Comparison of Delta-Shape Anastomosis and Extracorporeal Billroth I Anastomosis after Laparoscopic Distal Gastrectomy for Gastric Cancer: A Systematic Review with Meta-Analysis of Short-Term Outcomes

Objective The aim of this systematic review and meta-analysis is to evaluate the safety and relative benefits of delta-shape anastomosis (DA) by comparing to conventional laparoscopy-assisted distal gastrectomy with Billroth I gastroduodenostomy (LADG BI). Methods Studies and relevant literature regarding DA versus LADG BI were searched in the electronic databases. Operation time, postoperative complications, estimated blood loss, number of retrieved lymph nodes, time to first flatus, time to oral intake, length of postoperative hospitalization in DA and LADG BI were pooled and compared using meta-analysis. Weighted mean differences (WMDs) and odds ratios (ORs) were calculated with 95% confidence intervals (CIs) to evaluate the effect of DA. Results Eight studies of 1739 patients were included in the meta-analysis. Compared with LADG BI, DA had shorter postoperative hospitalization (WMD = -0.47, 95%CI: -0.69 to -0.25, P<0.01), less blood loss (WMD = - 25.90, 95%CI: -43.11 to -8.70, P<0.01), shorter time to oral intake (WMD = -0.25, 95%CI: -0.49 to -0.01, P = 0.04), and more retrieved lymph nodes (WMD = 1.36, 95%CI: 0.30 to 2.43, P = 0.01). Operation time (WMD = -0.07, 95%CI -15.58 to 15.43, P = 0.99), overall postoperative complication rate (OR = 1.05, 95%CI: 0.74 to 1.49, P = 0.63), surgical complication rate (OR = 1.02, 95%CI: 0.70 to 1.49, P = 0.90), nonsurgical complication rate (OR = 1.21, 95%CI: 0.54 to 2.72, P = 0.64), leakage rate (OR = 2.54, 95%CI: 0.92 to 7.01, P = 0.07), stricture rate (OR = 0.36, 95%CI: 0.09 to 1.44, P = 0.15), wound complication rate (OR = 0.71, 95%CI: 0.33 to 1.55, P = 0.39), time to first flatus (WMD = -0.10, 95%CI: -0.27 to 0.07, P = 0.26), and proximal surgical margin (WMD = -0.25, 95%CI: -1.14 to 0.65, P = 0.59) was not statistically different. Conclusion Compared with LADG BI, DA is a safe and feasible procedure, with significantly reduced blood loss, time to oral intake, and postoperative hospitalization.

our standardized questionnaires; and 4) in the event that duplication of data was observed, more recent studies or those with larger sample sizes were preferentially considered. A manual search of the references of all retrieved articles was also carried out to identify publications for possible inclusion.

Exclusion criteria
The following studies or data were excluded: 1) Abstracts, letters, editorials, expert opinions, reviews without original data and case reports; 2) the outcomes and parameters of patients were not clearly reported; 3) it was impossible to extract the appropriate data of the primary outcomes from the published results; and 4) there was an overlap between authors or centers in the published literature.

Data extraction and synthesis
Two reviewers independently undertook literature searches, screened abstracts, and assessed articles met eligibility criteria. The quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp). Studies achieving six or more scores were considered to be of high quality and were included in the meta-analysis. Reviewers extracted the following parameters from each study: (1) author and study period; (2) study population characteristics, study design; (3) number of patients operated on with each technique; and (4) preoperative data, intraoperative data and postoperative data. Discrepancies between the two reviewers were resolved by discussion with all coauthors and consensus was reached.

Outcomes of interest
The following outcomes were used to compare the two operative techniques: (1) primary outcomes, which referred to operation time, postoperative stay, overall postoperative complication rate, surgical complication rate, nonsurgical complication rate, anastomotic leakage rate, anastomotic stricture rate, wound complications rate; (2) secondary outcomes, which included estimated blood loss (EBL), time to first flatus, time to oral intake earlier, number of retrieved lymph nodes and proximal surgical margins. The surgical and nonsurgical complication was defined as Jung et al. [15].

Statistical analysis
The meta-analysis was performed using the Review Manager software, version 5.3, provided by the Cochrane Collaboration (http://tech.cochrane.org/revman/download). We analyzed dichotomous variables using the estimation of odds ratios (ORs) with 95% confidence interval (95% CI) and continuous variables using weighted mean differences (WMDs) with a 95% CI. According to the Higgins I 2 statistic, heterogeneities <25%, 25% to 50%, and >50% were defined as low, moderate, and high, respectively [16]. A fixed-effects model was used for studies with low or moderate statistical heterogeneity [17]. Otherwise, a random-effects model was used for studies with high statistical heterogeneity [18]. A subgroup analysis of studies more than 50 cases in both DA and LADG BI was conducted. An estimate of potential publication bias was carried out using the funnel plot. P <0.05 was considered statistically significant.

