BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

Home , Gosogliptin

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay- per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected]

BMJ Open

Pharmacologic interventions for prevention of type 2 diabetes mellitus in people with prediabetes: a systematic review and network meta-analysis protocol ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-029073

Article Type: Protocol

Date Submitted by the 10-Jan-2019 Author:

Complete List of Authors: Zeng, Hai; The Second Clinical College of Guangzhou University of Chinese Medicine Wen, Junru; Shanghai University of Traditional Chinese Medicine He, Guoxin; The First College of Guangzhou University of Chinese Medicine Li, Zunjiang; The Second Clinical College of Guangzhou University of Chinese Medicine Jin, Yuelin; Shanghai University of Medicine & Health Sciences Fu, Wenbin; Guangdong Provincial Hospital of Chinese Medicine

diabetes, network meta-analysis, prevention, prediabetes, pharmacologic Keywords: interventions, systematic review

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 12 BMJ Open

1 2 3 4 Pharmacologic interventions for prevention of type 2 5 6 7 diabetes mellitus in people with prediabetes: 8 9 a systematic review and network meta-analysis protocol 10 11 Hai Zeng,1 Junru Wen,2,3 Guoxin He,4 Meng Luo,1 Zunjiang Li,1 Yueling Jin,3 Wenbin 12 Fu 1,5* 13 14 15 ABSTRACT 16 Introduction Type 2 diabetes mellitus (T2DM) is a substantial health problem worldwide. 17 18 Prediabetic state isFor associated peer with increased review risk for the development only of diabetes. There are 19 various pharmacologic therapies for diabetes prevention. Of those, most are being compared with 20 placebo instead of active agents. The relative effects and safety of different pharmacologic 21 interventions still remains uncertainty. To address this gap, we will conduct a systematic review 22 23 and network meta-analysis to evaluate comparative efficacy and safety of pharmacologic therapies 24 for T2DM prevention in patients with prediabetes to generate reliable evidence. 25 Methods and analysis PubMed, the Cochrane library, and EMBASE will be utilized to search for 26 27 relevant RCTs of pharmacologic therapies for diabetes prevention in participants with prediabetes 28 from inception until December 2018. Two reviewers working independently will screen titles, 29 abstracts, and full papers. Data extraction will also be completed by two independent authors. 30 31 Primary outcome will be incidence of T2DM in patients with prediabetes at baseline. Secondary 32 outcomes will include achievement of normoglycaemia, all-cause mortality, cardiovascular 33 mortality, and hypoglycaemic event. Pairwise meta-analysis and network meta-analysis will be 34 conducted for each outcome using a random-effects model within a frequentist approach. To 35 36 evaluate the robustness of our findings, subgroup analyses and sensitivity analyses will also be 37 performed. The comparison-adjusted funnel plot will be used to assess publication bias. The 38 overall quality of evidence of estimates will be rated with the recommendations assessment, 39 40 development and evaluation (GRADE) framework. Data analysis will be conducted using Stata 41 V.14.0. 42 Ethics and dissemination Ethics approval is not required. We plan to submit results of this study 43 44 to a peer-review journal. 45 PROSPERO registration number CRD42019119157. 46 47 Strengths and limitations of this study 48 49 ► This is a comprehensive network meta-analysis to evaluate the effectiveness and safety of 50 various pharmacologic therapies on diabetes prevention among people with prediabetic state. 51 ► Where possible, network meta-analysis will combine direct evidence with indirect evidence, 52 53 allowing comparisons of treatments without being compared to each other head-to-head in clinical 54 trials. 55 ► This research will generate clinically useful evidence to benefit patients, clinicians, and 56 57 guideline-makers. 58 ► The different frequencies, dosage, and routes of administration of pharmacological therapies 59 may lead to considerable heterogeneity. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 12

