Promising effects of 33 to 36 Fr. bougie calibration for laparoscopic sleeve gastrectomy: a systematic review and network meta-analysis

The standard size of bougie for laparoscopic sleeve gastrectomy (LSG) is not yet established. Therefore, a systematic review and network meta-analysis were conducted to assess the weight loss effects and associated complications of LSG for patients with morbid obesity, based on different bougie sizes. A total of 15 studies were reviewed in this systemic review and network meta-analysis (2,848 participants), including RCTs and retrospective studies in PubMed, and Embase until September 1, 2020. The effectiveness of different bougie calibration sizes was assessed based on excess weight loss (EWL), total complications, and staple line leak. Within this network meta-analysis, S-sized (≤ 32 Fr.) and M-sized (33–36 Fr.) bougies had similar effects and were associated with the highest EWL improvement among all different bougie sizes (S-sized: standardized mean difference [SMD], 10.52; 95% confidence interval [CI] − 5.59 to  − 26.63, surface under the cumulative ranking curve [SUCRA], 0.78; and M-sized: SMD, 10.16; 95% CI − 3.04–23.37; SUCRA, 0.75). M-sized bougie was associated with the lowest incidence of total complications (M-sized: odds ratio, 0.43; 95% CI, 0.16–1.11; SUCRA, 0.92). Based on our network meta-analysis, using M-sized bougie (33–36 Fr.) is an optimal choice to balance the effectiveness and perioperative safety of LSG in the clinical practice.

www.nature.com/scientificreports/ Not only changes in gastric motility and related hormonal secretion but also dominant mechanisms of gastric restriction, based on the essential intraoperative bougie calibration with longitudinal gastric transection of the fundus, body, and antrum along the lesser curvature, lead to limited eating volume of about 100 mL, subsequent dietary habit modification, and eventually weight reduction 9,10 . Although the surgical standardization and relative contraindications for LSG have been well documented, the standard bougie size used to calibrate the gastric sleeve still remains to be established, based on the weight loss effects and related complications, such as staple line leak (SLL), gastric stenosis, or de novo reflux esophagitis [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26] .
Although several systemic meta-analyzes, based on pooled data from case series and randomized controlled trials (RCTs), have been published, the standard bougie size for LSG is not yet established 16,17,21,[23][24][25][26][27][28] . Herein, we performed a network meta-analysis and cataloged results of these controlled trials into a comprehensive systematic review and meta-analysis of available data to determine the standard bougie size for calibration during LSG, based on excess weight loss (EWL), associated complications, and SLL percentage.

Materials and methods
Current network meta-analysis was performed after establishing guidelines from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Network Meta-Analyses (PRISMA-NMA) 29,30 and Meta-analyzes Of Observational Studies in Epidemiology (Table S1.1 and Table S1.2) 31 . Registered protocol was available in the Open Science Framework (https:// osf. io/ drhsb).

