Bariatric/Metabolic Surgery toTreat Type 2 Diabetes in PatientsWith a BMI,35 kg/m2

Diabetes Care 2016;39:924–933 | DOI: 10.2337/dc16-0350

OBJECTIVE

Global usage of bariatric surgery has been dictated for the past quarter century byNational Institutes of Health recommendations restricting these operations toindividuals with a BMI ‡35 kg/m2. Strong evidence now demonstrates that bar-iatric proceduresmarkedly improve or cause remission of type 2 diabetes mellitus(T2DM), in part through weight-independent mechanisms, and that baseline BMIdoes not predict surgical benefits on glycemic or cardiovascular outcomes. Thisimpels consideration of such operations as “metabolic surgery,” which is usedexpressly to treat T2DM, including among patients with a BMI <35 kg/m2 whoconstitute the majority of people with diabetes worldwide. Here, we review avail-able evidence to inform that consideration.

RESULTS

Ameta-analysis of the11published randomized clinical trials (RCTs) directly comparingbariatric/metabolic surgery versus a variety of medical/lifestyle interventions forT2DMprovides level 1Aevidence that surgery is superior for T2DMremission, glycemiccontrol, and HbA1c lowering. Importantly, this is equally true for patients whosebaseline BMI is below or above 35 kg/m2. Similar conclusions derive from meta-analyses of high-quality nonrandomized prospective comparisons. Meta-analysis ofall pertinent published studies indicates that T2DMremission rates followingbariatric/metabolic surgery are comparable above and below the 35 kg/m2 BMI threshold. Thesafety, antidiabetes durability, and benefits on other cardiovascular risk factors frombariatric/metabolic surgery appear roughly comparable among patients with a BMIbelowor above35 kg/m2. Further studies areneeded toextend long-termfindings andmeasure “hard” macrovascular/microvascular outcomes and mortality in RCTs.

CONCLUSIONS

Extant data, including level 1A evidence from numerous RCTs, support new guide-lines from the 2ndDiabetes Surgery Summit that advocate for the consideration ofbariatric/metabolic surgery as one option, along with lifestyle and medical ther-apy, to treat T2DM among patients with a BMI <35 kg/m2.

For the past quarter century, worldwide usage of bariatric surgery has largely beengoverned by a 1991 set of recommendations from the National Institutes of Health(NIH) that limit these operations to severely obese individuals (BMI $40 kg/m2) orto patients with a BMI$35 kg/m2 and serious obesity-related comorbidities, such astype 2 diabetes mellitus (T2DM) (1).In the time since those NIH recommendations were written, a large new evidence

base has been generated demonstrating powerful effects of most bariatric

1VA Puget Sound Health Care System and Diabe-tes and Obesity Center of Excellence, Universityof Washington, Seattle, WA2Center for Diabetes and Obesity, Oswaldo CruzHospital, São Paulo, Brazil

Corresponding author: David E. Cummings,davidec@u.washington.edu.

Received 17 February 2016 and accepted 22March 2016.

© 2016 by the American Diabetes Association.Readersmayuse this article as longas thework isproperly cited, the use is educational and not forprofit, and the work is not altered.

See accompanying articles, pp. 857,861, 878, 884, 893, 902, 912, 934, 941,949, and 954.

David E. Cummings1 and Ricardo V. Cohen2

924 Diabetes Care Volume 39, June 2016

METABOLICSU

RGER

Y

operations on T2DM (2,3). It has alsobecome very clear within the past de-cade that the antidiabetes impact ofsome bariatric procedures results fromnot only secondary consequences of re-duced food intake and body weight butalso additional weight-independentmechanisms (4–6).These findings have led to a paradigm

shift of thought in the field, propellingan increasingly popular view that someoperations should be viewed not just as“bariatric surgery” but also “metabolicsurgery” (7,8). A natural consequence ofthis change in mind-set is to considerthe use of bariatric/metabolic surgeryto treat T2DM in less obese or evenmerelyoverweight patients, with BMI levels belowexisting NIH cutoffs.Here, we discuss the conceptual logic

for contemplating the use of bariatric/metabolic surgery to treat T2DM in pa-tients with a BMI,35 kg/m2, along withavailable evidence pertinent to thatconsideration. Elsewhere in this issueof the Diabetes Care, new guidelinesfrom the 2nd Diabetes Surgery Summit(DSS-II) are published to inform theproper place for bariatric/metabolic sur-gery in the overall T2DM treatment al-gorithm (9). These guidelines, which areintended to replace the conspicuouslyoutdated 1991 NIH recommendations(1), advocate for the consideration ofsurgery as one option, along with life-style and medical approaches, to treatT2DM in patients with a BMI as low as30 kg/m2, or as low as 27.5 kg/m2 forAsian populations. This article evaluatesthe evidence supporting these new clin-ical practice guidelines.

