For the past two decades, the number of patients

When Shouldn t We Retransplant? Michael A. Zimmerman and R. Mark Ghobrial Key Points 1. In the setting of early graft failure after primary transplantation, orthotopic liver retransplantation (re-olt) should be undertaken within the first 7 days, but it should be discouraged within 8-30 days, since re-olt within this intermediate frame is associated with the worst results. 2. Late retransplantation should be cautioned in severely ill patients who exhibit Model for End-Stage Liver Disease (MELD) scores >25, require mechanical ventilation, have advanced renal insufficiency, and in advanced-age recipients. 3. Re-OLT should not be undertaken with extended and older donors particularly when retransplantation for recurrent disease is considered. 4. Prognostic models that take into account the severity of disease and the effect of the organ to be transplanted should be developed to better predict outcomes after re- OLT. 5. Accurate definitions of acceptable outcomes after retransplantation and futile re-olt are desperately needed. (Liver Transpl 2005;11:S14-S20.) Abbreviations: re-olt, orthotopic liver retransplantation; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; UNOS, United Network for Organ Sharing; HCV, hepatitis C virus. From the Department of Surgery, Division of Liver and Pancreas Transplantation, The Pfleger Liver Institute, The Dumont-UCLA Transplant Center, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, CA. Address reprint requests to R. Mark Ghobrial, MD, PhD, The Dumont-UCLA Transplant Center, 77-120 CHS, Box 957054, 10833 Le Conte Ave., Los Angeles, Ca 90095-7054. Telephone: 310-825- 2678; FAX: 310-825-0392; E-mail: rghobria@mednet.ucla.edu Copyright 2005 by the American Association for the Study of Liver Diseases Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/lt.20599 For the past two decades, the number of patients listed for orthotopic liver transplant (OLT) continues to rise, far exceeding cadaveric organ availability. Despite expanding the definition of acceptable cadaveric grafts and employing partial grafts from living donors, the number of deaths on the waiting list is also increasing. 1 This disparity between number of patients listed for OLT and the almost fixed number of transplantable organs fuels the debate surrounding organ allocation for primary OLT in general and orthotopic liver retransplantation (re-olt) in particular. The association of a second transplant with increased mortality is well documented. 2-6 Currently, this debate has focused more on retransplantation secondary to recurrent disease vs. early graft loss. Increased mortality following retransplantation, limited availability of cadaveric organs, and death of primary OLT candidates on the waiting list fuel the concern over retransplantation. Nevertheless, retransplantation remains the only viable option for patients with graft failure secondary to disease recurrence and may be accompanied by reasonable short-term survival when performed early. 5 Predictors of Survival Following Retransplantation To date, several large series have identified variables that may predict overall patient survival following a second liver transplant, some of which are demonstrated in Table 1. 3,4,7-10 One of the earliest clinical experiences was chronicled at the University of Pittsburgh analyzing the outcome of 418 patients who underwent re-olt over a 6-year period. 7 Several independent variables were associated with graft failure, including female gender, bilirubin, creatinine, mechanical ventilation, and recipient age. One-, 5-, and 10-year survival rates following re-olt at the University of California, Los Angeles were 62%, 47%, and 45%, respectively. In this cohort, survival decreased with the increasing number of transplanted grafts and a higher United Network for Organ Sharing (UNOS) status. 3 Compared against a series of patients matched for age and UNOS status who underwent a single OLT, variables that negatively impact survival include creatinine 1.6 and preoperative ventilator support. Rosen and colleagues reported one of the largest clinical series to date from the UNOS Scientific Registry for Liver Transplantation, including adults who underwent retransplantation from January 1990 to February 1996. 4 Overall, 1,356 patients comprised the study group, with 157 (9.8%) undergoing a third and 16 (1%) a fourth transplant, respectively. Sixty percent of this cohort were listed as UNOS status 1 with a mean follow-up of nearly 840 days. Using a Cox proportional hazard model, 5 variables were identified to have significant predictive influence on survival following re-olt: recipient age, bilirubin, creatinine, UNOS status, and cause of graft failure. Unfortunately, other factors that may be important, such as immunosuppression and donor characteristics, were not included. The authors make several interesting points that highlight the importance of the recipient severity of illness and its S14 Liver Transplantation, Vol 11, No 11, Suppl 1 (November), 2005: pp S14-S20

