Validity of preoperative volumetric analysis of congestion volume in living donor liver transplantation using three-dimensional computed tomography



Reconstruction of middle hepatic vein (MHV) tributaries is controversial in right-lobe living donor liver transplantation (LDLT). This study aimed to evaluate the appropriateness of reconstructing MHV tributaries by volumetry using 3-dimensional computed tomography (3D-CT). Between November 2003 and January 2005, 42 donor livers (right-lobe graft, n = 25; left-lobe graft, n = 17) were evaluated using this software. The total congestion volume (CV) associated with the MHV tributaries and the inferior right hepatic vein (IRHV), and graft volume (GV) were calculated. In recipients with right-lobe grafts, CV/(right liver volume [RLV]) and (GV − CV)/(standard liver volume [SLV]) were compared between 2 groups: with reconstruction (n = 16) and without reconstruction (n = 9). To evaluate the influence of CV on the remnant right lobe in donors, total bilirubin was compared between 2 groups: high CV (CV > 20%, n = 13) or low CV (CV ≤ 20%, n = 4). The mean CV/RLV ratio was 32.3 ± 17.1% (V5, 15.2 ± 9.9%; V8, 9.2 ± 4.1%; and IRHV, 8.5 ± 11.4%) and the maximum ratio was as high as 80.8%. The mean (GV − CV)/SLV ratio before reconstruction in patients with or without reconstruction resulted in 33.5 ± 12.8% and 55.4 ± 12.9%, respectively (P < 0.01). In donors, total bilirubin was significantly high in the high CV group on postoperative day 1 compared with the low CV group (P < 0.05). In conclusion, calculation of CV using 3D-CT software proved to be very useful. We concluded that this evaluation should be an integral part of procedure planning, especially for right-lobe LDLT. (Liver Transpl 2005;11:1556–1562.)

Living donor liver transplantation (LDLT) was developed to overcome the shortage of suitably sized organs from cadaveric donors for children and adults with end-stage liver disease.1 LDLT necessitates a right-lobe graft for adequate liver volume; however, a right-lobe graft without a middle hepatic vein (MHV) potentially has problems of hepatic venous congestion (HVC) caused by deprivation of drainage from the inferior right hepatic vein (IRHV) and MHV tributaries (V5 and V8).2–4 Lee et al.2 reported 2 cases of severe congestion of a graft without MHV; 1 resulted in sepsis due to congestive infarction and the other developed prolonged massive jaundice. We believe that impaired graft congestion venous outflow was the cause of previously unexplained graft failures during our initial experience of LDLT. Surgeons have agreed on the need of MHV reconstruction because of occasional massive congestion, and so there is an interest in criteria for MHV reconstruction.3–5 Further, HVC has been realized only after parenchymal transection and temporary arterial clamping of the donor liver; while preoperative prediction of congestion volume (CV) has been difficult.5 Recently, preoperative liver volumetry and the measurement of the hepatic vein diameter based on the 3-dimensional computed tomography (3D-CT) has resulted in a significantly improved outcomes, compared to the use of 2-dimensional computed tomography.6, 7 Furthermore, Kishi et al.8 reported that CV could be calculated and predicted applying new 3D-CT software by calculating the volume from the diameter and length of intrahepatic vascular branches. This study aimed to analyze the efficacy and accuracy of predicting CV and graft volume (GV) based on the software and to evaluate the appropriateness of the reconstruction of these tributaries.


LDLT, living donor liver transplantation; MHV, middle hepatic vein; IRHV, inferior right hepatic vein; CV, congestion volume; 3D-CT, 3-dimensional computed tomography; GV, graft volume; RLV, right-lobe volume; SLV, standard liver volume; HVC, hepatic venous congestion.