RATIONALE FOR CONSIDERINGMETABOLIC SURGERY FOR T2DMIN LOWER-BMI PATIENTS

Several lines of evidence and logic justifycontemplating the use of bariatric/metabolic operations in lower-BMIpatients who have T2DM that is not ad-equately controlled with behavioral/pharmaceutical interventions.First, the impact of bariatric/metabolic

surgery on T2DM, especially from oper-ations involving intestinal bypasses, isvery impressive. Although diabetes istraditionally considered a progressive,relentless disease in which mitigationof end-organ complications is the pri-mary therapeutic goal, a large majorityof patients with T2DM who undergo

bariatric/metabolic surgery experienceremission of this disease and thereaftermanifest nondiabetic glycemia off all di-abetes medications (2,3,10). For exam-ple, the T2DM remission rate afterRoux-en-Y gastric bypass (RYGB) is typi-cally 70–80%, and it is even higher forbiliopancreatic diversion (BPD). Suchpercentages vary depending on theHbA1c threshold used to define “remis-sion,” but by any definition, these oper-ations yield T2DM remission in mostcases. Across many studies, the best pre-operative predictors of failure to remitdiabetes are long duration of disease,use of insulin, high glycemia, and verylow C-peptide levels. As these probablyall reflect advanced diabetes with irre-versible b-cell destruction, the implica-tion is that surgery should not merely beconsidered as a salvage option to beused after failing many years of othertherapies. Although many people withT2DM who initially experience post-operative diabetes remission ultimatelydevelop recurrence, the median disease-free period among such individuals afterRYGB is 8.3 years, for example (11). Mostof this evidence derives from studies ofpeople with a BMI$35 kg/m2, but thereis no a priori reason to predict that theantidiabetes effects of surgery woulddisappear among patients below thatBMI level, which was defined relativelyarbitrarily in 1991 as a cutoff for bariatricsurgery (1).

Second, although high BMI has tradi-tionally been used as the primary crite-rion to select patients for bariatricsurgery, no data demonstrate that base-line BMI predicts the success of suchoperations on metabolic, cardiovas-cular, or other “hard” clinical outcomes(even though higher baseline BMI doespredict greater weight loss). Instead,strong evidence indicates that preoper-ative BMI, at least within the obeserange, does not predict the benefits ofsurgery on diabetes prevention (12,13),remission (11,14–19), and recurrenceafter initial remission (15) or the magni-tude of its effects on heart attacks,strokes (20,21), cancer (22), or death(11,12,17,18,20,22,23). In contrast,high levels of baseline fasting insulinand/or glucose (presumably reflectinginsulin resistance) do predict the be-nefits of surgery on most of these endpoints. This strongly suggests thatthe advantages of bariatric/metabolic

surgery on key clinical outcomes resultmore from improved glucose homeo-stasis than from weight loss per se(12,14,20,22–26). Thesedata also indicatethat high fasting insulin and glucose lev-els, or some other measure of insulin re-sistance, would be better evidence-basedcriteria for surgical selection than BMI is.

Third, use of metabolic surgery to treatT2DM in lower-BMI patients makes con-ceptual sense if it improves diabetes atleast in part through weight-independenteffects, and considerable evidence nowdemonstrates suchmechanisms (3,5). Re-garding RYGB, for example, the followingfive bodies of evidence attest to weight-independent antidiabetes mechanismsengaged by this operation, in addition tothe well-known glycemic benefits ofweight loss (4).

1. Diabetes remission frequently occursvery fast, long before substantialweight loss has occurred. At leastsome of this might result from peri-operative acute caloric restriction,which iswell known to improve insulinsensitivity and glycemia, although it isnot clear why it is observedmore afterbariatric/metabolic surgery than othergastrointestinal operations.

2. Glucose homeostasis improves moreafter a given amount of RYGB-inducedweight loss than with equivalent weightreduction achieved by diet/exercise orlaparoscopic adjustable gastric banding(LAGB).

3. There is an inconsistent correlationbetween the amount of weight lostafter RYGB and the degree of diabe-tes improvement.

4. Novel experimental operations thatreplicate some of the intestinal anat-omy and physiology of RYGB withoutcompromising the stomach can exertpowerful antidiabetes effects withlittle or no weight loss.

5. Rarecasesofextremehyperinsulinemic/hypoglycemia that occasionally de-velopmany years after RYGB (typicallyduring partial weight regain) suggestthe possible existence after surgery ofchronic b-cell–stimulatory effects un-related to weight change.

Potential mechanisms mediating di-rect antidiabetes effects ofmetabolic sur-gery include enhanced secretion of lowerintestinal hormones such as glucagon-likepeptide 1, altered physiology due to

care.diabetesjournals.org Cummings and Cohen 925

excluding ingested nutrients from theupper intestine, upregulation of oneor more putative “anti-incretins” or“decretins,” compromised ghrelin se-cretion, modulations of intestinalnutrient-sensing pathways that regu-late insulin sensitivity, changes in bileacid signaling, perturbations of gutmicrobiota, alterations of intestinalglucose transport and metabolism, at-tenuation of intestinal sodium–glucosecotransport, and other changes not yetfully characterized (5). Many of the ob-servations that identify these candidatemechanisms derive primarily from ani-mal experiments and need to be veri-fied in humans, but it is an active areaof research.Last, the 1991 NIH recommendations

that restrict bariatric surgery to peoplewith a BMI $35 kg/m2 were based al-most exclusively on data from Caucasianpatients, but all other large racial groupstend to develop T2DM at lower BMI lev-els than those in this population (27).Hence, the NIH standards deny accessto metabolic surgery for the large ma-jority of patients with diabetes world-wide who might benefit from this optionto treat their disease. For example, inTaiwan, the median BMI of patientswith T2DM is approximately 24 kg/m2,and ,2% have a BMI $35 kg/m2 (28).Thus, NIH recommendations exclude.98% of these East Asian patients fromconsidering metabolic surgery to treatT2DM. Similar comments pertain toSouth Asians. Even in the U.S., the peakof the BMI distribution curve for patientswith T2DM lies between 30 and 35 kg/m2

(29), so a very substantial proportion ofAmerican patients with diabetes have aBMI too low to qualify for surgery byexisting standards. In short, the 1991NIH recommendations exclude hun-dreds of millions of patients with diabe-tes from access to a highly effectiveT2DM treatment option.

