Antecolic reconstruction is associated with a lower incidence of delayed gastric emptying compared to retrocolic technique after Whipple or pylorus-preserving pancreaticoduodenectomy

Supplemental Digital Content is available in the text


Introduction
Pancreatoduodenectomy (PD) is the standard surgical treatment for pancreatic and other periampullary malignancies. [1][2][3] Pylorus-preserving PD (PPPD) is a procedure with equal short-and long-term outcomes compared to the classic PD, which has been demonstrated by several studies. [4][5][6][7] Both PD and PPPD procedures were considered the only possibly curative treatment.
Despite the mortality of such procedures have decreased to less than 5% in high-volume surgical institutions, but postoperative morbidity remains relatively high. [8] One of the most frequent postoperative complications after PD or PPPD is delayed gastric emptying (DGE), which ranges from 5% to 80% among published studies. [9][10][11][12][13] Generally, DGE is not a lethal complication; it is associated with significantly longer hospital stay and higher costs. Several retrospective comparative studies (RSCSs), as well as prospective randomized trials (RCTs) comparing antecolic (AC) versus classic retrocolic (RC) reconstruction, has provided controversial results and the influence of the chosen route of reconstruction is still a matter of discussion. [14][15][16] Tani et al [14] suggested that the route of gastrojejunostomy (GJ) after a classic Whipple's resection or a duodenojejunostomy (DJ) after a PPPD might be to help to prevent DGE and further revealed that an AC route leads to lower incidences of DGE, as compared with a RC route. However, a RCT conducted by Eshuis et al [16] demonstrated that the AC and RC route of GJ reconstruction after PD does not influence the postoperative incidence of DGE.
Currently, 2 systematic reviews and meta-analyses focused on this topic have been published. [17,18] RC was reported to have higher incidence of DGE than cases associated with AC procedure. However, these previous meta-analyses maybe limited by its small sample sizes; the pooled results of these studies may be unreliable and underpowered for comparison among surgical techniques because potential studies and publication biases are more likely to occur. Therefore, we conducted an update analysis that included the largest available database from RCTs and RSCSs to overcome these limitations, and to investigate the relationship between the AC versus RC route of GJ after PD or DJ after PPPD and the incidence of DGE using the meta-analytical method and sensitivity analyses.

The literature review
This study was performed under a human investigational protocol that was approved and monitored by the Institutional Review Board of The first affiliated hospital of Chongqing Medical University. We adhered to the 2009 preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement. [19] An electronic databases search of the Medline Ovid, PubMed, Cochrane Library, and the Controlled Trials Registry was performed, using the following Mesh search headings and their combinations "Pancreaticoduodenectomy," "Pylorus-preserving pancreaticoduodenectomy," "Delayed gastric emptying," "gastrojejunostomy," "duodenojejunostomy," articles published in English language as a limit. We gathered all RCTs and RSCSs between 1991 and 2018 and compared the outcomes of AC and RC route of GJ after PD or DJ after PPPD. Retrieval time was ended by December 2017. Title and abstracts of each identified publication were screened, and only publications that reported the clinical outcomes of this analysis were further retrieved.

Inclusion and exclusion criteria
Inclusion criteria for all eligible studies must have to compare the outcomes of AC to RC reconstruction after PD or PPPD and it reported the postoperative outcome of incidence of DGE or at least one of the measured outcomes that referred as follows: when 2 studies containing overlapping (more than 50%) patients were reported from the same institution or authors, either the one of higher quality or the most recent publication was included in the analysis unless an older publication had more measured outcomes or an RCT publication; and studies with limited information, cases reports, and results that were not published in English language were excluded from this analysis.

