In this retrospective cohort study, patients who underwent LLR or OLR were selected at two tertiary hospitals, the Eastern Hepatobiliary Surgery Hospital (EHBH) and Mengchao Hepatobiliary Hospital of Fujian Medical University (MHBHFMU), from January 2014 to August 2021. A total of 3171 HCC patients were included in this cohort, including 501 laparoscopic hepatectomy and 2670 open hepatectomy. This study was conducted in accordance with the Helsinki Declaration and approved by the Clinical Research Ethics Committees of the EHBH and MHBHFMU. Patients with BCLC stage A large HCC located in difficult segments who underwent hepatectomy were enrolled in our study. The inclusion criteria in this study were as follows: 1) cases with HCC in a difficult location (segment I, IVa, VII, VIII); 2) BCLC stage A; and 3) cases with HCC ≥5 cm. The exclusion criteria were as follows: 1) patients with vascular invasion or extrahepatic metastasis; 2) insufficient data; and 3) cases lost to follow-up. Our study divided patients into a LLR cohort and an OLR cohort based on surgical approaches. The choice of surgical approach was based on full discussion between the patient and the surgeon, in accordance with the wishes of the patient. All operations were performed by experienced surgeons.

As described in our previous study (12), routine preoperative assessments consisted of liver dynamic computed tomography (CT), chest CT and bone scanning, as well as serological indicators related to the disease. Multidisciplinary treatment (MDT) meetings (13) were held weekly to discuss the optimal treatment options of patients. Experts from all relevant disciplines will attend, including liver surgeons, gastroenterologists, imageologists and oncologists.

Laparoscopic hepatectomy was performed as previously reported (10, 14). When the lesions were located in segments VII and VIII, patients were placed in a supine position, and the operating table was tilted 15-45° to the left during the operation. For lesions located in segment I and the superior part of segment IV, the lithotomy position was adopted in the surgical procedure. Ultrasonic shears (harmonic scalpel; Ethicon Endo-Surgery Inc., Cincinnati, OH, USA), Cavitron ultrasonic surgical aspirator (CUSA; ValleyLab Inc., Boulder, CO, USA), and LigaSure (Valley-Lab Inc.) were applied in the parenchymal transection. Large blood vessels were ligated with Hem-O-lock clips (Teleflex Medical, Research Triangle Park, NC, USA). The liver section was haemostatic with fibrin glue sealant (Greenplast, Green Cross Corp., Seoul, Korea).

The right subcostal incision extending to the midline was applied in the OLR. Parenchymal transection was conducted with an ultrasonic scalpel. If necessary, Pringle’s manoeuvre and infrahepatic vena cava clamping were used.

In the first year after discharge, all patients were followed up every 2 months and then every 3 months until death or loss to follow-up. Routine postoperative follow-up consisted of liver dynamic CT, chest CT, and laboratory indices. HCC recurrence was identified according to CT and elevated serum AFP levels.

OS and RFS were the primary outcomes. Secondary outcomes were intraoperative outcomes (intraoperative blood loss; intraoperative blood transfusion; surgical margin) and postoperative short-term outcomes (hospital stay after surgery; 30-day mortality; postoperative complications).

Bile leakage was defined based on the International Study Group of Liver Surgery (15). Postoperative daily abdominal drainage of more than 10 ml/kg was defined as ascites (16). Fluid in the thoracic cavity with atelectasis requiring percutaneous drainage was defined as pleural effusion. OS was defined as the time from the date of surgery to the date of death or last recorded visit. The time from the date of surgery to the date of the first diagnosis of HCC recurrence or the last follow-up was calculated as RFS.

In this study, continuous data with a normal distribution were compared by two-sided Student’s t tests, while continuous data without a normal distribution were compared using the Mann–Whitney U test. The chi-square test or Fisher’s exact test was adopted in the comparison of categorical data.

Propensity score matching (PSM) analysis was performed in a 7:12 matching to overcome bias in the two cohorts and balance the baseline characteristics using the nearest-neighbour matching method with no replacement. The following variables were matched: age, sex, tumour location, maximum tumour diameter, platelets, total bilirubin, alanine aminotransferase (ALT), albumin, prothrombin time, γ-glutamyl transferase, alpha fetoprotein (AFP), protein-IIinduced by vitamin K absence (PIVKA-II), hepatitis B surface antigen (HBsAg), hepatitis B virus deoxyribonucleic acid (HBV DNA), satellite nodule, tumour capsule, tumour differentiation, microvascular invasion (MVI) and liver fibrosis. We set the calliper width to 0.2 standard deviations to ensure good matching.

OS and RFS are shown by means of the Kaplan–Meier method and compared with the log rank test. Independent risk factors were identified by univariate and multivariate Cox regression. P <0.05 was considered statistically significant. R software (version 4.2.0, Vienna, Austria; packages: Survival and Survminer) and SPSS software (version 19.0, IBM, Armonk, New York, USA) were applied in the statistical analysis.

