Usability of ringed polytetrafluoroethylene grafts for middle hepatic vein reconstruction during living donor liver transplantation§


  • Shin Hwang,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Dong-Hwan Jung,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Tae-Yong Ha,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Chul-Soo Ahn,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Deok-Bog Moon,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Ki-Hun Kim,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Gi-Won Song,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Gil-Chun Park,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Sung-Won Jung,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Sam-Youl Yoon,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Jung-Man Namgoong,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Chun-Soo Park,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Yo-Han Park,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Hyeong-Woo Park,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Hyo-Jun Lee,

    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Sung-Gyu Lee

    Corresponding author
    1. Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea
    • Division of Liver Transplantation and Hepatobiliary Surgery, Department of Surgery, Asan Medical Center, University of Ulsan College of Medicine, 388-1 Poongnap-Dong, Songpa-Gu, Seoul, Korea 138-736
    Search for more papers by this author
    • Telephone: 82-2-3010-3485; FAX: 82-2-474-9027

  • This study was registered at (NCT01205802). It was terminated in advance because the final goal was met early during the study period.

  • This study was supported by research funds from the Asan Institute of Life Science (grant 2007-246 to Shin Hwang).

  • §

    These authors contributed equally to this work.


Large vein allografts are suitable for middle hepatic vein (MHV) reconstruction, but their supply is often limited. Although polytetrafluoroethylene (PTFE) grafts are unlimitedly available, their long-term patency is relatively poor. We intended to enhance the clinical usability of PTFE grafts for MHV reconstruction during living donor liver transplantation (LDLT). Two sequential studies were performed. First, PTFE grafts were implanted as inferior vena cava replacements into dogs. Second, in a 1-year prospective clinical trial of 262 adults undergoing LDLT with a modified right lobe, MHV reconstruction with PTFE grafts was compared with other types of reconstruction, and the outcomes were evaluated. In the animal study, PTFE grafts induced strong inflammatory reactions and luminal thrombus formation, but the endothelial lining was well developed. In the clinical study, the reconstruction techniques were revised to make a composite PTFE graft with an artery patch on the basis of the results of the animal study. MHVs were reconstructed with cryopreserved iliac veins (n = 122), iliac arteries (n = 43), aortas (n = 13), and PTFE (n = 84), and these reconstructions yielded 6-month patency rates of 75.3%, 35.2%, 92.3%, and 76.6%, respectively. The overall 6-month patency rates for the iliac vein and PTFE grafts were similar (P = 0.92), but the 6-month patency rates with vein segment 5 were 51.0% and 34.7%, respectively (P = 0.001). The overall graft and patient survival rates did not differ among these 4 groups. In conclusion, ringed PTFE grafts combined with small vessel patches showed high patency rates comparable to those of iliac vein grafts; thus, they can be used for MHV reconstruction when other sizable vessel allografts are not available. Liver Transpl, 2012. © 2012 AASLD.

Middle hepatic vein (MHV) reconstruction with an interposition vessel graft has been established as a standard procedure for living donor liver transplantation (LDLT) with a right lobe graft when the donor's MHV trunk is preserved in the donor's remnant liver. Materials used to date for this type of venous vascular reconstruction have included various types of homologous and autologous vessel grafts.1-5 The recent increase in the number of adult LDLT procedures and the relatively limited number of vessel allografts have led us to search for new vessel substitutes.6 Although thin-walled, expanded polytetrafluoroethylene (PTFE) grafts have been used for MHV reconstruction, their patency rates appear to be insufficiently low.7 Using PTFE grafts for MHV reconstruction in a small number of patients undergoing LDLT and nontransplant surgery, however, we have found that although their long-term patency is much lower than that of allografts, the late occlusion of PTFE grafts is not significantly associated with patient survival outcomes.

Because of the advantage of prosthetic vessel grafts (ie, their unlimited availability), we performed 2 sequential studies to assess the clinical usability of ringed, expanded PTFE grafts for MHV reconstruction during LDLT. The first study was performed with a dog model to determine what occurs at these expanded PTFE grafts after implantation. The second was a prospective clinical trial designed to determine whether these grafts can maintain luminal flow for a sufficiently long period of time.


CT, computed tomography; GRWR, graft-to-recipient weight ratio; IVC, inferior vena cava; LDLT, living donor liver transplantation; MELD, Model for End-Stage Liver Disease; MHV, middle hepatic vein; PTFE, polytetrafluoroethylene; V5, hepatic vein branch segment 5; V8, hepatic vein branch segment 8.


