Laparoscopic vs Open Pancreaticoduodenectomy—an Updated Meta-Analysis of Randomized Control Trials

There is ongoing debate regarding the usefulness of laparoscopic pancreaticoduodenectomy. This study aimed to analyze all the randomized control trials published including the most recent one. The study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and MOOSE guidelines. Heterogeneity was measured using Q tests and I2. The random-effects models were used to summarize the relative risks, odds ratios, and mean differences as appropriate. Four RCTs were included consisting of 818 patients. Four hundred eleven patients were in the laparoscopic group and 407 in the open pancreaticoduodenectomy group. Weighted baseline patient characteristics were similar except more patients with pancreatic adenocarcinoma and more males were there in the open pancreaticoduodenectomy group. There was no difference in-hospital stay, 90-day complication rate, 90-day mortality, R1 resection, postoperative pancreatic fistula, delayed gastric emptying, postpancreatectomy hemorrhage, and bile leak between the two groups. Operative time was more in the laparoscopic group. Blood loss [mean difference − 132.12 ml (− 172.60, − 91.65)] and surgical site infection [risk ratio 0.41 (0.17–1.0)] were significantly lesser in laparoscopic group. There was no benefit in-hospital stay or clinical outcomes after laparoscopic pancreaticoduodenectomy. Blood loss and surgical site infection were lesser in laparoscopic pancreaticoduodenectomy.


Background
Pancreaticoduodenectomy is the only curative option for pancreatic ductal adenocarcinomas, duodenal carcinoma, ampullary carcinoma, lower common bile duct cholangiocarcinoma, and various periampullary regions. It is still associated with very high morbidity and mortality [1].
Gagner and Pomp [2] first described laparoscopic pancreaticoduodenectomy in 1994. There have been many technological advancements since then. Many centers are now performing laparoscopic pancreaticoduodenectomy. However, it remains one of the most technically challenging surgeries [3].
Although some retrospective cohort studies showed benefits in in-hospital stay [4], recent randomized control trials failed to show benefits about the hospital stay. The widely discussed LEOPARD II trial showed more harm than benefit with laparoscopic pancreaticoduodenectomy and raised a debate in the scientific world, regarding the usefulness of laparoscopic pancreaticoduodenectomy. LEOPARD II had to be stopped earlier due to more harm related to laparoscopic pancreaticoduodenectomy [5].
Ausania et al. [6] published a meta-analysis of randomized control trials which showed no difference in mortality, morbidity, and hospital stay between laparoscopic vs open pancreaticoduodenectomy. However, all the randomized control trials included in that were having a limited number of patients.
Recently, Wang et al. [7] published a multicenter openlabel randomized control trial with a good sample size which showed laparoscopic pancreaticoduodenectomy was associated with the significantly reduced hospital stay. However, the difference in the in-hospital stay was just 1 day.
This study aimed to do an updated meta-analysis of all the randomized control trials including the recently published randomized control trial [7] to look for the difference in

Statistical Analysis
The meta-analysis was conducted using Review Manager 5.4. Heterogeneity was measured using Q tests and I 2 , and P < 0.10 was determined as significant, using the randomeffects model. The risk ratios (RR)/odds ratios (OR) were calculated for dichotomous data wherever applied and weighted mean differences (WMD) were used for continuous variables. Both differences were presented with 95% CI. For continuous variables, if data were presented with medians and ranges, then we calculated the means and standard deviations according to Hozo et al. [11,12]. If the study presented the median and interquartile range, the median was treated as the mean, and the interquartile ranges were calculated using 1.35 SDs, as described in the Cochrane handbook. Assessments of bias were done according to the Cochrane handbook [11]. Publication bias was assessed using funnel plots. The review was not registered.

Results
Four randomized control trials [5,7,13,14] consisting of 818 patients were included in the final analysis (Fig. 1). Four hundred eleven were in the laparoscopic arm and 407 in the open pancreaticoduodenectomy group. Study characteristics are described in Table 1. The risk of bias summary is described in Fig. 2.
We also did a weighted pooled analysis to compare the basic characteristics of the patients between laparoscopic pancreaticoduodenectomy and open pancreaticoduodenectomy (Table 2)

Other Parameters (Table 3) (Supplement Fig. 1)
There was no significant difference in reoperations, readmissions, lymph node harvest, and R1 resection between both groups. Blood loss was significantly less in the laparoscopic pancreaticoduodenectomy group [mean difference − 132. 12

