Improving Outcomes of Liver Retransplantation: An Analysis of Trends and the Impact of Hepatitis C Infection

Authors


Corresponding author: Rolland Dickson, dickson.rolland1@mayo.edu

Abstract

Retransplantation (RT) in Hepatitis C (HCV) patients remains controversial. Aims: To study trends in RT and evaluate the impact of HCV status in the context of a comprehensive recipient and donor risk assessment. The UNOS database between 1994 and October 2005 was utilized to analyze 46  982 LT and RT. Graft and patient survival along with patient and donor characteristics were compared for 2283 RT performed in HCV and non-HCV patients during 1994–1997, 1998–2001 and 2002–October 2005. Overall HCV prevalence at RT increased from 36% in the initial period to 40.6% after 2002. In our study group, 1-year patient and graft survival post-RT improved over the same time intervals from 65.0% to 70.7% and 54.87% to 65.8%, respectively. HCV was only associated with decreased patient and graft survival with a retransplant (LT-RT) interval (RI) >90 days. Independent predictors of mortality for RT with RI >90 days were patient age, MELD score >25, RI <1 year, warm ischemia time ≥75 min and donor age ≥60 (significant for HCV patients only). Outcomes of RT are improving, but can be optimized by weighing recipient factors, anticipation of operative factors and donor selection.

Introduction

Hepatitis C infection (HCV) is the leading indication for primary liver transplantation (LT) and liver retransplantation (RT) (1–3). RT is associated with decreased patient and graft survival compared with LT regardless of HCV status, with sepsis and postoperative complications accounting for the majority of early morbidity and mortality (3,4). However, RT in HCV is controversial due to concerns for aggressive disease recurrence post-RT (5) and worse patient and graft survival (2,5–7).

Studies analyzing predictors of outcomes of RT utilizing the United Network for Organ Sharing (UNOS) data from 1990 to 2002 focused primarily on recipient factors. These factors included serum creatinine and bilirubin, model for end-stage liver disease (MELD) score, age, LT to RT interval, intensive care at RT and HCV status. Donor and operative factors considered included donor age and cold ischemia time (CIT) (2,3,5–7). There have been a number of significant developments in LT since the prior analysis. The MELD-based organ allocation was initiated in 2002 and there has been improved understanding of donor and operative risk factors, including warm ischemia time (WIT), donor race, donor cause of death, donation after cardiac death and reduced or small for size grafts (8–10). Additionally donor age has been shown to carry disproportionately greater risk in HCV patients undergoing LT (8).

RT outcomes improved in 1997–2002 compared with 1990–1997 (3). However, more recent trends in RT outcomes are unknown and the impact of currently defined recipient and donor risk factors including HCV status is unclear. The aims of this study were to determine trends in RT outcomes and to evaluate the predictors of RT graft survival, including HCV status, in the context of a comprehensive recipient and donor risk assessment.

Methods

This is a registry study based on the results of a data query processed by the Organ Procurement and Transplant Network on October 11, 2006. All LT and RT (as classified by UNOS) performed between 1994 and October 11, 2005, were analyzed according to HCV status. HCV status was determined by results of HCV testing (ELISA, RIBA or polymerase chain reaction (PCR). All RT utilizing whole grafts from deceased adult donors in adult recipients were further analyzed. RT were excluded for (i) multiorgan transplantation, (ii) donation after cardiac death, (iii) hepatitis B surface antigen positive recipients, since prophylaxis during the study period was variable between different periods and centers. The lack of or inadequate dosing of hepatitis B immunoglobulin may have introduced potentially confounding variables in this small group of patients. (iv) Live donor LT.

A total of 57  572 liver transplant procedures were registered with UNOS between 1994 and October 11, 2005. HCV serologies were available in 42  978 of the 51 918 primary LT and 4004 of the 5654 RT during the study period, with positive serologies in 40.1% and 38.3%, respectively. Of the 4004 RT with known HCV status, 2891 RT were performed in adults using deceased adult donors. Patients were excluded from the analysis for missing data on recipient hepatitis B surface antigen status (185), Hepatitis B surface antigen positive recipients (160), donors after cardiac death (31), split grafts (20) and multiorgan transplants (212). The remaining 2283 RT were included for further analysis.

