Effect of the Duodenal-JejunalBypass Liner on Glycemic Controlin Patients With Type 2 DiabetesWith Obesity: A Meta-analysisWith Secondary Analysis onWeight Loss and HormonalChangesDiabetes Care 2018;41:1106–1115 | https://doi.org/10.2337/dc17-1985

OBJECTIVE

Duodenal-jejunal bypass liner (DJBL) is an endoscopic device that may mimic smallbowel mechanisms of Roux-en-Y gastric bypass (RYGB). Previous studies have dem-onstrated the efficacy of DJBL at inducingweight loss.We assessed the effect of DJBLon glycemic control in patients with type 2 diabetes (T2D) with obesity.

RESEARCH DESIGN AND METHODS

Data sources included MEDLINE, EMBASE, and Web of Science through 1 July 2017.Includedwere published studies that assessed DJBL outcomes in obese T2D patients.

RESULTS

Primary outcomeswere change in HbA1c and HOMA of insulin resistance (HOMA-IR).Secondary outcomes were change in weight and gut hormones glucose-dependentinsulinotropic peptide (GIP), glucagon-like peptide 1 (GLP-1), peptide YY (PYY), andghrelin.Seventeenstudieswereincluded.Atexplant,HbA1cdecreasedby1.3%[95%CI1.0, 1.6] and HOMA-IR decreased by 4.6 [2.9, 6.3]. Compared with control subjects,DJBL subjectshadgreaterHbA1c reductionby0.9%[0.5, 1.3]. Sixmonthsafterexplant,HbA1cremained lower thanbaselineby0.9%[0.6,1.2].Atexplant,patients lost11.3kg[10.3, 12.2], corresponding to aBMI reductionof 4.1 kg/m2 [3.4, 4.9], totalweight lossof18.9%[7.2,30.6],andexcessweight lossof36.9%[29.2,44.6].Theamountofweightloss remained significant at 1 year postexplantation. After DJBL, GIP decreased,whereas GLP-1, PYY, and ghrelin increased.

CONCLUSIONS

DJBL improves glycemic control and insulin resistance in T2D patients with obesity.DJBL also appears to induce significant weight loss in this population. Additionally,changes in gut hormones suggest mechanisms similar to RYGB. Study limitationsincluded heterogeneity among studies.

1Division of Gastroenterology, Hepatology andEndoscopy, Brigham and Women’s Hospital,Boston, MA2Harvard Medical School, Boston, MA3Division of Endocrinology, Diabetes and Hy-pertension, Brigham and Women’s Hospital,Boston, MA

Corresponding author: Christopher C. Thompson,cthompson@hms.harvard.edu.

Received 24 September 2017 and accepted 6February 2018.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc17-1985/-/DC1.

© 2018 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.More infor-mation is available at http://www.diabetesjournals.org/content/license.

Pichamol Jirapinyo,1,2 Andrea V. Haas,2,3

and Christopher C. Thompson1,2

1106 Diabetes Care Volume 41, May 2018

META-ANALYSIS

Type 2 diabetes (T2D) is one of the majorcomorbidities associated with obesity. Astheprevalenceofobesity rises, thenumberof patients affected by these comorbiditieswill continue to increase (1). In 2015, 30.3million Americans, or 9.4% of the U.S.population, had T2D, which resulted in a245 billion U.S. dollar cost to the healthcare system (2).Traditionally,T2Dhasbeentreatedwith

oral and injectable medications. Never-theless, despite maximal doses and acombination of medical therapy, a propor-tion of patients fail to achieve adequateglycemic control. More recently, it hasbeendemonstrated thatRoux-en-Y gastricbypass (RYGB) is effective at inducing T2Dremissionin;84–90%ofpatientsat1yearand 29–50% at 5 years (3–7). Specifically,Mingrone et al. (8) demonstrated in arandomized controlled trial (RCT) thatmore patients who underwent RYGBwere able to maintain T2D remission at5 years than those who were medicallymanaged(37%vs.0%,respectively).Similarly,a more recent RCT (the Surgical Treatmentand Medications Potentially Eradicate Di-abetes Efficiently [STAMPEDE] trial) dem-onstrated similar findings. At 5 years, morepatients who underwent RYGB had hemo-globin A1c (HbA1c) of#6% comparedwiththosewhoreceived intensiveantidiabetesmedical therapy alone (29% vs. 5%, re-spectively) (9,10). In addition, the RYGBgroup had a greater reduction in HbA1cthan those who received medical therapyalone (2.1% vs. 0.3%, respectively). As aresult, a recent joint statement by in-ternational diabetes organizations hassuggested that metabolic surgery be rec-ommended to treat T2D in patients withclass III obesity (BMI$40 kg/m2) regard-less of glycemic control and class II obesity(BMI 35–39.9 kg/m2) when hyperglyce-mia is inadequately controlled with life-styleandoptimalmedical therapy.Surgerymay also be considered for patients withT2Dandclass I obesity (BMI30–34.9 kg/m2)if hyperglycemia remains inadequatelycontrolled despite optimal treatmentwith either oral or injectable medications(11,12).It has been hypothesized that some of

themetaboliceffectofRYGBappearstobeindependent ofweight loss. In the foreguthypothesis, it is thought that RYGB ex-cludes the proximal small intestine frombeing in contact with ingested nutrients.Thismay lead to a decrease in the secretionofunidentifiedduodenal factorsthatusually

promote insulin resistance and T2D (13).Alternatively, the hindgut hypothesisproposes that RYGB leads toearly deliveryof nutrients to the distal small bowel,which then stimulates L cells in the termi-nal ileum and proximal colon to secretemore glucagon-like peptide 1 (GLP-1),peptide YY (PYY), and other hormones(14). These ultimately lead to enhancedinsulin secretion by pancreaticb-cells anddecreased insulin resistance (15,16).

Given a known impact of RYGB onglycemic control, an endoscopic bariatricandmetabolicdevice called theduodenal-jejunal bypass liner (DJBL) was developedto mimic some of the proposed smallbowel mechanisms of RYGB. Specifically,DJBL is a 60-cm fluoropolymer liner that isanchored at the duodenal bulb and endsat the jejunum. It is placed and removedendoscopically and allows nutrients topass directly from the stomach into thejejunum.Similar tothe foreguthypothesis,the duodenum and proximal jejunum areexcluded from direct contact with chyme,potentially downregulating production ofanti-incretins. At the same time, undi-gested nutrients and bile reach the distalsmall intestine before mixing, potentiallytriggering hindgut mechanisms includingupregulation of incretin production.