Materials and Methods

The animal experiment involved the replacement of the infrarenal inferior vena cava (IVC) through the interposition of a cold-preserved IVC allograft or an expanded PTFE graft, 4 weeks of waiting, and gross and histological examinations of the vessel grafts after harvesting. This study protocol was approved by our institutional ethics board for animal experimentation.

We used 11 male dogs weighing 25 to 30 kg each. After laparotomy under general anesthesia, the infrarenal IVC was isolated by the ligation of multiple lumbar veins. After the clamping of both ends of the infrarenal IVC, a 5- to 6-cm-long segment was excised from each. This segmental defect was replaced by an IVC allograft in 3 dogs and by an expanded PTFE graft in 5 dogs. The IVC allografts had been harvested 1 week earlier from 3 other dogs and had been preserved in a cold histidine-tryptophan-ketoglutarate solution (Alsbach-Hähnlein Co., Germany). The size of the IVC graft was matched to each native IVC. Each PTFE graft was made of thin-walled, expanded PTFE with a 14-mm inner diameter and a 15-mm outer diameter (Gore-Tex, WL Gore & Associates, Inc.). The anastomosis was made with continuous running sutures (6-0 monofilaments). None of these animals received antiplatelet or anticoagulation therapy, and none experienced any serious surgical complications. After 4 weeks, the dogs were sacrificed, and their IVCs, including the interposed grafts and anastomoses, were harvested.

The luminal patency of each graft portion was evaluated with Doppler ultrasonography during the heart-beating state, and a morphometric assessment was performed after IVC harvesting. The mean outer and inner diameters were calculated after multiple measurements along the longitudinal lengths of the grafts. Tissue samples were fixed in a 10% formalin solution and were histologically evaluated after routine hematoxylin-eosin staining, Masson trichrome staining for collagen and muscle fibers, and Verhoeff–Van Gieson staining for elastic fibers. The tissue samples were also assessed immunohistochemically by incubation with a mouse antibody to proliferating cell nuclear antigen (NeoMarkers, Fremont, CA), which is a marker for cellular proliferation; with a mouse antibody to α-actin (Dako, Carpinteria, CA), which is a marker for smooth muscle cells within the vessel; and with a rabbit antibody to von Willebrand factor (Dako), which is a marker for vascular endothelial cells. Median values were determined, and the groups were compared statistically with Mann-Whitney U tests. A P value < 0.05 was regarded as statistically significant.


Blood flow through the IVC grafts was maintained in all 8 dogs, and this resulted in a 1-month patency rate of 100%. Anastomotic and luminal stenoses were greater in the PTFE group versus the allograft group (Fig. 1). More than half of each PTFE graft lumen was filled with thrombus. Morphometric changes in the graft diameters and histological findings are summarized in Table 1. More severe inflammation was observed in the PTFE group. Immunohistochemical staining showed scant cellular proliferation at allografts but more active replication at PTFE grafts. The endothelial lining at the luminal surface was well developed in both groups.

Figure 1.

Gross and microscopic findings of the IVC replacement grafts. (A) Implantation of the interposition grafts. (B) Harvesting of the shrunken grafts after 4 weeks. (C) Gross views of the harvested grafts after longitudinal and transverse cutting. The images show the formation of a thickened luminal thrombus at the lumen of the PTFE graft. (D) Microscopic intimal findings. The left images in each column, which were taken after Masson trichrome staining, reveal the differences in the thickness of the intima around the anastomotic site; the right images, which were taken after staining with von Willebrand factor antibodies, show integrated endothelial cell lining at the graft luminal surface. Green bars represent 0.1 mm; the magnification ratios were ×100 and ×25.

Table 1. Morphometric Changes and Histological Findings at the Interposed Grafts
 IVC Allograft Group (n = 3)PTFE Graft Group (n = 5)P Value
  • *

    Subendothelial fibrous hyperplasia.

Morphometric changes   
 Median outer diameter (mm)   
  At implantation15.815.00.41
  After 4 weeks8.413.80.001
 Median inner luminal diameter (mm)   
  At implantation13.314.00.31
  After 4 weeks7.96.30.61
Histological findings   
 At anastomotic site   
  Wall thickeningModerateSevere* 
  Inflammatory cell infiltrationModerateSevere 
  Intimal thickeningMildModerate 
 At graft lumen   
  Luminal thrombusAbsentPresent 
  Intimal thickeningMildMild 
  Endothelial liningIntegratedIntegrated 


