Since the beginning of 2002, the priority of adult. Pretransplant MELD Score and Post Liver Transplantation Survival in the UK and Ireland

Transcription

1 Pretransplant MELD Score and Post Liver Transplantation Survival in the UK and Ireland Mathew Jacob, 1 Lynn P. Copley, 1 James D. Lewsey, 1,2 Alex Gimson, 3 Giles J. Toogood, 4 Mohamed Rela, 5 and Jan H. P. van der Meulen, 1,2 on behalf of the UK & Ireland Liver Transplant Audit It has been shown that the model for end-stage liver disease (MELD) score is an accurate predictor of survival in patients with liver disease without transplantation. Four recent studies carried out in the United States have demonstrated that the MELD score obtained immediately prior to transplantation is also associated with post-transplant patient survival. Our aim was to evaluate how accurately the MELD score predicts 90-day post-transplant survival in adult patients with chronic liver disease in the UK and Ireland. The UK and Ireland Liver Transplant Audit has data on all liver transplants since We studied survival of 3838 adult patients after first elective liver transplantation according to United Network for Organ Sharing categories of their MELD scores (< 10, 11 18, 19 24, 25 35, >36). The overall survival at 90-days was 90.2%. The 90-day survival varied according to the United Network for Organ Sharing MELD categories (92.6%, 91.9%, 89.7%, 89.7%, and 70.8%, respectively; P < 0.01). Therefore, only those patients with a MELD score of 36 or higher (3% of the patients) had a survival that was markedly lower than the rest. As a consequence, the ability of the MELD score to discriminate between patients who were dead or alive was poor (c-statistic 0.58). Re-estimating the coefficients in the MELD regression model, even allowing for nonlinear relationships, did not improve its discriminatory ability. In conclusion, in the UK and Ireland the MELD score is significantly associated with post-transplant survival, but its predictive ability is poor. These results are in agreement with results found in the United States. Therefore, the most appropriate system to support patient selection for transplantation will be one that combines a pretransplant survival model (e.g., MELD score) with a properly developed post-transplant survival model. (Liver Transpl 2004;10: ) Since the beginning of 2002, the priority of adult patients with chronic liver disease on the waiting list to receive a cadaveric transplant in the United States is determined on the basis of the model for end-stage liver disease score (MELD), an accurate predictor of survival of patients with end-stage liver disease without transplantation. 1 The MELD was originally developed by Malinchoc and co-workers to predict survival in patients undergoing transjugular intrahepatic portosystemic shunts. 2 Its current use as a prioritization tool is intended to enhance the overall effectiveness and fairness of the organ allocation system. An important advantage of the MELD is that it is based on only three standardized biochemical assessments (i.e., serum bilirubin, serum creatinine, and international normalized ratio [INR] that are objective and rapidly available. A number of recent studies carried out in the United States have demonstrated that the MELD score obtained immediately prior to transplantation is also associated with post-transplant patient survival. 3 7 Patients with high MELD scores have a poorer posttransplant survival. This indicates that patients with high MELD scores have a poorer outcome with as well as without transplantation. The question is therefore to what extent prioritization based on the highest MELD scores identifies patients who will benefit most from transplantation, as this is dependent on the difference in survival with and without transplantation. To study these issues further, we evaluated how accurately the pretransplant MELD score predicts posttransplantation survival in chronic liver disease patients in the UK and Ireland, where the MELD score is not used to prioritize patients for transplantation. It has been suggested that the outcome of liver transplantation depends especially on the functional integrity of the extrahepatic organs. 8 If this were true then it is likely that the relative importance of the three biochemical parameters included in the MELD score differs for the prediction of pre- and post-transplant survival. Therefore, we also explored whether the accuracy of post-transplant survival could be improved by re-es- Abbreviations: MELD, model for end-stage liver disease; UNOS, United Network for Organ Sharing; IQR, interquartile range; INR, international normalized ratio. From the 1 Clinical Effectiveness Unit, Royal College of Surgeons of England, London, UK; 2 Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London, UK; 3 Liver Transplant Unit, Addenbrooke s Hospital, Cambridge, UK; 4 Transplant and Hepatobiliary Unit, St James University Hospital, Leeds, UK; 5 Institute of Liver Studies, King s College Hospital, London, UK. Address reprint requests to Mathew Jacob, Clinical Effectiveness Unit, The Royal College of Surgeons of England, Lincoln s Inn Fields, London, WC2A 3PE. Telephone: ; Fax ; mjacob@rcseng.ac.uk Copyright 2004 by the American Association for the Study of Liver Diseases Published online in Wiley InterScience ( DOI /lt Liver Transplantation, Vol 10, No 7 ( July), 2004: pp