1 2 3 4 5 Keywords: diabetes, network meta-analysis, pharmacologic interventions, prediabetes, prevention, 6 protocol, systematic review 7 8 INTRODUCTION 9 10 Type 2 diabetes is a chronic and complex disease related to insulin secretory defects frequently on 11 the background of insulin resistance, and its progression is associated with genetic factors, 12 metabolic stress, and inflammation.1 The global prevalence of Type 2 diabetes mellitus (T2DM) 13 2 14 has reached alarming proportions with an estimated 463 million people in 2017. People with 15 T2DM are at elevated risk for chronic kidney disease, heart failure, atherosclerotic cardiovascular 16 disease, polyneuropathy, cognitive impairment, anxiety disorder, and depression.1,3 The term 17 18 prediabetes is usedFor to describe peer a blood glucose review level higher thanonly is considered normal but below 19 the cut-off value for T2DM.4 Different glycaemic measurements to define the prediabetic stage 20 exist, including impaired fasting glucose (IFG), impaired glucose tolerance (IGT), and elevated 21 glycosylated haemoglobin A1c (HbA1c).1 According to standards of medical care in diabetes of 22 23 American Diabetes Association (ADA), prediabetes is defined as a FBG of at least 5.6 mmol/L 24 but lower than 6.9 mmol/L, a HbA1c of 5.7-6.4%, or IGT (a 2-hour plasma glucose value of 25 7.8-11.1 mmol/L during oral glucose tolerance test).1 These measurements are considered to 26 27 predict a different risk spectrum for progression of prediabetes to diabetes. The International 28 Diabetes Federation (IDF) estimated that, in 2017, approximately 352 million persons globally 29 had IGT, which is projected to exceed half a billion people before 2045.5 Dysglycaemia is a well 30 31 described risk factor for all-cause mortality, total numbers of all-age deaths attributed to high 32 fasting plasma were 6 million people in 2017,6 with type 2 diabetes accounting for 1 million 33 deaths.7 Moreover, the economic burden of diabetes is large, economic costs of which increased to 34 126% from 2012 to 2017 in the US, with a total estimated cost of $327 billion in 2017.8,9 Diabetic 35 36 patients incurred average medical expenditures of $16,750 yearly, with diabetes accounting for 37 $9,600.9 Thus, there is an urgent need to address huge burden of this worldwide disease with a 38 growing number of suffers. Early interventions for preventing type 2 diabetes are warranted.8 39 40 Persons diagnosed with prediabetes are thought to be at increased risk for developing T2DM, the 41 approximated incidence rate of diabetes among people defined as “prediabetic stage” by 42 measurements of IFG, IGT, or HbA1c in the following 10 years is more than one third.10 These 43 44 people are ideal candidates for T2DM prevention efforts. To prevent progression of prediabetes to 45 type 2 diabetes, an intensive behavioral lifestyle intervention program is recommended, including 46 individualized medical nutrition therapy, physical activity, and no tobacco use.1 However, to date, 47 whether any pharmacologic intervention should be recommended for persons with prediabetes or 48 49 not has not yet to be clarified clearly.11 Importantly, in recent years, an increasing number of 50 clinical trials have investigated several groups of pharmacological therapies for T2DM prevention, 51 including insulin secretagogues, glucagon-like peptide (GLP)-1 analogues, alpha-glucosidase 52 53 inhibitors, dipeptidyl-peptidase (DPP)-4 inhibitors, biguanides, and thiazolidinediones. Some 54 findings suggest that using these pharmacological agents could reduce or delay the progression to 55 T2DM. Nevertheless, head-to-head comparisons of different pharmacologic therapies have rarely 56 57 been performed by previous clinical trials. Evidence regarding the overall and comparative 58 efficacy of these pharmacological interventions for T2DM prevention is limited, while important 59 for clinical decision-making. Conventional pairwise meta-analyses are limited to pool the results 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 12 BMJ Open

1 2 3 of trials comparing two interventions directly while a network meta-analysis (NMA) method is 4 5 able to combine direct and indirect evidence and assess comparative efficacy and safety of various 6 interventions.12-14 Therefore, to bridge this knowledge gap, we plan to conduct a network 7 meta-analysis to assess comparative effectiveness and safety of several medications for preventing 8 T2DM in participants with prediabetes, which may provide beneficial information for clinical 9 10 decision-making and further clinical trials. 11 12 METHODS 13 14 Study design and registration 15 This systematic review protocol is reported in line with the Preferred Reporting Items for 16 Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines.15,16 This study will be 17 12 18 performed in accordanceFor with peer PRISMA extension review statements foronly network meta-analysis. 19 20 Eligibility criteria 21 Population 22 23 Adults (older than 18 years) who have prediabetes will be eligible for inclusion. In this study, 24 prediabetic state involves separate impaired fasting glucose (IFG), separate impaired glucose 25 tolerance (IGT), separate elevated glycosylated haemoglobin A1c (HbA1c) or combinations 26 27 thereof. Diagnostic criteria for prediabetes should be established and described in eligible trials. 28 29 Intervention and comparator 30 31 This study will investigate comparisons of pharmacological interventions versus another active 32 agent, lifestyle interventions (diet, exercise, or both), placebo or no intervention. Pharmacological 33 therapies include alpha-glucosidase inhibitors, sulphonylureas, meglitinide analogues, 34 dipeptidyl-peptidase (DPP)-4 inhibitors, glucagon-like peptide (GLP)-1 analogues, biguanides, 35 36 thiazolidinediones, alone or in combination. 37 38 Outcomes 39 40 Primary outcome will be incidence of T2DM in patients with prediabetes at baseline. Secondary 41 outcomes will include achievement of normoglycaemia, all-cause mortality, cardiovascular 42 mortality, and hypoglycaemic event. Classification and definition of T2DM could be based on any 43 44 recognised standard diagnosis criteria (ie, the American Diabetes Association guidelines). 45 46 Type of studies 47 All randomised controlled trials (RCTs) comparing pharmacological agents with other drugs, 48 49 lifestyle interventions, placebo or no intervention for T2DM prevention in patients with 50 prediabetes will be included in this study. Duration of intervention has to be with a minimum 51 duration of 12 weeks. 52 53 54 Search strategy 55 Various databases will be utilized to search for RCTs of pharmacologic therapies for preventing 56 57 diabetes among patients with prediabetes from inception date of databases until December 2018. 58 The databases will include PubMed, Embase, and the Cochrane Library. In addition, the language 59 of publication will be limited to English. Any potentially-relevant article will be retrieved for 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 12