Survival and Retransplantation S15 Table 1. Predictive Models of Post-OLT Survival Reference Patient No. Predictive Variables Comment Markmann et al. 3 299 Age, interval to transplantation, number of grafts, Single center UNOS status Rosen et al. 4 1356 Age, bilirubin, creatinine, UNOS status, cause of UNOS database graft failure Doyle et al. 7 418 Donor gender, recipient age, mechanical Single center ventilation, creatinine, bilirubin Kim et al. 8 447 Time interval to re-olt Single center Ghobrial et al. 9 510 Age, creatinine, donor gender, donor age, bilirubin, Single center warm/cold ischemia time, prothombin time, previous transplant Azoulay et al. 10 139 Age, serum creatinine, urgency of transplantation Single center Abbreviation: PNF, primary nonfunction; OLT, orthotopic liver transplantation. contribution to post-olt mortality following retransplant. Employing this 5-variable model, retransplantation in low-risk patients was similar to patients undergoing primary OLT. However, since most investigators attempt to construct re-olt models based on preoperative data elements, the effects of the utilized liver allograft on retransplantation outcomes have not been fully studied. Ironically, even the most experienced surgeons may not attempt redo transplants unless a high-quality graft is available. Timing and Retransplantation A persistant question is whether the risk of death after early re-olt due to initial graft failure is different from that of delayed transplantation undertaken from disease recurrence. Further, deciding when to retransplant a patient with early graft dysfunction can be extremely difficult. Few studies addressed such questions. The temporal association between graft failure and the interval to retransplant was illustrated in over 70 patients. 11 Mortality dramatically escalated with an increasing time interval between the first and second transplant (Fig. 1A). Overall, survival was 57% if re-olt was within 3 days of the primary transplant. Unfortunately, this dropped to 24% if re-olt was between 4 and 30 days. Interestingly, if retransplantation was performed after a time interval 1 year from the first OLT, survival increased to 83%. A large series from the Mayo Clinic, limited to patients with primary biliary cirrhosis and primary sclerosing cholangitis, suggested that retransplantation after an interval of 30 days from the primary OLT is associated with a 6.7-fold increase in Figure 1. Patient survival, graft failure, and time to retransplant. (A) One-year survival estimates of re-olt recipients based on time interval from first transplant (adapted from Powelson et al. 11 ). (B) Probability of graft failure following re-olt as a function of time elapsed from primary transplant (adapted from Doyle et al. 7 ).

S16 Zimmerman and Ghobrial Table 2. Effect of Time Interval to Retransplantation Interval After OLT (Days) Death Rate Mortality Risk Ratio 60-Month Survival (%) P Value 0-7 0.802 1.00 52.5 0.01 8-30 1.337 1.668 39.7 30 0.688 0.858 57 NOTE: Adapted from Busuttil et al. 12 the risk of death. 8 Results of re-olt prior to the 30-day time point was similar to that of primary OLT. A study from the University of Pittsburgh similarly revealed that the probability of graft failure after retransplantation steadily increased form 0.58 at day 0 to a maximum of 0.8 at day 38, followed by a slow decline thereafter (Fig. 1B). 7 Thus, overall mortality and graft failure are dependent upon the time interval to retransplant. Recently, the cumulative experience at the University of California, Los Angeles with 3,200 liver transplants was reviewed by Busuttil and colleagues. 