Selected studies characteristics
After selection, eight studies were eventually included in our research [13,[19][20][21][22][23][24][25] (Fig 1). In total, 755 patients treated with DA and 984 patients treated with LADG BI were included in the analyses ( Table 1). All the patients were from Eastern Asia, including Korea, Japan and China. The majority of patients underwent DA were with early gastric cancer (520 out of 755). All of the eight studies were retrospective, nonrandomized. All eight studies were considered to be of adequate quality for the meta-analysis according to NOS assessment (score >5 points) ( Table 2).

Primary outcomes
As showed in Fig 2, there was no significant difference between DA and LADG BI regarding the operation time (WMD = -0.07, 95%CI -15.58 to 15.43, P = 0.99). Patients underwent DA had a shorter postoperative stay than those underwent LADG BI (WMD = -0.47, 95%CI: -0.69 to -0.25, P<0.01) (Fig 3). Sixty-seven out of 755 patients in the DA group and 86 out of 984 patients in LADG BI group had postoperative morbidities (Table 3). There was no significant difference between two groups (OR = 1.05, 95%CI: 0.74 to 1.49, P = 0.63) (Fig 4A). The pooled effect showed no significant difference between DA and LADG BI in surgical complication

Nutritional status and postgastrectomy symptoms
Four pooled studies reported the postoperative nutritional status and postgastrectomy symptoms [13,19,22,25], as showed in Table 5. Lee et al. reported DA group had more food intake   [13,25]. No difference was revealed between the DA group and LADG BI group regarding nausea, dyspepsia, diarrhea and dumping syndrome.

Subgroup analysis
Subgroup analysis of the pooled studies was performed to evaluate whether the pooled primary outcomes and secondary outcomes altered in different case volume subgroup. Subgroup analysis regarding the surgical outcomes had similar results as above (Table 6).

Publication bias
A funnel plot was constructed for the overall postoperative complications and showed symmetry, suggesting that publication bias was acceptable and was unlikely to drive conclusions (Fig 6).