Data sources and search strategy.
A systematic publication review without language restrictions was performed and retrieved from PubMed and Embase from inception until September 1, 2020. Gray literature and manual searches for potentially eligible articles from review articles were reviewed. The US Government Clinical Trials database (www. Clini calTr ials. gov) was searched for ongoing clinical trials. The search terms comprised "laparoscopic sleeve gastrectomy, " "bougie calibration, " and "bougie size, " along with a list of all interventions and possibly relevant keywords. (Table S2).
Inclusion and exclusion criteria. Both RCTs and observational cohort studies were included in the study design. Targets of comparison were patients with morbid obesity (aged at least 18 years) received LSG with bougie calibration to create the neogastric tube. Morbid obesity is defined as body mass index (BMI) of ≥ 40 kg/m 2 or 35 kg/m 2 generally associated with comorbidities, except for lower BMIs among East Asians 20,32 . The comparison included two or more different bougie sizes; studies reporting the relationship between bougie sizes and weight loss were enrolled in this study.
The exclusion criteria included (1) studies that evaluated adolescents or pediatrics participants, (2) those in which interested outcomes were not reported, (3) those not specific to patients with morbid obesity, and (4) single-arm studies without comparators.
Two authors (KH Chen, TW Chang) independently selected trials that met the inclusion criteria, and another author (KH Chen) adjudicated differences. In case of disagreement, the same authors consulted with another one (PC Chang) to achieve decisions after a deliberate group discussion.
Data extraction and bias assessment. Two reviewers (HJ Jhou, TW Chang) independently screened the studies, extracted relevant data, and assessed the risk of bias among included studies using the Cochrane risk of bias tool (Table S3.1 and Table S3.2) 33 . Data extraction was performed using a special designed sheet obtained from reports of a previous meta-analysis 28 . Study information regarding studies, participants, and treatment characteristics was obtained. If available data are lacking, corresponding authors were contacted for data collection.
Outcome definition.
1. Percentage of EWL (% EWL): (weight loss/baseline excess weight) × 100, where weight loss = preoperative weight -the initial weight; baseline excess weight = initial weight − ideal weight (X) where X was calculated using an ideal BMI, and the ideal BMI cutoff point has been used differently in enrolled studies (Table S4). 2. Overall complications: All complications related to LSG, such as de novo gastroesophageal reflux disease (GERD), postoperative bleeding, nonspecific abdominal pain, nausea, dehydration, surgical site infection, portal vein thrombosis, or as defined by the study authors, were obtained (Table S4). 3. SLL: Postoperative neogastric tube leak or as defined by study authors (Table S4). We used the frequentist network meta-analysis (NMA) model to compare effect sizes among studies with the same interventions. All frequentist approach network meta-analyzes were performed using the statistical package Netmeta (Version 1.2-1) 34,35 in R Project 3.6.1 (R Core Team, Vienna, Austria) and Stata version 16 (Stata Corp, College Station, Texas). The symmetry and geometry of the evidence were examined by producing a network plot with nodes for the number of study participants and connection sizes corresponding to the number of studies 36 . For continuous data, summary standardized mean differences (SMDs) with 95% confidence intervals (CIs) were calculated using a random-effects model (21)  www.nature.com/scientificreports/ ratios (ORs) with 95% CIs were estimated with 0.5 zero-cell correction 37 . All comparisons were set as two-tailed, and a p value statistical significant cutoff point was set at 0.05. For generating high resolution figure, we applied Adobe Photoshop CC 2018 and Sketch Version 45.1 software. The rank of treatment within defined groups measured was measured using the surface under the cumulative ranking curve (SUCRA) 38 , which is the relative percentage of probabilities as the best treatment, in a scale from 0 (worst) to 1 (best) 39 . SUCRA can be clinically used to compare treatment effects of all treatments for the target outcomes.
Potential inconsistencies between direct and indirect evidences were compared within the network model. Moreover, global inconsistencies were examined using a design-by-treatment interaction model, whereas local inconsistencies between the included comparators were examined using a node-splitting method 40 . Statistical significance was set at 5% for analyzes. If the inconsistency existed, sensitivity analysis was performed to determine possible reasons.
The assumption of network transitivity was examined by visually inspecting tables with patient's population across included studies, study methodologies, design intervention details, and outcome measurement differences 41 .
Comparison-adjusted funnel plots and Egger's test regression were used to assess possible publication bias or potential small study effects for available interventions 42 .
Ethical statement. There was no human trial in this systematic review and network meta-analysis, and the approval from the ethics committee does not apply.

Consent statement.
Because this study is a systematic review with network meta-analysis, the informed consent does not apply.

