Abstract
Background
The Model for End-stage Liver Disease (MELD) has been used as a prognostic tool since 2002 to predict pre-transplant mortality. Increasing proportions of transplant candidates with higher MELD scores, combined with improvements in transplant outcomes, mandate the need to study surgical outcomes in patients with MELD scores of ≥40.
Methods
A retrospective longitudinal analysis of United Network for Organ Sharing (UNOS) data on all liver transplantations performed between February 2002 and June 2011 (n = 33 398) stratified by MELD score (<30, 30–39, ≥40) was conducted. The primary outcomes of interest were short- and longterm graft and patient survival. A Kaplan–Meier product limit method and Cox regression were used. A subanalysis using a futile population was performed to determine futility predictors.
Results
Of the 33 398 transplant recipients analysed, 74% scored <30, 18% scored 30–39, and 8% scored ≥40 at transplantation. Recipients with MELD scores of ≥40 were more likely to be younger (P < 0.001), non-White and to have shorter waitlist times (P < 0.001). Overall patient survival correlated inversely with increasing MELD score; this trend was consistent for both short-term (30 days and 90 days) and longterm (1, 3 and 5 years) graft and patient survival. In multivariate analysis, increasing age, African-American ethnicity, donor obesity and diabetes were negative predictors of survival. Futility predictors included patient age of >60 years, obesity, peri-transplantation intensive care unit hospitalization with ventilation, and multiple comorbidities.
Conclusions
Liver transplantation in recipients with MELD scores of ≥40 offers acceptable longterm survival outcomes. Futility predictors indicate the need for prospective follow-up studies to define the population to gain the highest benefit from this precious resource.
Introduction
The Model for End-stage Liver Disease (MELD) laboratory score was adopted in 2002 to quantify the clinical status of potential liver transplant recipients and to prioritize candidates for the allocation of donor organs.1 The MELD score is calculated using three objective values: the international normalized ratio (INR); serum creatinine, and serum bilirubin. It reflects a patient's 3-month mortality risk on a scale of 6–40. Priority for liver allocation is given to individuals with the highest scores.1,2
The United Network for Organ Sharing (UNOS), a private non-profit organization, governs the allocation of donor organs to transplant candidates within the USA. In 2011, median MELD scores in adults receiving deceased donor livers varied by region from 19.5 to 36.0. The national median MELD score was 27.4 When the MELD score was implemented, patients with scores of >40 were frequently considered to be too ill to undergo transplantation.5 Since then, transplants among patients with MELD scores of >40 have increased in frequency as patient outcomes have improved.5,6 Between 2006 and 2010, the proportion of transplant recipients with MELD scores of ≥40 increased from 6.8% to 10.7% across the nation.5 Given their high waitlist mortality rate and short life expectancy, these transplant candidates have also demonstrated in small reports the most significant benefit from liver transplant, regardless of donor allograft quality. Even in patients with very high MELD scores, transplantation has been shown to provide a short-term survival benefit.7
Although the MELD score stratifies patients based on their risk for mortality without a transplant, it does not necessarily identify which patients will have the best longterm outcomes from liver transplantation. This has become increasingly important in an era of increased metrics and oversight across health care fields. The present literature addressing liver transplantation in patients with MELD scores of >40 and longer-term post-transplant outcomes among this group of patients is scant. The issue has been evaluated only in a single-centre context8 and no national multicentre studies exist.
This retrospective cohort study uses the UNOS database to examine post-transplant survival, morbidity and prognostic indicators of survival in liver allograft recipients with MELD scores of ≥40 at the time of transplantation. It also explores futility predictors of early mortality.
Materials and methods
A population-based, longitudinal retrospective review of patients who became transplant recipients during the period from 27 February 2002 to 30 June 2011 was conducted using UNet , a secure UNOS online database that comprehensively shares nationwide information on organ matching.
Inclusion criteria required the patient to have undergone a transplant performed on or after 27 February 2002 and to have had a MELD score of 6–60 (n = 56 470). Exclusion criteria removed patients with approved MELD exception (n = 16 460), a history of one or more liver allograft failures (n = 4380), and simultaneous kidney, pancreas, intestine, lung or heart (n = 4688) transplant. The resulting cohort (n = 33 398) was further divided into three groups based on MELD score at transplant: <30 (n = 24 804); 30–39 (n = 5984), and ≥40 (n = 2610). The primary outcomes of interest were short- and longterm outcomes in patients with scores of ≥40 compared with those in the other groups. Short-term outcomes were defined as length of hospitalization, and 30- and 90-day post-transplantation survival of grafts and patients. Longterm outcomes were defined as 1-, 3- and 5-year survival. To identify potential predictors, multiple univariate Cox regression analyses were completed for clinically suspected risk factors including: donor and recipient characteristics (including age, gender and ethnicity); the medical condition of the patient immediately before transplantation; the cause of the donor's death; allograft type, and organ share type. Covariates including donor and recipient age, gender, body mass index (BMI), waiting time, total warm ischaemia time, share type and cause of donor death were used in the multivariate analysis. Table1 gives a full list of the variables examined.
Table 1.
