Eversion Thrombectomy for Portal Vein Thrombosis During Liver Transplantation



Portal vein thrombosis (PVT) has been seen as an obstacle to orthotopic liver transplantation (OLT), but recent data suggest that favorable results may be achieved in this group of patients. The aim of this study was to analyze the incidence, management, and survival of patients with PVT undergoing primary OLT with thrombectomy. Between October 1990 and August 2000, 468 liver transplantations were performed in our center and portal vein thrombosis was present in 38 patients (8.1%). Preoperative diagnosis, extension, intraoperative management, postoperative recurrence of portal vein thrombosis, and 1-year actuarial survival rates were retrospectively studied. Preoperative diagnosis was made in 17 cases (44.7%). In all patients, portal flow was restored after portal vein thrombectomy, followed by usual end-to-end portal anastomosis. All patients received preventive low-weight heparin from day 2 to hospital discharge, and then aspirin. Rethrombosis was observed in one patient with extended splanchnic thrombus. The 1-year actuarial patient survival rate was 83.7%, and did not significantly differ from the patients without portal vein thrombosis (86.7%). Our results suggest that portal vein thrombosis is often partial and thus difficult to diagnose preoperatively; it can be managed successfully during surgery by thrombectomy, except when there is complete splanchnic veins thrombosis; and it did not affect 1-year survival.


Orthotopic liver transplantation (OLT) is now an accepted and efficient therapy for end-stage liver diseases. Portal vein thrombosis (PVT) is a complication of chronic liver diseases that occurs in approximately 5–15% of these patients (1). At the beginning of OLT history, PVT was considered an absolute contra-indication (2), and is still a relative contra-indication in some centers. Several initial reports indicated a bad prognosis with PVT (3–5). Screening for PVT diagnosis is routinely performed in all candidates for OLT, mainly using Doppler ultrasonography. Despite exhaustive radiological evaluation before OLT, cases of undiagnosed PVT continue to be encountered during surgery. Different approaches have been proposed to restore portal vein patency at the time of OLT, such as thrombophlebectomy, portal vein resection with or without venous graft interposition, portal revascularization from the superior mesenteric vein using venous graft, or cavoportal hemitransposition (6). The aim of this study was to retrospectively review our experience of performing OLT in patients with PVT, in order to evaluate the feasibility of thrombectomy, and to compare 1-year patient survival between patients with or without PVT.

Patients and Methods

Patient population

Records of patients transplanted at the Liver Transplant Unit of the Edouard Herriot Hospital, Lyon, France, were retrospectively retrieved. From October 1990 to August 2000, 468 OLT were performed and 38 patients with intraoperatively confirmed nontumoral PVT formed the study group. There were 31 males and 7 females, representing 8.1% of the whole cohort of liver transplanted patients in our center, with a median age of 50 years (range 35–62). Indications for OLT were: alcoholic cirrhosis (n = 16), viral cirrhosis (n = 17), cryptogenetic cirrhosis (n = 2), autoimmune cirrhosis (n = 1), hemochromatosis (n = 1), and nodular regenerative hyperplasia of the liver (n = 1). An associated hepatocellular carcinoma was present in eight patients; none of the cancer patients had tumor thrombus in the portal vein. The patients were classified as Child-Pugh A, B or C in 13, 8 and 17 cases, respectively. Previous transjugular intrahepatic portosystemic shunt had been performed in two patients, and two patients had a history of surgical splenorenal shunt. These patients were included in the study because of the main repercussion of thrombosis on liver graft implantation. The median waiting time for the patients in the study was 3 months (range 0–27 months).

PVT characteristics

Assessment of liver transplant candidates included both Doppler ultrasonography and CT scan in all cases. Angiography was performed only in cases of suspected complete PVT or portal cavernoma (n = 24). PVT was diagnosed preoperatively in 17 patients (44.7%). Doppler ultrasonography failed to show PVT in 28 cases (73.7%), CT scan in 28 cases (73.7%) and angiography in 19 cases (79.2%). The median delay between the last Doppler ultrasound examination and the transplantation was 2 months.

