The oncological safety in minimally invasive versus open distal pancreatectomy for pancreatic ductal adenocarcinoma: a systematic review and meta-analysis

The safety of minimally invasive distal pancreatectomy (MIDP) and open distal pancreatectomy (ODP) regarding oncological outcomes of pancreatic ductal adenocarcinoma (PDAC) remains inconclusive. Therefore, the aim of this study was to examine the oncological safety of MIDP and ODP for PDAC. Major databases including PubMed, Embase, Science Citation Index Expanded, and the Cochrane Library were searched for studies comparing outcomes in patients undergoing MIDP and ODP for PDAC from January 1994 to August 2018. In total, 11 retrospective comparative studies with 4829 patients (MIDP: 1076, ODP: 3753) were included. The primary outcome was long-term survival, including 3-year overall survival (OS) and 5-year OS. The 3-year OS (hazard ratio (HR): 1.03, 95% confidence interval (CI): 0.89, 1.21; P = 0.66) and 5-year OS (HR: 0.91, 95% CI: 0.65, 1.28; P = 0.59) showed no significant differences between the two groups. Furthermore, the positive surgical margin rate (weighted mean difference (WMD): 0.71, 95% CI: 0.56, 0.89, P = 0.003) was lower in the MIDP group. However, patients in the MIDP group had less intraoperative blood loss (WMD: −250.03, 95% CI: −359.68, −140.39; P < 0.00001), a shorter hospital stay (WMD: −2.76, 95% CI: −3.73, −1.78; P < 0.00001) and lower morbidity (OR: 0.57, 95% CI: 0.46, 0.71; P < 0.00001) and mortality (OR: 0.50, 95% CI: 0.31, 0.81, P = 0.005) than patients in the ODP group. The limited evidence suggested that MIDP might be safer with regard to oncological outcomes in PDAC patients. Therefore, future high-quality studies are needed to examine the oncological safety of MIDP.

