Laparoscopic versus open surgery for liver resection: a multicenter cohort study Authors: Yesheng Li, Longrong Wang, Yibin Guo, Jiamin Zhou, Ning Zhang, Xigan He, Yixiu Wang, Weiping Zhu, Miao Wang, Hongxu Zhu, Zhiwen Ding, Yibin Wu, Ti Zhang, Qi Pan, Yun Feng, Zhenhai Lin, Anrong Mao, Yongfa Zhang, Yilin Wang, Bin Zhang, Yangqing Huang, Yiming Zhao, Lu Wang Published: 2024 ABSTRACT The relative risk and benefits of Laparoscopic liver resection (LLR) are still controversial. This study aimed to evaluate perioperative and survival outcomes for three primary malignant liver tumors related outcomes following LLR. Data from three Homogeneous-High-Volume-Expert-Centers were collected. Multivariable logistic regression analysis was performed for the association between LLR, and major complications defined as Clavien-Dindo Classification grade ≥ II. Multivariable Cox proportional hazards regression was used to explore the impact of LLR on primary malignant liver tumor survival. Overall, 5886 patients underwent liver resection, of which 1991 underwent LLR. After adjusting for covariables, LLR had significantly lower rate of CDC grade ≥ II (OR = 0.56, 95% CI: 0.48–0.67, P < 0.001). Interaction analysis showed that LLR had significant association with sex, drinking status, history of abdominal surgery, and platelet count (all P < 0.05). The overall survival for each primary malignant liver tumor shown potential risk but without significantly different between LLR and OLR (all P > 0.05). LLR was associated with lower rate of major complications, especially in women, non-drinkers, those without a history of abdominal surgery, and those with normal platelet. For primary malignant liver tumors, the survival outcomes did not differ significantly between LLR and OLR.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-024-76260-w. Introduction Liver resection is the optimal treatment option for many liver diseases, especially malignant liver tumors1,2. Since Reich et al. reported the first case of laparoscopic liver resection (LLR) in 19913, LLR has been used as an alternative to traditional open liver surgery worldwide4–6. Furthermore, laparoscopy is increasingly used for donor liver resection in living donor liver transplantation7. Previous studies have demonstrated that laparoscopic appendectomy and cholecystectomy are more effective than open surgery8,9. In addition, laparoscopic surgery carries a lower surgical risk than open surgery for gastric, colon, and rectal cancer10–12. The LEOPARD-2 trial indicated that patients who underwent laparoscopic pancreatoduodenectomy had a higher mortality rate due to postoperative complications than those who underwent open pancreatoduodenectomy13, and the LACC trial also showed that laparoscopic surgery was detrimental to the long-term survival of patients with cervical cancer compared with open surgery14. Hence, laparoscopic surgery may have different outcomes depending upon the type of surgery. Due to the complex anatomy of the liver and the wide range of malignant tumors that can affect it, liver resection has many different surgical approaches and is considered one of the most complex surgeries15. This complexity makes it difficult to compare the relative risks of LLR and open liver resection (OLR)16, such as in the ORANGE-II trial, which failed to enroll enough patients on schedule17. Other reports on the partial benefits of LLR were based on expert centers with different volumes18–25. These studies suggest that the experience, skill, surgical volumes, and instruments used for LLR are not uniform among different centers worldwide, resulting in a wide range of outcomes26,27. To the best of our knowledge, there has been no comprehensive analysis of the relative risks of LLR and OLR based on large, high-volume expert center databases. Therefore, this study aimed to explore the impact of LLR and OLR on the perioperative outcomes for all patients and the long-term survival of patients with primary malignant liver tumors (MLTs) by conducting a large population study based on the original hospital database of three homogeneous high-volume expert centers (HHVECs). Methods Study design and patients This population-based, multicenter, retrospective cohort study followed the STROBE statement28. The data of patients treated between January 2015 and July 2020 were extracted from the intent-to-treat hospital database