Propensity Score Analysis of Outcomes Following Laparoscopic or Open Radical Resection for Gallbladder Cancer in T2 and T3 Stages

Although laparoscopic radical resection (LRR) has long been contraindicated in gallbladder cancer (GBC), recent studies have demonstrated laparoscopic surgery did not adversely affect the perioperative and survival outcomes of GBC patients. However, these literatures are mainly focused on GBC of relatively early stages or incidental GBC. This study aimed to investigate the perioperative and long-term outcomes of LRR versus open radical resection (ORR) for GBCs in T2 and T3 stages. A retrospective study was conducted on 99 patients with GBC of T2 and T3 stages who underwent radical resection at Zhejiang Provincial People’s Hospital from January 2010 to December 2020. A 1:1 propensity score matching (PSM), which is widely used to reduce selection bias, was performed to compare the surgical outcomes and long-term prognosis between LRR and ORR. A logistic regression analysis was implemented to identify the predictive risk factors of postoperative overall survival. By using PSM, the baseline characteristics of two groups (with 30 patients in each group) were generally well balanced. In the LRR group, the length of operation was significantly longer than the ORR group, but the intraoperative bleeding and postoperative days of hospital stay were significantly decreased compared to the ORR group. The two groups showed comparable outcomes regarding the incidence of biliary reconstruction, lymph node yield, the incidence of postoperative morbidities, the incidence of Clavien–Dindo (C–D) grades III–IV, the days of drainage tubes indwelling, mortality at 30 postoperative days and 90 postoperative days, and the incidence of port-site metastasis. The 1-, 2-, and 3-year overall survival rates were 61.2, 40.1, and 30.1%, respectively, in the LRR group, and 53.3, 40.1, and 40.1%, respectively, in the OLR group (P = 0.644). On multivariate analysis, T stage, vascular invasion, and tumor differentiation were found to be the independent risk factors for overall survival of GBC in T2 and T3 stages. For GBC in T2 and T3 stages, LRR can achieve comparable perioperative outcomes and similar long-term survival benefit compared to ORR. LRR tends to show advantages over ORR regarding intraoperative bleeding and postoperative days of hospital stay.


Introduction
Gallbladder cancer (GBC), the most common biliary tract malignancy, is known to have a dismal prognosis. 1 Curative radical resection remains the gold standard treatment for GBC patients, despite the median survival was 26 months, and 5-year survival was 35% after surgical resection. 2,3 Although laparoscopic surgery is widely used in gastrointestinal malignancies, 4-7 the application of laparoscopic approach in the curative resection of GBCs remains controversial. Peritoneal dissemination of cancer cells, port-site metastasis, and inadequacies of radical resection are the Changwei Dou and Chunxu Zhang contributed equally to this study.
Recently, increasing literatures reported encouraging results regarding the application of laparoscopic surgery in GBC. Agarwal et al. demonstrated that there was no significant difference in morbidity, mortality, lymph node yield, and recurrence rate by retrospectively comparing outcomes of 24 cases of laparoscopic radical cholecystectomy and 46 cases of open surgery. 8 Itano et al. showed no significant difference regarding postoperative morbidity, mortality, lymph node yield, and recurrence rate between laparoscopic and open cases of T2 GBC. 9 Yoon et al. reported favorable long-term oncologic results of laparoscopic radical cholecystectomy by including 2 cases of Tis GBC, 10 cases of T1a GBC, 8 cases of T1b GBC, and 25 cases of T2 GBC. 10 However, these literatures are mainly focused on GBC of relatively early stages or incidental GBC. 11 Therefore, investigations examining the efficacy of laparoscopic surgery for advanced GBC remain scare. In this study, the results of laparoscopic radical resection for GBC of T2 and T3 were compared with those carried out by the open approach, focusing on perioperative outcomes and long-term survival.

