The model for end-stage liver disease allocation system for liver transplantation saves lives, but increases morbidity and cost: a prospective outcome analysis

Authors


Abstract

We analyzed the first 100 patients who underwent liver transplantation by Model for End-Stage Liver Disease (MELD) allocation, and compared the outcome of patients on the waiting list and after orthotopic liver transplantation with the last 100 patients who underwent transplantation prior to the introduction of the MELD system in July 2007. MELD allocation resulted in decreased waiting list mortality (386 versus 242 deaths per 1000 patient-years, P < 0.0001) and the transplantation of sicker recipients (uncorrected median MELD score 13.5 versus 20, P = 0.003). Recipient posttransplant morbidity was significantly higher, mainly caused by increased percentage of renal failure requiring renal replacement therapy (13 versus 46%, P < 0.0001). However, kidney function recovered in most cases within 6 months after OLT. Hospital mortality remained similar in both groups (6% versus 9%). Patient 1-year survival was 91% versus 83% (pre-MELD versus MELD era, P = 0.2154), graft 1-year survival was 88% versus 78% (P = 0.1013), respectively. Costs accumulated were significantly higher after introduction of the MELD policy (US $81,967 versus US $127,453, a 55% increase, P = 0.02) with a strong correlation with the individual MELD score (P < 0.0001). The MELD system addresses the goal of fairness well. However, the postoperative course appears more difficult in the MELD era with increased financial burden, but reasonable patient and graft survival. This is the inevitable price to balance justice and utility in liver graft allocation. Liver Transpl 17:674–684, 2011. © 2011 AASLD.

See Editorial on Page 631

There is currently an intense debate about whether liver grafts should be offered directly to a patient (the sickest one) or rather to a center with the freedom to use an organ for the patient of their choice. The discussion about this topic is driven by the still unsolved challenge for the transplant community to balance the high number of patients in need of a liver transplant with the lack of available grafts. When the demand far exceeds the supply, it commonly raises conflicts about allocation of the scarce resources. A new system to distribute grafts to specific recipients by estimating the disease severity of transplant candidates was introduced in the United States in February 2002. Such a policy, based on the Model for End-Stage Liver Disease (MELD) score, which combines recipient kidney function, coagulation time and serum bilirubin, has been proven to be a reliable parameter to predict waiting list mortality.1 In the US, allocation of liver grafts through the MELD system resulted in a substantial decrease in median time to transplant from 981 days in 2002 to 306 days in 2006.2 Simultaneously, waiting list mortality decreased from 150 deaths in 1997 to 117 deaths in 2006, calculated per 1000 patient-years at risk.2 Despite this change leading to sicker patients at the time of transplantation, first outcome analysis showed excellent 1 year survival after liver transplantation in the MELD era.3-7 These results convinced several European countries to follow this policy (North Italian Transplant in March 2003, Eurotransplant in November 2006, “Etablissement Francais des Greffes” in March 2007, and Swiss Transplant in July 2007).8

In Switzerland, as in many other European countries such as Spain, Portugal, UK, Sweden, Norway, Finland, and Denmark, there has been a long tradition over 3 decades of center organ allocation. Thus, the introduction of the new policy induced many discussions and claims about the poorer patient outcome by MELD allocation, like the correlating increasing cost. The present study was therefore undertaken to investigate the impact of the MELD allocation policy in our transplant population. The same number of transplant candidates was evaluated before and after the change of allocation.

Abbreviations:

ELTR, European Liver Transplant Registry; FFP, fresh frozen plasma; HCC, hepatocellular carcinoma; ICU, intensive care unit; ITBL, ischemic type biliary lesions; LDLT, Living donor liver transplantation; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; RBC, red blood cell; UNOS, United Network for Organ Sharing; UW, University of Wisconsin solution.

MATERIALS AND METHODS

After July 1, 2007, we analyzed the first 100 liver transplanted recipients during the new allocation law (MELD era) and compared this group to the last 100 cases receiving an organ during the center specific allocation policy (pre-MELD era).