A recent meta-analysis has demon-stratedtheefficacyoftheDJBLoninducingweight loss in patients with obesity re-gardless of diabetes status (17). However,the study did not find significant meta-bolic effects in the secondary analysis.We therefore aim to conduct a systematicreview and meta-analysis with a primaryfocusonassessingtheeffectoftheDJBLonglycemic control in patients with obesityand concomitant T2D. In addition, ameta-regression was performed to assess pre-dictors of response.

RESEARCH DESIGN AND METHODS

Data Sources and SearchesThesearchstrategy,studyeligibilitycriteria,selection process, data collection process,primary and secondary outcomes, andanalyses were defined a priori and aredescribed below.

WesearchedthreedatabasesdMEDLINE,EMBASE, and Web of Sciencedfrom in-ception to1 July2017without languageorstudy design restriction. An extensivesearch strategy was employed to findarticles that relate to changes in HbA1cor weight or gut hormones after DJBL.

Specifically, the following were searchedby text words in the title and abstract:Endobarrier, duodenal sleeve*, jejunalsleeve*, gastrointestinal liner*, bypassliner*, duodenal jejunal bypass sleeve*,GI dynamics sleeve*, GI dynamic*.

Duplicates were removed. Two of theauthors (P.J. and A.V.H.) then indepen-dently reviewed titles and abstracts pro-duced by the search. Studies deemedpotentially relevant were reviewed in fullto determine eligibility. Disagreementsregarding final study inclusion were re-solved by discussion with the senior au-thor (C.C.T.).

Study Selection

Study Design and Population

RCTs, observational cohort studies, andcase series that were published and peerreviewed were included. Reviews, edito-rials, case-control studies, case reports,conference abstracts, and studies usingnonhuman subjects were excluded, aswere articles without full text availabilityorEnglish translation.Onlyonestudyfromthe same research group was selected topreserve independence of observations.Studies were included if there were adultsubjects (defined as age .18 years)with obesity (defined as BMI$30 kg/m2)and T2D (defined as HbA1c $6.5%, or48 mmol/mol), who underwent DJBL im-plantation. Studies that assessed changesin HbA1c as either a primary or second-ary outcome were included. If a studyincluded patients with obesity with andwithout T2D, only those with T2D wereincluded in the analysis. Studies that didnot contain a variance parameterwere ex-cluded. Corresponding authors were con-tactedforadditional informationifneeded.

Outcomes

The primary outcomes were the changein HbA1c and HOMA of insulin resistance(HOMA-IR) at time of DJBL explantationcompared with the levels at time ofimplantation.

Secondary outcomes were the changeinHbA1cpostexplantationandthechangesin weight and gut hormones at timeof explantation. In this study, the guthormones of interest included ghrelin,glucose-dependent insulinotropic peptide(GIP), GLP-1, and PYY. If a study includedboth fasting and postprandial values,fasting ghrelin, postprandial GIP, post-prandialGLP-1,andpostprandialPYY levelswere used for our analysis. Additionally,

care.diabetesjournals.org Jirapinyo, Haas, and Thompson 1107

pooled serious adverse events (SAEs) werereported. In this study, SAEs were definedas severe abdominal pain, dehydration,ulceration, gastrointestinal bleeding, acutepancreatitis, liver abscess, obstruction,and perforation.Weight changeswerereported using absolute weight loss inkilograms (kg), percent total weight loss(TWL), and percent excess weight loss(EWL).

Data Extraction, Risk of BiasAssessment, and Quality AssessmentStudycharacteristics,patientcharacteristics,andpredefined outcomeswere collected.The quality of cohort studies and RCTswas evaluated using the Newcastle-OttawaScale (NOS)and Jadadscore, respectively. Inthis study,highqualitywasdefinedasaNOSscore of$6 or a Jadad score of $3. Twoauthors (P.J. and A.V.H.) independentlyextracted data and assessed study qualityforeachofthearticles.Anydisagreementswere resolved by discussion with thesenior author (C.C.T.).

Data Synthesis and AnalysisOur primary analysis was the change inpercentage HbA1c at time of DJBL explan-tation compared with the level at time ofimplantation. Traditional forest plots withtwo-sided 95% CIs were constructed.Subgroup analysis of the RCT studies wasalso performed.Secondary analyses were defined a

priori and were performed to assess thechange in percentage HbA1c postexplan-tation compared with the level at time ofimplantation. Additionally, the effect ofDJBL on weight loss was assessed andreported using absoluteweight loss, TWL,or EWL. Gut hormonal change at time ofexplantation compared with baselinewasreported using Hedges g. Additionally,pooled SAEs were summarized.Meta-regression analyses were per-

formed to assess the influence of prede-finedclinicalcharacteristicsontheprimaryoutcome. Significant predictors from theunivariable regression analysis were in-cludedinthemultivariablemeta-regressionanalysis.For studies that only provided CIs or

interquartile ranges and those for whichwewere unable to obtain SD or SEM fromthe authors, a normal distribution wasassumed inorder tocalculateSDandSEM.Heterogeneity was assessed for the individ-ualmeta-analyses using thex2 test and theI2 statistic. Significant heterogeneity was

defined as P , 0.05 using the x2 or I2 .50%. A random-effects model was usedexcept when statistical heterogeneitywas not significant. Differences in sub-groups were assessed using a x2 test forinteraction with a P , 0.05 defined asstatistically significant. To assess for pub-licationbias, a funnelplotwascreatedandvisually inspectedforasymmetry.Thetrimand fill method was used to correct forfunnel plot asymmetry and provide anadjusted effect. The classic fail-safe testwas also applied to assess risk of biasacross studies. Analyses were performedusing Comprehensive Meta-Analysis, ver-sion 3.0 (Englewood, NJ).

RESULTS

Search ResultsA total of 1,064 studies were identified,147ofwhichwereduplicates.After abstractreview, 741 studies were excluded, leaving176 articles for full manuscript review. Fullarticle review resulted in 17 studies thatsatisfied all criteria and therefore were in-cluded in the systematic review and meta-analysis (Fig. 1) (18–34).

