Short-term outcomes and risk factors of full-size split liver transplantation: a single-center experience of 113 cases

Abstract

Background:

Full-size split liver transplantation (FSLT) offers an attractive way to expand the donor pool for adult patients. However, the efficacy and safety of FSLT remain under debate.

Materials and Methods:

One hundred and three recipients who underwent FSLT were enrolled, and 124 recipients who received whole-liver transplantation (WLT) were taken as a control group. Clinical data and short-term outcomes within 1 year posttransplantation from 2019 to 2023 were compared between the two groups. Survival rates were calculated using the Kaplan–Meier method and compared by the log-rank test. Meanwhile, propensity-score matching (PSM) using key prognostic factors was conducted and the results were compared. Univariate and multivariate logistic regression analyses were performed to identify risk factors for serious complications in FSLT.

Results:

Thirty-three (29.2%) and 25 (20.2%) recipients in FSLT and WLT experienced postoperative complications graded ≥ III, with no significant differences (P = 0.106). However, biliary and vascular complications were more frequent in FSLT than in WLT (11.5% vs. 4.0% and 12.4% vs. 1.6%). The 3- and 6-month mortality rates in FSLT did not differ significantly from those in WLT (6.2% and 8.8% vs. 3.2% and 4.8%). The 1-year recipient (OS) and graft survival (GS) rates were 87.6% and 85.8% for FSLT, and 92.7% and 91.9% for WLT, with no significant differences. After PSM, the 1-year OS and GS were 81.3% and 81.3% for FSLT, and 90.6% and 90.6% for WLT (P = 0.081). Multivariate analysis revealed that a graft with steatosis of 5%–10%, a MELD score ≥ 20, sarcopenia, a history of hepatectomy, and an operation time > 420 min were independent risk factors for the incidence of postoperative complications graded ≥ III in FSLT.

Conclusions:

FSLT remains a challenging procedure. However, the short-term outcomes of FSLT were acceptable compared with those of WLT, encouraging the use of this procedure to increase the donor pool whenever possible.

Introduction

Liver transplantation (LT) has significantly contributed to the successful treatment of terminal liver diseases and malignancies. However, waitlist mortality rates remain high due to the scarcity of donor organs and the increasing number of patients requiring LT. Split liver transplantation (SLT), a strategy to increase the liver graft supply, is categorized into two main types: conventional SLT (CSLT) and full-size SLT (FSLT)^[1–3^].

Currently, CSLT is a significant source of pediatric grafts without compromising graft availability for adults and has been widely accepted^[3–7^]. FSLT involves splitting the donor liver into full right and left hemiliver grafts, thereby expanding the donor pool for adult patients. Since the first FSLT was reported in the 1990s, its use has been limited to a few experienced centers^[2,8–12^]. FSLT is technically challenging and requires the immediate availability of skilled surgeons to perform the three complex procedures simultaneously. Compared with whole-liver transplantation (WLT), FSLT is associated with an increased risk of postoperative complications, particularly vascular and biliary complications^[6,9–12^]. Consequently, the safety and feasibility of FSLT remain a topic of debate, causing reluctance for the wider adoption of this procedure^[5,13–15^].

Most studies have reported that graft and patient survival after FSLT were primarily compromised in the early postoperative period, and the most significant differences between FSLT and WLT were observed in 90-day and 1-year outcomes, with no significant long-term differences^[6–12^]. However, most studies on FSLT were performed over a long time span and involved a limited number of cases^[3,6,8–12^]. For example, Chen et al reported 100 FSLTs performed at a single center over 16 years, while Aseni et al detailed 64 cases from a 12-year collaborative study involving seven centers^[11,12^]. A meta-analysis of FSLT studies from 2001 to January 2021 identified only 397 hemiliver grafts with suitable data^[6^]. The extended time frame and the limited number of cases, averaging fewer than 20 cases per year worldwide, have diminished the impact of these studies and hindered the accumulation of experience needed to construct a learning curve for FSLT. Besides, most studies either lack a risk factor analysis for serious postoperative complications, which are strongly correlated with postoperative mortality, or the conclusions of such analyses were weakened by the small sample size^[3–12^]. Therefore, the safety and effectiveness of FSLT for adults within 1 year of posttransplantation warrant further exploration.

Thus, this study aimed to evaluate the short-term outcomes of FSLT performed from 2021 to 2023 at a single center, using a cohort of patients who underwent WLT from 2019 to 2023 as a control group.

Methods

Consent statement

All prospectively registered data were extracted from the electronic medical record system of the Hospital. All LT recipients were registered in the China Organ Transplant Response System (COTRS), www.cot.org.cn)^[16^]. The organ donation and transplantation procedures were strictly implemented under the regulations of the China Organ Donation Committee and the Organ Transplant Committee^[16,17^]. Written informed consent for voluntary donation was obtained from the families of deceased donors between January 2019 and December 2023, documented and stored by the National Organ Transplantation Coordinator in China. No organs from prisoners were used in this study. Organ donation was approved by the Ethics Committee of the Organ Procurement Organization (OPO)^[16,17^]. All enrolled recipients provided written informed consent upon registration in COTRS before LT at the Hospital, and the consented forms were stored by the chief physician.

This study was approved by the Ethics Committee of the Hospital and was conducted in accordance with the Declaration of Helsinki, with no vulnerable populations involved. The study was retrospectively registered at ClinicalTrials.gov. This work has been reported in line with the STROCSS criteria^[18^].

Study population and cohorts

Between January 2019 and December 2023, a total of 296 patients underwent LT at Hospital. All recipients who underwent FSLT or WLT were considered for inclusion. Of these, pediatric cases, re-transplantation, CSLT, living donor liver transplantation (LDLT), and grafts with steatosis > 10% were excluded from this study. According to the inclusion and exclusion criteria, 113 consecutive adult FSLT recipients (2021–2023) were enrolled as the study group, with 124 WLT cases (2019–2023) serving as controls. All patients were followed up until December 2024, with a median follow-up of 23.0 months (Q1–Q3: 15.5–33.0) in the FSLT group and 39 months (Q1–Q3: 21.0–53.8) in the WLT group. Figure 1 presents the study flowchart, detailing patient selection for the pre- and post-propensity-score matching (PSM) FSLT and WLT cohorts.

Figure 1.

Flowchart of study population.

HIGHLIGHTS

• The short-term outcomes of FSLT were comparable with WLT, encouraging the use of this procedure to increase the donor pool.

• Patients with a history of hepatectomy, graft with steatosis 5%–10%, MELD≥20, sarcopenia and operation time > 420 min who received FSLT are at high risk of serious postoperative complications.

• This study highlights the importance of refining surgical techniques for vascular and biliary reconstruction to reduce the incidence of postoperative complications and mortality.

Criteria for split donor screening

After brain death confirmation, donors underwent enhanced upper/middle abdominal CT with three-dimensional reconstruction for liver volumetry when feasible. The general criteria for donor selection for FSLT were as follows:

(1) Age between 18 and 60 years;

(2) Body mass index (BMI) ranging from 19 to 28 kg/m²;

(3) Hemodynamic stability or the use of a single vasoactive drug at a low dosage (norepinephrine ≤ 0.1 μg/kg/min or dopamine ≤ 5 μg/kg/min);

(4) Graft without steatosis or with steatosis ≤ 10%;

(5) No uncontrollable infection;

(6) Liver functional indicators (alanine aminotransferase, aspartate transaminase, total bilirubin quantification) ≤ 3 times the normal value;

(7) Blood sodium ≤ 160 mmol/L;

(8) Absence of major vascular/biliary anomalies.

