Patient selection
First, we reviewed the medical records of consecutive patients who underwent OPD at Zhongshan Hospital, Fudan University (Shanghai, China) or LPD at Zhejiang Provincial People’s Hospital (Hangzhou, Zhejiang, China) between January 2013 and September 2017. Patients with pathologically confirmed PDAC and without any evidence of distant metastasis by preoperative examinations were included. All of the included cases met the resectable criteria laid down by the National Comprehensive Cancer Network guidelines for preoperative assessments [17]. We (YP Mou, RC Zhang and YC Zhou) began to perform LPD in 2012, and more than 10 LPDs were completed for less aggressive pancreatic tumors, such as neuroendocrine tumors and cystic neoplasms, in that year. OPDs were performed by five surgeons (DY Jin, WH Lou, DS Wang, WC Wu and TT Kuang). This research was approved by the Ethics Committee of both Zhongshan Hospital and Zhejiang Provincial People’s Hospital. Both hospitals are high-volume pancreatic surgical centers, and the surgical teams are both experienced in open and laparoscopic surgery.
Variables and definitions
Demographic, clinical, and pathological data were extracted from corresponding medical records. Baseline characteristics included patient age, gender, body mass index, American Society of Anesthesiologists score, Charlson comorbidity index [18], year of operation, tumor differentiation, nerve invasion, T stage, N stage, TNM stage, and history of adjuvant treatment. TNM stage was classified according to the American Joint Committee on Cancer staging system (8th edition) [19]. Adjuvant treatment comprised of postoperative chemotherapy (e.g., gemcitabine, S−1) or chemoradiotherapy (e.g., gemcitabine plus radiotherapy). The primary endpoint was median overall survival (OS). OS was defined as the duration from the first day after surgery to either the date of death or the last follow-up. Secondary endpoints included postoperative complications, digital subtraction angiography (DSA) intervention, reoperation, in-hospital mortality, readmission, postoperative length of stay, and time to adjuvant chemotherapy. Complications were evaluated based on the Clavien-Dindo classification system [20], and the highest grade for each patient was analyzed for overall postoperative complications. Postoperative pancreatic fistula (POPF) [21], delayed gastric emptying (DGE) [22], and postpancreatectomy hemorrhage (PPH) [23] were defined and classified according to the criteria set out by the International Study Group of Pancreatic Surgery (ISGPS). Similarly, bile leakage (BL) was recorded and graded according to the standard definitions of the International Study Group of Liver Surgery [24]. Wound infection was defined as purulent drainage from the incision or/and positive findings of culture of the fluid or tissue aseptically obtained from the incision. Operative details, such as duration of the operation, estimated blood loss, intraoperative blood transfusion, vascular resection, number of resected lymph nodes, and R0 resection rate, were also analyzed. R0 resection was defined as the absence of tumor cells on the pancreatic neck margin, the retroperitoneal margin, and the bile duct margin. The definitions for all these parameters were unified by both teams at the beginning of this study.
Surgical technique
The technique we used for LPD was as described in a previous publication [25]. Briefly, five trocars were placed in the abdomen in a V shape. If there was no sign of metastasis upon laparotomy, the gastrocolic ligament was divided to enter the lesser omental sac and expose the anterior surface of the pancreas. Then, the right gastroepiploic vessels were isolated and excised. After dissecting along the superior border of the pancreas, the common hepatic artery, gastroduodenal artery, and proper hepatic artery were identified, and the gastroduodenal artery was further ligated. The inferior border of the pancreas was then dissected to expose the portal vein and superior mesenteric vein, and a retropancreatic tunnel was established prior to the Kocher maneuver. The proximal jejunum and distal stomach or proximal duodenum were divided with liner staplers, then the gallbladder was isolated, and the common hepatic duct was transected with scissors. Subsequently, the pancreatic neck and uncinate process were divided using ultrasonic shears. Finally, the specimen was placed into an endoscopic bag for retrieval. For reconstruction, the Child’s procedure was used, involving pancreaticojejunostomy, hepaticojejunostomy, and gastrojejunostomy in a sequential order. An internal stent was then used to maneuver an end-to-side, duct-to-mucosa pancreaticojejunostomy. This was followed by an end-to-side hepaticojejunostomy using a 4-0 absorbable suture. Subsequently, an end-to-side gastrojejunostomy was performed in an antecolic type using a stapler.
Our OPD procedure resembled the LPD procedure except for two aspects. First, we mainly performed the Kocher maneuver as an initial step after negative abdominal exploration. Second, various fashions of pancreaticojejunostomy were adopted by our OPD team depending on the surgeon’s individual preferences.
Postoperative treatment
For LPD patients, we routinely stopped using antibiotics at 2 days after surgery if there were no definite POPF, BL, or infections. The nasogastric tube was usually removed on the first or second postoperative day if the volume of digestive juice was less than 200 mL/day and had a normal appearance. The patients were then encouraged to take a liquid diet, followed by a semi-liquid diet. Amylase measurements of the drainage fluid were conducted since the first postoperative day, and the drainage tubes were removed if the volume was less than 50 mL/day for three consecutive days and the amylase level was lower than three times the upper normal serum amylase level and had a normal appearance.
The postoperative treatments in the OPD group shared similarities with those in the LPD group. However, our OPD team usually removed the nasogastric tubes on the third or fourth postoperative day in consideration of the relatively late recovery of gastrointestinal motility following open surgery. In addition, abdominal computed tomography (CT) scan was routinely performed before the removal of drainage tubes for patients in the OPD group, but not routinely performed for patients in the LPD group.
Follow-up
Patients were recommended to return to the outpatient department for follow-up 1 month after being discharged, every 3–6 months for the first 2 years, then annually. We routinely performed a variety of tests, including blood tests, liver and kidney function tests, carbohydrate antigen 19-9 detection, and chest X-rays; abdominal CT scans were performed when appropriate. Survival data were collected by searching the electronical outpatient system or by telephone interviews. The last follow-up was in November 2017.
Statistical analyses
To minimize the effect of confounding factors and potential bias between the OPD and LPD groups, propensity score was calculated using logistic regression, and a 2:1 patient matching was performed using the nearest-neighbor matching method without replacement. A caliper radius equal to a standard deviation of 0.1 was set to prevent poor matching. Variables included in the matching model were gender, tumor differentiation, nerve invasion, T stage, and adjuvant treatment; these were distributed differently between the two original groups.
Continuous variables are described as medians and interquartile ranges (IQR), while categorical variables are expressed as whole numbers and percentages. Two-tailed unpaired t tests were performed to compare the continuous variables that followed normal distributions; otherwise, the Mann–Whitney U test was used. The distribution differences of categorical variables between the two groups were analyzed using the Pearson Chi square tests or Fisher’s exact tests. Survival analyses were conducted using the Kaplan–Meier method with log-rank tests. Univariable and multivariable Cox regression analyses were used to identify independent risk factors of OS. All statistical analyses were performed with SPSS software (version 22.0, IBM Corp., Armonk, NY, USA). Propensity score matching (PSM) was carried out using the “PS MATCHING 3.04”, “SPSS Statistics R Essentials 22.0”, and “R-2.15.3-win” R packages. The GraphPad Prism software (version 5.01, GraphPad Software Inc., San Diego, CA, USA) was used to plot the OS curves. All P values were based on 2-sided statistical analyses, and P < 0.05 was considered significant.