EVIDENCE REGARDING THE USEOF BARIATRIC/METABOLICSURGERY TO TREAT T2DM INPATIENTS WITH A BASELINEBMI <35 KG/M2

Effects of Bariatric/Metabolic Surgeryon Diabetes in Patients With aBMI <35 kg/m2

Several excellent, recent systematic re-views and meta-analyses help summa-rize and interpret findings from the

large, growing number of publicationsreporting data on bariatric/metabolicsurgery for people with a preoperativeBMI ,35 kg/m2.

Before discussing these, it is impor-tant to note that evidence in this fieldis muddied by the fact that there is nouniversally agreed-upon standard formeasuring the success of bariatric/metabolic surgery to treat T2DM, eventhough standard definitions for diabe-tes “remission” have been publishedby prominent authorities (30). Variousinvestigators define remission differ-ently, typically as an HbA1c level belowsome threshold, off diabetes medica-tions. However, remission rates differgreatly even within the same study de-pending on whether the required HbA1cthreshold is 6.0%, 6.5%, or 7.0%. In ad-dition, many physicians commonly leavepatients onmetformin even after normo-glycemia is achieved, using it for preven-tion of relapse, hoped-for cardiovascularbenefits independent of glycemia, poly-cystic ovarian syndrome treatment, andso forth. This practice confounds any def-inition of diabetes remission that requirespatients to be off all diabetes medica-tions, and there is no widely acceptedstandard in bariatric/metabolic researchfor how to deal with this issue.

Muller-Stich et al. (31) recentlypublished a high-quality systematic re-view and pooled meta-analysis of onlylevel 1 and level 2 evidence from studiesdirectly comparing surgical versusmedical/lifestyle interventions for T2DM amongpatients, at least some of whom in eachstudy had a baseline BMI ,35 kg/m2.This included seven randomized clinicaltrials (RCTs) and six high-quality prospec-tive observational comparisons, encom-passing 818 participants with diabetes,with follow-up of 1–3 years. No deathswere reported.

Every one of these studies found thatvarious surgical interventions werestatistically significantly superior to avariety of nonsurgical interventions incausing either diabetes remission (i.e.,nondiabetic HbA1c levels off all diabetesmedications) (Fig. 1) and/or glycemiccontrol (i.e., nondiabetic HbA1c with orwithout diabetes medications) (31). Theoverall odds ratio (OR) for surgical supe-riority in diabetes remission was 14.1among all studies and 22 among thosethat exclusively examined patientswith a preoperative BMI ,35 kg/m2.

These results persisted with fixed- versusrandom-effects models, in subgroupanalyses of only RCTs or only prospectiveobservational comparisons, and with orwithout adjustment for potential publi-cation biases. The overall average per-cent HbA1c dropped by 1.5 points moreafter surgical compared with nonsurgicalinterventions, even though patients inthe former group used far fewer diabe-tes medications compared with the lat-ter at the end of these studies. The ORsfor surgical superiority over medical/lifestyle interventions regarding diabe-tes remission were similar for each in-dividual operation in this meta-analysiscompared with a prior meta-analysis ofRCTs examining surgical versus nonsur-gical T2DM approaches among patientswith a baseline BMI $35 kg/m2 (32).For example, the OR for surgical superi-ority in diabetes remission after LAGBwas 12 versus 5 in the former versuslatter analysis, respectively, and ap-proximately 30–50 in both analyses forRYGB, vertical sleeve gastrectomy (VSG),and BPD.

Not surprisingly, Muller-Stich et al.(31) found that BMI fell much morewith surgery than medical/lifestyle in-terventions in every case except one.The exception was an investigation ofan experimental operation that repli-cates the proximal intestinal bypass ofRYGB without affecting the stomach,i.e., duodenal-jejunal bypass surgery. Itcaused substantially greater glycemiccontrol than did nonsurgical care,despite equal weight change in bothgroups, further demonstrating weight-independent antidiabetes effects ofproximal intestinal bypass (33). As iscommonly observed, the effects of sur-gery on blood pressure and plasma lip-ids were less impressive than those onglycemia. Nevertheless, the surgicalgroups overall were four times lesslikely to have hypertension and fivetimes less likely to have dyslipidemiacompared with medical groups at theend of these studies (31).

A smaller systematic review andmeta-analysis recently published byRao et al. (34) examined the effects ofRYGB on T2DM among studies whoseparticipants exclusively had a baselineBMI ,35 kg/m2. This encompassednine publications, describing a total of343 participants (baseline BMI range19–35 kg/m2, follow-up 1–7 years.)

926 Bariatric/Metabolic Surgery in Lower-BMI Patients Diabetes Care Volume 39, June 2016

Again, there were no deaths, and surgicalcomplication rates were 6–20%, which issimilar to published rates for patientswith a baseline BMI $35 kg/m2 (9). Allnine articles reported significant HbA1creductions after surgery, with an averagepercent HbA1c lowering of 2.8 points(34). Overall, surgery reduced fastingblood glucose by 60 mg/dL more thandid the various nonsurgical comparatorinterventions. Rates of diabetes remis-sion (defined here as HbA1c ,6.5% offall diabetes medications) ranged from65 to 93%, which is at least as high as isreported historically among patientswith a baseline BMI$35 kg/m2 (2,3,10).