Measured outcomes and definitions
The primary outcome measure was the postoperative incidence of DGE, according to the International Study Group of Pancreatic Surgery (ISGPS) consensus definition, [20] grade A was defined as nasogastric intubation (NGT) lasting more than 3 postoperative of days (POD) or the inability to tolerate a solid diet by POD 7; grade B was defined as NGT lasting for 8 to 14 days, the need for reinsertion of the NGT after 7 days, or the necessity of prolonged gastric drainage and a delayed return to solid food intake; grade C was defined as NGT lasting more than 14 days, the need for NGT reinsertion after 14 days, or the inability to tolerate a solid diet by POD 21 (Table 1).
The following secondary outcomes were used to compare AC with RC route: (1) Operative parameters: operative time (minutes); blood loss (ml), transfusion rate. (2) Postoperative outcomes: days until to liquid diet (days); days until to solid diet (days); lengths of hospital stay (days).
A pancreatic fistula was defined as drainage of fluid with an amylase concentration 3 times the upper limit of normal serum as per the ISGPF definition. [20] 2.4. Data extraction and quality assessment Two authors (QJQ and LM) examined the studies independently and extracted data according to a predefined criteria. If data were not presented in the articles, the corresponding authors were contacted by email to specifically ask for the missing information. If all required numbers were obtained, the study was included. Any discrepancy between the 2 reviewers was assessed and resolved by panel consensus. The methodological quality of each RCT was evaluated by individual components based on the Cochrane risk of bias tool. [19] For RSCSs, the quality of these studies was assessed by using the Modified Newcastle-Ottawa Score, [21] which allocates a maximum of 9 points each to patient selection, the comparability of the 2 groups (AC and RC), and outcome assessment.

Statistical analysis
This meta-analysis was performed in line with recommendations from the Cochrane Collaboration and the Quality of Reporting of Meta-analyses guidelines. [22,23] The statistical software Review Manager version 5.0 (The Cochrane Collaboration, Oxford, United Kingdom) was used to perform all statistical analyses. For the analysis, continuous data with weighted mean differences (WMDs) and corresponding standard deviations were presented as weighted WMDs with 95% confidence intervals (CIs). However, odds ratios (ORs) with 95% CIs as the summary statistics were used to perform statistical analysis for dichotomous variables. Clinical heterogeneity was tested by means of the I 2 value; a value exceeding 50% was considered to represent a significant difference. A random-effects model was used to report the results of heterogeneous data; otherwise, a fixed-effects model was used. Funnel plots were constructed to detect and assess publication bias and any associations between treatment estimates and sample size. Forest plots were constructed, and the value of P < .05 was considered to indicate statistical significance.

Study selection
The PRISMA flow chart of literature search strategies is illustrated in Figure 1. Initially, a total of 456 potential articles published until 2017 were identified from literature searches after screening all titles and abstracts; no other eligible studies were found from other sources. Overall, 31 articles were included for a full-text evaluation. Of these, 7 were excluded for not meeting the inclusion criteria. In addition, 4 authors were contacted for additional information or to correct inaccurate information, and 2 provided data that were incomplete or not extractable from the original report. Subsequently, 22 studies were potentially included in this study. Of these, 5 studies, in which mixed groups of surgeries were reported and from which data could not be extracted separately and 2 meta-analysis studies were excluded. Finally, in total, 15 [14,16,[24][25][26][27][28][29][30][31][32][33][34][35][36] studies met our inclusion criteria and were retrieved for more detailed evaluation.

Study characteristics and methodological quality assessment
The baseline characteristics of the 15 included studies and the study design are summarized in  Table 3. The quality assessment of RSCSs [24][25][26][27][28]30,35,36] is also presented in Table 2.

Primary outcome measure
Basic demographics and treatment characteristics did not differ among the study populations. The overall observed incidence of DGE in the 2270 analyzed patients was 27.2% (Table 4). There was a significant difference in the overall incidence of DGE between the AC and the RC group (OR = 0.29; 95% CI, 0.16-0.52; P < .0001; Fig. 2). However, no significant differences were seen in any grade A (P = .14), grade B (P = .41), and grade C (P = .78) DGE between AC and RC reconstruction by subgroup analysis, respectively.