Outof17studies,14reportedtheeffectof DJBL on glycemic control in patientswith obesity and T2D; therefore, they wereincluded in the primary meta-analysis(Table 1). None of the 14 studies thatevaluated the glycemic effect of DJBLused a treat-to-target protocol. The re-maining three studies reported the effectof DJBL on at least one of the followingsecondary outcomes: postexplantationglycemic control, weight loss, or changes ingut hormones. These studies were includedin the secondary analyses. Out of the14 studies included in the primary meta-analysis, 9 were observational studies and5 were RCTs, with 2 comparing DJBL to asham procedure and 3 comparing DJBL tolifestyle modification. Only the DJBL armof the RCTs was included in the primaryanalysis, while both the DJBL and controlarms of the RCTs were included in asubgroup analysis of the RCTs.

Primary Outcome

Effect of DJBL on HbA1c at Time

of Explantation

A total of 14 included studies yielded atotal of 412 obese subjects with diabeteswho underwent attempted DJBL implan-tation. Of these, 388 underwent success-ful placement of DJBL (technical successrate of 94.2%). Reasons for unsuccessful

implantation included abnormalities ofthe small intestine (such as polyps, sharpangulations, or short duodenal bulbs)and anesthetic side reaction. Mean ageof the subjects ranged from 36 to 54 yearsold. Mean BMI and HbA1c at the time ofDJBL implantation ranged from 30.0 to48.9 kg/m2 and from 6.7 to 9.2% (50 to77 mmol/mol), respectively. On average,duration of T2D at the time of deviceimplant rangedfrom3.0 to14.8years. Themajority (75–100%) of the subjects wereonmetformin at the time of implantation,while 0–63% were on insulin. Baselinediabetes medication usage is shown inSupplementary Table 1. On average, thedevicewas implanted for8.464.0months.At the time of DJBL explantation, HbA1cdecreased by 1.3% [95% CI 1.0, 1.6] (P,0.0001) (Fig. 2A), which corresponded to13.3 mmol/mol [8.1, 18.5] (P , 0.0001).Heterogeneity across studies was high,with an I2 of 62 (P = 0.001) and a Q-valueof 34. Similarly, HOMA-IR significantlydecreased by 4.6 [2.9, 6.3] (P , 0.0001)(five studies with 91 patients; averagetimeof9.663.5monthsfromimplantation).Fasting insulin decreased by 4.8 mU/L[2.7, 6.8] (P , 0.0001) (six studies with122 patients; average time of 9.1 6 3.4months), and fasting glucose decreasedby 44.6 mg/dL [31.2, 57.9] (P , 0.0001)(nine studies with 164 patients; averagetime of 8.26 3.5 months).

In studies with baseline HbA1c$8%, or64 mmol/mol, there was a greater re-duction in HbA1c and fasting glucose com-paredwith studieswith lower initialHbA1c(change in HbA1c 21.4% [21.7, 21.0]vs.20.5%[21.3,0.2]or215.3mmol/mol[218.6, 210.9] vs. 25.5 mmol/mol[214.2, 2.2], P for the difference = 0.03)(change in fasting glucose 249.2 mg/dL[256.8, 241.5] vs. 25.4 mg/dL [239.5,28.7], P for the difference = 0.01). Therewere no significant differences in theamount of change in HOMA-IR and fastinginsulin in studies with higher versus lowerbaseline HbA1c. For the observationalstudies, the NOS scores were 8 in fivestudies and 7 in four studies. The majorshortcoming was lack of blinding of out-comeassessors inall studies. For theRCTs,the Jadad scores were 3 in two studiesand 2 in three studies. None of the RCTstudies were double-blind.

Risk of bias across studieswas assessedusing a funnel plot. Visual inspectiondemonstrated that smaller and statisti-cally significant studies appeared to be

1108 Duodenal-Jejunal Bypass Liner and Glycemic Control Diabetes Care Volume 41, May 2018

missing. The Duval and Tweedie trim andfill method resulted in a greater decreasein HbA1c at the time of explantation com-pared with baseline (1.4% [1.0, 1.7] vs.1.3% [1.0, 1.6]) (15.3 mmol/mol [10.9,18.6] vs. 14.2mmol/mol [10.9, 17.5]). Theclassic fail-safe method suggested that598 studies would be required to showno change in HbA1c at the time of DJBLexplantation.

Subgroup Analysis of RCTs

One of the five RCT studies had only oneobese patient with diabetes in the controlarm and therefore was excluded from thesubgroup analysis of the RCTs (33). Theremaining four RCT studies included atotal of 116 patients with obesity andconcomitant T2D (63 DJBL subjects vs.53 control subjects). Of these, two in-tended to treat T2D with DJBL with HbA1cbeing a primary or coprimary outcome,while the remaining two intended to treat

obesity with DJBL with changes in HbA1cbeing one of the secondary outcomes.

On average, the device was implantedfor 4.56 1.7 months. At the time of DJBLexplantation, theDJBLgrouphadagreaterdecrease in HbA1c by 0.9% [0.5, 1.3] (P,0.0001) (Fig. 2B), or 9.7 mmol/mol [4.8,14.7], compared with the control group(P, 0.0001). Heterogeneity across stud-ies was low with an I2 of 0 (P = 0.848) andaQ-value of 0.8. The remove-one analysisshowed consistent results with a greaterdecrease in HbA1c in the DJBL arm com-pared with the control arm by 0.9% [0.5,1.3], or 9.7 mmol/mol [4.8, 14.7].

Secondary Outcomes

Effect of DJBL on HbA1c Postexplantation

Six studies reported follow-upHbA1c afterDJBLexplantation.Ofthese, two, four,andtwo studies assessed HbA1c levels at 3, 6,and 12 months postexplantation, respec-tively. Of note, some studies reported

postexplantation HbA1c at more than onetime point. Supplementary Table 2 dem-onstrates T2Dmedication usage through-out the studyperiod, including at the timeof postexplantation follow-ups.