Patients' Cohort and Data Collection
Data for a consecutive series of patients who underwent radical resection for GBC, by either open or laparoscopic approach between January 2010 and December 2020 were retrieved from Zhejiang Provincial People's Hospital from a prospectively collected database. The inclusion criteria for this study were as follows: (1) histopathologically confirmed advanced GBC (stages T2-T3), (2) patients underwent radical surgery (open or laparoscopic approach), (3) no preoperative antitumor therapies, (4) had complete perioperative and follow-up data. Exclusion criteria were as follows: patients who received simple cholecystectomy, patients who had another primary malignancy at the time of surgery, or patients who were lost during follow-up. The primary aim of the analysis was to compare short-and longterm outcomes of patients receiving a laparoscopic radical resection (LRR) with those receiving an open radical resection (ORR). The secondary aim was to explore, in the entire cohort of patients, factors affecting the overall survival of GBC patients. Approval for the study was obtained from the Institutional Review Board of Zhejiang Provincial People's Hospital.

Preoperative Management
No neoadjuvant locoregional or systemic treatments were applied before operation. Abdominal ultrasonography and contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) were routinely performed in patients with advanced GBC. Magnetic resonance cholangiopancreatography (MRCP) was performed in patients with obstructive jaundice. CT angiography was used to evaluate patients with suspected vessel invasion. Positron emission tomography was performed in selected patients to rule out systemic metastasis before radical surgery. Laboratory examinations, including routine blood tests, liver function tests, as well as serum carbohydrate antigen (CA) 19-9 levels were routinely performed in all the patients. Endoscopic or percutaneous biliary drainage was performed in the patients with obstructive jaundice with total serum bilirubin (TB) > 200 µmol/L until TB levels decreased to < 80 µmol/L. Indocyanine green (ICG) clearance test at 15 min along with Child-Pugh (CTP) classification was employed to evaluate the liver function and reserve.

Surgical Procedure
The patients undergoing laparoscopic surgery were placed in the reverse Trendelenburg position. Routinely, the fiveport method was utilized for extended resection of GBC (Fig. 1A). For cases requiring bilioenteric anastomosis, the six-port method was used (Fig. 1B). Careful inspection of the peritoneal cavity was performed after establishment of pneumoperitoneum to rule out peritoneal or visceral metastasis. Frozen section examination of para-aortic LNs was performed to decide whether radical resection should be terminated. Wedge resection of the GB bed (≥ 2 cm) ( Fig. 1C) or segment IVb/V resection (Fig. 1D) or major hepatectomy was chosen depending on the extent of liver parenchyma invasion. The portal inflow was blocked using the Pringle maneuver when severe intraoperative bleeding occurred. Extra-hepatic bile duct resection and bilioenteric anastomosis (end-to-side method) were performed if the margin of cystic duct was positive. LNs along the extrahepatic bile ducts, the hepatoduodenal ligament, the hilar areas, and the pancreatic head region were dissected ( Fig. 1E and F). Pancreatoduodenectomy was performed for cases with pancreatic parenchymal or intra-pancreatic bile duct infiltration. Resection of adjacent organs including the colon, stomach, and duodenum, or vessel construction was performed to achieve R0 resection when necessary. Resected specimens were inserted into vinyl bags and extracted through the umbilical port.

Postoperative Management and Follow-up
Blood tests and liver function tests were performed every 3 days after operation. The level of bilirubin in drainage fluid was measured. Bile leakage was defined according to the criteria proposed by ISGLS. 12 Morbidity was classified according to the Clavien-Dindo classification. Drainage tubes were removed if contrast-enhanced CT did not show any fluid accumulation in peritoneal cavity. Postoperative chemotherapy was recommended to all GBC patients (T2 and T3) without contradiction to chemotherapy. The postoperative follow-up protocol included clinical examination, liver function tests, serum tumor marker levels (CA199), abdominal ultrasonography, and enhanced CT or MRI every month during the first 3 months, followed by every 3 months during the initial 2 years and every 6 months during the next 3 years. Telephone follow-up was conducted to obtain the updated survival status of every patient. OS was calculated as the interval from the date of resection until the date of death, with censoring at the date of the last follow-up.

PSM
A 1:1 propensity score matching (PSM) was performed to reduce bias resulting from lack of randomization due to the different co-variable distribution among the patients who received either of the surgical approaches to radical resection. The caliper of 0.05 was used in the model. A consensus regarding which variables should be included in the model was reached among all the authors. The balanced variables included age, gender, body mass index (BMI), the history of gallbladder stone, CA199 level, total bilirubin, ALT, T stage, N stage, M stage, vascular invasion, neural invasion, pathological type, tumor differentiation, and postoperative chemotherapy.

