This was a retrospective cohort design study. We included adult (18 years of age and older) patients diagnosed with HCC preoperatively who underwent CLT or IFLT at The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China, between June 2017 and August 2020. Split LT, liver-kidney combined transplantation or multivisceral transplantation were excluded, and patients who died within 30 days of LT were also excluded. Because patients who underwent IFLT only received donors from donation after brain death (DBD), donation after circulatory death (DCD) or living donors were excluded from the current study. Thus, patients identified as having HCC were chosen for the study population (n=226); 30 of 226 patients received IFLT, and the remaining 196 patients received CLT. All 226 donors were enrolled in a voluntary organ donation program for deceased Chinese citizens, and informed consent was obtained from relatives of the donors. No organ donations were from executed prisoners. This study was approved by the Institutional Review Board of the First Affiliated Hospital of Sun Yat-sen University.

The donor variables were age, BMI, donor serum creatinine, donor total bilirubin, donor serum sodium, and cold ischemia time. The recipient variables were age at transplant, BMI, preoperative laboratory MELD score, and positive hepatitis B surface antigen. Tumor parameters were pretransplant AFP, most radiologic tumor diameter, number of lesions (1,2, 3+), Milan criteria, liver resection history, neoadjuvant therapy (RFA or TACE), tumor differentiation (well, moderate, poor), and microvascular invasion. Intraoperative and posttransplantation data included operation duration, anhepatic phase, total blood loss, blood transfusion, ICU stay, hospital stay, early allograft dysfunction (EAD), serum INR, lactate, ALT, AST and creatinine level on the first day posttransplantation, recurrence, date of recurrence, and site of recurrence.

To evaluate whether the outcomes of HCC recipients who underwent IFLT or CLT were different, propensity-matched analyses were performed. IFLT recipients were matched 1:3 with patients who had undergone CLT during the same time period utilizing a propensity match score, matching for the following variables: pretransplant tumor characteristics (serum AFP, the most radiologic tumor diameter, multiple/single, portal vein tumor thrombosis, liver resection history, RFA/TACE neoadjuvant therapy) and explant tumor characteristics (differentiation and microvascular invasion). The caliper width was 0.2 standard deviations of the logit-transformed propensity score. The absolute standardized differences method was used to diagnose the balance after matching, and all were confirmed to be less than 0.25.

The primary endpoint of this study was tumor recurrence. Secondary endpoints included operation time, intensive care and hospital stay, EAD, serum ALT and AST levels on the first day posttransplantation, histological analysis of liver tissues before procurement, at the end of preservation and after revascularization, and overall survival. EAD was defined by the presence of one or more of the following after LT (16): bilirubin of ≥10 mg/dL on Day 7, international normalized ratio of ≥1.6 on Day 7, and alanine aminotransferase or aspartate aminotransferase >2000 IU/L within the first 7 days. To estimate overall survival, survival time was calculated from the date of LT to death or last known follow-up, and status was recorded at the last point of contact with the patient who died or lived. To estimate recurrence-free survival (RFS), patients with no evidence of recurrence were censored at the time, of last follow-up or death.

The CLT procedure included organ procurement after rapid cold flush, subsequent static cold storage (SCS) and back-table preparation, and then implantation. The surgical procedures of IFLT were as described in our previous study (14, 15). Briefly, after the liver was fully mobilized, a 4 cm-long segment of the external iliac vein from the blood donor was harvested and end-to-side anastomosed to the portal vein of liver donors, serving as an access point for portal vein cannulation while still permitting native blood flow through the portal vein. A 12 Fr cannula was inserted into the splenic artery (or gastroduodenal artery) without interruption of arterial supply to the liver from the celiac artery. A 32 Fr cannula was placed in the infrahepatic inferior vena cava for outflow. A straight 24 Fr cannula connected to the portal vein perfusion line of Liver Assist (Organ Assist, Groningen, The Netherlands) was inserted into the portal vein via the interposition vein. The arterial cannula was then connected to the hepatic artery (HA) perfusion line of Liver Assist. After the in situ normothermic mechanical perfusion (NMP) circuit for the livers was established, the livers were harvested and moved to the organ reservoir under continuous NMP. So that the graft did not suffer ischemia during procurement. Hereafter, the liver underwent ex situ NMP. The perfusate contained approximately 1.3 L cross-matched leucocyte-depleted washed red cells, 1.4 L Succinylated gelatinor, 30 mL 5% sodium bicarbonate, 0.5 g metronidazole, 37500 U heparin, 1.5 g cefoperazone sodium and sulbactam sodium, 30 mL 10% calcium gluconate, 3 mL 25% magnesium sulfate and 250 mL compound amino acid injection. In IFLT, no back-table preparation is needed, and the liver is implanted under continuous NMP. Briefly, the splenic artery (or gastroduodenal artery) and interposition vein and all vascular anastomoses (donor suprahepatic vena cava to recipient suprahepatic vena cava, donor portal vein to recipient portal vein, and donor celiac artery/common hepatic artery to recipient common hepatic artery) were performed without interruption of the blood supply to the graft under continuous NMP. After the native blood supply from the recipient’s portal vein and hepatic artery to the graft was re-established, the in situ NMP was stopped, and the cannulas were removed. The donor splenic artery (or gastroduodenal artery) was ligated closed, and the interposition vein was sutured closed. The donor infrahepatic vena cava and common bile duct were anastomosed to the recipient counterparts. Graphic rendering of the ischemia-free liver transplantation procedure is shown in Figure 1 .

