Donor hemodynamic profile presages graft survival in donation after cardiac death liver transplantation

Authors


Abstract

Obligatory exposure to a period of warm ischemia is the defining feature of liver allografts from donation after cardiac death (DCD) donors. We explored novel methods for characterizing the dynamic aspects of donor warm ischemia that might be useful in assessing organ quality. The hemodynamic profile during donor warm ischemia was retrospectively studied for 110 Maastricht category III DCD donors. Three methods were used to summarize the hemodynamic changes after extubation: (1) the area under the systolic blood pressure curve (AUCSBP), (2) the slope of the systolic blood pressure regressed onto the time from extubation until cross-clamping, and (3) the slope of the systolic blood pressure regressed onto the time from extubation but calculated with only the values during the first 10 minutes after extubation (SBP10). Stepwise multivariate Cox models were created to study the association of these measures with graft survival. The duration of the donor warm ischemia time (23.6 ± 8.5 minutes) was not associated with graft survival (P = 0.35), although AUCSBP and SBP10 demonstrated significant associations (P = 0.02 and P = 0.05, respectively) in a univariate analysis. Multivariate regression models incorporating donor and recipient covariates indicated that among all covariates, SBP10 had the closest association with graft survival (hazard ratio = 1.08, P = 0.01). This association was even stronger when SBP10 was dichotomized into values above or below the median (−7.2 mm Hg/minute). Patients with SBP10s steeper than the median had an estimated 5-year graft survival rate of 76%, whereas patients with slopes less than the median had a 5-year survival rate of 45% (P < 0.007). In conclusion, the incorporation of novel methods for characterizing the donor warm ischemia time may help in selecting DCD liver allografts with favorable outcomes. Liver Transpl 20:165-172, 2014. © 2013 AASLD.

Abbreviations
AIC

Akaike information criterion

AUCSBP

area under the systolic blood pressure curve

DCD

donation after cardiac death

DNDD

donor with neurological determination of death

HR

hazard ratio

ICU

intensive care unit

MELD

Model for End-Stage Liver Disease

OPTN

Organ Procurement and Transplantation Network

ROC

receiver operating characteristic

SBP slope

slope of the systolic blood pressure regressed onto the time from extubation until cross-clamping

SBP10

slope of the systolic blood pressure regressed onto the time from extubation but calculated with only the values during the first 10 minutes after extubation

MEDSBP10

median value of SBP10

Donation after cardiac death (DCD) donors have been promulgated as a resource for abrogating the unresolved mortality rate among patients awaiting a liver transplant. Although the number of DCD donors and the number of kidney transplants from DCD donors in the United States have increased markedly within the last decade, the utilization of livers from DCD donors has plateaued, and this attests to continuing concerns about these organs.[1, 2] Several analyses using Organ Procurement and Transplantation Network (OPTN) data have demonstrated inferior survival among recipients of DCD liver allografts versus recipients of allografts from donors with neurological determination of death (DNDDs).[3-6] Although numerous individual center-specific analyses are congruent with national findings, some centers have reported allograft survival rates equivalent to those with transplants from DNDDs.[7-12]

Inferior liver allograft survival with DCD donors has been ascribed to the period of obligatory donor warm ischemia, which is the primary element differentiating DCD donors from DNDDs (Fig. 1). Recipients of DCD livers have increased rates of primary nonfunction, hepatic artery thrombosis, and biliary complications (particularly ischemic cholangiopathy); these factors likely underlie the graft survival statistics.[11-14] Previous publications have attempted to delineate the association between donor warm ischemia and posttransplant complications and graft failure. In an analysis of OPTN data, Lee et al.[5] identified a donor warm ischemia time greater than 15 minutes as a risk for graft loss [hazard ratio (HR) = 1.37, P = 0.03], and the risk increased beyond 30 minutes (HR = 1.78, P = 0.01); however, in a subsequent analysis of OPTN data by Mathur et al.,[15] an association with graft loss was not observed until the donor warm ischemia time exceeded 35 minutes (HR = 1.84, P = 0.003). Single-center analyses have provided conflicting findings about the impact of the duration of the donor warm ischemia time on graft outcomes.[8, 9, 16, 17] Likewise, the data regarding the association of the donor warm ischemia time with biliary complications remain contradictory.[8, 9, 11, 12]

Figure 1.

Timeline of the events during DCD organ procurement. The donor warm ischemia time is defined as the period from the planned withdrawal of care by extubation until aortic cross-clamping and flushing of the organs with a chilled preservation solution.

The failure of the collective literature to demonstrate an unequivocal association between warm ischemia and untoward events in DCD liver transplantation may be related to a variety of factors, including nonstandardized criteria for certifying cardiopulmonary death, variability in defining the duration of the donor warm ischemia time, inadequate statistical power, and the use of static measures to characterize donor warm ischemia. It has been recognized that the hemodynamic perturbations occurring after the withdrawal of care are dynamic and that a wide range of profiles are encountered among donors that are discernible by variations in organ perfusion and not solely in duration. Without consideration of the quality of the donor warm ischemia time, an assessment of quantity measured by time alone may fail to recognize the importance of donor warm ischemia to the outcome of a DCD allograft. We hypothesized that an investigation of the dynamic nature of the donor warm ischemia time might offer an opportunity for stratifying the risk of failure among DCD allografts. This is an exploratory analysis using 3 different techniques to measure donor hemodynamic profiles for a cohort of Maastricht category III donors.

PATIENTS AND METHODS

Study Population

This is a retrospective analysis of DCD liver transplants performed with organs from Maastricht category III donors recovered within the Gift of Life donor service area, which includes eastern Pennsylvania, Delaware, and southern New Jersey. Gift of Life prospectively maintains a database of all organ recoveries performed within the donor service area. The linkage of donor data to recipient outcomes was conducted through the United Network for Organ Sharing. Approval to perform the analysis was obtained from the institutional review board of the University of Pennsylvania.

Maastricht category III DCD liver donors and transplant recipients between June 1995 and September 2007 constituted the population for this study. Serial measurements of donor hemodynamic changes during the donor warm ischemia time were recorded by the procurement coordinator. All DCD organ recoveries were performed according to a predetermined algorithm. Extubation occurred in the operating room after authorization for organ donation had been documented. All donors received systemic heparin before extubation. The donor warm ischemia time was defined as the period from extubation until aortic cross-clamping. Donor death was determined by a physician not associated with the organ procurement organization or the surgical team recovering the organs. The criteria for determining death were at the discretion of the certifying physician. After the determination of death but before incisions, a 5-minute stand-down period was observed in accordance with recommendations from the Institute of Medicine.[18] The surgical technique for recovery included abdominal aortic cannulation and thoracic or supraceliac aortic cross-clamping. University of Wisconsin solution was used as a perfusate for all donors except for 5 donors who received histidine-tryptophan-ketoglutarate (Custodiol).

Analytical Methods

We constructed 3 different methods for characterizing the donor hemodynamic profile during the period of warm ischemia: (1) the area under the systolic blood pressure curve (AUCSBP), which was truncated at a maximum of 60 minutes; (2) the slope of the systolic blood pressure regressed onto the time from extubation until cross-clamping (SBP slope); and (3) the slope of the systolic blood pressure regressed onto the time from extubation but calculated with only the values during the first 10 minutes after extubation (SBP10). SBP10 was chosen to assess the impact of the initial trajectory of the systolic blood pressure after extubation. The analysis for exploring the relationship between the hemodynamic profile and graft survival then proceeded as follows:

  1. The specificity and sensitivity of each summary measure (ie, AUCSBP, SBP slope, and SBP10) for predicting graft survival were quantified with a receiver operating characteristic (ROC) analysis as detailed by Heagerty et al.[19]
  2. Cox proportional hazards models were fitted with the time to graft failure as an independent variable. The multivariate models were initially constructed with donor-related variables to explore the utility of the donor hemodynamic profile in the assessment of donor organ quality. The models included the following donor-related variables: age, final aspartate aminotransferase level, final alanine aminotransferase level, number of days at the hospital before organ donation, number of vasopressors, height, cold ischemia time, and time from extubation until cross-clamping. Subsequent models included the recipient-related covariates of age, diagnosis, and hospitalization status. Hospitalization status was used as a surrogate for medical urgency. The Model for End-Stage Liver Disease (MELD) score was not included because it was unavailable for a large number of patients undergoing transplantation before the adoption of MELD-based allocation. Additional recipient covariates were not included because of the potential for model overfitting. For each multivariate model, the Akaike information criterion (AIC), a measure of the relative quality of a statistical model for a given set of data, was calculated so that we could compare the multivariate models.[20]
  3. A visual inspection of residual scatter plots from the Cox regression model indicated that the most promising hemodynamic profile measure (SBP10) might be better represented if it were dichotomized. A new summary measure, median value of SBP10 (MEDSBP10), was created with 2 categories (ie, a subject's SBP10 was above or below the median for all subjects). The multivariate models were refitted with this summary statistic.
  4. Recognizing that there were collinearities present between the donor and recipient variables and MEDSBP10, we used a stepwise procedure to construct a parsimonious predictive model for graft survival. The stepwise procedure included covariates as long as all covariates in the model were significant at a level of 0.05. When no additional covariates could be included at this level of significance, the resulting model was judged to be final. Statistical analyses were conducted with SAS 9.1.

RESULTS

Donor Demographics

One hundred ten DCD liver allografts with hemodynamic data were recovered and transplanted from Maastricht category III donors (Table 1). The average age of the donors was 34.6 ± 15.0 years, and the length of the hospital stay before organ recovery was 5.8 ± 6.4 days. On the whole, the donors were thin with an average body mass index of 25.4 kg/m2, were hemodynamically stable (with only 39.1% requiring vasopressor support), and showed little evidence of hepatocellular injury at the time of recovery (with aspartate aminotransferase and alanine aminotransferase levels of 93.7 and 70.1 IU/mL, respectively). Most of the livers (67.3%) were transplanted within the Gift of Life donation service area, 21.8% were transplanted within United Network for Organ Sharing region 2, and 10.9% were transplanted nationally. The time from extubation until cross-clamping (the donor warm ischemia time) was 23.6 ± 8.5 minutes. The cold ischemia time was 8.7 ± 3.2 hours.

Table 1. Donor Characteristics
  1. a

    The data are presented as means and standard deviations.

Age (years)a34.6 ± 15.0
Sex: female (%)42.2
Race (%) 
White76.4
Black18.2
Other5.4
Body mass index (kg/m2)a25.4 ± 5.8
Length of hospitalization (days)a5.8 ± 6.4
Final aspartate aminotransferase level (IU/mL)a93.7 ± 110.8
Final alanine aminotransferase level (IU/mL)a70.1 ± 99.6
Cause of injury (%) 
Head trauma40
Cerebrovascular accident21.8
Anoxia31.8
Central nervous system tumor0.9
Other5.5
Number of vasopressors at time of withdrawal (%) 
060.9
120
214.5
34.5
Allocation (%) 
Local67.3
Regional21.8
National10.9
Cold ischemia time (hours)a8.7 ± 3.2
Extubation to death (minutes)a14.7 ± 8.0
Death to cross-clamping (minutes)a9.0 ± 3.1
Extubation to cross-clamping (minutes)a23.6 ± 8.5

Recipient Demographics

The average recipient was Caucasian and 51.8 ± 10.2 years old at the time of transplantation (Table 2). The most common primary etiology of end-stage liver disease was noncholestatic cirrhosis. Thirty percent of the patients were hospital inpatients before transplantation, and approximately half of these were located within the intensive care unit (ICU). For those patients undergoing transplantation since the introduction of the MELD score, the average score at transplant was 20.8 ± 8.5. The mean graft survival time was 4.5 ± 3.5 years, and 13.8% of the patients required retransplantation subsequently.

Table 2. Recipient Characteristics
  1. a

    The data are presented as means and standard deviations.

  2. b

    For patients undergoing transplantation since the initiation of the MELD system.

Age (years)a51.8 ± 10.2
Sex: female (%)42.7
MELD scoreab20.8 ± 8.5
Ethnicity (%) 
White79.1
Black10.9
Hispanic8.2
Asian0.9
Unknown0.9
Location at time of transplant (%) 
ICU14.6
Hospitalized, non-ICU15.4
Not hospitalized70
Diagnosis (%) 
Noncholestatic69.1
Cholestatic4.5
Acute hepatic necrosis3.6
Metabolic1.8
Primary hepatic malignancy10.0
Other10.9
Graft survival 
Currently functioning graft (%)47.4
Graft survival (years)a4.5 ± 3.5
Retransplantation (%)13.8

Donor Hemodynamics and Graft Survival

In a univariate analysis, the duration of the donor warm ischemia time was not associated with graft survival (HR = 0.99, P = 0.35); however, AUCSBP and SBP10 were associated with graft survival (P = 0.02 and P = 0.05, respectively; Table 3). In an ROC analysis, SBP10 had the strongest association with graft survival (C statistic = 0.76; Fig. 2).

Table 3. Summary Statistics for the 3 Measures of the Systolic Blood Pressure During the Donor Warm Ischemia Time
 AUCSBP (mm Hg × Minutes)SBP Slope (mm Hg/Minute)SBP10 (mm Hg/Minute)
  1. a

    Calculated in accordance with Heagerty et al.[19]

Median730−1.3−7.2
Mean ± standard deviation946 ± 873−1.5 ± 0.9−7.9 ± 7.4
Minimum to maximum0-5151−4.66 to 0−28.2 to 9.5
Univariate analysis   
HR1.000.761.04
P value0.020.060.05
95% CI1.00-1.000.57-1.011.00-1.08
C statistica0.650.490.76
Figure 2.

ROC plot for SBP10 as a predictor of 5-year graft survival.

In order to determine whether the putative relationship of AUCSBP and SBP10 with graft survival was due to confounding factors, they were investigated jointly with other potential explanatory variables in multivariate Cox regression analyses. Proportional hazards estimates were computed along with P values for significance first with donor covariates (in order to explore the utility of using the donor hemodynamic profile to assess donor quality) and subsequently with donor and recipient covariates (Table 4). In a model that contained only donor variables, AUCSBP was not associated with graft survival. In a separate model with donor covariates, SBP10 was found to have the most significant association (P = 0.02), and this was followed by the cold ischemia time (P = 0.03).

Table 4. Cox Regression Proportional Hazards Estimates and P Values for the Association Between AUCSBP, SBP10, and MEDSBP10 and Graft Failure
VariableMultivariate Analysis of Donor Covariates
AUCSBPSBP10
HRP ValueHRP Value
Hemodynamic measurement1.000.101.070.02
Cold ischemia time (hours)1.130.021.120.03
Extubation to cross-clamping (minutes)0.990.901.000.84
Donor height (meters)0.330.270.380.35
Number of vasopressors1.270.251.340.15
Length of donor hospital stay (days)1.020.541.060.19
Donor age (years)1.010.331.010.24
Donor body mass index (kg/m2)1.020.581.050.17
Final aspartate aminotransferase level (IU/mL)0.990.590.990.76
Final alanine aminotransferase level (IU/mL)1.000.231.000.63
VariableMultivariate Analysis of Donor and Recipient Characteristics
AUCSBPSBP10MEDSBP10
HRP ValueHRP ValueHRP Value
Hemodynamic measurement1.000.041.080.013.580.006
Cold ischemia time (hours)1.130.041.120.041.090.15
Extubation to cross-clamping (minutes)0.990.711.010.791.010.77
Donor height (meters)0.150.120.210.200.200.19
Number of vasopressors1.290.241.370.141.160.48
Length of donor hospital stay (days)1.030.551.060.171.050.24
Donor age (years)1.000.801.010.521.000.74
Donor body mass index (kg/m2)1.010.851.050.151.040.23
Final aspartate aminotransferase level (IU/mL)0.990.400.990.561.000.75
Final alanine aminotransferase level (IU/mL)1.000.311.000.731.000.93
Recipient diagnosis      
Cholestatic cirrhosis (reference)
Noncholestatic cirrhosis1.300.641.140.851.250.74
Acute hepatic necrosis2.840.432.730.472.540.51
Other1.890.421.270.791.050.95
Metabolic1.110.930.680.761.280.84
Hepatic neoplasm4.550.073.820.133.900.12
Recipient age (years)1.050.0041.040.021.040.06
Recipient location at time of transplant      
ICU (reference)
Hospitalized, non-ICU1.240.781.620.501.870.40
Not hospitalized1.680.341.710.311.790.26

We then proceeded to study the relationship of AUCSBP and SBP10 with graft survival in 2 separate multivariate Cox regression models that included both donor and recipient covariates. Proportional hazards estimates were computed along with P values for significance (Table 4). In the model using AUCSBP, the following variables were found to be associated with graft survival: AUCSBP (P = 0.04), cold ischemia time (P = 0.04), and recipient age (P = 0.004). In the model using SBP10, the following variables were associated with P values less than 0.05: SBP10 (P = 0.01), cold ischemia time (P = 0.04), and recipient age (P = 0.02). According to the AIC, the model containing SBP10 with a score of 338.7 was preferable to the one containing AUCSBP with a score of 350.4.

To assess whether the relationship between SBP10 and graft survival was linear, we constructed a deviance Martingale residual plot. Deviance residual plots are a customary way of assessing the goodness of fit for lifetime regression models.[21] The residual plot was augmented with an adaptive scatter plot smoother (SAS PROC LOESS) in order to capture the functional relationship between SBP10 and graft survival. As can be seen in Fig. 3, the trend was not linear but appeared to be better modeled by a step function in SBP10 with different levels depending on whether the SBP10 value was above or below the median value. We hence proceeded to dichotomize SBP10 into an indicator variable (MEDSBP10) above or below the median (−7.2 mm Hg/minute), and we reran the multivariate analysis. Table 4 presents the results of the Cox regression analysis including MEDSBP10 and other potential explanatory variables. MEDSBP10 was the most significant variable (HR = 3.58, P = 0.006). The AIC for the model declined to 337.5, and this indicated that MEDSBP10 provided a slightly better fit for the data than SBP10 itself.

Figure 3.

Deviance residual plot for graft failure times plotted against SBP10. The bump after the median SBP10 value (−7.2 mm Hg/minute) indicates that the dependence on SBP10 is not linear but is better described by 2 levels.

Finally, recognizing that the multivariate model had to contain collinearities between the covariates, we reduced the number of parameters through a stepwise variable selection procedure (SAS PROC PHREG). The multivariate analyses, including donor and recipient variables, was rerun and again included the systolic blood pressure as AUCSBP, SBP10, or MEDSBP10. The results are summarized in Table 5. In the final model with AUCSBP, AUCSBP (P = 0.01), recipient age (P = 0.02), and diagnosis of hepatic malignancy (P = 0.002) were included. Similarly, with SBP10 in the model, SBP10 (P = 0.02), recipient age (P = 0.02), and hepatic malignancy (P = 0.009) were all included. In the model using MEDSBP10, only MEDSBP10 itself was found to be significant (HR = 2.67, P = 0.003); no other covariates yielded additional explanatory power. The AIC for the model with MEDSBP10 was the smallest of the 3 values (319.3), although the AIC for the model with SBP10 was close (AIC = 319.5). Both were clearly better descriptions of the data than the model with AUCSBP (AIC = 330.9).

Table 5. Stepwise Regression Models for AUCSBP, SBP10, and MEDSBP10
 HRP Value
AUCSBP1.000.01
Recipient age (years)3.180.02
Hepatic malignancy1.050.002
SBP101.060.02
Recipient age (years)3.500.02
Hepatic malignancy1.040.009
MEDSBP102.670.003

To illustrate the predictive power of MEDSBP10, Kaplan-Meier curves of the time to graft failure by the SBP10 status (ie, above or below the median of −7.2 mm Hg/minute) are presented in Fig. 4. Patients who received allografts from donors with steeper slopes (<−7.2 mm Hg/minute) had an estimated 5-year graft survival rate of 76%; patients with allografts from donors with less steep slopes (>−7.2 mm Hg/minute) had a 5-year survival rate of 45% (P = 0.007).

Figure 4.

Kaplan-Meier curves of the time to graft failure by the SBP10 status (ie, above or below the median of −7.2 mm Hg/minute). Patients who received allografts from donors with steeper slopes (<−7.2 mm Hg/minute) had an estimated 5-year graft survival rate of 76%. For patients with less steep donor slope values (>−7.2 mm Hg/minute), the 5-year survival rate was 45% (P < 0.007).

DISCUSSION

The hemodynamic changes that occur during DCD donor warm ischemia may offer opportunities for assessing organ quality before transplantation. This study has explored several methods for characterizing donor warm ischemia that consider the dynamic nature of organ perfusion after extubation because the duration alone is unlikely to accurately reflect the impact of warm ischemia on donor organs.

In the present analysis, we found that the duration of warm ischemia was not associated with graft survival; however, measures providing a dynamic characterization of donor warm ischemia (AUCSBP and SBP10) were related to graft outcomes in stepwise multivariate analyses. The relationship between SBP10 and graft survival appeared to be nonlinear, and this suggested that the median value of SBP10 could designate grafts with inferior outcomes. These findings require substantiation in an additional cohort, but they can serve as a starting point for considering a refined approach to understanding donor warm ischemia as it relates to transplant outcomes and potentially organ selection from DCD donors.

Previous analyses using static measures of the warm ischemia time have demonstrated a variable association between warm ischemia and graft survival. OPTN data indicate that an increasing duration of warm ischemia is detrimental to the graft. An initial analysis by Lee et al.[5] found that a warm ischemia time greater than 15 minutes was associated with an increased risk of graft failure; however, a subsequent analysis with a larger cohort by Mathur et al.[15] demonstrated that the risk of graft failure did not increase until 35 minutes. Single-center reports have provided conflicting results. de Vera et al.[9] described the experience at the University of Pittsburgh and noted inferior outcomes when the warm ischemia time exceeded 20 minutes. Ho et al.[17] were unable to find an association between the donor warm ischemia time and an inferior outcome, but they demonstrated that the duration of a systolic blood pressure < 50 mm Hg to cold flushing predicted poor graft survival as defined by a composite endpoint of death, graft loss, and biliary strictures within 1 year of transplantation. Taner et al.,[8] on the other hand, did not find an association between the warm ischemia time and graft loss. Likewise, the data regarding the association of the donor warm ischemia time with biliary complications remain contradictory[7, 8, 11, 12, 16]

The 3 methods presented in this article reflect an attempt to characterize different aspects of the dynamic nature of the donor warm ischemia time (Fig. 1). When considering these methods, one should recognize that the length of donor warm ischemia is related not only to the intrinsic agonal phase distinguished by the demise and cessation of cardiopulmonary activity but also to the variability within the process of DCD organ recovery, which is not in and of itself a reflection of the donor physiology. One source of variability is incurred through the utilization of diverse criteria for cardiopulmonary death, which are related to physician and/or hospital-based practices. Certification of death for DCD donors is commonly established with one of several criteria, including auscultation of heart tones, a pulse pressure detected manually or via an arterial line, and cessation of cardiac electrical activity. The total warm ischemia time is a direct reflection of the criteria chosen by an independent physician and cannot be dictated by the organ procurement organization or the recovery surgeon. A second source of variability, though limited, is the mandatory stand-down time for monitoring a potential donor for the return of spontaneous cardiopulmonary activity. The 5-minute respite was proposed by the Institute of Medicine, but other organizations have supported a 2-minute interlude.[18, 22] A third contributor to the donor ischemia time that is not reflective of the donor physiology is the time needed by the recovering team to make an incision, cannulate the aorta, and cross-clamp; this variable is associated with the development of biliary complications.[8] As for the measures studied in this article, AUCSBP captures the entire donor exposure to warm ischemia, and the SBP slope reflects the entire trajectory of the systolic blood pressure from the withdrawal of care until cross-clamping. Thus, AUCSBP and the SBP slope reflect the physiology during the agonal phase and also directly capture contributors to the warm ischemia time that are dictated by external sources. SBP10 captures the early trajectory of the systolic blood pressure immediately after the withdrawal of care. In most donors, SBP10 will reflect only the agonal phase, and it is not influenced by the method for determining death, the definition of warm ischemia time, or the technical aspects of the surgical team that may influence the time to aortic cross-clamping.

We speculate that unlike the association between the donor warm ischemia time and graft survival, the association between SBP10 and graft survival is significant because SBP10 subselects for donors with a favorable blood pressure trajectory during the agonal phase. The exposure of the donor organ to ischemic damage is influenced not only by the length of ischemia but also by the perturbations of perfusion that occur during the agonal phase. The donor warm ischemia time, even if it is relatively brief, does not provide any information about the blood pressure trajectory during the agonal phase; it just provides information about the duration. Among donors, there is heterogeneity not only in the length of the warm ischemia time but also in the patterns of perfusion during the agonal period. SBP10 may be effective in identifying organs from donors with more favorable survival characteristics through the exclusion of some of the donor heterogeneity that is incurred when the donor warm ischemia time is used to solely assess donor quality.

The current study was designed as an exploratory analysis to consider several new methods for evaluating the contribution of DCD donor warm ischemia to liver allograft outcomes. There are several limitations inherent to this analysis, including the small sample size, the nonuniform criteria for establishing donor death, and a pretransplant selection bias by the transplant surgeons to opt for organs felt to have superior survival characteristics. The study is also limited by an inability to determine the association of the SBP slope, AUCSBP, and SBP10 with the development of biliary complications or recipient causes of death.

In summary, this report provides evidence that a more complete characterization of the donor warm ischemia time may offer an opportunity for informing the selection of organs from DCD liver donors. The validation of this study's findings in a separate cohort, the inclusion of additional recipient covariates in future models, and an effort to identify an association between these findings and ischemic-type biliary strictures are a logical evolution for future inquiry.

Ancillary