone, an adjustment would need to be made to allow for the anticipated risk of dropout and death due to HCC.
The risk of dropout at one year for patients with HCC approached 40%, prompting recent changes in organ allocation policy designed to better serve patients with HCC (19).
The Milan criteria (20) adopted by the UNOS in 1998, has proposed OLT as a treatment for unresectable intrahepatic HCC.  Under the organ allocation system implemented by the UNOS between January 1998 and February 2002, patients with HCC wwere eligible for status 2B listing for OLT (21).  The one year survival rate for OLT in patients with HCC was around 80, similar to that in patients without HCC.  Corresponding balues for 5 year survival rate were 46.7% versus 70.6% (22).
The current UNOS staging classification for HCC, implemented on February 2002, has incorporated Milan criteria (20) into stages T1 and T2 (table 5).

Patients with T1 HCC were given a MELD score of 24, corresponding to an expected 15% 3-month mortality.  These with T2 HCC were assigned a MELD score of 29, with an expected 30% 3-month mortality. These patients were given an increase in the MELD score, corresponding to 10% increase in mortality for every 3 month on the waiting list (23).  So, the initial HCC adjusted MELD scheme was designed to allow candidates with stages T1 and T2 a priority MELD score beyond the degree of hepatic decompensation, so as to limit the dropout rate of these patients when placed on the waiting list for OLT. (24)
On February 2003, the UNOS implemented a modification, in which the initial MELD score for HCC, was reduced from 29 to 24 points for patients with T2 HCC, and from 24 to 20 points for these with T1 lesion (23).  The decision to downgrade the initial MELD score may be further justified by the apparent increase in mortality for these without HCC, but a MELD score between 25 and 30 based on preliminary UNOS data (25).  By giving patients too high a priority score UNOS data showed 90% OLT rate within 3 months on the waiting list (25).  Access to timely OLT for patients with HCC has significantly improved in the first year between February 2002 and February 2003, when compared to the pre-MELD era (26).  Refinements adopted since February 2003 proved to be more consistent with the natural history of tumour progression and the risk of dropout (19, 23).
In the latest proposal by UNOS, patients with T1 HCC are no longer eligible for MELD priority listing, but should be monitored closely with serial imaging studies to evaluate for evidence of tumour progression and appropriate priority listing during follow-up.  This policy takes in consideration to protect T2 HCC patients at higher risk of dropout. The current UNOS selection criteria for HCC based on tumour burden has proved to provide adequate discriminatory power in predicting outcome of OLT (27).
Further research need to continue to improve the accuracy and prediction of this clinical model because it is apparent that a patient with a MELD score of 20 may have more severe disease and a higher risk of death than a patient with a MELD score of 24.
The ABO blood group was found to be a significant determinant of the probability of OLT under MELD, favouring non-O blood groups in comparison with blood group O (27).  Patients with blood group O, generally, have a longer median waiting time and higher mortality rate, irrespective of their HCC status (28, 29).  This is a significant disadvantage for these patients in terms of OLT under MELD.  According to current UNOS guidelines, blood group O donors do not cross the blood group barrier for OLT, with the exception that an O liver may go to a blood group B patient with fulminant hepatic failure or graft non-functioning listed status 1 patient for OLT (27, 30).

The PELD Scoring System
A pediatric end-stage liver disease (PELD) score, currently replaces UNOS statuses for candidates below 18 years of age (1).  It is similar to the MELD score, with a different statistical formula to recognize the needs for children growth and development.  The items are: bilirubin (mg/ dl) which measures how effectively the liver excretes bile, albumin (g/L) which measure the liver ability to maintain nutrition, INR which measures the liver ability make blood clotting factors, growth failure (based on gender, height and weight) together with age at listing specially whether the child is less than one year.  The PELD score accurately predicts three month probability of waiting list death for children with chronic liver disease.