12 Of 13 donor and recipient variables analyzed, retransplant was among the strongest indicators that negatively impacted survival. Four hundred and fifty patients were retransplanted, 73 underwent 3 retransplants, and 13 patients received a fourth. Survival declined with the increasing number of grafts transplanted with 1-year survival of 31% for patients receiving 4 organs. Importantly, the risk of mortality following retransplantation varied based on the interval from the primary transplant. The highest death rate and mortality risk ratio for retransplant was between 8 and 30 days from the first graft (Table 2). Taken together, the above-mentioned data argue that retransplantation within the first week may exhibit results that are similar to chronic transplantation. MELD scores in re-olt patients and mortality. A report of over 2,000 retransplant patients from the UNOS database evaluated the impact of MELD in the patient population undergoing re-olt. 15 Approximately 40% of patients in this series that had a MELD score between 11 and 20. A reduction in short-term survival to less than 60% was observed in all re-olt patients with a MELD score over 25 (Table 3). While mortality was increased in all groups with a concomitant rise in MELD score, patients with a score greater than 30 had an alarming 20%-40% survival rate. These trends were not disease specific. More recently, Watt and colleagues from the University of Nebraska reported a single-center experience documenting the ability of the MELD score to predict death while awaiting retransplant. 16 Of 63 patients in the study group, 43 (68%) underwent re-olt. MELD scores were significantly higher at the time of the second listing and at the time of re-olt compared to the first listing and primary transplant. Importantly, the MELD score was predictive of death on the waiting list. These data suggest that at a given MELD score, retransplant may exhibit a significantly higher mortality rate than the initial transplant. To improve outcomes, retransplantation must be attempted at a lower MELD score than that used for the primary operation. This principle was illustrated in study by Burton et al. 20 Using weighted utility curves, the maximal value of retransplantation was achieved at a MELD score of 21 for hepatitis C virus (HCV)-positive patients and 24 for non-hcv patients (Fig. 2). Collectively, these data indicate that retransplantation at MELD scores 25 are associated with high mortality rates that far exceed the mortality of primary OLT undertaken at a similar MELD score. It may therefore be prudent to restrict re-olt for patients with high MELD scores to avoid futile transplantations. Retransplantation in the MELD Era Contrary to the UNOS status classification, the MELD scoring system for organ allocation provides an objective stratification of retransplant candidates based on the severity of illness. 13 Initially employed to predict the short-term survival of patients with cirrhosis following a transjugular intrahepatic portosystemic shunt procedure, 14 the MELD model is now used to prioritize patients on the transplant waiting list. Several groups have drawn a direct correlation between increasing Table 3. Survival After Re-OLT Based on MELD 1 Year 5 Year MELD 10 83% 55% MELD 11-20 65% 55% MELD 21-25 62% 47% MELD 26-30 57% 37% MELD 30 42% 21% NOTE: Adapted from Watt et al. 15

Survival and Retransplantation S17 Figure 2. Maximum utility in retransplantation by MELD score. Using weighted utility curves, the maximal value of retransplantation was achieved at a MELD score of 24 for non-hcv patients (adapted from Burton et al. 20 ). Retransplant and Hepatitis C HCV reinfection is universal following OLT, and HCV-associated graft cirrhosis is reportedly as high as 30% in 5 years. 17,18 Retransplantation for HCVpositive patients accounts for approximately 40% of all re-olts in the United States. 19 This number is predicted to increase as more patients are transplanted for HCV and subsequently develop allograft failure secondary to viral recurrence. 20 Initial reports suggested that outcomes after re-olt in HCV patients are poor and that viral recurrence has a negative influence on survival. 5,6 Analyzing the UNOS registry, Rosen et al. chronicled 357 patients transplanted for HCV, with 207 patients requiring re- OLT. 6 They note that survival is significantly lower in HCV-positive patients. Of concern was the marked increase in the prevalence of HCV in the study time interval. In 1990 the prevalence of HCV was 6.5% compared to 38.4% in 1995. Interestingly, only 7 patients of the 207 requiring re-olt died of HCV recurrence in the second graft. A second study suggesting inferior outcomes in HCV-positive patients analyzed the predictors of mortality after late retransplantation ( 6 months after initial OLT). They demonstrated that patients retransplanted for recurrent HCV had poorer survival (57% and 43%) compared to patients retransplanted for other indications (81% and 74%) at 90 days and 1 year, respectively. 