Results
Systematic literature review. Figure 1 presents the whole flowchart of the current NMA. After the initial screening procedure, a total of 25 articles were considered for full-text review; 10 of which were excluded for various reasons. These 15 trials were included in our study, and a total of 2848 participants receiving LSG using different bougie sizes calibrated as XL, L, M, and S were included.
Study characteristics are summarized in Table 1. The sample size of enrolled studies ranges from 24 to 1395 patients with morbid obesity. The total numbers of RCTs and observational cohort studies are 687 and 2161 patients respectively. Among these participants, the mean age of enrolled patients was 37.46 years. The mean BMI and body weight of participants were 46.97 kg/m 2 and 121.16 kg, respectively. The bougie calibration sizes ranged from 27 to 60 Fr.
Inconsistency and sensitivity analysis. In the EWL outcome, global inconsistency existed with statistical significance between design inconsistency in design-by-treatment interaction model. Local inconsistency was also observed with statistical significance in the node-splitting model.
Thus, a sensitivity analysis was performed to evaluate possible reasons of inconsistency. As heterogeneity may exist in the follow-up period between studies, Hady et al. 's study was excluded with a 6-month follow-up 23 . The remaining seven included studies had constant follow-up period of at least 1 year and enrolled in our sensitivity analysis. In the sensitivity analysis, S-, M-, and L-sized bougies were associated with significantly better EWL than that in the XL-sized. According to the SUCRA value, the S-sized bougie was associated with the greatest EWL among all of the different sizes of bougie. (S-sized: SMD, 7.76; 95% CI 1.75-13.77; SUCRA, 0.84). Furthermore, the L-and M-sized bougies had similar effect and were ranked as second and third, respectively (L-sized: SMD,  Figure S1). In outcomes of total complication or SLL, no inconsistency was observed between evidence derived from direct and indirect comparisons, including either global inconsistency as assessed using the design-by-treatment interaction model or local inconsistency as assessed using the node-splitting model (Table S6).

Risk of bias, inconsistency, and publication bias.
We found that 57.1% (100/175 items), 27.4% (48/175 items), and 15.4% (27/175 items) of included studies were assessed as low, unclear, and high risk of bias, respectively. Funding sources and concealing procedure after randomization mainly contributed to the high and unclear risk of bias, respectively (Table S3.1 and Table S3.2).
The assumption of network transitivity was established by visually inspecting tables with patient population across included studies (Table 1). Formal assessments of funnel plots across included studies were conducted for all outcomes and revealed general symmetry without publication bias. Results of the Egger's test indicated were not statistically significant, which also suggested no publication bias in the present NMA ( Figure S2.1 to S2.3).

