LIVER TRANSPLANTATION 20:823–830, 2014
ORIGINAL ARTICLE
Perioperative Complications in Liver Transplantation Using Donation After Cardiac Death Grafts: A Propensity-Matched Study Xiongxiong Pan,1,3 Worapot Apinyachon,1 Wei Xia,1 Johnny C. Hong,4 Ronald W. Busuttil,2 Randolph H. Steadman,1 and Victor W. Xia1 Departments of 1Anesthesiology and 2Surgery, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA; 3Department of Anesthesiology, First Affiliated Hospital of Nanjing Medical University, Nanjing, People’s Republic of China; and 4Department of Surgery, Medical College of Wisconsin, Milwaukee, WI
Donation after cardiac death (DCD) is an important source for expanding the donor pool for liver transplantation (LT). Although the long-term outcomes of LT using DCD grafts have been extensively studied, perioperative complications related to DCD grafts are rarely reported. The aim of this study was to determine whether DCD grafts were associated with a higher incidence of postreperfusion complications and worse outcomes in adult LT patients. After institutional review board approval, the medical records of all adult patients who underwent LT at our medical center between 2004 and 2011 were reviewed. Postreperfusion complications and posttransplant outcomes were compared between patients receiving DCD grafts and patients receiving donation after brain death (DBD) grafts. In all, 74 patients received DCD grafts during the study period, and 1369 patients received DBD grafts. An initial comparison showed that many preoperative, prereperfusion, and donor variables in the DCD group differed significantly from those in the DBD group. Propensity matching was chosen so that adjustments could be made for the differences. A postmatching analysis showed that the preoperative, prereperfusion, and donor variables no longer differed between the 2 groups. The postreperfusion requirements for blood products and vasopressors, the posttransplant ventilation times, the incidence of posttransplant acute renal injury, and the 30-day and 1-year patient and graft survival rates were comparable between the 2 groups. However, patients receiving DCD grafts experienced significantly higher rates of hyperkalemia (33.8% versus 18.9%, P < 0.05) and postreperfusion syndrome (PRS; 25.7% versus 12.3%, P < 0.05). In conclusion, after adjustments for preoperative and prereperfusion risks via propensity matching, DCD grafts remained a risk factor for postreperfusion hyperkalemia and PRS. A prophylactic regimen aimed at decreasing postreperfusion hyperkalemia and PRS is recommended for the management of LT using DCD grafts. Liver C 2014 AASLD. Transpl 20:823-830, 2014. V Received December 16, 2013; accepted March 27, 2014.
The demand for organs for liver transplantation (LT) continues to exceed the supply. The discrepancy between the number of available organs and the number of patients waiting for transplantation remains significant.1 To address this urgent shortage and the rising wait-list mortality rate, the expansion of the
donor pool through the use of donation after cardiac death (DCD) grafts has been promoted.2 Although donation after brain death (DBD) remains the predominant source of deceased donor grafts, the use of DCD grafts in LT is being increasingly accepted worldwide. Now LT using DCD grafts accounts for approximately 5% of all
Abbreviations: ARI, acute renal injury; BMI, body mass index; CIT, cold ischemia time; DBD, donation after brain death; DCD, donation after cardiac death; DWIT, donor warm ischemia time; FFP, fresh frozen plasma; GWIT, graft warm ischemia time; ICU, intensive care unit; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; PRS, postreperfusion syndrome; RBC, red blood cell; UCLA, University of California Los Angeles. Address reprint requests to Victor W. Xia, M.D., Department of Anesthesiology, David Geffen School of Medicine, University of California Los Angeles, 757 Westwood Plaza, Suite 3325, Los Angeles, CA 90095. Telephone: 310-825-3664; FAX: 310-794-2141; E-mail: [email protected] DOI 10.1002/lt.23888 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
824 PAN ET AL.
LT procedures in the United States and as many as 50% of all LT procedures in other countries.1,3,4 The procurement of DCD grafts follows a specific procedure that makes DCD grafts different in many ways from DBD grafts.5 After the withdrawal of life support, DCD grafts often have a period during which organ perfusion is poor. The warm ischemia occurring during this period is unique to DCD grafts and may be responsible for exacerbated ischemia/reperfusion injury and posttransplant graft dysfunction.5 Several posttransplant complications, including ischemic cholangitis, nonanastomotic biliary strictures, primary nonfunction, acute renal injury (ARI), and mortality, have been shown to be associated with DCD grafts.6-11 Because of these factors, DCD livers are considered high-risk grafts for LT.12 Although the long-term posttransplant complications of DCD grafts have been extensively studied, the perioperative complications of DCD grafts have not been thoroughly investigated. In this retrospective study, we investigated whether DCD grafts were associated with a higher incidence of perioperative complications in patients undergoing LT. The perioperative complications in this study included complications during the postreperfusion period and adverse outcomes during the immediate posttransplant period.
PATIENTS AND METHODS The institutional review board of the University of California Los Angeles (UCLA) approved this study. Adult patients (18 years) who underwent LT at the UCLA Medical Center between 2004 and 2011 were identified and included in the study. Perioperative data related to LT were prospectively collected and stored in the UCLA transplant database and were retrieved for this study. Donor data and immediate postoperative outcomes were retrospectively collected for this study. DCD donors and recipients were selected according to our center’s protocol as previously described.5,13 The DCD donors were typically younger than 45 years with a body mass index (BMI) less than 30 kg/m2, with hospitalization for 5 days or less, and with serum transaminase levels at procurement less than twice the normal values. DCD grafts should have excellent parenchymal quality during the intraprocurement assessment and a projected cold ischemia time (CIT) less than 8 hours. The DCD procurement followed the Institute of Medicine guidelines and the UCLA procurement protocol.14 All potential DCD donors were controlled through the planned withdrawal of ventilator and organ-perfusion support. Artificial life support was withdrawn in the operating room or intensive care unit (ICU). An independent physician from the donor hospital was assigned to provide end-of-life care, withdraw life support, and declare death. After a 5-minute mandatory waiting period after asystole, procurement started. Livers were stored at 4 C for transport after they had been flushed with 4 to 6 L of cold University of Wisconsin solution via both the
LIVER TRANSPLANTATION, July 2014
abdominal aorta and the inferior mesenteric vein. The DBD procurement was performed according to the standard technique after brain death was declared, and the standard protocol was initiated. Livers from DBD donors were flushed via the aortic and portal system with University of Wisconsin solution and were preserved in a cold cooler until use. The donor warm ischemia time (DWIT) started when life support was withdrawn and ended when cold perfusion started. DWIT occurred only for DCD grafts. The graft warm ischemia time (GWIT) started with the live graft being taken from the ice box and ended with reperfusion. GWIT occurred for both DBD and DCD grafts. The anesthetic and surgical management of the LT recipients followed the protocol of the UCLA team. The anesthetic and surgical management of patients receiving DCD grafts did not differ from that for patients receiving DBD grafts. At our center, we accepted patients who were intubated and on vasopressors before transplant surgery for either DCD or DBD grafts. The anesthetic techniques typically consisted of intravenous induction and combined inhalational and intravenous maintenance. In addition to American Society of Anesthesiologists standard monitors, invasive monitors included intra-arterial and intrapulmonary arterial catheters. Intraoperative transesophageal echocardiography was at the discretion of the anesthesiologist. Stored red blood cells (RBCs) were administered via a rapid transfusion device to maintain a hematocrit of approximately 30%. The conventional bicaval technique was usually used in adult patients. Intraoperative venovenous bypass was selectively used and was determined by the attending surgeon and the anesthesiologist. Indications for venovenous bypass usually included severe portal hypertension, hemodynamic instability, and cardiopulmonary comorbidities that might make patients unable to tolerate cross-clamping of the vena cava or reperfusion. Lactated Ringer’s solution with albumin was used to flush out University of Wisconsin solution before reperfusion of the liver graft. The liver graft was reperfused first via portal flow, which was followed by hepatic artery anastomosis and reperfusion. Blood gas and electrolytes were routinely checked every hour. Hyperkalemia was treated with intravenous insulin, hyperventilation, urine output augmentation with furosemide, alkalizing agents (trishydroxymethylaminomethane solution or sodium bicarbonate), calcium chloride, and intraoperative hemodialysis. The intraoperative course was divided into 2 periods separated by reperfusion of the liver graft: the prereperfusion and postreperfusion periods. Prereperfusion factors were considered potential risk factors for events or complications occurring in the postreperfusion period or posttransplant period. The following perioperative complications were investigated in this study. Postreperfusion complications included increased requirements for blood products and vasopressors, hyperkalemia (defined as a serum K1 concentration 5.5 mmol/L), intraoperative cardiac events requiring cardiopulmonary resuscitation, and
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
PAN ET AL. 825
TABLE 1. Select Preoperative Variables Before Propensity Matching
Age (years)* Sex: male (%) Weight (kg)* BMI (kg/m2)* MELD score* Diabetes mellitus (%) Encephalopathy (%) Etiology (%) Hepatitis C Hepatitis B Hepatocellular carcinoma Acute liver failure Nonalcoholic steatotic hepatitis Baseline laboratory values* Potassium (mmol/L) Creatinine (mg/dL) International normalized ratio Hematocrit (%) Preoperative intubation (%) Preoperative vasopressor (%)
DBD (n 5 1369)
DCD (n 5 74)
P Value
53.6 6 10.9 65.2 79.6 6 22.3 28.5 6 10.8 30.9 6 8.2 25.5 43.3
56.2 6 10.0 60.8 80.0 6 18.8 27.6 6 5.0 27.9 6 7.9 28.4 31.1
0.048 0.441 0.634 0.566 0.002 0.587 0.039
42.5 7.1 32.5 6.0 5.4
37.8 6.8 48.6 2.7 5.4
0.434 0.901 0.004 0.236 0.993
3.9 6 0.6 1.9 6 2.0 1.8 6 1.3 29.4 6 5.8 23.3 13.3
3.9 6 0.7 1.6 6 1.2 1.6 6 0.4 28.7 6 6.5 8.1 2.7
0.789 0.106 0.083 0.270 0.002 0.008
*The data are presented as means and standard deviations. The list of etiology is not complete and one patient may have one or more etiologies.
