Introduction

Global obesity rates have increased significantly in recent decades, with adverse health impacts such as diabetes mellitus (DM), obstructive sleep apnea, stroke, coronary heart disease (CHD), hypertension (HTN), gastroesophageal reflux disease (GERD), and cancer highlighting this obesity epidemic as a central public health concern [1, 2]. Bariatric surgery is the most effective and lasting treatment for obesity and its associated comorbidities [3]. Over the past few decades, laparoscopic adjustable gastric banding (LAGB) has become one of the most common surgical procedures for morbid obesity because it is less invasive than other options and not a permanent operation [4, 5]. However, LAGB is no longer so commonly performed because enthusiasm for LAGB has been tempered by the frequent occurrence of late-onset complications and poor long-term weight loss. However, a considerable number of patients have gastric bands which may require revision surgery [6]. In addition, the percentage of people requiring reoperation due to failed weight loss and complications has been reported to be in the range of 30–60% [6,7,8,9,10]. Current strategies for failed LAGB consist of removal of the gastric band with or without revision. However, a number of published studies have reported that the removal of gastric bands is not effective since weight is later regained [11,12,13]. Therefore, many surgeons have chosen to convert a failed LAGB to an alternative revision surgery such as Roux-en-Y gastric bypass (RYGB) or sleeve gastrectomy (SG) [14, 15]. Recently, SG has instead gained acceptance worldwide as it promotes effective weight loss with a sustained resolution of comorbidities when performed as either primary or revision surgery [16, 17].

The safety and efficacy of SG and RYGB as revision surgery have been reported by a number of studies, but the most suitable revision surgery for patients with LAGB failure remains controversial. This meta-analysis was conducted to compare the efficacy and safety of SG with RYGB for failed LAGB.

Materials and Methods

The principles of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2009 Guidelines [18] were used as the standard in this review.

Literature Search

Two reviewers searched the following keywords for relevant articles: “failed adjustable gastric banding,” “revision surgery,” “sleeve gastrectomy,” “Roux-En-Y gastric bypass,” “operative time,” “hospital stay,” and “complications” using the PubMed, Scopus, the Cochrane Central Register of Controlled Trails (CENTRAL), and Embase databases. The literature search publication date ranged from the inception of each database until January 11, 2019. In addition, the references of all relevant articles were searched to identify other eligible studies. This search was performed from January 5, 2019, to January 12, 2019. Two reviewers conducted the complete search process independently. For any disagreement, a third reviewer would examine the study until consensus was reached.

Inclusion Criteria

Two reviewers selected relevant articles independently in accordance with the following criteria: (i) articles were published in English; (ii) articles included patients who required conversion of failed LAGB to SG or RYGB; (iii) more than 10 patients were included in the two groups; (iv) the reported results comprised one or more of the following parameters: percent excess weight loss (%EWL), BMI, complications, remission (effectiveness of revision surgery on medical comorbidities), reintervention, readmission, hospital stay, and operative time; (v) the articles included available data; and (vi) the full text was available. Disagreements were resolved by discussion with a third reviewer until a consensus was reached.

Data Extraction

Two reviewers extracted the following information independently from each eligible article: the last name of the author, publication year, country or region, revision procedure, patient demographics, complications, remission, reintervention and readmission, hospital stay, operative time, and %EWL.

Statistical Analysis

Dichotomous variables were analyzed and presented as odds ratio (OR) with a 95% confidence interval (95% CI). In addition, weighted mean difference (WMD) was calculated with a 95% CI to assess the difference in continuous variables. If OR > 1 or WMD > 0, it means that values in the RYGB group were higher than those of SG patients. For dichotomous variables, a value of 0.5 was added to each cell if any of the cells were 0 in a fourfold table. Heterogeneity was assessed using Q test and I2 test value. A value of I2 < 50% or the p (heterogeneity) > 0.05 indicated that little statistical heterogeneity was presented in the results and thus the data were analyzed using a fixed-effects model (Mantel-Haenszel method) [19]. Conversely, heterogeneous data were analyzed using the random-effects model (DerSimonian and Laird method) [20]. Results were considered statistically significant when the p value was less than 0.05. All statistical analyses were conducted in Stata 14.0 (StataCorp, College Station, TX, USA).