Allocation systems: Overview
Living related donor liver transplantation is being increasingly adopted in USA and Europe because of the long waiting lists for cadaveric organs (31).  This is also helpful to expand organ availability in areas where the concept of brain death is mostly not adopted.
For OLT, the MELD and PELD scores are the allocation systems that now identify patients at highest risk of early death, while being on the waiting list. These patients are, thus, given a chance of appearing at the top of the list when they need to.
The MELD and PELD formulas are objectable and verifiable yielding consistent results whenever the scores are calculated. They use specific programmable calculators with screens to collect data of the scores.  The complexity of the formulas has been considered, by some authors, as a barrier for their bed-side clinical use (11, 32).  
Kamath et al (6), established the MELD, mostly in patients with advanced liver disease.  Overall 3 month morality based on the data used in the first report ranged between 2% and 21%.  Thus, it is mandatory to warrant further validation of the MELD score in patient with early stage liver disease.
It may be useful to design the formula of a NMELD system which modifies MELD by adding three more variables; serum sodium, wbc and an oxygenation parameter. Again it seems mandatory to search for a method for accommodation of special cases as patients with metabolic disorders and those with symptoms as intractable purities.
Overall, application of MELD and PELD models provides an evidence-based approach on which to base further refinements (33).  This is because these models prioritize recipients at early risk of death on the waiting list, but they do not identify patients who benefit the most from liver transplantation. So, although these models are here, the question is whether they are here to stay.

References 
Wiesner RH, Mediarmid SV, Kamath PS, et al.  MELD and PELD: Application of survival models to liver allocation. Liver Transpl 2001; 567-580.
Pugh RNH, Murray LIM, Dawson JL, et al.  Transaction of the oesophagus for bleeding oesophageal varices. Br J Surg 1973; 60 646-649.
Forman LM, Lucey MR. predicting the prognosis of chronic liver disease: an evolution from Child to MELD.  Hepatology 2001; 33: 473-475.
Mark SR, Derek CA, Cindy LB, et al.  Survival after liver transplantation in United States: a disease-specific analysis of the United Network of Organ sharing database.  Liver Transpl 2004; 10: 886-897.
Lucey MR, Brown KA, Everson GT, et al.  Minimal criteria for placement of adults on liver transplant waiting list: a report of a national conference organized by the American Association for the study of Liver Disease.  Liver Transpl Surg 1997; 3: 628-637.
Kamath PS, Wiesner RH, Malinchoc M, et al.  A model to predict survival in patients with end-stage liver disease. Hepatology 2001; 33: 464-470.
Malinchoc M, Kamath PS, Gordon FD, et al.  A model to predict poor survival in patients undergoing transjuglar intrahepatic portosystemic shunts. Hepatology 2000; 31: 846-871.
Merion RM, Wolfe RA, Dykstra DM, et al.  Longitudinal assessment of mortality risk among candidates for liver transplantation.  Liver Transpl 2003; 9: 12-18.
Sammy S, Victor W, Ayman BI, et al.  MELD score predicts one-year patient survival post-orthotopic liver transplantation.  Liver Tranplant vol 9, issue 5: 473-476.  Published online 30 December 2003.
Maddrey WC, Boitnott JK, Bedine MS, et al.  Corticosteroid therapy of alcoholic hepatitis.  Gastroenterology 1978; 75: 196-199.
Sheth M, Riggs M, Patel T.  Utility of the Mayo End-stage Liver Disease (MELD) score in assessing prognosis of patients with alcoholic hepatitis.  BMC Gastroenterology 2002; 2: 2-13.
Everson GT. MELD: the answer to just more questions? Editorial. Gastroenterology 2003; 124: 1-6.
Weisner R, Edwards E, freeman R, et al, and the United Network for Organ Sharing.  Liver Disease Severity Score Committee. The model for end-stage liver disease (MELD) and allocation for donor livers.  Gastroenterology 2003; 124: 91-96.