2 904 Jacob et al. timating the coefficients of the regression model from which the MELD score is derived. Materials and Methods UK and Ireland Liver Transplant Audit The UK and Ireland Liver Transplant Audit was initiated in March All eight transplant centers in these countries are obliged to collect prospective data on the characteristics of recipients and donors, the transplantation procedure and post-transplant outcome. Both computer checks and a comparison against medical notes are carried out at regular intervals to ensure the completeness and accuracy of the database. A note validation exercise, covering the period between October 2000 and October 2002, found that 98% of the data was complete, and that 94% of the completed data fields were accurate. Previous note validation exercises produced similar results first liver transplants were recorded in our database between 1 March 1994 and 30 September We excluded patients younger than 17 years (n 695). We also excluded patients who received a super-urgent liver transplantation 9 (n 641) or a multi-organ transplant (n 79). We included patients for whom the MELD score could not be calculated due to missing covariate information by categorizing them separately. The final study group included 3838 patients. MELD scores were calculated, using the biochemical results collected immediately before transplantation: MELD (0.957 ln(creatinine) ln(bilirubin) 1.12 ln(inr) 0.643) 10. We followed the steps outlined by the United Network for Organ Sharing (UNOS) in the United States 10 except that we did not award additional points for hepatocellular cancer patients and the MELD score was not limited to 40. Patients were stratified into 5 UNOS MELD categories according to their pretransplant MELD scores: 10, 11 18, 19 24, and 36, similar to the MELD categories used by UNOS and described by Saab et al. 5 Our main outcome measure was patient survival after transplantation. Statistical Methods Categorical variables were expressed as percentages and continuous variables were represented as medians and inter quartile ranges. Patient survival as a function of time after transplantation was measured using Kaplan-Meier methods. 11 Patients who were alive at the end of follow-up were considered to be censored observations. We compared patient survival according to the UNOS categories of the MELD score. The area under the receiver operating characteristic curve, also known as the c-statistic, was used to evaluate the ability of the MELD score to distinguish between patients who were dead or alive at some point after transplantation. 12 This c-statistic represents the probability that in a pair of patients, one randomly selected from those who died and one from those who survived, the patient who died has a higher MELD score than the patient who survived. A value of 0.5 for the c-statistic would indicate that the MELD does not predict survival at all, and a value of 1 that the MELD score predicts survival perfectly. We used the log-rank test to determine whether the survival differs according to the UNOS categories of the MELD score. We also used log-rank test to assess linear trend of the survivor function across the MELD categories. 13 A P-value of 0.05 or less was considered to indicate a statistically significant result. Cox regression 14 was used to re-estimate the coefficients of the regression model from which the MELD score is derived, and we also explored whether the accuracy of the model would improve by adding nonlinear terms (e.g., squares of the three log transformed biochemical variables) in the regression model. All analyses were carried out with Stata 7 software. 15 Results A total number of 3838 adult patients who received a first elective transplantation could be included of whom 3493 (91%) had complete data for the variables included in the MELD score. Table 1 shows the characteristics of the patients and donors in relation to the MELD score. Alcoholic and cholestatic liver disease accounted for slightly more than half of all transplants. The median serum bilirubin was 3.2 mg/dl; the median serum creatinine 1.0 mg/dl, the median INR 1.4, and the median MELD score was There is an increase in the proportion of inpatients, patients on renal support, and ventilated patients with increasing MELD scores, which indicates a strong association between the MELD score and pretransplant disease severity. Patients with higher MELD scores also had shorter waiting for transplantation (median waiting time of 21 days for patients with MELD scores 36 compared to 57 days for those with MELD score 10). Patient Survival The overall patient survival at 24 months was 80.7% (Table 2). The post-transplant survival varied according to the UNOS MELD categories [log rank test for equality (for linear trend): P 0.01 (P 0.01)]. However, the differences among the patients with a MELD score of less than 36 are small and not statistically significant [log rank test for equality (for linear trend): P 0.28 (P 0.14)]. Only patients with a MELD of 36 or higher had a survival that was markedly lower than the rest (3-month survival 70.8% and 91.3%, and 2-year survival 62.0% and 82.1%, for MELD scores 36 and 36 respectively).