Hepatocellular carcinoma and other liver tumors were reported in only 3.3% of the RT study group. Liver tumor data was most likely inconsistently recorded, as the number reported was lower than expected for this population, and therefore could not be uniformly excluded. Causes of death and graft failure were not recorded so it was difficult to know if tumor recurrence accounted for an increased number of deaths and graft failures. While tumor recurrence would be expected to account for potentially more deaths and graft failures following primary LT, it would be expected to play less of a role in RT as patients after primary LT with recurrence would not be expected to be retransplanted.

Patient characteristics were determined including age, gender, race, HCV status, laboratory values at RT [bilirubin, creatinine, international normalized ratio (INR), albumin], calculated MELD and Childs Turcotte Pugh (CTP) scores at RT, hospitalization and intensive care (ICU) status at RT, UNOS status 1, retransplant interval (RI) (interval from prior transplant to RT) and year of RT. The causes of graft failure were not available for the majority of patients. Donor and operative factors including age, gender, race, cause of death, CIT and WIT were determined. Continuous variables were described as mean ± standard deviation, and categorical variables as percentages. Prothrombin time values were mixed in with INR values in the same data field, therefore only INR values between 0.8 and 8 were utilized for MELD score calculation. Cases with creatinine and bilirubin values of 0 were excluded. MELD scores were calculated as previously described (11). Serum creatinine, bilirubin and INR values <1 were assigned a value of 1, and patients receiving dialysis at RT were assigned a creatinine value of 4.

Similar to prior studies, this study included patients undergoing multiple RT (2,3,6,7). The end point for graft survival in the UNOS data was a subsequent RT or death, whereas patient survival was censored at subsequent RT. Patient and donor characteristics, patient and graft survival were compared between RT performed during three periods, delineated as 4-year intervals from 1994 to 1997, 1998 to 2001 and 2002 to 2005. Patient and donor characteristics of HCV and non-HCV patients were compared. Two RI thresholds were individually important in this study. An RI of <90 days was likely to reflect perioperative complications leading to graft failure, whereas an RI cutoff of 1 year was statistically the most discriminating interval for predicting survival. Therefore, RT patient and graft survival in HCV and non-HCV patients were also compared by RI of <90 days, 90 days to 1 year and >1 year.

Statistical analysis was performed with SPSS 14.0 (SPSS Inc, Chicago, IL). Patient groups were compared using Student's t-test or Pearson's chi-square test. Trends in variables over the three time periods were analyzed by linear or logistic regression. Patient and graft survival were determined by the Kaplan–Meier method and life tables, and significance of survival differences was determined by the log-rank test. The influence of recipient and donor variables on RT patient and graft survival was analyzed by Cox proportional hazard regression in patients with RI >90 days, the group more likely to contain HCV patients with graft failure due to HCV recurrence. Factors achieving p < 0.2 on univariate analysis were entered into a stepwise multivariate Cox regression analysis. No adjustments were made for multiple testing. P-values were 2-tailed and significant when <0.05.

Recipient factors in the regression model included age, gender, HCV status, MELD and CTP scores, RI, ICU status, UNOS status 1 and year of RT. Donor and operative factors included age, black ethnicity, nontraumatic cause of donor death, CIT and WIT. Recipient age was categorized by 20-year increments using age 19 to 39 as a baseline for comparison. Other continuous variables were dichotomized using threshold values determined by exploration of the data and confirmed by sensitivity testing relative to graft survival. RI was analyzed by 3-month increments up to 2 years or more, and by 1-year increments up to 5 years or more. The most discriminating threshold values were MELD 25, CTP score 9 (Childs class C), RI 1 year, donor age 60 years, WIT 75 min and CIT 12 h. MELD was analyzed in lieu of creatinine and bilirubin. Though year of RT ≥2002 was associated with significantly improved survival by univariate analysis, complete data sets for other variables in the model were only available from 2002 and later, therefore year of RT was not included in the final model.

Results

The number of LT progressively increased by period of transplant (1994–1997, 1998–2001 and 2002–2005), but the volume of RT remained unchanged (Figure 1). The proportion of LT and RT patients with HCV (without applying RT study selection criteria) increased during the study period, but there was a relatively greater increase in HCV prevalence at LT than at RT (Table 1).