Robert SB, Shiva K, Mark WR, et al.  Model for end-stage liver disease and Child-Turcott-Pugh score as predictors of pre-transplantation disease severity, post-transplantation outcome, and resource utilization in UNOS status 2A patients.  Liver Transplant vol 8, issue 3: 278-284, Published online 30 December 2003.
Moemen ME, Gaafar TY, Farag WA; et al. Prognostic categorization in cirrhotic patients undergoing abdominal surgery: a randomized trial. EgJA 2004; 20: 7-14.
Webster NR. Scoring systems and surgery in cirrhotic patients. Editorial. EgJA 2004; 20: 1-3.
Moser MAJ, Kneteman NM, Minuk GY.  Review: research toward safer resection of the cirrhotic liver.  HPB Surg 2001; 11: 285-297.
Abdel-Wahab A, El-Eneim AA, Abou-Zeid M, et al.  Hepatocellular carcinoma in Mansoura-Egypt: Experience of 385 patients at a single center.  Hepato-Gastroenterol 2000; 47: 663-668.
Yao FY, Bass NM, Nikolai B, et al.  A follow-up analysis of the pattern and predictors of dropout from the waiting list for liver transplantation among patients with hepatocellular carcinoma : implications for the current organ allocation policy.  Liver Transpl 2003; 9: 684-692.
Mazzaferro V, Regalia E, Doci R, et al.  Liver transplantation for the treatment of small hepatocellular carcinoma in patients with cirrhosis.  N Engl J Med 1996; 334 693-699.
Keeffe EB.  Summary of guidelines on organ allocation and patient listing for liver transplantation. Liver Tranpl Surg 1998; 4(Suppl1): S 108-S114.
Jessica YL, Andrew XL, Fredric DG, et al.  Liver transplantation outcomes for early end-stage hepatocellular carcinoma: results of a multicenter study.  Liver Transpl. Vol 10, issue 11: 1343-1354.  Published online 20 October 2004.
United Network for Organ Sharing. Policy 3.6. Available at:http://www.unos.org. Accesed September 19, 2003.
Pratima SH, Vijayan B, Jose LH, et al.  Liver transplantation for hepatocellular carcinoma : the MELD impact. Liver Transpl 2004; 10: 36-40.
Roberts JP. Prioritization of patients with liver cancer within the MELD system.  Liver Transpl 2002; 8: 329-330.
Freeman FB, Harper A, Edwards EB, et al.  The MELD/PELD system and hepatocellular carcinoma [Abstract].  Liver Transpl 2003; 9; C31.
Yao FY, Bass NM, Ascher N, et al.  Liver transplantation for hepatocellular carcinoma : lessons from the first year under the Model of End-stage Liver Disease (MELD) organ allocation policy. Liver Transpl 2004; 10: 621-630.
Yao FY, Bass NM, Nikolai B, et al. Liver transplantation for hepatocellular carcinoma : analysis of survival according to the intention-to-treat principle, and dropout from the waiting list. Liver Transpl 2002; 8: 873-883.
Evehart RH, Lake JR. Increased waiting time for liver transplantation results in higher mortality.  Transplantation 1997; 64: 130-1306.
Kirli SH, Kate T, Calleen R, et al.  Liver transplantation for hepatocellular carcinoma: validation of present selection criteria in predicting outcome. Liver Transpl 2004; 10: 911-918.
Williams R, Alisa A, Karani J, et al.  Adult-to-adult living donor liver transplant : UK experience. Eur J Gastroenterol Hepatol 2003; 15: 7-14.
Robert S, Brown JR, Shiva K, et al. Model for end-stage liver disease and Child-Pugh score as predictors or pre-transplantation disease severity: post-transplantation outcome and resource utilization in United Network for Organ Sharing status 2A patients – Liver Transpl 2002; 8: 278-286.
Freeman RB, Weisner RH, Harper A, et al.  The new liver allocation system: moving toward evidence-based transplant policy. Liver Transpl 2002; 8: 851-858.