This study was conducted in compliance with the approval of our Institutional Review Board (IRB). The authors have no commercial or other association that would pose a conflict of interest in the writing of this manuscript.
The donor organ shortage is a well-recognized problem in liver transplantation (LT). The increasing liver transplant wait-list mortality in the United States prompted the United Network for Organ Sharing (UNOS) to revise the organ allocation system in 2002 in an attempt to address this problem.1 Since February 2002, allocation of cadaveric livers in the United States has been based on the Model for End-stage Liver Disease (MELD) score rather than time spent on the waiting list. The MELD score gives individuals with the highest predicted short-term mortality the top priority for organ allocation.2, 3 Specifically, the MELD scoring system gives priority for transplantation to patients with an elevated serum creatinine. This feature of the MELD system prioritizes LT in cirrhotic individuals with renal dysfunction.4
High pre-transplant MELD score has been associated with increased mortality in the first 2 years after LT in several cohort studies.5–7 Pre-transplant renal function has also been demonstrated to be a major determinant of patient survival after LT.8 The development of post-LT renal disease may negatively impact patient survival. In this regard, serum creatinine elevation during the first year after transplantation is a reliable predictor of post-LT renal failure.9, 10 Recently, Pawarode et al. suggested that pre-transplant renal dysfunction predicts chronic renal disease (CRD) after LT.11 As the MELD score is heavily weighted by pre-LT renal dysfunction, there is concern that organ allocation based on the MELD score could adversely affect long-term patient survival and/or increase the requirement for hemodialysis or kidney transplantation after LT. The aims of our study were to determine the impact of the MELD score implementation on the occurrence of post-LT renal disease in the pre-MELD as compared to the post-MELD cohort.
MELD, model for end-stage liver disease; UNOS, United Network for Organ Sharing; LT, liver transplantation; HCV, hepatitis C virus; HCC, hepatocellular carcinoma; ACR, acute cellular rejection; DM, diabetes mellitus; CRD, chronic renal disease; MDRD, modification of diet and renal disease; HTN, hypertension.
PATIENTS AND METHODS
We conducted a non-concurrent single-center cohort study all potential transplant recipients listed for solitary LT at Shands Hospital at the University of Florida Transplant Program (Gainesville, FL) between February 2001 and January 2003. The study was approved by the University of Florida Institutional Review Board. Patients who underwent LT during the year prior to the implementation of MELD scoring system (February 2001 to January 2002, pre-MELD cohort) were compared to those patients who underwent transplantation the year following the implementation of MELD (March 2002 to January 2003, MELD cohort). Of the 184 patients listed for LT, a total of 174 consecutive adult patients (66 women and 108 men, mean age 50 ± 11 years) underwent solitary LT during the study period. Patients who received a prior solid organ transplant or a combined liver and kidney transplant were excluded from the study. Patients were followed for a minimum of 24 months after LT or until graft loss or death occurred. All data analyzed were recorded prospectively in the University of Florida Liver Transplant Database but were reviewed retrospectively.
The MELD score was calculated for each subject using the UNOS formula based on laboratory values obtained within six hours prior to transplant surgery.12 The calculated MELD score used for the purposes of this analysis did not necessarily equal the actual MELD score used for wait listing [(the listing MELD score may take into account exceptions such as hepatocellular carcinoma (HCC) or hepatopulmonary syndrome)].
Clinical conditions frequently associated with CRD were identified in each patient at time of onset, either pre- or post-LT. Diabetes mellitus (DM) was defined by a fasting serum glucose concentration above 126 mg/dl on two separate occasions, or the requirement of treatment with insulin or oral hypoglycemic medication.13 Arterial hypertension (HTN) was defined by a sustained elevation in the systolic blood pressure above 140 mmHg, a sustained elevation of the diastolic blood pressure above 90 mmHg, and/or the use of antihypertensive medications.14
Organ procurement was carried out with aortic and portal perfusion with University of Wisconsin solution.15 All transplants were performed using the piggyback technique, without veno-venous bypass.16 LT was considered with a concomitant diagnosis of HCC only if the validated Milan criteria were met.17 Patients with concomitant HCC received pre-transplant chemoembolization as per protocol.18
Assessment of Renal Function
Renal function was assessed by measurement of fasting serum creatinine within 6 hours prior to LT, and at 6, 12, and 24 months after LT. Creatinine clearance (ClCr) estimation was performed in every patient using the abbreviated Modification of Diet and Renal Disease (MDRD) equation, corrected for patient's gender and race.19 Twenty-four hour urine collection for ClCr measurement was performed in any patient with an estimated ClCr lower than 50 mL/min/1.73m2 or a rising serum creatinine. Stage 3 Chronic Renal Disease (CRD-3) was defined by an estimated ClCr below 60 mL/min/1.73m2.20 Stage 4 Chronic Renal Disease (CRD-4) was defined by an estimated ClCr below 30 mL/min/1.73m2.20 Additional end-points evaluated during the study period included requirement of kidney transplantation and need for hemodialysis support after LT. All patients who had at least one session of hemodialysis or had been continued on veno-venous hemofiltration after LT were designated as having undergone hemodialysis.