Figure 1.

Number of LT (unshaded box) and RT (shaded box) during 1994–1997, 1998–2001 and 2002–2005.

Table 1.  Trends in the proportions of LT and RT in HCV patients by period of transplantation
 % of LT in HCV patients (number of HCV/ total number)% of RT in HCV patients (number of HCV/ total number)
1994–199733%36%
(3698/11  176)(441/1219)
1998–200141.7%37.9%
(6198/14  861)(523/1379)
2002–200543.4%40.6%
(7356/16  941)(571/1406)
p<0.001<0.001

Further analysis found that LT 90 day graft survival improved 4% from the first to the last period (86.3–90.3%. p = 0.003) while the percentage of patients undergoing RT within 90 days of LT decreased by 4% overall from 8% to 4% (6.1–2.9% in HCV patients). The percentage of patients undergoing RT after 90 days was unchanged for all patients (5.7–5.3%) and HCV patients (5.8–4.9%).

Patient and donor characteristics for the study group were compared by the transplant period (Table 2). The prevalence of patients with HCV in the RT study group increased from 44% to 47% over the study period, but did not reach statistical significance. Over the three time periods, there was a progressive significant decline of mean serum albumin and patients with UNOS status one. The proportion of patients in the ICU at RT declined only in the 2002–2005 period. There was a progressive increase in the mean RI over the three time periods. In analysis of donor and operative factors, there were significant progressive declines in WIT, CIT and percentage of patients receiving grafts with donors age ≥60 years. Other measured factors were not significant. There was insufficient data to calculate CTP scores prior to 2002, which prevented meaningful trend analysis of this factor. Patient and graft survival were significantly improved for the period after 2002 for all RT (Table 3). Though a similar period effect was seen in HCV patient and graft survival, it only reached significance for graft survival (Table 3).

Table 2.  Patient and donor characteristics for the RT study group by period of RT
Number of RT meeting study criteria1994–1997 n = 6761998–2001 n = 7862002–2005 n = 821p
  1. *(Number of cases shown if < 240) these values were also excluded from comparison.

  2. NA = not applicable.

Recipient age (years)48 ± 1148 ± 1148 ± 11  0.34
Recipient age  0.62
 18–3920%18%18% 
 40–5964%68%68% 
 60–7916%14%14% 
Male gender (recipient)61%60%65%  0.17
Caucasian (recipient)78%77%73%  0.08
Study patients with HCV44%44%47%  0.37
Bilirubin mg/dL15.4 ± 14.813.7 ± 13.815.7 ± 14.7  0.56
Creatinine mg/dL2.0 ± 1.22.1 ± 1.32.0 ± 1.2  0.81
INRNA *(n = 9)1.8 ± 0.91.8 ± 0.9  0.93
Albumin g/dL2.9 ± 0.72.81 ± 0.7 2.75 ± 0.7 <0.001
MELDNA *(n = 9)26 ± 1026 ± 10  0.77
MELD >25NA *(n = 9)50%48.7%  0.60
ICU49%57%37%<0.001
UNOS status 147%42%34%<0.001
RI (days)
 Mean674 ± 311782 ± 596821 ± 770<0.001
 Median3046188 
RI <1 year78.8%68%60.2%<0.001
Donor age (years)41 ± 1542 ± 1640 ± 15  0.24
Donor age ≥6013.8%14.6%10.4%   0.026
Male gender (donor)58%53%57%  0.17
Caucasian/Black (donor)74%/14%73%/12%71%/12%  0.60
Donor cause of death (nontrauma related)57%59%59%0.4 
CIT (h)9.0 ± 4.78.1 ± 4.97.5 ± 3.8<0.001
CIT ≥1216.3%12.9%8.5%<0.001
WIT (min)55 ± 2442 ± 2040 ± 18<0.001
WIT ≥7515.4%5.3%3.3%<0.001
Table 3.  One- and 3-year RT patient and graft survival by period of RT for the overall study group, and for HCV patients. The p-values correspond to log-rank test comparing outcomes of RT performed before and since 2002
Period of RT1994–1997 n = 6761998–2001 n = 7862002–2005 n = 821p
  1. 1Comparing survival before and since 2002.