Quality and Publication Bias Assessment

The methodological quality of the articles included in the review was appraised in accordance with the Newcastle-Ottawa Quality scale (NOS) [21]. In accordance with the NOS guidelines, two reviewers independently evaluated the quality of each article according to the following parameters: selection, comparability, and exposure. The results are shown in Table 1. In addition, studies which achieved 5–9 stars were defined as high-quality articles [22]. Publication bias was assessed using Egger’s and Begg’s tests and presented using a funnel plot [23]. The “leave one-out” approach was used to perform a sensitivity analysis, if required.

Table 1 Characteristics of the studies finally included in the systematic review and meta-analysis
Full size table

Results

Literature Search and Characteristics

A total of 584 articles were retrieved from the 4 databases of which 155 duplicate articles were excluded using the document manager. Finally, 37 full-text articles were selected for further evaluation. Based on the established inclusion criteria, 16 articles were included in this meta-analysis. The search process described above and the PRISMA checklist are shown separately in Fig. 1 and Table S1. This meta-analysis included 5131 patients, 2141 of whom underwent revision surgery using SG and 2990 with RYGB, extracted from the 16 articles for the assessment of safety and efficacy of SG versus RYGB after failed LAGB (Table 1). The selected articles described patients from the USA [24,25,26,27,28], UK [29, 30], Spain [31], Turkey [32], Greece [33], Taiwan [34, 35], Poland [36], Italy [37], Singapore [38], and France [39]. The sample sizes for the RYGB and SG procedures ranged from 9 to 1354 patients, with SG performed in 41.7% of cases. The proportion of female patients was 83.2% and 83.0% in the SG and RYGB groups, respectively. The mean age of SG patients ranged from 35.6 to 49.8 years and from 33.9 to 50.7 years for RYGB patients. Mean BMI at the time of conversion surgery in the two groups was 41.10 ± 6.24 (SG) and 41.22 ± 35.34 kg/m2 (RYGB).

Fig. 1
figure 1

Flow chart for searching articles

Full size image

Quantitative Synthesis

Complications

Comparisons of complications between the SG and RYGB groups were presented in 8 articles, with 7 that analyzed early complications (within 30 days) and 5 articles that included late complications (over 30 days). However, no statistical difference was found regarding early complications (OR = 1.53 [95% CI 0.54–4.30] p = 0.421), late complications (OR = 2.17 [95% CI 0.94–5.02] p = 0.070), and total complications (OR = 2.01 [95% CI 0.99–4.12] p = 0.055) between the RYGB and SG groups (Table 2, Fig. 2). In addition, 5 studies reported stenosis data, showing no significant difference between the RYGB and SG groups (OR = 0.50 [95% CI 0.15–1.64] p = 0.249). Eight studies included data on leaks, also finding no significant difference between the RYGB and SG groups when analyzed together (OR = 1.60 [95% CI 0.95–2.69] p = 0.076). Furthermore, no significant difference between the two groups was reported regarding bleeding (OR = 3.64 [95% CI 1.89–7.00] p = 0.000), ulcer (OR = 1.94 [95% CI 0.20–19.07] p = 0.570), or incisional hernia (OR = 1.16 [95% CI 0.29–4.62] p = 0.831). However, 5 studies that reported obstruction indicating that RYGB as a revision surgery for failed LAGB increased the incidence of obstruction (OR = 4.16 [95% CI 1.001–17.28] p = 0.050) (Table 2, Fig. 3).

Table 2 Summary relative risks for the association between SG and RYGB for failed LAGB by study characteristics
Full size table
Fig. 2
figure 2

The forest plot showed the OR (95% CI) of early, late, and total complications between RYGB and SG groups after failed LAGB

Full size image
Fig. 3
figure 3

The forest plot showed the OR (95% CI) of special complications between RYGB and SG groups after failed LAGB

Full size image

Reintervention and Readmission Within 30 Days

Reintervention was analyzed in 5 articles, demonstrating that the number of patients who underwent reintervention in the RYGB and SG groups was 74 and 23, respectively. Readmission was analyzed in 4 articles, with the number of readmissions in the RYGB and SG groups being 63 and 124 patients, respectively. The incidence of reintervention (OR = 3.13 [95% CI 1.95–5.02] p = 0.000) and readmission (OR = 2.02 [95% CI 1.47–2.76] p = 0.000) was significantly higher in RYGB group (Table 2, Fig. 4).