At 3 months after device removal (twostudies with 46 patients; average totaltime of 12.3 months from implantation),HbA1c remained significantly lower thanthe baseline level by 2.2% [1.6, 2.7] (P ,0.0001), or 19.1mmol/mol [8.0, 30.2] (P =0.001). Heterogeneity across studieswas low with an I2 of 42 (P = 0.189) anda Q-value of 1.7.

At 6 months after device removal (fourstudieswith 120 patients; average time of15.1 months from implantation), HbA1cremained significantly lower than thebaseline level by 0.9% [0.6, 1.2] (P ,0.0001),or10.0mmol/mol [7.0,13.1] (P,0.0001). Heterogeneity across studieswas low with an I2 of 47 (P = 0.129) anda Q-value of 5.7.

Figure 1—PRISMA flow diagram. Search and selection process used for studies included in the meta-analysis.

care.diabetesjournals.org Jirapinyo, Haas, and Thompson 1109

Tab

le1—

Charac

teristicsofstudiesincluded

intheprimarymeta-an

alysis

Cou

ntry

N*

NwithT2D

Age

(years)

Female

sex(%)

Baseline

BMI(kg/m

2)

Durationof

T2D(years)

Implant

duration

(weeks)

Pre-DJBL

HbA

1c(%)

Post-DJBL

HbA

1c(%)

Pre-DJBL

weight(kg)

Post-DJBL

weight(kg)

Coh

ortstud

ies

Betzel2017

theNetherlands

185

185

526

849

35.16

4.3

8[1–36]

468.36

3.6

7.76

3.6

107.16

17.7

94.36

16.7

Coh

en2013

Sweden

1616

506

2738

30.06

3.6

528.66

0.8

7.56

1.6

82.16

18.0

76.26

17.2

deJonge2013

theNetherlands

1717

516

818

37.06

5.4

248.46

0.8

7.06

0.8

116.06

23.9

103.36

21.3

deMou

ra2012

Brazil

2222

466

1186

44.86

7.4

428.96

1.7

6.66

1.4

121.86

27.4

101.66

20.6

Escalona

2012

Chile

396

366

1080

43.76

5.9

52Decreased

by1.06

1.5

Decreased

by17.16

10.5

Kavalkova2016

Czech

Repub

lic30

3033–65

3342.76

6.6

13.06

4.2

409.06

3.9

7.26

3.4

129.76

24.1

117.36

23.6

Roh

de2017

Denmark

199

506

2038.66

6.6

3.06

4.0

266.76

3.1

6.86

3.2

109.46

18.4

103.66

18.0

Stratm

ann2016**

Germany

1616

506

819

48.86

8.5

9.06

6.1

528.26

3.0

7.06

3.0

Vilarrasa2017

Spain

2121

546

1043

33.46

1.9

14.86

8.5

529.16

1.3

7.86

1.7

91.36

12.3

78.46

11.5

RCTs Gersin2010**

U.S.

47DJBL

219

456

771

46.06

5.0

127.76

2.5

7.26

2.1

Sham

266

436

1089

46.06

6.0

127.66

1.3

6.86

0.2

Koehestanie2014

theNetherlands

73DJBL

3434

506

1238

34.66

4.2

5.06

3.7

248.36

1.0

7.06

0.8

105.46

13.3

94.86

18.1

Con

trol

3939

496

836

36.86

7.0

5.06

3.0

248.36

0.9

7.96

1.3

110.86

21.7

105.56

33.9

Rod

riguez

2009

Chile

18DJBL

1212

456

767

38.96

5.9

3.56

2.5

24Decreased

by2.46

2.4

Decreased

by10.26

3.9

Sham

66

516

1350

39.06

7.2

4.26

2.1

24Decreased

by0.86

1.0

Decreased

by7.36

8.6

Scho

uten

2010**

theNetherlands

37DJBL

268

40.96

10.0

8548.96

5.4

128.86

1.7

7.76

1.8

Con

trol

112

41.26

9.7

8249.26

5.9

127.36

0.1

6.96

0.6

Tarnoff2009**

Chile

39DJBL

253

38.06

10.1

6042.06

5.1

127.26

0.9

6.66

0.8

Con

trol

141

43.06

10.6

5740.06

3.5

1212.6

7.8

Dataarepresentedas

means,m

eans

6SD

,median[range],or

rangeun

less

otherw

iseindicated.

*Num

berof

patientswho

successfullyun

derw

entDJBLimplantation

.**N

oweightdata

onthediabetes

subgroup

.

1110 Duodenal-Jejunal Bypass Liner and Glycemic Control Diabetes Care Volume 41, May 2018

At12monthsafterdevice removal (twostudies with 80 patients; average time of23.0 months from implantation), therewasnostatisticallysignificantdifferenceinHbA1c compared with the baseline level(P = 0.865). Heterogeneity across studieswas high with an I2 of 68 (P = 0.078) and aQ-value of 3.1.

Effect of DJBL on Weight Loss in Patients

With Obesity and Concomitant Diabetes

Ten studies with a total of 352 patientsassessed the effect of DJBL on theamount of weight loss in patients withobesity and concomitant T2D. At thetime of explantation (9.2 6 3.1 monthsfrom device implantation), patients lost11.3 kg [10.3, 12.2] (P , 0.0001) (Fig. 3).Heterogeneity across studies was lowwith an I2of25(P=0.209)andaQ-valueof12.1.ThiscorrespondedtoadecreaseinBMIof 4.1 kg/m2 [3.4, 4.9] (P , 0.0001) (nine

studieswith 340 subjects), TWLof 18.9%[7.2, 30.6] (P = 0.002) (four studies with305 subjects), and EWL of 36.9% [29.2,44.6] (P , 0.0001) (four studies with301 subjects). Six studies reported follow-up weights after DJBL explantation. Ofthese, one, three, and two studies as-sessed follow-up weights at 3, 6, and12 months postexplantation, respec-tively.

At6monthsafterdeviceremoval (threestudieswith 104 subjects; average time of15.3 months from implantation), patientweights remained significantly lower thanthe baseline weights by 7.1 kg [5.2, 9.0](P , 0.0001). This corresponded to adecreaseinBMIof2.1kg/m2[1.1,3.1] (P,0.0001) (three studies with 104 subjects),TWL of 8.0% [5.9, 10.2] (P, 0.0001) (twostudies with 131 subjects), and EWL of25.5% [13.5, 37.4] (P , 0.0001) (twostudies with 131 subjects).