A sample biopsy was performed for each graft before splitting. The OPO team at the hospital carried out the procurement operations^[19^].

Criteria for recipient screening and graft allocation

The criteria for potential recipients of FSLT were as follows:

(1) Graft-to-recipient weight ratio (GRWR) ≥ 1%;

(2) No significant vascular anomalies.

Once the decision to proceed with FSLT was made, hemiliver graft allocation was carried out according to the Model for End-stage Liver Disease (MELD) scores of patients on the LT waiting list using the COTRS with the allocation policy of regional priority. The potential primary candidate, determined by the highest MELD score, was given priority for the hemiliver graft. If this recipient did not meet the criteria for FSLT, they would be bypassed until a suitable one was identified. All hemiliver grafts were allocated using COTRS^[16,17^].

Surgical techniques for FSLT

The surgical techniques for graft procurement and implantation are described in detail in the supplementary data to this article online at Supplemental Digital Content Figures S1–S2, available at: http://links.lww.com/JS9/H156. Briefly, for donors who donated after brain death (DBD) with stable hemodynamics (systolic pressure ≥ 90 mmHg), hemiliver graft procurement was performed in situ using the modified LDLT technique described by Fan et al^[20^]. The liver parenchyma was split using a Cavitron Ultrasonic Surgical Aspirator without the Pringle maneuver for vascular inflow control. The segment V/VIII outflow veins were disconnected after being protected by the traction band. After splitting the liver parenchyma, intraoperative cholangiography was performed to accurately locate the cutoff point of the bile duct. After perfusion and preservation in the University of Wisconsin solution, a liver graft was routinely obtained. In the case of multi-organ procurement including heart and/or lung, liver parenchyma splitting in situ was performed first, followed by the thoracic team in the usual fashion for the perfusion and procurement of the chest organs^[21^]. For donors with either circulatory death [DCD or DBCD (donors after brain and circulatory death)] or DBD with hemodynamic instability (systolic pressure < 90 mmHg), hepatic parenchymal splitting was performed ex vivo after rapid procurement of liver grafts using a technique modified from Broering et al^[22^]. The celiac trunk was usually retained in the left hemiliver, and the common bile duct (CBD) was on the right side. The allocation of the main portal vein (MPV) was determined based on the anatomical characteristics of the donor and recipient. The inferior vena cava (IVC) was retained in the left hemiliver or longitudinally split and shared with both grafts using a modified technique reported by Gundlach et al^[23^]. The left or right hemiliver and accessory IVC were completely separated, and each hemicaval was reconstructed to create a wide, inverted triangular patch, allowing for a convenient anastomosis. Reconstruction was performed when the diameter of the severed veins in segments V and VIII was ≥ 3 mm. The donor’s “Y”-type iliac vessel was frequently used as a conduit draining into the IVC to restore the outflow of segments V and VIII.

Following the surgical principles of LDLT, the left and right branches of the hepatic artery (HA) and portal vein (PV), as well as the bile ducts, were severed close to the hilar plate while preserving the IVC. Implantation of a right or left hemiliver graft was mainly performed using the modified piggyback technique, transverse cuts along the three hepatic veins (HVs), and longitudinal division of the anterior wall of the IVC to form a large triangular opening matching the size of the hemicaval patch of the graft. For hepatic arterial anastomosis, microsurgical techniques were performed using continuous (diameter ≥ 3 mm) or intermittent (diameter < 3 mm) sutures. In most cases, the bile ducts were anastomosed end-to-end without stenting for right hemiliver grafts, whereas T-tube stenting was frequently used for left hemiliver grafts. Choledochojejunostomy was occasionally performed as appropriate.

Immediately after graft implantation and vascular anastomosis, HV, PV, and HA flows were measured using color Doppler ultrasonography. In cases in which the actual GRWR was less than 1.0%, intentional portal pressure monitoring via the right gastroepiploic vein or inferior mesenteric vein with an 18-G central venous catheter during the recipient operation was performed to guide pressure modulation, such as splenic arterial ligation or splenectomy, to avoid the occurrence of small-for-size syndrome (SFSS), and somatostatin may be used postoperatively^[24–26^].

Immunosuppressive protocols

The routine immune induction scheme for both groups of patients (FSLT and WLT) is initiated by administering 20 mg of basiliximab within 2 h of surgery, followed by 500 mg of methylprednisolone intraoperatively and another 20 mg of basiliximab on postoperative day 4^[27^]. Sustained immunosuppression typically employs a calcineurin inhibitor (predominantly Tacrolimus) plus mycophenolate mofetil (MMF) dual regimen^[28^]. For patients with autoimmune liver diseases, a triple regimen of corticosteroids plus Tacrolimus and MMF is used. The mammalian target of rapamycin inhibitors (mTORi) (predominantly sirolimus) plus MMF regimen could be considered for patients with hepatocellular carcinoma (HCC) at a higher risk of tumor recurrence 3 months postoperatively^[29^].

Since there are reports that LDLT recipients have significantly decreased tacrolimus dosing requirements compared with deceased donor LT with whole-liver graft (DDLT) recipients during the first 3 months posttransplantation^[30,31^], the initiating dose of Tacrolimus in our center is reduced to 1.0–2.0 mg/day for FSLT according to the patients’ weight and is adjusted according to its concentration monitored through serum trough levels and the clinical status of the recipients.

Short-term outcome assessment and study endpoints

The demographic data and clinical variables of donors and recipients in the two groups were recorded for analysis. The actual weight of the graft, measured after procurement, was used to calculate the graft-recipient weight ratio (GRWR), and only laboratory data were used for MELD scores. Grafts with steatosis from 0 to <5% were defined as having none, and those from 5% to 10% were defined as having it. Sarcopenia was defined as the skeletal muscle index (SMI) at the L3 vertebra level < 40.2 cm²/m² for males and < 31.6 cm²/m² for females according to Kong et al^[32^]. Operative data, including graft split techniques, cold ischemic time (CIT), anhepatic phase, and operation time, were compared between the two groups. The optimal operation time cutoff (420 min) was determined by ROC curve analysis using the maximum Youden index^[33^].

The primary study endpoints for short-term outcome assessment were recipient mortality within 3 and 6 months, overall survival (OS), and graft survival (GS) within 1 year after transplantation. The secondary endpoints included the incidences of postoperative complications graded ≥ III according to the Clavien–Dindo classification system^[34^] and reoperation or re-transplantation within 1 year. All clinical and follow-up data were prospectively registered and had to be uploaded to the China Liver Transplant Registry (CLTR) system^[16,17^]. In this study, the grade of postoperative complications was blindly checked and reevaluated by two independent assessors, and the opinion of a third assessor may be introduced in case of disagreement. If multiple complications occurred in the same patient, only the complication with the highest grade was included in the morbidity calculation for postoperative complications.

