Despite the decrease in perioperative mortality observed during orthotopic liver transplantation (OLT), intraoperative major hemodynamic events remain a serious concern for the anesthesiologist. These events seem to be indeed associated with adverse outcomes during the postoperative period.1 The period of greatest hemodynamic disturbances during OLT occurs at graft reperfusion. After unclamping of the portal vein, postreperfusion syndrome (PRS) is characterized by a marked decrease in systemic blood pressure, systemic vascular resistance, and cardiac output and a moderate increase in pulmonary arterial pressure.2 The underlying mechanisms of these severe hemodynamic changes are complex. Cold and acidotic components released at the time of reperfusion have been reported to play an important role in this phenomenon.3 Proinflammatory cytokines such as interleukin 6 or tumor necrosis factor alpha are produced during ischemia and are also possibly involved in the production of hypotension at that time.4, 5 Moreover, ischemia/reperfusion syndrome, occurring in every liver transplant procedure, could be correlated with the hemodynamic changes. Indeed, the cellular response to ischemia and reperfusion can lead to a systemic inflammatory response and subsequent cellular damages.6 Clinical factors associated with the occurrence of PRS are not fully understood. Moreover, the impact of these intraoperative hemodynamic changes on postoperative morbidity or mortality are a matter of debate.3, 7 Therefore, the primary aim of our study was to determine the clinical independent predictors of the occurrence of intraoperative PRS. A secondary objective was to examine the postoperative consequences associated with PRS after OLT in patients with cirrhosis.
During orthotopic liver transplantation (OLT), a marked decrease in blood pressure following unclamping of the portal vein and liver reperfusion is frequently observed and is termed postreperfusion syndrome (PRS). The predictive factors and clinical consequences of PRS are not fully understood. The goal of this study was to identify predictors of PRS and morbidity/mortality associated with its occurrence during OLT in patients with cirrhosis. During a 3-year period, all consecutive OLT procedures performed in patients with cirrhosis were studied. Exclusion criteria were OLT for acute liver failure, early retransplantation, combined liver/kidney transplantation, and living-donor related transplantation. PRS was defined as a decrease in the mean arterial pressure of more than 30% of the value observed in the anhepatic stage, for more than 1 minute during the first 5 minutes after reperfusion of the graft. Transplantation was performed with preservation of the inferior vena cava with or without temporary portocaval shunt. Associations between PRS and donor and recipient demographic data, recipient operative and postoperative outcomes were tested with bivariate statistics. Independent predictors of PRS were determined in multivariable logistic regression analysis. Of the 75 patients included in the study, 20 patients (25%) developed PRS. In a multivariable analysis, absence of a portocaval shunt [odds ratio (95% confidence interval) = 4.42 (1.18-17.6)] and duration of cold ischemia [odds ratio (95% confidence interval) = 1.34 (1.07-1.72)] were independent predictors of PRS. Patients who experienced PRS displayed more postoperative renal failure and lower early (<15 days after OLT) survival (80% versus 96%; P = 0.04). In conclusion, the absence of portocaval shunt and the duration of cold ischemia were independent predictors of intraoperative PRS. PRS was associated with significant adverse postoperative outcome. These results provide realistic clinical targets to improve patient outcome after OLT for cirrhosis. Liver Transpl 15:522–529, 2009. © 2009 AASLD.
PATIENTS AND METHODS
After institutional review board approval, this observational study was conducted on all consecutive patients with cirrhosis undergoing cadaveric OLT in a single adult liver transplantation center. Exclusion criteria were OLT for noncirrhotic liver diseases, acute liver failure, early retransplantation, transplantation with living-related graft donation, and simultaneous kidney and liver transplantation.