In ex vivo perfusion, samples from the perfusate were collected for analysis of the blood gas parameters (pO2, pCO2, PH and lactate). Perfusate samples were also collected and centrifuged, and the supernatant was stored at -80°C for liver function tests, such as aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total bilirubin (Tbil), using standard biochemical methods. Bile production was collected every 60 minutes from the biliary draining tube for bile production calculation and detection of PH and HCO3^- of bile. Liver tissue biopsy was performed for histology studies before procurement, at the end of preservation and after revascularization, including hematoxylin and eosin (HE) staining for histological scoring and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay for hepatocyte apoptosis.

After CLT or IFLT, all patients were admitted to the intensive care unit. The immunosuppressive regimen was 20 mg basiliximab induction therapy administered during the operation and on postoperative Day 4. Both a calcineurin inhibitor and mycophenolate mofetil were administered for immunosuppressive maintenance therapy beginning on postoperative Day 4.

The donor and recipient characteristics were expressed as the means ± standard deviation (SD) for metric parameters and as percentages for nominal parameters. Continuous data were compared using t-tests, whereas categorical variables were compared using the chi-square test or Fisher’s exact test. We used Kaplan-Meier statistics and the log-rank test to analyze the recurrence-free survival (RFS) and overall survival (OS) rates of the patients both in the entire cohort and propensity-matched cohorts. To identify independent predictors of RFS in entire cohort, an univariate and multivariate Cox proportional hazard regression was conducted for all candidate predictors of RFS. Pretransplant serum AFP levels were categorized into two groups according to interquartile range 3 (IQR3). Variables with a P < 0.05 in univariate analysis were subjected to multivariate Cox proportional hazards regression via the forward stepwise method; the results were presented as hazard ratios (HRs) with 95% confidence intervals (CIs). The data analysis was performed using STATA 14.0 software (Stata Corp). P <0.05 was considered to be statistically significant.

Patients identified as having HCC were chosen for the study population (n=226) between June 2017 and August 2020 in the entire cohort; 30 of 226 patients received IFLT, and the remaining 196 patients received CLT. To correct selection biases and confounding factors, propensity score matching was performed at a 1:3 ratio. After matching, 30 IFLT recipients and 85 matched CLT patients from our center were enrolled in propensity-matched cohorts during the same period. In the entire cohort, the duration of follow-up was 22.9 ± 11.2 months in the IFLT group and 22.6 ± 11.4 months in the CLT group with no significant differences, securing a minimal follow-up of 6 months in the two groups.

There were no significant differences in donor and recipient characteristics between the IFLT group and the CLT group both in the entire cohort and in the propensity-matched cohort, including donor characteristics (age, gender, donor BMI, donor serum creatinine, donor serum total bilirubin and donor serum sodium) and recipient characteristics (age, gender, BMI, laboratory MELD scores and positive hepatitis B surface antigen). The cold ischemia time in the CLT group was 6.7 ± 2.1 hours in the entire cohort and 6.8 ± 2.2 hours in the propensity-matched cohort. Donor and recipient characteristics were summarized in Table 1 .

In the entire cohort, the pretransplant AFP level was higher in the CLT group than in the IFLT group (P=0.016). The percentage of LT recipients within the Milan criteria and microvascular invasion was higher in the CLT group than in the IFLT group (P=0.032 and P = 0.042, respectively). There were no differences in the size largest HCC lesion, number of lesions, tumor differentiation, liver resection history and neoadjuvant therapy (RFA or TACE) history between the two groups. In the propensity-matched cohort, there were no differences in all tumor parameters between the two groups. Comparison of tumor parameters were summarized in Table 1 .

