ORIGINAL ARTICLE
When and Why Portal Vein Thrombosis Matters in Liver Transplantation A Critical Audit of 174 Cases Taizo Hibi, MD, PhD,∗ † Seigo Nishida, MD, PhD,∗ † David M. Levi, MD,§ Gennaro Selvaggi, MD,∗ † Akin Tekin, MD,∗ † Ji Fan, MD,∗ † Phillip Ruiz, MD, PhD,∗ ‡ and Andreas G. Tzakis, MD, PhD¶ Objective: To identify complications associated with different techniques utilized to treat portal vein thrombosis (PVT) during primary liver transplantation and their impact on survival. Background: PVT remains an intricate problem in liver transplantation, and the long-term outcomes of patients with PVT who undergo transplantation are not well defined. Methods: We performed a retrospective cohort analysis of all consecutive adult patients who underwent primary isolated liver transplantation from 1998 to 2009 (median follow-up period, 89 months). The outcomes of patients with PVT were compared with those without PVT. Results: Among 1379 recipients, 174 (12.6%) had PVT at the time of transplantation [83 (48%) complete and 91 (52%) partial]. Among PVT patients with reestablished physiological portal inflow (PVT: physiological group; n = 149), 123 underwent thrombectomies, 16 received interpositional vein grafts, and 10 received mesoportal jump grafts. In 25 patients, physiological portomesenteric venous circulation was not reconstituted (PVT: nonphysiological group; 18 underwent cavoportal hemitranspositions, 6 renoportal anastomoses, and 1 arterialization). The PVT: nonphysiological group suffered a significantly increased incidence of rethrombosis of the portomesenteric veins and gastrointestinal bleeding, with a marginal 10-year overall survival rate of 42% (no PVT, 61%; P = 0.002 and PVT: physiological, 55%; P = 0.043). The PVT: physiological and no PVT groups exhibited comparable survival rates (P = 0.13). No significant differences in survival were observed between complete and partial PVT as long as physiological portal flow was reestablished. Conclusions: The subset of PVT patients requiring nonphysiological portal vein reconstruction was associated with higher complication rates and suffered diminished long-term prognoses. For the most severe PVT cases, a comprehensive approach is critical to further improve outcomes. Keywords: cavoportal hemitransposition, liver transplantation, mesoportal jump graft, portal vein thrombosis, renoportal anastomosis (Ann Surg 2014;259:760–766)
P
ortal vein thrombosis (PVT) in liver transplantation remains an intricate problem in patients requiring liver transplantation. Although it is no longer an absolute contraindication because of From the ∗ Miami Transplant Institute, University of Miami and Jackson Memorial Hospital, Miami, FL; †DeWitt Daughtry Family Department of Surgery, University of Miami Leonard M. Miller School of Medicine, Miami, FL; ‡Department of Pathology, University of Miami Leonard M. Miller School of Medicine Miami, FL; §Transplant Center, Carolinas Medical Center, Charlotte, NC; and ¶Transplant Center, Cleveland Clinic Florida, Weston, FL. Disclosure: There is no funding to declare. The authors declare no conflicts of interest. Reprints: Andreas G. Tzakis, MD, PhD, Transplant Center, Cleveland Clinic Florida, 2950 Cleveland Clinic Blvd, Weston, FL 33331. E-mail: tzakisa@ ccf.org. C 2013 by Lippincott Williams & Wilkins Copyright ISSN: 0003-4932/13/25904-0760 DOI: 10.1097/SLA.0000000000000252
760 | www.annalsofsurgery.com
important technical breakthroughs,1–6 surgical strategies depend on the extent of PVT.7 Complete PVT is generally associated with an increased risk of complications and a poor prognosis7–9 ; moreover, many transplant centers believe that patients with diffuse PVT cannot undergo transplantation.10 Only a few studies of more than 5 cases have reported the longterm outcomes of cavoportal hemitranspositions, renoportal anastomoses, or mesoportal jump graft placement for PVT.11–14 The 5-year survival rate of 23 PVT patients who underwent cavoportal hemitransposition at our center was 38%.11 Likewise, a French study group demonstrated a survival rate of 60% after 5 years in 20 recipients with caval inflow to the graft (3 underwent cavoportal hemitranspositions and 17 renoportal anastomoses).12 Nonetheless, their prognoses were not compared with those of patients undergoing other portal vein reconstruction techniques. Furthermore, both data sets included cases of retransplantation and/or combined organ transplantation. An Italian study group reported an excellent 5-year survival rate of 68% in 41 recipients with complete PVT (6 were managed with mesoportal jump grafts and 6 with cavoportal hemitranspositions), which was comparable with that of 51 patients with partial thrombosis; however, this article also included retransplantation cases and combined organ transplantation cases.13 Nikitin et al reported that portal vein conduits had no negative impact on long-term patient and graft survival, but precise information regarding PVT was lacking.14 To accurately delineate the sequelae of surgical innovations over the past 2 decades, the present study focused exclusively on adult, primary isolated liver transplantation. For the first time, the impact of PVT on postoperative comorbidities and long-term survival (up to 14 years) was analyzed on the basis of the type of portal inflow.