Anesthesia was induced with intravenous thiopental (5 mg/kg), fentanyl (3 μg/kg), and succinylcholine (0.1 mg/kg). After intubation of the trachea, anesthesia was maintained with isoflurane and additional fentanyl and atracurium. Mechanical ventilation was begun with a tidal volume of 10 mL/kg, a respiratory rate of 14, and an inspired oxygen fraction of 0.5 in air. Patients were monitored with radial and pulmonary artery catheters. Intravenous fluids (gelatins) were infused to correct hypovolemia. Losses related to ascites were corrected by the intravenous administration of 4% albumin for half of the volume. A continuous infusion of norepinephrine was started if the systolic arterial blood pressure remained under 90 mm Hg despite adequate filling pressure. Packed red blood cells were administered in order to maintain a hemoglobin blood level above 90 g/L. Fresh frozen plasma was administered to treat clinically significant bleeding, and platelets were infused to maintain a platelet count above 50 × 109/L. Antifibrinolytic therapy (aprotinin) was used to treat documented fibrinolysis. The intraoperative immunosuppressive regimen included a 5 mg/kg bolus of methylprednisone at the beginning of the anhepatic stage.
Liver allografts were harvested by senior surgeons and were preserved with cold University of Wisconsin solution. Venovenous bypass was never used. The piggyback technique and preservation of the inferior vena cava were used for all transplants with or without temporary portocaval shunting according to the attending surgeon. Similarly, initial revascularization of the graft was performed with the portal vein or hepatic artery first.
PRS was considered when the mean arterial blood pressure was 30% lower than the previous value immediately at the end of the anhepatic stage and lasted for at least 1 minute within the 5 minutes after unclamping.2, 4, 8 The hemodynamic data were collected contemporaneously. In the period preceding revascularization and reperfusion of the graft liver, a rapid infusion of colloid (500 mL) was administered. The arterial pH value was checked and corrected if necessary to be greater than 7.28 with sodium bicarbonate and an increase in mechanical ventilation. If the systolic blood pressure remained under 95 mm Hg despite volume loading, norepinephrine was administered in a continuous infusion before revascularization of the liver. Moreover, during this period, hypocalcemia was checked and corrected if necessary. There was no systematic pretreatment with CaCl2, methylene blue, or mannitol. After unclamping, patients whose mean arterial pressure was lower than 55 mm Hg were treated first with a bolus of 250 μg of phenylephrine intravenously. Patients who did not recover promptly (less than 5 minutes) received a continuous administration of norepinephrine titrated to maintain a mean arterial pressure above 65 mm Hg. Moreover, the fluid infusion was titrated according to the hemodynamic data. Patients who displayed PRS syndrome as previously mentioned were classified as PRS+, and patients who did not were classified as PRS−. Patient data were then analyzed according to the occurrence of PRS.
Recipient preoperative variables (age, etiology of cirrhosis, Child-Pugh classification, Model for End-Stage Liver Disease score, and preoperative β-blocker therapy) and donor criteria (age, body mass index, and duration of cold ischemia) were recorded. Intraoperative variables included blood product transfusion, vasoconstrictor and norepinephrine requirements, use of a temporary portocaval shunt, and duration of the surgical procedure. Postoperative variables included the duration of postoperative mechanical ventilation, intensive care unit stay, early death occurring during the first 15 days after OLT, and death in the intensive care unit. Severe postoperative renal failure was defined as creatinine clearance under 30 mL/minute according to the Cockroft formula.
Statistical analysis was performed with Excel version 11.3.5 (Microsoft Corp., 2004) and JMP version 7 (SAS Institute, Inc., Cary, NC, 1989-2007). Data are expressed as median (25-75 interquartile range) or number and percentage. The normality of distributions and equality of variances of interval variables were checked with the Shapiro-Wilk W test and Levene's test (univariate analysis) before parametric analysis was undertaken. Bivariate analysis tested the associations between 2 variables. Associations between both binary nominal variables were tested with Fisher's exact test (2-tailed). Associations between nominal and ordinal variables or nonnormally distributed continuous variables were tested with the Kruskall-Wallis or Wilcoxon rank sum test. When continuous variable distributions were not different from the normal distribution and variances were not noticeably different, we tested associations with nominal variables using analysis of variance and the Student t test (2 groups). Pearson-Yates chi-square tests (qualitative, independent variables) and simple logistic regressions (continuous, independent variables) were, moreover, performed for nominal dependent variables. When significant associations (P < 0.05) were found with bivariate statistics between PRS as the outcome and dependent and independent variables, we retained these factors for a multivariable analysis. Multivariate statistics were then performed with nominal logistic models (with PRS as a dependent variable). Survival analysis using the Kaplan-Meier product-limit method, log-rank test, and Wilcoxon's test for censored data (because early events were particularly important) was performed. P < 0.05 was considered the threshold for significance.
