Evolution of GERD symptoms after bariatric surgery: A dose-response meta-analysis

Importance Obesity is associated with increased prevalence of gastroesophageal reux disease (GERD); with recent reports suggesting improvement in GERD symptoms and weight loss following bariatric surgical intervention. However, the exact impact of the type of bariatric surgery on the evolution of GERD symptoms have remained unexamined. Objective To characterize the exact evolution of GERD symptoms, post bariatric surgery. Data sources We systematically searched electronic databases (PubMed, EMBASE, Web of Science, and the Cochrane Library from inception to December 2018) for eligible studies that satisfy pre-specied inclusion criteria. Study selection We included clinical trials of all designs (prospective and retrospective) that reported on GERD outcomes following Laparoscopic Sleeve Gastrectomy (LSG) or laparoscopic Roux-en-Y gastric bypass (LRYGB). Data extraction and synthesis Two independent reviewers extracted relevant data based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline. Data were pooled using a random-effects model. Results A total of 39 studies were analyzed and a robust-error meta-regression model was used to conduct a Dose-Response Meta-Analysis (DRMA) synthesizing data on 39 studies that reported GERD outcomes after bariatric surgery. Out of 43,994 patients who underwent either LSG (N = 9,547 patients) or LRYGB (N = 34,447 patients), 32.4% experienced improvement in symptoms (95% CI 20.62 to 45.45); The DRMA demonstrated a window period of two years for sustained improvement after which symptoms began to recur in those that were asymptomatic. Conclusion and relevance Bariatric surgery may improve GERD symptoms in obese patients who underwent LSG, however, the most favorable effect is likely to be found after Roux-en-Y gastric bypass surgery. The effects were not sustained and returned to baseline within 4 years.


Introduction
The burden of obesity is on the rise across different continents and populations around the world. 1 Although different socio-demographic groups are involved, the burden appears more pronounced amongst adolescent and adult populations. 1 Uncontrolled obesity has been associated with preventable morbidities across a wide range of cardiovascular and metabolic risks; these include cardiovascular disease, hypertension, diabetes mellitus, venous thrombo-embolic (VTE) disease, obstructive sleep apnea, and cancer amongst others. 2 For most patients, bariatric surgery remains the only option for the treatment of obesity when dietary interventions and pharmacotherapy fail. Amongst a range of bariatric surgical options currently in use includes [ Laparoscopic Sleeve Gastrectomy (LSG), and laparoscopic Roux-en-Y gastric bypass (LRYGB)], both of which have been shown to result in varying degrees of weight loss. 3 However, there has been a rising incidence in complications associated with these procedures. Gastroesophageal re ux disease (GERD) in particular has been the subject of recent concerns. 4 The prevalence of clinically relevant GERD associated with bariatric surgery is variable, but has been reported as ranging between 7% and 14%. [5][6][7] Uncertainty however remains regarding the exact evolution of symptoms especially among patients with different clinical phenotypes such as metabolic syndrome, obesity and multi-morbidities. In this Dose-Response Meta-Analysis (DRMA), we therefore intend to assess GERD symptoms over time with the view to clarify the timeframe of evolution of GERD symptoms among patients that have undergone these procedures.

Database Search
Literature search strategies was developed using medical subject headings (Mesh) and text key words related to bariatric surgery and GERD.
The speci c search strategies were created by a health sciences librarian with expertise in systematic reviews. Following the recommendations of the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement, 8 a systematic literature search was performed using PubMed, Medline, EMBASE, Web of Science, and the Cochrane Library from inception to December 2018, combining the key words obesity, high BMI, weight loss and gastroesophageal re ux with bariatric surgery OR LSG OR laparoscopic sleeve gastrectomy OR sleeve gastrectomy OR SG OR LRYGB OR laparoscopic Roux-en-Y gastric bypass OR gastric bypass OR GB. Moreover, we individually observed the reference lists of the selected articles to nd other potentially relevant studies. GERD was used as the observation index of outcome after all types of bariatric surgery (i.e., improvement of GERD after bariatric surgery or/and the number of cases of newly onset or worsened GERD after bariatric procedure). The means and measures of dispersion were approximated from the gures given in the reports.