Similarly, at 12 months after deviceremoval (two studies with 80 subjects;average time of 23.0 months from im-plantation), patientweights remained sig-nificantly lower than the baselineweightsby 10.7 kg [5.0, 16.4] (P , 0.0001). Thiscorresponded to a decrease in BMI of 3.9kg/m2 [1.4, 6.3] (P = 0.002) (two studieswith 80 subjects), TWL of 7.2% [5.5, 8.9](P , 0.0001) (two studies with 80 sub-jects), and EWL of 27.7% (one study with59 subjects).

Effect of DJBL on Gut Hormones at Time

of Explantation

Seven studies with a total of 124 patientsreported the changes in gut hormonesafter DJBL implantation. Of these, six, five,four, and three studies assessed changesinGLP-1,GIP,ghrelin,andPYY,respectively.GLP-1. Of the six studies, five assessedchanges in postprandial GLP-1 (22,26,30,

Figure2—Forestplotsofstudiesassessingthechanges inHbA1cafterDJBL inpatientswithobesityandconcomitantT2DinRCTsandobservationalstudies(A)and in RCTs only (B). LS, lifestyle intervention.

care.diabetesjournals.org Jirapinyo, Haas, and Thompson 1111

32,34), while one evaluated changes infastingGLP-1afterDJBL (28).Overall, GLP-1 level did not significantly change afterDJBL (Hedges g of 0.33 [20.09, 0.75], P =0.13, I2 = 77). However, when only studiesthat evaluated postprandial GLP-1 wereincluded (the preferred method of analy-sis) (14), therewas a significant increase inpostprandialGLP-1levelsafterDJBL(Hedgesg of 0.46 [0.02, 0.90], P = 0.04, I2 = 75).GIP. Of the five studies, four assessedchanges inpostprandialGIP (21,26,30,32),whileoneevaluatedchanges in fastingGIPafter DJBL (28). GIP level was significantlylower than the baseline level (Hedges gof 20.36 [20.57, 20.15], P = 0.001, I2 =26 when all five studies were included)(Hedges g of 20.33 [20.56, 20.10], P =0.006, I2 = 37 when only four studies withpostprandial GIP were included).Ghrelin.All four studies evaluated changesin fasting ghrelin (21,28,32,34). AfterDJBL, fasting ghrelin significantly increased(Hedges g of 1.45 [0.25, 2.65], P = 0.018,I2 = 92).PYY. Of the three studies, two assessedchanges inpostprandial PYY (30,32),whileone evaluated changes in fasting PYY (34).Overall, PYY significantly increased afterDJBL (Hedges g of 0.57 [0.08, 1.05], P =0.023, I2 = 61 when all three studies wereincluded), although this change becamenonsignificantwhenonly two studieswithpostprandial PYY were included in theanalysis (Hedges g of 0.86 [20.42, 2.13],P = 0.19, I2 = 53).

SAEs

Nine studies reported adverse events inobese patients with diabeteswho received

DJBL treatment. Most common adverseevents were abdominal pain, nausea, andvomiting, which occurred more commonlyimmediately after DJBL implantation. SAEsoccurred in 55 out of 350 DJBL cases(15.7%). These included gastrointestinalbleeding (16), hypoglycemic events (8),acute pancreatitis (6), sleeve migration(5), severe abdominal pain (4), liver ab-scesses (4), anchor perforation next tothe pylorus into the stomach (4), sleeveobstruction (3), esophageal perforationduringexplantation(2),acutecholecystitisand duodenal fistula (1), ulceration atthe duodenal bulb (1), and dehydration(1). In the control group, 19 hypoglycemicevents were reported in 47 patients.Supplementary Table 3 details each SAEand its corresponding management.

Meta-Regression AnalysisOnaunivariablemeta-regressionanalysis,baseline HbA1c and the amount of weightlossweresignificantpredictorsoftheamountof HbA1c decrease at the time of DJBLexplantation (b =20.78 [21.28,20.27],P=0.002forbaselineHbA1candb=20.08[20.16, 20.01], P = 0.027 for amountof weight loss). Baseline BMI, durationof T2D, baseline insulin use, change inHOMA-IR, and duration of device implan-tationwerenot associatedwith theamountof HbA1c change (P . 0.05 for all). On amultivariable meta-regression analysis,baseline HbA1c had a trend toward beingasignificantpredictorof the improvementin glycemic control after DJBL after control-ling for amount of weight loss (b = 20.72[21.50, 0.05], P = 0.067) (Supplementary

Fig. 1). The amount of weight loss nolonger predicted the amount of HbA1c

changeoncecontrolled forbaselineHbA1c(P = 0.14).

CONCLUSIONS

This systematic review and meta-analysisis thefirst to focus onapatientpopulationwith obesity and concomitant T2D. Priorstudies included subjects with obesityregardless of diabetes status encompass-ing different phenotypes that may re-spond differently to a specific therapy.Our study demonstrates that DJBL is asso-ciated with significant improvementsin glycemic control and weight loss, anassociation also notable in the poorlycontrolled subgroup. These metabolicandweight-losseffectsappeartolastuptoat least 6 months and 12 months afterdeviceremoval, respectively.Additionally,DJBL is associatedwithan increase inGLP-1and PYY and a decrease in GIP, supportinga possible incretin mechanism. The safetyprofile of DJBL appears to be acceptable,with a 15.7% SAE rate and with ,1% ofpatients requiring surgical intervention toaddress adverse events.

Given the rising number of patientswith obesity and concomitant T2D, effec-tive treatments for these diseases areurgently needed. While medications andlifestyle modification are effective, somepatients still fail to respond and requiremore aggressive therapy. Bariatric sur-geries have been shown to be effectiveat treating obesity and T2D, with recentrecommendations suggesting their useas a treatment option for T2D in patients

Figure 3—Forest plot of studies assessing the changes in weight after DJBL in patients with obesity and concomitant T2D.