Comparing the Effects of Surgery inPatients With a Preoperative BMIBelow Versus Above 35 kg/m2

In considering whether to lower the BMIthreshold for contemplating the use ofbariatric/metabolic surgery to treat in-adequately controlled T2DM in lessobese patientsdas recommended bythe new DSS-II guidelines publishedthis issue of the Diabetes Care (9)dacrucial question is whether the antidia-betes effects of surgery are attenuated in

lower-BMI patients compared with se-verely obese individuals, who have beenmore extensively studied to date. Intui-tively, one might speculate that rates ofdiabetes remission and/or glycemic controlwould be lower among leaner patients be-cause such individuals lose less bodyweight after surgery (both in percent andabsolute terms) thandopeoplewith higherBMI values. Indeed, we have heard thisview expressed by prominent figures at sci-entific meetings for some time. However,recent evidence from large meta-analysesand RCTs does not support that assertion.

Panunzi et al. (18) performed an ex-tensive systematic review searching forpredictors of diabetes remission afterbariatric/metabolic surgery. They exam-ined all publications up through 2015reporting postsurgical diabetes remis-sion rates: a total of 94 articles describ-ing 94,579 surgical patients with T2DM(Fig. 2). Notably, they found that theoverall rate of diabetes remission wasequivalent among the 60 studies in whichmean preoperative BMI was $35 kg/m2

compared with the 34 studies with meanpreoperative BMI ,35 kg/m2 (71% vs.72%, respectively). Rates of diabetes

remission were also similar within eachindividual operation among patientswith a baseline BMI above versus below35 kg/m2 (overall remission 89% for BPD,77% for RYGB, 62% for LAGB, and 60%for VSG). Surprisingly, among manybaseline patient characteristics exam-ined, the only significant predictor ofthe magnitude of postoperative fall inHbA1c was lower preoperative waist cir-cumference. A major strength of this sys-tematic review is that it included allextant publications on the topic and wasthus very large. However, the authors didnot limit their analyses to only high-qualitystudies.

Accordingly, a meta-analysis was per-formed for the DSS-II conference exam-ining only level 1 evidence from the 11published RCTs directly comparing surgi-cal versus nonsurgical approaches to di-abetes care, including among manypatients with a baseline BMI ,35 kg/m2

(9). These trials analyzed 1,090 random-ized participants. Together they exam-ined all four clinically practiced bariatric/metabolic operations (RYGB, VSG, LAGB,and BPD), as well as a variety of behavioral/medical approaches, including very

Figure 1—Forest plot of T2DM remission rates after bariatric/metabolic surgery compared with medical/lifestyle interventions. The effect of eachsurgical vs. nonsurgical intervention is shown as the OR for T2DM remission with its 95% CI. Overall random effect OR 14.1 (95% CI 6.7–29.9, P ,0.001). AGB, adjustable gastric banding; SG, sleeve gastrectomy. Reprinted with permission from Muller-Stich et al. (31).

care.diabetesjournals.org Cummings and Cohen 927

intensive lifestyle interventions (35)modeled after Look AHEAD (Action forHealth in Diabetes) and Diabetes Preven-tion Program (DPP). As shown in Fig. 3A,all 11 RCTs reported superior results fromsurgery compared with medical/lifestyleinterventions for diabetes remission and/or glycemic control, with an overall OR forsurgical superiority of about 10. This con-stitutes unanimous level 1A evidence(i.e., meta-analysis of only RCTs) demon-strating that surgery improves diabetesmore than medical/lifestyle interven-tions do. The only study in which thePeto OR confidence intervals crossed 1was for LAGB, which is generally foundto be the least effective of these fouroperations for T2DM treatment.

Importantly, the magnitude of surgi-cal superiority over medical/lifestyle in-terventions for diabetes remission and/or glycemic control was similar amongthe trials in which the average baselineBMI of the study cohort was below ver-sus above 35 kg/m2 (Fig. 3A) (9). Therewas no trend toward reduction in therelative benefit of surgical comparedwith nonsurgical interventions on theseglycemic parameters based on decreas-ing preoperative BMI.Moreover, amongthe RCTs that have now reported bothearly and later follow-up data, the mag-nitude of surgical superiority overmedical/lifestyle interventions for glycemic out-comes is similar at 1–2 years and at 2–5years (Fig. 3B). As with the end points ofdiabetes remission and glycemic control,the degree of superiority for loweringHbA1c levels with surgical comparedwith nonsurgical interventions is similaramong RCTs wherein the study cohortsstarted with a mean baseline BMI belowor above 35 kg/m2 (Fig. 4). This finding isclearly displayed in the data from Surgi-cal Therapy And Medications Poten-tially Eradicate Diabetes Efficiently(STAMPEDE) trial, arguably the bestRCT in this arena to date. At all timepoints over the course of 3 years, surgi-cal patients consistently displayedgreater HbA1c lowering compared topatients treated with medical/lifestyleinterventions, but this finding wasequivalent among participants whoseaverage baseline BMI was below versusabove 35 kg/m2 (Fig. 5) (36).

A very important point to emphasizein interpreting all of the above studiescomparing surgical versus nonsurgicalapproaches to diabetes is that in most

Figure 2—Forest plots from a systematic review and meta-analysis of all published articlesreporting T2DM remission rates following bariatric/metabolic surgery. Data points display theeffectiveness of surgery to promote diabetes remission. Studies are divided into two groupsdepending on whether the average preoperative BMI for the study cohort was,35 kg/m2 (A) or$35 kg/m2 (B). DM, diabetes. Reprinted with permission from Panunzi et al. (18).