postreperfusion syndrome (PRS; defined as a 30% decrease in blood pressure from the baseline lasting more than 1 minute within the first 5 minutes after reperfusion of the graft). Postoperative outcomes included ARI within the first week after LT, ventilator duration, and graft and patient survival. Prolonged postoperative ventilation was defined as the need for mechanical ventilation for more than 3 days after LT. The Risk, Injury, Failure, Loss, and End-Stage Kidney Disease criteria were used to define postoperative ARI. ARI was defined as a 2-fold or greater increase in the serum creatinine level in the first week after LT in comparison with the preoperative levels. ARI was analyzed after the exclusion of patients with pretransplant renal failure.15 Data are expressed as means and standard deviations for continuous variables and as proportions for categorical variables. Univariate analyses were performed with the Student t test or Mann-Whitney U test for continuous variables and with the chi-square test for categorical variables. Significant variables from the univariate analyses were selected for multivariate analyses using logistic regression. The propensity-matching method was used to control confounding factors and selection bias between the DCD and DBD groups. Variables that had the potential to influence the use of DCD grafts and outcomes were selected with the multivariate analyses. These selected variables were then used in a logistic regression to generate a propensity score for each patient. The 2 groups were matched with the nearest propensity scores. Postreperfusion complications and posttransplant outcomes were compared for the 2 postmatched groups. One-year graft
and patient survival rates were also estimated with Kaplan-Meier survival curves. All analyses were performed with Statistical Package for Social Science 21.0 for Windows (IBM Corp., Armonk, NY).
RESULTS Between 2004 and 2011, adult LT was performed 1443 times at our center. Seventy-four of these patients (5.5%) received DCD grafts, and 1369 received DBD grafts. The mean age of the patients in the DCD group was 56.2 6 10.0 years (range 5 19-74 years), and the majority of these patients were male (60.8%). The mean Model for End-Stage Liver Disease (MELD) score was 27.9 6 7.9, and approximately half of the patients (48.6%) had a pretransplant diagnosis of hepatocellular carcinoma. The mean DWIT was 27.3 6 10.6 minutes (range 5 12-76 minutes). The recipient and donor variables for the DBD and DCD groups were compared, and the results for select variables are shown in Tables 1 and 2. In comparison with the DBD group, many variables in the DCD group were significantly different. Recipients in the DCD group were older, had a higher frequency of hepatocellular carcinoma and lower MELD scores, and required less preoperative endotracheal intubation and vasopressor use. Intraoperatively, the DCD group received significantly less RBCs and fresh frozen plasma (FFP) before reperfusion of the liver graft in comparison with the DBD group. Because many preoperative, intraoperative, recipient, and donor variables were significantly different between the DCD and DBD groups, propensity
826 PAN ET AL.
LIVER TRANSPLANTATION, July 2014
TABLE 2. Select Intraoperative/Prereperfusion and Donor Variables Before Propensity Matching
Liver-kidney transplantation [n (%)] Venovenous bypass [n (%)] Piggyback technique [n (%)] Prereperfusion RBCs (U)* Prereperfusion FFP (U)* Prereperfusion hyperkalemia [n (%)]† Surgical time (minutes)* Donors Age (years)* Sex: male [n (%)] Weight (kg)* Height (cm)* BMI (kg/m2)* Macrosteatosis 20% [n (%)] ICU before procurement (days)* Graft CIT (minutes)* GWIT (minutes)*
DBD (n 5 1369)
DCD (n 5 74)
P Value
110 (8.0) 497 (36.3) 152 (11.1) 10.0 6 8.5 13.0 6 12.3 181 (13.2) 430.5 6 276.5
3 (4.1) 22 (29.7) 6 (8.1) 7.6 6 6.5 9.5 6 7.6 12 (16.2) 379.5 6 214.4
0.222 0.250 0.613 0.004 0.017 0.465 0.058
39.8 6 16.9 850 (62.1) 76.7 6 19.0 170.2 6 11.3 26.2 6 5.5 126 (9.2) 4.3 6 3.0 371.9 6 175.5 40.9 6 9.9
37.9 6 13.8 52 (70.3) 83.5 6 27.1 172.7 6 10.0 26.7 6 6.5 8 (10.8) 5.6 6 3.1 358.5 6 156.9 40.2 6 9.8
0.350 0.261 0.016 0.147 0.538 0.479 0.004 0.526 0.556
*The data are presented as means and standard deviations. † Defined as a serum potassium > 5.5 mmol/L should be 5.5 mmol/L.
matching was selected to make adjustments for the differences. In order to properly select factors for the propensity score generation, the preoperative, recipient, donor, and prereperfusion variables were analyzed with univariate tests and then with multivariate logistic regression against each studied complication and outcome (detailed data are not shown). A total of 16 preoperative and prereperfusion variables were identified as significant risk factors for postreperfusion complications and adverse posttransplant outcomes. They included the following: age, MELD score, sex, body weight, history of diabetes, cirrhosis caused by hepatitis C, baseline K1 concentration, baseline creatinine concentration, preoperative intubation, preoperative pressor use, prereperfusion RBCs, prereperfusion FFP, prereperfusion K1 concentration, CIT, GWIT, and surgical time. All 16 factors had odds ratios between >1 and 26 with P values < 0.05. Each perioperative complication had 2 to 6 risk factors, and each variable was confirmed to be a risk factor for 1 to 4 perioperative complications. All 16 variables were included in a logistic regression model to generate a propensity score for each patient. The DCD and DBD groups were matched in a 1:2 ratio with the nearest propensity scores. Table 3 shows the results of comparisons of 20 variables for the DCD and DBD groups after propensity matching. The propensity matching eliminated all significant differences between the 2 groups that existed before the matching. In addition, many nonsignificant differences between the 2 groups became smaller after the matching in comparison with the differences based on prematching data. After the propensity matching, postreperfusion complications and posttransplant outcomes were com-
pared between the DCD and DBD groups (Tables 4 and 5). The requirements for postreperfusion RBCs and FFP were similar for the DCD and DBD groups. The use of continuous vasopressor infusion immediately after reperfusion and at the end of surgery in the DCD group was not significantly different from that in the DBD group. Posttransplant ventilation requirements and the incidences of posttransplant ARI were also similar between the 2 groups. The 1year recipient and graft survival rates for the DCD group were comparable to those for the DBD group. However, patients receiving DCD grafts had a significantly higher incidence of hyperkalemia in the postreperfusion period in comparison with the DBD group (Table 4). The mean K1 concentrations at multiple time points were significantly higher for the DCD group versus the DBD group (4.6 6 0.6 versus 4.2 6 0.8 mmol/L, 4.4 6 0.6 versus 4.1 6 0.7 mmol/L, and 4.4 6 0.5 versus 4.0 6 0.5 mmol/L 2 hours after reperfusion, 3 hours after reperfusion, and at the end of surgery, respectively; all P < 0.05). The higher incidence of hyperkalemia in the DCD group did not result from less use of insulin because the administration of intravenous insulin either in boluses or as continuous infusions was similar for the 2 groups. In addition, patients receiving DCD grafts had a significantly higher incidence of PRS in comparison with those receiving DBD grafts. Six patients experienced a cardiac event requiring cardiopulmonary resuscitation immediately after reperfusion of the liver graft: 4 (5.4%) in the DCD group and 2 (1.4%) in the DBD group (P 5 0.272). The postreperfusion cardiac events included severe hypotension, ventricular tachycardia, and asystole. Prereperfusion hyperkalemia was a risk factor for severe cardiac events after reperfusion (odds
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
PAN ET AL. 827
TABLE 3. Comparison After Propensity Matching Variables
DBD (n 5 148)
DCD (n 5 74)
P Value
Age (years)* Sex: male [n (%)] Weight (kg)* MELD score* Baseline hematocrit (%)* Baseline K1 (mmol/L)* Baseline creatinine (mg/dL)* Baseline international normalized ratio* Hepatocellular carcinoma [n (%)] Encephalopathy [n (%)] Hepatitis C cirrhosis [n (%)] Diabetes mellitus [n (%)] Preoperative intubation [n (%)] Preoperative vasopressor [n (%)] Prereperfusion hyperkalemia [n (%)] Prereperfusion RBCs (U)* Prereperfusion FFP (U)* GWIT (minutes)* CIT (minutes)* Surgical time (minutes)*
56.2 6 9.6 89 (60.1) 80 6 19.9 28.5 6 7.4 28.8 6 7.1 3.9 6 0.6 1.7 6 1.4 1.6 6 0.5 70 (47.3) 48 (32.4) 58 (39.2) 42 (28.4) 14 (9.5) 10 (6.8) 20 (13.5) 7.2 6 5.5 8.9 6 6.5 40.1 6 10.0 359.6 6 173.4 360.7 6 230.7
56.2 6 10.0 45 (60.8) 80.8 6 18.8 28.2 6 7.9 28.7 6 6.5 3.9 6 0.7 1.6 6 1.2 1.6 6 0.4 36 (48.6) 23 (31.1) 28 (37.8) 21 (28.4) 6 (8.1) 2 (2.7) 12 (16.2) 7.6 6 6.5 9.5 6 7.6 40.2 6 9.8 358.6 6 156.9 379.5 6 214.4
0.981 0.923 0.764 0.802 0.907 0.553 0.604 0.855 0.849 0.839 0.846 1.000 0.740 0.218 0.589 0.588 0.510 0.951 0.968 0.557
*The data are presented as means and standard deviations.