Expanded, and the Cochrane Library to identify and retrieve studies published from January 1994 to August 2018 that examined distal pancreatectomy for PDAC (last search on August 8, 2018). The following medical search headings were used: ("Pancreatic Cancer" or "Pancreatic ductal adenocarcinoma" or "Pancreatic adenocarcinoma") and ("Left pancreatectomy" or "Distal Pancreatectomy" or "Pancreatectomy") and ("Laparoscopy" or "Laparoscopic-assisted" or "Laparoscopic" or "Robotic" or "Robot-assisted" or "Mini-invasive" or "Minimally invasive"). The language of full text articles was limited to English. In addition, the references of all selected articles were screened for any potential eligible studies. study selection. Studies were included in the meta-analysis if they met the following criteria: (1) human study; (2) published in English; (3) if studies were reported by the same institution (and/or authors), the study with either the larger sample size or higher quality was included. Studies meeting the following criteria were excluded: (1) case reports, letters, editorials, expert opinions and abstracts; (2) benign or other malignant tumors were included without reporting PDAC separately.
Qualitative assessment of the studies selected. The risk of bias of included nonrandomized studies was evaluated according to the risk of bias in nonrandomized studies of interventions (ROBINS-I) tool 13 .
Data extraction and synthesis. Each study was evaluated by two independent reviewers (Du-Jiang Yang and Jun-Jie Xiong) for inclusion or exclusion from the review. Disagreements between the reviewers were resolved by consensus and by consultation with a third reviewer (Hui-Min Lu) when necessary. Data were collected by two independent researchers using standardized forms. The study characteristics, quality assessment, and intraoperative and postoperative outcomes were included. The means of outcomes were used for the meta-analysis unless otherwise mentioned. Furthermore, the means and standard deviations or medians and ranges were reported 14 .
The following data were extracted from each study: author, year, country, study design, study duration, number of patients, age, sex, body mass index, tumor size, operation time, intraoperative blood loss, hospital stay, morbidity, postoperative pancreatic fistula (POPF) occurrence, mortality, positive surgical margin rate, lymph nodes harvested, perineural and lymphovascular invasion, multivisceral resection, positive lymph nodes, recurrence, adjuvant therapy and follow-up time.
outcomes of interest and definitions. Minimally invasive was defined as a laparoscopic, robotic, laparoscopic-assisted or robot-assisted procedure. The primary outcome was overall survival (OS) time, which was defined as the time from the operation until death or the final follow-up evaluation. The secondary outcomes included operative time, intraoperative blood loss, hospital stay, morbidity, POPF occurrence, mortality, positive surgical margin, lymph nodes harvested, recurrence, perineural and lymphovascular invasion, multivisceral resection, positive lymph nodes, and adjuvant therapy. The operative time was defined as the interval from incision to suturing of the skin. Intraoperative blood loss was defined as the blood loss during surgery. The hospital stay was defined as the length from patient admission to discharge from the hospital. Morbidity was defined as all complications that occurred during the hospital stay or within 90 days after surgery. POPF occurrence was defined according to the International Study Group of Pancreatic Fistula (ISGPF) 15 . Mortality was defined as the number of deaths occurring during hospitalization or within 30 days after surgery. A positive surgical margin was defined as tumor in the transection and circumferential margins with a distance from the margin to the tumor of <1 mm 12,16,17 or <0 mm 18,19 . Lymph nodes harvested was defined as the number of lymph nodes obtained during the operation. Recurrence was defined as local recurrence or distant metastasis. Adjuvant therapy was defined as the number of patients who received adjuvant therapy including postoperative radiation or chemotherapy. statistical analysis. A meta-analysis was performed using Review Manager Version 5.3 software (The Cochrane Collaboration, Oxford, UK). Variables are expressed as weighted mean differences (WMDs) or odds ratios (ORs) as appropriate, with their corresponding 95% confidence interval (CI). For continuous and categorical variables, treatment effects are expressed as WMDs and ORs with corresponding 95% CIs. For the survival analysis, we extracted data from the survival curve using a method reported in a previous study, and hazard ratios (HRs) were used for the quantitative analysis 20 . A Chi-square test was used to assess heterogeneity, with P < 0.1 considered significant. I 2 values were used to evaluate statistical heterogeneity; an I 2 value of 50% or more indicated the presence of heterogeneity 21 . The fixed-effects model was initially used for all outcomes, while the random-effects model was used if the test suggested rejection of the assumption of homogeneity 22 . Descriptive methods were used if the data were inappropriate for meta-analysis. A sensitivity analysis was performed by removing individual studies from the data set and analyzing the effect on the overall results to identify sources of heterogeneity. A funnel plot was constructed to evaluate potential publication bias based on the OS and positive surgical margin rate 23 .

Results
Description of the included studies. The reporting of this systematic review was in accordance with the PRISMA Statement 24 . A flow diagram of the study is shown in Fig. 1. In total, 2612 studies were identified from the electronic databases, and 954 studies were excluded because they were duplicate publications. Finally, the full texts of 40 studies 11,12,[16][17][18][19] were screened for eligibility; however, 28 studies were excluded for various reasons 11 19 , and Europe 12 . The tumor size, which was reported in 10 studies, was larger in patients who underwent ODP than in those who underwent MIDP (WMD: −0.45, 95% CI: −0.85, −0.05; P = 0.03). All included studies were retrospective comparative studies. The risk of bias in included studies was evaluated by the ROBINS-I tool ( Table 2). Based on the ROBINS-I assessment, two studies 12,37 were considered to have low risk, four studies 17,18,26,40 were considered to have moderate risk, and five studies 16,19,36,44,56 were considered to have serious risk of bias. Perioperative and tumor outcomes are shown in Tables 3 and 4. The results of the meta-analysis are shown in Table 5   Sensitivity analysis. A sensitivity analysis was conducted by changing the type of effects model or excluding individual studies from the outcome analysis. The results in operation time were different in the fixed and random effects models. Although high heterogeneity was found for blood loss, this result was presented in both the fixed and random effects models. The heterogeneity was large for hospital stay; however, the heterogeneity was zero when two studies (Sulipice 40 and Bauman 59 were excluded.
Publication bias. Funnel plots based on the 3-year OS, 5-year OS and positive surgical margins are shown in Fig. 2. No study was outside the limits of the 95% CI; therefore, no evidence of publication bias was present.