21 Although this difference approached, but did not reach, statistical significance, it is clinically relevant as most of the deaths after re-olt occurred within the first 90 postoperatively. Independent predictors identified for 90-day mortality included preoperative creatinine 2 mg/dl, recipient age 50 years, and use of intraoperative blood products. Thus, the initial high mortality observed with re-olt for recurrent HCV may reflect the poor preoperative condition of the patient and failure to identify the predictors of a poor prognosis. 22 Despite previous observations that HCV may be an independent risk factor for increased mortality, several studies demonstrate that reasonable survival can be achieved following retransplantation. Probing the UNOS registry, the group from the University of Nebraska stratified the non-hcv recipients into subgroups based on the etiology of liver disease. Only hepatitis B recipients and patients with autoimmune disease exhibited better survival rates than HCV patients after re-olt. There were no significant survival differences following retransplantation between HCV-positive, cryptogenic, cholestatic, or alcoholic liver disease patients when adjusted for age and MELD scores. These results are in support of other single-center experiences where the recipient primary diagnosis was not found to influence survival after re-olt. 15 A series from the University of California, Los Angeles, evaluated 298 patients undergoing cadaveric OLT for HCV. 2 Seventy-six patients were retransplanted. Twoyear survival in this group approached 60%. These findings suggest that the severity of illness in the recipient is a contributing factor related to post-olt mortality. Additionally, it is possible that a lack of early recognition of recurrent HCV and delay in retransplantation may account for poor initial results of re-olt in these patients. Survival Models for Retransplant Clinical deterioration of patients awaiting retransplantation continues to be a troubling obstacle in patients with primary nonfunction or initial poor function after transplant. The severity of illness and functional reserve of the recipient are increasingly being implicated as the cornerstone to appropriate patient selection for retransplantation. 23 As such, numerous predictive models have been proposed to guide patient selection for re- OLT within that crucial window of time after the primary transplant. Importantly, these models will facilitate identification of patients with poor predicted outcomes after a second transplant. As a result, futile

S18 Zimmerman and Ghobrial operations can be avoided. However, before such models can be applied clinically, several issues remain to be settled. Primarily, a consensus must be developed to determine the acceptable survival range after retransplantation. Secondly, exactly which criteria should be employed for patient selection remains to be established. Rosen et al. evaluated survival outcomes of re-olt by assigning a mortality risk score based on preoperative variables. 4 These factors included recipient age, creatinine, bilirubin, cause of graft failure (versus non-), and a UNOS status. Calculated risk scores for individual patients stratified re-olt recipients into low-, medium- and high-risk groups. While the model predicted survival was not statistically different from observed survival after re-olt, it has not undergone stringent validation. Another retransplant model similarly employed a 5-point scoring system that incorporated recipient age, creatinine, bilirubin, cold ischemia time, and ventilatory status. 24 Although originally developed in a small cohort, it has been validated by UNOS datasets and the Baylor University Medical Center patient population. Evaluating multiple recipient and donor variables, Ghobrial and colleagues have defined a universal model for overall survival. 9 This model attempts to calculate survival for both primary transplant and retransplant. Variables identified by multivariate analysis were incorporated into the survival model and mortality was calculated. These factors include recipient age, creatinine, warm and cold ischemia times, bilirubin, and prothrombin time. The authors validated the model by demonstrating identical predicted survival rates to actuarial rates in both HCVpositive and HCV-negative recipients from the UNOS database. Two modifications were adopted in this model. The MELD score was substituted for preoperative serum creatinine, serum bilirubin, and prothrombin time. Second, to increase its sensitivity, the risk of death imposed by re-olt was reassessed, since retransplantation was