ratio 5 8.520, 95% confidence interval 5 2.78-26.06, P < 0.001) in the multivariate logistic analysis.
DISCUSSION In this retrospective study, we found that LT patients receiving DCD grafts experienced hyperkalemic episodes and PRS after reperfusion of the liver graft approximately twice as often as those receiving DBD grafts. In addition, there was a trend that DCD grafts were associated with a higher incidence of severe cardiac events immediately after reperfusion, although the difference was not statistically significant. It is likely that our study was underpowered for detecting such relatively rare postreperfusion cardiac incidents. In contrast to previous studies, our study showed that postreperfusion complications (massive blood
transfusion and vasopressor requirements) and posttransplant outcomes (ARI, ventilation time, and graft and patient survival) were not significantly different between patients receiving DCD grafts and patients receiving DBD grafts after adjustments of the risks with the propensity score. Although DCD has been promoted as an important source for expanding the donor pool, the use of such grafts has been surrounded by controversies. In addition to ethical issues and procurement procedures, the outcomes of LT using DCD grafts have been continuously debated in the transplant community.16,17 Early studies using national registry data showed 30% lower graft survival for LT using DCD grafts versus LT using DBD grafts.6,16-18 Therefore, DCD livers are considered high-risk grafts.7,12 Later, several single-center studies reported comparable
TABLE 4. Comparison of Postreperfusion Complications After Propensity Matching DBD (n 5 148)
DCD (n 5 74)
P Value
6.3 6 7.4 29 (19.6) 7.3 6 8.0 2 (1.4) 75 (50.7)
7.6 6 8.4 17 (23.0) 8.7 6 8.3 4 (5.4) 41 (55.4)
0.213 0.558 0.231 0.272 0.560
37 (25.0) 28 (18.9) 18 (12.2)
20 (27.0) 25 (33.8) 19 (25.7)
0.697 0.014 0.019
Postreperfusion RBCs (U)* >10 U of RBCs [n (%)] Postreperfusion FFP (U)* Postreperfusion cardiac events [n (%)] Pressor use immediately after reperfusion [n (%)] Pressor use at end of surgery [n (%)] Postreperfusion hyperkalemia [n (%)] PRS [n (%)] *The data are presented as means and standard deviations.
828 PAN ET AL.
LIVER TRANSPLANTATION, July 2014
TABLE 5. Comparison of Postoperative Outcomes After Propensity Matching
Posttransplant prolonged ventilation: >3 days Posttransplant ARI* 30-day graft survival 30-day patient survival 1-year graft survival 1-year patient survival
DBD
DCD
P
(n 5 148)
(n 5 74)
Value
47 (31.8)
26 (35.1)
0.609
13 (12.5) 142 (95.9) 143 (96.6)
8 (13.8) 71 (95.9) 71 (95.9)
0.814 1.000 0.799
142 (95.9)
70 (94.6)
0.652
131 (88.5)
67 (90.5)
0.660
NOTE: The data are presented as numbers and percentages. *After the exclusion of patients with preoperative hemodialysis, there were 104 patients in the DBD group and 58 patients in the DCD group.
posttransplant outcomes for DCD and DBD grafts.19,20 It is not entirely clear why these studies are so different. It is reasonable to assume that the patient populations, DCD selection criteria, and clinical practices at various centers were not identical and, therefore, resulted in different outcomes. Recent publications on this subject add additional elements to the controversy. Several recent studies have indicated that DCD grafts are associated with a more complicated intraoperative course, higher costs, longer ICU stays, and posttransplant ARI.4,10,11,21,22 Comparing LT procedures using different grafts is not an easy task. DCD grafts differ in many aspects from DBD grafts. In addition, DCD recipients are considerably different from DBD recipients. Our study showed that patients receiving DCD grafts tended to have lower MELD scores and less severity of the disease before transplant surgery. This is consistent with many previous findings.18 Furthermore, we demonstrated that the DCD group had a less complicated intraoperative course before reperfusion than the DBD group. The differences between DCD LT and DBD LT likely resulted from a matching practice commonly used by surgeons.18 Matching DCD grafts with recipients with low MELD scores has been promoted by several previous studies.18 Such matching may improve the outcomes but creates a difficulty in studying the true effects of DCD grafts on postreperfusion complications and posttransplant outcomes. Many methods have been used to reduce selection bias. Pair matching is a method for pairing 2 groups with various factors.23 Although pair matching can reduce selection bias, it is severely limited by the few factors that can be used in pairs. Propensity matching is a powerful way of eliminating selection bias and
confounding factors.10,24 However, propensity matching requires a large number of study patients and a proper selection of the factors used for the generation of the propensity score. Without a large sample size and a proper selection of factors, the power of the matching will be dramatically reduced. In this study, we performed extensive prematching analyses in an attempt to select appropriate factors for the generation of the propensity score. All 16 factors used to generate the propensity score were independent risk factors for the studied complications or outcomes in this study. The propensity scores generated from these factors provided a wide range of equalization before we analyzed postreperfusion complications and posttransplant outcomes for the DCD and DBD groups. Intraoperative requirements for blood products and vasopressors were topics of several previous studies.4 One of the severe limitations of these previous studies was that intraoperative requirements—not postreperfusion requirements—were used in the analysis. Because liver grafts can be responsible for blood product and vasopressor requirements only after reperfusion, an analysis of the combined requirements of the prereperfusion and postreperfusion periods for blood products and vasopressors may not be accurate. In our study, the requirements for blood products and vasopressors in the postreperfusion period were separated from those in the prereperfusion period. The analysis of the former (rather than the latter or a combination of the two) made sure that the true effects of the liver grafts were studied. In addition, prereperfusion factors have a significant influence on requirements for blood products and vasopressors. In this study, we carefully investigated these factors with multivariate logistic analysis and adjusted for them with propensity matching. Donor variables have been linked to hyperkalemia in previous studies. We have previously shown that DCD, a longer donor hospital stay, and a longer GWIT are associated with postreperfusion hyperkalemia.25 Now, in this study, we confirmed DCD as a risk factor for postreperfusion hyperkalemia after adjustments for many preoperative and prereperfusion factors. We suspect that severe cellular breakdown in DCD grafts may be responsible for this high incidence of postreperfusion hyperkalemia. Previous studies have indicated that DWIT is a risk factor for many posttransplant adverse outcomes18; however, a correlation between DWIT and hyperkalemia could not be confirmed in our study. The exact mechanisms by which DCD grafts are associated with PRS are not known. It is plausible that this is related to ischemia/reperfusion injury in DCD grafts. Many previous studies have suggested that DCD grafts are prone to severe ischemia/reperfusion injury.26 Additional warm ischemia occurring before donor procurement may be responsible for this susceptibility.5 In addition, previous studies have suggested that there is a link between severe ischemia/reperfusion injury and PRS.27 Unfortunately, the
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
degree of ischemia/reperfusion cannot be currently detected with a clinical test. It is well known that the administration of vasopressors immediately before reperfusion will reduce the incidence and severity of conventional PRS.28 The difference in the incidence of PRS between the 2 groups unlikely resulted from the administration of vasopressors because the propensity matching had adjusted for such information before the analysis. The identification and confirmation of DCD grafts as a risk factor for postreperfusion complications have significant clinical implications because many postreperfusion complications, including postreperfusion hyperkalemia and PRS, can be treated or prevented. The administration of a large bolus of insulin during the anhepatic stage has been approved as an effective method for preventing postreperfusion hyperkalemia.29 In a previous study, we also showed that an insulin regimen targeting risk factors (high baseline K1 concentration and RBC transfusions) could be effective in reducing the incidence of postreperfusion hyperkalemia.30 On the basis of the findings of this study, we recommend the inclusion of DCD grafts on the list of risk factors used in the target regimen for hyperkalemia. In fact, we used insulin more frequently later in the study period (2007-2008 and 2009-2011) than earlier (2004-2006), and this reflected more experience in dealing with DCD grafts gained over time. In agreement with the insulin administration, the overall incidence of postreperfusion hyperkalemia decreased over time during the study period. In addition to insulin, many prophylactic and therapeutic measures, including alkalizing drugs (trishydroxymethylaminomethane solution or sodium bicarbonate), hyperventilation, furosemide, calcium chloride, and even intraoperative hemodialysis, can be used for hyperkalemia. Stored RBCs can be washed to reduce the K1 content before administration. We rarely use this technique in adult patients at our center because it is time-consuming and may cause damage to RBCs during the centrifugation process. Many methods, including modified reperfusion techniques, small vasopressor boluses, and adequate volumes before reperfusion, can be used to prevent or treat PRS.31 Several limitations are worth mentioning. First, the study was retrospective in design with many inherent shortcomings. Second, the management of postreperfusion and posttransplant complications was not standardized, and this inevitably introduced some bias into analysis. Third, although we made our best effort to select important risk factors before we compared the 2 groups, a complete list of risk factors could not be guaranteed. Despite the limitations, our study is one of the largest studies using propensity matching to investigate perioperative complications of LT using DCD grafts. Furthermore, we have found that several postreperfusion complications are associated with DCD grafts, and we have made recommendations for the management of these complications.