Pretransplant recipient and liver allograft characteristics by Model for End-stage Liver Disease (MELD) score
| MELD score category | ||||
|---|---|---|---|---|
| Recipient characteristics | <30 (n = 24 804) | 30–39 (n = 5984) | ≥40 (n = 2610) | P-value |
| Age, years, median (IQR) | 53 (13) | 51 (16) | 50 (17) | <0.001 |
| Male, n (%) | 15 916 (64%) | 3567 (60%) | 1626 (63%) | <0.001 |
| Obese (BMI ≥30 kg/m2), n (%) | 7129 (29%) | 1519 (25%) | 695 (27%) | <0.001 |
| Ethnicity, n (%) | ||||
| White | 18 811 (76%) | 3916 (65%) | 1611 (62%) | <0.001 |
| African-American | 2214 (9%) | 727 (12%) | 317 (12%) | |
| Hispanic/Latino | 2885 (12%) | 1022 (17%) | 496 (19%) | |
| Asian | 636 (3%) | 233 (4%) | 141 (5%) | |
| Othera/unknown/mixed | 258 (1%) | 86 (1%) | 45 (2%) | |
| Viral infections, n (%) | ||||
| HBsAg positive | 904 (4%) | 329 (6%) | 207 (8%) | <0.001 |
| HBcAb positive | 3858 (16%) | 1015 (17%) | 501 (19%) | <0.001 |
| Hepatitis C positive | 8171 (33%) | 1795 (30%) | 769 (30%) | |
| EBV positive | 14 325 (58%) | 3096 (52%) | 1336 (51%) | <0.001 |
| Other comorbidities at transplant, n (%) | ||||
| Diabetes | 4876 (20%) | 983 (16%) | 376 (14%) | <0.001 |
| Portal vein thrombosis | 1365 (6%) | 283 (5%) | 132 (5%) | <0.001 |
| PRBC prior to transplantb | 317 (1%) | 137 (2%) | 67 (3%) | <0.001 |
| Prior upper abdominal surgery | 9299 (38%) | 1894 (32%) | 733 (28%) | <0.001 |
| TIPS | 2332 (9%) | 431 (7%) | 199 (8%) | <0.001 |
| Encephalopathy at transplant, n (%) | ||||
| None | 6998 (28%) | 851 (14%) | 252 (10%) | <0.001 |
| Grade I or II | 15 000 (61%) | 3298 (55%) | 1250 (48%) | |
| Grade III or IV | 2101 (9%) | 1680 (28%) | 1044 (40%) | |
| Ascites at transplant, n (%) | ||||
| Absent | 3845 (16%) | 732 (12%) | 354 (14%) | <0.001 |
| Mild | 13 673 (55%) | 2356 (40%) | 904 (35%) | |
| Moderate | 6607 (27%) | 2740 (46%) | 1297 (50%) | |
| Waiting time, days, median (IQR) | 82 (250) | 14 (85) | 8 (37) | <0.001 |
| Distance from donor to transplant hospital, miles, median (IQR) | 58 (176) | 70 (194) | 74 (2037) | <0.001 |
| MELD score, mean (SD) | 18.4 (5.4) | 33.9 (2.8) | 44.2 (3.7) | <0.001 |
| Pre-transplant laboratory results | <0.001 | |||
| Total bilirubin, mg/dl, median (IQR) | 3.3 (3.9) | 15.9 (21.3) | 32.7 (14.3) | |
| Creatinine, mg/dl, median (IQR) | 1.0 (0.6) | 2.0 (1.5) | 2.5 (2.5) | <0.001 |
| INR, median (IQR) | 1.6 (0.6) | 2.4 (1.2) | 3.1 (1.8) | <0.001 |
| Albumin, g/dl, mean (SD) | 2.9 (1.0) | 2.6 (1.0) | 2.8 (1.0) | <0.001 |
| Dialysis in week prior to transplant, n (%) | 395 (2%) | 1126 (20%) | 1221 (47%) | <0.001 |
| Dialysis while on waitlist, n (%) | 199 (1%) | 300 (5%) | 273 (11%) | <0.001 |
| Hospitalized pre-transplant,c n (%) | 2968 (12%) | 1144 (19%) | 459 (18%) | <0.001 |
| Transplant candidate TIPS, n (%) | 1865 (8%) | 311 (5%) | 139 (5%) | <0.001 |
| Transplant candidate on life support/ventilator, n (%) | 428 (2%) | 586 (10%) | 393 (15%) | <0.001 |
| Medical condition at transplant, n (%) | ||||
| Hospitalized in ICU | 1650 (7%) | 2060 (35%) | 1424 (55%) | <0.001 |
| On life support | 828 (3%) | 1090 (18%) | 832 (32%) | |
| On ventilator | 728 (3%) | 954 (16%) | 711 (27%) | |
| Hospitalized not ICU | 3304 (13%) | 2393 (40%) | 949 (37%) | <0.001 |
| Not hospitalized | 19 759 (80%) | 1489 (25%) | 216 (8%) | <0.001 |
| Known malignancy since waitlist | 410 (2%) | 71 (1%) | 31 (1%) | <0.001 |
| Share type, n (%) | ||||
| Local | 17 622 (71%) | 3940 (66%) | 1650 (63%) | <0.001 |
| Regional | 5149 (21%) | 1792 (30%) | 860 (33%) | |
| National | 2030 (8%) | 251 (4%) | 98 (4%) | |
| Foreign | 3 (0.01%) | 1 (0.02%) | 2 (0.08%) | |
| Liver graft type, n (%) | ||||
| Split | 2602 (11%) | 186 (3%) | 84 (3%) | <0.001 |
| Whole | 22 202 (90%) | 5798 (97%) | 2526 (97%) | |
Includes American Indian, Alaskan Native, Native Hawaiian, other Pacific Islander.
Received ≥5 units of packed red blood cells 48 h prior to transplant as a result of spontaneous portal hypertensive bleeding.
Hospitalized within 90 days before transplantation.
BMI, body mass index; EBV, Epstein–Barr virus; ICU, intensive care unit; INR, international normalized ratio; IQR, interquartile range; PRBC, packed red blood cells; SD, standard deviation; TIPS, transjugular intrahepatic portocaval shunt.
For the futility analysis, a nested case–control study was designed within the cohort of patients with MELD scores of ≥40 (n = 2610). Transplant recipients with MELD scores of ≥40 who died during the study period were included in this sub-study (n = 660); those with MELD scores of ≥40 who were lost from follow-up or who remained alive at the end of the study (censored) were excluded (n = 1950). Futility was defined as death within 3 months of transplantation. Thus, patients who died within 3 months were designated as futile (n = 282). Patients who died after 3 months were considered as non-futile (n = 378). Multiple logistic regression analyses were performed to identify potential predictors of futility.
Statistical analysis
Descriptive statistics of clinical and demographic characteristics were summarized using the chi-squared test or Fisher's exact test for categorical variables. Normally distributed continuous variables such as BMI were tested using one-way analysis of variance (anova). Non-normally distributed numerical variables such as hospital length of stay (LoS) were tested using the Kruskal–Wallis test. Survival curves for three groups were obtained using the Kaplan–Meier product limit method and compared using the Cox–Mantel log rank statistic. Patient death and graft failure represented the endpoints in the survival analysis. Right censoring was performed for patients who entered the study at a definite time (e.g. transplantation), but ceased to remain in follow-up before the endpoints of the study were observed (e.g. graft failure and death).