METHODS We retrospectively analyzed all adult (≥18 years) patients who underwent primary isolated liver transplantation from 1998 to 2009 at the University of Miami/Jackson Memorial Hospital (Miami, FL). This study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was conducted with approval from the Institutional Review Board of the University of Miami. The collected and analyzed data included recipient/donor demographics, use of partial/split grafts, intraoperative data, the extent of PVT confirmed at the time of transplantation [ie, partial (Yerdel grades 1 and 2) or complete (grades 3 and 4)7 ], the portal vein reconstruction technique, posttransplant complications, retransplantation information, causes of graft loss/patient death, and survival. Orthotopic liver transplantation was performed using standard techniques, with the recipient’s inferior vena cava preserved in a vast majority of cases.15,16 Our first choice for surgical intervention for PVT was eversion thrombectomy and end-to-end anastomoses.17,18 If the thrombus could not be extracted but could be resected with the thrombosed portal vein, and if anastomosis was feasible between the recipient splenomesenteric confluence and the donor portal vein, interpositional vein grafting was our second preference. If the thrombus Annals of Surgery r Volume 259, Number 4, April 2014
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 259, Number 4, April 2014
extended below the splenomesenteric confluence but the proximal superior mesenteric vein was patent, mesoportal jump grafting (extraanatomical) was selected as the third option.1 If these techniques were not readily achievable, then we looked for any large portal venous tributaries or dilated collateral veins that could be utilized. Only after exhausting all these measures did we consider complex reconstruction involving renoportal anastomosis with interpositional vein grafting3,5 in patients with a patent surgical or spontaneous splenorenal shunt. Cavoportal hemitransposition was considered as the last option.2 Portal vein arterialization6 was also a reasonable alternative under rare circumstances; for example, when a patient had an aberrant hepatic artery that was not used for arterial anastomosis. Consequently, the patients with PVT were divided into 2 groups: a “PVT: physiological” group, in which portomesenteric venous circulation was physiologically reestablished, and a “PVT: nonphysiological” group, in which portal blood flow to the liver graft could not be reinstituted, that is, cavoportal hemitransposition, renoportal anastomosis, or portal vein arterialization. Recipients with no intraoperative evidence of PVT were classified as a third group termed the “no PVT” group. Arterial revascularization of the liver allograft was achieved using standard techniques.19 Patient background information, postoperative complication profiles, and patient and graft survival rates were compared among the 3 groups. The patency of vascular anastomoses was assessed by serial intra- and postoperative Doppler ultrasonography.20 If there was clinical suspicion of compromised arterial or portal flow, either computed tomography (CT) or formal angiography was performed for further investigation. A diagnosis of hepatic arterial or complete portal vein (re-) thrombi was determined by the absence of hepatic arterial enhancement or portal flow on the angiogram or identification of complete occlusion of the artery or portal vein during surgical exploration. Hepatic artery stenosis (≥50% vessel narrowing) or partial portal vein (re-) thrombosis was confirmed by CT or angiography. Biliary complications included leaks, strictures (anastomotic and nonanastomotic), and others (sludges/stones, casts, and bilomas). Documented bleeding from varices on endoscopy was defined as “gastrointestinal bleeding,” whereas postoperative hemorrhage necessitating surgical reexploration was labeled as “surgical bleeding.” Patients requiring temporary renal replacement therapy or those on postoperative mechanical ventilation for more than 1 week were designated as “hemodialysis” or “prolonged ventilation,” respectively. Kidney function was assessed by the estimated glomerular filtration rate using the 4-variable Modification of Diet in Renal Disease Study equation.21 The 10-year cardiovascular risk of each patient who died of cardiopulmonary disease in the entire cohort was assessed by the Framingham Risk Score system and further categorized into 3 groups: low, 10% or less; moderate, more than 10%–20%; and high, more than 20%.22–24 In the PVT: nonphysiological group, anticoagulation with intravenous heparin infusion was initiated within 24 hours after transplantation, followed by long-term anticoagulation with warfarin. In addition, aspirin was administered as a prophylactic drug if the risk of hepatic arterial thrombosis was anticipated. The mainstay of immunosuppression maintenance was tacrolimus (Prograf; Astellas Pharma US, Northbrook, IL) as part of a dual- or triple-agent regimen with prednisone and mycophenolate mofetil (CellCept; Genetech, San Francisco, CA).
