Roux-en-Y and Billroth II Reconstruction after Pancreaticoduodenectomy: A Meta-Analysis of Complications

Objective To evaluate Roux-en-Y and Billroth II reconstruction following pancreaticoduodenectomy (PD). Methods PubMed, Embase, the Cochrane Library, and the Web of Science were searched to identify randomized controlled trials (RCTs) and controlled clinical trials that compared Roux-en-Y and Billroth II reconstruction following PD up to December 2019. RevMan 5.3 software was used for the statistical analysis. Results Four RCTs and five controlled clinical trials were included, with a total of 1,072 patients (500 and 572 patients in the Roux-en-Y and Billroth II groups, respectively). No significant differences in delayed gastric emptying (DGE), A-grade DGE, B-grade DGE, or C-grade DGE were observed between the Roux-en-Y and Billroth II reconstruction groups after PD (odds ratio [OR] = 1.01, 95% confidence interval [CI]: 0.50–2.03, P = 0.98; OR = 0.49, 95% CI: 0.17–1.45, P = 0.20; OR = 0.63, 95% CI: 0.29–1.38, P = 0.25; and OR = 2.13, 95% CI: 0.38–11.99, P = 0.39). No significant difference in the incidence of postoperative pancreatic fistula, abscess, bile leaks, infection, postoperative bleeding, or the length of the postoperative hospital stay was observed between the Roux-en-Y and Billroth II groups (P > 0.05), but the operation time was significantly different (mean difference [MD] = 31.65, 95% CI: 7.14–56.17, P = 0.01). Conclusions Billroth II reconstruction after PD did not significantly reduce the incidence of DGE or other complications but shortened the operation time compared to Roux-en-Y reconstruction. However, the results must be verified by further high-quality, large RCTs or controlled clinical trials.


Introduction
The incidence of pancreatic cancer is rising, and it is estimated that pancreatic cancer will become the second most deadly cancer in the world by 2030 [1]. Scholars have performed a considerable amount of research on pancreaticoduodenectomy (PD) since Whipple et al. first proposed the concept in 1935 [2]. Complications after PD have a significant impact on the postoperative quality of life of patients, for whom the total and clinically relevant delayed gastric emptying (DGE) incidence rates are 27.7% and 14.3%, respectively. The incidence of DGE is associated with the pancreatic resection type, pylorus preservation status (yes or no), antecolic and retrocolic gastrojejunal anastomosis, and gastrojejunal anastomosis type [3]. The incidence of DGE after Billroth II and Roux-en-Y gastrojejunal reconstruction has remained controversial. Based on the hypothesis that Roux-en-Y anatomy can prevent gastric contents from activating trypsin, Machado et al. [4] proposed that Roux-en-Y "protects" the pancreatojejunostomy. Results of single-center randomized controlled trials (RCTs) comparing the incidence of DGE between Roux-en-Y and Billroth II reconstruction are not consistent [5][6][7][8]. Meta-analyses have also reported contrasting results. Yang et al. [9] proposed that Billroth II reconstruction reduces the incidence of clinical DGE compared to Roux-en-Y; however, Li et al. [10] reported no difference between the two reconstruction types.
The purpose of this study is to evaluate the ability of Roux-en-Y and Billroth II reconstruction following PD to prevent DGE and other complications.       RCTs and CCTs comparing Roux-en-Y and Billroth II reconstruction following PD were searched. Two researchers independently screened the studies, cross-checked their quality, and asked a third researcher to settle any controversies regarding whether to include a study.