During the period of the study, 188 patients underwent OLT in our institution. Seventy-five patients received OLT for cirrhosis and were included in this observational study. Demographic data of recipients and donors are depicted in Table 1. PRS occurred in 20 patients (25%). The preoperative data of both groups showed that there was no significant difference in gender, weight, etiology of cirrhosis, use of β-blocker therapy, Child Pugh classification, or Model for End-Stage Liver Disease score. Similarly, there were no significant differences in the graft variables, that is, age and percentage of steatosis. The only significant difference in bivariate analysis was the duration of cold ischemia, which was significantly longer in the patients with PRS [630 (480-730) versus 495 (420-615) minutes; P = 0.01]. No reperfusion syndrome occurred in patients with a duration of cold ischemia less than 6 hours.
|Total (n = 75)||PRS− (n = 55)||PRS+ (n = 20)||P|
|Age (years)||52 (48-56)||53 (46-56-62)||51 (48-54-62)||0.2|
|Recipient body mass index||26 (23-29)||26 (24-29)||26 (22-29)||0.85|
|Alcoholic cirrhosis||35 (47%)||26 (47%)||9 (45%)||0.6|
|Viral cirrhosis||46 (61%)||37 (67%)||9 (45%)||0.16|
|β-Blocker therapy||48 (64%)||34 (62%)||14 (70%)||0.25|
|Child-Pugh classification (A/B/C)||20/44/16||14/33/13||6/11/3||0.63|
|MELD score||15 (10-21)||16 (11-22)||11 (8-21)||0.12|
|Preoperative GFR (mL/minute)||108 (91-133)||108 (91-133)||110 (90-135)||0.73|
|Preoperative creatinine (μmol/L)||73 (66-87)||72 (66-87)||74 (68-86)||0.74|
|Donor graft age (years)||49 (36-57)||50 (39-57)||44 (28-55)||0.07|
|Graft steatosis > 30% (n = 68)||21 (31%)||15/50 (30%)||6/18 (33%)||0.78|
|Graft steatosis > 60% (n = 68)||4 (6%)||3 (6%)||1 (5.5%)||0.85|
|Duration of warm ischemia (minutes)||62 (50-80)||62 (50-76)||65 (49-80)||0.93|
|Duration of cold ischemia (minutes)||510 (440-650)||495 (420-615)||630 (480-730)||0.01|
Intraoperative data (Table 2) showed that patients with PRS had a significantly longer duration of surgery and less frequent placement of a portocaval shunt. Absence of a shunt was associated with a relative risk (95% confidence interval) of PRS of 2.3 (1.1, 4.8; P = 0.04). The frequency of initial portal revascularization was similar in both groups. Packed red blood cells, fresh frozen plasma, and platelet transfusion requirements were similar in both groups. Hemodynamic and biological profiles are depicted in Table 3. There was no significant difference in the cardiac index or pulmonary wedge pressure at any time between PRS+ and PRS− patients. Vasoconstrictors were more often used as a bolus in the PRS group. However, there was no significant difference in the continuous administration of vasopressors in the neohepatic stage.