1 2 3 review. Details of search strategy of PubMed database is shown in supplemental material. The 4 5 literature search will be conducted using the following keywords: alpha-glucosidase inhibitors, 6 sulphonylureas, glinide, dipeptidyl-peptidase (DPP)-4 inhibitors, glucagon-like peptide (GLP)-1 7 analogues, biguanides, thiazolidinediones, diabetes, T2DM, prediabetes, prediabetic state, glucose 8 intolerance, impaired glucose, diabetic, dysglycaemia, hyperglycaemia, conversion, delay, and 9 10 prevent. Moreover, all drug names in each drug class will be included in key search terms, for 11 instance, acarbose, voglibose, metformin, alogliptin, saxagliptin, liraglutide, and albiglutide. To 12 identify other eligible studies, reference lists of relevant publications (including trials, reviews, 13 14 and meta-analyses) will be reviewed for a manual search. 15 16 Selection of studies 17 18 In accordance withFor the prespecified peer inclusion review criteria, two reviewers only working independently will 19 evaluate all titles and abstracts to eliminate papers deemed irrelevant. The remaining articles will 20 be included in the further assessment. Reviewers will scrutinize full text for each 21 potentially-relevant article. The study identification and exclusion process will be depicted using 22 23 the PRISMA flow diagram. Discrepancies in study selection will be resolved by negotiation. 24 25 Data collection process 26 27 Two independent reviewers will use a standardized data form to extract trial information. All 28 disagreements will be settled via discussion with the third reviewer. The data extracted will be as 29 follows: 30 31 ► Patient characteristics (age, gender, race, weight and glycemic parameters). 32 ► trial characteristics (author, publication year, study design, country setting, and funding 33 information). 34 ► 35 Details of intervention and control (dosage, frequency, and treatment duration). 36 ►Outcome data for all endpoints of interest. 37 38 Assessment of methodological quality 39 40 The Cochrane risk of bias assessment tool will be used to assess risk of bias for individual studies. 41 The method includes following domains: random sequence generation, allocation concealment, 42 blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, 43 17 44 selective reporting, and other bias. Each item will be classified into one of three categories as 45 follows: unclear, high or low risk. All discrepancies in quality assessment will be resolved after 46 mutual agreement and discussion. 47 48 49 Data synthesis and statistical analysis 50 Initially, we will use a random-effects approach to pool effect estimates for all treatment 51 comparisons in conventional pairwise meta-analyses. For categorical outcomes, the pooled 52 53 estimates as risk ratios (RRs) with 95% confidence intervals (CIs) will be reported. Continuous 54 data will be reported as mean differences (MDs) with their respective 95% CIs. Statistical 55 heterogeneity across trials will be examined using the I2 statistic. An I2 statistic of 75%, 50%, or 56 18 57 25% indicates high, moderate, or low heterogeneity, separately. Then, network meta-analyses 58 will be carried out in a frequentist environment. Local inconsistency between direct and indirect 59 evidence within each closed loop will be assessed using a node-splitting test.19,20 In addition, a 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 12 BMJ Open

1 2 3 “design-by-treatment” model will be applied to evaluate the assumption of consistency in the 4 19 5 whole network. We will generate the surface under the cumulative ranking curve (SUCRA) to 6 assess probabilities of interventions in superiority regarding efficacy and safety outcomes, with 7 higher SUCRA values indicating better effects or safety.21 The level of significance will be set at 8 an alpha of 0.05. All analyses will be performed with Stata version 14.0 (StataCorp, College 9 10 Station, TX). 11 Sensitivity and subgroup analyses 12 We will perform additional sensitivity analyses. Where possible, analyses will be stratified by age 13 2 2 14 (18-64 years and at least 65 years), gender, ethnicity, and BMI (25-29.9 kg/m and ≥30 kg/m ). 15 Moreover, we will also perform subgroup analyses according to diagnostic criteria (IFG, IGT, and 16 HbA1c). 17 18 For peer review only 19 Publication bias 20 We will employ the comparison-adjusted funnel plot to assess small study effects including 21 publication bias at the network level.22 22 23 24 Quality of evidence 25 The quality of evidence of estimates derived from NMA will be rated using the recommendations 26 27 assessment, development and evaluation (GRADE) framework. The GRADE approach 28 characterises the quality of evidence according to publication bias, study limitations, inconsistency, 29 imprecision, and indirectness.23 Evidence of efficacy outcomes will be rated from high quality to 30 31 very low quality. 32 33 Patient and public involvement 34 No patients or public will participate in the study. 35 36 37 Ethics and dissemination 38 Since confidential patient data will not be involved in this study, formal ethics approval is not 39 40 required. The framework of the PRISMA statements for NMA will be applied to guide review 41 authors to perform this study. The results will be disseminated by a peer-reviewed publication. 42 43 44 45 DISCUSSION 46 This study is a comprehensive NMA comparing and ranking a variety of pharmacological 47 interventions for preventing T2DM in patients at high risk for the development of T2DM. Our 48 49 study will provide a summary of the best available evidence concerning pharmacological therapies 50 for T2DM prevention in patients with prediabetic state, benefitting for clinicians, 51 guideline-makers, and policy-makers to generate higher quality recommendations for these 52 24 53 patients. Although a relevant NMA published, the study was based on clinical trials before 2014. 54 Additionally, included pharmacological interventions in the study were limited, 55 dipeptidyl-peptidase (DPP)-4 inhibitors, glucagon-like peptide (GLP)-1 analogues, and some other 56 57 glucose-lowering drugs that have been tested by later trials clinically were not involved. It is 58 essential to contain these commonly prescribed agents in multiple comparisons of medications for 59 the prevention of T2DM. Moreover, the definition of adults at high risk for T2DM was based on 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 12

