Objectives: Scarcity of liver grafts has led to the use of marginal donors, consequently increasing the number of complications posttransplant. To prevent this situation, several indicators have been developed. However, important differences remain among countries. Here, we compared an early-risk liver transplant indicator based on the Spanish Liver Transplant Registry, called the Graft Risk Index, versus the US donor risk index and the Eurotransplant donor risk index.

Materials and Methods: The new indicator was based on prospectively collected data from 600 adult liver transplants performed in our center. We considered 2 events to compare the indexes: graft survival and rejection-free graft survival, with Cox proportional regression for analyses. Power to predict graft survival was evaluated by calculating the receiver operating characteristic area under the curve.

Results: We found no differences between the US and Eurotransplant donor risk indexes in prediction of patients with and without early graft failure. With regard to early survival, only the Graft Risk Index allowed better survival discrimination, in which survival progressively decreased with values ≥ 3 (with probability of graft survival at 1 month of 68%; 95% confidence interval, 46.2-82.5). This increase in risk was significant compared with the standard group (hazard ratio of 10.15; 95% confidence interval, C 3.91- 26.32; P < .001). We calculated powers of prediction of 0.52 (95% confidence interval, 0.43-0.62), 0.54 (95% confidence interval, 0.45-0.65), and 0.69 (95% confidence interval, 0.61-0.77) for donor risk index, Eurotransplant donor risk index, and early Graft Risk Index, respectively.

Conclusions: Neither the US donor risk index nor the Eurotransplant donor risk index was valid for our Spanish liver donation and transplant program. Therefore, an indicator to predict posttransplant graft survival that is adapted to our environment is necessary. This national Graft Risk Index can be a useful tool to optimize donor-recipient matching.

Key words : Donor Risk Index, Early complications, Eurotransplant Donor Risk index, Graft survival, Organ Procurement and Transplantation Network, Rejection-free graft survival

Introduction

In Spain, more than 1000 transplant procedures have been performed each year in the past decade; more than half of these patients will have at least 1 significant complication during their first year posttransplant.¹ Specifically, in the immediate postoperative period, they are admitted to the intensive care unit (ICU) and may have systemic complications, including respiratory, cardiovascular, renal, and hematologic, as well as graft complications (technical and nontechnical), including primary graft failure, arterial or portal thrombosis, and rejection. These complications can lead to longer ICU and hospital stays and can result, in many cases, in graft failure (retransplant) or even patient death.

High complication rates may be associated with the use of expanded criteria donors and/or recipients with comorbidities and poor clinical conditions pretransplant. An important way to prevent these complications is appropriate donor-recipient matching. For this purpose, several indicators have been described: Model for End-Stage Liver Disease (MELD), Child-Turcotte-Pugh score,² Newly Derived Discrimination Function,³ Score of Liver Donor,⁴ the Donor Risk Index (DRI),⁵ Survival Outcome Following Liver Transplantation (SOFT),⁶ donor-MELD (D-MELD),⁷ Balance of Risk score,⁸ Eurotransplant Donor Risk Index (ET-DRI),⁹ Neural Network-Correct Classification Rate, Neural Network-Minimum Sensitivity,¹⁰ Neural Network,¹¹ and the Donor Liver Index.¹²

There are important differences among organ donation and transplant programs in different countries and organizations in terms of both donor characteristics and transplant results. These differences are more striking regarding donor age and ethnicity, number of donors after brain death versus donors after cardiac death, causes of brain death, controlled hepatic bipartition, and other factors. For example, for the French donation program, Winter and associates have shown that DRI and ET-DRI were not useful.¹³

All of the above factors indicate that donors and recipients can differ between countries and/or regions; therefore, a region-specific scoring system that could be used for liver selection and allocation would be appropriate. The Spanish Liver Transplant Registry (RETH; for its acronym in Spanish) was created with the cooperation of professional teams incorporated in the Spanish Liver Transplantation Society and the Spanish National Transplant Organization. The RETH records the experiences of liver transplant in Spain; the database contains structured data of all liver transplants performed in the country, thus serving as a knowledge repository of our own experience.

In this study, our main objective was to determine graft quality when we compared the DRI, the ET-DRI, and our region-specific Graft Risk Index (GRI).14 We used data from 22 846 liver transplants from RETH, which included liver transplant procedures done at our center. The GRI, a new early risk indicator in liver transplant, was based on outcomes recorded in RETH. This Spanish-specific indicator, which combines donor- and recipient-related factors, could facilitate patient-specific decision-making in organ selection and allocation.

Materials and Methods

The description of the new indicator was based on data collected from the annual report from RETH,¹⁵ which describes donor characteristics and liver transplant results in Spain. In particular, the structured data include 22 846 liver transplants, which were performed in 24 transplant centers from 1984 to 2016. This report is in the public domain and is available online.

For validation of described indicators, we included prospectively collected data of 600 adult liver transplants (recipients were older than 15 years) performed consecutively in our center. The study complied with Organic Law 15/1999 on Personal Data Protection. It was also approved by the Aragon ethics committee (CodeCEICA PI18/0097) on April 24, 2018.

The RETH report also included a Cox proportional hazards regression analysis of the factors associated with overall graft survival. Results are expressed as Cox regression values with their 95% confidence intervals (95% CI).

Based on the Cox regression equation itself, ln (λt) = a + b₁x₁ + b₂x₂ + … + bnxn, an indicator can be defined as the exponential (inverse logarithm) of a linear risk score. By grouping variables with statistically significant differences for multivariate analysis, 2 GRIs can be defined, one overall (global) and one early (< 1 month).

To evaluate the indicators, 2 types of events were considered. The first event was graft survival, defined as the period from date of transplant to date of retransplant or death for any reason, whichever occurred first. The second event was rejection-free graft survival, defined as the period from date of transplant to date of retransplant or death due to chronic graft dysfunction, whichever occurred first.

The Hosmer-Lemeshow goodness of fit test (P < .05) was used to evaluate model fit. We then obtained the values used to generate the logistic regression curve: regression coefficients (β0 and β1) with their corresponding standard errors, the Wald statistic, and odds ratio and confidence intervals.

To compare indicators, graft survival curves were calculated by indicator-related risk group using Kaplan-Meier estimations. The hazard ratio (HR) for each risk group was estimated versus a standard group using a Cox regression model.

The accuracy of DRI, ET-DRI, and GRI to predict graft survival was assessed using the area under the curve in a receiver operating characteristic curve analysis, where 1 represents perfect discrimination and 0.5 represents discrimination not above chance level.

A Wald P value < .05 was considered significant. All analyses were performed with IBM SPSS Statistics version 22.0 (IBM Corporation, Chicago, IL, USA).

Results

To describe and calculate the GRI, global and early GRIs were obtained using Cox logistic regression analysis on global graft survival data from the RETH (Table 1). Table 1 shows an example of a global GRI with risk 1 and an example of maximum risk (GRI = 81.04); the examples shown are accompanied by DRI and ET-DRI scores.

To validate new indicators besides DRI and ET-DRI, data from 600 adult liver transplants performed at our center were included. Validation series data are described in Tables 2, 3, and 4.

With reference to the logistic function of the GRI, the Hosmer-Lemeshow test revealed that the probabilities observed were similar to those expected for the 3 indicators concerned: DRI (P = .20), ET-DRI (P = .25), global GRI (P = .20), and early GRI (P = .28). Figure 1 shows the logistic equation curve of the early GRI, together with the values observed in the early GRI intervals. An early GRI value of 1 corresponded to a probability of graft failure (5.07%), and each point of early GRI increase indicated that the probability of having a graft failure was multiplied by 2.06 (95% CI, 1.57-2.96).

In our comparison of indicators, no statistically significant differences in mean DRI results were seen between groups with and without graft survival. However, the GRI (global and early) results showed significant differences between both groups (Table 5).

Regarding early survival, risk group stratification of the DRI and ET-DRI values did not show significant differences in early graft survival between these indicators and global GRI, except for the group with extreme values. Only early GRI showed better survival discrimination, with an observation of a progressive decrease in survival in which an early GRI value ≥ 3 indicated a probability of graft survival at 1 month of 68.0% (95% CI, 46.2-82.5) and a significant increase in risk compared with the standard group (HR = 10.15, 95% CI, 3.91-26.32; P < .001) (Table 6). With regard to global survival, all indexes showed differences between extreme groups, with GRI having the best discrimination.

Our power of prediction calculation using receiver operating characteristic area under the curve showed values of 0.52 (95% CI, 0.43-0.62) and 0.54 (95% CI, 0.45-0.65) for DRI and ET-DRI, respectively, indicating that their capacity to predict graft survival was not better than chance. The receiver operating characteristic area under the curve analysis of global GRI was 0.68 (95% CI, 0.60-0.76) and that of early GRI was 0.69 (95% CI, 0.61-0.77) (Figure 2).

Discussion

Recently, benchmark¹⁶ indicators have been described regarding outcomes after liver transplant, including surgical outcomes, patient and graft survival, patient care in the ICU, complications, length of stay, and other factors.

Regarding length of stay in the ICU, Pedersen and associates,¹⁷ in a study of factors associated with a longer stay in the ICU, found that pretransplant MELD was the most potent predictor of prolonged ICU stay. Stratigopoulou and colleagues¹⁸ confirmed that MELD and transplant duration are independent predictors of prolonged ICU stay. However, Rana and associates¹⁹ described a new index to predict the length of posttransplant hospitalization; of 22 independent risk factors associated with longer hospital stay, the 2 most significant were shown to be previous admission of the recipient to the ICU (odds ratio of 1.75; 95% CI, 1.58-1.95) and previous transplant (odds ratio of 1.60; 95% CI, 1.47-1.75).

Regarding morbidity and mortality in the ICU, preventing complications is a key factor in the evolution of these patients and begins with appropriate donor-recipient matching. Traditional indicators used to assess the prognosis of cirrhosis (Child-Pugh) or mortality on the liver transplant wait list (MELD) are not valid to predict posttransplant patient survival. Recent indicators based on neural networks are complex, based on limited experience, have low generalization capacity, and are difficult to extrapolate to daily clinical practice. Finally, the external validations of some indicators (DRI, MELD, vs D-MELD; SOFT, DRI, vs ET-DRI; and DRI vs SOFT)20 could not be confirmed when applied in different scenarios from the one originally described, such as in our study. The latter is likely because of the large differences between countries and regions regarding donor population characteristics and the donation-transplant process and less to recipient characteristics.

With regard to indicators, donor-related variables may be associated with specific graft characteristics and general characteristics of the “donor per se.” Most described indicators share some common variables, such as age and cause of brain death, among others. However, some indicators are those whose use is not based on a solid scientific basis, such as height. Although steatosis is a main determinant of posttransplant liver function (especially in the first year), as described by Kulik and associates²¹ in a recent study, it is striking that it has not been included in any of the indicators described.

Donor age has been included in many indicators.²² However, in RETH, it is an independent risk factor regarding overall graft survival but not with respect to short-term graft survival (< 1 mo), which has been previously concurred by an international group²³ and by our group.^24,25

Risk factors include advanced age, and, traditionally, donor hypernatremia has been as-sociated with primary graft dysfunction or initial poor graft function.26 Subsequent studies have not confirmed these findings; however, donor hyper-natremia may be associated with marginal livers, but only alanine aminotransferase levels (> 65 IU/L) were associated with a higher incidence of primary graft dysfunction.²⁷

Variables associated with the donor per se should be included as indicators of graft survival. Some of these variables have been grouped internationally under the concept of expanded criteria donors. These include drug abuse, tumors, infections, hepatitis B/C, age > 65 years, and high transaminase/bilirubin levels.

Regarding recipient-related variables, we have already stated how the previous clinical condition of the recipient could affect length of stay in the ICU (these include high MELD score and United Network for Organ Sharing status). Allocating an organ to the most appropriate recipient is one of the most difficult decisions. We agree with Feng and associates5 that an ideal graft is homogeneous, allowing allocation of an optimal organ to anyone; however, nonideal grafts are heterogeneous, covering a wide spectrum of risk. In the latter case, allocating a suboptimal organ to a recipient with high MELD,^28,29 specific conditions,³⁰ or conditions with higher wait list mortality could interfere with the risk of patient survival against wait list mortality.³¹ Therefore, an indicator should include this aspect in its design.

With regard to process-related variables, most indicators have included cold ischemia time in their design. Cold ischemia time may depend in turn on donor location, available means of transportation, surgical variables, and other factors. We believe a good indicator should be known at the time of organ donation and should not include other added variables (eg, donor location and cold ischemia time).

Another much debated issue is the influence of the number of transplants performed at the center on morbidity and mortality rates. Muller and associates16 analyzed this effect, concluding that centers with a higher proportion of reference transplants had more biliary complications. However, in our RETH analyses, the number of transplants performed at the center did not constitute an independent risk factor for graft survival.

The present study has some limitations. Although the indicator described is based on a national registry of more than 22 000 transplants, it has been validated with a limited series of a single center. These indexes aim to predict and simplify complex situations depending on many factors and, in many cases, simplify quantitative variables (eg, time of cardiac arrest, cold ischemia time) to categorical ones.

In the future, decision-making should be based on complex artificial neural networks³² that will eventually replace medical doctors. However, until then, we will continue the search for tools that decrease errors in our donor-recipient matching decisions.

Conclusions

We found that neither DRI nor ET-DRI adequately predicted graft risks in our setting. Therefore, it is necessary to have an indicator to predict post-transplant graft survival that is adapted to the environment in which it will be used. A national GRI can be a useful tool to optimize donor-recipient matching. Therefore, a national study would allow delineation of indicators and a more extensive validation to be performed.

References:

1. Bates E, Martin D. Immediate postoperative management and complications on the intensive care unit. Br J Hosp Med (Lond). 2017;78(5):273-277.
   CrossRef - PubMed
   
2. Brown RS, Jr., Kumar KS, Russo MW, et al. Model for end-stage liver disease and Child-Turcotte-Pugh score as predictors of pretransplantation disease severity, posttransplantation outcome, and resource utilization in United Network for Organ Sharing status 2A patients. Liver Transpl. 2002;8(3):278-284.
   CrossRef - PubMed
   
3. Desai NM, Mange KC, Crawford MD, 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(1):99-106.
   CrossRef - PubMed
   
4. Ioannou GN. Development and validation of a model predicting graft survival after liver transplantation. Liver Transpl. 2006;12(11):1594-1606.
   CrossRef - PubMed
   
5. 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.
   CrossRef - PubMed
   
6. Rana A, Hardy MA, Halazun KJ, et al. Survival outcomes following liver transplantation (SOFT) score: a novel method to predict patient survival following liver transplantation. Am J Transplant. 2008;8(12):2537-2546.
   CrossRef - PubMed
   
7. 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.
   CrossRef - PubMed
   
8. Dutkowski P, Oberkofler CE, Slankamenac K, et al. Are there better guidelines for allocation in liver transplantation? A novel score targeting justice and utility in the model for end-stage liver disease era. Ann Surg. 2011;254(5):745-753; discussion 753.
   CrossRef - PubMed
   
9. Braat AE, Blok JJ, Putter H, et al. The Eurotransplant donor risk index in liver transplantation: ET-DRI. Am J Transplant. 2012;12(10):2789-2796.
   CrossRef - PubMed
   
10. Briceno J, Cruz-Ramirez M, Prieto M, et al. Use of artificial intelligence as an innovative donor-recipient matching model for liver transplantation: results from a multicenter Spanish study. J Hepatol. 2014;61(5):1020-1028.
    CrossRef - PubMed
    
11. Lau L, Kankanige Y, Rubinstein B, et al. Machine-learning algorithms predict graft failure after liver transplantation. Transplantation. 2017;101(4):e125-e132.
    CrossRef - PubMed
    
12. 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.
    CrossRef - PubMed
    
13. 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):15109.
    CrossRef - PubMed
    
14. Araiz Burdio JJ, Serrano Aullo MT, Garcia Gil A, et al. Graft survival after liver transplantation: an approach to a new Spanish risk index. Rev Esp Enferm Dig. 2018;110(12):782-793.
    CrossRef - PubMed
    
15. Spanish Liver Transplant Registry. http://www.sethepatico.org.
    
    
16. Muller X, Marcon F, Sapisochin G, et al. Defining benchmarks in liver transplantation: a multicenter outcome analysis determining best achievable results. Ann Surg. 2018;267(3):419-425.
    CrossRef - PubMed
    
17. Pedersen MR, Choi M, Brink JA, Seetharam AB. Pretransplant factors and associations with postoperative respiratory failure, ICU length of stay, and short-term survival after liver transplantation in a high MELD population. J Transplant. 2016;2016:6787854.
    CrossRef - PubMed
    
18. Stratigopoulou P, Paul A, Hoyer DP, Kykalos S, Saner FH, Sotiropoulos GC. High MELD score and extended operating time predict prolonged initial ICU stay after liver transplantation and influence the outcome. PLoS One. 2017;12(3):e0174173.
    CrossRef - PubMed
    
19. Rana A, Witte ED, Halazun KJ, et al. Liver transplant length of stay (LOS) index: A novel predictive score for hospital length of stay following liver transplantation. Clin Transplant. 2017;31(12).
    CrossRef - PubMed
    
20. Jadlowiec CC, Taner T. Liver transplantation: Current status and challenges. World J Gastroenterol. 2016;22(18):4438-4445.
    CrossRef - PubMed
    
21. Kulik U, Lehner F, Klempnauer J, Borlak J. Primary non-function is frequently associated with fatty liver allografts and high mortality after re-transplantation. Liver Int. 2017;37(8):1219-1228.
    CrossRef - PubMed
    
22. Ghinolfi D, Marti J, De Simone P, et al. Use of octogenarian donors for liver transplantation: a survival analysis. Am J Transplant. 2014;14(9):2062-2071.
    CrossRef - PubMed
    
23. Bertuzzo VR, Cescon M, Odaldi F, et al. Actual risk of using very aged donors for unselected liver transplant candidates: a European single-center experience in the MELD era. Ann Surg. 2017;265(2):388-396.
    CrossRef - PubMed
    
24. Franco CC, Martinez JM, Bellido CB, et al. Results of liver transplants from donors aged 70 plus: analysis of Andalusian transplant register. Transplant Proc. 2013;45(10):3647-3649.
    CrossRef - PubMed
    
25. Seller-Perez G, Herrera-Gutierrez ME, Lebron-Gallardo M, Moreno-Quintana J, Banderas-Bravo E, Quesada-Garcia G. [Liver transplantation: influence of donor-related factors]. Med Intensiva. 2008;32(8):378-384.
    CrossRef - PubMed
    
26. Gonzalez FX, Rimola A, Grande L, et al. Predictive factors of early postoperative graft function in human liver transplantation. Hepatology. 1994;20(3):565-573.
    CrossRef - PubMed
    
27. Sirivatanauksorn Y, Taweerutchana V, Limsrichamrern S, et al. Analysis of donor risk factors associated with graft outcomes in orthotopic liver transplantation. Transplant Proc. 2012;44(2):320-323.
    CrossRef - PubMed
    
28. Grat M, Wronka KM, Patkowski W, et al. Effects of donor age and cold ischemia on liver transplantation outcomes according to the severity of recipient status. Dig Dis Sci. 2016;61(2):626-635.
    CrossRef - PubMed
    
29. Schlegel A, Linecker M, Kron P, et al. Risk assessment in high- and low-MELD liver transplantation. Am J Transplant. 2017;17(4):1050-1063.
    CrossRef - PubMed
    
30. Schoening W, Helbig M, Buescher N, et al. Eurotransplant donor-risk-index and recipient factors: influence on long-term outcome after liver transplantation - A large single-center experience. Clin Transplant. 2016;30(5):508-517.
    CrossRef - PubMed
    
31. Araiz JJ, Serrano MT, Garcia-Gil FA, et al. Intention-to-treat survival analysis of hepatitis C virus/human immunodeficiency virus coinfected liver transplant: Is it the waiting list? Liver Transpl. 2016;22(9):1186-1196.
    CrossRef - PubMed
    
32. Ayllon MD, Ciria R, Cruz-Ramirez M, et al. Validation of artificial neural networks as a methodology for donor-recipient matching for liver transplantation. Liver Transpl. 2018;24(2):192-203.
    CrossRef - PubMed