that was established by three HHVECs [including the hepatic surgery department of Fudan University Shanghai Cancer Center (FUSCC), the hepatobiliary surgery department of Shanghai Public Health Clinical Center (SPHCC) Fudan University, and the Digestive Diseases Research Center of the Affiliated Hospital of Xuzhou Medical University (AHXMU)]. All consecutive patients were grouped into the LLR and OLR groups according to the intent-to-treat principle before surgery. The exclusion criteria were (1) missing key data (as grouping, primary outcome), (2) duplicate registration, (3) with synchronous major extrahepatic organ surgery (defined as enterotomy, hemicolectomy, proctocolectomy, pancreaticoduodenectomy, or cholangiojejunostomy), or (4) only undergoing minor liver surgery (defined as biopsy or tumor ablation). Those who underwent liver resection with a synchronous minor operation (defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues) were still included. All patients in the three HHVECs were routinely advised to undergo regular follow-ups every 3 months. In this study, patients with three primary MLTs [hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), and colorectal cancer liver metastases (CRLM)] who underwent R0 resection were included in the survival analysis. Covariables of the baseline, clinical, and oncological characteristics were extracted from the database. All covariables were adjusted in different multivariable models, and the bias of unmeasured confounders was minimized using a consistent clinical treatment protocol in the three HHVECs (Supplementary Materials, part 1). Outcomes The primary outcome was major complications (defined as Clavien-Dindo Classification [CDC] grade ≥ II)24,29. The secondary outcomes were all-in complications [defined as comprehensive complication index® (CCI®)]30, 90 days mortality associated with liver resection, and post-hepatectomy liver failure [PHLF, defined as International Study Group of Liver Surgery (ISGLS) grade ≥ B]31. In addition, the other four outcomes [operation time (minutes, defined as the time from skin cutting to the end of stitching), postoperative hospital stay (days, defined as from the operation to discharge), the 7th day ECOG score after operation, and hospitalization costs (thousands of Chinese yuan) were also evaluated. For long-term survival outcomes, overall survival (OS) and disease-free survival (DFS) were calculated (Supplementary Materials, part 2). Statistical analysis The statistical software packages R (http://www.R-project.org, The R Foundation) and Free Statistics software versions 1.8 were used to perform the statistical analyses32. Data were presented as means ± standard deviations (SD) for continuous variables and as percentages for categorical variables. For baseline, clinical, and oncological characteristics analysis, the statistical differences between the LLR and OLR groups were tested using Student’s t-test for continuous variables and Pearson’s chi-square or Fisher test for categorical variables. The OLR group was used as the reference group to assess the relative risk of LLR. Multivariable logistic regression models (Model 1: Adjusted for the demographic characteristics. Model 2: Adjusted for the demographic characteristic and medical history. Model 3: Adjusted for the demographic characteristic, medical history, and preoperative indicators. Model 4: Adjusted for the demographic characteristic, medical history, preoperative indicators, and surgical procedure during surgery) were used for categorical outcomes (CDC grade ≥ II, mortality, and ISGLS grade ≥ B), multivariable linear regression models were used for continuous outcomes (CCI®, operation time, postoperative hospital stay, 7th day ECOG score, and hospitalization costs), and a multivariable Cox proportional-hazards model was used for survival outcomes (OS and DFS). Subgroup analyses for major complications were performed using multivariable logistic regression, and interaction analyses for subgroups were performed using the likelihood ratio test. Patients with missing key data were excluded from the analyses, whereas those lost to follow-up were censored during the last follow-up cycle. Six propensity score matchings (PSMs) were used for sensitivity analyses of the primary outcomes. Statistical significance was set at P < 0.05. The E-value test of the primary outcome was used to test