Sensitivity and Subgroup
Analyses. The heterogeneity data for DGE were P = 0:0002, I 2 = 74%; the results did not change after the subgroup analysis, indicating high reliability thereof. The heterogeneity values of A-grade DGE and Cgrade DGE were P = 0:03, I 2 = 70%, and P = 0:10, I 2 = 52%, respectively. We eliminated each study one by one, and the results showed good stability. The heterogeneity data for operation time and length of postoperative hospital stay were P < 0:00001, I 2 = 93%, and P < 0:0001, I 2 = 77%, respectively. We conducted a subgroup analysis based on the RCT and CCT groups. Operation time in the RCT group was not significantly different between the two reconstructions, while in the CCT group Roux-en-Y reconstruction had a longer operation time (MD = 47:28, 95% CI: 17.58-76.98, P = 0:002). Subgroup analysis did not reveal a significant difference in length of postoperative hospital stay between the Roux-en-Y group and Billroth II groups (P = 0:44 and P = 0:12 for RCT and CCT, respectively). Therefore, operation time was associated with study type. A high-quality RCT or CCT is still needed to verify the operation time difference between the two reconstructions.
3.5. Bias Analysis. We drew funnel plots based on the POPF data. The results showed that the 95% CI data were similar among the studies, which were distributed in a roughly symmetrical manner between the two sides of the midline, suggesting that the results were unaffected by publication bias and were thus highly reliable ( Figure 13).

Discussion
PD is a classic operation for benign and malignant lesions around the head of the pancreas. The incidence and complexity of postoperative complications are high due to the loss of organs and tissues [2,21,22]. DGE is a common complication after PD [23,24]. Among the factors affecting DGE, the method used to reconstruct the digestive tract has been controversial. The use of Billroth II or Roux-en-Y reconstruction following PD to prevent DGE is also controversial [5][6][7][8][13][14][15][16][17]. Meta-analyses that investigated the two reconstruction methods could not reach an agreement regarding which was superior [9,10].
Consequently, we conducted this meta-analysis to analyze the characteristics of the two reconstruction methods and provide evidence-based guidance for clinical work. The results showed that traditional Billroth II reconstruction shortened the operation time compared to Roux-en-Y reconstruction, but no significant differences in any other complications were observed. Our results also indicated no significant differences between Roux-en-Y reconstruction and Billroth II reconstruction in DGE    BioMed Research International In 2015, Yang et al. [9] included three high-quality RCTs in their meta-analysis and reported that Billroth II reconstruction lowered the incidence of B-and C-grade DGE, although the small number of included studies was a limitation. Li et al. [10] conducted a case-control study that included 43 patients undergoing Roux-en-Y and 43 patients undergoing traditional Billroth II reconstruction after PD. The operation time of the Roux-en-Y group was longer than that of the traditional reconstruction group. A subsequent systematic review evaluated a series of postoperative complications. Klaiber et al. [25] systematically evaluated the postoperative complications of the two methods, and the results were largely consistent with those of the present study. However, it remains to be determined whether carrying out pancreatogastrostomy after PD had an impact on the overall results [26].
Based on previous studies, our study systematically screened the literature for studies of PD and excluded those in which PD was followed by pancreatogastrostomy. Articles with high heterogeneity were also excluded, so our analyses were characterized by relatively high homogeneity. We conducted this new systematic evaluation to overcome the deficiencies of previous research, and the results are reliable. However, this study also had some shortcomings. Although homogeneity was high, the number of included studies was small; hence, a larger study is needed.
We performed subgroup and sensitivity analyses of the outcomes with high heterogeneity; the results remained unchanged, except for operation time, further indicating high reliability. We also performed a subgroup analysis of operation time based on the RCTs and CCTs, and the results indicated that study type was the primary influencing factor. Therefore, an RCT or CCT with a large sample size is needed to further compare the operation times of Roux-en-Y and Billroth II digestive tract reconstruction.
In conclusion, we demonstrated that Roux-en-Y reconstruction took longer than Billroth II reconstruction after PD. However, complications were not different between the two reconstruction types. Therefore, it is suggested that consideration of the difference in operation time and patients' condition is needed to ensure that a suitable personalized surgical plan is implemented.

Data Availability
The datasets generated or analyzed during the current study are available from the corresponding author on reasonable request.