3 Pretransplant MELD Score and Post Transplant Survival 905 Table 1. Pre-transplant Patient and Donor Characteristics in Relation to the MELD Score Variable Categories 10 (n 532) 11 to 18 (n 1701) 19 to 24 (n 741) 25 to 35 (n 399) 36 (n 120) Missing (n 345) All (n 3838) Age (years) Median (IQR) 53 (14.5) 53 (14) 51 (15) 50 (14) 47 (16.5) 52 (15) 52 (15) Gender Male Female Inpatient No Yes Renal No support* Yes Pre-operative No ventilation Yes Diagnosis Alcoholic/ Cholestatic disease Others Ascites No Yes Abdominal No surgery Yes Varices No Yes Creatinine (mg/dl) Median (IQR) 0.9 (0.2) 0.9 (0.3) 1.0 (0.4) 1.2 (0.6) 2.0 (1.7) 1.0 (0.4) (n 254) 1.0 (0.4) (n 3747) Bilirubin (mg/dl) Median (IQR) 1.0 (0.7) 2.7 (2.4) 6.4 (7.3) 9.1 (12.9) 27.2 (24.1) 3.5 (5.1) (n 200) 3.2 (5.1) (n 3693) INR Median (IQR) 1.1 (0.2) 1.3 (0.3) 1.6 (0.5) 1.9 (1.0) 3.1 (2.9) 1.5 (0.6) 1.4 (0.5) (n 82) (n 3575) Organ Normal appearance Abnormal CIT (min) Median (IQR) 685 (267) 691 (248) 702 (244) 693 (235) 675 (325) 689 (287) 691 (252) (n 493) (n 1601) (n 706) (376) (n 114) (n 229) (n 3519) Organ type Whole Partial Waiting time (days) Median (IQR) 57 (114) (n 387) 55 (94) (n 1477) 40 (76) (n 626) 33 (78) (n 333) 21 (47) (n 74) 28 (63) (n 254) 46 (87) (n 3151) Donor cause of death Non traumatic Trauma Others Abbreviations: IQR, interquartile range; UNOS, united network for organ sharing; MELD, model for end-stage liver disease; CIT, cold ischemia time. *Renal support: hemodialysis or hemofiltration in the week before transplantation. Cholestatic disease: primary biliary cirrhosis, primary sclerosing cholangitis, and secondary biliary cirrhosis. Abdominal surgery: previous laparotomy with a supraumbilical incision. NOTE: Percentage may not add up to 100% due to missing patient data. Accuracy of the MELD Score to Predict Post- Transplant Survival The ability of the MELD score to identify patients who were dead or alive 90 days after transplantation was poor (c-statistic was 0.58, 95% confidence interval 0.54 to 0.61). Re-estimating the coefficients in the MELD model, even considering nonlinearity (see Materials and Methods), did only marginally improve the model s accuracy (c-statistic always less than 0.60). Discussion The ability of the MELD score to predict post-transplant survival in patients who underwent an elective liver transplantation in the UK and Ireland appeared to be poor. Only patients with a MELD score of 36 or higher, which accounted for only 3.1% of the cases, had a survival that was markedly decreased. The survival of all other MELD categories was similar. We only considered patients who had an elective

4 906 Jacob et al. Table 2. Survival at Different Intervals Within the First 2 Years after Liver Transplantation According to UNOS Categories of the MELD Score MELD Score No. of Patients 1 Month 3 Months 6 Months 12 Months 24 Months ( ) 92.6 ( ) 90.8 ( ) 87.1 ( ) 82.1 ( ) ( ) 91.9 ( ) 89.3 ( ) 86.1 ( ) 83.1 ( ) ( ) 89.7 ( ) 86.8 ( ) 83.5 ( ) 79.9 ( ) ( ) 89.7 ( ) 87.0 ( ) 84.5 ( ) 81.3 ( ) ( ) 70.8 ( ) 69.0 ( ) 68.0 ( ) 62.0 ( ) Missing MELD ( ) 86.4 ( ) 82.8 ( ) 79.1 ( ) 74.1 ( ) All ( ) 90.2 ( ) 87.6 ( ) 84.4 ( ) 80.7 ( ) transplantation. Patients who were classified as superurgent 9 (broadly similar to patients with UNOS status 1) were excluded. In the UK and Ireland, there is no national waiting list for elective transplantations, and the median waiting time for an elective transplantation for our cohort was only 46 days, which indicates that the pretransplant MELD scores used in this study might not be too different from those at entry on the waiting list. In the UK and Ireland, patient selection for transplantation is determined by the individual transplant centers. 