The study design was adapted in terms of the following considerations. First, all living donor liver transplantations (LDLT) performed during both observation periods (n = 7, pre-MELD era and n = 11, MELD era) were excluded, because these patients do not assume the same waiting list mortality risks that candidates waiting for deceased donor livers face.

Second, all retransplantations (n = 18) were excluded due to less accuracy of the MELD score compared with patients who are de novo transplanted.

The 2 observation periods lasted from January 2003 to June 2007 and from July 2007 to May 2010, respectively. Median follow-up was cumulated to 58.5 months for the pre-MELD and to 13.5 months for the MELD era. Clinical data were entered real time into an established database and analysis was performed following approval by the local ethics committee. All patients gave written informed consent for data analysis before transplantation.

With the implementation of the MELD system in Switzerland by July 2007, a consensus was reached to give all patients with HCC 14 exception points at the time of listing (median MELD in Switzerland), added by 1 extra point per each following waiting month (standard deviation). For cholangiocarcinoma, the same policy was used if patients were within criteria for the Mayo protocol.

Before the MELD era there were no specific rules or prioritization for candidates with HCC or cholangiocarcinoma. Centers were free to choose HCC candidates at any stage. We respected, however, in both eras the UCSF critera. Drop-out rates due to tumor growth beyond UCSF criteria are shown in Table 1 and were equal in both groups (6% versus 7%).

Table 1. Patient Population on Waiting List
 Center Allocation Jan 2003-Jun 2007MELD Allocation Jul 2007-May 2010P Value
Underwent transplantation100100 
 Alive after OLT8181 
 Dead after OLT1919 
Withdrawal from waiting list   
 Recovered1012 
 Refused OLT1 
 Tumor progress67 
 Too sick95 
 Living donor liver transplantation711 
 Totaldropout3236 
Death while on list4320 
Total number of listed patients175156 
Death rate while on list43/175 (25%)20/156 (13 %)0.008
Death rate (per 1000 patient years at risk)386242<0.0001
Death total (on list and after OLT)62/175 (35 %)39/156 (25 %)0.043
Table 2. Clinical Data of Transplant Candidates Prior to OLT
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Data are given as median (range).

MELD uncorrected at listing*12.5 (6-40)16 (6-40)0.113
MELD uncorrected at transplant*13.5 (6-40)20 (6-40)0.003
MELD > 25 (n) at listing8250.002
MELD > 25 (n) at transplant14320.004
MELD > 35 (n) at transplant4100.1640
Hepatorenal syndrome (n)1435<0.001
Hospitalized before OLT (n)18350.01
Hemodialysis or hemofiltration (n)217<0.001
Waiting list time (days)*255 (1-1925)192 (1-2636)0.0693
Waiting list time for patients with HCC(days)*334 (7-1330)204 (21-1541)0.036

Organ procurement was done by aortic and portal perfusion. The preservation solution first used was University of Wisconsin (UW) solution. In 2006, Swiss Transplant replaced UW solution with Celsior®. The first 68 grafts of the pre-MELD era were consecutively preserved with UW in contrast to the following grafts. Bile duct and hepatic artery were routinely flushed on the back table. We preferred the classic implantation technique in most cases. However, in an attempt to protect kidney function, we performed during the MELD era more frequently a cava preserving implantation (piggyback technique)(pre-MELD era group 3% versus the post-MELD era group 23%, P = 0.0001).

Extended criteria donor grafts (ECD) were defined as either donor age ≥ 65 years, or cold ischemia time ≥ 12 hours, or biopsy-proven graft steatosis (≥60% microsteatosis or macrosteatosis, or ≥ 30% macrosteatosis).9, 10 Primary graft nonfunction was defined as death or retransplantation within the first postoperative week after exclusion of technical, immunological, and infectious causes.11-13 Primary dysfunction was defined as the presence of at least 1 of the following at 7 days after OLT: serum bilirubin ≥10 mg/dL and international normalized ratio (INR) ≥1.6 or alanine aminotransferase (ALT) >2000.12, 14

