Patients
Data of 497 patients who underwent LDLT at Severance Hospital from January 2016 to December 2021 were retrospectively collected. Pediatric patients and patients who received grafts from two separate donors or underwent re-liver transplantation were excluded. The donors of each recipient were paired during data collection. Ten patients received unusual grafts (Figure 1). All recipients and donors underwent a series of evaluations to investigate liver anatomy, including dynamic computed tomography (CT), magnetic resonance imaging (including magnetic resonance cholangiopancreato-graphy), and liver fibrosis scanning.
Graft Selection
The graft weight was estimated using CT volume analysis. After the measurement of the volumes of each conventional liver graft volumes (extended RL, modified RL, extended LL, or left lateral segment grafts), we calculated the graft-recipient weight ratio (GRWR). Then, we assessed the anatomical variances in the portal vein (PV), hepatic artery (HA), and hepatic duct (HD). Many types of HA branching have been reported, from the standard anatomy to the left HA arising from the left gastric artery and the right HA arising from the superior mesenteric artery. [8] Additionally, there is a variant in which the segment 4 artery branches out from the right HA. PVs are usually of three different types. Type 1 PV variant is the standard type, which shows the bifurcation of the right and left PVs from the main PV. Type 2 PV variant is the trifurcation type, in which the main PV branches into the right anterior, right posterior, and left PVs. Type 3 PV variant shows separate right posterior PV from the main PV and bifurcation of the right anterior and left PVs. [9] The intra-HD also has many variations. [10] The relation between the HA, PV, and HD of each section is also a major factor for consideration.
Considering the GRWR and the anatomical variances, donors who were unsuitable for donating conventional grafts were investigated for their eligibility to be the donors of unusual grafts. Based on their anatomical variances, possible unusual graft was assessed with GRWR. Each unusual graft has its favorable anatomy, and it will be described in detail later. We visualized each unusual graft using Synapse 3D Liver Analysis (FUJIFILM Medical Systems U.S.A. Inc., Valhalla, NY, USA).
Right Anterior Section Graft (Figure 2-A)
Suitable Anatomy
The extrahepatic second-order bifurcation of the right HA favors right anterior section graft procurement because it can ensure a longer stump of the right anterior HA. [4] No segment 4 artery branching from the right HA is also helpful. Type 3 PV variant, in which the posterior PV is separate from the main PV, is the favorable choice for right anterior section grafts. Separate bile ducts from the anterior and posterior sections are also important. For donors who have dominant volumes of the anterior section compared to those of the posterior section, the right anterior section grafts are favored.
Donor Graft Sectionectomy
After ligamentum teres, falciform, coronary, and right triangular ligament divisions, hilar dissection was initiated posterolaterally. The right HA, right PV, and right HD were identified and isolated. The anterior branches of the right HA and PV were identified and isolated separately. The right anterior HA and right anterior PV were clamped to draw a demarcation line. Intraoperative ultrasonography was performed to identify the location of the right and middle HVs. After confirming the dissection plane, the Cavitron ultrasonic surgical aspirator (CUSA) was used to transect the liver. While transectioning along the right HV, branches of segments 5 and 8 were ligated gently and divided for reconstruction. Another transection was performed along the middle HV. The right anterior HD was identified and isolated using the Glissonian approach and divided, leaving a 3-mm HD stump for reconstruction. The right anterior HA and right anterior PV were divided gently. The middle HV was sealed with an endovascular stapler and transected. Intraoperative cholangiography was performed to confirm the integrity of the right posterior HD. The graft was flushed with a histidine–tryptophan–ketoglutarate solution. The right HV branches of segments 5 and 8 was reconstructed using the vascular interposition graft (reconstruction of RHV). The proximal end of the interposition graft was anastomosed to the middle HV for a common channel outflow.
Recipient Surgery
The stump of the middle HV was left to reconstruct anastomosis. A bigger orifice was created in the middle HV using venotomy to ensure adequate venous outflow. The conjoined graft HV and enlarged recipient middle HV were anastomosed. The recipient PV was connected to the donor right anterior PV, and the recipient HA was connected to the donor right anterior HA. Intraoperative duplex ultrasonography was performed to confirm a successful flow. Finally, the right HD of the recipient and the right anterior HD of the donor were anastomosed using the duct-to-duct method. In some cases, hepaticojejunostomy was used to reconstruct the bile duct.