Discussion
Surgeons are always attracted by technology. Minimal invasive techniques like laparoscopy and robotics have revolutionized many fields of surgery. However, there is an ongoing debate about the feasibility and usefulness of laparoscopic pancreaticoduodenectomy with various observational studies showing conflicting results with some showing increased mortality and morbidity. Whereas some are showing lack of benefits in terms of hospital stays, some show benefits [18][19][20][21][22][23].
Our meta-analysis aimed to do an updated meta-analysis of randomized control trials including a recently published meta-analysis by Wang et al. [7] which was not included in  an earlier meta-analysis published (6), but the randomized control trial published by Wang et al. [7] was the largest until now. Both PLOT and PADULAP trials [13,14] were single-center analyses and also they had a limited sample size. LEOPARD II [5] terminated their trial earlier owing to higher mortality in the laparoscopic group but their primary outcome was earlier recovery and that reduced power of the study. In that matter, the recently published meta-analysis by Wang et al. [7] was far better evidence; the issue with the trial by Wang et al. [7] was the significant crossovers from laparoscopic to open hence the difference in sample size in their intention to treat and per-protocol analysis. All the trials involved expert surgeons. Out of 4 randomized control trials included in our metaanalysis, LEOPARD II [5] had functional recovery as the primary outcome rest all had a hospital stay as the primary outcome. However, LEOPARD II mentioned hospital stay separately. There was no significant difference in hospital stay between laparoscopic and open pancreaticoduodenectomy in the meta-analysis. This was an important finding as PLOT, PADULAP, and the largest trial by Wang et al.     [7] showed that hospital stay was significantly less in laparoscopic pancreaticoduodenectomy; only LEOPARD II showed no difference. However, Wang et al. [7] showed just a 1-day difference, but a weighted meta-analysis showed no difference in hospital stay between laparoscopic and open pancreaticoduodenectomy. ICU stay was significantly less in the laparoscopic group, but the difference was just half to 1 day. There was no difference in mortality, total complications, clinically significant pancreatic fistula, delayed gastric emptying, postpancreatectomy hemorrhage, and bile leak between the laparoscopic and open pancreaticoduodenectomy groups in our meta-analysis.
Surgical site infections were lesser in the pancreaticoduodenectomy group but due to wider confidence intervals and borderline significance, the quality of this evidence can be termed low. Blood loss was significantly less in the laparoscopic group but in the pooled analysis of patients' characteristics, open pancreaticoduodenectomy had larger size tumors, more male patients, and more pancreatic ductal adenocarcinoma and hence pancreatic head tumors, which can be the cause.
There was no significant difference in reoperation, readmissions, R1 resections, or lymph node harvest between the group but all the surgeons involved in each trial were very experienced surgeons so real-world implications are remained to be seen. Operative time was significantly high in the laparoscopic group without visible benefits in clinical outcomes as per our meta-analysis.
If we evaluate the components of pancreatic fistula risk score [24] in included studies, three studies evaluated pancreas gland structure (754 patients), all four evaluated histology (818 patients), three studied evaluated blood loss (758 patients), and two studied duct diameter (655 patients). Blood loss was significantly more in the open pancreaticoduodenectomy group; more patients were operated on for pancreatic ductal adenocarcinoma in the open pancreaticoduodenectomy group. There was no difference in duct diameter or gland structure. So, despite higher pancreatic fistula risk in the open pancreaticoduodenectomy group, still, there was no difference in postoperative pancreatic fistula rates. These findings need to be evaluated in larger studies with pancreatic fistula, complications, and mortality as primary outcomes.
The strength of this meta-analysis is that it is an updated meta-analysis including the most recent and largest randomized control trial. Publication bias was non-significant in Eager's test in every analysis. A limitation is very few randomized control trials available. Heterogeneity was high and significant in in-hospital stay, operative time, blood loss, and lymph node harvest. For the other analysis, heterogeneity was mild to moderate as per I 2 analysis and non-significant.
In conclusion, laparoscopic pancreaticoduodenectomy had a little extra benefit over open pancreaticoduodenectomy. There may be slight benefits in surgical site infections, ICU stays, and blood loss with low certainty, but it seems there are hardly any benefits in in-hospital stay or clinically important outcomes but also it seems it is non-inferior to open pancreaticoduodenectomy concerning mortality, complications, pancreatic fistula rates, delayed gastric emptying, and postpancreatectomy hemorrhage.