Patients and Methods

Conclusions from the Preceding Animal Study

The results of the animal study led to 3 important conclusions. First, the PTFE grafts induced strong inflammatory tissue reactions at the anastomotic sites, and this may have led to early severe narrowing of the anastomosis. Because this type of anastomotic stenosis appears to be inevitable with PTFE grafts, a wider anastomosis would likely alleviate the stenosis-inducing effects of severe inflammation. Second, progressive luminal thrombus formation led to early narrowing of the effective lumen, although the endothelial lining was well formed. Thrombus formation appears to be alleviated by the application of sufficient antiplatelet and anticoagulation therapy. The effective lumen may be enlarged with a larger caliber PTFE graft, although this may increase the risk of thrombus formation because of the diameter-dependent decrease in the flow velocity. Third, expanded PTFE grafts are flexible and are, therefore, vulnerable to buckling, compression, and even shrinkage. The area of MHV reconstruction at the cut surface of the graft liver is usually compressed by the stomach and by fluid collection or a hematoma. Ringed PTFE grafts appear more resistant to extrinsic forces than the usual PTFE grafts without outer rings.

Study Design

Since 2009, the relative supply of vessel allografts at our institution has decreased because of gradual increases in the demand for LDLT.6 We have, therefore, decided to routinely use a prosthetic vascular graft when an adequate vessel allograft is unavailable.

Before we made the decision to routinely use PTFE grafts, we prudently performed a 3-month preliminary study to develop effective surgical techniques, which were integrated with the 3 conclusions from the preceding animal study. The early outcomes of the first 8 patients were satisfactory and included no occurrences of MHV stenosis, so we thought the use of PTFE grafts should be applied to more patients through a prospective clinical trial.

The primary goal of this clinical study was to compare the short- and mid-term luminal patency of PTFE grafts and cryopreserved iliac vein allografts, with the latter grafts usually having patency rates > 90% at 1 month and > 60% at 6 months. Because MHV occlusion after 6 months was usually not associated with graft failure or patient death, the minimal follow-up period was set at 6 months.

To assess the clinical usability of PTFE grafts for MHV reconstruction, we performed a prospective clinical trial for 12 months from March 2010 to February 2011; follow-up continued until the end of August 2011. This study protocol was approved by our institutional review board for clinical trials.

Indication for MHV Reconstruction

Since we raised the concept of MHV reconstruction in 1997, we have tried to reconstruct most sizable MHV branches (≥5 mm in diameter). The indications for MHV reconstruction have been described in detail previously.2, 4, 6

Selection of the Graft Materials for MHV Reconstruction

After parenchymal transection of the donor liver, the size and shape of a vessel allograft suitable for MHV reconstruction were determined. On the basis of a list of all available allografts stored in our institutional tissue bank, we decided whether or not to use a PTFE graft in a particular patient. Each LDLT recipient consented to the potential use of either a PTFE graft or an allograft during the transplantation of a right lobe graft.

The graft material was selected primarily by the graft-to-recipient weight ratio (GRWR) and the Model for End-Stage Liver Disease (MELD) score. Because we did not know the actual risk of PTFE grafts at the time of the study design, we intended to avoid their use in high-risk patients (eg, patients with low GRWRs or high MELD scores). Iliac vein allografts were preferentially chosen when the GRWR was close to 0.8 or the MELD score was >30, whereas PTFE grafts were chosen when the GRWR was >1.0 and the MELD score was <20. Aorta grafts and large iliac artery grafts were also used according to their availability.

Surgical Techniques

The guidelines for the use of PTFE grafts in MHV reconstruction were based on the 3 conclusions of our preceding animal study and our clinical experience with LDLT. Detailed surgical techniques were devised and refined in the preliminary clinical study.

We used only ringed PTFE grafts (Gore-Tex) of larger calibers (most had an internal diameter of 10 mm, but for a few, the diameter was 12 mm). After we inserted a niche to enlarge the orifices of hepatic vein branch segments 5 (V5) and 8 (V8), we applied an intervening allograft patch for an end-to-side anastomosis of MHV branches. This was followed by a large anastomosis at the recipient-side stump of the middle-left hepatic vein trunk. The surgical techniques for using PTFE grafts are illustrated in detail in Figs. 2 and 3.

Figure 2.

Intraoperative photographs showing the standardized techniques for MHV reconstruction with a composite graft of ringed PTFE and a cryopreserved iliac artery patch. (A) A hepatic vein branch orifice at the cut surface of the liver was widened by a ventral cut, and then an arterial patch was sutured (inset). (B) A 10-mm ringed PTFE graft was prepared, and holes (dotted lines) were excised to match the size of the arterial patches. (C) An end-to-side anastomosis was made between the PTFE graft and the arterial patch; a funnel-shaped intervening arterial patch can be observed (inset). A leakage test was performed after the closure of the distal end with a large Hem-o-lok clip. (D) The anastomosis of the PTFE graft to the middle-left hepatic vein stump with an end-to-side method (dotted, ovoid line) was followed by the closing of the proximal end with a large Hem-o-lok clip.