PAN ET AL. 829
In summary, after adjustments for risk factors, DCD grafts remain a significant risk factor for postreperfusion hyperkalemia and PRS. Prophylactic regimens aimed at decreasing postreperfusion hyperkalemia and PRS are recommended for the management of LT using DCD grafts.
REFERENCES 1. Kim WR, Stock PG, Smith JM, Heimbach JK, Skeans MA, Edwards EB, et al. OPTN/SRTR 2011 annual data report: liver. Am J Transplant 2013;13(suppl 1):73-102. 2. Merion RM. What can we learn from discarded deceased donor kidneys? Transplantation 2006;81:973. 3. DeOliveira ML, Jassem W, Valente R, Khorsandi SE, Santori G, Prachalias A, et al. Biliary complications after liver transplantation using grafts from donors after cardiac death: results from a matched control study in a single large volume center. Ann Surg 2011;254:716-722. 4. Broomhead RH, Patel S, Fernando B, O’Beirne J, Mallett S. Resource implications of expanding the use of donation after circulatory determination of death in liver transplantation. Liver Transpl 2012;18:771-778. 5. Hong JC, Yersiz H, Kositamongkol P, Xia VW, Kaldas FM, Petrowsky H, et al. Liver transplantation using organ donation after cardiac death: a clinical predictive index for graft failure-free survival. Arch Surg 2011;146:10171023. 6. Abt PL, Desai NM, Crawford MD, Forman LM, Markmann JW, Olthoff KM, et al. Survival following liver transplantation from non-heart-beating donors. Ann Surg 2004; 239:87-92. 7. Braat AE, Blok JJ, Putter H, Adam R, Burroughs AK, Rahmel AO, et al.; for European Liver and Intestine Transplant Association (ELITA) and Eurotransplant Liver Intestine Advisory Committee (ELIAC). The Eurotransplant donor risk index in liver transplantation: ET-DRI. Am J Transplant 2012;12:2789-2796. 8. Burroughs AK, Sabin CA, Rolles K, Delvart V, Karam V, Buckels J, et al.; for European Liver Transplant Association. 3-month and 12-month mortality after first liver transplant in adults in Europe: predictive models for outcome. Lancet 2006;367:225-232. 9. Taner CB, Bulatao IG, Keaveny AP, Willingham DL, Pungpapong S, Perry DK, et al. Use of liver grafts from donation after cardiac death donors for recipients with hepatitis C virus. Liver Transpl 2011;17:641-649. 10. Leithead JA, Tariciotti L, Gunson B, Holt A, Isaac J, Mirza DF, et al. Donation after cardiac death liver transplant recipients have an increased frequency of acute kidney injury. Am J Transplant 2012;12:965-975. 11. Croome KP, Wall W, Chandok N, Beck G, Marotta P, Hernandez-Alejandro R. Inferior survival in liver transplant recipients with hepatocellular carcinoma receiving donation after cardiac death liver allografts. Liver Transpl 2013;19:1214-1223. 12. Feng S, Goodrich NP, Bragg-Gresham JL, Dykstra DM, Punch JD, DebRoy MA, et al. Characteristics associated with liver graft failure: the concept of a donor risk index. Am J Transplant 2006;6:783-790. 13. Cameron AM, Ghobrial RM, Yersiz H, Farmer DG, Lipshutz GS, Gordon SA, et al. Optimal utilization of donor grafts with extended criteria: a single-center experience in over 1000 liver transplants. Ann Surg 2006; 243:748-753. 14. Hong JC, Koroleff D, Xia V, Chang CM, Duarte SM, Xu J, et al. Regulated hepatic reperfusion mitigates ischemiareperfusion injury and improves survival after prolonged liver warm ischemia: a pilot study on a novel concept of
830 PAN ET AL.
15.
16.
17.
18.
19.
20.
21.
22.
organ resuscitation in a large animal model. J Am Coll Surg 2012;214:505-515. Tan WF, Steadman RH, Farmer DG, Hong JC, Busuttil RW, Apinyachon W, Xia VW. Pretransplant neurological presentation and severe posttransplant brain injury in patients with acute liver failure. Transplantation 2012; 94:768-774. Mateo R, Cho Y, Singh G, Stapfer M, Donovan J, Kahn J, et al. Risk factors for graft survival after liver transplantation from donation after cardiac death donors: an analysis of OPTN/UNOS data. Am J Transplant 2006;6: 791-796. Mathur AK, Heimbach J, Steffick DE, Sonnenday CJ, Goodrich NP, Merion RM. Donation after cardiac death liver transplantation: predictors of outcome. Am J Transplant 2010;10:2512-2519. Lee KW, Simpkins CE, Montgomery RA, Locke JE, Segev DL, Maley WR. Factors affecting graft survival after liver transplantation from donation after cardiac death donors. Transplantation 2006;82:1683-1688. Dubbeld J, Hoekstra H, Farid W, Ringers J, Porte RJ, Metselaar HJ, et al. Similar liver transplantation survival with selected cardiac death donors and brain death donors. Br J Surg 2010;97:744-753. Grewal HP, Willingham DL, Nguyen J, Hewitt WR, Taner BC, Cornell D, et al. Liver transplantation using controlled donation after cardiac death donors: an analysis of a large single-center experience. Liver Transpl 2009; 15:1028-1035. Callaghan CJ, Charman SC, Muiesan P, Powell JJ, Gimson AE, van der Meulen JH; for UK Liver Transplant Audit. Outcomes of transplantation of livers from donation after circulatory death donors in the UK: a cohort study. BMJ Open 2013;3:e003287. van der Hilst CS, Ijtsma AJ, Bottema JT, van Hoek B, Dubbeld J, Metselaar HJ, et al. The price of donation after cardiac death in liver transplantation: a prospective cost-effectiveness study. Transpl Int 2013;26:411-418.
LIVER TRANSPLANTATION, July 2014
23. Pine JK, Aldouri A, Young AL, Davies MH, Attia M, Toogood GJ, et al. Liver transplantation following donation after cardiac death: an analysis using matched pairs. Liver Transpl 2009;15:1072-1082. 24. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127(pt 2): 757-763. 25. Xia VW, Ghobrial RM, Du B, Chen T, Hu KQ, Hiatt JR, et al. Predictors of hyperkalemia in the prereperfusion, early postreperfusion, and late postreperfusion periods during adult liver transplantation. Anesth Analg 2007; 105:780-785. 26. Cameron A, Busuttil RW. AASLD/ILTS transplant course: is there an extended donor suitable for everyone? Liver Transpl 2005;11(suppl 2):S2-S5. 27. Hilmi I, Horton CN, Planinsic RM, Sakai T, NicolauRaducu R, Damian D, et al. The impact of postreperfusion syndrome on short-term patient and liver allograft outcome in patients undergoing orthotopic liver transplantation. Liver Transpl 2008;14:504-508. 28. Ryu HG, Jung CW, Lee HC, Cho YJ. Epinephrine and phenylephrine pretreatments for preventing postreperfusion syndrome during adult liver transplantation. Liver Transpl 2012;18:1430-1439. 29. De Wolf A, Frenette L, Kang Y, Tang C. Insulin decreases the serum potassium concentration during the anhepatic stage of liver transplantation. Anesthesiology 1993;78: 677-682. 30. Xia VW, Obaidi R, Park C, Braunfeld M, Neelakanta G, Nourmand H, et al. Insulin therapy in divided doses coupled with blood transfusion versus large bolus doses in patients at high risk for hyperkalemia during liver transplantation. J Cardiothorac Vasc Anesth 2010;24: 80-83. 31. Fukazawa K, Pretto EA. The effect of methylene blue during orthotopic liver transplantation on post reperfusion syndrome and postoperative graft function. J Hepatobiliary Pancreat Sci 2011;18:406-413.