|Total (n = 75)||PRS− (n = 55)||PRS+ (n = 20)||P|
|Length of surgery (minutes)||570 (522-660)||570 (490-630)||660 (540-725)||0.02|
|Anhepatic duration (minutes)*||120 (90-140)||115 (85-140)||134 (114-155)||0.06|
|Temporary portocaval shunt||61 (81%)||48 (87%)||13 (65%)||0.043|
|Initial portal revascularization||61 (81%)||44 (80%)||17 (85%)||0.74|
|Intraoperative blood loss (L)||1.55 (1-2.8)||1.5 (1-2.5)||1.9 (1.2-5)||0.11|
|Transfusion of PRC (units)*||3 (0-5)||3 (0-5)||3 (0-9)||0.38|
|Transfusion of FFP (units)*||1 (0-5)||0 (0-4)||4 (0-6)||0.11|
|Transfusion of platelets (units)*||0 (0-1)||0 (0-1)||1 (0-1)||0.27|
|Total (n = 75)||PRS− (n = 55)||PRS+ (n = 20)||P|
|MAP at the initiation of surgery (mm Hg)||75 (67-85)||76 (67-86)||71 (67-85)||0.8|
|MAP before reperfusion (mm Hg)||76 (70-85)||77 (68-79)||80 (75-86)||0.1|
|MAP 5 minutes after reperfusion (mm Hg)||69 (60-75)||70 (65-76)||60 (58-69)||0.01|
|MAP at the end of surgery (mm Hg)||76 (68-81)||75 (68-80)||73 (65-85)||0.56|
|CI at the initiation of surgery (L/minute/m2)||3.6 (3-4.4)||3.6 (3-4.5)||4.1 (3-4.1)||0.26|
|CI before reperfusion (L/minute/m2)||3.1 (2.5-4.)||3.3 (2.5-4.1)||2.8 (2.3-3.9)||0.5|
|CI 5 minutes after reperfusion (L/minute/m2)||4.5 (3.2-5.5)||4.6 (3.3-5.8)||3.6 (3.1-4.7)||0.31|
|CI at the end of surgery (L/minute/m2)||4 (3.3-5.3)||4.1 (3.3-5.5)||4. (3.1-5)||0.6|
|PWP at the initiation of surgery (mm Hg)||11 (8-14)||11 (8-15)||11 (8-13)||0.43|
|PWP before reperfusion (mm Hg)||11 (7-14)||11 (7-14)||11 (7-14)||0.73|
|PWP 5 minutes after reperfusion (mm Hg)||10 (6-14)||10 (6-14)||8 (5-12)||0.3|
|PWP at the end of surgery (mm Hg)||10 (7-12)||10 (7-12)||10 (8-13)||0.74|
|Arterial pH before reperfusion||7.33 (7.29-7.39)||7.33 (7.29-7.39)||7.33 (7.24-7.39)||0.33|
|Temperature before reperfusion (°C)||35.1 (34.3-35.6)||35.1 (34.4-35.6)||34.9 (34-35.6)||0.078|
|Prereperfusion VP infusion||16 (21%)||13 (26%)||3 (15%)||0.53|
|Postreperfusion VP bolus||35 (47%)||21 (38%)||14 (70%)||0.02|
|Postreperfusion VP infusion||37 (49%)||27 (49%)||10 (50%)||1|
Only 2 independent variables were found to be significantly different in bivariate analysis and were included in a multivariable analysis. In nominal logistic regression, the duration of cold ischemia and absence of a portocaval shunt were independent predictors of reperfusion syndrome (Table 4). After 6 hours of cold ischemia, the incidence of PRS increased [odds ratio (95% confidence interval) = 1.34 (1.07, 1.72); P = 0.035] for each 60-minute step of cold ischemia.
|Total (n = 75)||PRS− (n = 55)||PRS+ (n = 20)||Multivariable Analysis||P|
|Cold ischemia duration||10 (10%)||5 (6%)||5 (50%)||OR (95% CI): 1.34 (1.07-1.72)||0.01|
|No PCS||14 (19%)||7 (13%)||7 (45%)||OR (95% CI): 4.42 (1.18-17.6)||0.03|
Postoperative data (Table 5) showed that patients with PRS displayed severe renal failure more frequently [14 (70%) versus 20 (36%); P = 0.03]. PRS increased the incidence of severe renal failure with a relative risk of 3.3 (1.7, 6.4; P = 0.01). The occurrence of PRS was associated with significantly greater early postoperative mortality in the 15 first postoperative days [relative risk = 2.87 (1.41, 5.85); P = 0.04]. The intensive care unit mortality of patients who experienced PRS was not statistically different from that of those who did not. The overall 5-year survival rate of the 75 patients was 80%: 66% in the PRS+ group and 85% in the PRS− group (Wilcoxon test, P = 0.09; Fig. 1).