1112 Duodenal-Jejunal Bypass Liner and Glycemic Control Diabetes Care Volume 41, May 2018

with BMI$40 kg/m2 or$35 kg/m2 if T2Dremains inadequately controlled despiteoptimal medical therapy. Nevertheless,despite its efficacy, only;1% of patientswhoareeligibleundergosurgery, leavingamajority of the patient population sub-optimally managed (35,36).Endoscopic bariatric and metabolic

therapies (EBMTs) have risen to fill thetreatment gap between medical andsurgical therapies. EBMTs may be furthercategorized into gastric and small boweltherapies (37). Previous meta-analysisstudies have demonstrated a significantimpactofgastricEBMTsonweightlossandobesity-relatedcomorbidities.Specifically,a meta-analysis of 17 studies including1,683 patients shows that the ORBERAintragastric balloon (IGB) (Apollo Endosur-gery, Austin, TX) is associated with EWL of25.44% and TWL of 11.27% at 12 monthsafter implantation (36). Additionally, IGBsare demonstrated to be associatedwith adecreaseinHbA1cby0.6%,or6.6mmol/mol(12 observational studies; 470 subjectswith mean baseline HbA1c of 6.1% [5.5,6.6], or 43 mmol/mol [37, 49]) (38). Froma small bowel therapy standpoint, DJBLsremain the most widely studied deviceworldwide. A recent meta-analysis showsthat DJBLs are associated with weightloss of 5.1 kg (four RCTs; 151 subjects)andEWLof12.6%(fourRCTs;166subjects)in patients with obesity (with or with-out T2D) (17). In this meta-analysis, onlytwo studies were included in the diabetessubgroup analysis and no improvement inglycemic indexeswere identified. Further-more, long-term efficacy data of theseEBMT devices remain to be determined.With the aim to assess the effect of

DJBLs on T2D, our study includes onlypatients with obesity and concomitantT2D. Out of the 14 studies included in theprimary analysis, 8 include only patientswithobesity anddiabetes. For the remain-ing six, four report glycemic outcomes indiabetes subgroups, and we were able toreceive the diabetes subgroup data fromthe authors of the remaining two studies.In ourmeta-analysis, at 1 year after deviceimplantation, HbA1c decreases by 1.3% or14.2 mmol/mol (nine observational stud-ies and five RCTs; 412 subjects) comparedwith baseline, which is approximatelytwice the decrease seen in observa-tional studies with IGBs, although base-line HbA1c was lower in these studies.Additionally, our study also demon-

strates that DJBL is associated with

significant weight loss in patients withobesity andconcomitantT2D. Specifically,at the time of device explantation, pa-tients lose 11.3 kg (ten studies; 352 pa-tients), which corresponds to a decrease inBMI of 4.1 kg/m2 (nine studies; 340 sub-jects),TWLof18.9%(fourstudies;305sub-jects), and EWL of 36.9% (four studies;301subjects). Interestingly, theamountofweight loss associated with DJBL appearsto be greater in patients with obesity andconcomitantdiabetescomparedwiththosewith obesity alone. This observation sug-gests that these two patient populationsmayrepresentdifferentphenotypes,whichmay cause them to have a different re-sponse to theprocedure. Further studies toinvestigate this observation arewarranted.

In our study, DJBL is also shown tobe associated with changes in incretinhormones in a similar direction as thosefollowingRYGB, suggesting that it touchesupon some analogous small bowel mech-anismsofaction. Specifically,postprandialGLP-1 and PYY increase after DJBL. GLP-1and PYY are anorectic incretin hormonesthat are secreted postprandially in thedistal ileum/proximal colon. In addition toinducingweight loss (via increasing satietyand delaying gastric motility), both hor-mones are thought to play a key role inimproved glucose homeostasis by induc-ing a release of insulin from pancreaticcells following ingestion of oral glucose.Previous studies have shown that afterRYGB, GLP-1 and PYY increase comparedwith the presurgical levels (39–41). Ofnote, in our study, a subgroup analysis ofpostprandial GLP-1was statistically signif-icant from the baseline pre-DJBL levels.For PYY, a subgroup analysis on thepostprandial level yielded a nonsignifi-cant result. This could be due to a lack ofstatistical power, with only two studiesreporting postprandial PYY values. In ourstudy,GIPdecreasedafterDJBL.Literatureon GIP levels after RYGB remain hetero-geneous (42,43); therefore, no compari-son of GIP changes between DJBL andRYGB can be made. Furthermore, unlikeRYGB that affects both gastric and smallbowel anatomy, DJBL only affects smallbowelmechanisms. Therefore, asexpected,the gastric gut hormone, i.e., ghrelin, whichis secreted in the fundus of the stomach,does not decrease after DJBL (unlike RYGB).In fact, our study shows that ghrelin in-creases after DJBL, a finding similar to thatseen after dieting and exercise, likely as aresponse to weight loss (44).

TreatmentwithDJBL isaccompaniedbyadverse events, with some being serious.In our study, the pooled SAE rate was15.7%, which is comparable to otherEBMTs. Most common SAEs were GIbleeding, the majority of which weretreated with supportive care. There werealso four hepatic abscess cases, with nonerequiring surgical intervention. In the17 included studies, no deaths were re-ported.Ofnote, theU.S. SafetyandEfficacyof EndoBarrier in Subjects With Type 2DiabetesWhoAreObese (ENDO) trialwasstoppedearlyduetoahepaticabscessrateof 3.5%, which was above the predeter-mined threshold. Similarly, none of thesepatients required intensive care unit stayor surgical intervention, and there wereno deaths.