ORIGINAL ARTICLE
Perioperative Complications in Liver Transplantation Using Donation After Cardiac Death Grafts: A Propensity-Matched Study Xiongxiong Pan,1,3 Worapot Apinyachon,1 Wei Xia,1 Johnny C. Hong,4 Ronald W. Busuttil,2 Randolph H. Steadman,1 and Victor W. Xia1 Departments of 1Anesthesiology and 2Surgery, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA; 3Department of Anesthesiology, First Affiliated Hospital of Nanjing Medical University, Nanjing, People’s Republic of China; and 4Department of Surgery, Medical College of Wisconsin, Milwaukee, WI
Donation after cardiac death (DCD) is an important source for expanding the donor pool for liver transplantation (LT). Although the long-term outcomes of LT using DCD grafts have been extensively studied, perioperative complications related to DCD grafts are rarely reported. The aim of this study was to determine whether DCD grafts were associated with a higher incidence of postreperfusion complications and worse outcomes in adult LT patients. After institutional review board approval, the medical records of all adult patients who underwent LT at our medical center between 2004 and 2011 were reviewed. Postreperfusion complications and posttransplant outcomes were compared between patients receiving DCD grafts and patients receiving donation after brain death (DBD) grafts. In all, 74 patients received DCD grafts during the study period, and 1369 patients received DBD grafts. An initial comparison showed that many preoperative, prereperfusion, and donor variables in the DCD group differed significantly from those in the DBD group. Propensity matching was chosen so that adjustments could be made for the differences. A postmatching analysis showed that the preoperative, prereperfusion, and donor variables no longer differed between the 2 groups. The postreperfusion requirements for blood products and vasopressors, the posttransplant ventilation times, the incidence of posttransplant acute renal injury, and the 30-day and 1-year patient and graft survival rates were comparable between the 2 groups. However, patients receiving DCD grafts experienced significantly higher rates of hyperkalemia (33.8% versus 18.9%, P < 0.05) and postreperfusion syndrome (PRS; 25.7% versus 12.3%, P < 0.05). In conclusion, after adjustments for preoperative and prereperfusion risks via propensity matching, DCD grafts remained a risk factor for postreperfusion hyperkalemia and PRS. A prophylactic regimen aimed at decreasing postreperfusion hyperkalemia and PRS is recommended for the management of LT using DCD grafts. Liver C 2014 AASLD. Transpl 20:823-830, 2014. V Received December 16, 2013; accepted March 27, 2014.
The demand for organs for liver transplantation (LT) continues to exceed the supply. The discrepancy between the number of available organs and the number of patients waiting for transplantation remains significant.1 To address this urgent shortage and the rising wait-list mortality rate, the expansion of the
donor pool through the use of donation after cardiac death (DCD) grafts has been promoted.2 Although donation after brain death (DBD) remains the predominant source of deceased donor grafts, the use of DCD grafts in LT is being increasingly accepted worldwide. Now LT using DCD grafts accounts for approximately 5% of all
Abbreviations: ARI, acute renal injury; BMI, body mass index; CIT, cold ischemia time; DBD, donation after brain death; DCD, donation after cardiac death; DWIT, donor warm ischemia time; FFP, fresh frozen plasma; GWIT, graft warm ischemia time; ICU, intensive care unit; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; PRS, postreperfusion syndrome; RBC, red blood cell; UCLA, University of California Los Angeles. Address reprint requests to Victor W. Xia, M.D., Department of Anesthesiology, David Geffen School of Medicine, University of California Los Angeles, 757 Westwood Plaza, Suite 3325, Los Angeles, CA 90095. Telephone: 310-825-3664; FAX: 310-794-2141; E-mail: [email protected] DOI 10.1002/lt.23888 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
824 PAN ET AL.
LT procedures in the United States and as many as 50% of all LT procedures in other countries.1,3,4 The procurement of DCD grafts follows a specific procedure that makes DCD grafts different in many ways from DBD grafts.5 After the withdrawal of life support, DCD grafts often have a period during which organ perfusion is poor. The warm ischemia occurring during this period is unique to DCD grafts and may be responsible for exacerbated ischemia/reperfusion injury and posttransplant graft dysfunction.5 Several posttransplant complications, including ischemic cholangitis, nonanastomotic biliary strictures, primary nonfunction, acute renal injury (ARI), and mortality, have been shown to be associated with DCD grafts.6-11 Because of these factors, DCD livers are considered high-risk grafts for LT.12 Although the long-term posttransplant complications of DCD grafts have been extensively studied, the perioperative complications of DCD grafts have not been thoroughly investigated. In this retrospective study, we investigated whether DCD grafts were associated with a higher incidence of perioperative complications in patients undergoing LT. The perioperative complications in this study included complications during the postreperfusion period and adverse outcomes during the immediate posttransplant period.
PATIENTS AND METHODS The institutional review board of the University of California Los Angeles (UCLA) approved this study. Adult patients (18 years) who underwent LT at the UCLA Medical Center between 2004 and 2011 were identified and included in the study. Perioperative data related to LT were prospectively collected and stored in the UCLA transplant database and were retrieved for this study. Donor data and immediate postoperative outcomes were retrospectively collected for this study. DCD donors and recipients were selected according to our center’s protocol as previously described.5,13 The DCD donors were typically younger than 45 years with a body mass index (BMI) less than 30 kg/m2, with hospitalization for 5 days or less, and with serum transaminase levels at procurement less than twice the normal values. DCD grafts should have excellent parenchymal quality during the intraprocurement assessment and a projected cold ischemia time (CIT) less than 8 hours. The DCD procurement followed the Institute of Medicine guidelines and the UCLA procurement protocol.14 All potential DCD donors were controlled through the planned withdrawal of ventilator and organ-perfusion support. Artificial life support was withdrawn in the operating room or intensive care unit (ICU). An independent physician from the donor hospital was assigned to provide end-of-life care, withdraw life support, and declare death. After a 5-minute mandatory waiting period after asystole, procurement started. Livers were stored at 4 C for transport after they had been flushed with 4 to 6 L of cold University of Wisconsin solution via both the
LIVER TRANSPLANTATION, July 2014
abdominal aorta and the inferior mesenteric vein. The DBD procurement was performed according to the standard technique after brain death was declared, and the standard protocol was initiated. Livers from DBD donors were flushed via the aortic and portal system with University of Wisconsin solution and were preserved in a cold cooler until use. The donor warm ischemia time (DWIT) started when life support was withdrawn and ended when cold perfusion started. DWIT occurred only for DCD grafts. The graft warm ischemia time (GWIT) started with the live graft being taken from the ice box and ended with reperfusion. GWIT occurred for both DBD and DCD grafts. The anesthetic and surgical management of the LT recipients followed the protocol of the UCLA team. The anesthetic and surgical management of patients receiving DCD grafts did not differ from that for patients receiving DBD grafts. At our center, we accepted patients who were intubated and on vasopressors before transplant surgery for either DCD or DBD grafts. The anesthetic techniques typically consisted of intravenous induction and combined inhalational and intravenous maintenance. In addition to American Society of Anesthesiologists standard monitors, invasive monitors included intra-arterial and intrapulmonary arterial catheters. Intraoperative transesophageal echocardiography was at the discretion of the anesthesiologist. Stored red blood cells (RBCs) were administered via a rapid transfusion device to maintain a hematocrit of approximately 30%. The conventional bicaval technique was usually used in adult patients. Intraoperative venovenous bypass was selectively used and was determined by the attending surgeon and the anesthesiologist. Indications for venovenous bypass usually included severe portal hypertension, hemodynamic instability, and cardiopulmonary comorbidities that might make patients unable to tolerate cross-clamping of the vena cava or reperfusion. Lactated Ringer’s solution with albumin was used to flush out University of Wisconsin solution before reperfusion of the liver graft. The liver graft was reperfused first via portal flow, which was followed by hepatic artery anastomosis and reperfusion. Blood gas and electrolytes were routinely checked every hour. Hyperkalemia was treated with intravenous insulin, hyperventilation, urine output augmentation with furosemide, alkalizing agents (trishydroxymethylaminomethane solution or sodium bicarbonate), calcium chloride, and intraoperative hemodialysis. The intraoperative course was divided into 2 periods separated by reperfusion of the liver graft: the prereperfusion and postreperfusion periods. Prereperfusion factors were considered potential risk factors for events or complications occurring in the postreperfusion period or posttransplant period. The following perioperative complications were investigated in this study. Postreperfusion complications included increased requirements for blood products and vasopressors, hyperkalemia (defined as a serum K1 concentration 5.5 mmol/L), intraoperative cardiac events requiring cardiopulmonary resuscitation, and
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
PAN ET AL. 825
TABLE 1. Select Preoperative Variables Before Propensity Matching
Age (years)* Sex: male (%) Weight (kg)* BMI (kg/m2)* MELD score* Diabetes mellitus (%) Encephalopathy (%) Etiology (%) Hepatitis C Hepatitis B Hepatocellular carcinoma Acute liver failure Nonalcoholic steatotic hepatitis Baseline laboratory values* Potassium (mmol/L) Creatinine (mg/dL) International normalized ratio Hematocrit (%) Preoperative intubation (%) Preoperative vasopressor (%)
DBD (n 5 1369)
DCD (n 5 74)
P Value
53.6 6 10.9 65.2 79.6 6 22.3 28.5 6 10.8 30.9 6 8.2 25.5 43.3
56.2 6 10.0 60.8 80.0 6 18.8 27.6 6 5.0 27.9 6 7.9 28.4 31.1
0.048 0.441 0.634 0.566 0.002 0.587 0.039
42.5 7.1 32.5 6.0 5.4
37.8 6.8 48.6 2.7 5.4
0.434 0.901 0.004 0.236 0.993
3.9 6 0.6 1.9 6 2.0 1.8 6 1.3 29.4 6 5.8 23.3 13.3
3.9 6 0.7 1.6 6 1.2 1.6 6 0.4 28.7 6 6.5 8.1 2.7
0.789 0.106 0.083 0.270 0.002 0.008
*The data are presented as means and standard deviations. The list of etiology is not complete and one patient may have one or more etiologies.