Patients and Methods


From October 1996 to January 2005, 177 consecutive LDLTs were performed at Kyushu University Hospital (Fukuoka, Japan). Our preoperative volumetric evaluation for donors consisted of two-dimensional computed tomography (1996-1999) and conventional 3D-CT (2000).6, 7 In October 2003, the 3D-CT examination using new software (Region Growing Software Version 0.5a; Hitachi Medical Corporation, Chiba, Japan) was introduced for preoperative evaluation and 42 adult donors (right-lobe graft, n = 25; left-lobe graft, n = 17) have since been evaluated. The indication for 42 LDLTs included fulminant hepatic failure (n = 4), primary biliary cirrhosis (n = 7), hepatitis C-cirrhosis (n = 21), hepatitis B-cirrhosis (n = 5), alcoholic cirrhosis (n = 1), cryptogenic cirrhosis (n = 3). The donors were a father (n = 1), a mother (n = 1), husbands (n = 5), wives (n = 2), brothers (n = 2), sisters (n = 2), sons (n = 21), daughters (n = 8), and a cousin (n = 1). All patients provided written informed consent. Donors were 30 males and 12 females, mean age 35 yr (range 20-58). Donor mean height and weight was 164 cm (range 147-180) and 60 kg (range 37-83), respectively. Mean blood loss and operation time were 568 gm (range 100-725) and 417 minutes (range 320-515), respectively.

Measurement of Actual Graft Weight

Actual graft weight was measured on the back table after flushing with University of Wisconsin solution (ViaSpan®; Bristol-Myers Squibb, New York, NY) and trimming. A total of 1 cm3 of liver was estimated as 1 gm.7 The error ratio (%) was expressed as |E − A|/A × 100, where E is the estimated GV (mL) and A is the actual graft weight (gm).7

3D-CT Volumetry

Preoperative multidetector helical computed tomography (MDCT) images were made using 2-mm-thick slices represented on a computed tomography machine. Enhancement was achieved by an intravenous bolus of contrast nonionic medium (Iopamion™; Schering, Erlangen, Germany) at a speed of 5 mL/second. This method allows for clear visualization of the hepatic arteries, portal veins, and hepatic veins including IRHV and MHV tributaries. Three-dimensional reconstructions of the liver and the graft were rendered by multidetector helical competed tomography using the new 3D-CT software, which was able to calculate total liver volume and the volume of each vessel's (both portal vein branches and hepatic venous branches) territories from their diameter and length. The 3-dimensional image reconstructed by this software could reflect the actual congestion area (Fig. 1). The right-lobe volume (RLV) was calculated from the right portal vein territories and the CV of each hepatic venous branch was calculated automatically (Fig. 2).

Figure 1.

3D-CT image of the congestion area and an intraoperative finding. (A) The construction of the 3-dimensional image shows the drainage area of the MHV tributaries (V5, V8; indicated by red). (B) Discoloration of the donor liver surface after clamping of the right hepatic artery and MHV tributaries, which completely matched the predicted area by 3D-CT (A,B: same donor).

Figure 2.

3D-CT image of a liver. The volume of each vessel branch can be automatically calculated before an operation using the new software. (A,B) Construction of a 3-dimensional image shows the perfusion area (orange color) of the right portal vein. (C,D) The construction of a 3-dimensional image shows the drainage area of middle hepatic vein (MHV) tributaries (V5, V8; indicated by red).

GV Excluding CV

The percentage of CV/RLV was also calculated. Further, we developed a new parameter: (GV − CV)/(standard liver volume [SLV]). The SLV was calculated using Urata's formula: SLV (mL) = 706.2 × body surface area (m2) + 2.4.9 In our institution, SLV calculated using this formula was found to be more suitable compared with other formulas (data not shown).10 Recipients with right-lobe grafts (n = 25) were divided into 2 groups: with reconstruction of MHV tributaries or IRHV (n = 16), and without reconstruction of any these veins (n = 9) groups. (GV − CV)/SLV was compared between the 2 groups.