For multivariate analysis, predetermined covariates (age, gender, ethnicity, donor and share type) and variates with a P-value of ≤0.1 were included in the final model. All probability values were two-tailed. The type I error rate was set at 0.05. All statistical analyses were performed using IBM spss Statistics for Mac OS Version 21.0 (IBM Corp., Armonk, NY, USA).
Results
Of the 33 398 transplant recipients analysed, 74% (n = 24 804) had MELD scores of <30, 18% (n = 5984) had MELD scores of 30–39, and 8% (n = 2610) had MELD scores of ≥40 at transplantation. Overall mortality was directly proportional to MELD score (26%, 28% and 30%, respectively, in recipients with MELD scores of <30, 30–39 and ≥40).
There were 1735 futile transplants in recipients who died within 3 months after transplantation. Analysed according to MELD score subgroup, 4.2% of patients with MELD scores of <30 (n = 1031), 7.1% of patients with MELD scores of 30–39 (n = 422), and 10.8% of patients with MELD scores of ≥40 (n = 282) underwent futile transplants.
Transplant recipient characteristics
The characteristics of transplant recipients are displayed and compared in Table1. Gender distributions were similar across the groups. At transplantation, recipients with MELD scores of ≥40 were slightly younger (median age: 53 years, 51 years and 50 years, respectively, in recipients with MELD scores of <30, 30–39 and ≥40; P < 0.001). Additionally, the group with MELD scores of ≥40 had a higher proportion of non-White recipients, including patients of African-American, Hispanic, Asian, American Indian, Alaskan Native, Native Hawaiian and other Pacific Islander ethnicities (P < 0.001). At transplantation, recipients with MELD scores of ≥40 had serum chemistry measurement values higher than those with MELD scores of <30: total bilirubin was 10 times higher; creatinine was 2.5 times higher, and INR was twice as high (P < 0.001). Compared with patients with MELD scores of <30 or of 30–39, these high MELD score recipients were also significantly more likely to be dialysed within 1 week prior to transplantation (2%, 20% and 47%, respectively, in recipients with MELD scores of <30, 30–39 and ≥40; P < 0.001) and at some other time while on the waiting list (1%, 5% and 11%, respectively, in recipients with MELD scores of <30, 30–39 and ≥40; P < 0.001).
A greater proportion of recipients with MELD scores of ≥40 had grade III or IV encephalopathy prior to transplantation. However, fewer patients had a history of diabetes and portal vein thrombosis. They were also less likely to have a history of upper abdominal surgery or a transjugular intrahepatic portocaval shunt (TIPS) (P < 0.001). Despite spending significantly less time on the waiting list (P < 0.001), they were more likely to have been hospitalized within the 90 days prior to transplantation and to have required mechanical ventilation (P < 0.001). The most common indication for transplantation was hepatitis C infection (30%). Recipients with MELD scores of ≥40 were more likely to receive an organ of regional share type (P < 0.001). Recipients with MELD scores of <30 were more likely to receive a split liver (P < 0.001) or a liver from a local and national share (P < 0.001).
Donor characteristics
Donor characteristics are displayed in Table2. Patients with MELD scores of ≥40 were more likely to receive liver allografts from donors who were younger (median donor age: 41 years, 39 years and 39 years, respectively, in recipients with MELD scores of <30, 30–39 and ≥40; P < 0.001), non-White (P < 0.001), and obese (P = 0.001). These donors were less likely to have a medical history of hypertension and diabetes (P < 0.001) or to be extended criteria donors (defined as those aged >60 years or those aged 50–59 years with two of history of hypertension, serum creatinine >1.4 mg/dl and death resulting from stroke). The most common cause of donor death was stroke, followed by head trauma and anoxia, across all three MELD score-stratified recipient groups (P = 0.404).
Table 2.
Demographic and clinical characteristics of liver allograft donors by recipient Model for End-stage Liver Disease (MELD) score
| Characteristics | MELD score categories of transplant recipients | |||
|---|---|---|---|---|
| <30 | 30–39 | ≥40 | P-value | |
| (n = 24 804) | (n = 5984) | (n = 2610) | ||
| Age, years, median (IQR) | 41 (30) | 39 (29) | 39 (29) | <0.001 |
| Gender, male, n (%) | 14 537 (59%) | 3558 (60%) | 1570 (60%) | 0.189 |
| Race, n (%) | ||||
| White | 17 473 (70%) | 3924 (66%) | 1659 (64%) | <0.001 |
| African-American | 3809 (15%) | 843 (14%) | 365 (14%) | |
| Hispanic/Latino | 2771 (11%) | 1006 (17%) | 478 (18%) | |
| Asian | 452 (2%) | 129 (2%) | 73 (3%) | |
| Othera/unknown/mixed | 298 (1%) | 82 (2%) | 35 (2%) | |
| Obesity (BMI ≥30 kg/m2), n (%) | 10 292 (42%) | 2592 (43%) | 1162 (45%) | 0.001 |
| ECD,b n (%) | 6028 (27%) | 1223 (21%) | 540 (21%) | <0.001 |
| Hypertension, n (%) | 7451 (30%) | 1658 (28%) | 730 (28%) | <0.001 |
| Diabetes mellitus, n (%) | 2282 (9%) | 487 (8%) | 214 (8%) | <0.001 |
| HBV core Ab positive, n (%) | 1224 (5%) | 245 (4%) | 101 (4%) | <0.001 |
| Controlled donor, n (%) | 1075 (4%) | 170 (3%) | 71 (3%) | <0.001 |
| Liver biopsied, n (%) | 7793 (31%) | 1617 (27%) | 657 (25%) | <0.001 |
| Cause of death, n (%) | ||||
| Anoxia | 3787 (17%) | 1034 (18%) | 434 (17%) | 0.404 |
| CVA/stroke | 9441 (42%) | 2363 (40%) | 1055 (41%) | |
| Head trauma | 8850 (39%) | 2365 (40%) | 1017 (40%) | |
| CNS tumour | 143 (1%) | 37 (1%) | 16 (1%) | |
| Other | 498 (2%) | 114 (2%) | 49 (2%) | |
Includes American Indian, Alaskan Native, Native Hawaiian, other Pacific Islander.
ECD: extended criteria donors according to the kidney allocation system includes donors aged >60 years or aged >50 years with two of the following: hypertension, serum creatinine >1.5 mg/dl, and death resulting from a stroke.
BMI, body mass index; CNS, central nervous system; CVA, cerebrovascular accident; HBV, hepatitis B virus; IQR, interquartile range.