This study has several limitations.First, the primary outcome of the study,i.e., the change in HbA1c at time of DJBLexplantation, does not take into accountthe changes in antidiabetes medicationsthroughoutthestudyperiod.Additionally,many studies were initiated prior to theavailability of sodium–glucose cotrans-porter 2 inhibitors; therefore, they didnot include patients who were on theseagents. Similarly, patients on GLP-1 ago-nists were included only in the laterstudies. Therefore, not all patients hadmaximized the medical therapy for T2D.Nevertheless,moststudiesreportadecreasein medication dosages and/or discontin-uation of antidiabetes medication includ-ing insulin. Therefore, the change inHbA1creported in this study is likely conserva-tive and possibly underestimates the trueeffect on glucose homeostasis. Second,the majority of the included studies wereobservational studies. To account forthis, a subgroup analysis of RCTs wasperformed. Moreover, none of the RCTsusedatreat-to-targetprotocol,whichmayhaveaffectedtheamountofHbA1cchanges.Third, each of the included studies hasvaried loss–to–follow-uprates,whichmayhave introduced a bias. If available, theoutcome reported using an intention-to-treat analysis is used in the meta-analysisinordertominimizethis loss–to–follow-upbias. Furthermore, due to a limited numberof included studies, only three covariatesareallowedinourmeta-regressionmodel.Therefore, these covariates are defined apriori usingexpertopinionwith avoidanceof collinearity. Last, heterogeneity amongstudies remains relatively high, especiallyfor studies that evaluate the effectofDJBL

care.diabetesjournals.org Jirapinyo, Haas, and Thompson 1113

on gut hormones. These differences arelikely due to different methods, assays, andtiming of measurement. To control for this,Hedges g is used to report the directionof changes in gut hormones. However, theinterpretation of themagnitude of changesremains limited. Additionally, in order toaddress the heterogeneity of the includedpatient population among different stud-ies, a subgroup analysis of studies withaverage baselineHbA1c of$8% vs.,8% isalso conducted, confirming a significanteffect for those subjects with poorly con-trolled diabetes.In summary, this systematic reviewand

meta-analysis suggests that DJBL is asso-ciated with significant improvement inglycemic indexes for patientswith obesityand concomitant T2D. Additionally, sub-stantial weight loss is experienced in thispopulation,whichpersists for at least 1 yearafterdeviceremoval.Movingforward, smallbowel EBMTs may have implications as anadjunct therapy to pharmacotherapy andlifestyle intervention for the care of patientswith obesity and concomitant T2D.

Duality of Interest. C.C.T. has contracted re-search for Aspire Bariatrics, USGI Medical, Spatz,andApolloEndosurgery;hasservedasaconsultantfor Boston Scientific, Covidien, USGI Medical, Olym-pus, and Fractyl; holds stock and royalties for GIWindows and EndoSim; and has served as an expertreviewerforGIDynamics.Nootherpotentialconflictsof interest relevant to this article were reported.Author Contributions. P.J. contributed to thestudy design, acquisition of data, analysis andinterpretation of data, drafting of themanuscript,and final approval of the manuscript. A.V.H. con-tributed to the acquisition of data and finalapproval of the manuscript. C.C.T. contributedto the study design, critical review of the manu-script, and final approval of the manuscript.Prior Presentation. Parts of this study were pre-sented as anoral presentation atDigestiveDiseaseWeek (DDW) 2017, Chicago, IL, 6–9 May 2017.

References1. Ogden CL, Carroll MD, Fryar CD, Flegal KM.Prevalence of obesity among adults and youth:United States, 2011-2014. NCHS Data Brief 2015;219:1–82. AmericanDiabetesAssociation.Economiccostsofdiabetes in theU.S. in2012.DiabetesCare2013;36:1033–10463. Buchwald H, Avidor Y, Braunwald E, et al.Bariatric surgery: a systematic review and meta-analysis. JAMA 2004;292:1724–17374. PournarasDJ, AasheimET, Søvik TT, et al. Effectof the definition of type II diabetes remission inthe evaluation of bariatric surgery for metabolicdisorders. Br J Surg 2012;99:100–1035. Brethauer SA, Aminian A, Romero-Talamas H,et al. Can diabetes be surgically cured? Long-term

metabolic effects of bariatric surgery in obesepatients with type 2 diabetes mellitus. Ann Surg2013;258:628–636; discussion 636–6376. Buse JB, Caprio S, Cefalu WT, et al. How do wedefine cure of diabetes? Diabetes Care 2009;32:2133–21357. Mingrone G, Panunzi S, De Gaetano A, et al.Bariatric surgery versus conventional medicaltherapy for type 2 diabetes. N Engl J Med 2012;366:1577–15858. Mingrone G, Panunzi S, De Gaetano A, et al.Bariatric-metabolic surgery versus conventionalmedical treatment in obese patients with type 2diabetes: 5 year follow-upof anopen-label, single-centre, randomised controlled trial. Lancet 2015;386:964–9739. Schauer PR, Bhatt DL, Kirwan JP, et al.;STAMPEDE Investigators. Bariatric surgery versusintensive medical therapy for diabetesd5-yearoutcomes. N Engl J Med 2017;376:641–65110. Schauer PR, Bhatt DL, Kirwan JP, et al.;STAMPEDE Investigators. Bariatric surgery versusintensive medical therapy for diabetes–3-yearoutcomes. N Engl J Med 2014;370:2002–201311. Brito JP, Montori VM, Davis AM. Metabolicsurgery in the treatment algorithm for type 2diabetes: a joint statement by international di-abetes organizations. JAMA 2017;317:635–63612. Rubino F, Schauer PR, Kaplan LM, CummingsDE. Metabolic surgery to treat type 2 diabetes:clinicaloutcomesandmechanismsof action.AnnuRev Med 2010;61:393–41113. Rubino F, ForgioneA, CummingsDE, et al. Themechanism of diabetes control after gastrointes-tinal bypass surgery reveals a role of the proximalsmall intestine in the pathophysiology of type 2diabetes. Ann Surg 2006;244:741–74914. JirapinyoP,JinDX,QaziT,MishraN,ThompsonCC.Ameta-analysisofGLP-1afterRoux-en-Ygastricbypass: impact of surgical technique andmeasure-ment strategy. Obes Surg 2018;28:615–62615. Mason EE. The mechanisms of surgical treat-ment of type 2 diabetes. Obes Surg 2005;15:459–46116. Meek CL, Lewis HB, Reimann F, Gribble FM,Park AJ. The effect of bariatric surgery on gastroin-testinal and pancreatic peptide hormones. Peptides2016;77:28–3717. RohdeU,HedbackN,GluudLL,VilsbøllT,KnopFK. Effect of the EndoBarrier Gastrointestinal Lineron obesity and type 2 diabetes: a systematic reviewandmeta-analysis. Diabetes Obes Metab 2016;18:300–30518. Betzel B, Homan J, Aarts EO, et al. Weightreduction and improvement in diabetes by theduodenal-jejunalbypass liner:a198patientcohortstudy. Surg Endosc 2017;31:2881–289119. Betzel B, Koehestanie P, Homan J, et al.Changes in glycemic control andbodyweight afterexplantation of the duodenal-jejunal bypass liner.Gastrointest Endosc 2017;85:409–41520. Cohen R, le Roux CW, Papamargaritis D, et al.Role of proximal gut exclusion from food on glu-cosehomeostasis inpatientswithType2diabetes.Diabet Med 2013;30:1482–148621. deJongeC,RensenSS,VerdamFJ,etal. Impactof duodenal-jejunal exclusion on satiety hor-mones. Obes Surg 2016;26:672–67822. de Jonge C, Rensen SS, Verdam FJ, et al.Endoscopic duodenal-jejunal bypass liner rapidlyimproves type 2 diabetes. Obes Surg 2013;23:1354–1360

