We retrospectively reviewed the data for patients who received LR for HCC in Southwest hospital, Army medical university, Chongqing, China from January 2009 to December 2017.The inclusion criteria was as follows: (1) patients aged 18-75 years; (2) liver function classified as Child-Pugh A or B; (3) the remaining liver volume was adequate with the preoperative evaluation; (4) histopathological confirmation of HCC; and (5) no extrahepatic metastasis. The exclusion criteria were as follows: (1) patients with recurrent HCC; (2) liver resection combined with abdominal organ resection other than gallbladder resection; and (3) patients with HCC who underwent prior RFA or TACE.

A total of 1773 patients with HCC were collected in this study: 683 patients in LLR group and 1090 patients in OLR group ( Supplementary Table S1 ). Preoperative evaluations were similar in two groups and included routine blood tests, liver function, coagulation examinations, serum tumor markers, indocyanine green retention test at 15 minutes (ICG-R15), and triphasic enhanced computed tomography (CT) and/or magnetic resonance imaging (MRI). Routine blood and hepatic function tests were performed after surgery. Abdominal ultrasonography was routinely conducted for patients before discharge. LMLR (≤ 2 segments) or LMH (≥ 3 contiguous segments) were defined according to the previous study (8).The severity of postoperative complications was graded by the Clavien–Dindo classification, and the severe complications were defned as Clavien–Dindo grade III and above (17). This study was approved by the Ethics Committee of Southwest Hospital of Army Medical University.

For LLR, either 3D or 2D laparoscopic system was used. The patient was placed in the supine position. Carbon dioxide pneumoperitoneum was established with a pressure of 11–13 mmHg. Two 12 mm ports and two 5 mm ports were applied for the operation, and one 12mm port was used for the laparoscope. Laparoscopic ultrasonography was routinely performed to confirm the positions of tumors and guide the dissection line. The Pringle maneuver was applied to control blood loss. The liver parenchyma was dissected using the ultrasonic dissector. Intraparenchymal vascular and biliary structures (≥3 mm) were ligated by titanium clips or Hem-o-lock clips. The main Glissonian pedicles or hepatic veins were transected by the laparoscopic linear stapler or Hem-o-lock clips after ligation. The specimen was placed into a sterile bag and extracted through a suprapubic incision or an upper abdominal midline incision.

For OLR, a reverse L-incision or right subcostal incision was conducted. The operating procedure was similar to LLR. CUSA or clamp crushing was used as the main method for liver parenchyma dissection.

The follow-up was conducted every one month within three months after operation, and then every three months within two years and three or six months afterwards. Routine investigations at each follow-up included routine blood tests, liver function, tumor markers, abdominal ultrasonography, and even CT or MRI if necessary. Overall survival (OS) was defined as the time from the surgery date to death from any cause or the last follow-up. Disease free survival (DFS) was defined as the time from the surgery date to tumor recurrence. All patients were followed up with the same protocol.

PSM with 1:1 ratio was used to eliminate the selection bias between LLR and OLR groups based on the nearest neighbor matching method without replacement. The propensity score covariates in this study included age, gender, HBV, HCV, liver cirrhosis, Child–Pugh score, American Society of Anesthesiologists score (ASA score), preoperative blood tests (ALT, TBIL, ALB, PT, Platelet count and AFP), ICG-R15, tumor location, tumor number, largest tumor diameter, type of LR, range of LR, resection tumor margin, margin status, histological grade, satellite nodule, portal vein invasion, bile duct invasion and TNM stage ( Table 1 ). After matching, P values for the group samples were all greater than 0.05, indicating a good balance. Continuous variables were compared using a t test or Mann–Whitney test. Categorical variables were compared using the χ² or Fisher exact test. A two-tailed P value <0.05 was considered significant. Survival curves were estimated by the Kaplan-Meier method for OS and DFS, the log-rank test was used for between-group comparisons. P values < 0.05 were considered significant. All statistical analyses were performed using SPSS software version 25.0 (IBM SPSS Inc. Chicago, IL).

A total of 1773 patients with HCC were collected in this study, 683 patients received LLR and 1090 patients received OLR. The baseline characteristics were shown in Table 1 . Between LLR and OLR groups, age, HBV-DNA and ASA scores were significantly different. Besides, the preoperative serum levels of TBIL, ALT, PT, Platelet count and AFP were different. Among the operative characteristics, tumor location, largest tumor diameters and the range of LR were significantly different between two groups. The characteristics of histological grade and portal vein invasion were also significantly different. Moreover, the TNM stage was different between two groups. After PSM with 1:1 ratio, there were 567 patients in each group with well-balanced baseline characteristics ( Table 1 ).