928 Bariatric/Metabolic Surgery in Lower-BMI Patients Diabetes Care Volume 39, June 2016

of them, the intensity of the lifestyle in-tervention and/or rigor of pharmaceuti-cal care (including use, or more typicallylack of standardized use, of medication-assisted weight loss) was not as aggres-sive as is possible. Although someof theseRCTs have involved quite intensive life-style/medical interventions (35,37,38),more work is needed in this domain.

Safety of Bariatric/Metabolic Surgeryin Patients With a BaselineBMI <35 kg/m2

The safety of bariatric/metabolic sur-gery in lower-BMI patients has been ex-amined most thoroughly in a very largesystematic review by the Agency forHealthcare Research and Quality (39).It examined the safety and comparativeeffectiveness of surgical versus nonsurgi-cal approaches to metabolic conditions

such as diabetes among patients with apreoperative BMI of 30–35 kg/m2. Thecomprehensive report confirmed that sur-gery caused greater reductions of BMI,HbA1c, hypertension, LDL, and triglyceridesthan did medical/lifestyle interventions.Importantly, the final summary statementreported that “rates of adverse events ofsurgery were relatively low,” surgical mor-tality was 0.0–0.3% (which is similar to his-torical data for patients with a BMI $35kg/m2 [2]), and “most surgical complica-tions were minor and tended not to re-quire major interventions” (39). They alsoconcluded that excessive (i.e., too much)weight loss is not a problem for standardproximal RYGB, VSG, or LAGB.

Demaria et al. (40) analyzed the BariatricOutcomes Longitudinal Database (BOLD)database from the American Society forMetabolic and Bariatric Surgery examining

66,264 patients who had undergoneRYGB; of whom, 235 had a baselineBMI,35 kg/m2, even though that is notyet approved for insurance coverage.There were no deaths within 90 days fol-lowing surgery in the low-BMI group, andthe complication rates in that cohort were3% for LAGB and 18% for RYGB, analogousto morbidity rates for patients with a BMI$35 kg/m2 (2). Clinical diabetes remissionrateswere also similar in thehigher- versuslower-BMI groups.

Long-term Effects of Surgeryin Patients With a BaselineBMI <35 kg/m2

Although long-term data regardingbariatric/metabolic surgery in lower-BMIpatients is relatively limited, some perti-nent evidence has begun to emerge inthis arena.

Figure 3—A: Forest plot of Peto ORs of main glycemic end points (Glyc. Endp.), as defined in each trial, from published RCTs of bariatric/metabolicsurgery compared withmedical/lifestyle treatments for diabetes. B: Forest plot of the trials depicted in panel A that have published both their initialshorter-term data and subsequent longer-term results from the same study. In both panels, data are arranged in order of ascending mean baselineBMI; the dotted line separates trials performed with cohorts exhibiting an average baseline BMI above or below 35 kg/m2. Study duration and HbA1cend point thresholds are shown in brackets in column 1, where “off meds” indicates a threshold achieved off all diabetes medications; otherwise,end points represent HbA1c thresholds achieved with or without such medications. ORs .1 indicate a positive effect of surgery compared withmedical/lifestyle treatment. For each study, the OR is shown with its 95% CI. The pooled Peto OR (95% CI) for all data were calculated under theassumption of a fixed-effects model. SG, sleeve gastrectomy.

care.diabetesjournals.org Cummings and Cohen 929

We prospectively studied the efficacyand safety of RYGB among 66 patientswith T2DM and a baseline BMI of 30–35 kg/m2, who were followed with100% retention for 6 years (25). The studycohort had severe, long-standingdiabetes(at baseline: average duration of diabetes13 years, mean HbA1c 9.7%, with 40% oninsulin and the rest on oral medications).Nevertheless, we observed a rapid de-crease of average HbA1c within the firstfew months, from nearly 10% down tonondiabetic levels, with subsequentmaintenance of that degree of improvedglycemia for 6 years (Fig. 6A). At the endof the study, 88% of participants still en-joyed diabetes remission (defined here asHbA1c ,6.5% off all diabetes medica-tions), another 11% clearly had improveddiabetes status, and only 1 patient out of66was unchanged.We found no relation-ship at any time point from 1 month to 6years between the magnitude of weightloss and the degree of improvement in anyglycemic variable (e.g., HbA1c, fastingplasma glucose, insulinogenic index duringa standardized meal test, and HOMA-insulin resistance). Systolic and diastolicblood pressure decreased progressivelythroughout the study, as did total choles-terol, LDL cholesterol, and triglycerides,and HDL cholesterol increased progres-sively for 6 years. These changes yieldedsubstantial, highly significant improve-ments in estimated 10-year risks of fataland nonfatal heart attacks and strokes.

A large, recent study by Hsu et al. (41)reported similar findings among EastAsian patients with T2DM and a baselineBMI ,35 kg/m2. Over 5 years, the au-thors examined the effects of eitherRYGB or VSG compared with medical/lifestyle diabetes care among 351 pa-tients with initial diabetes who were

matched between the surgical and non-surgical groups for age, BMI, and diabe-tes duration. Despite this matchingattempt, the surgical group had a higherbaseline average HbA1c (9.1% vs. 8.1%)and longer duration of diabetes (5.0 vs.2.7 years), both of which introduce con-servative biases against finding surgical

Figure 4—Forest plot of mean differences (MDs) of HbA1c serum levels after bariatric/metabolic surgery comparedwithmedical/lifestyle treatmentsin published RCTs related to diabetes. Data are arranged in order of ascending mean baseline BMI; the dotted line separates trials performed withcohorts exhibiting an average baseline BMI above or below 35 kg/m2. Study duration and HbA1c end point thresholds are shown in brackets incolumn 1, where “off meds” indicates a threshold achieved off all diabetes medications; otherwise, end points represent HbA1c thresholds achievedwith or without such medications. Negative MDs denote lower HbA1c levels following surgery than medical/lifestyle treatment. Data for each studyare shown as the MD with its 95% CI. A random-effects model was used to calculate the pooled standardized MD.