Sensitivity analysis
The sensitivity analysis included the following: (1) 7 RCTs; (2) 8 RSCSs; (3) 13 studies of high quality (7 RCTs and 6 RSCSs with quality score of 6 or more using the modified Newcastle-Ottawa scale); (4) AC versus RC route of GJ after PD procedure; (5) AC versus RC route of DJ after PPPD procedure; (6) 9 studies with more than 30 patients in each group; and (7) 6 studies with DGE was defined as ISGPS consensus definition.
The results from sensitivity analysis were summarized in Table 5. The forest plots of DGE by sensitivity analysis were illustrated in Figure S1-7, http://links.lww.com/MD/D196.

RCTs
When only RCTs [14,16,29,31,32,33,34] were analyzed, DGE rate was still significantly fewer in the AC group by 0.71% (95% CI, 0.50-1.01; P = .05). Hospital stay and days until to liquid diet became not significantly between the AC and RC group. The rest of the outcomes were consistent with the overall analysis, including operative time (P = .84), blood loss (P = .29), overall complications (P = .30), incidence of pancreatic fistula (P = .76), bile leakage (P = .97), and mortality (P = .32). There was generally a reduction in the degree of heterogeneity.

High-quality studies (scores of >6 and RCTs)
The outcomes showed similar results as the original analysis. The heterogeneity was once again reduced.

AC versus RC route of GJ after PD procedure
There were 6 studies [16,[30][31][32][33]36] comparing AC with RC route of GJ after PD procedure. Hospital stay became no significant (MD = 1.32 days; 95% CI, À 2.7 to 0.06 days; P = .06), whereas operative time became significantly shorter in the AC group than RC group (MD = 32.92 minutes; 95% CI, À 47.93 to À 17.92 minutes; P < .00001). The remaining results were similar to the original analysis, and heterogeneity was once again reduced.

AC versus RC route of DJ after PPPD procedure
There were 9 studies [14,[24][25][26][27][28][29]34,35] comparing AC with RC route of DJ after PPPD procedure. Operative time and estimated blood loss became significantly lower in the AC group than RC group (P = .01 and P = .02), respectively. The remaining results were similar to the original analysis, and heterogeneity was once again reduced.

Publication bias
The funnel plot of this study based on postoperative incidence of DGE is shown in Figure 8. All studies except 2 lay inside the limits of the 95% CIs and distributed more evenly about the vertical, showing no evidence of publication bias.

Research frontiers and summary of the study
Since the first report of PPPD was published in 1978, [37] there has a boom in the number of PPPD being performed during the past few years. Several studies have shown that PPPD had a higher incidence of DGE versus the standard PD. [38][39][40] At the same time, other studies have revealed no difference in the incidence of DGE after PPPD or PD. [41][42][43] Regardless of the type of PD, DGE is still a frustrating complication following PD or PPPD and is seen in a significant proportion of patients leading to prolonged hospital stay, decreased patient comfort, increased morbidity, and medical costs. [42,43] Many interventions have been tried in an attempt to reduce the high incidence of DGE. One of the most commonly advocated techniques is to perform an AC route of GJ/ DJ instead of a RC reconstruction. [16][17][18] Murakami et al [28] showed that AC reconstruction was better than RC reconstruction in terms of DGE. Another 2 randomized studies also found a significant benefit in favor of AC reconstruction following PPPD. [14,29] However, the influence of the chosen route of reconstruction is still controversial. Therefore, the aim of this study was to compare the incidence of DGE and other parameters between AC and RC route of reconstruction after PD or PPPD using meta-analytical techniques. AC reconstruction was associated with a lower incidence of DGE, shorter hospital stay, and faster recovery to regular diet compared to that in RC group. No statistically significant differences were found between the 2 groups regarding operative time, incidence of pancreatic and bile leakage, and postoperative mortality. These results are similar to majority of the previous published studies. [14,17,18,28,29] Due to the included number of RCTs was too small to make confirm conclusion, we performed several sensitivity analyses including all the RCTs and the high-quality nonrandomized trials and so on, which greatly increase the credibility of the results without necessarily compromising the quality of the study. Results from sensitivity analysis of high-quality and large trials are in accordance with original analysis, which suggesting the reliability of our results.