Liver allograft and preservation conditions
The majority of transplanted allografts were whole livers. Average cold ischaemia time was similar across all groups (8 h; P = 0.082). Total warm ischaemia time, including anastomotic time, was significantly longer in the higher MELD score group (38 min, 39 min and 41 min, in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001).
Split- versus whole-liver allografts
Among recipients with MELD scores of <40, short-term patient survival was similar for split- and whole-liver allografts (97%, 95%, and 97%, 94% at 30 days and 90 days, respectively); however, longterm patient survival was better in those who received split grafts (91%, 86%, 82%, and 89%, 81%, 75%, at 1, 3 and 5 years, respectively; P < 0.001). Among recipients with MELD scores of ≥40, short-term survival was better in those who received split grafts (93%, 88% and 92%, 87% at 30 days and 90 days, respectively), whereas longterm survival was similar (79%, 79%, 69% and 80%, 73%, 69% at 1, 3 and 5 years, respectively).
Univariate regression analysis showed a split graft to confer no benefit in recipients with MELD scores of ≥40 [hazard ratio (HR) 0.80, 95% confidence interval (CI) 0.50–1.29; P = 0.357]. However, it showed a reduced risk for mortality in recipients with MELD scores of <40 (HR 0.70, 95% CI 64–0.78; P < 0.001). In multivariate analysis adjusted for donor and recipient age, gender, BMI and waiting time, a split graft showed no significant effect in terms of patient survival in recipients with MELD scores of ≥40 (HR 1.56, 95% CI 0.92–2.66; P = 0.102) and some survival benefit in recipients with MELD scores of <40 (HR 0.89, 95% CI 0.80–0.99; P = 0.032). After controlling for confounders, graft survival in split-liver transplantations did not differ significantly between the groups [HR 0.72, 95% CI 0.90–1.07 (P = 0.718) in recipients with MELD scores of <40; HR 1.59, 95% CI 1.00–2.52 (P = 0.049) in recipients with MELD scores of ≥40]. Overall, the survival outcomes of split-liver and whole-liver allografts were similar.
Short-term graft and patient survival
Post-transplant hospital LoS was significantly longer in patients with higher MELD scores (median LoS: 11 days, 14 days and 15 days in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). Incidences of acute rejection episodes between transplantation and discharge were similar across groups (P = 0.350).
Graft survival at 30 days varied inversely with increasing MELD score (95%, 93% and 92% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). The same relationship was observed for graft survival at 90 days (92%, 88% and 87% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001), patient survival at 30 days (97%, 95% and 92% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001), and patient survival at 90 days (95%, 91% and 87% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). Figure1 depicts graft and patient survival curves.
Figure 1.
(a) Kaplan–Meier curves depicting liver allograft survival by Model for End-stage Liver Disease (MELD) score category (Mantel–Cox statistic 133.4, P < 0.001). (b) Kaplan–Meier curves for patient survival by MELD score category (Mantel–Cox log rank statistic 199.8, P < 0.001)
All subgroups had similar rates of graft non-function from all causes, regardless of increasing MELD score (12%, 11% and 12% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P = 0.009). Rates of treatment for rejection within 6 months or within 1 year of transplantation did not vary directly with increasing MELD score (10% and 10%, respectively, 7% and 11%, respectively, and 10% and 8%, respectively, in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). The most common cause of graft failure across the groups was primary graft failure (3%, 3% and 4% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.041). The incidence of recurrent hepatitis leading to graft failure was higher in the subgroup with MELD scores of <30 (3%, 2% and 2% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). The frequency of graft failure caused by infection was higher in the group with MELD scores of ≥40 (1%, 1% and 2% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). There was no difference across the groups in the frequency of graft failure caused by acute and chronic rejection (1% for all).
Longterm graft and patient survival
Graft survival at 1 year varied inversely with increasing MELD score (86%, 80% and 77% in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001). Graft survival at 3 years and 5 years showed the same trend (77% and 71%, 72% and 67%, and 69% and 64%, respectively; P < 0.001). This overall pattern was repeated for patient survival at 1 year (90%, 84% and 80% in recipients with MELD scores of <30, 30–39 and ≥40, respectively), 3 years (83%, 77% and 73%, respectively) and 5 years (77%, 72% and 69%, respectively) (all: P < 0.001) (Fig.1).
Survival analysis by MELD categories
A separate regression analysis to evaluate the survival hazard of increasing MELD score was carried out using the entire sample stratified by MELD score. In univariate Cox regression analysis, the hazard of graft failure (HRs 1.46, 1.24 and 1.00 in recipients with MELD scores of <30, 30–39 and ≥40, respectively; P < 0.001) and patient mortality (HRs 1.66, 1.34 and 1.00, respectively; P < 0.001) was significantly higher in the subgroup with MELD scores of 40–49. After adjusting for waiting list time, donor and recipient factors including age, gender and BMI, and graft characteristics such as a human leukocyte antigen (HLA) mismatch, share type, and total warm ischaemia time, the subgroup with MELD scores of 40–49 had a higher risk for mortality than the group with MELD scores of 30–39 (P = 0.006) and that with MELD scores of <30 (P = 0.004). Graft survival did not differ significantly among groups with MELD scores of <30 and 30–39 (P = 0.086), or 50–60 (P = 0.890).
Potential mortality risk predictors
Multiple univariate Cox regression analyses were performed in groups with MELD scores of ≥40 and <40, respectively, to determine potential predictors of mortality. As expected, the risk for mortality was higher in patients receiving livers from older donors and with MELD scores of ≥40 (P < 0.001). Donor obesity (P = 0.005) and diabetes (P < 0.001) increased the risk significantly. Patients with MELD scores of ≥40 who received livers from African-American donors had a slightly increased risk for mortality (HR 1.27, P < 0.001), whereas Hispanic ethnicity in the donor decreased the risk (HR 0.92, P = 0.011). Cause of donor death and type of liver (split or whole) were not significant predictors (P > 0.05). In both groups, recipient age of 40–59 years doubled (P < 0.001 and P < 0.001, respectively) and age of ≥70 years (P < 0.001) almost quadrupled the patient mortality risk. A requirement for ICU care between admission for transplantation and hospital discharge also increased the risk for mortality by 36% (P = 0.045). Diabetes for a duration of longer than 5 years (P < 0.001) and 10 years (P = 0.002) in the transplant recipient increased the mortality risk (HRs 1.69 and 6.28, respectively) compared with that in recipients without diabetes.