for unmeasured confounders (Supplementary Materials, part 3). Results A total of 9936 consecutive patients who underwent liver resection were identified at the three HHVECs. Among these, 4617 were excluded (377 were missing key data, 17 were duplicated registrations, 3165 were synchronous major extrahepatic organ surgeries, and 491 only underwent minor liver surgery). The remaining 5886 patients (1991 with LLR and 3895 with OLR) were included in the perioperative analyses. For survival analyses, 595 patients were further excluded (228 had missing data for R0 resection or tumor type, 318 were non-R0 resection, 39 were non-primary MLTs, and 10 were benign disease) and 5291 patients (3451 with HCC, 668 with ICC, and 1172 with CRLM) were included (Fig. 1). There were 3682 patients from the FUSCC, 1331 from the SPHCC, and 873 from the AHXMU (Table 1). Fig. 1Flow chat. Flow chat. Table 1Baseline, clinical and oncological characteristics of participants before liver resection.VariatesTotalLaparoscopicOpenP value(n = 5886)(n = 1991)(n = 3895)Centers, n (%)0.275 FUSCC3682 (62.6)1237 (62.1)2445 (62.8) SPHCC1331 (22.6)472 (23.7)859 (22.1) AHXMU873 (14.8)282 (14.2)591 (15.2)Gender, n (%)0.011 Men4389 (74.6)1525 (76.6)2864 (73.5) Women1497 (25.4)466 (23.4)1031 (26.5)Age (years), Mean ± SD58.9 ± 11.358.4 ± 11.059.1 ± 11.40.036BMI (kg/m2), Mean ± SD23.3 ± 3.223.8 ± 3.323.1 ± 3.1< 0.001Current smoker, n (%)0.128 No4335 (73.6)1442 (72.4)2893 (74.3) Yes1551 (26.4)549 (27.6)1002 (25.7)Drinker, n (%)0.27 No4587 (77.9)1535 (77.1)3052 (78.4) Yes1299 (22.1)456 (22.9)843 (21.6)Hypertension, n (%)0.322 No4926 (83.7)1653 (83)3273 (84) Yes960 (16.3)338 (17)622 (16)Diabetes, n (%)0.086 No5103 (86.7)1705 (85.6)3398 (87.2) Yes783 (13.3)286 (14.4)497 (12.8)Abdominal surgery history, n (%)< 0.001 No3873 (65.8)1490 (74.8)2383 (61.2) Yes2013 (34.2)501 (25.2)1512 (38.8)Hepatitis, n (%)< 0.001 No2673 (45.4)787 (39.5)1886 (48.4) Yes3213 (54.6)1204 (60.5)2009 (51.6)Cirrhosis, n (%)< 0.001 No3063 (52.0)915 (46)2148 (55.1) Yes2823 (48.0)1076 (54)1747 (44.9)Portal vein hypertension, n (%)0.008 No5751 (97.7)1931 (97)3820 (98.1) Yes135 ( 2.3)60 (3)75 (1.9)WBC (*109/L), Mean ± SD5.8 ± 2.05.5 ± 1.85.9 ± 2.0< 0.001HGB (g/L), Mean ± SD137.4 ± 17.4140.9 ± 15.9135.6 ± 17.8< 0.001PLT (*109/L), Mean ± SD182.8 ± 73.0169.2 ± 64.3189.7 ± 76.1< 0.001MELD score, Mean ± SD7.5 ± 3.07.7 ± 2.97.5 ± 3.10.011ECOG score, n (%)< 0.001 05601 (95.2)1927 (96.8)3674 (94.3) 1285 ( 4.8)64 (3.2)221 (5.7)Type of Tumors, n (%)< 0.001 Benign10 ( 0.2)10 (0.5)0 (0) HCC3660 (62.2)1377 (69.2)2283 (58.6) ICC744 (12.6)149 (7.5)595 (15.3) CRLM1433 (24.3)452 (22.7)981 (25.2) Other malignant tumors39 ( 0.7)3 (0.2)36 (0.9)Number of tumor lesions, n (%)< 0.001 14599 (78.1)1707 (85.7)2892 (74.2) 2666 (11.3)190 (9.5)476 (12.2) 3249 ( 4.2)40 (2)209 (5.4) >=4372 ( 6.3)54 (2.7)318 (8.2)Max diameter of tumors (cm), Mean ± SD5.5 ± 4.13.9 ± 2.86.3 ± 4.4< 0 001Portal vein tumor thrombosis, n (%)< 0.001 No5529 (93.9)1945 (97.7)3584 (92) Yes357 ( 6.1)46 (2.3)311 (8)Microvascular invasion, n (%)< 0.001 M03636 (61.8)1343 (67.5)2293 (58.9) M11482 (25.2)464 (23.3)1018 (26.1) M2768 (13.0)184 (9.2)584 (15)Lymph node metastasis, n (%)< 0.001 No5565 (94.5)1975 (99.2)3590 (92.2) Yes321 ( 5.5)16 (0.8)305 (7.8)AFP (ng/ml), Mean ± SD242.2 ± 786.7151.1 ± 614.5287.9 ± 856.8< 0.001CEA (ng/ml), Mean ± SD9.2 ± 47.54.8 ± 15.811.4 ± 56.9< 0.001CA 19 − 9 (U/ml), Mean ± SD60.1 ± 161.138.7 ± 116.370.7 ± 178.1< 0.001Tumor rupture before operation, n (%)< 0.001 No5802 (98.6)1985 (99.7)3817 (98) Yes84 ( 1.4)6 (0.3)78 (2)#n (%); Mean ± SD.$Student`s t-test; Pearson`s Chi-squared test; Fisher test.¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues.&Abbreviations: FUSCC, Fudan University Shanghai Cancer Center; SPHCC, Shanghai Public Health Clinical Center; AHXMU, Affiliated Hospital of Xuzhou Medical University; BMI, body mass index; WBC, white blood cell; HGB, hemoglobin; PLT, platelets; MELD, model for end-stage liver disease; ECOG, Eastern Cooperative Oncology Group; HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; AFP, alpha- fetoprotein; CEA, carcinoembryonic antigen. Table 1Baseline, clinical and oncological characteristics of participants before liver resection.VariatesTotalLaparoscopicOpenP value(n = 5886)(n = 1991)(n = 3895)Centers, n (%)0.275 FUSCC3682 (62.6)1237 (62.1)2445 (62.8) SPHCC1331 (22.6)472 (23.7)859 (22.1) AHXMU873 (14.8)282 (14.2)591 (15.2)Gender, n (%)0.011 Men4389 (74.6)1525 (76.6)2864 (73.5) Women1497 (25.4)466 (23.4)1031 (26.5)Age (years), Mean ± SD58.9 ± 11.358.4 ± 11.059.1 ± 11.40.036BMI (kg/m2), Mean ± SD23.3 ± 3.223.8 ± 3.323.1 ± 3.1< 0.001Current smoker, n (%)0.128 No4335 (73.6)1442 (72.4)2893 (74.3) Yes1551 (26.4)549 (27.6)1002 (25.7)Drinker, n (%)0.27 No4587 (77.9)1535 (77.1)3052 (78.4) Yes1299 (22.1)456 (22.9)843 (21.6)Hypertension, n (%)0.322 No4926 (83.7)1653 (83)3273 (84) Yes960 (16.3)338 (17)622 (16)Diabetes, n (%)0.086 No5103 (86.7)1705 (85.6)3398 (87.2) Yes783 (13.3)286 (14.4)497 (12.8)Abdominal surgery history, n (%)< 0.001 No3873 (65.8)1490 (74.8)2383 (61.2) Yes2013 (34.2)501 (25.2)1512 (38.8)Hepatitis, n (%)< 0.001 No2673 (45.4)787 (39.5)1886 (48.4) Yes3213 (54.6)1204 (60.5)2009 (51.6)Cirrhosis, n (%)< 0.001 No3063 (52.0)915 (46)2148 (55.1) Yes2823 (48.0)1076 (54)1747 (44.9)Portal vein hypertension, n (%)0.008 No5751 (97.7)1931 (97)3820 (98.1) Yes135 ( 2.3)60 (3)75 (1.9)WBC (*109/L), Mean ± SD5.8 ± 2.05.5 ± 1.85.9 ± 2.0< 0.001HGB (g/L), Mean ± SD137.4 ± 17.4140.9 ± 15.9135.6 ± 17.8< 0.001PLT (*109/L), Mean ± SD182.8 ± 73.0169.2 ± 64.3189.7 ± 76.1< 0.001MELD score, Mean ± SD7.5 ± 3.07.7 ± 2.97.5 ± 3.10.011ECOG score, n (%)< 0.001 05601 (95.2)1927 (96.8)3674 (94.3) 1285 ( 4.8)64 (3.2)221 (5.7)Type of Tumors, n (%)< 0.001 Benign10 ( 0.2)10 (0.5)0 (0) HCC3660 (62.2)1377 (69.2)2283 (58.6) ICC744 (12.6)149 (7.5)595 (15.3) CRLM1433 (24.3)452 (22.7)981 (25.2) Other malignant tumors39 ( 0.7)3 (0.2)36 (0.9)Number of tumor lesions, n (%)< 0.001 14599 (78.1)1707 (85.7)2892 (74.2) 2666 (11.3)190 (9.5)476 (12.2) 3249 ( 4.2)40 (2)209 (5.4) >=4372 ( 6.3)54 (2.7)318 (8.2)Max diameter of tumors (cm), Mean ± SD5.5 ± 4.13.9 ± 2.86.3 ± 4.4< 0 001Portal vein tumor thrombosis, n (%)< 0.001 No5529 (93.9)1945 (97.7)3584 (92) Yes357 ( 6.1)46 (2.3)311 (8)Microvascular invasion, n (%)< 0.001 M03636 (61.8)1343 (67.5)2293 (58.9) M11482 (25.2)464 (23.3)1018 (26.1) M2768 (13.0)184 (9.2)584 (15)Lymph node metastasis, n (%)< 0.001 No5565 (94.5)1975 (99.2)3590 (92.2) Yes321 ( 5.5)16 (0.8)305 (7.8)AFP (ng/ml), Mean ± SD242.2 ± 786.7151.1 ± 614.5287.9 ± 856.8< 0.001CEA (ng/ml), Mean ± SD9.2 ± 47.54.8 ± 15.811.4 ± 56.9< 0.001CA 19 − 9 (U/ml), Mean ± SD60.1 ± 161.138.7 ± 116.370.7 ± 178.1< 0.001Tumor rupture before operation, n (%)< 0.001 No5802 (98.6)1985 (99.7)3817 (98) Yes84 ( 1.4)6 (0.3)78 (2)#n (%); Mean ± SD.$Student`s t-test; Pearson`s Chi-squared test; Fisher test.¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues.&Abbreviations: FUSCC, Fudan University Shanghai Cancer Center; SPHCC, Shanghai Public Health Clinical Center; AHXMU, Affiliated Hospital of Xuzhou Medical University; BMI, body mass index; WBC, white blood cell; HGB, hemoglobin; PLT, platelets; MELD, model for end-stage liver disease; ECOG, Eastern Cooperative Oncology Group; HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; AFP, alpha- fetoprotein; CEA, carcinoembryonic antigen. Baseline, clinical and oncological characteristics of participants before liver resection. #n (%); Mean ± SD. $Student`s t-test; Pearson`s Chi-squared test; Fisher test. ¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues. &Abbreviations: FUSCC, Fudan University Shanghai Cancer Center; SPHCC, Shanghai Public Health Clinical Center; AHXMU, Affiliated Hospital of Xuzhou Medical University; BMI, body mass index; WBC, white blood cell; HGB, hemoglobin; PLT, platelets; MELD, model for end-stage liver disease; ECOG, Eastern Cooperative Oncology Group; HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; AFP, alpha- fetoprotein; CEA, carcinoembryonic antigen. Table 2Baseline, clinical and oncological characteristics of participants after liver resection.VariatesTotalLaparoscopicOpenP value(n = 5886)(n = 1991)(n = 3895)Resection classification, n (%)< 0.001 Minor4335 (73.6)1755 (88.1)2580 (66.2) Major1551 (26.4)236 (11.9)1315 (33.8)Resection complexity, n (%)< 0.001 Easy1443 (24.5)767 (38.5)676 (17.4) Moderate1470 (25.0)510 (25.6)960 (24.6) Difficult2973 (50.5)714 (35.9)2259 (58)Portal vein occlusion, n (%)< 0.001 No3021 (51.3)1267 (63.6)1754 (45) Yes2865 (48.7)724 (36.4)2141 (55)Synchronous minor operation, n (%)< 0.001 No3873 (65.8)1461 (73.4)2412 (61.9) Yes2013 (34.2)530 (26.6)1483 (38.1)Intraoperative blood transfusion, n (%)< 0.001 No5316 (90.3)1921 (96.5)3395 (87.2) Yes570 ( 9.7)70 (3.5)500 (12.8)R0 resection, n (%)< 0.001 No357 ( 6.1)45 (2.3)312 (8) Yes5291 (89.9)1893 (95.1)3398 (87.2) NA238 ( 4.0)53 (2.7)185 (4.7)#n (%); Mean ± SD.$Student`s t-test; Pearson`s Chi-squared test; Fisher test.¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues. Table 2Baseline, clinical and oncological characteristics of participants after liver resection.VariatesTotalLaparoscopicOpenP value(n = 5886)(n = 1991)(n = 3895)Resection classification, n (%)< 0.001 Minor4335 (73.6)1755 (88.1)2580 (66.2) Major1551 (26.4)236 (11.9)1315 (33.8)Resection complexity, n (%)< 0.001 Easy1443 (24.5)767 (38.5)676 (17.4) Moderate1470 (25.0)510 (25.6)960 (24.6) Difficult2973 (50.5)714 (35.9)2259 (58)Portal vein occlusion, n (%)< 0.001 No3021 (51.3)1267 (63.6)1754 (45) Yes2865 (48.7)724 (36.4)2141 (55)Synchronous minor operation, n (%)< 0.001 No3873 (65.8)1461 (73.4)2412 (61.9) Yes2013 (34.2)530 (26.6)1483 (38.1)Intraoperative blood transfusion, n (%)< 0.001 No5316 (90.3)1921 (96.5)3395 (87.2) Yes570 ( 9.7)70 (3.5)500 (12.8)R0 resection, n (%)< 0.001 No357 ( 6.1)45 (2.3)312 (8) Yes5291 (89.9)1893 (95.1)3398 (87.2) NA238 ( 4.0)53 (2.7)185 (4.7)#n (%); Mean ± SD.$Student`s t-test; Pearson`s Chi-squared test; Fisher test.¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues. Baseline, clinical and oncological characteristics of participants after liver resection. #n (%); Mean ± SD. $Student`s t-test; Pearson`s Chi-squared test; Fisher test. ¥Synchronous minor operation was defined as tumor ablation, cholecystectomy, and repair of perihepatic organs or tissues. Among the 5886 included patients (aged 58.9 ± 11.3 years), 4389 (74.6%) were men and 1497 (25.4%) were women. The patients who underwent LLR had a higher body mass index (BMI). They were less likely to have an abdominal surgery history, hepatitis, cirrhosis, multiple tumors, a larger maximum tumor diameter, portal vein tumor thrombosis, or lymph node metastasis. Intraoperative, major, and difficult resection, minor synchronous operation, and blood transfusion were more frequently performed in OLR (Tables 1, 2). The rate of CDC grade ≥ II in all patients was 22.4%, including 12.5% in LLR and 27.4% in OLR (Supplementary Table 1). Before covariable adjustment, LLR had a lower rate of CDC grade ≥ II (OR = 0.38, 95% CI:0.33 ~ 0.44, P < 0.001). The results were similar after adjusting for baselines (model-1), with baselines and past history (model-2), and with baselines, past history, and preoperative examination (model-3) (all P < 0.001). After adjusting for baseline, past history, preoperative examination, and intraoperative operation (model-4), the negative association remained significant (OR = 0.56, 95% CI:0.48 ~ 0.67, P < 0.001). All-in complications such as CCI® and PHLF as ISGLS grade ≥ B, operation time, postoperative hospital stay, and 7th day ECOG score were also significantly lower in the LLR group (model-4, all P < 0.005). The 90-day mortality associated with liver resection and the hospitalization costs did not differ significantly between the two groups (model-4) (Table 3). In the sensitivity analyses, the rate of CDC grade ≥ II associated with LLR remained stable with six different PSM methods (all P < 0.001) (Supplementary Table 2), and the E-value showed that the results were stable after considering the effect of unmeasured confounders (E-value = 2.007). Table 3The multivariable analyses of complications for laparoscopic versus open liver resection.Model-1Model-2Model-3Model-4OR or β (95%CI)P valueOR or β (95%CI)P valueOR or β (95%CI)P valueOR or β (95%CI)P valuePrimary outcome CDC grade > = II0.39 (0.34 ~ 0.45)< 0.0010.38 (0.32 ~ 0.44)< 0.0010.41 (0.35 ~ 0.48)< 0.0010.56 (0.48 ~ 0.67)< 0.001Secondary outcomes CCI®-5.78 (-6.44~-5.11)< 0.001-5.96 (-6.63~-5.29)< 0.001-5.56 (-6.23~-4.9)< 0.001-3.78 (-4.45~-3.11)< 0.001 Mortality0.7 (0.36 ~ 1.34)0.2820.67 (0.35 ~ 1.3)0.2380.7 (0.36 ~ 1.38)0.3021.26 (0.61 ~ 2.61)0.537 ISGLS grade > = B0.45 (0.35 ~ 0.58)< 0.0010.45 (0.34 ~ 0.58)< 0.0010.44 (0.34 ~ 0.58)< 0.0010.63 (0.47 ~ 0.85)0.002Extended outcomes Operation time (min)-20.01 (-22.45~-17.57)< 0.001-18.78 (-21.23~-16.32)< 0.001-17.42 (-19.9~-14.94)< 0.001-5.6 (-7.99~-3.22)< 0.001 Postoperative hospital stays(days)-1 (-1.16~-0.84)< 0.001-1.08 (-1.24~-0.92)< 0.001-0.95 (-1.11~-0.79)< 0.001-0.76 (-0.93~-0.6)< 0.001 7th day ECOG score-0.79 (-0.82~-0.77)< 0.001-0.81 (-0.83~-0.78)< 0.001-0.8 (-0.82~-0.77)< 0.001-0.76 (-0.79~-0.73)< 0.001 Hospitalization costs (¥, thousand)-0 28 (-0 48~-0 09)0.004-0 35 (-0 55~-0 15)0.001-0 21 (-0 41~-0 01)0.036-0 01 (-0 21 ~ 0 2)0.959$Reference group: open liver resection.#Model-1 was adjusted with sex + age + BMI;.Model-2 was adjusted with Model-1 + current smoker + drinker + hypertension + diabetes + abdominal surgery history + cirrhosis + hepatitis + portal vein hypertension;.Model-3 was adjusted with Model-2 + WBC + HGB + PLT + MELD score + ECOG score;.Model-4 was adjusted with Model-3 + resection classification + resection complexity + portal vein occlusion + synchronous minor operation + intraoperative blood transfusion.&Abbreviations: CDC, Clavien - Dindo Classification; CCI, comprehensive complication index; ISGLS, International Study Group of Liver Surgery; ECOG, Eastern Cooperative Oncology Group. Table 3The multivariable analyses of complications for laparoscopic versus open liver resection.Model-1Model-2Model-3Model-4OR or β (95%CI)P valueOR or β (95%CI)P valueOR or β (95%CI)P valueOR or β (95%CI)P valuePrimary outcome CDC grade > = II0.39 (0.34 ~ 0.45)< 0.0010.38 (0.32 ~ 0.44)< 0.0010.41 (0.35 ~ 0.48)< 0.0010.56 (0.48 ~ 0.67)< 0.001Secondary outcomes CCI®-5.78 (-6.44~-5.11)< 0.001-5.96 (-6.63~-5.29)< 0.001-5.56 (-6.23~-4.9)< 0.001-3.78 (-4.45~-3.11)< 0.001 Mortality0.7 (0.36 ~ 1.34)0.2820.67 (0.35 ~ 1.3)0.2380.7 (0.36 ~ 1.38)0.3021.26 (0.61 ~ 2.61)0.537 ISGLS grade > = B0.45 (0.35 ~ 0.58)< 0.0010.45 (0.34 ~ 0.58)< 0.0010.44 (0.34 ~ 0.58)< 0.0010.63 (0.47 ~ 0.85)0.002Extended outcomes Operation time (min)-20.01 (-22.45~-17.57)< 0.001-18.78 (-21.23~-16.32)< 0.001-17.42 (-19.9~-14.94)< 0.001-5.6 (-7.99~-3.22)< 0.001 Postoperative hospital stays(days)-1 (-1.16~-0.84)< 0.001-1.08 (-1.24~-0.92)< 0.001-0.95 (-1.11~-0.79)< 0.001-0.76 (-0.93~-0.6)< 0.001 7th day ECOG score-0.79 (-0.82~-0.77)< 0.001-0.81 (-0.83~-0.78)< 0.001-0.8 (-0.82~-0.77)< 0.001-0.76 (-0.79~-0.73)< 0.001 Hospitalization costs (¥, thousand)-0 28 (-0 48~-0 09)0.004-0 35 (-0 55~-0 15)0.001-0 21 (-0 41~-0 01)0.036-0 01 (-0 21 ~ 0 2)0.959$Reference group: open liver resection.#Model-1 was adjusted with sex + age + BMI;.Model-2 was adjusted with Model-1 + current smoker + drinker + hypertension + diabetes + abdominal surgery history + cirrhosis + hepatitis + portal vein hypertension;.Model-3 was adjusted with Model-2 + WBC + HGB + PLT + MELD score + ECOG score;.Model-4 was adjusted with Model-3 + resection classification + resection complexity + portal vein occlusion + synchronous minor operation + intraoperative blood transfusion.&Abbreviations: CDC, Clavien - Dindo Classification; CCI, comprehensive complication index; ISGLS, International Study Group of Liver Surgery; ECOG, Eastern Cooperative Oncology Group. The multivariable analyses of complications for laparoscopic versus open liver resection. $Reference group: open liver resection. #Model-1 was adjusted with sex + age + BMI;. Model-2 was adjusted with Model-1 + current smoker + drinker + hypertension + diabetes + abdominal surgery history + cirrhosis + hepatitis + portal vein hypertension;. Model-3 was adjusted with Model-2 + WBC + HGB + PLT + MELD score + ECOG score;. Model-4 was adjusted with Model-3 + resection classification + resection complexity + portal vein occlusion + synchronous minor operation + intraoperative blood transfusion. &Abbreviations: CDC, Clavien - Dindo Classification; CCI, comprehensive complication index; ISGLS, International Study Group of Liver Surgery; ECOG, Eastern Cooperative Oncology Group. Additionally, the subgroup (adjusted for covariables as in model-4 without each stratification component) and interaction analyses showed that women (OR = 0.4, 95% CI:0.27 ~ 0.61, P < 0.001), non-drinkers (OR = 0.51, 95% CI:0.46 ~ 0.62, P < 0.001), without abdominal surgery history (OR = 0.49, 95% CI:0.39 ~ 0.6, P < 0.001), and normal platelet count (OR = 0.48, 95% CI:0.4 ~ 0.59, P < 0.001) had a lower rate of CDC grade > II associated with LLR (all P for interactions < 0.05) (Fig. 2). However, there were no interactions between LLR and the other variables (Supplementary Fig. 1). Fig. 2The relationship between laparoscopic and rate of major complications defined as CDC grade > = II. The relationship between laparoscopic and rate of major complications defined as CDC grade > = II. For survival analyses, each primary MLT was followed for more than two years. The OS and DFS of HCC, ICC, and CRLM are shown in Supplementary Table 3. Neither OS nor DFS was significantly different between LLR and OLR in each primary MLT (all P > 0.05) (Table 4). Table 4The survival analyses of HCC, ICC and CRLM for laparoscopic versus open liver resection.HCCICCCRLMHR (95%CI)P valueHR (95%CI)P valueHR (95%CI)P valueOS1 (0.9 ~ 1.11)0.9581.17 (0.87 ~ 1.58)0.2991.17 (0.94 ~ 1.47)0.156DFS0.98 (0.89 ~ 1.09)0.7311.23 (0.93 ~ 1.62)0.1421.18 (0.96 ~ 1.44)0.109$Reference group: open liver resection.#Model was adjusted with sex + age + BMI + current smoker + drinker + hypertension + diabetes + cirrhosis + hepatitis + portal vein hypertension + tumor rupture before operation + WBC + HGB + PLT + AFP + CEA + CA-19-9 + MELD score + ECOG score + type of tumors + number of tumors + diameter of tumors + portal vein tumor thrombosis + microvascular invasion + lymph node metastasis.&Abbreviations: HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; OS, overall survival; DFS, disease free survival. Table 4The survival analyses of HCC, ICC and CRLM for laparoscopic versus open liver resection.HCCICCCRLMHR (95%CI)P valueHR (95%CI)P valueHR (95%CI)P valueOS1 (0.9 ~ 1.11)0.9581.17 (0.87 ~ 1.58)0.2991.17 (0.94 ~ 1.47)0.156DFS0.98 (0.89 ~ 1.09)0.7311.23 (0.93 ~ 1.62)0.1421.18 (0.96 ~ 1.44)0.109$Reference group: open liver resection.#Model was adjusted with sex + age + BMI + current smoker + drinker + hypertension + diabetes + cirrhosis + hepatitis + portal vein hypertension + tumor rupture before operation + WBC + HGB + PLT + AFP + CEA + CA-19-9 + MELD score + ECOG score + type of tumors + number of tumors + diameter of tumors + portal vein tumor thrombosis + microvascular invasion + lymph node metastasis.&Abbreviations: HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; OS, overall survival; DFS, disease free survival. The survival analyses of HCC, ICC and CRLM for laparoscopic versus open liver resection. $Reference group: open liver resection. #Model was adjusted with sex + age + BMI + current smoker + drinker + hypertension + diabetes + cirrhosis + hepatitis + portal vein hypertension + tumor rupture before operation + WBC + HGB + PLT + AFP + CEA + CA-19-9 + MELD score + ECOG score + type of tumors + number of tumors + diameter of tumors + portal vein tumor thrombosis + microvascular invasion + lymph node metastasis. &Abbreviations: HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CRLM, colorectal cancer liver metastases; OS, overall survival; DFS, disease free survival. Discussion This study suggested that LLR was associated with a lower CDC grade ≥ II rate, which was confirmed by the subgroup and sensitivity analyses. Furthermore, the interaction analysis revealed more favorable outcomes in women, non-drinkers, individuals without a history of abdominal surgery, and those with normal platelet levels. These findings might help optimize the balance between the risks and benefits of LLR versus OLR, select the most suitable patients, and provide support for the expansion of the gold standard for LLR. This study, to our knowledge, is the first to identify an interaction between LLR and sex, possibly attributed to the effects of estrogen. These results are similar to those of a previous study that reported a relationship between estrogen and major liver surgery33. Indeed, it appears that estrogen induces a faster regenerative pace after major surgery33. Therefore, a careful assessment of males and postmenopausal females should be conducted33. As previously reported, alcohol consumption strengthened the association with liver diseases such as cirrhosis or HCC34, but the interaction between alcohol and liver resection, especially LLR, was reported for the first time in this study. It should be noted that many studies on liver resection excluded patients with high alcohol consumption35, limiting the possible conclusions. Nevertheless, two studies reported that hepatic steatosis is a risk factor for complications after major hepatectomy36,37, but it is unclear whether the patients underwent LLR or OLR. Alcohol consumption was included in a nomogram for complications after surgery for HCC38; again, the type of surgery was not considered. Contrary to primary resection, abdominal surgery was deemed safe and feasible during LLR39. A study reported no differences in the rate of abdominal surgery history between LLR and OLR40. Accordingly, we