Figure 3.

Modified techniques for MHV reconstruction with composite grafts of ringed PTFE and a cryopreserved iliac artery. (A,B) When 2 vein branches were adjacent, the unification of 2 small patches (white arrows) enabled a single long anastomosis. (A-C) An end-to-end anastomosis was made for a large V5 branch; (D) an end-to-side anastomosis could make the anastomotic orifice larger than an end-to-end anastomosis (dotted arrow). (C,D) A recipient-side anastomosis also could be made in an end-to-end fashion. (A-C) A 10-mm PTFE graft or (D) a 12-mm graft was used.

Small to medium cryopreserved iliac artery grafts were concurrently used as composite patch materials because a large number of these artery grafts were stored in our tissue bank. These grafts had been deemed inadequate for use as independent vascular conduits during MHV reconstruction.

The technical details for securing anastomoses to PTFE grafts included (1) maintaining the rings attached to the side anastomosis sites, (2) using a PTFE suture material (Gore-Tex Suture; ie, a nonabsorbable, white monofilament made of expanded PTFE) that enabled a 1:1 needle-to-thread ratio to minimize needle hole bleeding at anastomoses (we used 6-0 threads with 9-mm double-arm needles), (3) making a redundant composite patch plasty for an end-to-side branch anastomosis (especially for V8), and (4) spraying a fibrin glue over the PTFE graft to control minute bleeds at suture points and to stably fix the PTFE graft at the cut surface of the liver.

After they were maintained in a hypocoagulable state for 1 to 2 weeks, nearly all adult LDLT recipients were routinely administered antiplatelet therapy [aspirin with or without venlafaxine hydrochloride (Plavix)] for 3 to 6 months to protect the hepatic artery anastomosis. Warfarin administration was indicated only for previous coronary/cardiac procedures or posttransplant hepatic artery insufficiency and not to improve MHV patency.

Evaluation of the Graft MHV Patency and Indications for Stenting

Our LDLT management protocol included Doppler ultrasonography for monitoring graft inflow and outflow (including the interposed MHV graft); this was performed intraoperatively, once per day during the first week after surgery, and once per week until hospital discharge. Posttransplant dynamic computed tomography (CT) scans were routinely performed every week while the patient was in the hospital and 1, 3, 6, and 12 months after LDLT. CT scans were performed more frequently for recipients who had hepatocellular carcinoma (for cancer surveillance), especially during the first year after transplantation (usually every 3 months).8

In this study, occlusion was defined as nonvisualization of blood flow in the graft conduit between V8 (or V5 when only V5 was reconstructed) and the IVC on the venous phase CT liver scan. A combination of a thrombosed V5 and a patent V8 was regarded as patent. If there was progressive luminal narrowing with only faint contrast enhancement, the conduit was considered not patent because it would be occluded within a few weeks and its blood flow volume would be very small. Because CT scans were not performed monthly, the median time between CT scans showing visualization and nonvisualization of blood flow was used to calculate the timing of occlusion. When only pre-enhancement CT images were taken because of impaired renal function, information from Doppler ultrasonography was used instead.

Patients with definitely noticeable MHV stenosis or occlusion during the early posttransplant period (≤4 weeks) were indicated for stent insertion in principle, even though their graft liver function was not seriously impaired. Late stenting was indicated only for patients showing significant MHV occlusion–related perfusion abnormalities. MHV stenting was regarded as MHV occlusion, although MHV patency was regained after stenting. Zilver stents (Cook, Bloomington, IN) were inserted across the MHV anastomosis through the internal jugular vein.9, 10


All numerical data are presented as means and standard deviations with ranges. For 2-group comparisons, the chi-square test and Fisher's exact test were used for incidence variables, and Student t tests were used for continuous variables. For comparisons of 4 groups, Kruskal-Wallis analyses of variance and median tests were used. Survival and patency rates were determined with the Kaplan-Meier method and were compared with the log-rank test. Statistical significance was set at P < 0.05.