postreperfusion syndrome (PRS; defined as a 30% decrease in blood pressure from the baseline lasting more than 1 minute within the first 5 minutes after reperfusion of the graft). Postoperative outcomes included ARI within the first week after LT, ventilator duration, and graft and patient survival. Prolonged postoperative ventilation was defined as the need for mechanical ventilation for more than 3 days after LT. The Risk, Injury, Failure, Loss, and End-Stage Kidney Disease criteria were used to define postoperative ARI. ARI was defined as a 2-fold or greater increase in the serum creatinine level in the first week after LT in comparison with the preoperative levels. ARI was analyzed after the exclusion of patients with pretransplant renal failure.15 Data are expressed as means and standard deviations for continuous variables and as proportions for categorical variables. Univariate analyses were performed with the Student t test or Mann-Whitney U test for continuous variables and with the chi-square test for categorical variables. Significant variables from the univariate analyses were selected for multivariate analyses using logistic regression. The propensity-matching method was used to control confounding factors and selection bias between the DCD and DBD groups. Variables that had the potential to influence the use of DCD grafts and outcomes were selected with the multivariate analyses. These selected variables were then used in a logistic regression to generate a propensity score for each patient. The 2 groups were matched with the nearest propensity scores. Postreperfusion complications and posttransplant outcomes were compared for the 2 postmatched groups. One-year graft
and patient survival rates were also estimated with Kaplan-Meier survival curves. All analyses were performed with Statistical Package for Social Science 21.0 for Windows (IBM Corp., Armonk, NY).
RESULTS Between 2004 and 2011, adult LT was performed 1443 times at our center. Seventy-four of these patients (5.5%) received DCD grafts, and 1369 received DBD grafts. The mean age of the patients in the DCD group was 56.2 6 10.0 years (range 5 19-74 years), and the majority of these patients were male (60.8%). The mean Model for End-Stage Liver Disease (MELD) score was 27.9 6 7.9, and approximately half of the patients (48.6%) had a pretransplant diagnosis of hepatocellular carcinoma. The mean DWIT was 27.3 6 10.6 minutes (range 5 12-76 minutes). The recipient and donor variables for the DBD and DCD groups were compared, and the results for select variables are shown in Tables 1 and 2. In comparison with the DBD group, many variables in the DCD group were significantly different. Recipients in the DCD group were older, had a higher frequency of hepatocellular carcinoma and lower MELD scores, and required less preoperative endotracheal intubation and vasopressor use. Intraoperatively, the DCD group received significantly less RBCs and fresh frozen plasma (FFP) before reperfusion of the liver graft in comparison with the DBD group. Because many preoperative, intraoperative, recipient, and donor variables were significantly different between the DCD and DBD groups, propensity
826 PAN ET AL.
LIVER TRANSPLANTATION, July 2014
TABLE 2. Select Intraoperative/Prereperfusion and Donor Variables Before Propensity Matching
Liver-kidney transplantation [n (%)] Venovenous bypass [n (%)] Piggyback technique [n (%)] Prereperfusion RBCs (U)* Prereperfusion FFP (U)* Prereperfusion hyperkalemia [n (%)]† Surgical time (minutes)* Donors Age (years)* Sex: male [n (%)] Weight (kg)* Height (cm)* BMI (kg/m2)* Macrosteatosis 20% [n (%)] ICU before procurement (days)* Graft CIT (minutes)* GWIT (minutes)*
DBD (n 5 1369)
DCD (n 5 74)
P Value
110 (8.0) 497 (36.3) 152 (11.1) 10.0 6 8.5 13.0 6 12.3 181 (13.2) 430.5 6 276.5
3 (4.1) 22 (29.7) 6 (8.1) 7.6 6 6.5 9.5 6 7.6 12 (16.2) 379.5 6 214.4
0.222 0.250 0.613 0.004 0.017 0.465 0.058
39.8 6 16.9 850 (62.1) 76.7 6 19.0 170.2 6 11.3 26.2 6 5.5 126 (9.2) 4.3 6 3.0 371.9 6 175.5 40.9 6 9.9
37.9 6 13.8 52 (70.3) 83.5 6 27.1 172.7 6 10.0 26.7 6 6.5 8 (10.8) 5.6 6 3.1 358.5 6 156.9 40.2 6 9.8
0.350 0.261 0.016 0.147 0.538 0.479 0.004 0.526 0.556
*The data are presented as means and standard deviations. † Defined as a serum potassium > 5.5 mmol/L should be 5.5 mmol/L.
matching was selected to make adjustments for the differences. In order to properly select factors for the propensity score generation, the preoperative, recipient, donor, and prereperfusion variables were analyzed with univariate tests and then with multivariate logistic regression against each studied complication and outcome (detailed data are not shown). A total of 16 preoperative and prereperfusion variables were identified as significant risk factors for postreperfusion complications and adverse posttransplant outcomes. They included the following: age, MELD score, sex, body weight, history of diabetes, cirrhosis caused by hepatitis C, baseline K1 concentration, baseline creatinine concentration, preoperative intubation, preoperative pressor use, prereperfusion RBCs, prereperfusion FFP, prereperfusion K1 concentration, CIT, GWIT, and surgical time. All 16 factors had odds ratios between >1 and 26 with P values < 0.05. Each perioperative complication had 2 to 6 risk factors, and each variable was confirmed to be a risk factor for 1 to 4 perioperative complications. All 16 variables were included in a logistic regression model to generate a propensity score for each patient. The DCD and DBD groups were matched in a 1:2 ratio with the nearest propensity scores. Table 3 shows the results of comparisons of 20 variables for the DCD and DBD groups after propensity matching. The propensity matching eliminated all significant differences between the 2 groups that existed before the matching. In addition, many nonsignificant differences between the 2 groups became smaller after the matching in comparison with the differences based on prematching data. After the propensity matching, postreperfusion complications and posttransplant outcomes were com-
pared between the DCD and DBD groups (Tables 4 and 5). The requirements for postreperfusion RBCs and FFP were similar for the DCD and DBD groups. The use of continuous vasopressor infusion immediately after reperfusion and at the end of surgery in the DCD group was not significantly different from that in the DBD group. Posttransplant ventilation requirements and the incidences of posttransplant ARI were also similar between the 2 groups. The 1year recipient and graft survival rates for the DCD group were comparable to those for the DBD group. However, patients receiving DCD grafts had a significantly higher incidence of hyperkalemia in the postreperfusion period in comparison with the DBD group (Table 4). The mean K1 concentrations at multiple time points were significantly higher for the DCD group versus the DBD group (4.6 6 0.6 versus 4.2 6 0.8 mmol/L, 4.4 6 0.6 versus 4.1 6 0.7 mmol/L, and 4.4 6 0.5 versus 4.0 6 0.5 mmol/L 2 hours after reperfusion, 3 hours after reperfusion, and at the end of surgery, respectively; all P < 0.05). The higher incidence of hyperkalemia in the DCD group did not result from less use of insulin because the administration of intravenous insulin either in boluses or as continuous infusions was similar for the 2 groups. In addition, patients receiving DCD grafts had a significantly higher incidence of PRS in comparison with those receiving DBD grafts. Six patients experienced a cardiac event requiring cardiopulmonary resuscitation immediately after reperfusion of the liver graft: 4 (5.4%) in the DCD group and 2 (1.4%) in the DBD group (P 5 0.272). The postreperfusion cardiac events included severe hypotension, ventricular tachycardia, and asystole. Prereperfusion hyperkalemia was a risk factor for severe cardiac events after reperfusion (odds
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
PAN ET AL. 827
TABLE 3. Comparison After Propensity Matching Variables
DBD (n 5 148)
DCD (n 5 74)
P Value
Age (years)* Sex: male [n (%)] Weight (kg)* MELD score* Baseline hematocrit (%)* Baseline K1 (mmol/L)* Baseline creatinine (mg/dL)* Baseline international normalized ratio* Hepatocellular carcinoma [n (%)] Encephalopathy [n (%)] Hepatitis C cirrhosis [n (%)] Diabetes mellitus [n (%)] Preoperative intubation [n (%)] Preoperative vasopressor [n (%)] Prereperfusion hyperkalemia [n (%)] Prereperfusion RBCs (U)* Prereperfusion FFP (U)* GWIT (minutes)* CIT (minutes)* Surgical time (minutes)*
56.2 6 9.6 89 (60.1) 80 6 19.9 28.5 6 7.4 28.8 6 7.1 3.9 6 0.6 1.7 6 1.4 1.6 6 0.5 70 (47.3) 48 (32.4) 58 (39.2) 42 (28.4) 14 (9.5) 10 (6.8) 20 (13.5) 7.2 6 5.5 8.9 6 6.5 40.1 6 10.0 359.6 6 173.4 360.7 6 230.7
56.2 6 10.0 45 (60.8) 80.8 6 18.8 28.2 6 7.9 28.7 6 6.5 3.9 6 0.7 1.6 6 1.2 1.6 6 0.4 36 (48.6) 23 (31.1) 28 (37.8) 21 (28.4) 6 (8.1) 2 (2.7) 12 (16.2) 7.6 6 6.5 9.5 6 7.6 40.2 6 9.8 358.6 6 156.9 379.5 6 214.4
0.981 0.923 0.764 0.802 0.907 0.553 0.604 0.855 0.849 0.839 0.846 1.000 0.740 0.218 0.589 0.588 0.510 0.951 0.968 0.557
*The data are presented as means and standard deviations.