1 2 3 IFG and IGT, excluding people identified by HbA1c. Importantly, HbA1c is a biomarker of 4 5 long-term glycemic control when compared IFG and IGT, representing average blood glucose 6 levels during the preceding two to three months.25 Several strengths can be foreseen of this review, 7 but our network meta-analysis may have some possible limitations. The different frequencies, 8 dosage, and routes of administration of pharmacological therapies may result in considerable 9 10 heterogeneity. Differences in inclusion criteria of participants and definition of end-point events 11 may influence the quality of evidence. 12 13 14 Author affiliations 15 1 The Second Clinical College of Guangzhou University of Chinese Medicine, Guangzhou, China 16 2 Shanghai University of Traditional Chinese Medicine, Shanghai, China 17 3 18 Shanghai UniversityFor of Medicine peer & Health reviewSciences, Shanghai, Chinaonly 19 4 The First College of Guangzhou University of Chinese Medicine, Guangzhou, China 20 5 Department of Acupuncture and Moxibustion, Guangdong Provincial Hospital of Chinese Medicine, 21 Guangzhou, China 22 23 24 Correspondence to 25 Professor Wenbin Fu; 26 27 [email protected] 28 29 Contributors HZ conceived the review. JRW, GXH, and HZ wrote the first draft of this protocol. 30 31 WBF, ZJL, ML, and YLJ were responsible for revising the draft. HZ, ML and JRW contributed to 32 developing the search strategy and registering the protocol. WBF and YLJ were the guarantors. All 33 authors scrutinized and approved the final manuscript. 34 35 36 Funding No funding. 37 Competing interests None declared. 38 Patient consent Not required. 39 40 Provenance and peer review Not commissioned; externally peer review. 41 42 43 44 REFERENCES 45 46 1. American Diabetes Association. Classification and Diagnosis of Diabetes: Standards of 47 48 Medical Care in Diabetes-2019. Diabetes care 2019;42:S13-28. 49 2. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and 50 national incidence, prevalence, and years lived with disability for 354 diseases and injuries 51 for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of 52 53 Disease Study 2017. Lancet 2018;392:1789-858. 54 3. Callaghan BC, Xia R, Reynolds E, et al. Association Between Metabolic Syndrome Components 55 and Polyneuropathy in an Obese Population. JAMA neurology 2016;73:1468-1476. 56 57 4. Moelands SV, Lucassen PL, Akkermans RP, De Grauw WJ, Van de Laar FA. Alpha-glucosidase 58 inhibitors for prevention or delay of type 2 diabetes mellitus and its associated complications 59 in people at increased risk of developing type 2 diabetes mellitus. The Cochrane database of 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 12 BMJ Open

1 2 3 systematic reviews 2018;12:Cd005061. 4 5 5. International Diabetes Federation. IDF Diabetes Atlas. 8th Edition. Brussels, Belgium: 6 International Diabetes Federation, 2017. 7 6. GBD 2017 Risk Factor Collaborators. Global, regional, and national comparative risk 8 9 assessment of 84 behavioural, environmental and occupational, and metabolic risks or 10 clusters of risks for 195 countries and territories, 1990-2017: a systematic analysis for the 11 Global Burden of Disease Study 2017. Lancet 2018;392:1923-94. 12 7. GBD 2017 Causes of Death Collaborators. Global, regional, and national age-sex-specific 13 14 mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic 15 analysis for the Global Burden of Disease Study 2017. Lancet 2018;392:1736-88. 16 8. Shrestha SS, Honeycutt AA, Yang W, et al. Economic Costs Attributable to Diabetes in Each 17 18 U.S. State.For Diabetes carepeer 2018;41:2526-34. review only 19 9. Economic Costs of Diabetes in the U.S. in 2017. Diabetes care 2018;41:917-28. 20 10. Morris DH, Khunti K, Achana F, et al. Progression rates from HbA1c 6.0-6.4% and other 21 22 prediabetes definitions to type 2 diabetes: a meta-analysis. Diabetologia 2013;56:1489-93. 23 11. Yudkin JS, Montori VM. The epidemic of pre-diabetes: the medicine and the politics. BMJ 24 (Clinical research ed) 2014;349:g4485. 25 12. Hutton B, Salanti G, Caldwell DM, et al. The PRISMA extension statement for reporting of 26 27 systematic reviews incorporating network meta-analyses of health care interventions: 28 checklist and explanations. Annals of internal medicine 2015;162(11):777-84. 29 13. Rouse B, Cipriani A, Shi Q, Coleman AL, Dickersin K, Li T. Network Meta-analysis for Clinical 30 31 Practice Guidelines: A Case Study on First-Line Medical Therapies for Primary Open-Angle 32 Glaucoma. Annals of internal medicine 2016;164(10):674-82. 33 14. Mills EJ, Thorlund K, Ioannidis JP. Demystifying trial networks and network meta-analysis. 34 35 BMJ (Clinical research ed) 2013;346:f2914. 36 15. Moher D, Shamseer L, Clarke M, et al. Preferred reporting items for systematic review and 37 meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews 2015;4:1. 38 16. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and 39 40 meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ (Clinical 41 research ed) 2015;350:g7647. 42 17. Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration's tool for assessing risk 43 44 of bias in randomised trials. BMJ (Clinical research ed) 2011;343:d5928. 45 18. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. 46 BMJ (Clinical research ed) 2003;327:557-60. 47 48 19. Higgins JP, Jackson D, Barrett JK, Lu G, Ades AE, White IR. Consistency and inconsistency in 49 network meta-analysis: concepts and models for multi-arm studies. Research synthesis 50 methods 2012;3:98-110. 51 20. Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in mixed treatment 52 53 comparison meta-analysis. Statistics in medicine 2010;29:932-44. 54 21. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical summaries for presenting 55 results from multiple-treatment meta-analysis: an overview and tutorial. Journal of clinical 56 57 epidemiology 2011;64:163-71. 58 22. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network 59 meta-analysis in STATA. PLoS One. 2013;8(10):e76654. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 12