1-year patient/graft survival65%/54.8%66.3%/60.3%70.7%/65.8%1Patient 0.016
 (number at risk at 1 year)(n = 360)(n = 470)(n = 426)1Graft <0.001
3-year patient/graft survival59.3%/48.8%57.9%/51.3%63%/57.5% 
 (number at risk at 3 years)(n = 303)(n = 384)(n = 123) 
1-year HCV patient/graft survival61%/52.3%64.3%/57.3%67.6%/62.7%1Patient 0.14
 (number at risk at 1 year)(n = 149)(n = 195)(n = 184)1Graft 0.024
3-year HCV patient/graft survival55.9%/46.9%53.1%/46.8%56%/51% 
 (number at risk at 3 years)(n = 126)(n = 151)(n = 49) 

Patient and donor characteristics and 1- and 3-year RT patient and graft survival for HCV (n = 1034) and non-HCV (n = 1249) patients were compared (Table 4). HCV patients had a lower proportion of recipients with age <40, and a greater proportion of males and male donors. They also had a greater proportion of patients with MELD >25 and Childs class C, but were less frequently UNOS status 1 at RT. Although HCV patients had a shorter mean RI, they represented a relatively greater proportion of later RT, particularly with RI 90 days to 1 year.

Table 4.  HCV and non-HCV patient and donor characteristics at RT, with overall 1- and 3-year patient graft survival
 HCV n = 1034Non-HCV n = 1249p
Recipient age (years)50 ± 947 ± 12<0.001
Recipient age <0.001
 18–399%26.5% 
 40–5977.5%58.3% 
 60–7913.5%15.2% 
Male gender (recipient)73%53%<0.001
Caucasian (recipient)72%73% 0.25
% undergoing 2nd or more RT10.2%9.5% 0.62
Bilirubin mg/dL15 ± 1515 ± 14 0.21
Creatinine mg/dL2.1 ± 1.32.0 ± 1.2 0.001
INR1.8 ± 0.91.8 ± 0.9 0.42
Albumin g/dL2.8 ± 0.72.8 ± 0.7 0.71
MELD26 ± 1025 ± 10 0.02
MELD >2553.1%44.9% 0.008
CTP10.7 ± 2.010.3 ± 2.3 0.008
Child's C72.1%62.1% 0.002
ICU at RT47%48% 0.76
UNOS status 137.4%43.0% 0.007
RI (days)
 Mean545 ± 960600 ± 1110 0.01
 Median12641 
RI <90 days47%56%<0.001
 90 days to 1 year20%14% 
 >1 year33%30% 
Donor age (years)41 ± 1541 ± 16 0.65
Donor age ≥6012.2%13.4% 0.40
Male gender (donor)58%54% 0.03
Caucasian (donor)75%77% 0.76
CIT (h)8.2 ± 4.18.2 ± 4.8 0.22
CIT ≥1212%12.8% 0.59
WIT (min)45 ± 2046 ± 22 0.28
WIT ≥756.2%8.8% 0.037
1-year patient/graft64.5%/57.9%70.4%/62.9%<0.001
 survival (number at risk) (n = 528) (n = 723)<0.001
3-year patient/graft55.3%/48.8%64.4%/55.8% 
 survival (number at risk) (n = 326) (n = 484) 

Patient and graft survival at 1- and 3-year post-RT were compared between HCV and non-HCV patients for RT <90 days, between 90 days and 1 year and greater than 1 year since prior LT. HCV (n = 484) and non-HCV (n = 700) patients undergoing RT within 90 days of prior LT had similar patient and graft survival at 1 year (patient: 64% vs. 66%, graft: 57% vs. 57%) and 3 years (patient: 57% vs. 59%, graft: 50% vs. 50%), respectively (Figure 2). Patient and graft survival were significantly lower for HCV compared with non-HCV patients with RI 90 days to 1 year and >1 year (Figure 3) (Table 5).

Figure 2.

RT patient (2-1) and graft (2-2) survival in HCV (dashed line, n = 484) versus non-HCV (solid line, n = 700) patients undergoing RT within 90 days of prior transplant.

Figure 3.