Selection of studies (Inclusion and exclusion criteria)
Only clinical trials were included for this meta-analysis. The inclusion criteria were as follows: 1) studies reporting on the e cacy of LSG and/or LRYGB on GERD symptomatology; 2) new onset or worsening GERD after LSG and/or LRYGB. The exclusion criteria includes 1) studies with incomplete data for GERD symptoms following bariatric surgery ; 2) case series or case reports, 3) studies with follow-up periods of less than 6 months after bariatric surgery 4) Studies reporting in other languages other than English. Only the most relevant and comprehensive publications were included in the analysis to avoid duplicates, ambiguity; studies involving non-human subjects, or papers reporting data from the same study populations were excluded.

Data Extraction
A team of two investigators extracted the data using a standardized form, which was reviewed individually by a third investigator following analysis individually and in duplicate. To ensure standardization across the reviewers, a calibration exercise was carried out before starting the actual review. Any disagreement between reviewers was resolved by the consensus of two authors and the third reviewer. Data from each included study was extracted by two investigators and reviewed independently by a third investigator following essential indicators from each article into Table 1. These indicators included study author/year, study design, level of evidence, sample size, follow up period, type of bariatric surgery (LSG or LRYGB), GERD improvement rate, method of GERD evaluation, and study country.

Quality Assessment
The methodological quality of clinical trials was assessed by Cochrane Collaboration's tool for assessing risk bias which covers: sequence generation, allocation concealment, blinding, incomplete outcome data (e.g. drop-outs and withdrawals) and selective outcome reporting. 9 The observational studies were assessed using the Newcastle-Ottawa Quality Assessment Scale (NOS) checklist. 10

Statistical analysis
The effect size of interest was the difference in prevalence of GERD symptoms in patients immediately before and at any time point after bariatric surgery reported in the papers. This was modeled as a risk difference in a DRMA. The latter was conducted as a one-step procedure with time (as the "dose") along with change in risk as the outcome were t into an inverse variance weighted non-linear robust error metaregression model, 11 using restricted cubic splines with three knots in an effort to approximate the potential non-linear relationship. The weights were based on the inverse of the variance of the incremental risk difference in proportions of patients with GERD symptoms from the baseline at surgery. Stata MP 15 (StataCorp, College Station, TX) was used for the analysis, utilizing the remr package. 12 Con dence level was set at 95%.

Study Features
A total of 32 studies including 43,994 patients (9547 in LSG and in 34447 LRYGB) were enrolled for the dose-response meta-analysis from a cluster of 2500 studies pooled from all databases. Figure 1 shows the ow chart of studies selection with details of retrieval process and ltering. The 32 selected studies were including nine prospective studies, and 23 retrospective observational studies    Table 1.

Effect of Bariatric Surgery on GERD
A total of 32 studies (84.6%) showed improvement or remission for obesity-related GERD (78.6% in LSG, 100% in LRYGB). The proportion of subjects with GERD symptoms at baseline was 23.1% (IQR 12.0 -34%) and at follow-up was 8.8% (IQR 2.8 -13.6%). While this seems to suggest a decrease in GERD symptoms after bariatric surgery, the follow-up varied considerably with a median of 24 months (IQR 12 -60 months). When the incremental risk was modeled in a DRMA with time as the "dose" there was a decrease in prevalence of GERD symptoms till 2 years post-surgery and this gradually increased back to baseline at 4 years post-LSG surgery ( Figure 2) and post-LRYGB surgery ( Figure   3). The trend was a continuing increase in risk but the trend after 4 years was not reliable due to a paucity of data points. There was a much greater decrease after LRYGB compared to LSG, but this was driven by a single large study and after this was excluded, the trends were similar (Figure 4). In addition, there were fewer data-points for LRYGB.