Patient Profiles

During the 1-year study period, 262 patients who underwent LDLT with a modified right lobe graft required MHV reconstruction with an interposition vessel graft. The mean patient age was 51.9 ± 8.2 years, and the male-to-female ratio was 193/69. The clinical profiles of these patients are summarized in Table 2. Five patients required retransplantation with deceased donor liver grafts by the end of the follow-up period. The overall graft and patient survival rates were 93.1% and 94.3%, respectively, at 6 months and 91.6% and 93.5%, respectively, at 12 months. The overall graft (P = 0.56) and patient survival rates (P = 0.75) were similar in the 4 groups of patients using different vessel grafts.

Table 2. Clinical Profiles of the 262 Patients According to the Types of Interposition Grafts
 Iliac Vein Group (n = 122)Iliac Artery Group (n = 43)Aorta Group (n = 13)PTFE Group (n = 84)P Value
All GroupsIliac Vein Group Versus PTFE Group
  • *

    Hepatitis B virus patients versus non–hepatitis B virus patients.

Recipient age (years)50.2 ± 9.352.3 ± 9.154.2 ± 6.553.7 ± 5.30.0380.002
Recipient sex: male/female (n/n)86/3640/33/1064/200.0690.37
Primary diagnosis (n)      
 Hepatitis B virus–associated liver cirrhosis73288600.39*0.087*
 Hepatitis C virus–associated liver cirrhosis10523  
 Alcoholic liver disease184110  
 Acute liver failure11404  
Concurrent hepatocellular carcinoma [n (%)]50 (41.0)21 (48.8)6 (46.2)45 (53.6)0.350.075
MELD score18.1 ± 10.319.3 ± 10.218.0 ± 10.515.9 ± 8.20.320.042
Emergency transplantation [n (%)]12 (9.8)4 (9.3)1 (7.7)5 (6.0)>0.990.32
Donor age (years)27.1 ± 7.727.6 ± 7.328.3 ± 5.127.6 ± 7.70.420.64
Donor sex: male/female (n/n)87:3532:114:966:180.680.24
ABO incompatibility [n (%)]11 (9.0)2 (4.7)1 (7.7)11 (13.1)0.470.35
GRWR1.04 ± 0.31.02 ± 0.31.09 ± 0.21.11 ± 0.30.0190.025
6-month MHV graft patency rate (%)75.335.292.376.6<0.0010.92
1-year MHV graft patency rate (%)<0.0010.92
6-month liver graft patency rate (%)92.688.410095.20.560.51
6-month patient survival rate (%)93.490.710095.20.750.37

Configurations for MHV Reconstruction

Graft materials for MHV reconstruction in the 262 patients included cryopreserved iliac veins [n = 122 (46.6%)], cryopreserved iliac arteries [n = 43 (16.4%)], cryopreserved aortas [n = 13 (5.0%)], and PTFE with small vessel patches [n = 84 (32.1%)]. The sources of the vessel patches were cryopreserved iliac arteries (n = 75), cryopreserved iliac veins (n = 5), and autologous saphenous and portal veins (n = 4). No autologous vein graft was independently used in this series. The configurations for MHV branch reconstruction are summarized in Table 3. Two hundred twenty-six of the 262 modified right lobes (86.3%) required reconstruction of both V5 and V8. Patch unification of 2 or 3 small branches, which enabled a single anastomosis to the interposition vessel graft, was used in 17 patients; most of these patients were in the PTFE group because a small vessel patch, which facilitated patch unification, was attached to every V5 and V8 opening.

Table 3. Configurations for MHV Reconstruction
 Iliac Vein Group (n = 122)Iliac Artery Group (n = 43)Aorta Group (n = 13)PTFE Group (n = 84)
  1. NOTE: Numbers in parentheses indicate the patch unification of 2 or 3 small branches.

V5 reconstruction (n)    
 Single88 (2)35970 (7)
V8 reconstruction (n)    
 Single10937 (1)1269 (7)
Postoperative stent insertion (n)61101

Comparison of MHV Patency

During follow-up, 18 patients (6.9%) required MHV stenting, with 16 undergoing early stenting (within 2 weeks) and 2 requiring stenting after 3 or 4 months because of a new occurrence of hepatic venous congestion. Eleven of these 18 patients underwent MHV reconstruction with iliac artery grafts (Table 3). After 6 months, there were no episodes of such congestion-associated infarcts, regardless of the MHV patency.

The patency rates of the 4 types of interposed MHV grafts differed significantly (P < 0.001; Fig. 4), but the patency rates of the iliac vein and PTFE grafts did not differ significantly (P = 0.92). Although the overall 6-month MHV patency rates were quite similar (75.3% and 76.6%, respectively; Table 2), the 6-month patency rates for V5 reconstruction were significantly different: 51.0% for iliac vein grafts and 34.7% for PTFE grafts (P = 0.001). These rates implied that the 2 groups had similar risks of occlusion for V8 anastomoses, but the risk for V5 anastomoses was much higher with PTFE grafts.