Other complications such as early allograft dysfunction (EAD) defined according to Olthoff et al^[35^], SFSS defined according to Hernandez–Alejandro et al^[24^], graft-versus-host disease (GVHD)^[36^], acute renal failure with dialysis, and acute rejection were also compared between the two groups.

Statistical analysis

Continuous variables were described using the median and first-third quartiles (Q1, Q3) and compared using the Mann–Whitney U test. Categorical variables were reported as absolute values and relative frequencies and compared using the chi-square test or Fisher’s exact test, as appropriate. OS was calculated from the date of LT until the date of death or the end of follow-up, with all causes of death recorded as events within 1 year posttransplantation. GS was defined as the time from transplantation to either re-transplantation or death for any cause within 1 year posttransplantation. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test. Power analysis was performed for the comparison of OS/GS rates between the two groups based on the observed sample sizes (α = 0.05). Meanwhile, to control for confounding factors, a PSM using key prognostic factors of donor type (DBD vs. DCD or DBCD), cold ischemic time of the graft (CIT), recipient weight, laboratory MELD score, and sarcopenia was performed between the two groups. Comparative analyses were then conducted before and after PSM. The PSM procedure was conducted using the “MatchIt” package of the R 4.2.2 software (R Foundation for Statistical Computing, Vienna, Austria), with the caliper set to 0.01. To determine the risk factors for the incidence of postoperative complications (Clavien–Dindo classification ≥ grade III)^[34^], multivariable logistic regression analysis using the forward likelihood ratio (LR) was carried out on variables with P ≤ 0.1 at univariate analysis.

All statistical analyses were performed using SPSS version 25.0 (IBM Corp., Armonk, NY, USA). Statistical significance was set at P < 0.05.

Results

Rate of full-size graft splitting

Between January 2021 and December 2023, the OPO of Zhejiang Province approved 152 deceased donors for organ donation in the local region^[17^]. Among them, 61 whole-liver grafts underwent full-size splitting for adult LT, resulting in 122 hemiliver grafts (61 right–left pairs). Of these, 114 grafts were allocated to adult recipients listed for LT at our center, while the remaining eight hemiliver grafts were allocated to other centers through the COTRS, with no graft wastage (confirmed by personal communication). Therefore, the rates of liver graft splitting and utilization were 40.1% and 100%, respectively.

Demographic and clinical features of donors and recipients

According to the inclusion and exclusion criteria, there were 113 consecutive adult recipients in the FSLT group and 124 in the WLT group (Fig. 1). The demographic and clinical features of the donors and recipients in pre- and post-PSM FSLT and WLT were compared (Table 1).

Table 1.

Donor and recipient characteristics in pre- and post-PSM of FSLT and WLT groups.

Characteristics	Pre-PSM			Post-PSM
	FSLT (N = 113)	WLT (N = 124)	P	FSLT (N = 32)	WLT (N = 32)	P
Donor
Gender (%)			0.043			0.320
Male	101(89.4)	99(79.8)		28(87.5)	25(78.1)
Female	12(10.6)	25(20.2)		4(12.5)	7(21.9)
Age (years)	43.0(32.0, 52.0)	49.0(39.2, 54.0)	0.025	42.5(32.0, 50.0)	49.5(39.2, 55.7)	0.092
Body weight (kg)	70.0(65.0, 75.0)	67.5(60.0, 70.0)	0.002	70.0(60.0, 75.0)	65.0(60.0, 70.0)	0.031
BMI (kg/m2)	24.2(22.8, 25.2)	23.5(22.0, 24.9)	0.117	24.2(22.4, 25.2)	22.9(21.3, 24.7)	0.280
ICU stay (days)	8.0(6.0, 12.0)	9.0(6.0, 14.5)	0.604	9.5(7.0, 14.5)	10.5(5.7, 20.2)	0.476
CPCR history (%)	26(23.0)	20(16.1)	0.181	26(81.3)	28(87.5)	0.491
Donor type (%)			<0.001			0.802
DBD	80(70.8)	45(36.3)		16(50.0)	17(53.1)
DBCD (DCD)	33(29.2)	79(63.7)		16(50.0)	15(46.9)
Procurement time (min)	181.0(164.2, 220.0)	60.0(24.5, 69.0)	<0.001	115.0(62.0, 176.7)	61.0(50.0, 82.0)	0.001
Cause of death (%)			0.552			0.294
CVA	40(35.4)	48(38.7)		7(21.9)	12(37.5)
Traumatic head injury	65(57.5)	70(56.5)		22(68.8)	18(56.3)
Anoxia	8(7.1)	5(4.0)		3(9.4)	1(3.1)
Others	0	1(0.8)		0(0.0)	1(3.1)
Graft with steatosis (%)			0.378			1.000
None (0%– < 5%)	78(69.0)	92(74.2)		25(78.1)	25(78.1)
Yes (5%–10%)	35(31.0)	32(25.8)		7(21.9)	7(21.9)
CIT (minutes)	286(238, 351)	351(270, 500)	<0.001	305.5(261.0, 385.0)	321.0(270.0, 480.0)	0.307
Serum sodium (mmol/L)	148.1(143.2, 156.3)	149.0(139.9, 15.90)	0.576	147.4(141.4, 157.1)	151.0(142.3, 161.4)	0.390
ALT (U/L)	43.0(29.5, 77.5)	50.0(27.0, 98.0)	0.655	35.5(26.7, 89.0)	46.0(24.0, 97.0)	0.984
AST (U/L)	59.0(40.0, 105.0)	61.0(38.0, 127.0)	0.699	63.5(40.2, 102.0)	64.0(31.0, 106.0)	0.885
TB (µmol/L)	15.0(10.4, 24.7)	14.6(9.1, 24.5)	0.355	12.6(10.4, 20.1)	15.0(10.3, 23.7)	0.545
INR	1.2(1.1, 1.4)	1.3(1.1, 1.4)	0.651	1.3(1.1, 1.4)	1.3(1.0, 1.4)	1.000
Recipient
Gender (%)			0.003			0.320
Male	82(72.6)	109(87.9)		25(78.1)	28(87.5)
Female	31(27.4)	15(12.1)		7(21.9)	4(12.5)
Age (years)	56.0(48.0, 62.5)	53.0(46.2, 61.0)	0.244	54.5(49.0, 61.7)	53.0(45.2, 61.5)	0.605
Body weight (kg)	62.0(53.0,68.0)	70.2(61.3, 77.0)	<0.001	66.8(59.8, 77.0)	66.9(60.0, 75.0)	0.968
BMI (kg/m2)	22.1(19.9, 24.4)	24.3(21.9, 26.4)	<0.001	22.8(21.1, 26.1)	23.5(20.8, 25.5)	0.682
ABO incompatible (%)	2(1.8)	11(8.9)	0.016	0(0.0)	1(3.1)	1.000
Indications of LT (%)			0.438			0.302
Malignant	64(56.6)	64(51.6)		14(43.8)	10(31.3)
Non-malignant	49(43.4)	60(48.4)		18(56.3)	22(68.8)
History of hepatectomy (%)	36(31.9)	32(25.8)	0.304	10(31.3)	9(28.1)	0.784
HBsAg (+) (%)	73(64.6)	85(68.5)	0.520	17(53.1)	25(78.1)	0.035
PV thrombosis before LT (%)	25(22.1)	18(14.5)	0.129	6(18.8)	6(18.8)	1.000
GRWR (%)	1.3(1.1, 1.5)	2.1(1.8, 2.5)	<0.001	1.3(1.1, 1.6)	2.2(1.9, 2.5)	<0.001
ICU stay after LT (d)	2.0(2.0, 4.0)	2.0(1.0, 3.0)	0.290	2.5(1.0, 4.0)	2.0(1.0, 3.0)	0.286
Post-LT hospital stay (d)	21.0(15.5, 30.5)	19.5(15.0, 27.0)	0.113	21.0(15.0, 32.0)	18.5(14.2, 26.5)	0.270
Laboratory MELD score ≥20(%)	28(24.8)	51(41.1)	0.008	9(28.1)	6(18.8)	0.376
Sarcopenia	32(28.3)	20(16.1)	0.024	5(15.6)	6(18.8)	0.740

ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI, body mass index; CIT: cold ischemic time; CPCR, cardiopulmonary cerebral resuscitation; CVA, cerebrovascular accident; DBCD: donation after brain and circulatory death; DBD, donation after brain death; DCD, donation after circulatory death; FSLT, full-size split liver transplantation; GRWR, graft-to-recipient weight ratio. ICU, intensive care unit; INR: international normalized ratio; LT, liver transplantation; MELD, laboratory value of model for end stage liver disease; PV, portal vein; PSM, propensity-score matching; TB: total bilirubin.

There were more male donors in the FSLT group. They were younger and had a higher body weight than those in the WLT group (P < 0.05). Forty donors donated 80 hemiliver grafts after brain death (70.8%) in FSLT, compared with 45 (36.3%) in the WLT group, and a significant difference was observed (P < 0.001). The graft procurement time was significantly longer in FSLT than in WLT (181.0 vs. 60.0 min; P < 0.001), but the cold ischemic time (CIT) of the grafts was significantly shorter in FSLT than in WLT (286 vs. 351 min; P < 0.001). After excluding grafts with steatosis > 10% in the WLT group, the ratio of grafts with steatosis from 5% to 10% showed no significant difference between the two groups (Table 1). Regarding the characteristics of recipients, a higher proportion of women with lower body weight than in the WLT group (P < 0.05). The GRWR was significantly lower in FSLT than in WLT (P < 0.001), and there were 2 and 11 cases of ABO-incompatible LT in FSLT and WLT, respectively (P = 0.016). The ratio of recipients with laboratory MELD scores ≥ 20 in FSLT was lower than that in WLT (P = 0.008), but the ratio of recipients with sarcopenia was higher than that in WLT (P = 0.024). No significant differences in other donor and recipient parameters were observed between the two groups (Table 1).

After PSM, the sample size was reduced to 32 cases for each group. Only donor weight, graft procurement time, the proportion of recipients with HBsAg (+), and GRWR remained statistically different between the two groups (Table 1).

Differences in surgical techniques

Analyses of surgical technique data (Table 2) revealed significant differences between the two groups. For graft implantation, the piggyback technique was performed in 95.6% of cases in the FSLT group, compared to 69.4% in the WLT group (P < 0.001). There were more cases receiving splenectomy in the FSLT group than in the WLT group (P = 0.037). The times for HA anastomosis and the operation in the FSLT group were longer than those in the WLT group (P < 0.001). Moreover, the rate of T-tube drainage in the FSLT group (52.2%) was significantly higher than that in the WLT group (3.2%; P < 0.001). The other variables did not differ significantly between the two groups. Comparison between the two post-PSM groups revealed the same results as in the pre-PSM analyses.

Table 2.

Surgical characteristics of patients in pre- and post-PSM of FSLT and WLT groups.

Characteristics	Pre-PSM			Post-PSM
	FSLT (N = 113)	WLT (N = 124)	P	FSLT (N = 32)	WLT (N = 32)	P
Implantation type (%)			<0.001			0.005
Conventional	5(4.4)	38(30.6)		2(6.3)	11(34.4)
Piggyback	108(95.6)	86(69.4)		30(93.7)	21(65.6)
Splenectomy (%)	14(12.4)	6(4.8)	0.037	5(15.6)	1(3.1)	0.196
SA ligation (%)	7(6.2)	6(4.8)	0.647	3(9.4)	0	0.238
Anhepatic phase (min)	48(43, 55)	49(44, 60)	0.172	50(43, 57)	54(45, 65)	0.114
HA anastomosis (min)	15(12, 23)	13(10, 15)	<0.001	16(13, 23)	13(11, 15)	0.002
PV thrombectomy (%)	26(23.0)	19(15.3)	0.132	7(21.9)	5(15.6)	0.522
ORT >420 min (%)	71(62.8)	28(22.6)	<0.001	22(68.8)	10(31.3)	0.003
T-tube drainage (%)	59(52.2)	4(3.2)	<0.001	14(43.8)	1(3.1)	<0.001

FSLT, full-size split liver transplantation; HA, hepatic artery; ORT, operation time; PSM, propensity-score matching; PV, portal vein; SA, spleen artery; WLT, whole-liver transplantation.

Postoperative complications and mortality within 1 year

Table 3 presents details of the postoperative complications and mortality within 1 year of posttransplantation. A total of 76 events of operative complications graded ≥ III were recorded in 33 recipients in the FSLT group, resulting in a morbidity of 29.2% (33/113), compared with 25 recipients in the WLT group (20.2%) with 49 such events. The difference in morbidity between the two groups was not statistically significant (P = 0.106), although the rate of total events of serious complications in FSLT (76/113, 67.3%) was significantly higher than that in WLT (49/124, 39.5%) (P < 0.001). Among these events, the overall incidences of biliary (11.5% vs. 4.0%, P = 0.040) and vascular complications (12.4% vs. 1.6%, P = 0.001) in the FSLT group were significantly higher than those in the WLT group. Also, the incidences of HA complications including HA thrombosis/stricture, secondary HA bleeding, and outflow obstruction in the FSLT group were significantly higher than those in the WLT group (P < 0.050). No significant differences were observed in the incidences of other postoperative complications such as incision infection/dehiscence, postoperative hemorrhage, and gastrointestinal perforation/hemorrhage between the two groups. SFSS was found in four cases in the FSLT group compared with zero cases in the WLT group (P = 0.050). No significant differences were found in other complications such as EAD, GVHD, acute renal failure with dialysis, and acute rejection between the two groups. Twenty cases in the FSLT group and 14 in the WLT group underwent reoperation because of postoperative complications, with no significant differences between the two groups. Two recipients in the FSLT group and one in the WLT group underwent re-transplantation because of HA thrombosis (P = 0.607). The 3- and 6-month posttransplant mortality rates were 6.2% and 8.8% in the FSLT group, and 3.2% and 4.8% in the WLT group, with no significant differences. A total of 14 (12.4%) and 9 (7.3%) deaths occurred at the end of 1-year follow-up in the FSLT and WLT groups, respectively, with no significant differences between the two groups (P > 0.05).

Table 3.