shown to exhibit a different risk based on the time interval from the first transplant. Accordingly, re-olt within the first 7 days after the first transplant markedly increased (HR, 1.53), while re-olt within 7-30 or 30 days doubled (HR, 2.153 and 2.131) the risk of death. Therefore, by holding warm and cold ischemic times at median levels, post-re-olt survival can be accurately computed for an individual patient, at 3 and 12 months posttransplantation, based on the MELD score, recipient ages, and donor ages. Accordingly, as shown in Table 4, a 50-year-old Table 4. Predicted Patient Survival Based on MELD Scores and Recipient Age Following Re-OLT by a 50-Year-Old Deceased Donor Graft* Recipient Age MELD 30 40 50 60 70 20 77 73 68 64 58 24 75 71 66 61 55 28 73 69 64 58 53 32 71 67 62 56 50 36 69 65 59 54 48 40 67 63 57 52 45 *Calculations are based on median cold and warm ischemia times of 12 hours and 51 minutes, respectively. re-olt candidate with a MELD score of 28 exhibits a 1-year survival of 64%, if a 50-year-old donor organ is utilized. Further, the same 50-year-old patient retransplanted at a MELD score of 36 would exhibit a 59% survival estimate at 1 year. As demonstrated in Table 4, to achieve a survival benefit of 65%, the 50-year-old donor is best utilized in 30- to 40-year-old re-olt recipients at any MELD score, or 50-year-old candidates with a MELD score of 24 or less (Table 4). Re-OLT should be avoided in older recipients or 50-year-old patients with MELD 28. Thus, this model has the ability to tailor organ needs to recipients based on severity of disease and expected outcomes. When Shouldn t We Retransplant? With the universal application of the MELD score and an increasing availability of various survival outcome models, the focus of debate seems to be shifting. The question is no longer should we or shouldn t we retransplant a given patient. The questions now are who should be retransplanted and when? The aim is to avoid futile transplantation to maximize the benefits of a scarce resource. Nevertheless, to apply such models, a consensus among transplant physicians must be reached that defines futile transplantation in general and accepted specific retransplantation survival outcomes below which retransplantation is to be avoided. Further, only the MELD score is currently in use as an outcome predictor. This has obvious disadvantages since the MELD does not consider liver allograft characteristics. Despite absence of standardized criteria as to which factors are predictive of post re-olt mortality, several

Survival and Retransplantation S19 generalizations can be made. In the setting of early graft failure such as in primary nonfunction, re-olt should be undertaken early within the first 7 days of the primary operation. As shown by multiple studies reoperation at an intermediate time interval (i.e., 8 to 30 days) from the primary OLT is associated with a significantly worse prognosis and should be discouraged. Late re-olt may exhibit survival rates similar to primary transplant in select patients. Retransplantation is more likely to be successful in healthier recipients with a lower MELD score. Based on current outcome predictors, late re-olt should be cautioned in severely ill recipients, who may be defined as those with MELD scores 25, those with advanced renal insufficiency, those on mechanical ventilation, and older candidates. We do not believe that re-olt decisions should be based on etiology of liver disease. The effect of allograft quality is exceedingly recognized as one of the important parameters that determine success of transplantation in general and re-olt in particular. Since only a few studies have attempted to define parameters of donor organ variables, it is difficult to specifically define which organs should not be utilized. Nevertheless, caution should be exercised with extended-criteria donors, such as older donors ( 50 years), those with extended hospital stays ( 5 days), and those with long cold ischemia time ( 8 hours). Every effort should be made to shorten the warm ischemia time. Summary Retransplantation remains a formidable procedure that requires sophisticated decision making and extensive surgical expertise. It is exceedingly recognized that a combination of donor, recipient, and surgical factors are all important contributors to the overall outcome. Thus, relying exclusively on the MELD to predict survival is insufficient. Development of accurate models that take into consideration parameters other than the recipient will facilitate tailoring of organ distribution to benefit a larger group of patients. Additionally, it is ironic that retransplantation is commonly denied based on predicted poor outcomes, when accurate outcome predictors are lacking as well as definitions of futile re-olt and accepted survival parameters. It is rather imperative that a multicenter effort address these pressing issues in the near future. References 1. Lucey MR, Brown KA, Everson GT, Fung JJ, Gish R, Keefe EB, et al. Minimal criteria for placement of adults on the liver transplant waiting list: A report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Transplantation 1998;66:956-962. 