Patient (3-1) and graft (3-2) survival in HCV and non-HCV patients undergoing RT more than 90 days from prior transplant by retransplant interval (RI) of less or more than 1 year. The p-value shown for RI <1 year in HCV (nonbold dashed line, n = 204) versus non-HCV patients (nonbold solid line, n = 169) and RI >1 year in HCV (bold dashed line, n = 344) versus non-HCV (bold solid line, n = 374).

Table 5.  Patient, donor and operative characterisitics and 1- and 3-year patient and graft survival in HCV and non-HCV patients undergoing RT with RI 90 days to 1 year and more than 1 year
 RI 90 days to 1 yearRI >1 year
HCV n = 204Non-HCV n = 169pHCV n = 344Non-HCV n = 376p
Recipient age50 ± 848 ± 12  0.1348 ± 9 43 ± 13<0.001
MELD25 ± 922 ± 8   0.07725 ± 925 ± 9  0.65
% in ICU at RT32%23%  0.04425%17%  0.011
Donor age 41 ± 1539 ± 15  0.31 38 ±1440 ±16  0.17
WIT 47 ± 2347 ± 23  0.85 43 ± 20 46 ± 22  0.19
1-year patient/graft survival (number at risk)56%/50%69%/64%<0.001 for patient and graft survival70%/63%79%/73%<0.001 for patient and graft survival
(n = 94)(103) (n = 193)(n = 250) 
3-year patient/graft survival (number at risk)44%/37%65%/58% 59%/53%73%/66% 
(n = 51)(72) (n = 106)(n = 146) 

Overall, HCV patients required ICU care more frequently at RT. In the RI range of 90 days to 1 year, HCV patients more often had a MELD >25 (50% vs. 32%, p = 0.013), and at an RI >1 year were less frequently younger than 40 years (10% vs. 37%, p < 0.001). HCV patients had higher patient and graft survival rates when undergoing RT >1 year compared with 90 days to 1 year (p < 0.002), but this did not reach statistical significance in non-HCV patients (p = 0.28).

The results of univariate and stepwise multivariate Cox proportional hazards analysis of patient and graft survival post-RT with RI >90 days by recipient, donor and operative factors are outlined in Table 6. In similar posthoc multivariate analysis of these variables in HCV and non-HCV patients separately, donor age ≥60 was associated with a greater risk of death (hazard ratio 3.4, p = 0.001) and graft failure (hazard ratio 2.7, p = 0.004) in HCV patients but was not significant in non-HCV patients.

Table 6.  Results of univariate and stepwise multivariate (for factors with p < 0.2 on univariate analysis) Cox regression analysis for patient and graft survival by recipient and donor criteria
CriterionPatient survivalGraft survival
Hazard ratio95% confidence intervalpHazard ratio95% confidence intervalp
  1. *Missing data for some cases (number of cases with available data).

  2. NS = nonsignificant.

Univariate analysis (n = 1093)
 Recipient age by 20 year increments (compared with adult recipients under 40)1.331.13–1.57 0.0011.181.02–1.36 0.028
 Male recipient gender1.040.85–1.26 0.730.990.84–1.18 0.96
 HCV positive1.721.42–2.08<0.0011.581.33–1.87<0.001
 MELD >25 *(n = 559)1.491.11–2.0 0.0081.43 1.1–1.85 0.008
 Child's class C *(n = 450)1.290.90–1.850.171.36 0.98–1.88 0.066
 ICU status at RT1.611.31–1.98<0.0011.501.25–1.8  <0.001
 UNOS status 11.751.24–2.46 0.0011.6 1.17–2.2   0.003
 RI ≤1 year1.351.12–1.64 0.0021.29 1.09–1.53 0.003
 Donor age ≥60 2.01.54–2.6<0.0011.811.42–2.3 <0.001
 Donor black ethnicity1.391.06–1.82 0.0171.43 1.13–1.81  0.003
Donor cause of death (nontrauma)1.411.16–1.72<0.0011.42   1.2–1.69<0.001
 CIT ≥12 h *(n = 969)1.371.06–1.77 0.0171.321.05–1.66 0.02
 WIT ≥75 min *(n = 905)1.541.12–2.13 0.0081.561.18–2.07 0.002
Multivariate analysis (n = 352)
 Recipient age by 20 year increments (compared with adult recipients under 40)1.521.04–2.21 0.029NSNSNS
 MELD >251.621.03–2.55 0.0381.781.23–2.57 0.002
 RI ≤1 year1.661.08–2.56 0.022NSNSNS
 ICU status1.660.98–2.83 0.060NSNSNS
 Donor age ≥602.181.25–3.8 0.006NSNSNS
 Donor black ethnicityNSNSNS1.681.01–2.79 0.046
 WIT ≥75 min2.861.29–6.35 0.0102.861.38–5.92 0.005