Statistical Analyses Patients were censored if they were alive at the time of the last follow-up. A graft was considered to be lost when a patient died or underwent retransplantation. Survival rates were calculated as the interval between the time of transplantation and patient death or graft loss. Unadjusted survival rates were estimated using the C 2013 Lippincott Williams & Wilkins
Portal Vein Thrombosis in Liver Transplantation
Kaplan-Meier method, and the log-rank test was applied to compare survival distributions. To identify baseline recipient/donor characteristics associated with decreased patient and graft survival in the entire cohort, univariate log-rank analysis was applied. The following recipient variables were included: age, gender, race, malignancy in the explanted liver, hepatitis C positivity, and placement of an aortohepatic conduit. Moreover, the following donor variables were included: age, gender, donor risk index25 and cold and warm ischemia times. Factors that emerged in the entire cohort with a P value of less than 0.20 were considered as significant baseline covariates for survival and were adjusted for comparison between the PVT and no PVT groups using Cox regression analysis. The χ 2 test or Fisher exact probability test was used to compare categorical variables with Bonferroni corrections for multiple comparisons. For continuous variables that did not follow a normal distribution, the Kruskal-Wallis test was conducted to compare independent samples and the Steel-Dwass procedure was used to assess differences between each pair of groups. Data are presented as median values (interquartile range) or numbers (%). A P value of less than 0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics 20 statistical software (IBM Corp., Armonk, NY).
RESULTS Patient Demographics From January 1998 to December 2009, a total of 1379 adult primary liver transplantations were performed at our institution. Patients were divided into 3 groups: no PVT (n = 1205), PVT: physiological (n = 149), and PVT: nonphysiological (n = 25). Recipient and donor background data are outlined in Table 1. The PVT: nonphysiological group more often presented with a diagnosis of cryptogenic cirrhosis compared with the other groups and that of Budd-Chiari syndrome compared with the no PVT group. Cold and warm ischemia times and estimated blood loss were greater, and hepaticojejunostomy was more frequently employed in the PVT: physiological group than in the no PVT group. In addition, the estimated blood loss was higher in the PVT: Nonphysiological group than in the no PVT group, and the former group patients were more likely to undergo conventional caval anastomosis, placement of an aortohepatic conduit, or hepaticojejunostomy. The other variables were comparable among the 3 groups. The extent of PVT and the portal reconstruction methods used in each group are listed in Table 2. Complete thrombi were observed in 83 of 174 patients with PVT (47.7%). In the PVT: physiological group, a majority of PVT cases was managed via thrombectomy and end-to-end anastomosis (a dilated recipient left coronary vein was used in 1 patient). Short interpositional vein grafts were placed in 16 patients, and 10 patients required an extra-anatomical mesoportal jump graft. In the remaining 25 patients (partial PVT, 3; complete PVT, 22), decompression