The deaths on the waiting list were related to the sum of waiting months of all candidates during the observation period (patient-years at risk) and adjusted to 1000 patient-years according to UNOS (United Network for Organ Sharing) data.15 We classified postoperative complications according to the highest Clavien-Dindo score for each type of complication.16 Patient and graft survival were determined 3 months and 1 year after OLT. Social integration was evaluated by a survey in December 2009 using a newly designed score from 1 to 5 (1 = hospitalized patient; 2 = patient at home but immobilized, dependent on care systems; 3 = patient at home, not working; 4 = patient partially integrated in work; 5 = patient completely independent except regular controls and medication). This survey was assessed by recipients and verified in the outpatient clinic.

The competing risk analysis (Fig. 1) refers to all waiting list candidates during both observation periods before and after transplantation or patients withdrawn from list. Survival time was calculated from the time of listing to death. Accordingly, survival represents the sum of waiting list time and posttransplant time for transplanted patients. For patients who were withdrawn from the list, survival was the sum of waiting list time and survival time after withdrawal. For patients, who died on the list, survival was the time from listing to death on the list. The cumulative hazards for all candidates were adjusted for MELD score at different events (transplantation, drop out, death on list).

The cost analysis refers to all costs accumulating during admission for OLT and includes costs from the time of hospital admission prior to surgery until first posttransplant discharge. All cost data have been calculated by our financial department. In general, costs are refunded by a case based lump sum independent of the medical effort. The cost calculation is based on a cost-unit accounting system. Raw data are generated by care providers and documented real-time in the computer system. Afterward, the work performed (eg, surgery, patient care, etc.) is converted to costs by case-specific factors and finally totaled. The Swiss francs were converted to U.S. dollars without adjustment for inflation (Conversion factor: 0.97394, August 27, 2010).

Statistical Analysis

Metric data in tables represent median with range. All comparisons are performed using Fishers exact or unpaired U test (Mann-Whitney-Wilcoxon). Survival analysis was calculated according to Kaplan-Meier (log-rank test). Regression analyses were performed using logistic binomial regression or cox regression (stepwise forward). All calculations and analyses were completed with SPSS, version 18. Statistical significance was accepted with P < 0.05 (2-sided tests).

RESULTS

Did MELD Allocation Reduce Waiting Time and Mortality Before Transplantation?

The waiting time before liver transplantation depends on the number of available grafts and on the number of waiting candidates. During the pre-MELD era, candidates had a lesser chance to receive an organ with increasing size of the waiting list. This disadvantage was decreased by MELD allocation. Accordingly, the median waiting time decreased during MELD allocation but did not reach significance (255 versus 192 days, P = 0.0693) (Table 2). Median waiting time of subgroups, such as patients with HCC, decreased significantly (334 versus 204 days, P = 0.036) due to exception points at the time of listing.

MELD policy also caused a change in the waiting list population. Previously, the median MELD of transplant candidates at the time of listing was 12.5. During MELD allocation the median MELD of listed patients increased to 16 (P = 0.113), indicating a shift to sicker candidates already at the time of listing (Fig. 2). Despite this higher risk of morbidity during waiting, the death rate on our waiting list decreased. In the pre-MELD era, 43 deaths occurred on the waiting list (386 deaths per 1000 patient-years at risk) compared to 20 deaths (P = 0.008) in the MELD era (242 deaths per 1000 patient-years at risk, P < 0.0001). The dropout rate was similar in both periods (32 versus 36 cases) (Table 1).

Figure 1.

Competing risk analysis: The cumulative death risk of all listed patients was not different between both observation periods (pre-MELD versus MELD era) for MELD ≤ 25. In higher MELD candidates the overall risk to die decreased by the MELD policy (arrows).

Figure 2.

The MELD score of recipients is shown at the time of listing and at the time of transplant before and after the change of the allocation policy (July 2007). The box plots show median and interquartile range of values.

Did MELD Allocation Result in Sicker Transplant Candidates?