Our standard immunosuppressant regimen during this study consisted of a calcineurin inhibitor, either tacrolimus (Prograf®, Fujisawa Healthcare Inc, Deerfield, IL) or cyclosporine (Neoral®, Novartis Pharmaceutical Corporation, East Hanover, NJ) and prednisone. Tacrolimus initial dose was 0.08-0.12 mg/kg/day orally in two divided doses with goal target trough whole blood concentrations of 10-15 ng/mL for the first month post-LT followed by 5-10 ng/mL thereafter. Cyclosporine initial dose was 2-4 mg/kg/day orally in two divided doses with goal target trough whole blood concentrations of 200-250 ng/mL for the first month post-LT followed by 150-200 ng/mL thereafter. Prednisone was tapered and discontinued within 4 months following LT. Mycophenolate mofetil was initiated as a third agent in patients with autoimmune-induced liver disease (autoimmune hepatitis, primary biliary cirrhosis, and primary sclerosing cholangitis), history of recurrent acute cellular rejection, or calcineurin inhibitor related toxicity preventing attainment of therapeutic levels.
A change in the standard immunosuppressive regimen was implemented for those patients who had pre-existing or developed significant renal dysfunction after LT. Two calcineurin-inhibitor sparing protocols were used at our center. The first one included the use of mycophenolate mofetil in combination with low-dose tacrolimus, adjusted to a trough level of 3-5 ng/mL. The later included the use of sirolimus, followed by low-dose tacrolimus upon improvement on renal function targeted to similar trough levels.
Acute cellular rejection (ACR) was defined based upon histological assessment of liver tissue. ACR was graded as mild, moderate or severe based on the criteria for global assessment proposed by Demetris et al.21 Patients with biopsy-documented moderate to severe ACR were treated with steroids. Initial treatment for ACR consisted of intravenous bolus of methylprednisolone (total dose 1 to 3 gm) divided on three alternate-day doses. Muromonab (OKT3®) antibody infusion was reserved for patients with ACR resistant to intravenous corticosteroids.
Categorical variables were summarized as frequencies (percentages) and analyzed using chi-square test or Fisher's exact test for categorical data. Unless otherwise specified, baseline characteristics were summarized as median and interquartile range for continuous variables, and values were compared using the Mann-Whitney or t-test as appropriate. Patient and graft survival were estimated using the Kaplan-Meier method. For the purposes of this analysis, CRD-3 and CRD-4 were used as a composite endpoint. The time to occurrence of CRD-3/CRD-4 was then analyzed using the Kaplan-Meier method. We then analyzed the effect of baseline covariates on the time to occurrence of CRD-3/CRD-4 in Cox proportional hazards regression.22 This multivariate model was stratified by cohort (Pre-MELD vs. MELD). We chose as covariates for inclusion in this model 1) baseline serum creatinine, 2) baseline presence of diabetes mellitus, 3) baseline presence of hypertension, 4) hepatitis C, and 5) recipient age at time of LT. These covariates were chosen both on the basis of their presumed pathophysiologic relevance to CRD risk and attained levels of significance in the preceding univariate analyses. For the purposes of modeling, we divided baseline serum creatinines into two arbitrary groups (Cr ≤ 1.5 mg/dL and Cr > 1.5 mg/dL). Age was divided arbitrarily into two groups, age ≤ 60 vs. > 60 years. All statistical analyses were performed using were completed using SPPS Version 13 (SPSS, Chicago, IL).