Among patients with MELD scores of ≥40, multivariate analysis adjusting for donor, recipient and graft characteristics confirmed donor and recipient age, a history of upper abdominal surgery, and need for life support at transplantation as significant negative predictors of graft survival. Recipient and donor age, history of upper abdominal surgery, and life support at transplantation were also significant negative predictors of patient survival (Table3).
Table 3.
Multivariate Cox regression analyses of graft and patient survival in transplant recipients with Model for End-stage Liver Disease (MELD) scores of ≥40
| Graft survival | Patient survival | |||||
|---|---|---|---|---|---|---|
| Factors | Levels | n | AHR (95% CI)a | P-value | AHR (95% CI)a | P-value |
| Recipient age, years | <18 | 109 | Ref | |||
| 18–39 | 276 | 1.78 (0.98–3.24) | 0.059 | 1.95 (0.95–4.03) | 0.070 | |
| 40–59 | 908 | 1.93 (1.08–3.45) | 0.027 | 2.41 (1.20–4.86) | 0.014 | |
| 60–69 | 277 | 2.63 (1.44–4.82) | 0.002 | 3.72 (1.81–7.63) | <0.001 | |
| ≥70 | 22 | 3.31 (1.34–8.20) | 0.010 | 4.65 (1.68–12.93) | 0.003 | |
| Donor age, years | <18 | 140 | Ref | Ref | ||
| 18–39 | 704 | 1.61 (1.00–2.58) | 0.048 | 1.53 (0.91–2.60) | 0.108 | |
| 40–59 | 564 | 2.08 (1.30–3.35) | 0.002 | 1.86 (1.10–3.13) | 0.020 | |
| 60–69 | 139 | 2.57 (1.51–4.40) | 0.001 | 2.28 (1.26–4.12) | 0.006 | |
| ≥70 | 45 | 2.59 (1.32–5.08) | 0.005 | 1.64 (0.72–3.74) | 0.241 | |
| Recipient ethnicity | White | 963 | ||||
| African-American | 201 | 1.12 (0.84–1.50) | 0.426 | 1.31 (0.95–1.89) | 0.099 | |
| Hispanic | 314 | 1.21 (0.94–1.57) | 0.148 | 0.79 (0.58–1.07) | 0.126 | |
| Asian | 85 | 1.47 (0.88–2.48) | 0.144 | 0.39 (0.20–0.76) | 0.006 | |
| Otherb | 29 | 1.37 (0.60–3.10) | 0.453 | 2.29 (1.23–4.25) | 0.009 | |
| Previous abdominal surgery | No | 1056 | Ref | |||
| Yes | 469 | 1.24 (1.00–1.54) | 0.052 | 1.27 (0.99–1.61) | 0.056 | |
| Gancyclovir | Yes | 542 | Ref | Ref | ||
| No | 1050 | 0.79 (0.64–0.98) | 0.008 | 0.72 (0.56–0.92) | 0.009 | |
| Valgancyclovir | Yes | 907 | Ref | Ref | ||
| No | 685 | 1.25 (1.00–1.54) | 0.044 | 1.18 (0.93–1.50) | 0.176 | |
| Life support at transplant | No | 1042 | Ref | Ref | ||
| Yes | 550 | 1.49 (1.21–1.83) | <0.001 | 1.61 (1.28–2.02) | <0.001 | |
| Share type | Local | 988 | Ref | |||
| Regional | 547 | 0.95 (0.77–1.18) | 0.668 | 0.94 (0.74–1.19) | 0.612 | |
| National | 56 | 1.42 (0.83–2.44) | 0.204 | 1.28 (0.67–2.44) | 0.456 | |
AHR: adjusted hazard ratio for donor and recipient age, gender, ethnicity, body mass index, diabetes status, graft type, share type, cold ischaemia time, waiting time, valgancyclovir, gancyclovir, previous abdominal surgery, life support at transplantation.
Other: American Indian, Alaskan Native, Native Hawaiian, Pacific Islander, unknown and mixed race.
Post-transplant futility analysis
Inclusion criteria for the futility analysis were a MELD score of ≥40 and death during the study period (n = 660). Patients who were lost from follow-up or who survived for longer than the study period were excluded (n = 1950). Those who died within 3 months of transplant were designated as ‘futile’ (n = 282), and the others were designated as ‘non-futile’ (n = 378). Table4 shows the characteristics of patients in the futile and non-futile groups.
Table 4.