The independent predictors of mortality on multivariate analysis of RT with RI <90 days were Childs class C (hazard ratio 2.51, p = 0.001), donor age ≥60 (hazard ratio 1.8, p = 0.023), black donor ethnicity (hazard ratio 1.88, p = 0.033) and WIT ≥75 (hazard ratio 3.18, p = 0.014). The independent predictors of graft failure were Childs class C (hazard ratio 2.34, p < 0.001) and WIT ≥75 (hazard ratio 2.56, p = 0.026).

Discussion

This review of the 1994 to 2005 UNOS liver transplant database uniquely encompasses the periods before and after adoption of MELD. MELD-based allocation may be expected to result in worse LT and RT outcomes as patients with higher disease severity are prioritized for transplantation (3,12). However, studies of UNOS data have not demonstrated increased rates of primary nonfunction or 1-year graft failure post-LT under MELD (13,14). Similarly, this study found an improvement in post-RT patient and graft survival for the period since 2002. The improvement in survival could be due to a more careful selection of patients and exclusion of sicker patients in the later period. This is supported by a progressive decline in UNOS Status 1 patients with each period and a decreased rate of RT in ICU patients in the period after 2002. However, MELD scores were not significantly lower in the later period. The decline in patients requiring ICU care after 2002 may have been an indirect effect of MELD, which unlike the previous allocation system did not allow an increase in status based on the patient being in the ICU. UNOS status 1 was also likely to have decreased due to the UNOS requirement of status 1 criteria certification since 2002. In addition, the improvement in 90-day graft survival post first LT appears to correlate with the isolated decrease in the proportion of patients undergoing RT within 90 days of LT. These factors may explain the decrease in ICU and UNOS status 1 at RT in the most recent period and may argue against a patient selection bias.

More likely the improvement in RT graft survival was related to continued improvement in organ allocation, surgical management and early postoperative care. Mean CIT and WIT decreased significantly over the study period. These decreases may reflect improving surgical techniques such as preservation of the vena cava and possibly more widespread surgical experience leading to improved immediate postoperative outcomes. In both this study and others, prolonged WIT was demonstrated to be a strong independent predictor of graft and patient survival (15–18). Arguably RT WIT would be minimized at centers with higher transplant volumes and experience with the technical demands of this procedure. This raises the issue of whether RT should be referred to one of these centers.

Despite the improvement in survival noted in the post-MELD era, the risk factors for poor outcome were similar. As in previous studies, patient age, MELD >25, advanced donor age, WIT and black donor ethnicity were found to be independent risk factors (3,8,9) (6–8,19). The risk associated with donor ethnicity and cause of death is recognized but not well understood (9). These factors may reflect donor comorbidities, but selecting donors accordingly may be prohibitively challenging in practice. The effects of other donor factors such as living donor-related LT, split grafts and donation after cardiac death cannot be assessed by this study as they were excluded by design. ICU status, which had been shown to be predictive in previous analysis approached but did not reach statistical significance as a predictor of mortality on our multivariate analysis. Perhaps this is due to exclusion of patients undergoing RT within 90 days in the multivariate analysis.

One of the major considerations of this study was the impact of HCV on RT outcomes. Despite the improvements in overall outcomes, HCV patients continued to have worse patient and graft survival in each of the three time periods, The worse outcomes are in conflict with a recent multicenter study (19), as well as recent data from our own center (20), which did not find a difference in graft or patient survival between HCV and non-HCV patients. However, of significance, the current study and others did not demonstrate HCV as an independent risk factor for graft or patient survival (8). This suggests that there may be factors that have a greater impact on HCV patients than non-HCV patients, or that HCV patients may have greater severity of illness at the time of RT leading to poorer outcomes. This warrants further analysis.