Fig. 4
figure 4

The forest plot showed the OR (95% CI) of reintervention and readmission between RYGB and SG groups after failed LAGB

Full size image

Complication Remission Rate Following Revision Surgery

Only 2 studies presented data concerning the effectiveness of revision surgery on medical comorbidities [28, 29]. There were 68 GERD patients before revision surgery, of which 13 (19.12%) received SG and 55 (80.88%) RYGB. The remission rates in the SG and RYGB groups were 23.08% and 45.45%, respectively. The two studies reported that 41.18% (OR = 1.47 [95% CI 0.14–14.98] I2 = 58.3%, p = 0.375) of the patients experienced remission from GERD. In addition, 33 patients that had DM preoperatively underwent revision, of which 6 (18.18%) received SG and 27 (81.82%) received RYGB. The remission rates in the SG and RYGB groups were 33.33% and 40.7%, respectively. It was reported that 39.39% (OR = 1.18 [95%CI 0.24–5.88] I2 = 0.0%, p = 0.844) of the patients experienced remission from DM. Furthermore, 69 patients exhibited HTN before revision surgery of which 8 (11.59%) underwent SG and 61 received (88.41%) RYGB. The remission rates in the SG and RYGB patients were 25% and 59.02%, respectively. The two studies reported that 55.07% (OR = 2.09 [95% CI 0.48–9.09] p = 0.324) of the patients achieved remission from HTN. Overall, RYG B had a better trend than SG in terms of remission in preoperative GERD, DM, and HTN, although no significant statistical difference was found regarding effectiveness of revision surgery on GERD (OR = 1.47 [95%CI 0.14–14.98] p = 0.375), HTN (OR = 2.09 [95%CI 0.48–9.09] p = 0.324), or DM (OR = 1.18 [95%CI 0.24–5.88] p = 0.844). The results of the remission rate following revision surgery are shown in Table 2 and Fig. 5.

Fig. 5
figure 5

The forest plot showed the OR (95% CI) of remission rate of GERD, DM, and HTN between RYGB and SG groups after failed LAGB

Full size image

%EWL

The %EWL at 3, 6, 12, 24, and 36 months was reported in 13 of the articles included in the review. There was no significant difference after 3 months (WMD = − 0.89 [95% CI − 11.56–9.79] p = 0.871) and 6 months (WMD = − 1.80 [95% CI − 12.31–8.71] p = 0.737). However, %EWL after 12 months was significantly higher in the RYGB group (WMD = 7.00 [95% CI 2.48–11.52] p = 0.002). In addition, %EWL was also significantly higher after 24 months (WMD = 12.37 [95% CI 6.20–18.54] p = 0.000) in the RYGB group compared with SG patients (Table 2, Fig. 6). There was no statistical difference in %EWL after 36 months (WMD = 3.67 [95% CI = − 4.35–11.69] p = 0.370) (Fig. 6).

Fig. 6
figure 6

The forest plot showed the WMD (95% CI) of %EWL at 3, 6, 12, 24, and 36 months between RYGB and SG groups after failed LAGB

Full size image

Mean Length of Hospital Stay and Operative Time

The mean length of hospital stay in the RYGB group ranged from 1.2 to 5 days and from 1.5 to 5.7 days in the SG group. There was no significant difference in length of hospital stay (WMD = 0.25 [95% CI − 0.32–0.82] p = 0.387) (Table 2, Figure S1). The mean operative time ranged from 140 to 218.9 min in the RYGB group and from 108.4 to 172.7 min in the SG group. This difference was statistically significant, indicating that RYGB requires more operative time than SG (WMD = 38.81 [95% CI 30.46–47.16] p = 0.000) (Table 2, Figure S1).

Assessment of Publication Bias and Quality

Publication bias was assessed using an Egger’s formal statistical test and a Begg’s correlation test then presented using a funnel plot when the number of articles analyzed was not less than 10. Therefore, we conducted an assessment of publication bias for %EWL after 12 months. The results (Egger P > |t| = 0.721 Begg, Pr > |z| = 0.837) showed that there was no publication bias (Figures S2, S3, and S4). In addition, the quality assessment of the articles ranged from 6 to 7 stars (Table 1), thus classifying them as high-quality articles.