Demographic and clinical characteristics of transplantations in recipients with Model for End-stage Liver Disease (MELD) scores of ≥40 by futility status
| Characteristics | Futile | Non-futile | P-value |
|---|---|---|---|
| (n = 282) | (n = 378 | ||
| Recipient age, years, median (IQR) | 53 (16) | 53 (12) | 0.953 |
| Recipient age categories, n (%) | |||
| ≤17 years | 14 (5%) | 8 (2%) | 0.007 |
| 18–39 years | 48 (17%) | 47 (12%) | |
| 40–59 years | 145 (51%) | 237 (63%) | |
| 60–69 years | 67 (24%) | 83 (22%) | |
| ≥70 years | 11 (4%) | 3 (1%) | |
| Recipient gender male, n (%) | 181 (64%) | 248 (66%) | 0.704 |
| Donor age, years, median (IQR) | 43 (29) | 42 (29) | 0.831 |
| Donor gender male, n (%) | 168 (60%) | 219 (58%) | 0.673 |
| Recipient race, n (%) | |||
| White | 187 (66%) | 246 (65%) | 0.762 |
| Hispanic | 39 (14%) | 64 (17%) | |
| African-American | 39 (14%) | 45 (12%) | |
| Asian | 8 (3%) | 13 (3%) | |
| Mixed/unknown/other | 9 (3%) | 10 (3%) | |
| Recipient BMI, kg/m2, median (IQR) | 27.3 (8.8) | 28.2 (7.4) | 0.186 |
| Recipient obesity (BMI ≥30 kg/m2), n (%) | 89 (32%) | 97 (26%) | 0.096 |
| Donor BMI, 30 kg/m2, median (IQR) | 26.2 (8.3) | 25.8 (6.7) | 0.200 |
| Donor obesity (BMI ≥30 kg/m2), n (%) | 131 (47%) | 186 (49%) | 0.484 |
| Dialysis in week prior to transplant, n (%) | 147 (52%) | 204 (54%) | 0.639 |
| Previous upper abdominal surgery, n (%) | 91 (32%) | 131 (35%) | 0.555 |
| Diabetes mellitus, n (%) | 52 (18%) | 96 (25%) | 0.336 |
| Recipient on life support, n (%) | 138 (49%) | 122 (32%) | <0.001 |
| Medical condition at transplant, n (%) | |||
| ICU | 181 (64%) | 197 (52%) | 0.002 |
| Hospitalized but not ICU | 87 (31%) | 144 (38%) | |
| Not hospitalized | 13 (5%) | 37 (10%) | |
| Recipient on gancyclovir | 75 (51%) | 88 (37%) | 0.007 |
| Recipient on valgancyclovir | 59 (40%) | 126 (53%) | 0.013 |
| Extended criteria donors, n (%) | 74 (27%) | 94 (25%) | 0.673 |
| Living donors, n (%) | 7 (2%) | 10 (3%) | 0.896 |
| Donor race, n (%) | |||
| White | 159 (56%) | 247 (65%) | 0.043 |
| Hispanic | 57 (20%) | 67 (18%) | |
| African-American | 16 (6%) | 8 (2%) | |
| Asian | 16 (6%) | 8 (2%) | |
| Mixed/unknown/other | 6 (2%) | 4 (1%) | |
| Donor diabetes mellitus, n (%) | 34 (12%) | 39 (10%) | 0.405 |
| Deceased donor cause of death, n (%) | |||
| Head trauma | 38 (4%) | 68 (18%) | 0.538 |
| Stroke | 138 (49%) | 165 (44%) | |
| Anoxia | 95 (34%) | 132 (35%) | |
| Other/unknown | 7 (2%) | 10 (3%) | |
| Share type, n (%) | |||
| Local | 172 (61%) | 264 (70%) | 0.014 |
| Regional | 91 (32%) | 105 (28%) | |
| National | 18 (6%) | 9 (2%) | |
| Regional | 1 (0.4%) | 0 | |
BMI, body mass index; ICU, intensive care unit; IQR, interquartile range.
Futile transplant recipients were more likely to be aged <40 years (22% versus 15%; P = 0.007) or >60 years (27% versus 23%; P = 0.007) than their non-futile counterparts. They were more likely to have received liver allografts of regional (32% versus 28%; P < 0.001) or national (6% versus 2%; P = 0.014) share type than a local share type (61% versus 70%; P = 0.014). Recipient gender (P = 0.704), ethnicity (P = 0.762) and BMI (P = 0.597) were equally distributed across the futile and non-futile groups. Futile donors were more likely to be of non-White ethnicity (African-American: 16% versus 14%; Hispanic: 20% versus 18%; Asian: 6% versus 2%; Other: 2% versus 1%; P = 0.043). Futile and non-futile groups were similar in terms of other donor characteristics including age, gender, BMI, presence of diabetes or hypertension, and cause of death.
At the time of listing, futile transplant recipients (n = 282) were less likely to be hospitalized, but more likely to be mechanically ventilated if hospitalized (21% versus 12%; P = 0.002). There was no difference in grade III and IV encephalopathy, risk for portal vein thrombosis, spontaneous bacterial peritonitis, marked muscle wasting, history of TIPS, or previous upper abdominal surgery. Furthermore, futile patients were more likely to have required ICU admission (64% versus 52%; P = 0.002) and mechanical ventilation (43% versus 26%; P < 0.001). However, they were less likely to have mild to moderate ascites (83% versus 89%; P = 0.002), or suffer from spontaneous bacterial peritonitis (8% versus 10%; P = 0.023) (Table4).
In the futile group, 21 patients experienced primary graft failure. Other causes of graft failure were infection (n = 12), vascular thrombosis (n = 8), biliary complications (n = 4), and hepatitis recurrence (n = 4). In the non-futile group, the most common cause of graft failure was hepatitis recurrence (n = 42), followed by infection (n = 24), primary graft failure (n = 30), vascular thrombosis (n = 12), non-hepatitis disease recurrence (n = 17), acute rejection (n = 9) and biliary complications (n = 7). The average length of post-transplant hospitalization was slightly longer in the futile group (26 days versus 24 days; P < 0.001). Of the futile transplants, 23.8% of recipients (n = 67) underwent retransplantation after experiencing graft failure. As expected, discharge laboratory values revealed significantly increased levels of aspartate aminotransferase (AST) (P < 0.001) and alanine aminotransferase (ALT) (P < 0.001) in the futile group.
In univariate Cox regression analysis, patients with futile transplants had a significantly higher risk for graft loss (HR 133.06, 95% CI 92.00–192.45; P < 0.001) and mortality (HR 133.87, 95% CI 77.82–166.62; P < 0.001) In a univariate binomial logistic regression analysis, recipients aged <40 years had a significantly higher risk for futility than those aged 40–60 years (P = 0.004). Compared with liver allografts from White donors, receipt of a graft from an Asian donor conferred a 3.12-fold increased futility risk (95% CI 1.30–7.43; P = 0.011). Liver allografts from the national share were associated with significantly increased futility [odds ratio (OR) 3.07, 95% CI 1.35–6.99; P = 0.008]. Recipients requiring ICU care prior to transplantation had a more than two-fold greater risk for futility than those not hospitalized at the time of transplant (OR 2.62, 95% CI 1.35–5.08; P = 0.005). In the same vein, requirements for mechanical ventilation or inotropic drugs at transplantation were significantly associated with futility (OR 2.09, 95% CI 1.46–2.76; P < 0.001). Patients on gancylovir were more likely to have futile transplants (OR 1.76, 95% CI 1.16–2.68; P = 0.008), whereas those on valgancyclovir were associated with the opposite outcome (OR 0.59, 95% CI 0.39–0.90; P = 0.002). At discharge, increased INR (OR 4.48, 95% CI 2.83–7.09; P < 0.001) significantly increased futility. Cold ischaemia time, total warm ischaemia time, and waitlist time did not affect futility. Donor age, gender, diabetes status, cause of death, and distance from the transplant hospital also did not predict futility.
Discussion
Over the past decade, the gradual rise in the number of liver transplant registrants with very high MELD scores has led to increasing interest in outcomes among this patient population. Currently, 8% of liver transplants are performed in patients with MELD scores of ≥40. The few single-centre studies to have addressed this topic have demonstrated mixed outcomes. Some report outcomes comparable with those of transplants in patients with lower MELD scores,3 whereas others note inferior survival.7–10 This is the first nationwide longitudinal study to explore survival outcomes and futility predictors in a cohort of patients with MELD scores of ≥40 who were transplanted during the period from 2002 to 2011. The study confirms the current literature on graft and patient survival among patients with high MELD scores. Although findings showed that overall mortality was statistically higher among patients with MELD scores of ≥40, the difference was clinically less significant: the cohort with MELD scores of ≥40 recorded a 30% mortality rate, whereas the cohort with MELD scores of <30 recorded a 26% mortality rate.
In the present sample, the better than expected survival in the cohort with MELD scores of ≥40 may be attributed to the younger age of recipients, and lower prevalences of diabetes, portal vein thrombosis, hepatitis C virus and Epstein–Barr virus (EBV) infection. These patients were also less likely to have required a TIPS procedure for portal hypertensive bleeds, a marker that is not reflected in the MELD score. They were less likely to have had previous upper abdominal surgery. The cohort also had a higher proportion of allografts of regional share type, indicating the prioritization of these patients over other candidates on the basis of acuity. Despite having fewer preoperative comorbid conditions, these recipients required more health care resources, represented by pre-transplant hospitalization, ICU admission, mechanical ventilation and a longer LoS.
The benefits of transplantation in a patient with a MELD score of >40 must be weighed against both the greater risks and increased resource utilization required. Patients with high MELD scores have been found to have increased incidences of post-transplant infection, longer ICU and general hospital stays, and imply increased overall costs.3,8 Several single-institution reports have demonstrated poorer post-transplant survival among recipients stratified by increasing MELD scores.10–13 However, Alexopoulos et al. reported a 1-year survival rate of 89% among patients with MELD scores of >40, which was similar to the 1-year survival rate among patients with MELD scores of <40.5 The 1-year patient survival rate identified in the present study is, at 80%, inferior to that reported by Alexopoulos et al.,5 which implies that nationally patients with MELD scores of >40 do slightly worse overall than those with lower MELD scores.
Many studies demonstrate increased perioperative complications and resource utilization in transplant recipients with higher MELD scores. Axelrod et al.9 found that patients with MELD scores of 30 utilized 10 times more Medicare spending than those with MELD scores of 20.10 Given that average annual spending on an end-stage liver disease patient is reported to be US$22 424,10 prioritizing of the most critically ill patients for transplantation is increasingly important. In the present cohort of recipients with MELD scores of ≥40, nearly half of all patients received pre-transplant dialysis. More than half were in the ICU prior to transplantation, and one-third needed some modality of life support, including mechanical ventilation and inotropic drugs, during their hospital stay. This finding is consistent with those of other studies5 and confirms that patients with MELD scores of >40 utilize a disproportionately higher amount of health care resources. Nonetheless, the rate of retransplantation attributable to graft failure, the rate of post-transplant infection leading to graft failure, and postoperative LoS were all lower in the present study than in that reported by Alexopoulos et al.5 These findings suggest that transplantation in patients with very high MELD scores may be sound and necessitate further investigation.
It is important to identify donor and recipient outcome predictors to determine which of the patients with MELD scores of ≥40 will survive transplantation. The present regression analysis revealed that a recipient age of 60–69 years, diabetes duration of >10 years, African-American ethnicity, donor diabetes and obesity were significant negative predictors. Although Alexopoulos et al. attributed poor outcomes among patients with MELD scores of ≥40 to the deleterious effects of prolonged pre-transplant waiting,5 the present study did not find waiting time to be a predictor of futility.
Although the difference in overall and 1-, 3- and 5-year mortality among MELD subgroups was not large, 30-day mortality was found to increase significantly as MELD scores increased. Patients with MELD scores of ≥40 were more than twice as likely to die within 30 days of transplant as those with MELD scores of <30. Thus, the present futility analysis reveals that there is a subgroup of patients with MELD scores of ≥40 who will not benefit from transplantation, and whose allocated organs may be better used in patients with more favourable prognoses. The present study found that recipient age of >60 years, a BMI of >30 kg/m2, requirements for ICU care or life support on the waiting list, and liver allograft of national share, were predictors of futility. Furthermore, high INR, ALT and AST values at the time of discharge were also positive predictors of futility. As these were postoperative laboratory values, they will not help in patient selection, but will be useful in helping the clinician to monitor such patients more closely. In Cox regression analysis, after controlling for age, gender, BMI, HLA mismatch, cold ischaemia time, total warm ischaemia time, waiting time, share type, and cause of donor death, patients with MELD scores of 40–49 were found to have a risk for mortality that was increased by 95% and 29% over those with MELD scores of <30 and of 30–39, respectively. After adjusting for confounders, patients with MELD scores of 40–49 had a graft failure rate that was 60% and 28% higher than those with MELD scores of <30 and 30–39, respectively. Furthermore, patients in this futile transplant subgroup were more likely to experience preoperative acute comorbidities such as portal vein thrombosis, renal failure requiring dialysis prior to transplantation, transfusions for a portal hypertensive bleed, history of TIPS and prior upper abdominal surgery, as is representative of a sicker population with increased utilization of health care resources.
Overall, the present findings suggest that candidates with MELD scores of ≥40 on the liver transplant waiting list can derive benefit from the receipt of an allograft. Despite significantly higher futility (43%, 32% and 21% in patients with MELD scores of ≥40, 30–39 and <30, respectively; P < 0.001), increased overall mortality (25%, 22% and 20% in patients with MELD scores of ≥40, 30–39 and <30, respectively) and a probable increase in health care expenses as a result of the provision of longer post-transplant inpatient care, these patients meet the expected survival of >50% at 5 years (64% graft survival and 69% patient survival) and therefore warrant attention.14 In characterizing the subset of patients with MELD scores of ≥40 among whom transplantation may be futile, the present authors conclude that patients aged >60 years, patients with obesity, those who are hospitalized in the ICU requiring ventilation, and those with multiple comorbidities are at risk for futile transplantation. Additionally, postoperative laboratory studies should be used to identify patients at high risk for adverse outcomes, and should demand greater clinician attention. Despite having fewer preoperative comorbid conditions, the futile group required more health care resources such as peri-transplant hospitalization, ICU care and life support measures.