A total of 182 patients were listed for LT during the study period, 102 patients prior to implementation of MELD allocation system, and 82 patients during the year following the implementation of the MELD allocation system. Death and removal from the wait-list for becoming “too sick to transplant” in the pre-MELD vs. MELD cohort occurred for 11 vs. 15 patients (P =NS) and 4 vs. 6 (P = NS), respectively. Two patients in each cohort had received a combined kidney and liver transplant and were excluded from the analysis due to inability to assess impact of LT on native kidney function. The study group analyzed consisted of 174 adult solitary LT recipients (66 female and 108 male) with a mean age of 50 ± 11 years (range 16-68 years). The pre-MELD cohort included 97 LT recipients and the MELD cohort included 77 LT recipients. Baseline demographic characteristics were comparable between both groups as depicted in Table 1. The main indication for LT in both groups was hepatitis C (HCV) related liver disease (50% vs. 53%, P = NS). The prevalence of HCC at time of transplantation in the pre-MELD cohort was 7% vs. 31% in the MELD cohort (P < 0.05). Mean calculated MELD score in the pre-MELD cohort was significantly lower than in the MELD cohort (16 vs. 19, P < 0.05). Post-LT variables, including immunosuppressive regimen were comparable in both cohorts, as shown in Table 2. The prevalence of co-morbidities associated to chronic renal disease (HTN and DM) before and after transplantation was similar between both groups.
Table 1. Pre-Transplantation Patient Characteristics According to MELD Cohort
Pre-MELD (n = 97)
MELD (n = 77)
Data presented as mean (range).
LT, liver transplantation; MELD, model for end-stage liver disease.
Table 2. Post-Transplantation Patient Characteristics According to MELD Cohort
Pre-MELD (n = 97)
MELD (n = 77)
*Data presented as mean (range).
MELD, model for end-stage liver disease.
Mycophenolate mofetil use
Diabetes mellitus at 2 years
Hypertension at 2 years
Hemodialysis and/or kidney transplantation at 2 years
No difference was observed in patient [Fig. 1] and graft survival (not shown) between pre-MELD and MELD cohorts at 2 years post-LT. Early mortality, defined as 90-day post-LT mortality, was also comparable between the two cohorts. Repeat transplantation was required in 4 individuals (4%) in the pre-MELD cohort compared with 2 (3%) in the MELD cohort (P = NS).
Impact of MELD Implementation on Renal Function After LT
Renal function between cohorts at baseline and during follow-up is depicted in Table 3. The mean serum creatinine at time of LT was higher in the MELD cohort in comparison with pre-MELD cohort (1.6 vs. 0.9, P < 0.05). Likewise, the estimated ClCr at time of LT was lower in the MELD cohort as compared with the pre-MELD cohort (75 vs. 95 mL/min/1.73m2, P < 0.01). During the follow-up period, serum creatinine and the estimated ClCr calculated by the MDRD equation were not statistically different at 6, 12 and 24 months after LT between the cohorts.
Table 3. Assessment of Renal Function according to MELD Cohort
Pre-MELD (n = 97)
MELD (n = 77)
Estimated creatinine clearance calculated using MDRD Equation.
All results expressed as mean ± S.D.
Serum creatinine (mg/dL)
0.9 ± 0.4
1.6 ± 1.5
6 months post-LT
1.2 ± 0.6
1.3 ± 1
12 months post-LT
1.2 ± 0.6
1.3 ± 1
24 months post-LT
1.4 ± 1.3
1.3 ± 1
Estimated creatinine clearance*
95 ± 42
75 ± 42
6 months post-LT
73 ± 33
71 ± 26
12 months post-LT
75 ± 33
71 ± 27
24 months post-LT
74 ± 36
70 ± 21
The prevalence of CRD-3 and CRD-4 in pre-MELD and MELD cohort at different time intervals are illustrated in Table 4. The prevalence of CRD-3 before LT was significantly lower in the pre-MELD cohort in comparison with the MELD cohort (15.5% vs. 40.2%, P < 0.02). However, the prevalence of CRD-3 at 6, 12, and 24 months after LT was similar between the cohorts. Similarly, the prevalence of CRD-4 before LT was significantly lower in the pre-MELD cohort compared with the MELD cohort (3% vs. 19.5%, P < 0.05), although the prevalence of CRD-4 at 6, 12, and 24 months after LT was comparable between both cohorts.
Table 4. Prevalence of Chronic Renal Disease Before and After Transplantation According to MELD Cohort
Pre-MELD (n = 97)
MELD (n = 77)
LT: liver transplantation; MELD, model for end-stage liver disease.