Postoperative complications and mortality in pre- and post-PSM of FSLT and WLT groups within 1 year.

Post-OR complicationsa	Pre-PSM			Post-PSM
	FSLT (N = 113)	WLT (N = 124)	P	FSLT (N = 32)	WLT (N = 32)	P
Morbidity of post-OR complications	33(29.2)	25(20.2)	0.106	9(28.1)	4(12.5)	0.120
Biliary complications	13(11.5)	5(4.0)	0.030	3(9.4)	1(3.1)	0.613
Biliary stricture	5(4.4)	2(1.6)	0.263	1(3.1)	1(3.1)	1.000
Biliary dehiscence/leakage	5(4.4)	1(0.8)	0.106	1(3.1)	0	1.000
Non-anatomic stricture	3(2.7)	2(1.6)	0.671	1(3.1)	0	1.000
Vascular complications	14(12.4)	2(1.6)	0.001	3(9.4)	0	0.238
HA complications	7(6.2)	1(0.8)	0.029	0	0	NA
PV thrombosis/stricture	5(4.4)	1(0.8)	0.106	0	0	NA
IVC thrombosis/stricture	2(1.8)	0	0.228	1(3.1)	0	1.000
HV outflow obstruction	5(4.4)	0	0.024	3(9.4)	0	0.238
Incision infection/dehiscence	6(5.3)	7(5.6)	1.000	3(9.4)	0	0.238
Postoperative hemorrhage	14(12.4)	10(8.1)	0.270	6(18.8)	2(6.3)	0.257
GI perforation/hemorrhage	6(5.3)	3(2.4)	0.316	2(6.3)	0	0.492
EAD	1(0.9)	4(3.2)	0.372	1(3.1)	1(3.1)	1.000
SFSS	4(3.5)	0	0.050	0	0	NA
GVHD	2(1.8)	1(0.8)	0.607	1(3.1)	0	1.000
Acute renal failure with dialysis	10(8.8)	6(4.8)	0.219	4(12.5)	0	0.113
Acute rejection	0(0.0)	4(3.2)	0.123	0	0	NA
Myocardial infarction	1(0.9)	0	0.477	0	0	NA
Total post-OR complications	76(67.3)	49(39.5)	<0.001	24(75.0)	5(15.6)	<0.001
Reoperation	20(17.7)	14(11.3)	0.160	6(18.8)	3(9.4)	0.474
Re-transplantation	2(1.8)	1(0.8)	0.607	0	0	NA
3-month mortality posttransplant	7(6.2)	4(3.2)	0.278	2(6.3)	1(3.1)	1.000
6-month mortality posttransplant	10(8.8)	6(4.8)	0.219	3(9.4)	1(3.1)	0.613
1-year mortality posttransplan	14(12.4)	9(7.3)	0.183	6(18.8)	3(9.4)	0.474

EAD, early allograft dysfunction; FSLT, full-size split liver transplantation; GI, gastrointestinal; GVHD, graft-versus-host disease; HA, hepatic artery; HV, hepatic vein; ITBL, ischemic-type biliary lesion; IVC, inferior vena cava; post-OR, post-operation; PSM, propensity-score matching; PV, portal vein; SA, spleen artery; SFSS, small-for-size syndrome; WLT, whole-liver transplantation.

^aAccording to the Clavien–Dindo scoring system graded ≥ III.

Table 4 presents the leading causes of posttransplant death within 1 year in both groups. Five recipients in the FSLT group died of graft failure related to vascular complications compared with 0 in the WLT group (P = 0.024). Of note, three deaths in the FSLT group and two in the WLT group were related to sarcopenia. No significant difference was observed for other causes of death between the two groups.

Table 4.

Leading cause of posttransplant death in pre- and post-PSM of FSLT and WLT groups within 1 year.

Causes	Pre-PSM			Post-PSM
	FSLT (N = 113)	WLT (N = 124)	P	FSLT (N = 32)	WLT (N = 32)	P
Vascular complications	5(4.4)	0	0.024	3(9.4)	0	0.238
Myocardial infarction	1(0.9)	0	0.477	0	0	NA
MOF secondary to sarcopenia	3(2.7)	2(1.6)	0.671	1(3.1)	1(3.1)	1.000
Sepsis secondary to new coronavirus infection	0	1(0.8)	1.000	0	0	NA
Intracranial hemorrhage secondary to postoperative hemorrhage	1(0.9)	0	0.477	0	0	NA
Rapture of liver hematoma secondary to acute rejection	0	1(1.6)	1.000	0	0	NA
GVHD	1(0.9)	1(1.6)	1.000	0	0	NA
Tumor recurrence	2(1.8)	2(1.6)	1.000	1(3.1)	2(6.3)	1.000
EAD	1(0.9)	2(1.6)	1.000	1(3.1)	0	1.000
Total (%)	14(12.4)	9(7.3)	0.183	6(18.8)	3(9.4)	0.474

EAD, early allograft dysfunction; GVHD, graft-versus-host disease; MOF, multiple organ failure; PSM, propensity-score matching.

Because of the reduced sample size, significant differences between FSLT and WLT in the morbidity of vascular and biliary complications could not be demonstrated by the post-PSM analyses, except for the difference in the total number of postoperative complications graded ≥ III between the two groups (P < 0.001, Table 3).

Overall and graft survivals within 1 year of posttransplantation

By Kaplan–Meier analysis, the 1-, 2-, and 3-year OS were 87.6%, 81.0%, and 79.2% in the FSLT group, respectively, and 92.7%, 84.6%, and 79.2% in the WLT group, respectively (Fig. 2A). The 1-, 2-, and 3-year GS were 85.8%, 81.0%, and 79.3% in the FSLT group, respectively, and 91.9%, 84.8%, and 78.4% in the WLT group, respectively (Fig. 2B). As a primary study endpoint for short-term outcome assessment, 1-year patient and graft survivals were compared and tested between the two groups. Though the 1-year OS and GS of 87.6% and 85.8% in the FSLT group were lower than those in the WLT group, the differences between the two groups tested by log-rank did not reach statistical significance (P = 0.186 for OS and P = 0.136 for GS). Power analysis revealed that with the available sample sizes, the statistical power to detect the observed differences in 1-year OS and GS was 26.4% and 32.6%, respectively.

Figure 2.

Comparison of Kaplan–Meier curves depicting patients and graft survivals within 1 year of posttransplantation between full-size split liver transplantation (FSLT) and whole-liver transplantation (WLT) groups. (A) Overall survival. (B) Graft survival. (C) Overall survival after PSM. (D) Graft survival after PSM.

After PSM, the 1-, 2-, and 3-year OS were 81.3%, 81.3%, and 81.3% in the FSLT group and 90.6%, 84.3%, and 76.6% in the WLT group, respectively (Fig. 2C). The 1-, 2-, and 3-year GS were the same as the OS (Fig. 2D). After PSM, the 1-year OS and GS of 81.3% and 81.3% in the FSLT group were lower than those of 90.6% in the WLT group, with borderline significant differences and a statistical power of 19.2% (P = 0.081).