Figure 5—Change in mean 6 SE HbA1c levels over 3 years in a large RCT comparing surgical(either RYGB or VSG) vs. intensive medical therapy for T2DM. Each treatment group is dividedinto two subgroups defined by an average baseline BMI,35 kg/m2 vs.$35 kg/m2, as indicatedin the figure. Mean values in each group are provided below the graph, with median values inparentheses. P = 0.008 for comparison between the surgical and medical groups within thesubgroup of patients with a baseline BMI,35 kg/m2; P, 0.001 for that comparison within thesubgroup with a baseline BMI$ 35 kg/m2. Reprinted with permission from Schauer et al. (36).

930 Bariatric/Metabolic Surgery in Lower-BMI Patients Diabetes Care Volume 39, June 2016

superiority regarding glycemia. Never-theless, HbA1c and BMI were both re-duced to a far greater degree in thesurgical group, and these changes werelargely stable from 6 months to 5 years(Fig. 6B), even though surgery patientsended up on fewer diabetes medica-tions, including insulin. Follow-up at 5years was 96% in the surgical groupand 84% in the medical/lifestyle group.Maintenance of an HbA1c ,6.5% off alldiabetes medications at the end of thestudy was achieved in 64% of surgerypatients compared with 3% of patientstreated with medical/lifestyle interven-tions. At 5 years, the surgical group alsodisplayed greater reductions in waistcircumference, central adiposity, LDLcholesterol, triglycerides, blood pressure,and the percent of participants with hy-pertension. Death rates were statisticallyequivalent (1.9% with surgery, 3.0% withmedical/lifestyle interventions).

A long-term study of South Asian pa-tients reported somewhat less durableeffects on diabetes than were observedin the two articles highlighted in Fig. 6.Lakdawala et al. (42) performed a pro-spective observational analysis of 52Asian Indian patients with a BMI of 30–35 kg/m2 and poorly controlled T2DMat baseline who underwent RYGB andwere followed for 5 years. Althoughthe rate of complete diabetes remissionat 1 year was high at 73% (similar to thattypically seen at this time point afterRYGB in patients with BMI $35 kg/m2

[2,3,10]), full remission had dropped to58% by 5 years. However, this type oferosion of diabetes remission ratesover time is compatible with what is ob-served among patients with a preoper-ative BMI$35 kg/m2, in whom 35–50%of individuals who initially achieve dia-betes remission also eventually experi-ence relapse (11,15,16,36). With orwithout diabetes recurrence, the largemajority of patients with a baselineBMI either above or below 35 kg/m2

who undergo bariatric/metabolic surgerymaintain substantial improvement ofglycemic control for many years, andLakdawala et al. (42) reported that 96%of their study participants had improvedmetabolic status at 5 postoperative years.

Overall, thesefindingsamong lower-BMIpatients compare favorably with long-termstudies of bariatric/metabolic surgery forindividuals with T2DM and a baseline BMI$35 kg/m2 (11–13,43). However, apart

aa

aa

a

a

a aa

a

Figure 6—Long-term studies of bariatric/metabolic surgery to treat T2DM in patients with apreoperative BMI ,35 kg/m2. A: Change in mean 6 SE HbA1c levels following RYGB among 66patients with a baseline BMI of 30–35 kg/m2, studied with 100% follow-up for 6 years. HbA1cdecreased from values representing poorly controlled diabetes, despite all patients being ondiabetes medications at baseline, to nondiabetic or normal-range levels from 6 months to 6years after RYGB, with 88% of participants off all diabetes medications at the end of the study.Reprinted with permission from Cohen et al. (25). B: Changes over 5 years in mean HbA1c andBMI among 351 Asian patients with T2DM and a BMI ,35 kg/m2 at baseline who underwentsurgical (RYGB or VSG) vs. medical/lifestyle care for T2DM. aP, 0.001 for comparison betweenthe surgical group and medical group, calculated from a repeated-measures model that con-siders data over time. Reprinted with permission from Hsu et al. (41).

care.diabetesjournals.org Cummings and Cohen 931

from the above-mentioned 3-year datafrom STAMPEDE (36), long-term resultsfrom RCTs of lower-BMI patients are stillpending. Another understudied area isthe relative cost-effectiveness of bariatric/metabolic surgery compared with con-ventional care among less obese patientswith T2DM, and RCTs powered to observe“hard” outcomes such as cardiovascularevents, cancer, and death are neededamong patients of any BMI level.

CONCLUSIONS

Numerous RCTs and high-quality non-randomized comparisons now demon-strate that bariatric/metabolic surgery ismore effective than a variety of medical/lifestyle interventions for weight loss,glycemic control, T2DM remission, andimprovements in other cardiovasculardisease risk factors, with acceptablecomplications for at least 1–5 years (2).Even though individuals with lower base-line BMI levels lose less weight after sur-gery than do more obese people, thesafety and efficacy of surgery for improv-ing T2DM and other metabolic disordersappear to be similar among patientswith abaselineBMIbelowversus above35kg/m2,the threshold used to determine surgicalcandidacy for the past 25 years. Availableevidence indicates that this rather arbitrarycut point should be lowered for patientswith T2DM, in accordance with newDSS-II guidelines published in this issueof Diabetes Care (9).