1 2 3 23. Puhan MA, Schunemann HJ, Murad MH, et al. A GRADE Working Group approach for rating 4 5 the quality of treatment effect estimates from network meta-analysis. BMJ (Clinical research 6 ed). 2014;349:g5630. 7 24. Stevens JW, Khunti K, Harvey R, et al. Preventing the progression to type 2 diabetes mellitus 8 9 in adults at high risk: a systematic review and network meta-analysis of lifestyle, 10 pharmacological and surgical interventions. Diabetes research and clinical practice. 11 2015;107(3):320-331. 12 25. Inzucchi SE. Clinical practice. Diagnosis of diabetes. The New England journal of medicine. 13 14 2012;367(6):542-550. 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 12 BMJ Open

1 2 3 4 5 6 Search strategy in PubMed. 7 Block 1: Prediabetes 8 9 #1 “Prediabetic state” [Mesh] 10 #2 “Glucose Intolerance” [Mesh] 11 #3 (prediabet* OR pre diabet*) [tiab] 12 #4 intermediate hyperglyc?emi* [tiab] 13 14 #5 ((impaired fasting adj2 glucose) OR IFG or impaired FPG) [tiab] 15 #6 ((impaired glucose adj (tolerance OR metabolism)) OR IGT) [tiab] 16 #7 (“dysglycaemia” OR “hyperglycaemia”) [tiab] 17 18 #8 ((risk OR progress*For ORpeer prevent* review OR inciden* OR only conversion OR develop* OR 19 delay*) adj4 (diabetes OR T2D* OR NIDDM OR “type 2” OR “type II”)) [tiab] 20 #9 OR #1-8 21 22 23 Block 2: Alpha-glucosidase inhibitors 24 #10 “Acarbose” [Mesh] 25 #11 (acarbos* OR glucobay OR precose OR prandase OR “bay g 5421” OR 26 27 BAYG5421) [tiab] 28 #12 (voglibos* OR glustat OR basen) [tiab] 29 #13 (miglitol* OR glyset) [tiab] 30 31 #14 (glucosidase* adj3 inhibitor*) [tiab] 32 #15 OR #11-14 33 34 35 Block 3: DPP-4 inhibitors 36 16# “Dipeptidyl-Peptidase IV Inhibitors” [Mesh] 37 17# “gliptin*”[tiab] 38 18# ((dipeptidyl peptidase or dipeptidylpeptidase or dpp) adj (“4” or IV) adj inhibitor?) 39 40 [tiab] 41 19# (alogliptin OR anagliptin OR bisegliptin OR carmegliptin OR denagliptin OR 42 dutogliptin OR evogliptin OR gemigliptin OR gosogliptin OR linagliptin OR 43 44 melogliptin OR omarigliptin OR sitagliptin OR saxagliptin OR teneligliptin OR 45 trelagliptin OR vildagliptin) [tiab] 46 #20 OR #16-19 47 48 49 Block 4: GLP-1 analogue 50 #21 “Glucagon-Like Peptide 1” [Mesh] 51 #22 ((glucagon like peptide* or GLP 1 or GLP1) adj3 (analog* or agonist*))[tiab] 52 53 #23 (exenatide OR liraglutide OR albiglutide OR elsiglutide OR lixisenatide OR 54 dulaglutide OR taspoglutide OR semaglutide OR teduglutide) [tiab] 55 #24 OR #21-23 56 57 58 Block 5: Sulfonylurea 59 #25 “Sulfonylurea Compounds” [Mesh] 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 12