9 The MELD score as a measure of disease severity is not being used for this purpose. It remains possible that our results might have been slightly different if candidates had been prioritized by MELD score. However, we found that the pretransplant MELD scores were strongly associated with known risk factors of pretransplant liver disease severity. Furthermore, waiting times were lower for those patients with higher MELD scores, and particularly short for patients with a MELD score of 36 or higher (median 21 days). Although we do not have information about the patients who were removed from the waiting list (6% in 2002) or who died before transplantation (6% in 2002), these associations might indicate that disease severity is a factor determining patient selection for transplantation and organ allocation in the UK and Ireland. 9 To our knowledge, the ability of the MELD score to predict post-transplant survival has been explored in four independent studies that were all performed in the United States before the MELD score was introduced. 3 6 All these studies included fewer patients than our study. Their results correspond with our observation that patients with a MELD score below a certain threshold have similar post-transplant survival. The exact location of this threshold is difficult to determine as the studies report the survival of patients according to slightly different MELD score categories (two studies used 25, one 36, and one 40). It has been argued that the use of the MELD score to prioritize patients could result in a decrease in posttransplant survival, as sicker patients will have priority for transplantation. 16 Our study based on patients transplanted in the UK and Ireland as well as the studies carried out in the United States indicate that posttransplant survival according to MELD score categories hardly differs for up to 95% of the patients. This indicates that the fear that the results of liver transplantation will deteriorate if the MELD score is introduced may be unjustified. This conclusion corresponds with the preliminary results of a study that demonstrated that waiting list mortality went down and post transplant mortality remained more or less the same after the introduction of a quantitative disease severity score system to prioritize patients for transplantation. 17 Patients with liver disease who have a high risk of dying without transplant but who have a high chance of success with a transplant should be prioritized to receive an organ for transplantation. 18 The MELD model was originally developed to predict survival without transplantation (or more precisely, for patients who received a transjugular intrahepatic portal systemic shunt. 2 As increasingly more models, attempting to predict posttransplant survival, become available 3,19 23 ; the most appropriate system to assist patient selection for transplantation will therefore be one that incorporates predictions about survival without as well with transplantation. In other words, the MELD score should be combined with a score derived from a properly developed and validated post-transplantation survival model. The first challenge in this context is to address the dilemma how the results from pre- and post-transplant models should be combined. Moreover, when assessing organ allocation systems one has to consider

5 Pretransplant MELD Score and Post Transplant Survival 907 survival from the point of entering a waiting list, as opposed to having the transplant, as this would take account of waiting list mortality. The second challenge is to improve the predictive ability of post-transplant survival models. We might have to accept that the ability to predict post-transplant outcome based on pretransplant characteristics will always be inadequate given the influence of chance events that occur in the perioperative period have on post-transplant survival. 