|Total (n = 75)||PRS− (n = 55)||PRS+ (n = 20)||P|
|Length of mechanical ventilation (days)*||2 (1-4)||2 (1-4)||2 (1-7)||0.3|
|Severe acute renal failure||34 (45%)||20 (36%)||14 (70%)||0.02|
|Extrarenal epuration therapy||6 (8%)||3 (5%)||3 (15%)||0.3|
|ICU length of stay (days)||6 (5-11)||6 (5-10)||6.5 (5-11)||0.56|
|Early ICU death (before day 15)||6 (8%)||2 (4%)||4 (20%)||0.04|
|Total ICU death||9 (12%)||4 (7.3%)||5 (25%)||0.051|
The main results of our observational study can be summarized in 2 findings: in patients with cirrhosis undergoing OLT, (1) the duration of cold ischemia and absence of a portocaval shunt were independent predictors of PRS, and (2) the occurrence of intraoperative PRS was associated with significantly more frequent postoperative renal failure and with more frequent early postoperative death.
The incidence of development of PRS during OLT varies greatly between studies. The first studies reported an incidence between 2% and 30%.2 However, more recently, the incidence has been reported to be around 50%.7, 9 Finally, in our current study, PRS occurred in 25% of the patients with cirrhosis undergoing OLT. Many differences between all these studies probably account for the wide variation of reported incidence. The most important difference seems to be within the definition of PRS. Variation of the mean arterial blood pressure has been used as a relative decrease from the level of pressure during the anhepatic stage or as an absolute value under 60 mm Hg, for example.9, 10 These criteria are easy to determine but are probably restrictive. Hemodynamic variations do not account for a possible systemic inflammation associated with graft revascularization and ischemia/reperfusion.11 Hilmi et al.7 proposed a new score to assess the severity of PRS based not only on the variation of blood pressure but also on the occurrence of fibrinolysis during the postreperfusion period. Another important discrepancy relies on the design of the studies. Previously reported incidences are based mostly on retrospective studies. One of the major strengths of the current study is its reliance on the contemporaneous determination of hemodynamic changes occurring during reperfusion. The percentage of the decrease in blood pressure between the mean arterial pressure immediately before unclamping of the portal vein or hepatic artery and the lowest blood pressure for at least 1 minute during the 5 minutes after unclamping was calculated, as proposed by many authors. Finally, previous studies included all patients who underwent OLT with various reasons for transplantation. The present study focused on patients with cirrhosis undergoing transplantation. The hemodynamic consequences of cirrhosis are well described and can potentially influence intraoperative hemodynamic variations.12 Similarly, surgical techniques also varied greatly, including various incidences of the use of venovenous bypass, preservation of vena cava, or piggyback techniques. These differences in surgical technique could influence the occurrence of PRS. Likewise, the potential effects of venovenous bypass on PRS remain controversial.13, 14 The addition of all these factors probably explains the wide variation of the reported incidence of PRS. From our observational study focused on a homogeneous population of patients with cirrhosis undergoing OLT with preservation of the vena cava and piggyback techniques, we reported an incidence of 25%, which confirms and extends the clinical relevance of this syndrome in these patients.
The knowledge of clinical predictors of PRS is particularly attractive because it could help to anticipate patients at risk. Potentially, predictors of PRS could be identified in preoperative recipient factors, donor factors, or surgical factors. Among potential recipient variables, no factor appeared to be a predictor of PRS in the current study. This is in accordance with most of the previously reported results.3, 7, 9, 10 Consequently, PRS appears to be unpredictable on the basis of only recipient factors.
Among donor variables and surgical factors, we identified the duration of cold ischemia and the absence of a portocaval shunt as predictors of PRS. Initial reports indicated that the severity of the reperfusion syndrome is closely related to the surgical technique.2 The most recent studies have shown that the preservation of caval flow does not prevent systemic arterial hypotension during graft reperfusion.15 The inferior vena cava preservation technique does not prevent splanchnic congestion secondary to portal vein occlusion. The potential benefits of a temporary portocaval shunt in patients with cirrhosis during the anhepatic stage is controversial because such patients with portal hypertension have developed portosystemic venous collaterals.15–17 This technique has been suggested to be beneficial in patients with cirrhosis who maintained a high portal venous flow with a significant reduction of transfusion requirements.18 Moreover, reducing splanchnic congestion could limit splanchnic ischemia and the subsequent release of toxic mediators at reperfusion.16 This hypothesis may account, at least in part, for the protective effect of a portocaval shunt toward the incidence of intraoperative PRS. These results may provide additional arguments for the use of a temporary portocaval shunt during OLT for cirrhosis.