Surgical Technique

Surgical procedures in donors have been described elsewhere.1, 11 Briefly, during mobilization of the liver, all the right accessory hepatic veins of a significant size (>5 mm in diameter) were preserved and reconstructed in the recipient. In right-lobe grafts, significant hepatic veins from segment 5 or 8 were preserved as long as was possible during parenchymal transection. The graft was procured and flushed via the portal vein on the back table using University of Wisconsin solution, and the weight of the graft measured. Interposition vein grafts from the donor or recipient (e.g., inferior mesenteric, greater saphenous, and intrahepatic portal veins of the recipient) were procured for reconstruction of V5 and V8, if their diameters were 5 mm or more.

Influence of CV in Donors' Remnant Right-Lobe

The influence of CV on the remnant right lobe in donors who underwent extended left lobectomy (n = 17) was evaluated. In cases in which CV was under 20%, the diameter of MHV tributaries and IRHV were all very small (the diameter <5 mm). Donors were divided into high CV (CV > 20%, n = 13) and low CV (CV ≤ 20%, n = 4) groups. Total bilirubin and aspartate aminotransferase level at postoperative days 1, 2, 3, 5, and 7 were compared between the 2 groups.

Statistical Analysis

Data are expressed mean ± standard deviation. Statistical analysis was performed using Student's t-test and the Mann-Whitney U-test. StatView™ (Version 4.11; Abacus Concepts, Berkeley, CA) software on a Macintosh computer was used for all analyses. P < 0.05 was considered to be significant.


Relationship Between Actual Weight of Grafts and Estimated Volume of Grafts Using 3D-CT

Mean estimated total liver volume and GV were 1,185 mL (range 803-2,004) and 592 mL (range 278-1,055), respectively. Mean graft weight was 531 gm (range 270-720). The relationship between estimated GV and actual graft weight using 3D-CT in 42 donors was linear: y = 282.179 + 0.419 × (R2 = 0.592, P = 0.0024). The mean error ratio was 20.4% (range 0.0-84.0) (Fig. 3).

Figure 3.

Relationship between actual volume of grafts and their estimated volumes using 3D-CT software: y = 282.179 + 0.419x (R2 = 0.592, P = 0.0024). The error ratio was 20.4 ± 15.7% in this 3D-CT software. The error ratio was calculated as follows: error ratio (%) = |E − A|/A × 100, where E is the estimated graft volume (mL) and A is the actual graft volume (gm).

Ratio of CV of V5, V8, and IRHV in Right-Lobe Grafts

The mean estimated CV/RLV ratio was 32.3 ± 17.1% (V5: 17.3 ± 9.5%; V8: 8.5 ± 4.4%; IRHV: 6.6 ± 9.7%) in 25 donors (Fig. 4). Twenty-five right-lobe grafts included reconstruction with all these tributaries (n = 3), with any 2 tributaries (n = 7), with only 1 tributary (n = 6), and without any tributaries (n = 9) (Fig. 4). The mean CV/RLV ratios in patients with (n = 16) or without (n = 9) a reconstruction of MHV tributaries or IRHV were 41.7 ± 19.4% and 17.3 ± 8.4%, respectively (P < 0.01).

Figure 4.

Estimated CV of V5, V8, and the IRHV and RLV ratios. CVs of V5, V8, and IRHV were 15.2 ± 9.9% (range 0.0-36.5), 9.2 ± 4.1% (range 1.9-18.4), and 8.5 ± 11.4% (range 0.0-33.8), respectively. Total CV of RLV was 32.3 ± 17.1% (range 5.6-80.8). A total of 16 patients were reconstructed with some branches and 9 without any branches (+, with reconstruction; −, without reconstruction).

(GV − CV)/SLV.