The distribution of the underlying disease, age or body mass index of our recipients was not different between the pre-MELD and the MELD era (Tables 3 and 4). The number of patients transplanted with hepatic tumors (HCC, cholangiocarcinoma) also did not change during both observation periods (Table 4).

Table 3. Baseline Characteristics
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Data are given as median (range).

Donor
Age (years)*49.0 (13-77)53.0 (14-86)0.098
Reanimation (n)18210.591
ICU stay (days)*1 (0-20)1 (0-13)1.0
Recipient
Male (n)75730.872
Female (n)2527 
Age (years)*52.0 (21-69)55.0 (13-70)0.071
Weight (kg)*75 (49-130)77 (43-136)0.927
Height (m)*1.73 (1.55-1.95)1.72 (1.17-1.91)0.597
BMI*26.0 (18-36)25.0 (16-43)0.973
Hematocrit*33.7 (15-47)31.1 (12-50)0.004
Platelets/μL*88,000 (22,000-285,000)87,000 (29,000-391,000)0.704
Table 4. Underlying Recipient Disease and Recipients With Additional Points on the List
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Autoimmune hepatitis, Budd-Chiari, M. Osler, hemochromatosis.

Underlying recipient disease
Hepatitis C (n)45370.3142
Hepatitis B (n)16120.5416
Alcoholic (n)15320.0072
PSC (n)420.6827
PBC (n)331.0
M. Wilson (n)410.3687
Amyloidosis (n)130.6212
Acute liver failure (n)520.4448
Alpha-1 antitrypsindeficiency (n)111.0
Cryptogenic (n)211.0
Other (n)*460.5371
Recipients with additional points on the list
HCC (n)30311.0
Cholangiocarcinoma (n)011.0

As expected, introduction of the MELD policy increased the uncorrected (laboratory) median MELD score of recipients from 13.5 to 20 (P = 0.003, Table 2, Fig. 2). One third of the transplanted patients (32%) had a MELD score > 25 compared to only 14% in the pre-MELD era (P = 0.004). Correspondingly, the preoperative incidence of hepatorenal syndrome increased in the MELD era from 14% to 35% (P < 0.001, Table 2). Significantly more patients in this group had to be hospitalized prior to OLT (18% versus 35%, P = 0.01) or needed renal replacement therapy already before OLT (2% versus 17%, P < 0.001, Table 2).

Importantly, despite sicker transplant candidates in the MELD era, the proportion of patients with an extreme MELD score of 36 or higher remained similar in both groups (4% versus 10%, P = 0.164, Table 2).

Did MELD Allocation Increase Graft Injury?

Before introduction of the MELD policy, many organs were used locally because of the preferred local distribution by a center specific system. The change to MELD allocation led to a higher percentage of traveling (>50 km) liver grafts (41% versus 63%, P = 0.0029, Table 5). Although this need to transport more grafts increased the risk of prolonged preservation, the median cold ischemia time did not increase due to short distances (maximum distance 300 km) in Switzerland (536 versus 520 minutes, P = 0.680) (Table 5).

Table 5. Graft Characteristics and Perioperative Data
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Travel of >50 km.

  • Data are given as median (range).

  • Donor age ≥65 years or cold ischemia time ≥720 minutes or biopsy-proven graft steatosis (≥60% microsteatosis or macrosteatosis or ≥30% macrosteatosis.9, 10

Traveling grafts (n)*41630.0029
Cold ischemia (minutes)536 (230-1197)520 (215-992)0.680
ECD grafts (n)31480.0204

In face of the scarce organ supply, the percentage of accepted so called “marginal” livers increased. Nearly half of the grafts in the MELD era fulfilled at least 1 extended criterion (ECD grafts) in contrast to one-third in the pre-MELD era (48 versus 31%, P = 0.0204) (Table 5). Despite that, the rate of graft dysfunctions and primary nonfunction (for definition, see Materials and Methods section) remained low in the pre-MELD and MELD era (10% versus 13% and 4% versus 2%, Table 6). Peak ALT and AST level after OLT appeared similar (897 versus 781 U/L and 1068 versus 1031 U/L, P = 0.529, P = 0.758) (Table 7). Based on this, MELD allocation did not increase the risk for graft injury in our study.