Graft characteristics may be predictors of PRS.19 The only graft factor identified in the current study was the duration of cold ischemia. This was previously reported in a retrospective study by Chui et al.10 The deleterious effects of the duration of cold ischemia have been previously suggested. At a microscopic level, cold ischemia/warm reperfusion injury of the liver is characterized by marked alterations of sinusoidal endothelial cells followed by hepatocyte apoptosis.20 An extended duration of cold ischemia induces activation of cellular nuclear factor kappa B in rat grafts after OLT, possibly by promoting the proinflammatory response.21 Furthermore, a sustained period of cold preservation significantly impairs the ability of the liver to regenerate.22 Finally, from a clinical perspective, the negative impact of cold ischemia has been reported to induce a decrease in the graft survival rate.23, 24 Our results emphasize that the duration of cold ischemia has negative effects during all the stages of OLT. Recognition of this predictor seems to be crucial today. Use of extended donor criteria grafts has indeed become commonplace to increase the rate of OLT and to reduce the rate of recipient death while patients are on the waiting list. Whether the use of these grafts can increase PRS syndrome remains debatable.3, 7, 9, 10 Criteria for extended donors are not universal. Threshold values for donor age, percentage of macrosteatosis on biopsy, and cold and warm ischemia times are not definitively fixed.7 Among them, duration of cold ischemia was the only significantly different criterion in this study. However, the impact of a combination of extended criteria on the occurrence of PRS has to be determined.
Another important finding of this current study is related to the negative impact of PRS on early postoperative outcome because PRS is associated with greater early mortality. This association does not mean a causative relation. Confounding factors associated with PRS or postoperative management could be responsible for this difference. Despite a trend toward lower long-term survival in the PRS group, there was no significant difference between the 2 groups. This could be related to the restrictive sample examined. The potential negative impact of PRS has been rarely reported. Hilmi et al.7 found that the intensive care unit and hospital length of stay and days on a ventilator are greater in patients with PRS. Moreover, these patients display increased allograft loss and a significantly higher retransplantation rate without excess long-term mortality. Reasons for the negative impact of PRS on early outcome are not clear. In our study, there was no difference in blood product transfusion requirement between patients with or without PRS. Thus, the poorest outcome observed in the PRS group cannot be related to an effect of transfusion.25, 26 Hypotension per se occurring during PRS can adversely affect renal function. Postoperative renal impairment has been shown to be a risk factor of death after OLT.27, 28 An alternative hypothesis could rely on the magnitude of PRS. It can be speculated that reperfusion syndrome could be considered a surrogate marker of ischemia/reperfusion injury occurring during every OLT.11 Ischemia/reperfusion can induce a systemic inflammatory response that may trigger multiorgan failure and subsequent death.6, 11
Finally, knowledge of clinical factors significantly associated with PRS may help to determine effective strategies to prevent this syndrome. Various approaches have been already advocated.5 For example, an interesting field of progress seems to have been opened with the administration of inhaled nitric oxide.29 However, scientific research in these fields is ongoing and will probably lead to a change in our clinical practice in a few years. Conversely, a reduction in the cold ischemia time and the creation of a portocaval shunt can be proposed as soon as today.
Our work highlights the potential role of the cold ischemia time and the absence of a portocaval shunt in the occurrence of PRS during liver transplantation for cirrhosis performed with the piggyback technique. Moreover, PRS appears to have clinically negative effects on postoperative outcome. These deleterious effects on clinical outcome emphasize the need to pay great attention to predictors of this syndrome. Interestingly, organ procurement organizations and surgeons can play a positive role in these 2 factors. Finally, knowledge of these predictors offers 2 simple ways to further improve results of liver transplantation in patients with cirrhosis.