Discussion
To the best of our knowledge, this is the first network meta-analysis that analyzes the effectiveness and safe range of bougie sizes to achieve reduced weight loss and lower complications for LSG. The network meta-analysis presented herein not only shows the relative treatment effect (EWL) and associated complications from all pairwise comparisons but also offered ranking of different bougie sizes 34,36,[39][40][41][42] . We comprehensively reviewed the major database and included only high-quality articles. Based on currently available evidences, our results suggest that S-and M-sized bougies both have the greatest EWL, and the latter was associated with the lowest www.nature.com/scientificreports/ incidence of total complications, including SLL. Within our network meta-analysis, using the M-sized bougie (bougie size between 33 and 36 Fr., including 36 Fr.) for intraoperative calibration is an optimal choice to balance the effectiveness and safety during LSG. LSG itself is considered as a purely restrictive bariatric surgical procedure and also has impacts on gastrointestinal motility, hormonal regulations, and gut microbiota. LSG has been demonstrated to increase the rate of gastric emptying and intestinal transit. Studies also found increased glucagon-like peptide 1 and peptide YY levels and increased endocrine functions for bile acids after LSG 9,43,44 . In regard to clinical effects, LSG can produce efficient weight loss and improve obesity-related comorbidities accordingly, such as type 2 diabetes mellitus, hypertension, dyslipidemia, or obstructive sleep apnea [43][44][45] .
Therefore, an ideal gastric sleeve with the proper size should be created to strike the balance between the acceptable weight loss and occurrence of complications. Intraoperative bougie calibration is an essential part for LSG via different bougie tube sizes to assist bariatric surgeons and to determine the expected gastric tube. Although the clinical significance of an ideal gastric sleeve cannot be overemphasized, the ideal bougie size used in LSG remains to be established. In 2013, a literature review to discuss the ideal bougie size reported the L-sized bougie could decrease the incidence of SLL with the similar EWL effect as the S-sized bougie 46 . Series of RCTs or retrospective studies to compare thinner and bigger size of bougie calibration were conducted in recent decades, and conflict results were presented 12,13,[17][18][19][20][21][22] . In 2018, Wang et al. conducted a meta-analysis that discovered thinner-sized bougie in LSG was more effective in augmenting weight loss, and overall complications were not increased 28 . With new evidences enrolled in this study, our results are in agreement with that of www.nature.com/scientificreports/ previous meta-analyzes, and we further determined the ideal range of bougie size balancing the effect of weight loss and safety in clinical practice. In addition to bougie sizes, some studies advocated that related surgical manipulation of the distance from the pylorus with associated antral resection/preservation might influence LSG outcomes [47][48][49][50] . For the restrictive purpose, antral resection with shorter distance from the pylorus limits more gastric volume to create a smaller  www.nature.com/scientificreports/ gastric tube and is hypothesized to increase intragastric pressure and decrease the distention ability after eating, which results in early satiety theoretically [47][48][49][50] . However, some surgeons suggested antral preservation to prevent possible gastric outlet stenosis and to decrease intragastric pressure in order to reduce the SLL risk [51][52][53][54] . In a 2018 meta-analysis, antral resection is associated with better effect of weight loss and without increased risk of surgical complications as compared with antral preservation 54 . Some adverse events reported to be associated with LSG were bleeding, nausea, wound infection, SLL, or de novo GERD 11,61 . However, the advantages and disadvantages of LSG for GERD remains controversial. In 2019, a systematic literature review found LSG is associated with an increased incidence of de novo GERD. Those with mild GERD might have improved symptoms after LSG; however, patients with morbid obesity, severe reflux and erosive esophagitis may have high possibility of persistent GERD thereafter 55 . Moreover, SLL is a catastrophic complication after LSG, and previous studies have postulated that larger bougie size may decrease the SLL risk 46,56 . Nevertheless, surgeons' personal experience might play a vital role in decreasing and even preventing this undesirable complication 57 . In 2018, Demusy et al. analyzed the nationwide data in the United States, which disclosed that the bougie size was not associated with postoperative leak rate, and the risk of bleeding and reoperation was decreased via concomitant staple line reinforcement intraoperatively 58 . In our network meta-analysis, overall complication rates or SLL did not significantly increase in the group with bougie size of < 32 Fr. (S-sized group), whereas the M-sized group was associated with the lowest incidence of total complications. Herein, we have categorized different bougie sizes into four groups for the following reasons. First, we routinely used a 32-Fr. oral gastric tube during vertical gastric sleeve stapling to format the gastric tube in our institution. Accordingly, we choose 32 Fr. as the first cut point. Moreover, we set 36 Fr. as the second cut point because it was found to be the optimal bougie size to augment EWL in recent studies 18,28,59 . A retrospective multicenter cohort study conducted by Sánchez-Santos et al. in 2016 concluded that a bougie size of > 40 Fr. had a protective effect to minimize the overall complication rate 60,61 . Therefore, 40-Fr. bougie was used as the third cut point to investigate the potential protective effect among the group with larger bougie sizes.
Some modifications have been made to increase the strength of the evidence in our network meta-analysis. First, we strictly followed standardized guidelines based on the PRISMA statement to improve reporting of systematic reviews 29,30 . Second, inconsistency and sensitivity analyzes were performed to evaluate possible reasons of inconsistency, and factors that could increase inconsistency were successfully identified and excluded. Nevertheless, the present network meta-analysis has three limitations. First, to increase patient numbers in our studies, some prospective and retrospective cohort studies, which may decrease the strength of the evidence were enrolled. Second, these four groups of different bougie sizes were categorized artificially, which may lead to some potential bias. Third, the criteria of reported complications in enrolled studies may be different, which may result in inaccurate complication rates. Therefore, results of this network meta-analysis should be cautiously interpreted.

Conclusion
Based on our network meta-analysis and current evidences, S-and M-sized bougies had similar effects of EWL, with the latter being associated with the lowest incidence of total complications, including SLL. Intraoperative calibration with M-sized bougie (33)(34)(35)(36) is an optimal choice to balance the effectiveness and safety for patients with morbid obesity undergoing LSG.