Gastric band is safe and effective at three years in a national study subgroup of non-morbidly obese patients

Aim To analyze the 3-year outcomes of lower body mass index (BMI) (<35 kg/m2) adjustable gastric band (AGB) recipients across multiple sites in the French health insurance system. Methods From prospectively collected data on a cohort of 517 morbidly obese Swedish Adjustable Gastric Band® (SAGB) patients (Clinical Trials Web database, #NCT01183975), a retrospective analysis of a subgroup of 29 low-BMI patients was conducted. Patients had a severe obesity-related comorbidity, had undergone a prior bariatric procedure requiring reintervention, or had a maximum adult BMI≥40. Safety (mortality, adverse events) and effectiveness (BMI change, excess weight loss [EWL, %], total body weight loss [%TBWL], quality of life [QoL], and comorbidities) were evaluated. Results Multiple surgical teams/sites enrolled patients and performed SAGB procedures between September 2, 2007 and April 30, 2008. Of 29 low-BMI patients (mean age, 41.3 ± 10.3 years), 89.7% were female, and obesity duration was 13.6 ± 7.3 years. Mean BMI was 31.5 ± 3.7; there were 37 comorbidities in 15/29 patients. At 3-year follow-up, BMI was 29.4 ± 4.9 (mean change, -2.3 ± 6.2; P = 0.069); total cohort EWL, 7.3 ± 74.8%; TBWL, 6.2 ± 18.8%; BMI≥30 to <35 EWL, 38.8 ± 48.0%; there were 7 comorbidities in 15/29 patients (P < 0.031). There were 20 adverse events in 13 patients (44.8%); SAGBs were retained in 25/29 (86.2%) at 3 years. Conclusions In a retrospective analysis of a subgroup of BMI<35 kg/m2 patients, some following a prior bariatric procedure, SAGB was found to be safe and effective at 3-year follow-up.

For over 2 decades, since publication of the 1991 National Institutes of Health (NIH) consensus conference statement (1), the cutoff point for bariatric surgery has been morbid obesity (body mass index [BMI, kg/m 2 ]≥40 or ≥35 with comorbidities), also termed class II obesity by the World Health Organization (WHO) (2). This demarcation of access to bariatric surgery was based on the observation that an increase in BMI leads to an increase in the risk of comorbid illness and premature death. Yet, investigation of the potential value of bariatric surgery as a safe and effective treatment for overweight (BMI 25-<30) and obesity class I (BMI≥30 to <35) patients has been under way since the publication of the NIH statement. In 1992 and 1995 landmark studies (3,4), Pories et al theorized that bariatric procedures might be safe and as beneficial for weight loss and comorbidity reduction in nonmorbidly obese patients as it was in the morbidly obese (5). In the last half decade, the least-invasive, lowest-risk restrictive procedures, such as adjustable gastric banding (AGB), have been employed at the forefront of exploring surgical options for the <35 BMI patient.
Adjustable gastric banding comprised the vast majority, nearly 90%, of bariatric procedures performed in morbidly obese patients in France prior to 2008 (6). To assess the national social insurance-supported use of the Swedish Adjustable Gastric Band (SAGB) (7,8), the French government commissioned a prospective, 31-center, "real-life, " observation of SAGB safety and effectiveness in class II and III obese patients (9). Between September 2, 2007 andApril 30, 2008, patients were selected and underwent SAGB implantation in rural and urban centers. SAGB weight-loss effectiveness analyzed on an intent-to-treat basis at the 3-year study endpoint was comparable to that of AGB findings summarized by global meta-analyses (10,11). Under the "real-world" SAGB study protocol requirement of consecutive recruitment and surgeon discretion, 29 patients (5.6% of 517) were included in the national SAGB study who presented with a BMI<35 and a severe obesity-related comorbidity, and/or had experienced a prior complicated bariatric surgery requiring revision, and/or had previously sustained a maximum adult BMI≥40. With the aim of contributing safety and effectiveness findings to the growing <35 BMI evidence base, we report 3-year outcomes for the French low-BMI SAGB study group.