Discussion
This meta-analysis represents the rst comprehensive attempt at exploring the comparative prevalence of GERD symptoms following either LSG or LRYGB surgery. We found a signi cant decrease in GERD related symptoms up to two years following surgery. This symptomatic improvement was regardless of the type of surgery. Beyond two years however, uncertainty remains as to the impact of surgery on GERD related symptom relief. The rate of symptomatic improvement was steeper and relapse of symptoms quicker with LRYGB surgery compared with LSG procedures or be it with an uncertain point estimates. This has signi cant implication for patient counseling before surgery as well commissioning of surgical procedures. The exact mechanism underpinning the improvement in GERD symptoms following surgery continues to generate intense mechanistic debate. Several reports have attributed this to signi cant and durable weight reduction evident in patients who underwent these procedures. 45,46 We found an unusually high prevalence of GERD symptoms at baseline before surgery. Previous reports have estimated this at 33%. 47 The difference in estimate from our review and that reported from previous studies may have to do with marked heterogeneity in the modalities employed to evaluate GERD. The change or improvement in GERD symptoms (ΔGERD) from 23% at baseline to 8.8% over a median period of two years is comparable or better than that seen with anti-secretory therapy. 48 The lack of difference in point estimates of residual GERD outcomes between SLG and LRYGB mirrors previous uncertainty regarding the exact impact of the type of bariatric surgery on GERD symptom improvement. The ndings from the largest systematic review by Stenard et al exploring this uncertainty in patients who underwent LSG were discordant. 49 About half of the studies in that review found symptomatic improvement in GERD symptoms following the procedure, with the remaining half showing signs of worsening GERD symptomatology. The apparent discordance in GERD outcomes in this review was attributable to heterogeneity in the mode of evaluation of GERD ranging from clinical evaluation, 24-hour ambulatory pH studies, esophageal manometry or contrast studies, to endoscopy. Additionally, only one study was prospective, the remaining studies were retrospective. Other subsequent reviews 50,51 including that by Himpens et al 52 also reported mixed outcomes regarding GERD symptoms improvement following LSG. Our review similarly found lack of superiority between different surgical modalities in GERD symptoms improvement following surgery. The confounding issues highlighted earlier including marked heterogeneity in the mode of evaluation of GERD may have accounted for our point estimates with regards to effect of bariatric surgical modality on GERD symptoms improvement.
Additionally, we found improvement in GERD symptoms plateauing at 4 years with signi cant uncertainty afterwards. Whilst this suggest need for more comprehensive studies to ascertain the exact impact of bariatric surgery in GERD symptoms beyond 4 years; it is probable that rapid weight gain reported from previous series may be the key driver to the is phenomenon.

Strengths and limitation
This DRMA represents the rst comprehensive attempt at exploring the comparative e cacy of LRYGB vs. LSG surgical procedures in achieving sustained and durable improvement in GERD symptoms. It provides the rst estimates of the average duration of GERD symptom improvement (2 years) following these procedures as well as raises the prospects for further studies to explore determinants of uncertainty in symptom relief after four years. As with previous systematic synthesis in this area, our DRMA is limited by differences in the modes of adjudication of GERD symptoms by various investigators. The reliance on clinical evaluation alone in some cases and paucity of data regarding utility of PH monitoring as well as endoscopy data in others may have accounted for the imprecision regarding some point estimates in previous studies (including ours).

Conclusion
In obese bariatric patients we found signi cant improvement in GERD symptoms at two years regardless of the type of bariatric surgery (LSG or LRYGB), but this is not sustained beyond four years. The rate of improvement in bariatric symptoms was faster but less durable with LRYGB compared to gastric sleeve surgery.

Declarations Acknowledgment
We acknowledge the Qatar national library for accepting to cover the cost of publication of this review Funding No infrastructural funding was received for this review.
Author contribution A-NE: Review conceptualization, protocol registration, independent reviewer, risk of bias assessment, data interpretation, and writing of the initial draft/revision of the nal manuscript MID: Independent reviewer, risk of bias assessment, data interpretation, writing of the initial draft/revision of the nal manuscript.     Laparoscopic Roux-en-Y gastric bypass (LRYGB) DRMA results after exclusion of a single large study. The gure depicts difference in prevalence from baseline over time of GERD symptoms. Dashed lines depict the 95% con dence intervals