Discussion

LAGB has gained worldwide acceptance for the treatment of obesity, thanks to the simplicity, good short-term results, and low early complication rate of the procedure [40]. However, several studies have reported that many patients eventually require revision surgery due to high failure rates, weight regain, and band-related complications in long-term follow-up [41,42,43]. In addition, RYGB and SG have both been proven effective in treating patients with failed LAGB [44, 45]. Moreover, different surgeons have varying opinions about which revision is preferable. Those in favor of SG were of the opinion that LSG had gained worldwide recognition as a result of sustained resolution of comorbidities and the significant reduction in weight as either primary or revision surgery [46, 47]. Conversely, those in favor of RYGB have argued that RYGB for failed vertical banded gastroplasty (VBG) has demonstrated to be the operation of choice for satisfactory weight loss in revisions, thus indicating that RYGB may have greater efficacy than SG as a revision surgery for failed LAGB [26].

Several similar reviews were published to discuss the clinical outcomes of revision surgery after failed LAGB, but their conclusions have not been consistent [45, 48, 49]. In this meta-analysis, the results demonstrated that the efficacy of RYGB for failed LAGB was better than of SG after12 months (WMD = 7.00 [95% CI 2.48–11.52] p = 0.002) and 24 months (WMD = 12.37 [95% CI 6.20–18.54] p = 0.000). However, there was no difference in their effect on weight loss after 36 months (WMD = 3.67 [95% CI = − 4.35–11.69] p = 0.370). This suggests that SG may have the additional benefit of regulation of hormones as revision procedure during the longer-term follow-up, an effect which has been previously reported in a clinical article by Cohen et al. [50]. In addition, the clinical results about %EWL after 24 months by Magouliotis et al. supported our final conclusion. In contrast to our results, Magouliotis, Sharples, and Zhou et al. conducted meta-analyses which reported no statistical difference in %EWL after 12 months. Sharples et al. suggested that there was no statistical difference in %EWL at 24 months [45, 48, 49]. The insufficient number of articles and short follow-up time may have caused their results to be unreliable. Moreover, a published review by Elnahas et al. suggested that RYGB was better than SG at promoting weight loss. However, no standardized outcome measurement and significant heterogeneity between studies made its results uncertain [51]. The remission rate of complications following revision surgery was slightly better in RYGB about preoperative GERD, DM, and HTN than in SG group, although the differences were not significantly different. When choosing revision surgery for patients with medical comorbidities, the actual situation of patients should be taken into consideration.

Based on our results, the incidence of complications, reintervention, and readmission was significantly higher in RYGB patients. Similarly, a study conducted by Spaniolas also suggested that RYGB was associated with a higher rate of uneventful recovery compared with SG [52]. There was no statistically significant difference in the mean length of hospital stay, but the mean operative time in the RYGB group was significantly higher. However, there are many factors affecting the length of hospital stay and operative time. Hospital stay is a poor metric of surgical outcomes. In addition, operative time may be increased if gastric band explantation and conversion to another procedure take place synchronously compared with two-step revision surgery, so surgeons should treat such results in the light of the actual situation. RYGB as revision surgery alters the structure of the intestine and its gastrointestinal anastomosis, possibly accounting for a higher complication rate. In addition, SG is essentially the first step of the BPD-DS procedure, the reason that SG requires less operative time.

Our meta-analysis included a higher number of patients than previous clinical studies and reviews. Nevertheless, the principal limitations of this meta-analysis were the small number of patients that contributes to the outcomes in each category and the methodological issues associated with research design, as all articles included in the review were observational studies. Therefore, limitations arising from the lack of randomized controlled studies are inevitable.

Conclusions

This meta-analysis indicated that RYGB was associated with better weight loss at 12 and 24 months; however, this appeared to be at the expense of increased post-operative complications and readmissions. There was a lack of detail regarding the type and severity of complications and the need for readmissions as well as a paucity of data on long-term outcomes. Thus, there is a pressing need for prospective controlled studies comparing surgical techniques to treat failed gastric banding.