Table 1 shows the baseline characteristics of patients before and after PSM. Before PSM, compared with the OLR cohort, the patients in the LLR cohort were older (57.5 vs. 52, p=0.035), had a smaller tumour maximum diameter (6 vs. 8.1, p<0.001), lower platelet counts (164.5 vs. 177, p=0.039), higher levels of total bilirubin (15.8 vs. 12.9, p=0.014), longer prothrombin time (12.7 vs. 11.7, p<0.001), lower levels of PIVKA-II (1143 vs. 2495, p=0.03), higher percent of HBsAg positivity (88.9% vs. 75.1%, p=0.022), less positive Satellite nodule status (29.6% vs. 44.8%, p=0.036), lower degree of tumour differentiation (Grade 2 percent: 16.7% vs. 4.8%, p=0.005), more positive MVI status (M1 and M0: 74% vs. 45.4%, p<0.001) and a lower percent of liver fibrosis (40.7% vs. 85.2%, p<0.001). After PSM, there was no significant difference in baseline characteristics for these two cohorts, and all of the characteristics were comparable (all p>0.05) ( Table 1 ).

Before PSM, the OS probabilities at 1, 2 and 3 years were 87.1%, 71.4% and 62.6%, respectively, and the cumulative recurrence probabilities at 1, 2 and 3 years were 23.4%, 36.0% and 43.5%, respectively, for the OLR cohort. In the LLR cohort, the OS probabilities at 1, 2 and 3 years were 96.2%, 87.6% and 85.2%, respectively, and the cumulative recurrence probabilities at 1, 2 and 3 years were 26.1%, 35.2% and 38.4%, respectively. These results indicated that the long-term survival outcomes between the LLR and OLR cohorts before PSM were not significantly different (both p>0.05) ( Figures 2A, B ; Table 2 ).

After PSM, the OS probability at 1, 2 and 3 years was 83.8%, 69.5% and 64.8%, respectively, and the cumulative recurrence probability at 1, 2 and 3 years was 26.6%, 36.5% and 50.3%, respectively, for the OLR cohort. In the LLR cohort, the OS probabilities at 1, 2 and 3 years were 97.1%, 82.8% and 69.7%, respectively, and the cumulative recurrence probabilities at 1, 2 and 3 years were 20.3%, 35.6% and 41.0%, respectively. These results indicated that the long-term survival outcomes between the LLR and OLR cohorts after PSM were not significantly different (both p>0.05) ( Figures 2C, D ; Table 2 ).

Survival analysis was performed to explore whether there were survival differences in the LLR and OLR subgroups cohorts after PSM. According to the clinical characteristic features, we divided the patients into 6 subgroups, which included age (≤60 years or >60 years), liver fibrosis (+ or -), satellite nodule (+ or -), tumour capsule (partial or intact), MVI (+ or -), and surgical margin (≤1 cm or >1 cm). However, no significant differences were found between these two cohorts in OS probability and RFS probability based on the Kaplan−Meier method (all p>0.05) ( Figures 3 , S1 ).

Table 2 shows the characteristics of intra- and postoperative outcomes of patients before and after PSM. Compared with the OLR cohort before PSM, the patients in the LLR cohort had shorter hospital stays after surgery (p<0.001). Other characteristics were comparable.

After PSM, compared with the OLR cohort, the patients in the LLR cohort had a higher first quartile (Q1) of intraoperative blood loss (p=0.036) and shorter hospital stay after surgery (p<0.001). However, there was no difference in intraoperative blood transfusion, surgical margin or prognostic outcomes between these two cohorts (all p>0.05).

Postoperative complications of patients were counted and analysed after PSM ( Table 2 ). The patient percent of complications in the LLR and OLR cohorts was not different (14.3% vs. 21.4%, p=0.369). In addition, complications, including ascites (5.7% vs. 4.8%, p=0.829), bile leakage (2.9% vs. 4.8%, p=0.637), operative bleeding (2.9% vs. 3.6%, p=0.844), pleural effusion (5.7% vs. 4.8%, p=0.829), surgical site infection (0.0% vs. 4.8%, p=0.189) and the Clavien–Dindo classification of complications (grade I–II: 8.6% vs. 15.8%, p=0.314; grade III–IV: 5.7% vs. 6.0%, p=0.96), were not different between these two cohorts ( Table 2 ). These results indicated that the postoperative complications of patients in the LLR and OLR cohorts after PSM were comparable.

Before PSM, the univariate and multivariate analysis results indicated that maximum tumour diameter, partial and intact tumour capsules, and MVI:M1 were independent risk factors for survival ( Table S1 ). Patient age, maximum tumour diameter, MVI:M2, γ-glutamyl transferase and surgical margin ≤1 cm were independent risk factors for tumour recurrence ( Table S2 ).

After PSM, as shown in Table 3 , maximum tumour diameter and the presence of satellite nodules were identified as independent risk factors for survival. In addition, HBV DNA level and surgical margin ≤1 cm were identified as independent risk factors for tumour recurrence.

Considering that surgical margin is an independent risk factor for tumour recurrence, which is the solely controllable variable during the operation, the long-term survival outcomes and intra- and postoperative short-term outcomes of patients with different surgical margins were compared. As shown in Figure S2 , no significant differences were found between surgical margins ≥1 cm and <1 cm in survival probability based on the Kaplan−Meier method before or after PSM (both p>0.05). However, the RFS probability in the surgical margin ≥1 cm group before and after PSM was significantly better than that in the surgical margin <1 cm group (both p<0.001). In the comparison of intra- and postoperative short-term outcomes, the results showed that these two groups were well balanced ( Table S3 ).