Methods study protocol
During 2007, the French Health Technology Assessment Body (HAS) (12,13) requested that Ethicon Endo-Sur-gery (Europe) GmbH sponsored and performed a countrywide health insurance study to assess reimbursement of the SAGB product in France. The study was registered (Clinical Trials Web database, #NCT01183975) (14) and a sponsor-developed protocol and case report form developed to direct implementation of HAS requirements and good clinical practices (GCPs) (ie, patient welfare in study design, ethical study conduct), defined by ISO EN 14155-1 and -2 (15,16). Ethical approval and protocol approval were given by HAS, the Commission Nationale de l'Informatique et des Libertés, and the Comité Consultatif sur le Traitement de l'Information en matičre de Recherche dans le domaine de la Santé. Treatment payments were covered by national health insurance (13).
A contract research organization, Medextens SARL, Paris, France, and an independent monitoring committee consisting of a non-participating bariatric surgeon, a pharmacologist, and a medical nutritionist supervised the study's progress and prepared an interim report for review by HAS and the sponsor. Patients were required to provide written informed consent before surgery per Declaration of Helsinki (17) and GCP guidelines.

design and setting
The prospective, multicenter, noncomparative study design aimed to facilitate observation and reporting of outcomes in a morbidly obese study cohort, of which the current low-BMI cohort was a subgroup (9). Primary HAS objectives were to asses SAGB safety (mortality, adverse event [AE] occurrence) and clinical effectiveness (changes in weight loss, quality of life [QoL], comorbid illness) in varied French hospital settings.
In order to incorporate "real-life" practice experiences across geographically diverse regions of France, surgeons were selected from academic, private, and public institutions with differing bariatric surgery volumes. Per GCP standards, surgeons were required to undergo training in the protocol; selected surgeons recruited SAGB patients consecutively.

Inclusion criteria
Study eligibility for the primary trial was based on a recruitment goal of >500 patients with <20% loss to follow-up after 3 years. General inclusion criteria stipulated patients with morbid obesity after failed medical treatment and no contraindications in accord with French (12), Europe-an (18), and American NIH bariatric surgery guidelines (1). French residents with a BMI<35 were permitted inclusion in the consecutive SAGB study enrollment if they had an adult maximum BMI≥40, and/or had a severe obesity-related comorbidity (thus, were receiving SAGB as a primary intervention in the current study [an "index SAGB"]), or if they required reintervention following a complicated prior bariatric procedure [a "PBP+SAGB"] (SAGB as a secondary intervention).

Variables
Safety variables analyzed for the low-BMI cohort were mortality and frequency of AEs. Effectiveness variables were evaluated as change over 3 years in absolute weight (AW); BMI; excess body weight (EW); percentage EW loss (%EWL), ie, baseline AW -follow-up AW/EW, calculated by Miller's formulas (19)(20)(21)(22)(23) for identifying ideal weight, corresponding to the midpoint value of the medium-frame range on the Metropolitan Life Insurance Height and Weight Tables ×100; and percentage total body weight loss (%TBWL), (ie, baseline AW -follow-up AW/baseline AW × 100) (24). Health-related QoL and changes in comorbid illness were also analyzed.

Quality of life instruments
The generic EuroQoL 5-Dimensions (EQ-5D), a psychometric instrument valued for its utility in calculating quality adjusted life years as well as the relative cost-effectiveness of obesity interventions, such as AGB (25), was used as a measure of QoL. The EQ-5D is a health-related QoL evaluation with 5 items and a visual analogue scale (EQ-VAS) (26-29) that provides a 5-dimensional profile: mobility, usual activities, self-care, anxiety/depression, and pain/discomfort. Dimensions are presented as 1 item with 3 response options: severe problems, some problems, and no problems. Item responses can be weighted normatively to derive a utility score (range -0.594 to 1, where 1 = ultimate health). A clinically important difference has been identified at ≥0.07 on the EQ-5D scale (29). The EQ-VAS module is a single-item global QoL evaluation in which patients rate their current health (scale from 0 = worst imaginable to 100 = best imaginable) (30).