24 The third challenge is that for certain indications, such as hepatocellular carcinoma, the need for liver transplantation can not be solely based on the severity of liver failure, which demonstrates the need for additional criteria for patient selection for transplantation. Acknowledgments The authors thank the following members of the UK and Ireland Liver Transplant Audit and their departments: Mr. Derek Manas and Liesl Smith (Freeman Hospital, Newcastle, UK), Mr. Steve Pollard and Christine Sutton (St James Hospital, Leeds, UK), Mr. N V Jamieson and Claire Jenkins (Addenbrooke s Hospital, Cambridge, UK), Mr. Keith Rolles and Dr. Nancy Rolando (Royal Free Hospital, London, UK), Mr. Nigel Heaton and Susan Landymore (King s College Hospital, London, UK), Mr. A D Mayer and Birdget Gunson (Queen Elizabeth Hospital, Birmingham, UK), Mr. O Traynor and Mr. Mashood Ahmed (St.Vincent s Hospital, Dublin, Republic of Ireland), Mr. John Forsythe and Maureen Cunningham (The Royal Infirmary at Edinburgh, Edinburgh, UK), and Kerri Barber (UK Transplant, Bristol, UK). The UK and Ireland Liver Transplant Audit is funded by the National Specialist Commissioning Advisory Group (NSCAG) of the Department of Health, London UK. References 1. Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001;33: 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: Desai NN, Mange KC, Crawford MD, Abt PL, Frank AM, Markmann JW, et al. Predicting outcome after liver transplantation: utility of the model for end-stage liver disease and a newly derived discrimination function. Transplantation 2004;77: Onaca NN, Levy MF, Sanchez EQ, Chinnakotla S, Fasola CG, Thomas MJ, et al. A correlation between the pretransplant meld score and mortality in the first two years after liver transplantation. Liver Transpl 2003;9: 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: Kim WR, Wiesner RH, Kamath PS, Malinchoc M, Kremers WK, Rosen CB, et al. Prediction of liver transplant outcome using the MELD scale [abstract]. Transplantation 2001; 71(suppl 1): Onaca N, Levy MF, Netto GJ, Thomas MJ, Sanchez EQ, Chinnakotla S, et al. Pretransplant MELD score as a predictor of outcome after liver transplantation for chronic hepatitis C. Am J Transpl 2003;3: Neuberger J, Altman DG, Polson R, Buckels J, Rolles K, Elias E et al. Prognosis after liver transplantation for biliary cirrhosis. Transplantation 1989;48: UK Transplant. Available at: Accessed July United Network for Organ Sharing. Available at: Accessed June Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Statistical Assoc 1958;53: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: Collett D. Modelling survival data in medical research. In Texts in Statistical Science, Chatfield C, Zidek JV Eds. 1 st ed., Chapt 2. London: Chapman & Hall, 1994: Cox DR. Regression models and life tables. J Royal Statistical Soc 1972;B,74: Statacorp Statistical Software: Release 7.0. College Station, TX: Stata Corporation. 16. Llado L, Figeras J, Memba R, Xiol X, Baliellas C, V azquez S, et al. Is MELD really the definitive score for liver allocation?. Liver Transpl 2002;8: Freeman RB, Rohrer RJ, Katz E, Lewis WD, Jenkins R, Cosimi AB et al. Preliminary results of a liver allocation plan using a continuous medical severity score that de-emphasises waiting time. Liver Transpl 2001;7: Freeman RB, Wiesner RH, Harper A, McDiarmid SV, Lake J, Edwards E, et al. The new liver allocation system: moving towards evidence-based transplantation policy. Liver Transpl 2002;8: Doyle HR, Marino IR, Jabbour N, Zetti G, McMichael J, Mitchell S, et al. Early death or retransplantation in adults after orthotopic liver transplantation. Can outcome be predicted? Transplantation. 1994;57: Adam R, Cailliez V, Majno P, Karam V, McMaster P, Calne RY et al. Normalised intrinsic mortality risk in liver transplantation: European Liver Transplant Registry Study. Lancet 2000;356: 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: Thuluvath PJ, Yoo HY, Thompson RE. A model to predict survival at one month, one year and five years after liver transplantation based on pretransplant clinical characteristics. Liver Transpl 2003;9: Bilbao I, Armadans L, Lazaro JL, Hidalgo E, Castells L, Margarit C. Predictive factors for early mortality following liver transplantation. Clin Transpl 2003 Oct; 17(5): Kim RW. Pretransplantation disease severity and post transplantation outcome. Liver Transpl 2003;9: