Comparison Between Minimally-Invasive Right Anterior and Right Posterior Sectionectomy versus Right Hepatectomy: An International Multicenter Propensity-Score Matched and Coarsened-Exact Matched Analysis of 1100 Patients

Abstract

Introduction: The role of minimally-invasive right anterior and right posterior sectionectomy (MI-RAS/MI-RPS) in right sided liver lesions remains debatable. However technically more demanding, these procedures might result in faster recovery and lower postoperative morbidity compared to minimally-invasive right hemihepatectomy (MI-RH). Methods: This is an international multicenter retrospective analysis of 1114 patients undergoing MI-RH, MI-RAS and MI-RPS at 21 centers between 2006-2019. MIS included pure laparoscopic, robotic or hybrid approach. A propensity matched and coarsened-exact matched analysis was performed. Results: 1100 cases met study criteria of whom 759 underwent laparoscopic, 283 robotic, 11 hand-assisted and 47 laparoscopic-assisted (hybrid surgery). There were 632 RH, 373 RPS and 95 RAS. There were no differences in baseline characteristics after matching. In the MI-RAS/MI-RPS group, median blood loss was higher (400 VS 300ml, p=0.001) as well as intraoperative blood transfusion (19.6% VS 10.7%, p=0.004). However, major morbidity was lower (7.1% VS 14.3%, p=0.007) and the need for reoperation was lower (1.4% VS 4.6%, p=0.029). The rate of close/involved margins was higher in the MI-RAS/MI-RPS group (23.4% VS 8.9%, p<0.001). These findings were consistent after both propensity and coarsened-exact matching. Conclusion: Although technically more demanding, MI-RAS/MI-RPS is a valuable alternative for MI-RH in right sided liver lesions with lower postoperative morbidity, possibly due to the preservation of parenchyma. However, the rate of close/involved margins is higher in these procedures. These findings might guide surgeons in selecting the appropriate procedure for their patients.Tabled 1EP01C-039TotalN = 1100Entire unmatched cohort1:1 Propensity matched cohort1:1 Coarsened-exact matched cohortMI-RA/RPSN = 468MI-RHN = 632P-valueMI-RA/RPSN = 280MI-RHN = 280P-value†MI-RA/RPSN = 188MI-RHN = 188P-value†Median blood loss (IQR), ml300 (200-600)400 (200-750)300 (200-500)0.001400 (200-750)300 (150-500)0.022400 (200-850)300 (120-500)0.001Intraoperative blood transfusion, n (%)169/1100 (15.4%)86/468 (18.4%)83/632 (13.1%)0.01755/280 (19.6%)30/280 (10.7%)0.00439/188 (20.7%)23/188 (12.2%)0.023Pringle maneuver applied, n (%)521/1084 (48.1%)276/466 (59.2%)245/618 (39.6%)<0.001179/278 (64.4%)102/277 (36.8%)<0.001114/187 (61.0%)67/187 (35.8%)<0.001Major morbidity (Clavien-Dindo grade> 2), n (%)129/1100 (11.7%)36/468 (7.7%)93/632 (14.7%)<0.00120/280 (7.1%)40/280 (14.3%)0.00714/188 (7.4%)22/188 (11.7%)0.139Reoperation, n (%)30/1100 (2.7%)5/468 (1.1%)25/632 (4.0%)0.0044/280 (1.4%)13/280 (4.6%)0.0293/188 (1.6%)10/188 (5.3%)0.02830-day mortality, n (%)12/1100 (1.1%)2/468 (0.4%)10/632 (1.6%)0.0682/280 (0.7%)2/280 (0.7%)1.0002/188 (1.1%)1/188 (0.5%)0.571Close/ involved margins (≤1mm) for malignancies, n (%)127/978 (13.0%)75/421 (17.8%)52/557 (9.3%)<0.00157/244 (23.4%)22/247 (8.9%)<0.00140/170 (23.5%)14/170 (8.2%)<0.001*P value in brackets were additionally obtained (as a sensitivity analysis) from mixed-effects negative binomial regression with robust variance, which treats length of stay as a count response, whereby a random-effects term was used to account for the paired data structure. The marginal model was used in the full (unmatched) cohort. The incidence rate ratios in the full cohort, PSM cohort, and CEM cohorts were 0.83 (95% CI 0.75 to 0.91), 0.87 (95% CI 0.77 to 0.98), and 0.85 (95% CI 0.75 to 0.96) respectively, indicating a shorter post-operative stay in the RAS/RPS arm compared to the RH arm.For categorical variables, denominators may differ from total numbers due to missing data Open table in a new tab *P value in brackets were additionally obtained (as a sensitivity analysis) from mixed-effects negative binomial regression with robust variance, which treats length of stay as a count response, whereby a random-effects term was used to account for the paired data structure. The marginal model was used in the full (unmatched) cohort. The incidence rate ratios in the full cohort, PSM cohort, and CEM cohorts were 0.83 (95% CI 0.75 to 0.91), 0.87 (95% CI 0.77 to 0.98), and 0.85 (95% CI 0.75 to 0.96) respectively, indicating a shorter post-operative stay in the RAS/RPS arm compared to the RH arm. For categorical variables, denominators may differ from total numbers due to missing data