Figure 4.

Patency of interposed MHV grafts with respect to the materials used for the interposition grafts.

Ten of the 262 patients (3.8%) required prolonged warfarin therapy because of earlier coronary/cardiac procedures or hepatic artery insufficiency. Five of these patients had PTFE grafts, and their 6-month MHV patency rate was only 40%; thus, their outcomes were not superior to the outcomes of patients not receiving anticoagulation therapy.

None of the patients with PTFE grafts experienced any PTFE graft–associated infectious complications.

Patterns of MHV Graft Occlusion

Serial follow-up CT scans showed different patterns of MHV graft occlusion. The short-term patency of the iliac vein grafts was excellent, but their diameter became narrower, especially at the V5 anastomosis, because of decreased outflow and extrinsic compression by the stomach. Over time, V5 was first occluded while the flow of V8 persisted; V8 occlusion occurred later (Fig. 5). Iliac artery grafts were occluded early, mainly because of their relatively small diameter: outflow was maintained for 1 to 3 months after stenting. The diameter of aorta grafts became narrower over time, but their diameter at 1 year remained comparable to the diameter of iliac vein grafts at 1 month (approximately 1 cm), so the resultant patency rate was high.

Figure 5.

CT images comparing the patterns of MHV occlusion. Iliac vein grafts showed excellent early patency, but their diameter became narrower and finally disappeared after total collapse (red arrow). Ringed PTFE grafts showed early partial thrombosis due to decreased outflow, and this finally resulted in the occlusion of the lumen by the thrombus with faint contrast visualization (green arrows).

The PTFE grafts often showed early thrombosis around the V5 anastomosis. V5 outflow was gradually reduced with resultant concentric thickening of the luminal thrombus. At this stage, a PTFE graft with a 10-mm inner diameter was converted into a narrow vessel with an inner diameter of 3 to 5 mm. Finally, the graft lumen between the V5 and V8 orifices was occluded (Fig. 5). In contrast, V8 outflow was maintained for a longer period, even after the complete occlusion of V5 and its intervening segment between the V5 and V8 orifices, probably because of its short pathway to the IVC and its protection from extrinsic compression. None of these ringed PTFE grafts were morphologically collapsed.


Clinical indications for PTFE grafts have been expanding, especially for thoracic and peripheral vascular surgery. To date, most PTFE grafts have been for the reconstruction of the aortic or arterial system, with fewer used for venous drainage systems (which have very different features). The indications for prosthetic vessel grafts for the splanchnic venous system have been especially limited because the blood flow is relatively slow and, is therefore, highly thrombogenic. Therefore, little clinical experience with the use of PTFE grafts in the splanchnic venous system has accumulated.7, 11, 12

MHV reconstruction during LDLT has resulted in new demands for vascular allografts, and the increased number of LDLT procedures has led to relative shortages of vessel allografts, especially in Asian countries in which the number of deceased organ/tissue donors is very limited.6

The need for vascular allograft substitutes prompted the aforementioned animal experiments, in which we assessed thrombosis-related changes at expanded PTFE grafts in the venous environment. Our experimental conditions were set to be more thrombogenic than those usually encountered to make the histopathological changes more evident. We found that the platelet and coagulation profiles were not impaired in the dog model (unlike in human LDLT recipients) with no administration of antiplatelet or anticoagulation therapy.13 The results of this animal study demonstrated that the anastomotic sites of the PTFE grafts became definitely stenotic because of severe subendothelial fibrous hyperplasia and concurrent graft luminal narrowing caused by luminal thrombi.14 These findings may explain the mechanisms underlying the previously reported low patency rate of PTFE grafts used in LDLT (ie, 1- and 4-month patency rates of 80.8% and 38.5%).7 Our previous clinical experience with PTFE grafts in venous reconstruction during LDLT and nontransplant hepatobiliary surgery also has shown the importance of maintaining the outer contour of each PTFE graft against extrinsic forces for luminal patency. These factors were reflected in the study design before we performed the prospective clinical trial.

The hematological condition of the patients undergoing LDLT was less thrombogenic than the condition of our dog model, as shown by the impairment of coagulation profiles, the moderate-to-severe thrombocytopenia, the low viscosity from lowered hematocrits, and the concurrent use of dual antiplatelet agents. The risk of tissue reaction–associated stenosis at the native vessel-to-PTFE interface may, therefore, have been alleviated by the interposition of the arterial patch. The outer rings of the PTFE grafts also acted as effective shields and protected the graft contours against all types of extrinsic compression. Our preceding preliminary study showed promising early outcomes comparable to the results of this clinical trial, so we decided to perform this trial because ringed PTFE grafts have the definite advantage of unlimited availability.