Risk Factors Affecting Overall Survival
The univariate and multivariate analyses were performed on the entire patients' cohort to identify the risk factors affecting overall survival. Variables that might affect the postoperative survival were incorporated into the univariate analysis including age, gender, body mass index (BMI), the history of gallbladder stone, CA199 level, preoperative total bilirubin, preoperative ALT, the surgical approach, T stage, N stage, M stage, vascular invasion, neural invasion, pathological type, tumor differentiation, operation time, blood loss, and postoperative chemotherapy. The variables (those with P < 0.05 in univariate analysis) were incorporated into the multivariate analysis of logistic regression to identify the independent risk factors affecting postoperative overall survival.

Statistical Analysis
SPSS statistical software (IBM SPSS, Inc., Chicago, IL, version 25) and Graphpad were used for statistical analysis. Categorical variables which were presented as absolute numbers (percentage) were compared between groups by the χ 2 test. Continuous variables presented as mean ± standard deviation were compared between groups using Student's t test. Logrank test were used to determine the difference of overall survival between two groups. P < 0.05 was considered to be statistically significant difference.

Results
During the study period, 99 GBC patients of T2 and T3 underwent radical resection. There were 56 patients who underwent LRR, and 43 patients received ORR. After a 1:1 PSM, there were 30 patients in each group.

Baseline and Pathological Characteristics Before PSM
There was no significant difference on age, gender, BMI,  15, P < 0.01). Lymph node yield was comparable between two groups. Two groups also showed comparable incidence of postoperative morbidities including bile leakage, postoperative bleeding, and abdominal abscess. The incidence of Clavien-Dindo (C-D) Grades III-IV was similar between groups. The LRR group showed a trend of decreased days of drainage tubes indwelling (7.46 ± 0.62 vs 9.30 ± 0.85), although the difference did not show statistical significance (P = 0.077). Compared with the ORR group, the postoperative days of hospital stay was significantly decreased in the LRR group (10.32 ± 0.60 vs 14.74 ± 0.91, P < 0.01). Mortality at 30 postoperative days and 90 postoperative days, as well as the incidence port-site metastasis did not differ significantly between two groups. As for long-term survival, the 1-, 2-, and 3-year overall survival rates were 70.1, 50.2, and 44.6%, respectively, in the LRR group, and 52.7, 40.7, and 38.2%, respectively, in the OLR group (P = 0.155, Fig. 2A).

Baseline and Pathological Characteristics After PSM
After PSM, the baseline and pathological characteristics including age, gender, BMI, history of gallstone, preoperative CA199 level, preoperative TB, preoperative ALT, T stage, N stage, M stage, pathological type, tumor differentiation, perineural invasion, and vascular invasion showed good comparability between two groups as shown in Table 3.  Fig. 2B).

Risk Factors of Overall Survival After Operation
As shown in Table 5, univariate analysis identified T stage, N stage, vascular invasion, tumor differentiation, and biliary reconstruction to be the risk factor for overall survival. Multivariate analysis further identified the T stage and tumor differentiation to be the independent risk factors of overall survival for GBC of T2 and T3 stages.

