Begin typing your search above and press return to search.
Volume: 19 Issue: 6 June 2021

FULL TEXT

ARTICLE
Combining Donor and Recipient Age With Preoperative MELD and UKELD Scores for Predicting Survival After Liver Transplantation

Abstract

Objectives: The end-stage liver disease scoring systems MELD, UKELD, and D-MELD (donor age × MELD) have had mediocre results for survival assessment after orthotopic liver transplant. Here, we introduced new indices based on preoperative MELD and UKELD scores and assessed their predictive ability on survival posttransplant.
Materials and Methods: We included 1017 deceased donor orthotopic liver transplants that were performed between 2008 (the year UKELD was introduced) and 2019. Donor and recipient characteristics, liver disease scores, transplant characteristics, and outcomes were collected for analyses. D-MELD, D-UKELD (donor age × UKELD), DR-MELD [(donor age + recipient age) × MELD], and DR-UKELD [(donor age + recipient age) × UKELD] were calculated.
Results: No score had predictive value for graft survival. For patient survival, DR-MELD and DR-UKELD provided the best results but with low accuracy. The highest accuracy was observed at 1 year posttransplant (areas under the curve of 0.598 [95% CI, 0.529-0.667] and 0.609 [95% CI, 0.549-0.67] for DR-MELD and DR-UKELD). Addition of donor and recipient age significantly improved the predictive abilities of MELD and UKELD for patient survival, but addition of donor age alone did not. For 1-year mortality (using receiver operating characteristic curves), optimal cut-off points were
DR-MELD >2345 and DR-UKELD >5908. Recipients with DR-MELD >2345 (P < .001) and DR-UKELD >5908
(P = .002) had worse patient survival within the first year, but only DR-MELD >2345 remained significant after multiva­riable analysis (P = .007).
Conclusions: DR-MELD and DR-UKELD scores provided the best, albeit mediocre, predictive ability among
the 6 tested models, especially at 1 year after posttransplant, although only for patient but not for graft survival. A DR-MELD >2345 was considered to be an additional independent risk factor for worse recipient survival within the first postoperative year.


Key words : Cirrhosis, Model for End-Stage Liver Disease, Risk factors, United Kingdom Model for End-Stage Liver Disease

Introduction

Malinchoc and associates introduced the Model for End-Stage Liver Disease (MELD) score in 2000 as a tool to predict 3-month survival of patients with cirrhosis undergoing transjugular intrahepatic portosystemic shunts, based on serum bilirubin, the international normalized ratio for prothrombin time (INR), and serum creatinine.1 Kamath and colleagues2 in 2001 and Wiesner and colleagues3 in 2003 demonstrated that MELD scores can also be used to predict 3-month survival of patients with cirrhosis and end-stage liver disease who are waiting for orthotopic liver transplant (OLT) and could be used to allocate organs to those most in need of transplant. Neuberger and colleagues4 introduced the United Kingdom Model for End-Stage Liver Disease (UKELD) score in 2008 as a new tool for predicting mortality of patients with cirrhosis who are on wait lists for OLT, based on serum bilirubin, INR, serum creatinine, and serum sodium. According to Barber and associates,5 the UKELD score is superior to the MELD score in predicting of 3-month and 1-year mortality of patients with cirrhosis who are waiting for transplant.

Both the MELD and UKELD scores have been used to predict graft and recipient survival after OLT; however, the accuracy is low with an area under the sensitivity/specificity curve (AUC) of less than 0.7.6-12 In 2009, Halldorson and colleagues13 described the D-MELD score, which is the product of donor age and preoperative MELD score; this score aimed to combine a major factor of liver graft quality (namely, donor age) with the severity of the recipient’s liver failure, as depicted by MELD score. However, AUC values remained less than 0.7,6-9,14 and comparative studies showed no significant differences among MELD, UKELD, and D-MELD scores.7,9-11

In the present study, our aim was to introduce new indices that combined preoperative MELD or UKELD scores with donor and recipient age. We assessed their predictive ability with regard to graft and patient survival after OLT versus the predictive ability of already existing aforementioned scores. In particular, we introduced D-UKELD score (the product of donor age and UKELD score), DR-MELD score (the product of MELD score and the sum of donor and recipient age), and DR-UKELD score (the product of UKELD score and the sum of donor and recipient age). We also assessed whether the addition of donor age or the sum of donor and recipient age to MELD or UKELD score improved their prognostic ability and tried to identify the optimal cut-off points of the scores that provided the best results.