1 2 3 #26 (sulfon?lurea* OR sulphon?lurea*)[tiab] 4 5 #27 (gl?benclamid* OR glyburid* )[tiab] 6 #28 (gl?bornurid* OR gluborid*)[tiab] 7 #29 (glipizid* OR gl?diazinamide OR glypidizine OR melizide OR napizide)[tiab] 8 9 #30 (gliquidon* OR glisoxepid* OR gl?clopyramid* OR glimepirid* OR gl?clazid* 10 OR gl?cazid*)[tiab] 11 #31 OR #25-30 12 13 14 Block 6: Glinide 15 #32 (glinide OR glinides) [tiab] 16 #33 (nateglinid* or senaglinid* OR repaglinid* OR mitiglinid*) [tiab] 17 18 #34 OR #32-33For peer review only 19 20 Block 7: Biguanides 21 22 #35 “Biguanides” [Mesh] 23 #36 (buformi* OR chloroguani* OR metformi* OR phenformi*) [tiab] 24 #37 OR #35-36 25 26 27 Block 8: Thiazolidinediones 28 #38 “Thiazolidinediones” [Mesh] 29 #39 (thiazolidinedione OR TZD OR glitazone OR glitazones) [tiab] 30 31 #40 (pioglit* OR rosiglit* OR troglit*) [tiab] 32 #41 OR #38-40 33 34 35 Block 9: RCT-filter 36 #42 "randomized controlled trial"[pt] OR "controlled clinical trial"[pt] OR 37 randomized[tiab] OR placebo[tiab] OR "drug therapy"[sh] OR randomly[tiab] OR 38 trial[tiab] OR groups[tiab] 39 40 41 #43 42 #9 AND #15 AND #20 AND #24 AND #31 AND #34 AND #37 AND #41 AND #42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 12 BMJ Open

1 2 3 PRISMA-P (Preferred Reporting Items for Systematic review and Meta-Analysis Protocols) 2015 checklist: recommended items to 4 address in a systematic review protocol* 5 6 Section and topic Item Checklist item Reported on 7 No Page # 8 9 ADMINISTRATIVE INFORMATION 10 Title: 11 Identification 1a Identify the report as a protocol of a systematic review 1 12 Update 1b If the protocolFor is for an update peer of a previous systematic review review, identify as such only 13 Registration 2 If registered, provide the name of the registry (such as PROSPERO) and registration number 1 14 15 Authors: 16 Contact 3a Provide name, institutional affiliation, e-mail address of all protocol authors; provide physical mailing address of corresponding 6 17 author 18 Contributions 3b Describe contributions of protocol authors and identify the guarantor of the review 6 19 Amendments 4 If the protocol represents an amendment of a previously completed or published protocol, identify as such and list changes; 20 otherwise, state plan for documenting important protocol amendments 21 Support: 22 Sources 5a Indicate sources of financial or other support for the review 6 23 24 Sponsor 5b Provide name for the review funder and/or sponsor 6 25 Role of sponsor 5c Describe roles of funder(s), sponsor(s), and/or institution(s), if any, in developing the protocol 6 26 or funder 27 INTRODUCTION 28 29 Rationale 6 Describe the rationale for the review in the context of what is already known 2 30 Objectives 7 Provide an explicit statement of the question(s) the review will address with reference to participants, interventions, 2, 3 31 comparators, and outcomes (PICO) 32 33 METHODS 34 Eligibility criteria 8 Specify the study characteristics (such as PICO, study design, setting, time frame) and report characteristics (such as years 3 35 considered, language, publication status) to be used as criteria for eligibility for the review 36 Information sources 9 Describe all intended information sources (such as electronic databases, contact with study authors, trial registers or other grey 3, 4 37 literature sources) with planned dates of coverage 38 Search strategy 10 Present draft of search strategy to be used for at least one electronic database, including planned limits, such that it could be 3, 4 39 repeated 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 12 of 12

1 2 3 Study records: 4 Data 11a Describe the mechanism(s) that will be used to manage records and data throughout the review 4 5 management 6 Selection 11b State the process that will be used for selecting studies (such as two independent reviewers) through each phase of the review 4 7 process (that is, screening, eligibility and inclusion in meta-analysis) 8 9 Data collection 11c Describe planned method of extracting data from reports (such as piloting forms, done independently, in duplicate), any 4 process processes for obtaining and confirming data from investigators 10 11 Data items 12 List and define all variables for which data will be sought (such as PICO items, funding sources), any pre-planned data 4 12 assumptions andFor simplifications peer review only 13 Outcomes and 13 List and define all outcomes for which data will be sought, including prioritization of main and additional outcomes, with 3 14 prioritization rationale 15 Risk of bias in 14 Describe anticipated methods for assessing risk of bias of individual studies, including whether this will be done at the outcome 4 16 individual studies or study level, or both; state how this information will be used in data synthesis 17 Data synthesis 15a Describe criteria under which study data will be quantitatively synthesised 4 18 15b If data are appropriate for quantitative synthesis, describe planned summary measures, methods of handling data and methods of 4 19 combining data from studies, including any planned exploration of consistency (such as I2, Kendall’s τ) 20 15c Describe any proposed additional analyses (such as sensitivity or subgroup analyses, meta-regression) 5 21 15d If quantitative synthesis is not appropriate, describe the type of summary planned 4, 5 22 23 Meta-bias(es) 16 Specify any planned assessment of meta-bias(es) (such as publication bias across studies, selective reporting within studies) 5 24 Confidence in 17 Describe how the strength of the body of evidence will be assessed (such as GRADE) 5 25 cumulative evidence 26 * It is strongly recommended that this checklist be read in conjunction with the PRISMA-P Explanation and Elaboration (cite when available) for important 27 clarification on the items. Amendments to a review protocol should be tracked and dated. The copyright for PRISMA-P (including checklist) is held by the 28 PRISMA-P Group and is distributed under a Creative Commons Attribution Licence 4.0. 29 30 From: Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, Shekelle P, Stewart L, PRISMA-P Group. Preferred reporting items for systematic review and 31 meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015 Jan 2;349(jan02 1):g7647. 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open