data collection
Protocol-prescribed safety and weight data collection and assessment measures were the only standardized requirements for the surgical centers, per the "real-life" observa-tional study design. Baseline characteristics (eg, gender, age) were collected; weight, obesity-related comorbid disease, and QoL were recorded on the day of surgery, and at 1, 3, 6, 12, 18, 24, and 36 months postoperatively. Comorbidity data were sought via questionnaire; diagnoses were established and recorded consonant with individual investigators' typical practice via the password-protected Medextens-Medalliance eCRF Manager (v.1.3) web database.
For the current low-BMI subgroup study, data were retrieved from the original HAS archive and sorted by script for the known 29 target patients. Coded variables that addressed identified study topics were chosen and manually exported to a dedicated SPSS database.
technique SAGB procedures were performed via pars flaccida technique (31), and band adjustments were accomplished at the discretion of the surgeons. Three SAGB model options were available: 2200-X Quick-Close; 2100-X (with locking ring and injection port); and the BD2XV Quick-Close with Velocity TM injection port.

statistical analysis
Statistical analyses were performed using the SPSS ® software package (ver. 20, IBM SPSS, Chicago, IL, USA). Quantitative demographic variables were generally reported as median and interquartile range (IQR); qualitative variables (demographic and outcome) were reported as number and percentage. Adverse events were also reported as number and percentage. Quantitative measures of change from baseline at 3 years were analyzed using the relatedsamples Wilcoxon signed rank test; between-group comparisons were made with the Mann-Whitney U test. The Fisher exact test was used to investigate relationships between qualitative variables. Multivariate modeling, linear regression, and logistic regression were used to explore relationships between patient characteristics, weight loss, and QoL. Alpha was set at P < 0.05.
Total cohort median changes in weight-related obesity indicators were not significant at 3 years; however, a high level of individual variation in weight-loss outcomes was noted. While some BMI<35 patients lost significant weight, some gained weight as indicated by negative %TBWL and %EWL values. Subgroup analyses, by intervention type and BMI category, were carried out. As detailed in     r = 0.631 (P = 0.001). A logistic regression model using BMI as the lone predictor was shown to correctly classify 83.3% of patients into their respective "weight-loss" vs "weightgain" groups. Logistic results are presented in the form of a probability curve in Figure 4

Comorbidities
Adhering to the study's observational design, no diagnostic tests for comorbidity assessment were required. At each visit, comorbidities were reported as present or absent. Significant variation in diagnostic methodology, terminology, and reporting regularity was noted. Despite this limitation, qualitative analysis indicated a continued reduction in the overall number of comorbidities over time and a gradual increase in those with no reported comorbidities. At baseline, there were 37 comorbidities in 15/29 BMI<35 patients. At 3-year follow-up, comorbidities were significantly reduced to 7 (P = 0.031); median number of comorbidities per patient fell significantly from 1.0 (0.0-2.0) at baseline to 0.0 (0.0-0.0), P = 0.002.