RT within the first 90 days led to equally poor outcomes in both HCV and non-HCV patients. These patients were more likely to be debilitated and critically ill after recent LT, while contending with graft (primary non function or delayed function), technical (arterial or biliary) or infectious complications prior to RT. These patients may have more advanced degrees of graft failure at RT, or insufficient time for patient recovery and reconditioning after the prior transplant (21).

In this series survival differences between HCV and non-HCV patients were demonstrated only in RT after 90 days, with the worst outcomes for RT in HCV patients between 90 days and 1 year after LT. A relatively greater proportion of HCV than non-HCV patients underwent RT within 90 days to 1 year of prior LT. This subset of HCV patients had greater disease severity than their counterparts without HCV, with increased MELD, ICU status and higher mortality and graft failure. Though the causes of death or graft failure could not be determined from the UNOS database, this finding may coincide with the timing of cholestatic or early and rapidly progressive HCV post-LT. It may also reflect the severity of illness in this patient subset with rapid HCV-related graft failure. A recent U.S. multicenter study demonstrated that early (within 6 months of LT) or cholestatic HCV recurrence were reasons for denial of RT in approximately 20% of all HCV patients evaluated for RT, while 1- and 3-year survival in HCV patients undergoing RT for HCV recurrence versus non-HCV patients were similar (22). The lack of defined RT indications in our series prevents the analysis of selection bias in patients with HCV recurrence.

Advanced donor age is a well-recognized risk factor in both LT and RT for all indications (7–10), but there is data to support an increased relative risk of graft failure in HCV patients following LT (8,23–25). In this study, donor age was a risk factor for mortality and graft failure only in HCV recipients, as it was not found to be significant factor in the non-HCV patients. The reason for the disproportionate risk with older donors in HCV patients is not known, but HCV recurrence is universal and is known to progress more rapidly in older grafts (23,26,27). Similar to non-HCV patients, HCV patients bare the risks of early graft failure related to advanced donor age, but in addition have longer term risks related to accelerated progression of recurrent HCV. A recent study demonstrated that when HCV patients were retransplanted in a large center with short waiting times, short WIT and cold ischemic times and virtual exclusion of donors >60 for HCV recipients (6%), 1- and 3-year outcomes were comparable to non-HCV RT (20).

The variation in risk of RT attributed to HCV in this and other studies of UNOS data may be explained by the limited sample size and variable inclusion of other factors in the analysis (3,6–8,28). The inclusion of WIT, components of the donor risk index (donor ethnicity and cause of death) (9) and novel inclusion of Childs class in this UNOS RT cohort, identified multiple non-HCV factors as significant predictors, underscoring the complexity of predicting RT outcomes. Though donor selection for RT may be optimized, imposing restrictions on an already limited donor pool has to be balanced against the risks of wait list mortality, particularly in patients with high disease severity. This remains a highly individualized process.

There is insufficient UNOS data to delineate the influence of viral (genotype), patient (functional status, rejections episodes, HCV recurrence and treatment) or center-specific factors (immunosuppression regimens, treatment protocols, practice patterns) on RT outcomes. Single center studies may hint at trends of decreasing rejection and ischemic biliary injury rates (29), but these findings may be prone to inter-center variation, and cannot be generalized. Ongoing study of RT outcomes is required to confirm these findings, and to define evolving trends. It is important to note that the results of multivariate analysis could be affected by the thresholds used for dichotomization of continuous variables. Though the most discriminating thresholds were used, continuous variables were in many cases associated with significant risk across a range of values, which should be considered in further analysis. Though early post-RT outcomes are improving, long-term follow-up and better posttransplant management options for HCV patients with disease recurrence are still needed.

Conclusions

Graft survival rates following RT are improving. Despite the controversy, HCV remains the most prevalent diagnosis in patients undergoing RT. HCV is superseded by multiple donor, operative and recipient factors as predicators of RT graft survival, and should not contraindicate RT. Donor and operative factors contribute significantly to RT outcomes, with donor age representing disproportionate risk in HCV patients. In addition to age and MELD, patient risk in RT may be estimated more accurately by considering the timing of graft failure.

Acknowledgments

The authors thank UNOS for providing the data. This work was supported in part by 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 mention of trade names, commercial products or organizations imply endorsement by the U.S. government.

Ancillary