Funding. D.E.C. is supported by NIH grants RO1DK103842, RO1 DK084324, RO1 DK089528, andU34 DK107917.Duality of Interest. D.E.C. is a principal investi-gator on the Comparison of Surgery vs. Medicinefor IndianDiabetes (COSMID) trial,which is fundedby Johnson & Johnson, and the Alliance of Ran-domized Trials of Medicine vs Metabolic Surgeryin Type 2 Diabetes (ARMMS-T2D) trial, which isfunded by Johnson & Johnson and Covidien, inconjunction with NIH. R.V.C. is principal investiga-tor for the Microvascular Outcomes After Meta-bolic Surgery (MOMS) trial, which is funded byJohnson& Johnson and theOswaldo Cruz GermanHospital Bioscience Institute. None of these stud-ies are discussed in this article. No other potentialconflicts of interest relevant to this article werereported.

References1. ConsensusDevelopment ConferencePanel. NIHconference. Gastrointestinal surgery for severeobesity. Ann Intern Med 1991;115:956–9612. SchauerPR,MingroneG, IkramuddinS,WolfeB.Clinical outcomes of metabolic surgery: efficacy of

glycemic control, weight loss, and remission ofdiabetes. Diabetes Care 2016;39:902–9113. Rubino F, Schauer PR, Kaplan LM, CummingsDE. Metabolic surgery to treat type 2 diabetes:clinical outcomes and mechanisms of action.Annu Rev Med 2010;61:393–4114. Thaler JP, Cummings DE. Minireview: Hor-monal and metabolic mechanisms of diabetesremission after gastrointestinal surgery. Endo-crinology 2009;150:2518–25255. Batterham RL, Cummings DE. Mechanismsof diabetes improvement following bariatric/metabolic surgery. Diabetes Care 2016;39:893–9016. Cohen R, Caravatto PP, Correa JL, et al. Gly-cemic control after stomach-sparing duodenal-jejunal bypass surgery in diabetic patients withlow body mass index. Surg Obes Relat Dis 2012;8:375–3807. Cummings DE, Cohen RV. Beyond BMI: theneed for new guidelines governing the use ofbariatric and metabolic surgery. Lancet Diabe-tes Endocrinol 2014;2:175–1818. Rubino F, Kaplan LM, Schauer PR, CummingsDE; Diabetes Surgery Summit Delegates. The Di-abetes Surgery Summit Consensus Conference:recommendations for the evaluation and use ofgastrointestinal surgery to treat type 2 diabetesmellitus. Ann Surg 2010;251:399–4059. Rubino F, Nathan DM, Eckel RH, et al.; Dele-gates of the 2nd Diabetes Surgery Summit. Met-abolic surgery in the treatment algorithm fortype 2 diabetes: a joint statement by interna-tional diabetes organizations. Diabetes Care2016;39:861–87710. Buchwald H, Estok R, Fahrbach K, et al.Weight and type 2 diabetes after bariatric sur-gery: systematic review and meta-analysis. Am JMed 2009;122:248–256.e511. Arterburn DE, Bogart A, Sherwood NE, et al.A multisite study of long-term remission andrelapse of type 2 diabetes mellitus followinggastric bypass. Obes Surg 2013;23:93–10212. Carlsson LM, Peltonen M, Ahlin S, et al.Bariatric surgery and prevention of type 2 dia-betes in Swedish obese subjects. N Engl J Med2012;367:695–70413. Adams TD, Davidson LE, Litwin SE, et al.Health benefits of gastric bypass surgery after6 years. JAMA 2012;308:1122–113114. Mingrone G, Panunzi S, De Gaetano A, et al.Bariatric surgery versus conventional medicaltherapy for type 2 diabetes. N Engl J Med2012;366:1577–158515. Mingrone G, Panunzi S, De Gaetano A, et al.Bariatric-metabolic surgery versus conventionalmedical treatment in obese patients with type 2diabetes: 5 year follow-up of an open-label,single-centre, randomised controlled trial.Lancet 2015;386:964–97316. Sjostrom L, Peltonen M, Jacobson P, et al.Association of bariatric surgery with long-termremission of type 2 diabetes and with microvas-cular and macrovascular complications. JAMA2014;311:2297–230417. Sjoholm K, Pajunen P, Jacobson P, et al.Incidence and remission of type 2 diabetes inrelation to degree of obesity at baseline and2 year weight change: the Swedish Obese Sub-jects (SOS) study. Diabetologia 2015;58:1448–1453