Table 6. Postoperative Complications
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Primary nonfunction defined according to Ploeg et al.,11 Dahm et al.,12 and Lo et al.13

  • Delayed graft function defined according to Lo et al.13 and Guarrera et al.14

Anastomotic bile leak (n)6130.1464
Anastomotic biliary stricture (n)16190.7102
 Late anastomotic stricture(> 3 months after OLT)14/1613/19 
 Repair by endoscopic stent13/1617/19 
 Repair by hepaticojejunostomy3/162/19 
Intrahepatic ischemic type biliary lesion (ITBL) (n)321.0
Hepatic artery thrombosis (n)270.1697
Anastomotic hepatic artery stenosis (n)651.0
Postoperative bleeding (n)4120.065
Retransplantation within 3 months (n)020.4975
Primary nonfunction (n)*420.6827
Delayed graft function (n)10130.6584
Hemofiltration or hemodialysis directly after OLT (n)1346<0.0001
Hemofiltration or hemodialysis 6 months after OLT (n)130.621
Rejection rate (n)33360.755
Septic complications (n)14180.365
Hospital mortality (n)690.5928
Table 7. Perioperative Data
 Center Allocation n = 100MELD Allocation n = 100P Value
  • *

    Data are given as median (range).

Warm ischemia (minutes)*42 (33-58)45 (31-61)0.827
Piggy back implantation (n)323<0.001
Classic implantation (n)9777 
Blood loss (cc)*1250 (300-10000)1500 (200-15000)0.689
RBC (U)*4 (0-39)3 (0-47)0.879
FFP (U)*13 (0-55)0 (0-77)<0.001
Fibrinogen (g)*0 (0-22)4 (0-30)<0.001
Cristalloids (cc)*3000 (770-10000)2500 (500-15000)0.136
Colloids (cc)*1500 (500-4500)2500 (500-10500)<0.001
Operation time (minutes)*365 (240-705)380 (200-660)0.704
Peak AST (U/L)*1068 (114-13805)1031 (60-34545)0.758
Peak ALT (U/L)*897 (95-8492)781 (110-12684)0.529

Did MELD Allocation Complicate the Transplant Procedure?

One argument of opponents of the MELD policy is an expected higher intraoperative bleeding risk due to worse coagulation with increasing MELD scores. However, despite sicker recipients, operation time (365 versus 380 minutes, P = 0.704), warm ischemia time, blood loss and transfusion of red blood cells (RBCs) during hepatectomy and implantation were comparable before and after the change of allocation policy (1250 versus 1500 mL, P = 0.689 and 4 versus 3 units of RBC, P = 0.879) (Table 7) despite lower hematocrit before OLT (33.7 versus 31.1, P = 0.004) (Table 3). The number of recipients needing postoperative lavage of hematoma increased from 4% to 12% during the MELD era (P = 0.065). This slight increase in bleeding events after transplant, however, should be interpreted with caution because our coagulation and fluid management during OLT have changed in 2008 due to a change in anesthesia management. Although previously transfusions of fresh frozen plasma were given on a routine basis, a thromboelastogram-guided substitution of isolated clotting factors (Beriplex, factor XIII, fibrinogen) is now standard of care.

Did MELD Allocation Increase Postoperative Morbidity?

It is well known that sicker transplant candidates have an increased risk for major morbidity. In this context, biliary complications occurred more often during the MELD era. Biliary anastomotic strictures were documented in 19% of cases in the post-MELD group compared to 16% in the pre-MELD group (P = 0.7102), biliary anastomotic leaks in 13% versus 6% (P = 0.1464) (Table 6). Most biliary stenoses occurred after 3 months and were successfully treated by temporary endoscopical stenting (Table 6). Intrahepatic ischemic type biliary lesions were rarely detected in both groups (2% versus 3%, P = 1.0) (Table 6). Vascular complications appeared similar in both observation groups in terms of hepatic artery thrombosis (2% versus 7%) and hepatic artery stenosis (6% versus 5%). We found also no difference in rejection rates and septic complications between both groups (Table 6).