Anti-diabetic agents for prevention of type 2 diabetes mellitus in people with prediabetes: a systematic review and network meta-analysis protocol ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-029073.R1

Article Type: Protocol

Date Submitted by the 21-Jun-2019 Author:

Complete List of Authors: Zeng, Hai; The Second Clinical College of Guangzhou University of Chinese Medicine Wang, Xianzhe ; The Second Clinical College of Guangzhou University of Chinese Medicine Zhang, Zexin ; The Second Clinical College of Guangzhou University of Chinese Medicine Wen, Junru; Shanghai University of Traditional Chinese Medicine He, Guoxin; The First College of Guangzhou University of Chinese Medicine Li, Zunjiang; The Second Clinical College of Guangzhou University of Chinese Medicine Luo, Meng; The Second Clinical College of Guangzhou University of Chinese Medicine Jin, Yuelin; Shanghai University of Medicine & Health Sciences Zhou, Peng; Shenzhen Bao'an Traditional Chinese Medicine Hospital Group Fu, Wenbin; Guangdong Provincial Hospital of Chinese Medicine

Primary Subject Diabetes and endocrinology Heading:

Secondary Subject Heading: Diabetes and endocrinology

diabetes, network meta-analysis, prevention, prediabetes, anti-diabetic Keywords: agents, systematic review

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 12 BMJ Open

1 2 3 4 Anti-diabetic agents for prevention of type 2 diabetes 5 6 7 mellitus in people with prediabetes: 8 9 a systematic review and network meta-analysis protocol 10 11 Hai Zeng,1 Xianzhe Wang,1 Zexin Zhang,1 Junru Wen,2,3* Guoxin He,4 Zunjiang Li,1 12 Meng Luo,1 Yueling Jin,3 Peng Zhou,5 Wenbin Fu6,7,8* 13 14 15 Correspondence to 16 Professor Wenbin Fu; [email protected]; Junru Wen; [email protected] 17 18 For peer review only 19 ABSTRACT 20 Introduction Type 2 diabetes mellitus (T2DM) is a substantial health problem worldwide. 21 Prediabetic state is associated with increased risk for the development of diabetes. There are 22 23 various pharmacologic therapies with glucose-lowering activity for diabetes prevention. Of those, 24 most are being compared with placebo instead of active agents. The relative effects and safety of 25 different glucose-lowering drugs still remain uncertain. To address this gap, we will conduct a 26 27 systematic review and network meta-analysis (NMA) to evaluate comparative efficacy and safety 28 of glucose-lowering agents for T2DM prevention in patients with prediabetes. 29 Methods and analysis PubMed, the Cochrane library, and Embase will be searched from 30 31 inception to December 2019 for relevant randomized controlled trials (RCTs) that examined 32 anti-diabetic drugs for diabetes prevention in patients with prediabetes. Two reviewers working 33 independently will screen titles, abstracts, and full papers. Data extraction will also be completed 34 by two independent authors. The primary outcome will be the incidence of T2DM in patients with 35 36 prediabetes at baseline. Secondary outcomes will include the achievement of normoglycemia, 37 all-cause mortality, cardiovascular mortality, and hypoglycemic event. Pairwise meta-analysis and 38 NMA will be conducted for each outcome using a frequentist random-effects model. Additionally, 39 40 subgroup analyses will also be performed. The comparison-adjusted funnel plot will be used to 41 assess publication bias. The overall quality of evidence will be rated with the recommendations 42 assessment, development and evaluation (GRADE) framework. Data analysis will be conducted 43 44 using Stata V.14.0. 45 Ethics and dissemination Ethics approval is not required. We plan to submit the results of this 46 study to a peer-review journal. 47 PROSPERO registration number CRD42019119157. 48 49 50 Strengths and limitations of this study 51 ► This is a comprehensive systematic review and network meta-analysis to evaluate the 52 53 effectiveness and safety of various glucose-lowering medications on diabetes prevention among 54 people with prediabetic state. 55 ► Where possible, a NMA will combine direct evidence with indirect evidence, allowing 56 57 comparisons of treatments without being compared to each other head-to-head in clinical trials. 58 ► This research will generate clinically useful evidence to benefit patients, clinicians, and 59 guideline-makers. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 12