23. de Moura EGH, Martins BC, Lopes GS, et al.Metabolic improvements in obese type 2diabetessubjects implanted for 1 year with an endoscop-ically deployed duodenal-jejunal bypass liner.Diabetes Technol Ther 2012;14:183–18924. Escalona A, Pimentel F, Sharp A, et al. Weightloss and metabolic improvement in morbidlyobese subjects implanted for 1 year with an en-doscopic duodenal-jejunal bypass liner. Ann Surg2012;255:1080–108525. Gersin KS, Rothstein RI, Rosenthal RJ, et al.Open-label,sham-controlledtrialofanendoscopicduodenojejunal bypass liner for preoperativeweight loss inbariatric surgery candidates.Gastro-intest Endosc 2010;71:976–98226. Kavalkova P, Mraz M, Trachta P, et al.Endocrine effects of duodenal-jejunal exclusionin obese patients with type 2 diabetes mellitus.J Endocrinol 2016;231:11–2227. KoehestanieP,deJongeC,BerendsFJ, JanssenIM, Bouvy ND, Greve JW. The effect of the endo-scopicduodenal-jejunalbypasslineronobesityandtype2diabetesmellitus,amulticenter randomizedcontrolled trial. Ann Surg 2014;260:984–99228. Koehestanie P, Dogan K, Berends F, et al.Duodenal-jejunal bypass liner implantation pro-vokes rapid weight loss and improved glycemiccontrol, accompanied by elevated fasting ghrelinlevels. Endosc Int Open 2014;2:E21–E2729. Rodriguez L, Reyes E, Fagalde P, et al. Pilotclinicalstudyofanendoscopic, removableduodenal-jejunal bypass liner for the treatment of type 2diabetes.DiabetesTechnolTher2009;11:725–73230. Rohde U, Federspiel CA, Vilmann P, et al. Theimpact of EndoBarrier gastrointestinal liner inobese patientswith normal glucose tolerance andin patients with type 2 diabetes. Diabetes ObesMetab 2017;19:189–19931. Schouten R, Rijs CS, Bouvy ND, et al. Amulticenter, randomized efficacy study of theEndoBarrier Gastrointestinal Liner for presurgicalweight loss prior to bariatric surgery. Ann Surg2010;251:236–24332. Stratmann B, Krepak Y, Schiffer E, et al.Beneficial metabolic effects of duodenal jejunalbypass liner for the treatment of adipose patientswith type 2 diabetes mellitus: analysis of res-ponders and non-responders. Horm Metab Res2016;48:630–63733. TarnoffM,RodriguezL,EscalonaA,etal.Openlabel, prospective, randomized controlled trial ofan endoscopic duodenal-jejunal bypass sleeveversus low calorie diet for pre-operative weightloss in bariatric surgery. Surg Endosc 2009;23:650–65634. Vilarrasa N, de Gordejuela AGR, Casajoana A,et al. Endobarrier� in grade I obese patients withlong-standing type 2 diabetes: role of gastroin-testinal hormones in glucose metabolism. ObesSurg 2017;27:569–57735. Buchwald H, Oien DM. Metabolic/bariatricsurgery worldwide 2011. Obes Surg 2013;23:427–43636. Abu Dayyeh BK, Kumar N, Edmundowicz SA,et al.; ASGE Bariatric Endoscopy Task Force andASGE Technology Committee. ASGE Bariatric En-doscopy Task Force systematic review and meta-analysis assessing the ASGE PIVI thresholds foradopting endoscopic bariatric therapies. Gastro-intest Endosc 2015;82:425–38.e537. Jirapinyo P, Thompson CC. Endoscopic bar-iatric and metabolic therapies: surgical analogues

1114 Duodenal-Jejunal Bypass Liner and Glycemic Control Diabetes Care Volume 41, May 2018

and mechanisms of action. Clin Gastroenterol

Hepatol 2017;15:619–63038. Popov VB, OuA, Schulman AR, Thompson CC.

The impact of intragastric balloons on obesity-

related co-morbidities: a systematic review and

meta-analysis. Am J Gastroenterol 2017;112:429–

43939. Korner J, Inabnet W, Febres G, et al. Pro-

spective study of gut hormone and metabolic

changesafteradjustablegastricbandingandRoux-

en-Y gastric bypass. Int J Obes 2009;33:786–795

40. Tsoli M, Chronaiou A, Kehagias I, KalfarentzosF, Alexandrides TK. Hormone changes anddiabetes resolution after biliopancreatic diversionand laparoscopic sleeve gastrectomy: a compara-tiveprospective study.SurgObesRelatDis 2013;9:667–67741. Guidone C, Manco M, Valera-Mora E, et al.Mechanisms of recovery from type 2 diabetes aftermalabsorptive bariatric surgery. Diabetes 2006;55:2025–203142. Jacobsen SH, Olesen SC, Dirksen C, et al.Changes in gastrointestinal hormone responses,

insulin sensitivity, and beta-cell function within2 weeks after gastric bypass in non-diabeticsubjects. Obes Surg 2012;22:1084–109643. Svane MS, Bojsen-Møller KN, Nielsen S, et al.Effects of endogenous GLP-1 and GIP on glucosetolerance after Roux-en-Y gastric bypass surgery.Am J Physiol Endocrinol Metab 2016;310:E505–E51444. Zhao X, Han Q, Gang X, et al. The role ofgut hormones in diet-induced weight change:a systematic review. Horm Metab Res 2017;49:816–825

care.diabetesjournals.org Jirapinyo, Haas, and Thompson 1115