2. Ghobrial RM, Farmer DG, Baquerizo A, Colquhoun S, Rosen HR, Yersiz H, et al. Orthotopic liver transplantation for hepatitis C: Outcome, effect of immunosuppression, and causes of retransplantation during an 8-year single-center experience. Ann Surg 1999;229:824-831. 3. Markmann JF, Markowitz JS, Yersiz H, Morrisey M, Farmer DG, Farmer DA, et al. Long-term survival after retransplantation of the liver. Ann Surg 1997;226:408-18; discussion 418-420. 4. Rosen HR, Madden JP, Martin P. A model to predict survival following liver retransplantation. Hepatology 1999;29:365-370. 5. Ghobrial RM. Retransplantation for recurrent hepatitis C. Liver Transpl 2002;8(Suppl):S38 S43. 6. Rosen HR, Martin P. Hepatitis C infection in patients undergoing liver retransplantation. Transplantation 1998;66:1612-1616. 7. Doyle HR, Morelli F, McMichael J, Doria C, Aldrighetti L, Starzl TE, et al. Hepatic Retransplantation an analysis of risk factors associated with outcome. Transplantation 1996;61: 1499-1505. 8. Kim WR, Wiesner RH, Poterucha JJ, Therneau TM, Malinchoc M, Benson JT, et al. Hepatic retransplantation in cholestatic liver disease: Impact of the interval to retransplantation on survival and resource utilization. Hepatology 1999;30:395-400. 9. Ghobrial RM, Gornbein J, Steadman R, Danino N, Markmann JF, Holt C, et al. Pretransplant model to predict posttransplant survival in liver transplant patients. Ann Surg 2002;236:3153-22. 10. Azoulay D, Linhares MM, Huguet E, Delvart V, Castaing D, Adam R, et al. Decision for retransplantation of the liver: An experience- and cost-based analysis. Ann Surg 2002;236:713-721. 11. Powelson JA, Cosimi AB, Lewis WD, Rohrer RJ, Freeman RB, Vacanti JP, et al. Hepatic retransplantation in New England a regional experience and survival model. Transplantation 1993; 55:802-806. 12. Busuttil RW, Farmer DG, Yersiz H, Hiatt JR, McDiarmid SV, Goldstein LI, et al. Analysis of long-term outcomes of 3200 liver transplantations over two decades: A single-center experience. Ann Surg 2005;241:905-916. 13. Charlton M, Kasparova P, Weston S, Lindor K, Maor-Kendler Y, Wiesner RH, et al. Frequency of nonalcoholic steatohepatitis as a cause of advanced liver disease. Liver Transpl 2001;7:608-614. 14. Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, ter Borg PC. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology 2000;31:864-871. 15. Watt KD, Lyden ER, McCashland TM. Poor survival after liver retransplantation: Is hepatitis C to blame? Liver Transpl 2003; 9:1019-1024. 16. Watt KD, Menke T, Lyden E, McCashland TM. Mortality while awaiting liver retransplantation: predictability of MELD scores. Transplant Proc 2005;37:2172-2173.

S20 Zimmerman and Ghobrial 17. Berenguer M, Ferrell L, Watson J, Prieto M, Kim M, Rayon M, et al. HCV-related fibrosis progression following liver transplantation: increase in recent years. J Hepatol 2000;32:673-684. 18. Testa G, Crippin JS, Netto GJ, Goldstein RM, Jennings LW, Brkic BS, et al. Liver transplantation for hepatitis C: Recurrence and disease progression in 300 patients. Liver Transpl 2000;6: 553-561. 19. McCashland TM. Retransplantation for recurrent hepatitis C: Positive aspects. Liver Transpl 2003;9(Suppl):S67 S72. 20. Burton JR, Jr., Sonnenberg A, Rosen HR. Retransplantation for recurrent hepatitis C in the MELD era: Maximizing utility. Liver Transpl 2004;10(Suppl):S59 S64. 21. Facciuto M, Heidt D, Guarrera J, Bodian CA, Miller CM, Emre S, et al. Retransplantation for late liver graft failure: Predictors of mortality. Liver Transpl 2000;6:174-179. 22. Sheiner PA. Hepatitis C after liver transplantation. Semin Liver Dis 2000;20:201-209. 23. Biggins SW, Beldecos A, Rabkin JM, Rosen HR. Retransplantation for hepatic allograft failure: Prognostic modeling and ethical considerations. Liver Transpl 2002;8:313-322. 24. Markmann JF, Gornbein J, Markowitz JS, Levy MF, Klintmalm GB, Yersiz H, et al. A simple model to estimate survival after retransplantation of the liver. Transplantation 1999;67: 422-430.