Short-term outcomes were compared between LLR and OLR groups after PSM ( Table 2 ). The operative time was similar between two groups. On the other hand, the operative blood loss and the rate of blood transfusion in LLR groups were significantly lower than them in OLR group (200.00 ml vs. 300.00 ml, P<0.001; 8.6% vs. 12.3%, P=0.042). The perioperative mortality was similar between two groups. However, the overall postoperative complications in LLR group were significantly lower than them in OLR group (23.8% vs. 32.8%, P=0.001). The complications mainly included seroperitoneum, hydrothorax, infection, hemorrhage, bile leak, liver failure, respiratory failure and renal failure. Moreover, the severe complications (Clavien–Dindo grade III and above) were also significantly lower in LLR groups (5.1% vs. 8.6%, P=0.019). Besides, the postoperative hospital stays were significantly shorter in LLR group (10.00 days vs. 13.00 days, P<0.001). Together, the results showed that patients in LLR group had better short-term outcomes compared with them in OLR group.

The long-term outcomes were compared between LLR and OLR groups after PSM. The OS and DFS curves were presented in Figure 1 . Survival analysis showed that the OS had no significant difference between LLR and OLR groups ( Figure 1 , P=0.973). The cumulative 1-, 3-, and 5-year OS rates were 87.9%, 68.9%, and 57.7% for patients in LLR group, and were 85.9%, 68.8%, and 58.8% for patients in OLR group, respectively. Consistently, the DFS had no difference between LLR and OLR groups ( Figure 1 , P=0.812). The cumulative 1-, 3-, and 5-year DFS rates were 73.0%, 51.5%, and 40.6% for patients in LLR group, and were 70.3%, 49.0%, and 42.4% for patients in OLR group, respectively. Together, the results showed that patients in LLR group had comparable long-term outcomes with them in OLR group.

A subgroup analysis was performed to assess the outcomes of LMLR-DS for HCCs compared with OMLR-DS. After PSM with 1:1 ratio, there were 178 patients in LMLR-DS or OMLR-DS group, and all baseline characteristics were well-balanced ( Supplementary Table S2 ). The operative time and the rate of blood transfusion were similar in two groups. While, the operative blood loss in LMLR-DS group was significantly lower than it in OMLR-DS group (200.00 ml vs. 300.00 ml, P<0.001). There was no perioperative death in both two groups. However, the overall and severe postoperative complications were significantly lower in LMLR-DS group compared with them in OMLR-DS group (27.5% vs. 39.9%, P=0.014; 3.9% vs. 11.2%, P=0.009; Table 3 ). The postoperative stays were significantly shorter in LMLR-DS group (10.00 days vs. 13.50 days, P<0.001). For long-term survival analysis, the OS had no significant difference between LMLR-DS and OMLR-DS groups (P=0.476, Figure 2 ). The cumulative 1-, 3-, and 5-year OS rates were 92.6%, 76.0%, and 61.6% for patients in LMLR-DS group, and were 90.9%, 78.2%, and 66.5% for patients in OMLR-DS group, respectively. Consistently, the DFS had no difference between LMLR-DS and OMLR-DS groups ( Figure 2 , P=0.536). The cumulative 1-, 3-, and 5-year DFS rates were 76.7%, 51.2%, and 43.9% for patients in LMLR-DS group, and were 75.7%, 54.1%, and 47.7% for patients in OMLR-DS group, respectively. Together, the results showed that LMLR-DS showed better short-term outcomes and similar long-term outocomes with OMLR-DS for HCCs.

Another subgroup analysis was performed to assess short-term and long-term outcomes of LMH for HCCs compared with OMH. There were 115 patients in either LMH or OMH group after PSM, and all baseline characteristics were well-balanced ( Supplementary Table S3 ). The operative time was similar in two groups. However, the operative blood loss and the rate of blood transfusion in LMH group were significantly lower than them in OMH group (300.00ml vs. 400.00ml, P=0.003; 13.0% vs. 24.3%, P=0.028; Table 4 ). There was no perioperative death in both two groups. Besides, the overall and severe postoperative complications had no significant difference between two groups ( Table 4 ). The postoperative hospital stays were significantly shorter in LMH group (11.00 days vs. 14.00 days, P<0.001). For long-term survival analyses, the OS had no significant difference between LMH and OMH groups (P=0.939, Figure 3 ). The cumulative 1-, 3-, and 5-year OS rates were 70.8%, 53.9%, and 44.3% for patients in LMH group, and were 75.5%, 49.2%, and 44.7% for patients in OMH group, respectively. Consistently, the DFS had no difference between two groups (P=0.681, Figure 3 ). The cumulative 1-, 3-, and 5-year DFS rates were 54.4%, 42.2%, and 29.9% for patients in LMH group, and were 55.7%, 33.5%, and 33.2% for patients in OMH group, respectively. Together, the results showed that LHM for HCCs had comparable short-term and long-term outcomes with OHM.