Stage 3 Chronic Renal Disease
6 months post-LT
12 months post-LT
24 months post-LT
Stage 4 Chronic Renal Disease
6 months post-LT
2 months post-LT
24 months post-LT
Univariate analysis did not reveal any difference in the cumulative incidence of CRD-3/CRD-4 (Fig. 2) between cohorts. Similarly, calculated MELD score at time of LT, HCV infection or HCC diagnosis were not associated with the development of CRD-3 during the first 2 years after LT by univariate analysis. Age at time of LT, baseline DM, baseline HTN, and serum creatinine at LT were associated with the development of CRD-3/ CRD-4 in the first 2 years after LT by univariate analysis. Interestingly, the presence of HCV infection was not associated with an increased risk for CRD 3/CRD-4 (Table 5). In the multivariate analysis, an elevated baseline creatinine (above 1.5 mg/dL) did not confer any increased risk for the development of CRD-3/CRD-4 (Hazard ratio, HR 0.43, 95% CI: 023-0.84; P = 0.014). In fact, an increased creatinine at time of transplant appears to have a protective effect against the risk for development of CRD-3/CRD-4. Notably, increasing age was associated in an increased risk for CRD-3/CRD-4 (HR =5.22; 95 % CI:1.25-21.91; P = 0.024) (Table 6). The requirement for renal replacement therapy (chronic dialysis or kidney transplantation) was similar between both groups (3% vs. 3%, P = NS). During the study period one patient in each group required combined liver and kidney transplantation. One patient in the pre-MELD cohort required kidney transplantation during follow-up. Three patients in the pre-MELD cohort and two patients in the MELD cohort required chronic hemodialysis.
Table 5. Univariate Analysis of Risk Factors for Development of CRD-3/CRD-4
The MELD score is a validated objective scoring system that predicts short-term mortality in cirrhotic patients.2, 3 The MELD score was originally designed to assess short-term prognosis in cirrhotic patients undergoing transjugular intrahepatic portosystemic shunt.23 The MELD scoring system has been also used to predict mortality in patients with alcoholic hepatitis and in cirrhotic individuals undergoing elective surgery.24, 25 In February 2002, the use of the MELD score was implemented by UNOS nationwide in an effort to improve the allocation system for LT and to reduce mortality among patients on the LT waiting list.4 One of the three variables used in this scoring system is serum creatinine. Based on its mathematical formula, the MELD scoring system heavily weighs in favor of LT candidates with concomitant renal dysfunction. This fact has raised concerns as to whether the implementation of the MELD allocation system would result in an increase in the occurrence of chronic renal disease following LT and associated decreases in post LT survival rates. To date, the impact of MELD on the occurrence of chronic renal disease after LT remains unclear.
Our single transplant center experience suggests that the implementation of MELD has not significantly influenced the occurrence of CRD in the first 2 years post-LT in comparison to the pre-MELD period. Although those patients transplanted at our center during the MELD period clearly had worse renal function at the time of LT, no difference in long-term renal sequelae such as the need for long-term hemodialysis and/or subsequent kidney transplant was observed between the two cohorts. Though a statistically significant difference in MELD scores (16 vs. 19) existed between cohorts, this difference does not appear to be associated with a clinical difference in clinical outcome. A valid hypothesis to explain our findings is that the MELD score may prioritize those patients for transplantation with reversible renal dysfunction. However, this assumption may be applicable only in the setting of a comprehensive pre-LT selection process, such as the one implemented at our transplant center, which excludes from solitary LT those individuals with preexisting chronic renal dysfunction unrelated to chronic liver disease. Most of our LT candidates with elevated serum creatinine before LT were likely related to either hepatorenal syndrome or to reversible causes of pre-renal azotemia such as intravascular volume-depletion, diuretic use, or contrast-induced nephropathy. In fact, a negative association between elevated baseline creatinine and risk of development of CRD was noted in our multivariate analysis. Reversibility and/or improvement of pre-transplant renal function could explain these findings. However, causes for pre-transplant azotemia could not be fully characterized in this retrospective study.