of the portomesenteric system was unsuccessful (PVT: nonphysiological group). Among these patients, the decision to perform nonphysiological portal vein reconstruction was made prior to surgery in 19 patients (76%) (cavoportal hemitransposition, n = 13, and renoportal anastomosis, n = 6). Of the remaining 6 patients who were found to have extensive portomesenteric thrombosis at the time of transplantation, 5 underwent cavoportal hemitransposition. In the remaining patient, thrombectomy and end-to-end anastomosis of the portal vein did not restore adequate portal inflow; therefore, we decided to connect the recipient replaced right hepatic artery, which was not used for arterial anastomosis, end-to-side to the portal vein (portal vein arterialization). www.annalsofsurgery.com | 761
Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 259, Number 4, April 2014
Hibi et al
TABLE 1. Patient Characteristics PVT No PVT (n = 1205)
Factor Age, yr Female sex Model for end-stage liver disease Indication for transplantation Hepatocellular disease Hepatitis C Alcohol Fulminant hepatic failure Cryptogenic Cholestatic liver disease Metabolic liver disease Budd-Chiari syndrome Other Hepatocellular carcinoma Donor age, yr Donor risk index Operation Cold ischemia time, min Warm ischemia time, min Estimated blood loss, L Caval anastomosis Piggyback or its variant Conventional Aortohepatic conduit Bile duct reconstruction Hepaticojejunostomy Duct to duct Others
Physiological (n = 149)
53 (46–60) 416 (34.5) 21 (16–24)
55 (49–62) 45 (30.2) 20 (16–24)
Nonphysiological (n = 25) 54 (47–60) 9 (36) 18 (15–26)
584 (48.5) 170 (14.1) 61 (5.1) 77 (6.4) 118 (9.8) 45 (3.7) 3 (0.2) 147 (12.2) 266 (22.1) 43 (26–56) 1.72 (1.51–2.09)
71 (47.7) 20 (13.4) 0 (0)∗ 13 (8.7) 5 (3.4)∗ 8 (5.4) 2 (1.3) 30 (20.1)∗ 37 (24.8) 41 (25–53) 1.71 (1.49–2.05)
7 (28) 3 (12) 0 (0) 7 (28)†‡ 0 (0) 2 (8) 2 (8)† 4 (16) 4 (16) 46 (32–58) 1.72 (1.52–2.08)
416 (347–490) 36 (31–42) 5.0 (2.5–9.0)
437 (381–511)∗ 39 (33–47)∗ 7.0 (3.5–11)∗
431 (373–501) 32 (27–38)‡ 10 (5–21)†
1,102 (91.5) 103 (8.5) 220 (18.3) 679 (56.3) 525 (43.6) 1 (0.1)
134 (89.9) 15 (10.1) 37 (24.8) 98 (65.8)∗ 49 (32.9)∗ 2 (1.3)∗
TABLE 2. Extent of Portal Vein Thrombosis and Portal Reconstruction Methods Reconstruction
Partial (n = 91)
Complete (n = 83)
Total (n = 174)
Physiological portal inflow End-to-end anastomosis Interpositional vein graft Mesoportal jump graft
79 (87) 6 (7) 3 (3)
44 (53) 10 (12) 7 (8)
123 (70.7) 16 (9.2) 10 (5.7)
88 (97)
61 (73)
149 (85.6)
1 (1) 2 (2) 0 (0)
17 (21) 4 (5) 1 (1)
18 (10.3) 6 (3.4) 1 (0.6)
3 (3)
22 (27)
25 (14.4)
Subtotal Nonphysiological portal inflow Cavoportal hemitransposition Renoportal anastomosis Portal vein arterialization Subtotal
Data presented as numbers (%) for categorical variables.
Postoperative Complications The PVT: nonphysiological group exhibited the longest median hospital stay of 24 (12–51) days [no PVT group, 12 (8–18) days and PVT: physiological group, 13 (9–24) days, P < 0.001] (Table 3) and the highest in-hospital mortality rate at the time of transplantation of 24% (no PVT, 5.1% and PVT: physiological group, 3.4%; P < 0.001). A total of 34 patients suffered portomesenteric (re-) thrombosis after transplantation. The PVT: nonphysiological group patients 762 | www.annalsofsurgery.com
0.13 0.98 0.003 0.001 0.006 0.23 0.001 0.022 0.61 0.53 0.97 0.011