ratio 5 8.520, 95% confidence interval 5 2.78-26.06, P < 0.001) in the multivariate logistic analysis.
DISCUSSION In this retrospective study, we found that LT patients receiving DCD grafts experienced hyperkalemic episodes and PRS after reperfusion of the liver graft approximately twice as often as those receiving DBD grafts. In addition, there was a trend that DCD grafts were associated with a higher incidence of severe cardiac events immediately after reperfusion, although the difference was not statistically significant. It is likely that our study was underpowered for detecting such relatively rare postreperfusion cardiac incidents. In contrast to previous studies, our study showed that postreperfusion complications (massive blood
transfusion and vasopressor requirements) and posttransplant outcomes (ARI, ventilation time, and graft and patient survival) were not significantly different between patients receiving DCD grafts and patients receiving DBD grafts after adjustments of the risks with the propensity score. Although DCD has been promoted as an important source for expanding the donor pool, the use of such grafts has been surrounded by controversies. In addition to ethical issues and procurement procedures, the outcomes of LT using DCD grafts have been continuously debated in the transplant community.16,17 Early studies using national registry data showed 30% lower graft survival for LT using DCD grafts versus LT using DBD grafts.6,16-18 Therefore, DCD livers are considered high-risk grafts.7,12 Later, several single-center studies reported comparable
TABLE 4. Comparison of Postreperfusion Complications After Propensity Matching DBD (n 5 148)
DCD (n 5 74)
P Value
6.3 6 7.4 29 (19.6) 7.3 6 8.0 2 (1.4) 75 (50.7)
7.6 6 8.4 17 (23.0) 8.7 6 8.3 4 (5.4) 41 (55.4)
0.213 0.558 0.231 0.272 0.560
37 (25.0) 28 (18.9) 18 (12.2)
20 (27.0) 25 (33.8) 19 (25.7)
0.697 0.014 0.019
Postreperfusion RBCs (U)* >10 U of RBCs [n (%)] Postreperfusion FFP (U)* Postreperfusion cardiac events [n (%)] Pressor use immediately after reperfusion [n (%)] Pressor use at end of surgery [n (%)] Postreperfusion hyperkalemia [n (%)] PRS [n (%)] *The data are presented as means and standard deviations.
828 PAN ET AL.
LIVER TRANSPLANTATION, July 2014
TABLE 5. Comparison of Postoperative Outcomes After Propensity Matching
Posttransplant prolonged ventilation: >3 days Posttransplant ARI* 30-day graft survival 30-day patient survival 1-year graft survival 1-year patient survival
DBD
DCD
P
(n 5 148)
(n 5 74)
Value
47 (31.8)
26 (35.1)
0.609
13 (12.5) 142 (95.9) 143 (96.6)
8 (13.8) 71 (95.9) 71 (95.9)
0.814 1.000 0.799
142 (95.9)
70 (94.6)
0.652
131 (88.5)
67 (90.5)
0.660
NOTE: The data are presented as numbers and percentages. *After the exclusion of patients with preoperative hemodialysis, there were 104 patients in the DBD group and 58 patients in the DCD group.
posttransplant outcomes for DCD and DBD grafts.19,20 It is not entirely clear why these studies are so different. It is reasonable to assume that the patient populations, DCD selection criteria, and clinical practices at various centers were not identical and, therefore, resulted in different outcomes. Recent publications on this subject add additional elements to the controversy. Several recent studies have indicated that DCD grafts are associated with a more complicated intraoperative course, higher costs, longer ICU stays, and posttransplant ARI.4,10,11,21,22 Comparing LT procedures using different grafts is not an easy task. DCD grafts differ in many aspects from DBD grafts. In addition, DCD recipients are considerably different from DBD recipients. Our study showed that patients receiving DCD grafts tended to have lower MELD scores and less severity of the disease before transplant surgery. This is consistent with many previous findings.18 Furthermore, we demonstrated that the DCD group had a less complicated intraoperative course before reperfusion than the DBD group. The differences between DCD LT and DBD LT likely resulted from a matching practice commonly used by surgeons.18 Matching DCD grafts with recipients with low MELD scores has been promoted by several previous studies.18 Such matching may improve the outcomes but creates a difficulty in studying the true effects of DCD grafts on postreperfusion complications and posttransplant outcomes. Many methods have been used to reduce selection bias. Pair matching is a method for pairing 2 groups with various factors.23 Although pair matching can reduce selection bias, it is severely limited by the few factors that can be used in pairs. Propensity matching is a powerful way of eliminating selection bias and
confounding factors.10,24 However, propensity matching requires a large number of study patients and a proper selection of the factors used for the generation of the propensity score. Without a large sample size and a proper selection of factors, the power of the matching will be dramatically reduced. In this study, we performed extensive prematching analyses in an attempt to select appropriate factors for the generation of the propensity score. All 16 factors used to generate the propensity score were independent risk factors for the studied complications or outcomes in this study. The propensity scores generated from these factors provided a wide range of equalization before we analyzed postreperfusion complications and posttransplant outcomes for the DCD and DBD groups. Intraoperative requirements for blood products and vasopressors were topics of several previous studies.4 One of the severe limitations of these previous studies was that intraoperative requirements—not postreperfusion requirements—were used in the analysis. Because liver grafts can be responsible for blood product and vasopressor requirements only after reperfusion, an analysis of the combined requirements of the prereperfusion and postreperfusion periods for blood products and vasopressors may not be accurate. In our study, the requirements for blood products and vasopressors in the postreperfusion period were separated from those in the prereperfusion period. The analysis of the former (rather than the latter or a combination of the two) made sure that the true effects of the liver grafts were studied. In addition, prereperfusion factors have a significant influence on requirements for blood products and vasopressors. In this study, we carefully investigated these factors with multivariate logistic analysis and adjusted for them with propensity matching. Donor variables have been linked to hyperkalemia in previous studies. We have previously shown that DCD, a longer donor hospital stay, and a longer GWIT are associated with postreperfusion hyperkalemia.25 Now, in this study, we confirmed DCD as a risk factor for postreperfusion hyperkalemia after adjustments for many preoperative and prereperfusion factors. We suspect that severe cellular breakdown in DCD grafts may be responsible for this high incidence of postreperfusion hyperkalemia. Previous studies have indicated that DWIT is a risk factor for many posttransplant adverse outcomes18; however, a correlation between DWIT and hyperkalemia could not be confirmed in our study. The exact mechanisms by which DCD grafts are associated with PRS are not known. It is plausible that this is related to ischemia/reperfusion injury in DCD grafts. Many previous studies have suggested that DCD grafts are prone to severe ischemia/reperfusion injury.26 Additional warm ischemia occurring before donor procurement may be responsible for this susceptibility.5 In addition, previous studies have suggested that there is a link between severe ischemia/reperfusion injury and PRS.27 Unfortunately, the
LIVER TRANSPLANTATION, Vol. 20, No. 7, 2014
degree of ischemia/reperfusion cannot be currently detected with a clinical test. It is well known that the administration of vasopressors immediately before reperfusion will reduce the incidence and severity of conventional PRS.28 The difference in the incidence of PRS between the 2 groups unlikely resulted from the administration of vasopressors because the propensity matching had adjusted for such information before the analysis. The identification and confirmation of DCD grafts as a risk factor for postreperfusion complications have significant clinical implications because many postreperfusion complications, including postreperfusion hyperkalemia and PRS, can be treated or prevented. The administration of a large bolus of insulin during the anhepatic stage has been approved as an effective method for preventing postreperfusion hyperkalemia.29 In a previous study, we also showed that an insulin regimen targeting risk factors (high baseline K1 concentration and RBC transfusions) could be effective in reducing the incidence of postreperfusion hyperkalemia.30 On the basis of the findings of this study, we recommend the inclusion of DCD grafts on the list of risk factors used in the target regimen for hyperkalemia. In fact, we used insulin more frequently later in the study period (2007-2008 and 2009-2011) than earlier (2004-2006), and this reflected more experience in dealing with DCD grafts gained over time. In agreement with the insulin administration, the overall incidence of postreperfusion hyperkalemia decreased over time during the study period. In addition to insulin, many prophylactic and therapeutic measures, including alkalizing drugs (trishydroxymethylaminomethane solution or sodium bicarbonate), hyperventilation, furosemide, calcium chloride, and even intraoperative hemodialysis, can be used for hyperkalemia. Stored RBCs can be washed to reduce the K1 content before administration. We rarely use this technique in adult patients at our center because it is time-consuming and may cause damage to RBCs during the centrifugation process. Many methods, including modified reperfusion techniques, small vasopressor boluses, and adequate volumes before reperfusion, can be used to prevent or treat PRS.31 Several limitations are worth mentioning. First, the study was retrospective in design with many inherent shortcomings. Second, the management of postreperfusion and posttransplant complications was not standardized, and this inevitably introduced some bias into analysis. Third, although we made our best effort to select important risk factors before we compared the 2 groups, a complete list of risk factors could not be guaranteed. Despite the limitations, our study is one of the largest studies using propensity matching to investigate perioperative complications of LT using DCD grafts. Furthermore, we have found that several postreperfusion complications are associated with DCD grafts, and we have made recommendations for the management of these complications.