To analysis whether IFLT has potential as an independent prognostic factor in HCC patients following liver transplantation, Cox regression analysis was performed to examine RFS in the entire cohort. Univariate analysis indicated that pretransplant AFP (≥ 300 ug/l vs. < 300 ug/l)(HR 3.830, 95% CI 2.447-5.997, P <0.001), biggest HCC diameter (≥ 5cm vs. < 5cm) (HR 1.753, 95% CI 1.119-2.746, P =0.014), poor tumor differentiation (HR 2.738, 95% CI 1.226-6.112, P = 0.014), microvascular invasion (HR 3.453, 95% CI 2.213-5.388, P <0.001) and surgical procedure (CLT vs IFLT) (HR 4.371, 95% CI 1.371-13.864, P =0.012) were associated with RFS. Furthermore, multivariate analysis demonstrated that pretransplant AFP (≥ 300 ug/l vs. < 300 ug/l) (HR 2.262, 95% CI 1.597-4.318, P <0.001), microvascular invasion (HR 2.309, 95% CI 1.403-3.801, P <0.001) and surgical procedure (CLT vs IFLT) (HR 3.728, 95% CI 1.172-11.861, P =0.026) were independent prognostic factors for DFS in HCC patients following liver transplantation ( Table 2 ).

In the entire cohort, Kaplan–Meier analysis showed the RFS rates at 1 and 3 years after LT in recipients with HCC in the IFLT group were 92.2% and 86.7%, respectively, which were significantly higher than those (73.0% and 46.3%) in the CLT group (P=0.006, Figure 2A ). The overall survival rates at 1 and 3 years after LT in recipients with HCC in the IFLT group were 96.7% and 90.6%, respectively, which tended to be higher than those (90.2% and 68.1%) in the CLT group, but with no significant differences (P=0.089, Figure 2B ). In the propensity-matched cohort, the RFS rates at 1 and 3 years after LT in recipients with HCC in the IFLT group were 92.2% and 86.7%, respectively, which were significantly higher than those (88.1% and 53.6%, respectively) in the CLT group (P=0.048, Figure 2C ). The overall survival rates at 1 and 3 years after LT in recipients with HCC in the IFLT group were 96.7% and 90.6%, respectively, which tended to be higher than those (94.1% and 70.6%) in the CLT group, but with no significant differences (P=0.442, Figure 2D ). These results indicate that IFLT provides greater benefits than CLT in terms of the reduction in post-LT HCC recurrence.

In the entire cohort, LT recipients with HCC in the IFLT group had a shorter operation duration than LT recipients with HCC in the CLT group (6.3 ± 1.4 hours vs. 6.9 ± 1.5 hours, P=0.016). The anhepatic time was not different between the IFLT and CLT groups (52.2 ± 16.9 mins vs. 53.3 ± 14.8 mins, P=0.980). There were no differences in total blood loss or blood transfusion between the two groups. The ICU stay (52.4 ± 50.7 hours vs. 53.8 ± 50.2 hours, P=0.895) and hospital stay (23.8 ± 17.6 days vs. 25.8 ± 15.1 days, P=0.772) were similar between the two groups. LT recipients with HCC in the IFLT group had a lower incidence of EAD than LT recipients with HCC in the CLT group (3.3% vs. 29.6%, P=0.002). Patients in the IFLT group had a lower serum lactate level (1.9 ± 1.2 mmol/L vs. 2.7 ± 1.3 mmol/L, P=0.005), lower serum ALT level (198.8 ± 157.9 U/L vs. 633.8 ± 706.2 U/L, P=0.001) and serum AST level (437.1 ± 328.9 U/L vs. 1571.6 ± 1764.6 U/L, P=0.001) on postoperative Day 1. There were no differences in serum creatine levels or serum INR levels between the two groups on postoperative Day 1.

In the propensity-matched cohort, the indicators mentioned above had the same trend. Operative and postoperative outcomes are summarized in Table 3 .

The NMP device provided adequate O2 and extraction of CO2 of the perfusion fluid with stable pressure and flow of both the portal vein and hepatic artery throughout the whole IFLT procedure ( Figures 3A, B ). The lactate levels in the perfusate dropped quickly from 4.73 ± 2.22 mmol/L to normal, and the pH value in the perfusate was within the normal physiological range, reflecting active metabolism by the liver grafts ( Figure 3C ). The liver grafts presented a vivid appearance during procurement, ex vivo NMP and implantation ( Figure 3D ). Continuous bile production with a high sodium bicarbonate and pH level indicated good quality of the bile ( Figures 3E, F ). Altogether, these results indicate the effectiveness of IFLT and suggest excellent organ viability.

Hematoxylin and eosin (HE) staining evaluation of IFLT allograft biopsies showed stable and low Suzuki scores, whereas increased Suzuki scores were observed in the CLT allograft biopsies at the end of preservation graft and after revascularization ( Figure 4A ). The TUNEL assay showed no significant increase in apoptotic hepatocytes throughout the whole IFLT procedure in the IFLT group, while a significant increase in apoptotic hepatocytes was observed at the end of the preservation graft and after revascularization ( Figure 4B ). These results suggest that IFLT can largely reduce IRI in LT recipients with HCC.