Discussion
MIDP has been performed for more than a decade, and the technical feasibility, safety and clinical benefit have been well documented by many studies comparing it to ODP 34,60,61 . However, some centers still favor adopting a conventional, open approach for PDAC 62,63 .This systematic review and meta-analysis suggested that the OS and oncologic outcomes are not significantly different between the MIDP and ODP groups. Although the pooled results were mainly based on nonrandomized data, the results suggested that MIDP did not adversely affect long-term survival in PDAC patients. Furthermore, the MIDP group also had a lower rate of positive surgical margins and less perineural and lymphovascular invasion. Therefore, MIDP might be feasible and safe regarding long-term survival and oncological outcomes in PDAC patients. Currently, the rate of long-term survival after distal resection for PDAC patients remains poor. Despite the significant improvement in operative skill for MIDP, the long-term survival is still not different between the two groups of PDAC patients. Radical surgical resection remains the only potentially curative treatment for patients with resectable PDAC 64 . A recent review from a large randomized trial involving resection of PDAC showed  positive margin rates ranging from 0% to 83% 65 . The finding in the current report is consistent with those in the published literature 30 . Importantly, the MIDP results in fewer resections with positive margins than ODP. This advantage might be due to the clearer view and more elaborate procedure of MIDP. However, the rate of positive margin resection in present study should be interpreted with caution because margin status was influenced by the use of different definitions and pathologic assessment methods. Furthermore, conclusions may vary by pathologists and institutions due to the absence of standardized pathology assessment and reporting. Moreover, surgeons determine which method is more suitable for the PDAC patients before surgery. Therefore, patients with less extensive cancer may undergo MIDP, and patients with more extensive cancer would undergo ODP 16,36 .
Regarding other oncological outcomes, no significant differences were found for recurrence, lymph nodes harvested and positive lymph nodes. The extent of the lymphadenectomy determined the number of retrieved lymph nodes. According to one report, at least 12 lymph nodes should be evaluated histologically to determine metastatic disease and adequately stage PDAC patients 66 . Moreover, according to the consensus of the International Study Group on Pancreatic Surgery (ISGPS), a standard lymphadenectomy with resection of lymph node stations 10, 11, and 18 is recommended for cancer of the body and tail of the pancreas 67 . In most included studies, the mean number of harvested lymph nodes was greater than 12. In present study, no significant difference was observed in the number of harvested lymph nodes between the two groups. Based on this finding, MIDP is a reasonable procedure for PDAC patients.
No difference was observed in some clinical outcomes. However, MIDP is associated with a steep learning curve. This observation potentially affected several outcomes, including operative time, blood loss and the length of hospital stay 16 . Therefore, the hospital stay and blood loss were decreased in the MIDP group. Moreover, MIDP was associated with significantly lower morbidity and mortality rates than ODP. Our results are consistent with those of many studies. The technical feasibility, safety and clinical benefit have been well confirmed by various matched comparison studies 34,60,61 . Thus, MIDP might be safer with regard to oncological outcomes in PDAC patients.   However, our study has some limitations. First, all of the included studies were retrospective, which could lead to inevitable selection bias toward resection of larger or locally advanced tumors via the open approach, especially in the earlier years of the study period. Second, the follow-up time differed and was relatively short in some studies. Therefore, the long-term survival was difficult to accurately estimate. Third, the definitions of some outcomes were different among studies.

Conclusion
In summary, the meta-analysis demonstrated that MIDP might be safer with regard to the oncological outcomes of PDAC patients. However, these results need to be confirmed in a future prospective randomized trial.