Materials and Methods

Patients
We retrospectively reviewed our prospectively maintained database in November 2019 for OLTs performed in our department between January 2008 and November 2019. This is the source of validated data for our department provided to the National Health Service Blood and Transplant organization. We chose January 2008 as the beginning of our study period because the UKELD score was introduced in that year. Over this period, 1038 liver-only transplants from deceased donors were performed: 870 (83.8%) from donation after brain death (DBD) and 168 (16.2%) from donation after circulatory death (DCD). We included 1017 OLTs (849 DBDs [83.5%] and 168 DCDs [16.5%]) in our analysis, after excluding 21 cases for which we did not have both preoperative MELD and UKELD scores available. Table 1 sum­marizes patient and transplant characteristics.

Data were collected concerning the following parameters: (1) donor characteristics (age, sex, DBD or DCD donor type, whole or split graft type, liver steatosis); (2) recipient characteristics (age, sex, ethnicity, body mass index, urgency for OLT [urgent or elective], indication for OLT, previous OLT, preoperative MELD score, preoperative UKELD score); (3) transplant technique (caval replacement or piggyback); (4) cold ischemia time (time between aortic cross-clamp with initiation of donor’s cold perfusion and liver reperfusion with recipient’s blood); and (5) transplant outcomes (graft survival, patient survival).

We calculated MELD score as follows: MELD = 9.57 × ln(serum creatinine [mg/dL]) + 3.78 × ln(serum bilirubin [mg/dL]) + 11.2 × ln(INR) + 6.43 × (0 if cholestatic or alcoholic etiology, 1 otherwise).1-3
D-MELD was estimated as the product of donor age and preoperative MELD score (D-MELD = [donor age] × MELD).13 DR-MELD was estimated as the product of preoperative MELD score and the sum of donor and recipient age (DR-MELD = [donor age + recipient age] × MELD).

We calculated UKELD as follows: UKELD = 5.395 × ln(INR) + 1.485 × ln(serum creatinine [μmol/L]) + 3.13 × ln(serum bilirubin [μmol/L]) – 81.565 × ln(serum sodium [mmol/L]) + 435.4,5 D-UKELD score was estimated as the product of donor age and preoperative UKELD score (D-UKELD = [donor age] × UKELD). DR-UKELD was estimated as the product of preoperative UKELD score and the sum of donor and recipient age (DR-UKELD = [donor age + recipient age] × UKELD).

Our study conformed to the 2000 Declaration of Helsinki, the 2008 Declaration of Istanbul, and the guidelines of the ethics committee of our institution.

Statistical analyses
Time-dependent receiver operating characteristic (ROC) analysis was performed to assess the ability of MELD, D-MELD, DR-MELD, UKELD, D-UKELD, and DR-UKELD scores to predict graft loss and death across time, using the “timeROC” R package. The AUC was calculated at 3 different time points: 1 year, 3 years, and 5 years post-OLT, and the change of AUC through time was graphically depicted for all scores in terms of graft loss and mortality. The Youden’s J statistic was applied to the ROC curves that provided the best results, in order for the optimal cut-off point to be identified. Based on the optimal cut-off points, patients were divided into groups. Kaplan-Meier curves and the log-rank test were used for the assessment and comparison of graft and patient survival between patient groups. Cox regression was used for multivariable graft and patient survival analysis. All tests were 2-tailed. The level of statistical significance was set at alpha = 0.05. Because the number of missing values was low (97/18 209 values [0.53%]) and the percentage of patients with at least 1 missing value was also low (62/1017 patients [6.1%]), we excluded cases pairwise for univariate analyses and listwise for multivariate analyses. Version 4.0.0 of the R Statistical Software (R Foundation for Statistical Computing) was used for the time-dependent ROC analysis, and version 25 of Statistical Package for Social Sciences (SPSS; IBM Corporation) was used for the rest of the statistical analysis.

Results

Time-dependent receiver operating characteristic analysis concerning graft survival
Time-dependent ROC analysis did not provide statistically significant results with regard to graft survival at 1, 3, or 5 years after OLT for any of the tested scores, as 95% CI included values on both sides of 0.5. Furthermore, 95% CI contained values on both sides of 0.5 for the whole length of the AUC graphs (Figure 1). In addition, the combination of donor age or of the sum of donor and recipient age with MELD or UKELD score did not improve their predictive abilities with regard to graft survival. No significant differences were detected when we compared MELD with UKELD score, D-MELD with D-UKELD score, or DR-MELD with DR-UKELD score. The exact results of time-dependent ROC analysis in terms of graft survival are shown in Table 2.

Time-dependent receiver operating characteristic analysis regarding patient survival
Time-dependent ROC analysis did not provide statistically significant results with regard to patient survival at 1, 3, or 5 years post-OLT for either MELD or UKELD score, as 95% CI included values on both sides of 0.5. This was also true for the whole length of their AUC graphs (Figure 2). However, outcomes were different when donor age or the sum of donor and recipient age were added to MELD and UKELD scores. In particular, D-MELD and D-UKELD scores resulted in a statistically significant, although low, AUC at 1 year post-OLT (for D-MELD, AUC was 0.588 [95% CI, 0.52-0.657]; for D-UKELD, AUC was 0.58 [95% CI, 0.517-0.643]). As shown in their AUC graphs, these scores lost statistical significance after the second postoperative year (Figure 2). With regard to DR-MELD and DR-UKELD scores, both provided higher AUCs compared with the other scores, although still low. Their highest AUC values corresponded to the first postoperative year (for
DR-MELD, AUC was 0.598 [95% CI, 0.529-0.667]; for DR-UKELD, AUC was 0.609 [95% CI, 0.549-0.67]), as shown in their AUC graphs (Figure 2). Figure 3 depicts ROC curves for DR-MELD and DR-UKELD scores for death within the first year after OLT.

The addition of donor age to either MELD or UKELD score did not significantly increase the predictive ability with regard to patient survival. The addition of the sum of donor and recipient age to either MELD or UKELD score increased the predictive ability significantly with regard patient survival at 1 year (P < .001) and 3 years (P = .038) post-OLT for DR-MELD score and at 1 year (P = .046), 3 years (P = .037), and 5 years (P = .031) for DR-UKELD score. Nevertheless, no significant differences were detected when we compared MELD with UKELD score, D-MELD with D-UKELD score, or DR-MELD with DR-UKELD score. The exact results of time-dependent ROC analysis in terms of patient survival are shown in Table 3.

Patient survival analysis based on DR-MELD score
The mean (SD) surveillance period was 1558.9 (1227.9) days, and the median surveillance period was 1323 days (minimum to maximum, 0-4318) for the entire cohort. After the application of the Youden’s J statistic to the ROC curve for DR-MELD score at 1 year post-OLT, a DR-MELD score of 2345 was chosen as the optimal cut-off point. In our study, 847 patients (83.3%) had DR-MELD ≤2345 and 170 patients (16.7%) had DR-MELD >2345. When follow-up was truncated at 1 year post-OLT, recipients with DR-MELD >2345 had shorter patient survival (mean [SE] of 314.9 [8.8] days; 95% CI, 297.6-332.2; median not yet reached) than recipients with DR-MELD ≤2345 (mean [SE] of 347.4 [2.5] days; 95% CI, 342.6-352.2; median not yet reached yet) (P < .001) by 32.5 days on average (Figure 4A). A DR-MELD >2345 remained an independent prognostic factor of worse patient survival within the first postoperative year in the multivariable Cox regression analysis (hazard ratio [HR] 2.263; 95% CI, 1.257-4.074; P = .007) (Table 4).

When the optimal cut-off point for predicting 1-year patient survival was applied to the overall patient survival, recipients with DR-MELD >2345 had still shorter patient survival (mean [SE] of 3227.1 [152.9] days; 95% CI, 2927.4-3526.7) than recipients with DR-MELD ≤2345 (mean [SE] 3535.6 [59.8] days; 95% CI, 3418.4-3652.7) (P = .005) by 308.5 days on average. Analyses of the Kaplan-Meier curves identified a deviation between the curves in the first postoperative year, but they were parallel after the second postoperative year (Figure 4B). Thus, the impact on patient survival during the initial postoperative period caused by the higher DR-MELD score was not enough to produce a significant difference in the overall patient survival, as it was also shown in the multivariable Cox regression analysis (HR of 1.523; 95% CI, 0.972-2.387; P = .067) (Table 5).

Patient survival analysis based on DR-UKELD score
After the application of the Youden’s J statistic to the ROC curve for DR-UKELD score at 1 year post-OLT, a DR-UKELD score of 5908 was chosen as the optimal cut-off point. In our study, 646 patients (63.5%) had DR-UKELD ≤5908 and 371 patients (36.5%) had DR-UKELD >5908. When follow-up was truncated at 1 year post-OLT, recipients with DR-UKELD >5908 had shorter patient survival (mean [SE] of 332.5 [4.9] days; 95% CI, 322.9-342.1; median not yet reached) than recipients with DR-UKELD ≤5908 (mean [SE] of 347.3 [2.9] days; 95% CI, 341.7-352.9; median not yet reached) (P = .002) by 14.8 days (Figure 5A). Nevertheless, a DR-UKELD >5908 did not remain an independent prognostic factor of worse patient survival within the first postoperative year in the multivariable Cox regression analysis (HR of 1.588; 95% CI, 0.966-2.611; P = .068) (Table 4).

When the optimal cut-off point for predicting 1-year patient survival was applied to the overall patient survival, recipients with DR-UKELD >5908 had still shorter patient survival (mean [SE] of 3274.2 [105.2] days; 95% CI, 3067.9-3480.4) than recipients with DR-UKELD ≤5908 (mean [SE] of 3583.8 [65.4] days; 95% CI, 3455.6-3711.9) (P = .007) by 309.6 days on average. Analyses of the Kaplan-Meier curves identified a deviation between the curves in the first postoperative year, but they were parallel after the second postoperative year (Figure 5A). Thus, the impact on patient survival during the initial postoperative period caused by the higher DR-UKELD score was not enough to produce a significant difference in the overall patient survival, as it was also shown in the multivariable Cox regression analysis (HR of 1.339; 95% CI, 0.946-1.894; P = .1) (Table 5).

Discussion

The aim of our study was to assess the ability of MELD and UKELD scores, as well as scores derived from them, to predict graft loss and mortality after OLT. We also aimed to investigate whether the predictive ability of MELD and UKELD scores improved after adding donor age or the sum of donor and recipient age to each model. Following the logic behind D-MELD score,13 we combined a major factor of liver graft quality (namely, donor age) with the severity of the recipient’s liver failure, as depicted by UKELD score this time, in order to create the D-UKELD score. Extending this logic of combining donor and recipient characteristics even further, we modified D-MELD and D-UKELD scores by also adding recipient age in each model and introduced the DR-MELD and DR-UKELD scores. We also tried to identify the optimal cut-off points of the scores that provided the best results leading to significant differences regarding survival after OLT.

Our findings are in accordance with other studies on MELD, UKELD, and D-MELD scores, reporting no or poor predictive ability in terms of graft loss and death.6-12,14 First, we found that none of the 6 tested scores had any actual ability to predict graft loss after OLT. In addition, MELD and UKELD scores did not have any ability to predict death after OLT. However, D-MELD, D-UKELD, DR-MELD, and DR-UKELD scores showed some predictive ability, albeit mediocre, with regard death, but mainly for the first 2 years post-OLT (3 years for the DR-UKELD score). The highest AUCs were detected at 1 year after OLT, with DR-MELD and DR-UKELD scores providing higher AUCs than D-MELD and D-UKELD, respectively. Another interesting finding was that the improvement in AUCs by the addition of just donor age to MELD or UKELD models was not statistically significant, which agrees with what a few other studies have already reported.7,9-11 However, the addition of both donor and recipient age to MELD and UKELD scores resulted in a statistically significant improvement in their AUCs. Moreover, MELD and UKELD did not seem to have different predictive ability, and the same was true when we compared the scores derived from them (namely, D-MELD with D-UKELD and DR-MELD with DR-UKELD). Thus, MELD and UKELD have the same predictive ability, which significantly improved in terms of patient survival when both donor and recipient age were added to each model; however, it only reached mediocre levels and was best for 1-year patient survival post-OLT.

The mediocre predictive ability of DR-MELD and DR-UKELD with regard to patient survival can be attributed to the fact that donor age, recipient age, and MELD and UKELD scores are not the only parameters that affect recipient survival after OLT. Nevertheless, DR-MELD and DR-UKELD could still provide useful information, and cut-off points were identified from ROC curves associated with shorter patient survival within the first year after OLT (2345 for DR-MELD and 5908 for DR-UKELD). Recipients with DR-MELD >2345 lost 1 month of survival within the first postoperative year and around 10 months of overall survival, corresponding to about one-fifth of the mean surveillance period of our study. Recipients with DR-UKELD >5908 lost only 2 weeks of survival within the first postoperative year, but they lost around 10 months of overall survival. Nonetheless, Kaplan-Meier curves showed that, after an initial deviation between the 2 curves during the first postoperative year, these were running practically in parallel and close to each other after the second postoperative year. In the multivariable survival analysis, only DR-MELD score provided statistically significant results and only for the first postoperative year. In particular, DR-MELD >2345 was an independent risk factor of worse patient survival within the first postoperative year, leading to a more than double risk of dying by the end of the first year after OLT (HR of 2.263), but this was not enough to sustain a long-term hit in overall survival. We also found that DR-UKELD >5908 was not an independent risk factor of worse 1-year or overall survival.

At this point, we would like to mention that there are some clear limitations to this study. Although we included a large cohort of more than 1000 consecutive deceased donor OLTs and retrieved the relevant information from our prospectively maintained database of transplant recipients, this is a retrospective study based on single-site data. An external validation of the newly described scores that used an independent cohort or a national data set would also be needed. The next step would be to combine MELD and UKELD scores with other already reported risk scores, combining additional donor and/or recipient parameters, such as donor risk index,15 donor quality index,16 UK donor liver index,17 UCLA score for DCD liver transplants,17 UK DCD risk score,19 and others, and test these on multicenter or national data.

Conclusions

Our analyses showed that DR-MELD and DR-UKELD scores provided the best, albeit mediocre, predictive ability among the 6 tested models, especially at 1 year post-OLT, but only for patient survival and not for graft survival. Because of the mediocre predictive ability, none of the tested models can be used on their own to predict death after OLT. However, a high DR-MELD score can be considered in clinical practice an additional independent risk factor for worse recipient survival within the first postoperative year, although not concerning overall survival. We found that DR-UKELD score did not seem to have a similar use regarding short-term or overall survival. Because this is the first study to introduce D-UKELD, DR-MELD, and DR-UKELD scores and to look into their predictive roles in post-OLT survival, more studies are needed for our results to be elucidated.


References:

  1. 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(4):864-871. doi:10.1053/he.2000.5852
  2. Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology. 2001;33(2):464-470. doi:10.1053/jhep.2001.22172
  3. Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology. 2003;124(1):91-96. doi:10.1053/gast.2003.50016
  4. Neuberger J, Gimson A, Davies M, et al. Selection of patients for liver transplantation and allocation of donated livers in the UK. Gut. 2008;57(2):252-257. doi:10.1136/gut.2007.131730
  5. Barber K, Madden S, Allen J, et al. Elective liver transplant list mortality: development of a United Kingdom end-stage liver disease score. Transplantation. 2011;92(4):469-476. doi:10.1097/TP.0b013e318225db4d
  6. Aberg F, Nordin A, Makisalo H, Isoniemi H. Who is too healthy and who is too sick for liver transplantation: external validation of prognostic scores and survival-benefit estimation. Scand J Gastroenterol. 2015;50(9):1144-1151. doi:10.3109/00365521.2015.1028992
  7. Ma Y, Wang Q, Yang J, Yan L. Comparison of different scoring systems based on both donor and recipient characteristics for predicting outcome after living donor liver transplantation. PLoS One. 2015;10(9):e0136604. doi:10.1371/journal.pone.0136604
  8. El Amir M, Gamal Eldeen H, Mogawer S, et al. Different score systems to predict mortality in living donor liver transplantation: which is the winner? The experience of an Egyptian center for living donor liver transplantation. Transplant Proc. 2015;47(10):2897-2901. doi:10.1016/j.transproceed.2015.10.022
  9. Badawy A, Kaido T, Hamaguchi Y, et al. Is Muscle MELD a more promising predictor for mortality after living donor liver transplantation? Prog Transplant. 2018;28(3):213-219. doi:10.1177/1526924818781571
  10. Chung IS, Park M, Ko JS, Gwak MS, Kim GS, Lee SK. Which score system can best predict recipient outcomes after living donor liver transplantation? Transplant Proc. 2012;44(2):393-395. doi:10.1016/j.transproceed.2012.01.064
  11. Ling Q, Dai H, Zhuang R, et al. Predicting short-term survival after liver transplantation on eight score systems: a national report from China Liver Transplant Registry. Sci Rep. 2017;7:42253. doi:10.1038/srep42253
  12. Jacob M, Copley LP, Lewsey JD, et al. Pretransplant MELD score and post liver transplantation survival in the UK and Ireland. Liver Transpl. 2004;10(7):903-907. doi:10.1002/lt.20169
  13. Halldorson JB, Bakthavatsalam R, Fix O, Reyes JD, Perkins JD. D-MELD, a simple predictor of post liver transplant mortality for optimization of donor/recipient matching. Am J Transplant. 2009;9(2):318-326. doi:10.1111/j.1600-6143.2008.02491.x
  14. Avolio AW, Cillo U, Salizzoni M, et al. Balancing donor and recipient risk factors in liver transplantation: the value of D-MELD with particular reference to HCV recipients. Am J Transplant. 2011;11(12):2724-2736. doi:10.1111/j.1600-6143.2011.03732.x
  15. Feng S, Goodrich NP, Bragg-Gresham JL, et al. Characteristics associated with liver graft failure: the concept of a donor risk index. Am J Transplant. 2006;6(4):783-790. doi:10.1111/j.1600-6143.2006.01242.x
  16. Winter A, Feray C, Audureau E, et al. A donor quality index for liver transplantation: development, internal and external validation. Sci Rep. 2018;8(1):9871. doi:10.1038/s41598-018-27960-7
  17. Collett D, Friend PJ, Watson CJ. Factors associated with short- and long-term liver graft survival in the United Kingdom: development of a UK donor liver index. Transplantation. 2017;101(4):786-792. doi:10.1097/TP.0000000000001576
  18. Collett D, Friend PJ, Watson CJ. Factors associated with short-and long-term liver graft survival in the United Kingdom: development of a UK donor liver index. Transplantation. 2017;101(4):786-792. doi:10.1097/TP.0000000000001576
  19. Schlegel A, Kalisvaart M, Scalera I, et al. The UK DCD Risk Score: A new proposal to define futility in donation-after-circulatory-death liver transplantation. J Hepatol. 2018;68(3):456-464. doi:10.1016/j.jhep.2017.10.034



Volume : 19
Issue : 6
Pages : 570 - 579
DOI : 10.6002/ect.2020.0513


PDF VIEW [1186] KB.
FULL PDF VIEW

From the Department of HPB Surgery and Liver Transplantation, Royal Free Hospital, Royal Free London NHS Foundation Trust, London, UK; and the Division of Surgery and Interventional Science, University College London, London, UK
Acknowledgements: The authors have not received any funding or grants in support of the presented research or for the preparation of this work and have no declarations of potential conflicts of interest.
Corresponding author: Ioannis D. Kostakis, Department of HPB Surgery and Liver Transplant, Royal Free Hospital, Royal Free London NHS Foundation Trust, Pond Street, NW3 2QG, London, UK
E-mail: i.d.kostakis@gmail.com