Despite several kidney protective strategies in the post-MELD era group (cava preserving implantation technique, low-level calcineurin inhibitor immunosuppression), recipients during the MELD era were significantly more in need of renal replacement therapy (13 versus 46%, P < 0.0001) (Table 6). The odds ratio for the development of renal failure > RIFLE class 2 (increase of creatinine by 2-fold and/or urinary output <0.5 mL/kg body weight over a 12-hour period) was 3.2 (CI: 1.1-8.9, P = 0.028). Importantly, 6 months after liver transplantation, the number of patients still requiring renal replacement therapy was comparable and low in both groups (1% versus 3%, P = 0.6212) (Table 6). In addition, median serum creatinine was not different in both groups after 6 months (96 versus 103 μmol/L).

Overall classification of complications showed more grade III (interventions) and grade IV (kidney failure) complications in the post-MELD era group (10 versus 19%, P = 0.107; 16 versus 32%, P = 0.013) (Table 8). Median ICU and hospital stay were 2 and 6 days longer during the MELD era, respectively, (3 versus 5 days and 18 versus 24 days, P = 0.017 P = 0.001) (Table 8). Also, the recipient MELD score correlated significantly with hospital stay (P = 0.0003).

Table 8. Morbidity According to Clavien-Dindo16 and Length of Stay
 Center Allocation n = 107MELD Allocation n = 105P Value
  • *

    Data are given as median (range).

Clavien I+II (n)6235<0.0001
Clavien III (n)10190.107
Clavien IV (n)16320.013
Clavien V (death) (n)19191.0
ICU stay (days)*3 (1-53)5 (1-93)0.017
Hospital stay (days)*18 (7-113)24 (8-128)0.001

Did MELD Allocation Lead to Poor Patient and Graft Survival?

One of the most discussed topics currently is the effect of MELD policy on recipient and graft survival. Hospital mortality (6% versus 9%) was not different between the pre-MELD and the MELD era. Patient survival at 3 months after OLT remained stable between both groups (93% ± 3% versus 89% ± 3%). Patient 1-year survival was also comparable (91% ± 3% versus 83% ± 4%, P = 0.2154) (Fig. 3).

Figure 3.

One-year patient and graft survival were not significantly different during MELD and pre-MELD allocation (P = 0.2154, P = 0.1013). Numbers of patients at risk are indicated.

The rate of retransplantation due to graft loss within 3 months after OLT was low in both groups (0% versus 2%) (Table 6). Also 1-year graft survival was not significantly different between both groups (88% ± 4% versus 78% ± 5%, P= 0.1013) (Fig. 3).

To evaluate the risk to die before and after OLT, we analyzed the cumulative hazards of all candidates on the waiting list during both observation periods with competing risk (transplantation, death on list, or withdrawal from list). For candidates with a MELD score ≤25, the MELD allocation policy resulted in a similar overall risk of death (Fig. 1). In contrast, for candidates presenting with a MELD > 25 the overall risk of death was higher in the pre-MELD era compared to the MELD era (Fig. 1)

Did MELD Allocation Increase Cost?

Introduction of the MELD system led to sicker transplant candidates and with increasing MELD score of recipients, more cost is expected due to the progress in liver disease. The calculation of the median cumulative cost per single case (from the time of admission to first discharge after OLT) confirmed an increase from US $81,967 during the pre-MELD era to US $127,453 per case (P = 0.02) in the MELD era. Cost correlated strongly with the individual MELD score (P = 0.001).

Did MELD Allocation Policy Delay Social Integration?

Performing transplantation on sicker candidates leads to delayed recovery and a number of rehospitalized patients. We investigated by a patient-assessed survey in December 2009 whether current disadvantages in terms of social integration occurred after OLT. Despite the methodological difference in follow-up between the pre-MELD and MELD era, we observed no differences in the percentage of currently hospitalized patients after OLT (14% versus 16%), the number of recipients currently dependent on regular care after OLT (4% versus 15%), the number of recipients currently being independent at home after OLT (57% versus 52%), and the number of recipients working full time (9% versus 14%).

DISCUSSION

In 2002, an organ allocation system by disease severity was introduced in the U.S. with the primary aim to reduce the mortality of those on the transplant list. First reports concerning this policy showed stable 1-year survival.3-7 Other countries and health care systems have subsequently adopted the principles of this MELD-based allocation, but with varying outcomes. Currently, many countries in Europe are critically evaluating whether the MELD allocation policy would decrease long-term outcome and where to find a certain cutoff for recipients not to be transplanted.1 In addition, few data have been published on morbidity during MELD allocation.17, 18 This study compared overall outcome of transplant candidates before and after liver transplantation in a pre-MELD and MELD era.

Although these data appear less new for the U.S. system, it provides information on experiences with MELD allocation in a European country where the debate is ongoing whether allocation based on the MELD system should be recommended.

In this study, we decided to consider deceased donor liver transplanted patients before and after the change in the allocation policy. Our population includes recipients who received extra points due to hepatic tumors (standard exception), as well as recipients with acute liver failure. Both observation periods are not different in terms of recipient age, sex, height, weight, and underlying disease. The shortcoming of our study is the different follow-up period in both groups (58.5 months versus 13.5 months). Because of this bias, we currently cannot predict long-term survival effects of MELD allocation. However, it is well known that most non–tumor related deaths after OLT occur within the first year.8

Under these circumstances, we demonstrate first, that starting MELD allocation in Switzerland led to a significant decrease of our waiting list mortality. Of note, MELD allocation did also change the listing policy. Previously, the number of patients with MELD >25 at the time of listing was very low (8%, Table 2), probably due to the fact that sick patients had no chance to receive a liver graft while waiting. At that time several end-staged candidates were not even placed on the list, and we therefore may postulate that true death rates were much higher in the pre-MELD era. The introduction of a disease oriented allocation policy changed this situation dramatically. The percentage of candidates with a MELD score > 25 at the time of listing increased significantly (Table 2) and further increased to 32% at the time of transplant (32/100 versus 14/100 pre-MELD, Table 2). The preferential liver allocation to the sickest candidates gives rise to some concerns regarding tumor patients on the list with low MELD. Our analysis, however, shows that HCC candidates were not disadvantaged by the MELD system (Table 2). In contrast, their waiting time decreased in the MELD era due to exception points. In addition, despite sicker transplant candidates, the overall intraoperative course in terms of blood loss, blood supply, operative time and also ICU stay did not differ between the pre-MELD and the MELD era.

One-year graft survival of recipients in the MELD era appeared 10% less compared to the previous allocation system (88% ± 4% versus 78% ± 5%, P = 0.1013). Patient 1-year survival decreased by 8% in the MELD era (91% ± 3% versus 83% ± 4%, P = 0.2154), and remained as high as survival rates reported by the European Liver Transplant Registry (ELTR; overall 1-year survival 82%).8

The MELD score is known to be influenced strongly by creatinine. MELD allocation, therefore, resulted in selection of candidates with pretransplant kidney injury. As a result, one major drawback of the MELD policy was the high rate of renal failure after OLT. In an attempt to adapt our transplant procedure to the expected kidney injury, we avoided cava clamping in recipients with high-MELD scores and preferred the piggy back implantation procedure in these patients (23/100).

Despite this, the number of patients requiring renal replacement therapy in the post group exceeded by far the number of comparable cases in the pre-group. However, 6 months after OLT, most kidneys recovered in both groups. Three patients in the MELD era required renal replacement therapy 6 months after OLT compared to 1 patient in the pre-MELD era (nonsignificant).