dIsCussIoN
Results suggest that the SAGB was safe and effective in French patients with a baseline BMI<35. There was no mortality and the AE rate was 0.23 AEs per patient-year, approximately similar to the 0.19 AE rate found in the main HAS cohort study. Adverse events were primarily confined to PBP+SAGB patients; whereas, 83.3% of index SAGB patients experienced no AE. SAGB device survival rate was also comparable to that found in the main co-  hort study (86.2% vs 87.0%). QoL was improved and a reduction in overall number of comorbidities was observed. BMI reduction was significantly related to positive changes in patient health status. On balance, weight loss trended toward significance at 3 years; however, some patients demonstrated weight gain. For example, those presenting with a BMI<30 (89.0% PBP+SAGB) experienced a median 9.0-kg AW gain (TBWL = -11.8%). Conversely, patients with BMI≥30 to <35 experienced significant AW loss (10.5 kg), median 33.1% EWL, and median 11.0% TBWL -more than double the 5.0% TBWL threshold associated with significant comorbidity improvement (32). First-intervention BMI≥30 to <35 patients experienced a median EWL of 51.0% (TBWL = 17.4%). In addition, within the BMI<35 cohort, logistic regression modeling suggested that a baseline BMI≥30 was the point above which weight loss was likely to occur 3 years post SAGB surgery.
Although weight-loss findings for the BMI≥30 to <35 firsttime SAGB patients derive from a very small subgroup (N = 10), their median weight-loss outcomes over 3 years were comparable to those of the 517 morbidly obese patients of the original HAS cohort (median BMI change, 6.1 vs 7.9; EWL, 51.0% vs 49.3%). The subgroup outcomes suggest that surgical weight loss in patients in the BMI≥30 to <35 category follows a pattern similar to that in patients with BMI>35. The observation lends support to the idea that lowering the 1991 NIH (1) bariatric surgery cutoff to 30 may be reasonable. In addition, obesity-related health risks, such as type 2 diabetes mellitus and cardiovascular disease, tend to arise at lower BMIs in certain non-Caucasian populations (eg, Asian Indians) due to a higher percentage and central distribution of body fat (33-35). The Asian Indian Consensus Group, for example, has moved to evaluate weight-related health risk with alternatives to the BMI metric in these patients and to lower the BMI cutoff for bariatric surgery to BMI>32.5 with a comorbidity or BMI>37.5 without comorbidities (36). In the current SAGB study, in which median weight loss in the BMI≥30 to <35 subgroup was significantly greater than in the BMI<30 subgroup, neither group lost an excessive amount of weight; in fact, mean AW increased slightly in the BMI<30 group (mostly prior bariatric procedure patients), as is typical for bariatric surgery patients after the point of their greatest weight loss. In 2007, Scopinaro et al found in their study of low-BMI biliopancreatic diversion (BPD) patients that, although the mildly obese group lost nearly twice the weight of the overweight group, weight loss was not excessive in either low-BMI category (45). Other surgical studies, including those using BPD, BPD with duodenal switch, AGB, sleeve gastrectomy, and RYGB, have observed the same phenomenon (46)(47)(48). Weight loss appears to stabilize within the postoperative year at a BMI>25 regardless of whether the procedure falls into the restrictive, malabsorptive/restrictive, or primarily malabsorptive surgical category (49), and regardless of the preoperative BMI. A homeostatic mechanism may exist that facilitates weight loss in proportion to procedure-specific caloric absorption capacity (5). An integrative analysis of the 16 then-existing bariatric surgery studies in low-BMI patients detected the same pattern of lesser weight loss in patients with BMI<30 than in those with a BMI≥30, suggesting a blunting of the weight-loss cascade at around 30 BMI.
Although the current study was limited by a restricted population of 29, the findings represent a small addition to the evidence base for bariatric surgery in the BMI<35 patients. As in results for the primary HAS "real-world" cohort study, the current low-BMI report contains an underreporting bias partially due to data recording by numerous surgical teams across diverse locations in France; calculating a quantitative measure of change in specific comorbidities was, therefore, not possible.
As early as 1997, Mason et al noted the dramatic trend toward increasingly higher weights in bariatric surgery candidates. They hypothesized that escalating obesity and life-threatening comorbidities should be prevented rather than treated in their full expression (50). Current study outcomes and those of a growing evidence base appear to support the value of lowering the BMI access point for bariatric surgery to permit earlier intervention in appropriate patients. Similar to findings in morbidly obese SAGB patients at 3 years, SAGB treatment for low-BMI patients in France, particularly those with BMI≥30 to <35, was found safe and effective.