Four of the 38 PVT were complete. The portal thrombus extended to the splenomesenteric confluence in 11 cases, into the superior mesenteric vein in one case, into the splenic vein in one case and into both the superior mesenteric and splenic veins in one case.

Surgical procedure

Orthotopic liver transplantation was performed using a whole cadaveric liver graft or partial liver graft obtained from a cadaveric split graft. Using the hilar approach, the portal vein was transected with the liver parenchyma once the liver was ready to be removed().

Thrombectomy was done under complete visual control. The hepatic artery and the bile duct were transected first. The portal vein was dissected in its entirety to determine the extent of the thrombus (Figure 1). By maintaining the portal vein open using four tonsil clamps, after having tied its right and left branches, the cleavage plane between the thrombus and the intima was found using the endarteriectomy Bengolea dissector. Then, the clot was progressively and circumferentially freed with the aid of a tonsil clamp by everting the venous wall, by clamping the free edge of the clot with a tonsil. This maneuver was extended to the splenic and/or superior mesenteric veins if necessary. After the clot had been pulled out, portal vacuity was assessed by introduction of the surgeon's index finger or a Hegar dilator. Usually, this technique allowed the entire clot material to be removed. Before completing the anastomosis, the blood flow in the recipient portal vein was verified by removing the clamp. The portal vein was flushed with blood in order to eliminate residual or newly formed clots. Subsequently, portal flow was restored by end-to-end portal anastomosis. Ligation of the collateral circulation, especially spontaneous or surgical splenorenal shunt, was necessary to optimize portal perfusion.

Figure 1.

Different steps of eversion thrombectomy in the case of portal vein thrombosis. The portal vein is transected as close to the liver as possible and the extent of the thrombus in the lumen is determined (A,B). The cleavage plane between the thrombus and the intima is found using the endarteriectomy Bengolea dissector (C,D). Then, the thrombus is progressively freed by everting the venous wall (E,F). Portal vacuity is assessed by introduction of the surgeon's index finger or a Hegar dilator (G,H).

Postoperative data and statistical analysis

Follow-up of our patients consisted of Doppler ultrasound at week 1, week 4, month 3, month 6 and then every year, in addition to CT scan at week 1, week 4, and then every year. The following were recorded after OLT: anticoagulation therapy, median hospital stay (in the intensive care unit and global), recurrence of PVT, early and late postoperative deaths and actuarial patient survival using the Kaplan-Meier method. The 1-year patient survival rate in the group of PVT patients was compared to the 1-year survival rate of the patients in the whole series, using the log-rank test. The difference was considered significant if p < 0.05.


Surgical procedure

Orthotopic liver transplantation was performed using a whole liver graft in 36 cases or using a right split liver lobe from cadaveric donor in two cases. The piggyback technique was used for six patients.

Portal venous thrombosis was total in four cases and partial (20–90% of the lumen) in 34 cases. Eversion thrombectomy was successful in all 38 cases. In one case, it was necessary to extend the thrombectomy to the superior mesenteric and the splenic veins. In three cases, thrombectomy was not total, leading residual mesenteric or splenic thrombus. Ligation of the splenorenal shunt was performed in six cases. Portal flow was restored in all cases.

The median operative time was 6 h (range 3 h 30 min−14 h 30 min). The median transfusion requirement was 8 red blood cell units (range 0–37).

Early and late postoperative period

The median intensive care unit stay was 9 days (range 2–24). The median hospital stay was 14 days (range 9–34). All patients received preventive low-weight heparin (2850 AXa IU/day) from day 2 to hospital discharge, and then aspirin at a dosage of 100–250 mg daily. Heparin was discontinued because of abdominal wall bleeding in two patients.

Early portal vein rethrombosis occurred in the patient who presented with extensive portal, mesenteric and splenic veins thrombus, 6 h after OLT. Despite relaparotomy, the patient died from liver graft infarction at day 2. There was no late diagnosis of PVT recurrence in our population.