Discussion
With the development of laparoscopic techniques and equipment, the laparoscopic surgery has been widely applied in gastric cancer, liver cancer, pancreatic cancer, and colon cancer. In the case of GBC, the development and application of laparoscopic surgery is still in initial stage compared with that of laparoscopic surgery in other gastrointestinal cancers. During recent years, several studies with limited number of patients reported the perioperative and survival outcomes of laparoscopic approach for early GBC or incidental GBC. 6,[8][9][10] A recent survey of surgical experts revealed that laparoscopic surgery has been gradually accepted for suspicious or early GBC. 13 However, the role of laparoscopic radical resection for GBC in relatively late stages remains uncovered.
In this study, we retrospectively analyzed the outcomes of 99 GBC cases in T2 and T3 stages who underwent laparoscopic (n = 56) or open (n = 43) radical resection. Our data showed that laparoscopic surgery achieved comparable perioperative outcomes and long-term survival compared with open surgery. Laparoscopic surgery even demonstrated advantages over open surgery regarding intraoperative bleeding and postoperative hospital stay. As in many other surgical centers, the LRR in our institution started with GBC in relatively early stages and moved progressively to advanced lesions. Therefore, the LRR and ORR groups in the current study were different regarding preoperative TB level, preoperative ALT level, T stage distribution, as a result of the decision-making in surgical approach that considered characteristics of individual GBC case, surgeons' attitude, and experience. To reduce this selection bias, PSM was performed and created two comparable groups of GBC patients in T2 and T3 stages who underwent LRR (n = 30) versus ORR (n = 30). After PSM, LRR and ORR groups showed better comparability in baseline and pathological characteristics. The data of perioperative and long-term survival further validated that compared with open surgery, laparoscopic approach did not adversely affect the perioperative safety, oncological outcomes, and overall survival of GBC in T2 and T3 stages.
One issue preventing the application of laparoscopic approach in GBC is the risk of port-site implantation or metastasis due to intraoperative gallbladder perforation. Although the previous study reported high incidence of portsite metastasis, there was no case of port-site metastasis in the present study. This data was consistent with recently published literatures. 14,15 Preoperative recognition of malignancy, precautionary operative manipulation, en-bloc resection of gallbladder and liver parenchymal, and use of plastic bag for specimen removal are probably attributable to this low incidence of port-site metastasis. Therefore, concern about port-site metastasis should not be an obstacle for the application of laparoscopic approach for GBC.
Another concern counting against the application of LRR for GBC is regarding its oncologic adequacy and intraoperative safety. Lymph node dissection is a key procedure during radical resection of GBC. 16 The extent of lymph node dissection for GBC includes removal of lymph nodes around the hepatoduodenal ligament and the posterior superior pancreaticoduodenal area. The experience of laparoscopic surgery for the treatment of gastrointestinal cancers has provided a foundation for complete lymph node dissection for GBC. Our data showed that laparoscopic approach was feasible for required lymph node dissection ( Fig. 1E-F), and the LRR group had comparable lymph node yield with the ORR group, which was consistent with other studies. Although the length of operation was increased in the LRR group, laparoscopic surgery was associated with significantly decreased intraoperative bleeding. The incidence of postoperative morbidity, the mortality at 30 and 90 postoperative days were comparable between the LRR and ORR groups. Moreover, laparoscopic resection was associated with decreased days of drainage tubes indwelling and postoperative days of hospital stay. These data demonstrate that laparoscopic approach can achieve satisfactory oncologic adequacy and intraoperative safety.
To explore the influence of various covariates on the overall survival of GBC in T2 and T3 stages, a multivariable logistic regression analysis was carried out on the entire cohort of patients. The analysis confirmed T stage and tumor differentiation to be the independent risk factors of overall survival for GBC in T2 and T3 stages. These indicate that the pathological characteristics, instead of the surgical approach (LRR vs. ORR), are the key factors determining the long-term prognosis of GBC patients.
There are some limitations to the present study. Firstly, this study is a retrospective, non-randomized, single-institution investigation. Although this is the largest retrospective study examining the perioperative and long-term outcomes of laparoscopic resection for GBC in T2 and T3 stages, the confidence level of this study will be improved by including more GBC cases from multiple surgical centers. Secondly, the intrinsic variability in baseline patient characteristics, pathological features could not be fully captured by the categorical variables described. This means other hidden differences could possibly exist between the groups, even in the propensity score-matched series, that may have influenced the choice of surgical approach and clinical outcomes. Finally, an intrinsic limitation of 1:1 PSM is that many control subjects (ORR) or treated subjects (LRR) are excluded from statistical analysis, which can result in loss of information and decrease the precision of the estimated association between surgical approach and clinical outcomes.
In all, even considering these limitations, this study suggests that laparoscopic resection for GBC in T2 and T3 stages shows perioperative outcomes and long-term survival comparable to those of open surgery and reduces the intraoperative bleeding and the duration of postoperative hospital stay.