18. Panunzi S, De Gaetano A, Carnicelli A,Mingrone G. Predictors of remission of diabetesmellitus in severely obese individuals undergo-ing bariatric surgery: do BMI or procedurechoice matter? A meta-analysis. Ann Surg2015;261:459–46719. Panunzi S, Carlsson L, De Gaetano A, et al.Determinants of diabetes remission and glyce-mic control after bariatric surgery. DiabetesCare 2016;39:166–17420. Sjostrom L, Peltonen M, Jacobson P, et al.Bariatric surgery and long-term cardiovascularevents. JAMA 2012;307:56–6521. Eliasson B, Liakopoulos V, Franzen S, et al. Car-diovascular disease and mortality in patients withtype 2 diabetes after bariatric surgery in Sweden: anationwide, matched, observational cohort study.Lancet Diabetes Endocrinol 2015;3:847–85422. Sjostrom L, Gummesson A, Sjostrom CD,et al.; Swedish Obese Subjects Study. Effectsof bariatric surgery on cancer incidence in obesepatients in Sweden (Swedish Obese SubjectsStudy): a prospective, controlled interventiontrial. Lancet Oncol 2009;10:653–66223. Sjostrom L, Narbro K, Sjostrom CD, et al.;Swedish Obese Subjects Study. Effects of bariatricsurgery on mortality in Swedish obese subjects.N Engl J Med 2007;357:741–75224. Schauer PR, Kashyap SR, Wolski K, et al.Bariatric surgery versus intensive medical ther-apy in obese patients with diabetes. N Engl JMed 2012;366:1567–157625. Cohen RV, Pinheiro JC, Schiavon CA, SallesJE, Wajchenberg BL, Cummings DE. Effects ofgastric bypass surgery in patients with type 2diabetes and only mild obesity. Diabetes Care2012;35:1420–142826. Sjostrom L. Review of the key results fromthe Swedish Obese Subjects (SOS) trialda pro-spective controlled intervention study of bariatricsurgery. J Intern Med 2013;273:219–23427. Chiu M, Austin PC, Manuel DG, Shah BR, TuJV. Deriving ethnic-specific BMI cutoff points forassessing diabetes risk. Diabetes Care 2011;34:1741–174828. LeeWJ,WangW, Lee YC, HuangMT, Ser KH,Chen JC. Effect of laparoscopic mini-gastric by-pass for type 2 diabetes mellitus: comparison ofBMI.35 and ,35 kg/m2. J Gastrointest Surg2008;12:945–95229. Bays HE, Chapman RH, Grandy S; SHIELD In-vestigators’ Group. The relationship of body massindex to diabetes mellitus, hypertension and dys-lipidaemia: comparison of data from two nationalsurveys. Int J Clin Pract 2007;61:737–74730. Buse JB, Caprio S, Cefalu WT, et al. How dowe define cure of diabetes? Diabetes Care 2009;32:2133–213531. Muller-Stich BP, Senft JD, Warschkow R,et al. Surgical versus medical treatment oftype 2 diabetes mellitus in nonseverely obesepatients: a systematic review and meta-analysis.Ann Surg 2015;261:421–42932. Gloy VL, Briel M, Bhatt DL, et al. Bariatric sur-gery versus non-surgical treatment for obesity: asystematic reviewandmeta-analysis of randomisedcontrolled trials. BMJ 2013;347:f593433. Geloneze B, Geloneze SR, Chaim E, et al.Met-abolic surgery for non-obese type 2 diabetes: in-cretins, adipocytokines, and insulin secretion/resistance changes in a 1-year interventional clin-ical controlled study. Ann Surg 2012;256:72–78

932 Bariatric/Metabolic Surgery in Lower-BMI Patients Diabetes Care Volume 39, June 2016

34. RaoWS, Shan CX, ZhangW, Jiang DZ, QiuM.A meta-analysis of short-term outcomes of pa-tients with type 2 diabetes mellitus and BMI #35 kg/m2 undergoing Roux-en-Y gastric bypass.World J Surg 2015;39:223–23035. Cummings DE, Arterburn DE, Westbrook EO,et al. Gastric bypass surgery vs. intensive lifestyleand medical intervention for type 2 diabetes: theCROSSROADS randomized controlled trial. Diabe-tologia 2016;59:945–95336. Schauer PR, Bhatt DL, Kirwan JP, et al.;STAMPEDE Investigators. Bariatric surgery versusintensive medical therapy for diabetes–3-yearoutcomes. N Engl J Med 2014;370:2002–201337. Ikramuddin S, Korner J, Lee WJ, et al.Roux-en-Y gastric bypass vs intensive medicalmanagement for the control of type 2 diabetes,hypertension, and hyperlipidemia: the Diabetes

Surgery Study randomized clinical trial. JAMA2013;309:2240–224938. Courcoulas AP, Goodpaster BH, Eagleton JK,et al. Surgical vs medical treatments for type 2diabetes mellitus: a randomized clinical trial.JAMA Surg 2014;149:707–71539. Maglione MA, Gibbons MM, Livhits M,et al. Bariatric surgery and non-surgical ther-apy in adults with metabolic conditions and abody mass index of 30.0 to 34.9 kg/m2. InAHRQ Comparative Effectiveness Reviews.Rockville, MD, Agency for Healthcare Researchand Quality, 2013 Jun. Report No. 12(13)-EHC139-E40. Demaria EJ, Winegar DA, Pate VW, HutcherNE, Ponce J, Pories WJ. Early postoperative out-comes of metabolic surgery to treat diabetesfrom sites participating in the ASMBS bariatricsurgery center of excellence program as

reported in the Bariatric Outcomes LongitudinalDatabase. Ann Surg 2010;252:559–566; discus-sion 566–56741. Hsu CC, Almulaifi A, Chen JC, et al. Effect ofbariatric surgery vs medical treatment on type 2diabetes in patients with body mass index lowerthan 35: five-year outcomes. JAMA Surg 2015;150:1117–112442. Lakdawala M, Shaikh S, Bandukwala S,Remedios C, Shah M, Bhasker AG. Roux-en-Ygastric bypass stands the test of time: 5-yearresults in low body mass index (30-35 kg/m(2))Indian patients with type 2 diabetes mellitus.Surg Obes Relat Dis 2013;9:370–37843. Brethauer SA, Aminian A, Romero-TalamasH, et al. Can diabetes be surgically cured? Long-term metabolic effects of bariatric surgery inobese patients with type 2 diabetes mellitus.Ann Surg 2013;258:628–636; discussion 636–637

care.diabetesjournals.org Cummings and Cohen 933