Strengths and limitations
The present study has several strengths. The most significant is that it is a population-based effectiveness study that used a robust national database and a follow-up period extending to almost a decade to elucidate an inadequately studied issue. The study addressed the timely topic of surgical outcomes in patients with MELD scores of ≥40 and the causes of futility in this particular group, in comparison with groups with lower MELD scores. The study also adds substantial information on prognosis after liver transplantation in the most critically ill patients as defined by UNOS, which may assist clinicians in making decisions on the allocation of liver allografts. This information has the potential to influence future guidelines on managing this particular type of patient. Additionally, because it is a longitudinal study, the present study is devoid of selection bias. Two control groups, consisting of patients with MELD scores of, respectively, <30 and 30–39, were used to compare survival.
The study has a few limitations. Because of its retrospective nature, confounding attributable to unobserved variables is more likely. This limits the present authors’ ability to comment accurately on the more detailed variables such as primary non-function, post-transplant ICU stay and utilization of rehabilitation services, or on direct causes of futility. Additionally, the study design was not that of a randomized prospective clinical trial. Hence, potential bias may have been introduced in addition to the instability of the univariate and multivariate models.
Conclusions
Although the benefits of liver transplantation in patients with MELD scores of ≥40 must be weighed against the increases in risk and resource utilization, liver transplantation in recipients with MELD scores of ≥40 demonstrates a significant increase in post-transplant survival that is greater than expected. The subset of patients with higher MELD scores in whom transplant may have inferior survival outcomes includes those aged >60 years, those with obesity, multiple comorbidities or pre-transplant ICU hospitalization, and those receiving allografts from obese or expanded criteria donors. Future prospective studies are required to further characterize the subset of patients with MELD scores of ≥40 who are most likely to benefit from transplantation in order to continue to improve the allocation of this vital resource and to avoid futile transplantation in the most critically ill patients.
Funding sources
This work was supported in part by the Health Resources and Services Administration (contract 234-2005-370011C). The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does any mention of trade names, commercial products or organizations imply endorsement by the US Government.
Conflicts of interest
None declared.
References
- Kim HJ, Lee HW. Important predictor of mortality in patients with end-stage liver disease. Clin Mol Hepatol. 2013;19:105–115. doi: 10.3350/cmh.2013.19.2.105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang VS, Saab S. Liver transplantation in the era of model for end-stage liver disease. Liver Int. 2004;24:1–8. doi: 10.1111/j.1478-3231.2004.00888.x. [DOI] [PubMed] [Google Scholar]
- Health Resources and Services Administration, US Department of Health and Human Services. 2013. OPTN: Organ Procurement and Transplantation Network. Available at http://optn.transplant.hrsa.gov/policiesAndBylaws/policies.asp (last accessed 9 November 2014)
- US Department of Health and Human Services, Organ Procurement and Transplantation Network. 2013. United States Organ Transplantation: Annual Data Report 2011, in OPTN/SRTR Annual Data Report… Available at http://srtr.transplant.hrsa.gov/annual_reports/2012/pdf/03_liver_13.pdf; Page# 71; Also Fig 4-9 on page# 82 (last accessed 9 November 2014)
- Alexopoulos S, Matsuoka L, Cho Y, Thomas E, Sheikh M, Stapfer M, et al. Outcomes after liver transplantation in patients achieving a model for end-stage liver disease score of 40 or higher. Transplantation. 2013;95:507–512. doi: 10.1097/TP.0b013e3182751ed2. [DOI] [PubMed] [Google Scholar]
- Trotter JF, Osgood MJ. MELD scores of liver transplant recipients according to size of waiting list: impact of organ allocation and patient outcomes. JAMA. 2004;291:1871–1874. doi: 10.1001/jama.291.15.1871. [DOI] [PubMed] [Google Scholar]
- Merion RM, Schaubel DE, Dykstra DM, Freeman RB, Port FK, Wolfe RA, et al. The survival benefit of liver transplantation. Am J Transplant. 2005;5:307–313. doi: 10.1111/j.1600-6143.2004.00703.x. [DOI] [PubMed] [Google Scholar]
- Petrowsky H, Rana A, Kaldas FM, Sharma A, Hong JC, Agopian VG, et al. Liver transplantation in highest acuity recipients: identifying factors to avoid futility. Ann Surg. 2014;259:1186–1194. doi: 10.1097/SLA.0000000000000265. [DOI] [PubMed] [Google Scholar]
- Axelrod DA, Dzebisashvili N, Lentine K, Segev DL, Dickson R, Tuttle-Newhall E, et al. Assessing variation in the costs of care among patients awaiting liver transplantation. Am J Transplant. 2014;14:70–78. doi: 10.1111/ajt.12494. [DOI] [PubMed] [Google Scholar]
- Asrani SK, Kim WR, Edwards EB, Larson JJ, Thabut G, Kremers WK, et al. Impact of the center on graft failure after liver transplantation. Liver Transpl. 2013;19:957–964. doi: 10.1002/lt.23685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Saab S, Wang V, Ibrahim AB, Durazo F, Han S, Farmer DG, et al. MELD score predicts 1-year patient survival post-orthotopic liver transplantation. Liver Transpl. 2003;9:473–476. doi: 10.1053/jlts.2003.50090. [DOI] [PubMed] [Google Scholar]
- Onaca NN, Levy MF, Sanchez EQ, Chinnakotla S, Fasola CG, Thomas MJ, et al. A correlation between the pretransplantation MELD score and mortality in the first two years after liver transplantation. Liver Transpl. 2003;9:117–123. doi: 10.1053/jlts.2003.50027. [DOI] [PubMed] [Google Scholar]
- Jacob M, Copley LP, Lewsey JD, Gimson A, Toogood GJ, Rela M, et al. Pretransplant MELD score and post liver transplantation survival in the UK and Ireland. Liver Transpl. 2004;10:903–907. doi: 10.1002/lt.20169. [DOI] [PubMed] [Google Scholar]
- Neuberger J, Gimson A, Davies M, Akyol M, O'Grady J, Burroughs A, et al. Selection of patients for liver transplantation and allocation of donated livers in the UK. Gut. 2008;57:252–257. doi: 10.1136/gut.2007.131730. [DOI] [PubMed] [Google Scholar]