Risk factors for postoperative complications graded ≥ III

Univariate analysis using the demographic and clinical data of donors and recipients in the FSLT group revealed that a graft with steatosis ranging from 5% to 10%, a laboratory MELD score ≥20, sarcopenia, a history of hepatectomy, an operation time >420 min, and red blood cell (RBC) transfusion ≥6 U were risk factors for postoperative complications graded ≥III. A logistic regression model was constructed, including variables with a P-value ≤0.1 in the univariate analysis. Multivariate logistic regression analysis using forward LR identified a graft with steatosis ranging from 5% to 10% (OR 3.097, P = 0.030), a MELD score ≥20 (OR 5.830, P = 0.001), sarcopenia (OR 6.414, P = 0.002), a history of hepatectomy (OR 4.233, P = 0.010), and an operation time >420 min (OR 5.217, P = 0.009) as independent risk factors for the incidence of postoperative complications graded ≥III (Table 5).

Table 5.

Risk factor analysis for postoperative complications ≥ Grade III according to the Clavien–Dindo classification in FSLT group within 1 year of posttransplantation.

Variables	Postoperative complications ≥ Grade III
	Univariate analysis				Multivariate analysis
	Yes (N = 33)	No (N = 80)	OR (95%CI)	P	OR (95%CI)	P
Donor factors
Age (years)	42.0(33.5, 53.0)	43.0(32.0, 50.0)		0.677
Gender				1.000
Male	30(90.9)	71(88.8)
Female	3(9.1)	9(11.2)
BMI (kg/m)	24.5(23.3, 25.7)	24.1(22.5, 24.8)		0.204
ICU stay (days)	8.0(6, 13)	8.0(6, 12)		0.817
CPCR history	4(12.1)	22(27.5)	0.364(0.115–1.154)	0.077
Type of graft				0.789
Left hemiliver	16(48.5)	41(51.2)
Right hemiliver	17(51.5)	39(48.8)
Graft weight (gm)	830(651, 952)	753(636, 984)		0.462
Type of donor				0.772
DBD	24(72.7)	56(70.0)
DBCD (DCD)	9(27.3)	24(30.0)
Splitting in situ: ex vivo	23:10	55:25		0.921
Cold ischemia time (min)	289(246, 356)	277(230, 350)		0.288
Graft with steatosis 5%–10% (%)	15(45.5)	20(25)	2.500(1.067–5.860)	0.032	3.097(1.117–8.587)	0.030
Recipient factors
Age (years)	58.0(50.5, 65.0)	54.0(48.0, 61.0)		0.122
Gender				0.367
Male	22(66.7)	60(75.0)
Female	11(33.3)	20(25.0)
BMI (kg/m2)	22.2(20.4, 25.2)	21.9(19.2, 24.2)		0.347
Indications of LT
Benign:malignant	18:15	31:49		0.123
ABO incompatible (%)	1(3.0)	1(1.2)		0.501
GRWR (%)	1.4(1.1, 1.6)	1.3(1.1, 1.5)		0.701
MELD score ≥20	15(45.5)	13(16.2)	4.295(1.734–10.636)	0.001	5.830(1.878–18.098)	0.001
Sarcopenia	15(48.5)	17(27.5)	3.088(1.294–7.396)	0.009	6.414(1.941–21.194)	0.002
History of hepatectomy	15(45.5)	21(26.3)	2.341(1.004–5.462)	0.046	4.233(1.420–12.619)	0.010
Surgical factors
HA anastomosis (min)	18(13, 27)	14(12, 20)		0.107
Anhepatic time ≥48 min	21(63.6)	37(46.3)	2.034(0.883–4.685)	0.093
Implantation type				0.148
Conventional	3(9.1)	2(2.5)
Piggyback	30(90.9)	78(97.5)
PV thrombectomy (%)	9(27.3)	17(21.2)		0.489
Splenectomy (%)	5(15.1)	9(11.2)		0.546
Spleen Artery ligation	2(6.1)	5(6.2)		1.000
“T” tube drainage (%)	17(51.5)	42(50.0)		1.000
ORT >420 min	26(78.8)	45(56.3)	2.889(1.124–7.427)	0.024	5.217(1.524–17.860)	0.009
RBC transfusion ≥ 6 U	23(69.7)	38(47.5)	2.542(1.073–6.022)	0.031

BMI, body mass index; CPCR, cardiopulmonary cerebral resuscitation; DBCD, donation after brain and circulatory death; DBD, donation after brain death; DCD, donation after circulatory death; GRWR, graft-to-recipient weight ratio; FSLT, full-size split liver transplantation; HA, hepatic artery; ICU, intensive care unit; LT, liver transplantation; MELD, laboratory value of model for end stage liver disease; ORT, operation time; PV, portal vein; RBC, red blood cell; WLT, whole-liver transplantation.

Discussion

Due to the scarcity of liver grafts, FSLT for two adults has been continuously performed in a few experienced centers worldwide. However, most studies on FSLT typically spanned long periods and involved a limited number of cases^[3,6,8–12^]. To address the gap between the supply and demand of donor livers, we initiated a program of CSLT followed by FSLT in 2021^[19^]. By the end of 2023, a total of 113 consecutive cases of FSLT, after excluding one case of re-transplantation, were completed. To the best of our knowledge, this study represents one of the largest series of FSLTs conducted within 3 years in a single center, with 1-year OS and GS of 87.6% and 85.8%, respectively, demonstrating relatively better short-term outcomes compared to previously published data^[6,8–12^]. Furthermore, the rate of graft full-size splitting was 40.1% (61/152) in this study, which we believe sets one of the highest graft splitting rates for deceased donors to date^[8–12,14,15^]. Meanwhile, this study identified independent risk factors for postoperative complications graded ≥ III, a finding of great significance for improving the efficacy of FSLT.

Due to the different inclusion and exclusion criteria for both donors and recipients of FSLT and WLT, as well as the inherent limitation of a retrospective study, there are significant baseline differences between the two groups in terms of gender, body weight, donor type, MELD scores, sarcopenia prevalence, etc. These imbalances directly undermine comparability. To address this issue, a PSM was performed using parameters such as donor type (DBD vs. DCD or DBCD), CIT of the graft, recipient weight, recipient laboratory MELD score, and sarcopenia. These parameters were shown to be key prognostic factors related to the postoperative morbidity and mortality of LT in previous studies^[6,7,11,12^]. After PSM, although the sample size was reduced to 32 cases for each group, only donor weight, graft procurement time, the proportion of recipients with HBsAg (+), and the GRWR remained statistically different between the groups. Among them, a lower GRWR is an inherent characteristic of FSLT, and the others have no significant impact on the results reported previously^[6,7,11,12^]. The PSM procedure enhances the interpretability and robustness of the results of this study (Table 1).