The mean GV/SLV ratio in patients with (n = 16) or without (n = 9) reconstruction of MHV tributaries or IRHV was 57.6 ± 9.2% (range 44.3-78.6) and 67.0 ± 14.1% (range 47.8-90.2) (P = 0.08), respectively (Fig. 5). (GV − CV)/SLV resulted in 33.5 ± 12.8% (range 12.6-58.3) and 55.4 ± 12.9% (range 44.4-81.4), respectively, which were significantly lower than the GV/SLV ratios. However, (GV − CV)/SLV recovered to 52.4 ± 10.2% (range 38.3-78.6) after reconstruction of these tributaries (Fig. 5). There were no postoperative complications associated with graft congestion in either group.

Figure 5.

A new parameter (GV − CV)/SLV. Mean GV/SLV was 60.3 ± 12.9% (n = 25), but (GV − CV)/SLV was 41.8 ± 17.1%. Deducing CV from GV was significantly decreased from GV/SLV. The mean GV/SLV ratio in patients with (n = 16) or without (n = 9) reconstruction of MHV tributaries or the IRHV was 57.6 ± 9.2% (range 44.3-78.6) and 67.0 ± 14.1% (range 47.8-90.2) (P = 0.08), but (GV − CV)/SLV was 33.5 ± 12.8% (range 12.6-58.3) and 55.4 ± 12.9% (range 44.4-81.4), respectively (P < 0.01). After reconstruction, (GV − CV)/SLV in reconstruction cases increased to 52.4 ± 10.2% (range 38.3-78.6). NS, not significant.

The Outcome of CV in Donor's Remnant Liver

To estimate the effect of liver congestion, the CVs of 17 donors who underwent extended left lobectomy and their postoperative liver function were examined. Total CV/RLV ratio estimated by 3D-CT was 27.7 ± 10.6% (V5: 20.5 ± 8.1%; V8: 7.2 ± 4.7%, respectively) (Fig. 6A). We did not reconstruct any MHV tributaries in the donors, who were divided into 2 groups: high CV (CV > 20%, n = 13) and low CV (CV ≤ 20%, n = 4), to evaluate the influence of CV on the remnant right lobe. Total bilirubin was significantly higher in the high CV group on postoperative day 1 (P < 0.05) (Fig. 6B). Aspartate aminotransferase was not seen to have any significantly changes (Fig. 6C). No complications associated with congestion of the remnant liver have been experienced.

Figure 6.

(A) Estimated CV of V5 and V8 in remnant right lobe of donor cases. In 17 donor cases, the CV/RLV ratio using 3D-CT was 27.7 ± 10.6% (range 12.1-50.8): V5, 20.5 ± 8.1% (range 7.9-37.3); V8, 7.2 ± 4.7% (range 0.0-16.7); and no donors had tributaries reconstructed. (B) Serial changes of postoperative total bilirubin and aspartate aminotransferase in left-lobe donors. Total bilirubin was significantly high in the high CV group (P < 0.05). (High CV group: CV > 20%, low CV group: CV ≤ 20%; NS, not significant.)


LDLT using the right lobe has become a standard procedure to overcome graft size problems for adults or larger pediatric patients.1 These grafts, especially without the MHV, can potentially lead to HVC in segments 5 and 8 due to insufficient venous drainage,2, 3 the importance of which has been recognized by many surgeons.4 However, prediction of CV is very difficult and management of V5, V8, and IRHV in LDLT has been a controversial area. The demarcation line of HVC is evident only after parenchymal transection and performing an intraoperative hepatic arterial clamping test (Fig. 1B).5 3D-CT examination has been extremely helpful for us to understand the anatomy of a donor's hepatic vessels,6, 7 so that when there is an anomaly with regard to hepatic or portal veins, it is relatively easy to develop a surgical strategy. Operative simulations using these anatomical images contribute to a reduction of not only the risk for donors but also the stress involved with donor surgery. In this study, we evaluated the usefulness of a newly developed 3D-CT software that can calculate the volume of each hepatic vessel from its diameter and length. Using this, we substituted the conventional preoperative evaluation of the graft for a more substantial evaluation of the graft. Three-dimensional images of the graft and CV were well visualized (Figs. 1 and 2). Our result for the CV/RLV ratio (32.3 ± 17.1%) was compatible with that reported by Shin et al.12 However, volumetry by this 3D-CT software produces an error ratio of approximately 20%. Some factors thus need to be considered in relation to this error. First, a mismatch exists between the cutting line in these simulations and that in the actual hepatectomy; a 2-cm discrepancy could represent a difference of as much as 200 gm.13 Second, we need to consider the reduction of the vascular bed of the graft.14 Total liver volume before and after fluid infusion showed an approximately 33% difference in a porcine model.15 Third, dehydration from the osmotic pressure of the University of Wisconsin solution could result in a decrease of graft weight by approximately 4%.7 Fourth, grafts from donors under 30 yr old were significantly overestimated as compared to grafts from those over this age (data not shown); graft compliance was seemingly the cause.