Another finding of this study was an unexpected higher incidence of anastomotic biliary complications in the MELD era. Notably, the reported incidence of biliary complications in historical pre-MELD series shows a large variation from 10%-30%.19 These complications have remained high despite several attempts to study techniques of biliary reconstruction. Multiple risk factors have been identified for biliary complications including technical issues, graft ischemia, and immunological factors.20 In this context, it has to be emphasized that institution of MELD policy was not the only innovation in our transplant center, as in many others. During the last 3 years we implemented changes in terms of procurement techniques (Celsior® instead of UW), and anesthesia management (isolated coagulation factors instead of fresh frozen plasma, increased use of catecholamines). The impact of all these factors on biliary complications remains unclear. However, a recent U.S. study also reported about 18% bile leaks and 23% anastomotic strictures in 256 cadaveric liver transplants allocated by MELD.21 These authors identified as independent factors only warm ischemia during implantation and the preservation solution (HTK being superior to UW). The recipient MELD score did not reach significance.21 A regression analysis in our population confirmed that biliary complications were not related to MELD scores but to the use of Celsior® and ECD grafts. Further investigations will evaluate this suspicion to prove our preservation concept.

Our results correlate well with several publications from the United States, underlining that MELD score cannot accurately predict short-term survival after OLT.4, 6, 7, 22-33 Also, data from a European center (Italy) show no change in survival rates by MELD allocation but a significant decrease (6%) in waiting list mortality.34 Median MELD was 19 in this study at the time of transplant. In contrast, recent results from Germany report significantly decreased 3-month patient survival in the MELD era (88.6 versus 79.5%), whereas median MELD at transplant was only 16.3.35 We failed to confirm this short-term survival disadvantage of MELD allocation in our study (3-month patient survival rates of approximately 90% in our pre-MELD and MELD era).

Such contradictory results highlight the difficulty to translate an allocation strategy in European countries despite former success in the United States. Several factors may be responsible. First, donation rates vary less among the 11 UNOS regions8 compared to Europe.36 For example, Spain has by far the highest donation rate (34.2 per million inhabitants in 2008), followed by Austria and Belgium. Only 11.8 and 8.9 donors per million inhabitants are reported in Switzerland and Greece, respectively.36 Second, the waiting time and deaths on the waiting list differ widely in European countries. Countries with very low donation rates are under much higher pressure to use grafts from extended criteria donors (ECD) than others. Third, the total number of liver transplants per million inhabitants in Europe is half of the number of liver transplants performed in the United States.8

MELD policy, therefore, may have different effects in Europe than in the United States. Local donation rates and the number and the stage of the liver disease of candidates will influence the individual MELD at transplant and thus postoperative outcome. The price of MELD allocation is an increase in postoperative morbidity, eventually resulting in longer hospital stay, temporary renal complications, and higher cost. More resources, both financial and personal, are urgently required to get sicker recipients through the first months after OLT. Further investigations in a high number of patients are necessary to confirm cutoff levels where graft and patient survival possibly deteriorate.

In this context, a separate analysis in our entire population confirmed that mortality rates after OLT (Clavien V complications) did not necessarily vary between different MELD groups up to a score of 30 in contrast to waiting list deaths (Fig. 4). Analysis of the cumulative risk to die in all waiting list candidates confirmed clear advantages for patients with MELD score > 25 (Fig. 1).

Figure 4.

The death rates on the waiting list for the different MELD groups were largely varying in contrast to mortality after OLT (dotted line). Survival benefit increased up to a MELD score of 35.

Data from the United Kingdom also suggest that the differences among recipients with a MELD score of less than 36 are small and not statistically significant (1-year survival between 87 and 83%). Only those patients with an extreme MELD score of 36 or higher had a markedly lower 1-year survival of 68%.29 Importantly, this group of patients accounts for less than 10% in our population (13/200) as in that in the United States.2 On the other hand, approximately one-third of our recipients present currently with MELD scores between 26 and 35 at the time of transplant (28/100). Considering the poor prognosis without liver transplantation, those patients have undoubtedly the highest survival benefit37 (Fig. 4).

On the basis of these results and considerations, we believe, therefore, that the MELD system addresses best the goal of fairness. Despite the expected higher postoperative effort, it still appears to be the most reliable tool for selecting liver transplant candidates.

Ancillary