Discussion
Totally laparoscopic gastrectomy is increasingly used since its debut in 1992 [26]. Several meta-analyses hinted totally laparoscopic distal gastrectomy (TLDG), pooling miscellaneous intracorporeal anastomosis including DA, BII and R-Y, has equivalent feasibility and safety as LADG [27][28][29]. More importantly, TLDG is seemingly less invasive and more suitable for the obese than LADG. Potential heterogeneity arises from different intracorporeal anastomosis in these studies and obscures the veritable efficacy of each method. To obtain a more reliable comparison on the feasible and safety of DA, we conducted this systematic review and metaanalysis of published articles that directly compared the DA with LADG BI.
There is no consensus on the indications of DA yet. The decision of DA usually resides with surgeons' or patients' preferences [13,20,23]. DA was mostly applied in early gastric cancer (EGC) located in the low third stomach, which combined indications of LG and conventional BI anastomosis [23,25]. Indications were also expanded to locally advanced gastric cancers (AGC) along with the accumulating surgeons' experiences [20,21,24]. Remarkably, radical treatment of AGC usually requires longer surgical margin than EGC, which increases the anastomosis tension and results in related complications. Lin et al. reported patients with AGC underwent DA had a higher risk of anastomotic leakage [24]. Identifying the lesion precisely during the operation and achievement of enough surgical margin is crucial in deciding the methods of anastomosis. Distinguishing from LADG BI, DA is deficient with tactility in detecting the lesion of EGC merely by the laparoscopic grasper. Preoperative or intraoperative endoscopy is quite practical to mark the lesion as recommended by quite a few of surgeons [19,20].
Prolonged operation time is a common concern of laparoscopic surgery. Our meta-analysis revealed DA wasn't inferior to LADG BI in this outcome. By using the liner stapler, DA has much saved the anastomosis time as well as the total operation time. In experienced hands, the operation time of LG with DA can be shorter than open gastrectomy and laparoscopy-assisted gastrectomy [13,20,23,24,30]. But this doesn't mean the DA is easy. Conversely, coordination between the surgeon and assistants is especially technically demanding. During joining the posterior walls of duodenum and stomach together, appropriate traction by the assistant is of great importance. Given the majority of studies included in our meta-analysis were conducted by surgeons well-trained in LG, or even in DA, outcomes may be discrepant in the non-specially trained surgeons. Kanaya et al. and Jeong et al. revealed the learning curve of DA was quite steep for laparoscopic surgeons [23,30]. After around 10~15 cases, time of completing DA would reach a plateau and maintain in 10~25 min.
Comparing with LADG BI, the potential benefits of DA are better cosmesis, less blood loss and less postoperative pain. Usually, the surgical incision in LADG BI is longer than DA. By hiding the largest surgical incision around the umbilicus, patients undergo DA appears to be scarless. Avoiding the mini-laparotomy for gastrointestinal anastomosis, DA is totally intracorporeal and more tense-free, which reduces the risk of injury around the anastomosis. These characteristics of DA contribute to less intraoperative blood loss, less postoperative pain and more comfortable experience presenting as lower pain score and less administration of    [20,21]. Kinoshita et al. reported the acute inflammatory response following operation was milder in DA with a placid elevation of C-reactive protein [19]. In the present study, data from pooled studies were insufficient to evaluate the postoperative pain and acute inflammatory response. The evaluation of these parameters after DA will be sufficiently assessed by prospective RCTs only. Acceptable postoperative complication of DA was observed in the present study, which was equivalent to LADG BI and even lower than open distal gastrectomy as historical reports [31,32]. Anastomosis-related complication such as leakage and stricture can be disastrous. There was a higher leakage rate in DA (DA vs. LADG BI, 10/755 vs.5/984), but it did not achieve a significant statistical difference. Kim et al. proposed mobilization of the duodenum until the gastroduodenal artery exposure to make sufficient duodenal stump, which eased the tension of anastomosis and might reduce the risk of anastomotic leakage [20]. DA contained several cutting edges, which may cause poor blood supply, yield more weak points around the anastomosis and result in leakage. Huang et al. modified the DA by resection the intersection of the duodenal cutting edge and the common closed edge at the same time to lessen the anastomotic weak point [33,34]. In contrast to leakage, DA group has less anastomotic stricture (2/ 755 vs.6/984). In LADG BI the size of anastomosis ring depends on the diameter of the circular stapler, while a large anastomosis ring can be easily achieved with the 45-mm linear stapler in DA. However, we should also admit the fact that the sample size was still too small to reveal the real benefits and drawbacks of DA because of the low risk of anastomotic leakage and stricture as reported.
Milder surgical trauma and comparable postoperative morbidity brought substantial clinical superiority. Patients in the DA group were able to resume oral intake earlier and have a shorter length of hospitalization, which could counteract the higher expense of DA technique itself. Enhanced recovery program was also applied in some included studies, which emphasized earlier oral intake and might contribute to these advantages of DA in some degrees. Apart from this, LADG BI had more severe complications demanded reoperation or other interventions, which postponed the discharge.
An unexpected result of our study was that the DA retrieved more lymph nodes than LADG BI. It didn't mean the DA was superior to LADG BI in lymphadenectomy because this procedure is supposed to be identical for both approaches. We deemed surgeons' experience played an important role as the majority of surgeons converted to DA after they had matured LADG BI [13,19]. Both DA and LADG BI achieved proximal resection margins more than 3cm that was believed to improve oncological outcomes [35]. Pooled analysis also demonstrated these two approaches achieved similar proximal surgical margins and overall survival rate. Large anastomotic ring and straightforward passageway from the esophagus to the duodenum allow fast gastric emptying and permit more food intakes. More food intake and better nutritional status was observed in DA group during long-term follow-up [22]. Clinically significant postgastrectomy symptoms will deteriorate postoperative QoL. Without the barrier of the pylorus, large anastomosis ring of DA has the concern of reflux gastritis, which makes patients uncomfortable and increases in risk of gastric remnant cancer [36][37][38]. Against with this hypothesis, several studies reported the rate of reflux gastritis after DA was acceptable and comparable with conventional LADG BI and R-Y. Both Kanaya et al. and Lee et al. reported DA had a prevalence of bile reflux but was not a clinical problem [22,30]. Other symptoms are also concerned. Twisting duodenum after DA appears to cause gastric stasis and gastroesophageal reflux. On the other hand, twisting duodenum happened to ward off dumping syndrome in some degree. According to the follow-up result, DA had a similar risk of these symptoms as LADG BI as showed in Table 4.
This meta-analysis had several limitations which should be taken into consideration in interpreting the conclusions of this study. First, this meta-analysis pooled eight retrospective studies which may bias the interpretation of their results. Though well-designed randomized clinical trials are suitable for meta-analysis, RCTs on this issue are rarely conducted because of ethical concern or practical difficulty. Nevertheless, all eight pooled studies were of high quality according to the NOS. Abraham et al. have found that meta-analysis of the well-designed nonrandomized clinical trial of surgical procedures was probably as accurate as that of RCTs [39]. Second, case volumes of the included studies varied greatly, which may lead to heterogeneity among studies. In such a case, comparisons of surgical results would be influenced by surgeons' experience. Third, DA and LADG BI were performed in different periods, from 2004-2014. Due to the development of laparoscopic instruments, perioperative management protocol and surgeons' surgical techniques, the clinical outcomes varied and may result in biases. Fourth, long-term functional outcomes in terms of nutritional status and postgastrectomy symptoms weren't directly compared in the present meta-analysis because four pooled studies report using inconsistent assessment scales with different follow-ups. Furthermore, our study was based on studies conducted in East Asia which should extrapolate these data to the Western population prudently, in where the patient population and disease biology may differ [40,41].

Conclusion
Our study suggests that the DA is a safe and feasible procedure as compared with LADG BI. DA broadens the options of gastrointestinal reconstruction following laparoscopic distal gastrectomy and might bring more minimally invasive benefits and better postoperative nutritional status. However, well designed large-scaled studies which balance the baseline of each arm are needed for further confirming the real benefits of DA.