further explored the interactions between a history of abdominal surgery and LLR and OLR. We divided the platelets into three grades according to the reference values in clinical settings, where the interaction with LLR was significant between the normal and abnormal values, which is supported by a previous study, reporting that plasma albumin level and blood platelet counts were proposed as the indicators for the preoperative evaluation of liver functional reserve after resection for HCC41. The subgroup and interaction analyses were noteworthy and showed that resection classification and complexity did not interact with LLR. These results are supported by previous research reporting different liver diseases and resection classification or complexity18–20,24,25. In the survival analyses for HCC, ICC, and CRLM, the OS and DFS did not significantly differ between the LLR and OLR groups, which is similar to previous studies20,23,42 but should still be further explored. Laparoscopic surgery has several advantages. It can offer surgeons better exposure to the lesions and surrounding vessels with video magnification and avoid the stress response of delicate tissues exposed to air6,43–45. These findings are supported by secondary and extended outcomes such as lower CCI®, ISGLS grade ≥ B, operation time, postoperative hospital stay, and 7th-day ECOG score in this study. The first comprehensive study on LLR has meaningful implications for future studies and surgical practice. Although liver resection is a high-risk surgery, using LLR instead of OLR may be beneficial for several patients. At present, there are no clear selection criteria for LLR vs. OLR. Most experts tend to prefer LLR for borderline, confined liver segments, and single tumors, but the technology itself is not limited. Hence, surgeons usually select LLR or OLR based on their experience, the characteristics of the disease, the expected complexity of the surgery, and the characteristics of the patients. Whether OLR should be selected from the start is difficult to evaluate based on the present study because it was retrospective. In practice, OLR is easier to perform than LLR from a technical point of view. Therefore, if the operation is evaluated as difficult before surgery, OLR may be directly considered instead of LLR. It is unavoidable in retrospective studies. A variety of statistical methods, such as logistic regression, subgroup analyses, and sensitivity analyses, were used in this study to reduce the impact of a possible selection bias. Nevertheless, establishing well-defined criteria for LLR or OLR could help standardize liver resection and could improve patient outcomes. The present study had limitations. It was based on a retrospective cohort without a uniform distribution of patients in the two groups, especially in subgroups such as anemia before surgery and intraoperative blood transfusion. The database is based on the intent-to-treat principle. All surgeries were performed by experienced surgeons, and LLR was already performed at the three hospitals when the database was implemented; therefore, the impact of the learning curve could not be evaluated. Multivariable and subgroup analyses were performed to address these limitations in a real-world study, and the results were confirmed by sensitivity analyses. Although the survival outcomes did not significantly differ in primary MLTs in this study, more studies are needed to explore the survival outcomes of other malignant liver tumors. Additionally, this study had not added the Textbook outcome (TO) to evaluate the perioperative and survival outcome. Though the TO might be a composite useful tool for assessing patient-level hospital performance and optimizing patient outcomes after LLR and OLR, it’s a new concept and has not been fully studied to date46. Conclusions In conclusion, this large-sample intent-to-treat study suggests that laparoscopic vs. open surgery might have advantageous perioperative outcomes for liver resection, especially for women, non-drinkers, those without abdominal surgery history, and those with normal platelets. It also showed non-inferior survival outcomes for primary MLTs in both groups. These findings support the promotion of LLR as a safe and feasible approach for any liver disease with different resection classifications and complexities. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5