Currently, there is no consensus among transplant organizations regarding the criteria for the selection of donors and liver grafts for splitting^[3^]. In Italy, a mandatory-split policy has been adopted since August 2015, defining “splittable” donors as those aged 10–50 years with stable hemodynamics, an intensive care unit (ICU) stay ≤ 5 days, transaminases ≤ 3 times the upper limit of normal, and the absence of steatosis on ultrasound scan^[5^]. These criteria are primarily intended for pediatric patients and may be stricter than those necessary for adults. The SLT rate varies worldwide. In Europe, SLT represents approximately 6% of all LT^[5,37^]. In the United States, it comprises approximately 1% of all LTs, although it is estimated that 20% of deceased donors meet the United Network for Organ Sharing guidelines for split livers^[14^]. In this study, we extended the donor age to 60 years and allowed for liver steatosis from 0 to 10%. As a prerequisite for excluding systemic infections, the ICU stay was not strictly limited. The rate of liver graft splitting for FSLT in this study was much higher than that reported in the literature, thereby providing more opportunities for patients on the waiting list for LT^[8–12,14,15^]. Subsequent univariate and multivariate analyses of donor factors in this study revealed that only a graft with steatosis from 5% to 10% was an independent risk factor (odds ratio [OR] 3.097, P = 0.030) for serious postoperative complications (Table 5). To date, the criteria for graft splitting lack a sound evidentiary foundation. Further prospective multicenter studies on the development of a risk-scoring model combining characteristics associated with graft failure for the selection of “splittable” liver grafts are of great significance.

Full-size SLT may especially benefit smaller adult patients who are at a disadvantage on the waiting list for receiving timely size-matched grafts^[38^]. East Asian female patients may particularly benefit from this technology because they are often short, of small stature, and have low body weight. The favorable outcomes observed in this study may be partially attributed to the high proportion of female recipients with low body weight. Eleven female recipients weighing ≤ 50 kg were matched with left hemiliver grafts during the study period. The first recipient, a 55-year-old woman weighing 45 kg, presented with refractory ascites and recurrent hepatic encephalopathy caused by cirrhosis secondary to autoimmune hepatitis. She had been on the LT waitlist for 26 months without finding a suitable graft until the FSLT technique was considered an opportunity. On 30 March 2021, the patient underwent FSLT using a left hemiliver graft weighing 451 g (GRWR 1.0). Refractory ascites persisted postoperatively; however, it disappeared synchronously with the gradual proliferation of the implanted hemiliver. The patient was discharged on postoperative day 25 and was followed up for 45 months with an excellent quality of life (Fig. 3).

Figure 3.

Clinical case illustration: A 55-year-old woman weighing 45 kg with end-stage cirrhosis due to autoimmune hepatitis underwent FSLT using a left hemiliver graft weighing 451 g on 30 March 2021. Refractory ascites persisted postoperatively but resolved gradually alongside the proliferation of the hemiliver graft. The patient was discharged on postoperative day 25 and followed up for 45 months with an excellent quality of life. (A, B) Preoperative computed tomography (CT) images showing cirrhosis, ascites, and portal hypertension with varices (white arrow). (C) Excised cirrhotic liver. (D) Left hemiliver graft weighing 451 g. (E, F) CT images showing the transplanted liver on postoperative day 7 with an estimated volume of 758 mL (+ 68.1%). (G, H) CT images showing the transplanted liver on postoperative day 160 with an estimated volume of 948 mL (+ 102.0%). (I, J) Magnetic resonance imaging (MRI) showing the implanted liver on postoperative day 760 with a stable volume compared to the CT images on postoperative day 160.

Splitting an intact graft into full hemiliver grafts for two adults predisposes recipients to a higher rate of complications, stemming from anatomic variation, smaller graft size, and difficulty in vascular and biliary reconstruction. Various postoperative complications graded ≥ III were observed in 33 recipients (29.2%) in the FSLT group, with no significant difference compared with 25 recipients (20.2%) in the WLT group (P = 0.106). However, the total number of such complications was much higher in the FSLT group than in the WLT group (67.3% vs. 39.5%, P < 0.001, Table 3). Among such events, the overall incidences of biliary (11.5% vs. 4.0%, P = 0.040) and vascular complications (12.4% vs. 1.6%, P = 0.001) in the FSLT group were significantly higher than those in the WLT group, leading to five deaths in the FSLT group (Table 4). The learning curve of the FSLT project in the early stage of this study may have significantly contributed to these events. The inappropriate application of the conventional orthotopic implantation technique to two recipients with left hemiliver grafts in the initial stage of this program resulted in compression and distortion of the IVC caused by Spiegel’s lobe, leading to the death of one recipient and multiple serious complications in another. Regardless of the type of graft or biliary reconstruction, the overall incidence of biliary complications in LDLT ranges from 7.4% to 39%^[39^]. As approximately half of the split hemiliver grafts lack a CBD, the high incidence of biliary complications, known as the “Achilles’ heel” in FSLT, is understandable. To address this issue, external biliary drainage with T-tube stenting was performed for small bile ducts ≤ 3 mm during the later stage of this study, and a downward trend in biliary complications was observed. Also, the accumulated technical knowledge and experience from LDLT and the initial phase of the FSLT program actively facilitate the reduction of vascular complications, particularly in reducing HA complications. Therefore, it must be recognized that FSLT remains a high-risk procedure and should be performed in experienced centers with skilled teams.

The literature consistently demonstrates that the short-term outcomes of FSLT were significantly worse than those of WLT because of the high incidence of serious postoperative complications^[3–12^]. Chen et al reported 3-month and 1-year mortality rates of 21% and 31% for 100 cases of FSLT^[11^]. Aseni et al presented a 3-month mortality rate of 15.6%, with 1-year graft and patient survival rates of 63.3% and 73.2% due to a high postoperative complication rate (64.1% grade III and IV) for 64 recipients who underwent FSLT. This was compared with 85.8% and 87.2% for a case–control group matched by donor age and recipient MELD score of 1199 patients who received a WLT in the same period^[12^]. Pooled proportions of vascular and biliary complications reported by a meta-analysis were 6.9% (95% CI: 3.1–10.7%, n = 273) and 25.6% (95% CI: 19–32%, n = 308), respectively. One-year graft and patient survival rates were 75.7% and 80.0% (95% CI: 74.3–85.7, n = 331)^[6^]. As the grafts used for FSLT are usually of good quality and the majority of these studies did not correct for confounding factors when comparing results after FSLT and WLT, risks associated with FSLT may be slightly underestimated compared with adequately matched WLT grafts^[4,6,8–10^]. The 1-year patient and graft survival rates of 87.6% and 85.8% for FSLT in this study have been confirmed by a follow-up with a median of 23.0 months (Q1, 15.5; Q3, 33.0). These have no statistical significance compared with the survival rates of 92.7% and 91.9% for the control group of WLT (P > 0.05). After PSM, the 1-year OS and GS of FSLT were 81.3% and 81.3% compared with 90.6% and 90.6% in the WLT group (P = 0.081).

It may raise the question of why the increased surgical complication rates did not translate into worse survival outcomes in this study. Actually, the leading cause of postoperative death was vascular complications for patients undergoing FSLT, compared with none in those undergoing WLT (Table 4). The absolute differences in 1-year OS and GS between the two groups increased from 5%–6% to 9% after PSM, which is clinically relevant. The statistical power to detect the observed differences was 26.4% for OS and 32.6% for GS, respectively, and it was reduced to 19.2% for both OS and GS after PSM. Taking all of this into consideration, differences in 1-year OS and GS between the two groups might reach statistical significance with larger sample sizes, demonstrating that the short-term outcomes of FSLT are somewhat inferior to those of WLT at present. In addition, prioritizing grafts that meet the established criteria for the split procedure may lead to a reduction in high-quality grafts available for WLT recipients and undermine the prognosis of WLT patients compared with the established benchmark of > 90% for both 1-year OS and GS in low-risk WLT cases^[40^]. Meanwhile, the enhanced perioperative management and complication-handling capabilities developed in recent years may also contribute to the improved 1-year survival compared to previously published data. Nevertheless, considering the considerable incidence of waitlist mortality, our encouraging results justify using FSLT whenever feasible to expand the donor pool and benefit patients on the waiting list.