Patient survival

Ten patients presenting with PVT died after OLT, with a median delay of 10 months (range 0–72), from recurrence of PVT (n = 1) at day 1, hepatic artery thrombosis (n = 1), intracerebral hemorrhage (n = 1), multiorgan failure (n = 1), cardiac failure (n = 2), recurrence of hepatocellular carcinoma (n = 2), salmonella infection (n = 1), and acute pancreatitis (n = 1).

The 1-year patient survival rates of patients with or without PVT were similar: 83.7% vs. 86.7%.


In historical series, the reported incidence of PVT in cirrhotic patients reached 0.6–64.1%, being highly dependent on the diagnostic methods used (7–9). The screening technique most often used for imaging the portal vein is now Doppler ultrasonography (10,11). In selected patients, referred for OLT, this incidence is lower (Table 1), ranging from 2.1 to 26% (4,5,7,12–18). Abnormalities of the portal vein, especially PVT, were initially considered a contraindication for OLT (4); operative deaths directly related to PVT were reported in the early 1980s (2). Despite surgical progress in the 1980s, the perioperative mortality rates remained high in the presence of PVT, ranging from 9.1 to 42% (4,5,13–15).

Table 1. : Incidence of PVT in liver transplanted patients, surgical approaches and recurrence
AuthorOLTPVTThrombectomyPortal vein resectionVenous bypassPVT recurrence
  1. OLT, orthotopic liver transplantation; PVT, portal vein thrombosis.

Cherqui (13)  6911 (15.9%) 90.9% 0% 9.1%0
Moreno Gonzalez (14) 17414 (8%) 78.7%14.2% 7.1%4 (28.5%)
Seu (16)142370 (4.9%) 87.1% 0.0%12.9%2 (3%)
Lerut (29) 32638 (11.7%) 63.2%10.5%26.3%4 (10.5%)
Yerdel (17) 70963 (8.4%) 30.1%40.6%29.3%3 (5%)
Molmenti (18)154685 (5.5%)100% 0.0% 0.0%2 (2.4%)
This series 46838 (8.1%)100% 0.0% 0.0%1 (2.6%)

Several factors have been found to be significantly associated with PVT occurrence: male sex, spontaneous portosystemic shunts, previous treatment for portal hypertension, including endoscopic therapy, TIPS, shunt surgery, splenectomy, Child-Pugh class C, alcoholic liver disease, repeated episodes of encephalopathy, severe ascites, Budd–Chiari syndrome and hypercoagulable status, and cancer (3,13,17,19–21). These factors reflect: (i) alcoholic disease (ii), the severity of portal hypertension (and the hemodynamic changes in portal vein flow), and (iii) prothrombotic states.

Imaging the portal vein is an important goal of pre-OLT patient evaluation and is usually based on Doppler ultrasonography, which is easily available, inexpensive and noninvasive. Its efficacy in the diagnosis of PVT varies between 26 and 87% (4,5,13–16,22). This agrees with the results of our report, Doppler ultrasonography having diagnosed PVT in only 26.3% of the patients. Extension of the thrombus is clearly a factor in misdiagnosis, leading to underestimation of the incidence of PVT. In addition, false positives with Doppler ultrasonography could be due to a decreased portal flow velocity. It has been hypothesized that PVT could occur in the pre-OLT waiting period. In our series, this is not highly relevant because the median delay between the last radiological imaging and OLT was only 2 months. Regular Doppler ultrasonographic evaluations are highly recommended in the waiting period. The other radiological methods include CT scan and angiography; in our patients these led to the diagnosis of PVT in 26.3% and 20.8% of the cases, respectively. Using a combination of these three techniques, the rate of pre-OLT diagnosis of PVT was 44.7% in our series. Imaging hepatic vasculature using angio-MR could be of interest, as recently reported (17). The deceptive results of diagnostic radiological methods in our patients were probably related to a majority of partial PVT (89.5%).