Right Posterior Section Graft (Figure 2-B)
Suitable Anatomy
The extrahepatic second-order bifurcation of right HA favors right posterior section graft procurement. The type 3 PV variant is the ideal choice for right posterior section grafts. Another favorable anatomy is an extrahepatic right posterior HD that drains into the common HD. Additionally, the right posterior HD that runs through the ventral side of the right posterior PV is the most suitable anatomy. [11, 12]
Donor Graft Sectionectomy
Division of ligamentum teres to hilar dissection was performed in the same manner as that for right anterior section grafts. The posterior branches of the right HA and PV were identified and isolated separately. The right posterior HA and right posterior PV were clamped to establish a demarcation line, and the right HV was identified using intraoperative ultrasonography. The CUSA was used to transect the liver after establishing the dissection plane. The Glissonian method was used to identify and isolate the right posterior HD, and parenchymal transection was performed toward the root of the right posterior HD. The right posterior HD stump length was 3 mm after division. The right posterior HA and right posterior PV were separated gently and the right HV was sealed with a stapler and transected. Intraoperative cholangiography was performed to confirm the integrity of the right anterior HD.
Recipient Surgery
The right HV of the donor and the right HV of the recipient were anastomosed. The donor right posterior PV was connected to the recipient PV, and the donor right posterior HA was connected to the recipient HA. A successful flow was confirmed using intraoperative duplex ultrasonography. Finally, the duct-to-duct approach was used to anastomose the right HD of the recipient with the HD of the donor. Hepaticojejunostomy was utilized to rebuild the bile duct in some situations.
Extended Left Liver Plus Caudate Lobe Graft (Figure 2-C)
Suitable Anatomy
The favorable anatomy for extended left liver plus caudate lobe grafts is the caudate vein draining into the inferior vena cava far from the orifices of the middle and left HVs because it is difficult to anastomose the caudate vein when it is situated near the root of the middle or left HV.
Donor Graft Lobectomy
The left PV was isolated at the bifurcation with the Glissonian pedicle method, while its transverse portion was left undissected to preserve the caudate lobe branches. Caution should be taken to avoid damage when securing the caudate vein. The left caval ligament was dissected, then the caudate lobe was mobilized from the vena cava while preserving the caudate vein. The transection line was made from the midpoint between the trunks of the right and middle HVs to the right margin of the retrohepatic inferior vena cava.
Recipient Surgery
A venotomy using an end-to-side method was performed on the vena cava to anastomose the caudate vein. [13] Using the conventional piggyback method, end-to-end anastomoses were established between the recipient common trunk and graft conjoined middle-left HV. End-to-end anastomosis of the left PV graft to the recipient PV and the left HA graft to the recipient HA was performed. However, in Case 7, the left HA was anastomosed to the recipient gastroduodenal artery due to the dissection of the recipient HA lumen. Usually, duct-to-duct anastomosis was favored for bile duct reconstruction, but hepaticojejunostomy was performed in one case (Case 6)
Postoperative Outcomes
We collected data on laboratory findings, CT findings, graft survival, overall survival, and complications of recipients and donors. We graded surgical complications using the Clavien–Dindo classification. [14] We classified Grade III or above as major complications. In addition, complications were categorized into the complication of HA, HV (including inferior vena cava), PV, bile duct, and others.
Statistical Analysis
Data are presented as numbers (proportions) for categorical variables and means and standard deviations for continuous variables. Due to the small sample size, non-parametric tests were required. To determine the significance of intergroup differences, Fisher’s exact test was used for categorical variables and the Mann–Whitney U test was used for continuous variables; data are summarized and reported as medians (interquartile ranges). Survival rates were evaluated using the Kaplan–Meier method, and the log-rank test was used for between-group comparisons.
Propensity score matching (PSM) was performed to minimize selection bias and balance variables between the unusual graft and the conventional graft groups. Unusual graft recipients were matched with conventional graft recipients in a ratio of 1:2 using the nearest neighbor matching algorithm without replacement with distances determined by logistic regression. The caliper was not applied because the unusual graft group only had 10 patients. PSM was performed based on the following variables: recipient model for end-stage liver disease (MELD) score, Milan’s criteria, and donor age, height, and weight.
Statistical analyses were performed using IBM SPSS 26 (SPSS Inc., Chicago, IL) and R 3.5.3 software (R Foundation for Statistical Computing, Vienna, Austria). P-values <0.05 were considered statistically significant.