The comparison of patency rates with the different graft materials in this study demonstrated definite differences that might reflect the unique features of each graft material, although this study was definitely limited by the lack of random allocation. When the discussion is confined to allografts only, the larger the vessel diameter is, the higher the patency rate is. We have previously reported that cryopreserved iliac arteries are useful for MHV reconstruction,2 but the patency rate has been lower than expected because a sufficiently large iliac artery (a proximal external iliac artery with an inner diameter ≥ 7 mm) is not commonly available in Korea and the use of a medium-sized artery (a proximal external iliac artery with an inner diameter of approximately 5 mm) is often associated with early stenting. In this study, approximately two-thirds of early stenting procedures were performed in the iliac artery graft group. In contrast, the long-term patency rate of aorta grafts was much higher than expected. Serial follow-up CT scans showed progressive shrinkage of aorta grafts, but after 1 year, their luminal diameter still appeared to be approximately 10 mm. The inner diameter of the 10-mm PTFE grafts was comparable to that of usual-size iliac vein grafts, and interestingly, the early patency rates of these 2 quite different materials were similar. Therefore, the diameter of interposition grafts appears to be an important factor for their patency.

The cryopreserved iliac vein graft was regarded as an ideal interposition material for MHV reconstruction after the consideration of its availability, diameter, and length. It showed a 3-month patency rate > 90%, but the 1-year patency rate decreased to 50%. In contrast, we observed that the native MHV trunks at the cut liver surface were patent after 1 year in most of our patients receiving extended right lobe grafts.4, 15 We wondered why half of iliac vein grafts exposed at the cut liver surface were occluded in 1 year. If we assume that the nature of cryopreserved iliac vein allografts is the same as the nature of native MHVs, 1 major difference is the exposure or lack of exposure at the cut liver surface. The MHV trunk of an extended right lobe graft is partially exposed at the cut liver surface, whereas the interposed MHV graft of a modified right lobe graft is fully exposed; the latter is thus vulnerable to extrinsic compression. We presume that this extrinsic compression will increase MHV outflow resistance and result in the gradual development of intrahepatic collaterals and a subsequent decrease in the outflow volume. After the repetition of these cycles, iliac vein grafts will be slowly occluded in the absence of significant hepatic venous congestion.

In this study, we investigated the underlying mechanisms for the progressive occlusion of the PTFE grafts despite the protection of their shape by the outer rings. Serial follow-up CT scans revealed that a concentric thrombus was formed within each PTFE graft, and this led to a narrow central lumen that was progressively occluded over time. Such luminal thrombus formation is uncommon when prosthetic grafts are used as conduits for large arteries, IVCs, and even portal veins.16-18 These vessels can be classified as high-flow vessels regardless of the pressure, so there is a low risk of luminal thrombus formation. In contrast, MHVs may be low-flow vessels. It is known that neointima is normally formed at the inner graft wall within 2 to 3 weeks and extends 2 to 3 cm from each PTFE anastomosis site.19 If the inner diameter of a PTFE graft is too large to fill with the MHV flow volume, the MHV flow becomes sluggish and thrombogenic, and this leads to progressive luminal narrowing via thrombus formation until the hemodynamic balance is met. During this remodeling process, new endothelial linings over the thrombus repeatedly develop, as shown in the animal model study, but this kind of neointima does not appear to be sufficiently antithrombogenic. Finally, approximately two-thirds of V5 reconstructions and approximately one-fourth of V8 reconstructions with PTFE grafts were occluded after 6 months. We guess that a long and narrow central lumen increases the outflow resistance and reciprocally decreases the MHV outflow volume. The repetition of these cycles finally leads to complete occlusion. The V5 anastomosis has a longer distance from the IVC than the V8 anastomosis, and this longer outflow pathway within the PTFE graft may result in a higher outflow resistance and lead to early occlusion. If a PTFE graft with no outer rings is used, the intraluminal hemodynamic balance will be met with PTFE graft collapse in addition to luminal thrombus formation. At this time, we guess that the progressive shrinkage or collapse of iliac vein grafts is also associated with a hemodynamic imbalance between the vein diameter and the outflow blood volume in addition to extrinsic compression.