Figure 7 shows an algorithm for graft selection for LDLT as used in our institution. The left-lobe is initially considered as the graft with respect to donor safety. We have reported favorable outcomes using such grafts.11, 16, 17 As left-lobe grafts are usually small, accurate assessment of graft volume before an operation is critical. Three methods of volumetry, such as formula, conventional 3D-CT, and this new 3D-CT software have been examined.6, 7, 18 Moreover, to avoid graft congestion due to excessive portal flow, either a splenectomy or a splenic artery ligation has been reported.19 Even for a right-lobe graft, graft-to-recipient weight ratio was sometimes under 1%. We selected a right-lobe graft when a left-lobe graft was insufficient for the recipient and remnant liver volume of the donor was over 35%. On the other hand, some reports have not reconstructed these tributaries if the diameter of these tributaries was under 5 mm.2, 13, 20 Certainly, a branch under 5 mm is not thought to be a significant branch, but our study revealed that some branches under 5 mm had high CV/RLV (>10%). Moreover, when many branches under 5 mm existed, as in V5 and V8, CV cannot be predicted only by the simple observation of a computed tomography image. Even if tributaries are under 5 mm, there are potential risks as such the CV being high. Surprisingly, our study shows that the maximum CV/RLV was over 80%. In such cases, careful planning of the reconstruction is critical, therefore it is currently thought that every effort should be made to reconstruct all significant MHV tributaries, as much as is possible because hepatic venous outflow is sometimes unpredictable. Since we clarified the importance of the each CV of V5, V8, and IRHV (Fig. 4), the use of the new parameter (GV − CV)/SLV has become a routine part of our graft selection algorithm (Fig. 7), and detailed simulations have become possible. Since the introduction of the criteria, we have not had any postoperative complications associated with graft congestion.

Figure 7.

Algorithm for the graft selection as used in our institution. Initially the left-lobe is considered as a graft, and generally is used. The right is chosen if the estimated extended left + caudate lobe volume of the donor is less than 35% of the SLV of the recipient. If a remnant liver volume is under 35% of the total liver volume, this donor will be rejected. If CV is over 25%, or the deducted CV from the GV is under 40%, reconstruction of these tributaries is needed.

There are some reports that HVC influences graft function and regeneration in LDLT recipients.2–4 However in healthy donors, there are few reports of how HVC influences the remnant liver.8, 21 In the donor's operation, we have not reconstructed these tributaries, except in 1 case. In this study, mean CV/RLV in the remnant right lobe was surprisingly high at 27.7 ± 10.6% and the maximum CV/RLV was as high as 50.8% (Fig. 6A). Moreover, total bilirubin was significantly high in cases with CV over 20% (Fig. 6B). These results indicate that CV is a latent risk and that estimation of CV using this software is very useful.

In conclusion, CV could be reliably predicted using this 3D-CT software. We believe that this new parameter [(GV − CV)/SLV] deserves to be an essential part of preoperative planning for hepatic vein reconstruction and graft selection.