The immunosuppressive strategy has a significant impact on the outcomes of recipients. According to the guidelines for LDLT recipients, modification of immunosuppression protocols for LDLT versus DDLT is not necessary^[41^]. In this study, the same immunosuppressive protocols are used for both groups of patients in FSLT and WLT. However, it has been shown that LDLT recipients with partial grafts achieve higher blood levels of tacrolimus for a given dose compared with DDLT recipients with whole-liver grafts^[30,31^]. Therefore, in this study, an initial tacrolimus dose reduction of about 50% in FSLT compared with WLT is implemented, and there is no significant difference in graft rejection between the groups. The alteration in the pharmacokinetics of blood tacrolimus levels in FSLT compared with WLT warrants further study^[42^].

Most reports in the literature lack risk factor analyses for serious postoperative complications of FSLT^[8–10^]. A multivariate analysis using a logistic regression model on 100 cases of FSLT only found that a graft weight less than 580 g was an independent risk factor for the occurrence of early mortality^[11^]. Aseni et al, using a Cox proportional hazard model on 64 recipients who underwent FSLT, indicated that donor age, female recipient sex, body weight, and HCV cirrhosis negatively affected patients’ 5-year survival^[12^]. The small sample size and long time span for FSLT performance reduced the impact of these studies. Our multivariate analysis using a logistic regression model, including variables with a P-value ≤ 0.1 in univariate analysis, identified that a graft with steatosis from 5% to 10% (OR 3.097, P = 0.030), MELD score ≥ 20 (OR 5.830, P = 0.001), sarcopenia (OR 6.414, P = 0.002), history of hepatectomy (OR 4.233, P = 0.010), and operation time > 420 min (OR 5.217, P = 0.009) were all independent risk factors for the incidence of postoperative complications graded ≥ III. As a marginal graft, the use of SLT in high-risk adult recipients with either high laboratory MELD scores (≥25–30) or sarcopenia remains controversial^[9,10,43–46^]. Since a MELD score greater than 30 was reported to be accompanied by the worst outcome after LT^[47^], recipients with a high MELD score (≥30) were prohibited from accepting hemiliver splitting in some studies^[9,45^]. The proportion of recipients with HCC was 56.6%, with a median (Q1, Q3) of the laboratory MELD score of 9.0 (7.0, 12.8), and the median of the FSLT group in total was 12 (8, 17). There were only two cases with scores over 30. Therefore, a MELD score ≥ 20 was considered an appropriate threshold in this study. Sarcopenia is associated with poor clinical outcomes in various clinical scenarios, including LT^[48–50^]. A multicenter retrospective study including 240 cases of SLT showed a 90-day mortality of 28.8% for recipients with sarcopenia^[51^]. Notably, decompensated cirrhotic patients with low body weight on the waiting list are often accompanied by sarcopenia. The ratio of recipients with sarcopenia defined according to Kong et al^[33^] in the FSLT group is significantly higher than that in the WLT group in this study. Under the current circumstances of an extreme lack of liver grafts and the “sickest first” allocation policy, patients with either high MELD scores or sarcopenia have more opportunities to be candidates for FSLT. Therefore, the performance of the FSLT for patients with high risks must be considered to maintain a balance between the benefits and risks of this procedure. Moreover, indications for high-risk recipients should be personalized, guided by the general indications of LT. Apart from recipient selection, another independent risk factor for postoperative morbidity is long operation time, which might be related to the history of hepatectomy or abnormalities of vascular anatomy and may serve as an early warning factor for the occurrence of severe postoperative complications, though it has limited value in the decision to perform a split. Donor age has no impact on the incidence of postoperative complications graded ≥ III, even though older donors (45–55 years) have been reported to be a negative factor for 1-year pediatric recipient and graft survival^[52^]. This study indicates that we are still in the learning curve phase, and further accumulation of experience and refinement of surgical techniques may improve FSLT outcomes.

The major limitations of this study are its retrospective design and the lack of an adequately matched case–control group, which undermine the comparability of outcomes. Second, the limited sample size, which decreased significantly after PSM using key prognostic factors, reduced the statistical power of this study. Third, these monocentric results from patients mainly in the southeast of China, which may not be generalized to centers with different populations (e.g., in terms of body weight and BMI). To reach a consensus on FSLT, a well-designed prospective study conducted by multiple centers with larger sample sizes and longer follow-up is required.

Conclusions

FSLT is a promising solution to the shortage of liver donors and can improve the availability of LT in adult patients with low body weight. While selection bias in donors and recipients may have affected the results, the short-term outcomes of FSLT in this study are superior to the published data and seem acceptable when compared with those of unmatched and inadequately matched patients who underwent WLT. It is essential to establish a risk assessment model for “splittable” liver grafts and recipient selection through a multicenter study under a strict mandatory split-liver allocation policy. Additionally, this study emphasizes the importance of refining surgical techniques for vascular and biliary reconstruction to reduce the incidence of postoperative complications and mortality. For FSLT, recipients with a history of hepatectomy, sarcopenia, a MELD score ≥ 20, and who have received grafts with 5%–10% steatosis are at high risk of serious postoperative complications. To ensure the success of FSLT, an optimal combination of recipient, donor, and transplant-center factors is crucial. Considering that the implementation of this project is still in the preliminary stage with immature technology, the learning curve persists, and many aspects remain to be improved.

Ethical approval

This retrospective study meeting the ethical standards of the World Medical Association Declaration of Helsinki was approved by the ethics committee of Ningbo Medical Center Lihuili Hospital (Reference number: KY2022PJ019).

Consent

Written informed consent was obtained from the patients.

Conflicts of interest disclosure

The authors declare no competing interests.

Sources of funding

This work was supported by the Ningbo Major Research and Development Plan Project (2024Z179), Ningbo Top Medical and Health Research Program (2024020818), Ningbo Major Research and Development Plan Project (2023Z160), General Program of National Natural Science Foundation of China (82470683).

Author contributions

C.D.L and S.D.W: conceptualization, liver transplantation performance, data analysis and interpretation, writing and revising the manuscript. X.X. conceptualization, data interpretation, writing – review. J.H and J.Z.F: liver transplantation performance, data collection. K.W and L.B.D: interpretation, and writing –revised. S.Q.M: data collection, analysis and interpretation, and writing –review and editing. C.J.L, K.W, W.J and H.D.Z.: data collection. All authors read and approved the final manuscript.

Research registration unique identifying number (UIN)

NCT06925828.

Guarantor

Caide Lu.

Provenance and peer review

Not commissioned.

Data availability statement

The data supporting the results in the manuscript can be obtained from the corresponding author based on reasonable request.