The patency of ringed PTFE grafts in this study was definitely superior to the patency in the previous report,7 and the improvement might be due to the following 3 differences: the protective effect of the outer rings against extrinsic compression, the offset of the stenosis-inducing effect from the tissue reaction after the placement of a composite artery patch between the recipient vessel and PTFE graft, and the construction of a streamlined endothelial cell–lined tunnel within the luminal thrombus of the PTFE graft. This tunnel acts like a narrow neointima-lined vessel, and the diameter is adjusted by the passing blood flow volume according to the principles of hemodynamics, including Bernoulli's equation.20, 21

After an interim analysis during this study, we recognized that add-on anticoagulation therapy, in addition to routine antiplatelet therapy for 3 to 6 months, was not effective in improving the patency of PTFE grafts. Once a neointima lining forms within a few weeks, the risk of acute luminal thrombosis may be much reduced. Our findings for the 5 patients who required warfarin therapy indicate that anticoagulation may have no definite beneficial effects on PTFE graft patency. Generally, anticoagulation therapy is indicated for cardiac valve replacement, but it is not recommended after arterial bypass surgery using a PTFE graft. Instead, strong antiplatelet therapy is necessary for all these patients. The results of this study reveal that PTFE grafts have a tendency of gradual occlusion for more than 1 year, and this implies that the period for antiplatelet therapy should be prolonged. After finishing this study, we modified the management protocol to prolong antiplatelet therapy for 1 year (previously 3-6 months) in patients with patent PTFE grafts. There is no reason to use antiplatelet therapy for a prolonged period in patients with MHV allografts or with already occluded PTFE grafts.

Vulnerability to bacterial colonization is one of the most important factors limiting the use of PTFE grafts in contaminated environments (eg, a bile leak at the cut surface of the liver).22, 23 Indeed, PTFE grafts per se may be susceptible to infections because their microporous structure provides an optimal environment for bacteria to survive and proliferate. Once an infection occurs, it can remain a significant problem along with the intractable consequences of bacterial propagation.24, 25 To date, however, none of our patients with PTFE grafts at the liver surface (n > 140) have experienced any episodes of significant infection, and this suggests that these expanded PTFE grafts may be less infection-prone than we thought. We presume that a diffuse spray of fibrin glue over PTFE grafts may effectively protect them against infection by sealing the microporous structure of the grafts against potential bile leaks.26 Moreover, we observed that this sealant remained intact during early re-exploration (<1 month).

To our knowledge, a cryopreserved iliac vein graft is the most suitable material for MHV reconstruction with an excellent early patency rate > 90% and with less effort required for benchwork. In fact, the branching pattern of the iliac vein and its large size are usually adequate for its use in MHV reconstruction. The only and most serious problem is its limited availability. The use of PTFE grafts has the definite advantage of an unlimited supply, and they appear more promising than before because of improved patency through technical refinements. Although PTFE grafts have some nonnegligible disadvantages such as a longer benchwork time and the need for a small vessel patch for a composite graft, most are acceptable or manageable. For us, the bench reconstruction of 1 V5 and 1 V8 with a PTFE graft required 30 minutes on average, whereas an iliac vein graft required 10 to 15 minutes. Because a lot of small iliac artery grafts are stored in our tissue bank, we prefer to use an iliac artery patch. Autologous veins such as greater saphenous veins and portal veins can be good and sometimes better substitutes. In countries in which vessel allografts are not readily available, the use of composite vessel grafts with a PTFE graft and a small autologous vein patch is a reasonable option for MHV reconstruction.

For MHV reconstruction, large autologous veins have been rarely harvested from recipients at our institution. This type of vein harvesting is a traditional method for portal vein replacement during surgery for perihilar cholangiocarcinoma.27 The use of the external iliac vein or the internal jugular vein from the recipient for the reconstruction of V5 is also feasible. At our institution, we prefer cryopreserved homologous vessel grafts or synthetic grafts to autologous vein grafts obtained from recipients. The patency rate of other large autologous veins used for MHV reconstruction (eg, portal veins, dilated paraumbilical veins, and in situ isolated MHVs) is comparable to the rate of cryopreserved iliac veins.28

In conclusion, ringed PTFE grafts combined with small vessel patches showed a high patency rate comparable to that of iliac vein grafts, so we suggest that this method can be reliably used for MHV reconstruction when other sizable vessel allografts are not available.


The authors thank Hyang-Woo Lee, R.N., for helping with the supply of PTFE grafts; Dr. Hea-Nam Hong for performing the animal experiment; and Mi-Kyoung Jeon, R.N., and Ju-Hee Bae, R.N., for maintaining the tissue bank.