PAN ET AL. 829
In summary, after adjustments for risk factors, DCD grafts remain a significant risk factor for postreperfusion hyperkalemia and PRS. Prophylactic regimens aimed at decreasing postreperfusion hyperkalemia and PRS are recommended for the management of LT using DCD grafts.
REFERENCES 1. Kim WR, Stock PG, Smith JM, Heimbach JK, Skeans MA, Edwards EB, et al. OPTN/SRTR 2011 annual data report: liver. Am J Transplant 2013;13(suppl 1):73-102. 2. Merion RM. What can we learn from discarded deceased donor kidneys? Transplantation 2006;81:973. 3. DeOliveira ML, Jassem W, Valente R, Khorsandi SE, Santori G, Prachalias A, et al. Biliary complications after liver transplantation using grafts from donors after cardiac death: results from a matched control study in a single large volume center. Ann Surg 2011;254:716-722. 4. Broomhead RH, Patel S, Fernando B, O’Beirne J, Mallett S. Resource implications of expanding the use of donation after circulatory determination of death in liver transplantation. Liver Transpl 2012;18:771-778. 5. Hong JC, Yersiz H, Kositamongkol P, Xia VW, Kaldas FM, Petrowsky H, et al. Liver transplantation using organ donation after cardiac death: a clinical predictive index for graft failure-free survival. Arch Surg 2011;146:10171023. 6. Abt PL, Desai NM, Crawford MD, Forman LM, Markmann JW, Olthoff KM, et al. Survival following liver transplantation from non-heart-beating donors. Ann Surg 2004; 239:87-92. 7. Braat AE, Blok JJ, Putter H, Adam R, Burroughs AK, Rahmel AO, et al.; for European Liver and Intestine Transplant Association (ELITA) and Eurotransplant Liver Intestine Advisory Committee (ELIAC). The Eurotransplant donor risk index in liver transplantation: ET-DRI. Am J Transplant 2012;12:2789-2796. 8. Burroughs AK, Sabin CA, Rolles K, Delvart V, Karam V, Buckels J, et al.; for European Liver Transplant Association. 3-month and 12-month mortality after first liver transplant in adults in Europe: predictive models for outcome. Lancet 2006;367:225-232. 9. Taner CB, Bulatao IG, Keaveny AP, Willingham DL, Pungpapong S, Perry DK, et al. Use of liver grafts from donation after cardiac death donors for recipients with hepatitis C virus. Liver Transpl 2011;17:641-649. 10. Leithead JA, Tariciotti L, Gunson B, Holt A, Isaac J, Mirza DF, et al. Donation after cardiac death liver transplant recipients have an increased frequency of acute kidney injury. Am J Transplant 2012;12:965-975. 11. Croome KP, Wall W, Chandok N, Beck G, Marotta P, Hernandez-Alejandro R. Inferior survival in liver transplant recipients with hepatocellular carcinoma receiving donation after cardiac death liver allografts. Liver Transpl 2013;19:1214-1223. 12. Feng S, Goodrich NP, Bragg-Gresham JL, Dykstra DM, Punch JD, DebRoy MA, et al. Characteristics associated with liver graft failure: the concept of a donor risk index. Am J Transplant 2006;6:783-790. 13. Cameron AM, Ghobrial RM, Yersiz H, Farmer DG, Lipshutz GS, Gordon SA, et al. Optimal utilization of donor grafts with extended criteria: a single-center experience in over 1000 liver transplants. Ann Surg 2006; 243:748-753. 14. Hong JC, Koroleff D, Xia V, Chang CM, Duarte SM, Xu J, et al. Regulated hepatic reperfusion mitigates ischemiareperfusion injury and improves survival after prolonged liver warm ischemia: a pilot study on a novel concept of
830 PAN ET AL.
15.
16.
17.
18.
19.
20.
21.
22.
organ resuscitation in a large animal model. J Am Coll Surg 2012;214:505-515. Tan WF, Steadman RH, Farmer DG, Hong JC, Busuttil RW, Apinyachon W, Xia VW. Pretransplant neurological presentation and severe posttransplant brain injury in patients with acute liver failure. Transplantation 2012; 94:768-774. Mateo R, Cho Y, Singh G, Stapfer M, Donovan J, Kahn J, et al. Risk factors for graft survival after liver transplantation from donation after cardiac death donors: an analysis of OPTN/UNOS data. Am J Transplant 2006;6: 791-796. Mathur AK, Heimbach J, Steffick DE, Sonnenday CJ, Goodrich NP, Merion RM. Donation after cardiac death liver transplantation: predictors of outcome. Am J Transplant 2010;10:2512-2519. Lee KW, Simpkins CE, Montgomery RA, Locke JE, Segev DL, Maley WR. Factors affecting graft survival after liver transplantation from donation after cardiac death donors. Transplantation 2006;82:1683-1688. Dubbeld J, Hoekstra H, Farid W, Ringers J, Porte RJ, Metselaar HJ, et al. Similar liver transplantation survival with selected cardiac death donors and brain death donors. Br J Surg 2010;97:744-753. Grewal HP, Willingham DL, Nguyen J, Hewitt WR, Taner BC, Cornell D, et al. Liver transplantation using controlled donation after cardiac death donors: an analysis of a large single-center experience. Liver Transpl 2009; 15:1028-1035. Callaghan CJ, Charman SC, Muiesan P, Powell JJ, Gimson AE, van der Meulen JH; for UK Liver Transplant Audit. Outcomes of transplantation of livers from donation after circulatory death donors in the UK: a cohort study. BMJ Open 2013;3:e003287. van der Hilst CS, Ijtsma AJ, Bottema JT, van Hoek B, Dubbeld J, Metselaar HJ, et al. The price of donation after cardiac death in liver transplantation: a prospective cost-effectiveness study. Transpl Int 2013;26:411-418.
LIVER TRANSPLANTATION, July 2014
23. Pine JK, Aldouri A, Young AL, Davies MH, Attia M, Toogood GJ, et al. Liver transplantation following donation after cardiac death: an analysis using matched pairs. Liver Transpl 2009;15:1072-1082. 24. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127(pt 2): 757-763. 25. Xia VW, Ghobrial RM, Du B, Chen T, Hu KQ, Hiatt JR, et al. Predictors of hyperkalemia in the prereperfusion, early postreperfusion, and late postreperfusion periods during adult liver transplantation. Anesth Analg 2007; 105:780-785. 26. Cameron A, Busuttil RW. AASLD/ILTS transplant course: is there an extended donor suitable for everyone? Liver Transpl 2005;11(suppl 2):S2-S5. 27. Hilmi I, Horton CN, Planinsic RM, Sakai T, NicolauRaducu R, Damian D, et al. The impact of postreperfusion syndrome on short-term patient and liver allograft outcome in patients undergoing orthotopic liver transplantation. Liver Transpl 2008;14:504-508. 28. Ryu HG, Jung CW, Lee HC, Cho YJ. Epinephrine and phenylephrine pretreatments for preventing postreperfusion syndrome during adult liver transplantation. Liver Transpl 2012;18:1430-1439. 29. De Wolf A, Frenette L, Kang Y, Tang C. Insulin decreases the serum potassium concentration during the anhepatic stage of liver transplantation. Anesthesiology 1993;78: 677-682. 30. Xia VW, Obaidi R, Park C, Braunfeld M, Neelakanta G, Nourmand H, et al. Insulin therapy in divided doses coupled with blood transfusion versus large bolus doses in patients at high risk for hyperkalemia during liver transplantation. J Cardiothorac Vasc Anesth 2010;24: 80-83. 31. Fukazawa K, Pretto EA. The effect of methylene blue during orthotopic liver transplantation on post reperfusion syndrome and postoperative graft function. J Hepatobiliary Pancreat Sci 2011;18:406-413.