DGE consensus definition and its influencing factors
The reported incidence of DGE after surgical procedure varies between 6% to 81.5% from the available literature database. [9][10][11][12][13]16,28,44] The lack of a uniform definition of DGE is largely responsible for the diversity. Some authors have defined DGE as requiring NGT decompression for more than 7 or 10 days postoperatively. [45][46][47] Others have defined DGE as the inability to tolerate a normal diet after the POD 10 or POD 14, [48,49] or a liquid diet after the 7th postoperative day. [14,50] It seems that to establish an internationally accepted uniform definition of DGE is imminent, therefore, a consensus definitions for DGE and other major pancreas-specific complications was proposed by the ISGPS in 2007, [51] which defined the DGE was based on whether there is a need for nasogastric tube placement (and if so, for how long), how soon after the operation the patient is able to tolerate solid oral intake, and whether prokinetic therapy is administered. [19,52] Welsch et al [13] evaluated the ISGPS definition of DGE after PD in a high-volume center and revealed that the ISGPS DGE definition is feasible and applicable in patients with an uneventful postoperative course. Although the ISGPS grading system for DGE is clearly measurable, it does not explain why DGE has occurred. Several studies have focused on its possible cause and pharmacologic strategies to reduce the high incidence of DGE have been published so far. Sato et al [53] and other researches [54][55][56] revealed that DGE have a correlation of intra-abdominal complications www.md-journal.com such as anastomotic leak or abscess. Gastric dysrhythmias, disruption of gastroduodenal neural connections, ischemia of the pylorus muscle, and ligation of the right gastric artery were all related to DGE, respectively. [57][58][59] In addition, preoperative diabetes mellitus condition, cholangitis, previous abdominal surgery as well as retromesenteric route of jejunal reconstruction were significantly associated with a higher incidence of DGE. [25] Park et al [24] found that the incidence of DGE was 31.7% in the  RC group, but only 6.5% in the AC group. Similay results were found by Sugiyama et al, [25] which reported that DGE occurred in 8% patients (8%) in the AC group, but in 72% patients in the RC group. Hartel et al [26] reported an incidence of 5% with the AC route and 24% with the RC route. Meta-analysis of these studies has suggested that the incidence of DGE is lower with the AC route than with the RC route. In the future, to research the exact mechanism of DGE will become more and more urgent and important.

Limitations and recommendations
This systematic review and meta-analysis still have several limitations that must be taken into account. First, the majority of our included studies were comparative studies, although there is evidence that estimates derived from high-quality nonrandomized comparative studies may be similar to those derived from randomized studies. Second, the strength and limitations of metaanalytic techniques have been a source of considerable debate. Third, despite a relatively large number of patients were included   in present study, but the sample size was still too small to allow firm conclusions. In future, it is necessary to conduct randomized trials using standardized unbiased methods and well-matched controls.

Conclusion
This is the very first and largest meta-analysis including RCTs and RSCSs comparing the incidence of DGE and other perioperative outcomes between AC and RC route of GJ after PD or DJ after PPPD. In this meta-analysis, AC route of GJ or DJ reconstruction shows a significantly lower incidence of DGE compared to the RC technique. Further prospective controlled studies are needed for a more comprehensive study between the 2 procedures in future. AC = antecolic, CI = confidence interval, DGE = delayed gastric emptying, HG = heterogeneity between studies, HQ = high quality, OR = odds ratio, PD = pancreaticoduodenectomy, PPPD = pylorus-preserving PD, RC = retrocolic, RCTs = randomized controlled trails, RSCSs = retrospective comparative studies, WMD = weighted mean difference.