Our study also suggests that implementation of MELD score has had no significant influence on 2-year patient and graft survival after LT. Although those patients transplanted in the MELD cohort had worse pre-transplant renal function, the implementation of MELD did not correlate with either patient or graft survival. This finding should be interpreted in the context of several potential confounding factors. First, the severity of the underlying liver disease at time of LT in our study may not reflect the U.S. LT wait-list pool. Our transplant center located in UNOS Region 3 routinely transplants patients with lower MELD scores as compared with other centers throughout the country; therefore, our data may not faithfully mirror the national experience. As expected, patients transplanted in the pre-MELD cohort had lower calculated MELD scores compared with the MELD cohort at time of LT. This finding implies the organ recipient population during the MELD cohort was sicker at the time of LT compared with the pre-MELD cohort. However, this difference only strengthens our observation that MELD score used as an allocation criterion does not result in higher post-LT mortality or increased renal morbidity during the first 2 years after LT. Another potential confounding factor that may impact our results is the inherent and unavoidable patient selection bias that may exist when patients undergo LT. Clearly, patients with significant medical co-morbidities (i.e., DM, HTN, cardiovascular disease, and even advance renal disease) may not be listed for transplantation. In addition, many patients may be de-listed if they become too sick to undergo LT. However, despite the MELD cohort being sicker at the time of LT than the pre-MELD cohort, there appears to be no difference in patient deaths or need for delisting for becoming “too sick to transplant” in the MELD vs. the pre-MELD cohort. In the absence of a difference between cohorts, pre-transplant deaths or delisting of patients for medical reasons is likely not a significant confounding factor to this data. Concern that the 2-year follow-up period comparing pre-MELD and post-MELD data is much too short surely exists. However, in a recent study of 1,602 patients who underwent LT by Paramesh et al.,26 elevated serum creatinine (creatinine level > 1.7 mg/dl) at one year was strongly associated with the development of CRD (P < 0.001) in a multivariate regression analysis. In our study, there was no difference in the serum creatinine level between the pre-MELD and MELD cohort at 1 year post-LT. Therefore, although no long-term data currently exists for those recipients in the MELD cohort, we suspect that no long-term difference in CRD for those patients transplant in the pre-MELD vs. the MELD era.
Although the MELD cohort preferentially selected those patients with decreased renal function as compared the pre-MELD cohort, the MELD score was not an independent predictor of CRD post-LT. A potential explanation for this finding is that patients meeting criteria for CRD-3 or CRD-4 at time of LT during the MELD era had a greater chance of renal recovery than those transplanted in the pre-MELD cohort as a consequence of shorter waiting times. Those patients with persistent renal dysfunction post-LT may have preexisting concomitant condition(s) associated to chronic renal disease (i.e., DM, HTN, intrinsic renal disease, and/or renal-vascular disease). Calcineurin inhibitor nephrotoxicity is a common complication in the post-transplant setting leading to renal dysfunction.27 Post-LT chronic renal disease, HTN, and hyperlipidemia have been associated more with cyclosporine than with tacrolimus.28 Notably, our immunosuppressive regimen remained unchanged during the study period and was comparable between cohorts thus reducing the confounding effect of immunosuppression regimen on CRD risk in these cohorts.
Uncontrolled HTN has been linked to the progression of CRD in the non-transplant setting.29 The prevalence of HTN is increased in the post-transplant population.30 Similarly, the diagnosis of DM before LT has been associated with renal disease post-LT, but results have been conflicting.30, 31 Our study revealed that the prevalence of both HTN and DM were similar in the pre-MELD and MELD cohorts. Although the presence of either HTN or DM at time of LT was associated with development of CRD-3/CRD-4 during the first 2 years after LT by univariate analysis, there was no association by multivariate analysis. The only significant predictor of development of CRD-3/CRD-4 during the first 2 years after LT was serum creatinine at time of transplantation.
The effect of advanced age in the progression of CRD has been well demonstrated in the non-transplant setting. Our findings of higher CRD-/CRD-4 incidence with increasing age was noted in both the univariate and multivariate analysis. Furthermore, older LT recipients in our cohort had an increased risk for mortality that did not allow sufficient patient survival to track the development of CRD-3/CRD-4 (data not shown). The overarching significance of baseline renal dysfunction on the future risk of CRD is not surprising and is consistent with the results of previous studies.31 This risk for CRD is observed even with serum creatinines in the normal range. The absence of significance of DM, HTN, or HCV infection in our multivariate model by no means detracts from their overall clinical significance. The relatively small sample size in our study may decrease the power to evaluate the effect of DM, HTN, or HCV on CRD risk owing both to a small number of patients and a relatively short follow-up period and to show significant survival differences over a short follow-up period. In addition, a selection bias where HCV positive transplant recipients with perceived severe HCV-associated renal dysfunction are less likely to be offered LT may exist. Our results require confirmation by a study involving a longer follow-up period and larger sample size. Analysis of the entire UNOS database may be necessary to assess the impact of individual risk factors on the progression of chronic renal disease after LT.
In summary, our single center experience noted essentially non-positive difference between risk of chronic renal disease as well as patient survival between the pre-MELD and MELD cohorts. This study also highlights that the MELD score is not an instrument that is meant to help predict survival post-transplant and specifically is not an instrument that is meant to help predict the need for renal replacement therapy or renal transplantation post-liver transplant. The MELD score is indeed a tool designed for morality predication in patients with liver failure awaiting liver transplant.