It is well known that the restoration of normal portal flow is absolutely necessary for good liver graft function (23). At present, there is no consensus on the surgical approach to treating PVT during OLT. Different surgical procedures have been proposed during the last 15 years (Table 1). Resection of the portal vein led to portoportal anastomosis with or without an iliac venous graft obtained from the donor (4). If the thrombus extends to the splenomesenteric confluence, restoration of portal perfusion may be obtained by anastomosing the donor portal vein to the superior mesenteric vein (3,4), to the splenic vein (24), to the left gastric vein (3,25), to choledochal (13,26,27), or mesenteric varices (28), with or without an iliac venous graft. The results of these complex surgical procedures are poor (14,16,29). In some exceptional cases, anastomosis between the donor portal vein and the recipient splanchnic system is not possible. This is the place for cavoportal hemitransposition (30). The preliminary results of this technique were promising, despite the deaths of two out of nine patients. An alternative procedure is to use a spontaneous splenorenal shunt to obtain a portal flow to the liver graft. A combined liver-intestine transplantation has been proposed in the case of complete splanchnic thrombosis (13).

Obstruction of the portal tract probably must be relieved, preferably by eversion thrombectomy, which represents the simplest way to restore portal flow. This might be difficult in the case of old or extensive or complete PVT. In our experience, thrombectomy with good portal flow restoration was possible in 37/38 patients (97.4%). If the thrombosis extends to the superior mesenteric vein or if the portal vein is reduced to a fibrotic vessel remnant, this approach must be ruled out. To our mind, eversion thrombectomy is the method of choice to obtain an optimal portal flow, and other procedures must be restricted to the failure of simple thrombectomy and usual end-to-end portal anastomosis.

It has been reported in the literature that OLT in patients with PVT is associated with a higher rate of complications, such as hepatic artery thrombosis, relaparotomy, pancreatitis, sepsis, and renal failure (17). Moreover, operative time could be longer and transfusion requirements could be higher (3–5,16,17). A higher incidence of primary nonfunction or dysfunction could be related to the more complex surgical procedure in fragile patients with severe portal hypertension (17).

Rethrombosis of the portal vein has been reported in up to 28.5% in the literature (Table 1). It is very interesting that the rate of rethrombosis could directly reflect the failure rate of thrombectomy (Table 1). This was observed in our series and all previously published series (13,14,16,29), except that of Yerdel et al. (17). This suggests that portal vein rethrombosis might be avoided by reducing complex surgical procedures (by performing a complete thrombectomy as often as possible), by ligation of portosystemic shunt during surgery, and by a regular postoperative follow-up using Doppler ultrasonography. As recently reported (17,31), mortality in the case of rethrombosis was 100% in our series. No systematic postoperative anticoagulation is recommended, but aspirin, low-molecular weight heparin, dextran, and coumadin derivatives have been used (14,16). Although the role of these prophylactic measures remains unclear, we used a combination of early low-molecular weight heparin and late aspirin treatment. Even though clinically evident rethrombosis was not documented in our series, some patients could still have developed rethrombosis that the follow-up studies failed to detect.

We observed that PVT did not affect patient survival (83.7% one-year survival rate in patients with PVT vs. 86.7% in patients without PVT), confirming most of the reports in the literature (Table 2). The patient survival rate might be altered in cases of extensive PVT, associated with high rates of rethrombosis and rethrombosis-related deaths (17,31).

Table 2. : Patient one-year survival rate after OLT with or without PVT
AuthorOne-year survival rate
With PVTWithout PVT
  1. OLT, orthotopic liver transplantation; PVT, portal vein thrombosis.

Seu (16)74%84%
Lerut (29)75.5%78.1%
Yerdel (17)65.6%84.2%
Molmenti (18)84.9%84.8%
This series83.7%86.7%

In conclusion, we confirm that PVT is not a contraindication to OLT at the present time. Our results suggest that: (i) the prevalence of preoperative PVT among transplanted patients may be high (8.1% in this series); (ii) PVT is often partial and thus difficult to diagnose preoperatively; (iii) PVT, even when total, can be managed successfully during surgery by thrombectomy, except when there is complete splanchnic veins thrombosis; and (iv) PVT did not affect survival.