1 2 3 ► The different frequencies, dosages, and routes of administration of pharmacological therapies 4 5 may lead to considerable heterogeneity. 6 7 Keywords: anti-diabetic agents, diabetes, network meta-analysis, prediabetes, prevention, 8 systematic review 9 10 11 INTRODUCTION 12 Type 2 diabetes mellitus (T2DM) is a chronic and complex disease, related to insulin secretory 13 14 defects frequently on the background of insulin resistance; the progression of the disease is 15 associated with genetic factors, metabolic stress, and inflammation.1 The global prevalence of 16 T2DM was estimated to be 463 million people in 2017.2 People with T2DM are at elevated risk 17 18 for chronic kidneyFor disease, peerheart failure, reviewatherosclerotic cardiovascular only disease, polyneuropathy, 19 cognitive impairment, anxiety disorder, and depression.3-5 The term prediabetes is used to describe 20 a blood glucose level higher than the normal range but below the cut-off value for T2DM.6 21 Different glycemic measurements to define the prediabetic stage exist, including impaired fasting 22 23 glucose (IFG), impaired glucose tolerance (IGT), and elevated glycosylated haemoglobin A1c 24 (HbA1c).1 According to the standards of medical care in diabetes of the American Diabetes 25 Association (ADA), prediabetes is defined as a fasting plasma glucose of at least 5.6 mmol/L but 26 27 lower than 6.9 mmol/L, a HbA1c of 5.7-6.4%, or IGT (a 2-hour plasma glucose value of 7.8-11.1 28 mmol/L during oral glucose tolerance test).1 These measurements are considered to predict a 29 different risk spectrum for the progression of prediabetes to diabetes. The International Diabetes 30 31 Federation (IDF) estimated that, in 2017, approximately 352 million persons globally had IGT, 32 which is projected to exceed half a billion people before 2045.7 Hyperglycemia is a well described 33 risk factor for all-cause mortality, total number of all-age deaths attributable to high fasting 34 plasma was 6.5 million people in 2017,8 with T2DM accounting for 1 million deaths.9 Moreover, 35 36 the economic burden of diabetes is large; in 2017, the ADA estimated the total economic costs 37 attributable to diabetes in the U.S. to be $327 billion.10,11 Diabetic patients incurred average 38 medical expenditures of $16,750 yearly, with diabetes accounting for $9,600.11 Thus, there is an 39 40 urgent need to address huge burden of this worldwide disease with a growing number of suffers. 41 Early interventions for preventing type 2 diabetes are warranted.10 Persons diagnosed with 42 prediabetes are thought to be at increased risk for developing T2DM, the estimated incidence rate 43 44 of diabetes among people defined as “prediabetic stage” by measurements of IFG, IGT, or HbA1c 45 in the following 10 years exceeds one-third.12 These people are ideal candidates for diabetes 46 prevention efforts. 47 To prevent the progression of prediabetes to T2DM, an intensive behavioral lifestyle 48 49 intervention program is recommended in the ADA guidelines, including individualized medical 50 nutrition therapy, physical activity, and no tobacco use.13 Besides lifestyle modification, a variety 51 of anti-diabetic agents have been investigated in clinical trials for diabetes prevention, including 52 53 insulin secretagogues, glucagon-like peptide (GLP)-1 analogues, alpha-glucosidase inhibitors, 54 dipeptidyl-peptidase (DPP)-4 inhibitors, biguanides, and thiazolidinediones. These pharmacologic 55 approaches with intrinsic glucose-lowering activity (e.g., improve the insulin resistance and 56 57 preserve pancreatic β -cell function) are recommended for glycemic treatment in patients with 58 T2DM in the ADA guidelines.14 Of these pharmacologic medications, only metformin therapy for 59 diabetes prevention is recommended as an option for patients with prediabetes.13 However, to date, 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 12 BMJ Open

1 2 3 whether other glucose-lowering agents should be considered in those patients or not has not yet to 4 5 be clarified clearly, even though some findings of recent studies have demonstrated that these 6 pharmacological agents could also exert benefits to prevent or delay the progression to T2DM. In 7 addition, head-to-head comparisons of different anti-diabetic agents have rarely been performed 8 by previous clinical trials. Evidence regarding the overall and comparative efficacy of these 9 10 anti-hyperglycemia agents for T2DM prevention is limited, while it is important for clinical 11 decision-making. Conventional pairwise meta-analyses are limited to pool the results of trials 12 comparing two interventions directly while a network meta-analysis (NMA) method is able to 13 14 combine direct and indirect evidence and assess comparative efficacy and safety of various 15 interventions.15-17 Therefore, to bridge this knowledge gap, we plan to conduct the systematic 16 review and NMA to assess comparative effects and safety of various anti-diabetic medications in 17 18 preventing T2DMFor in patients peer with prediabetes, review which may only provide beneficial information for 19 clinical decision-making and further clinical trials. 20 21 METHODS 22 23 Study design and registration 24 This systematic review protocol is reported in line with the Preferred Reporting Items for 25 Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines.18,19 This study will be 26 15 27 performed in accordance with the PRISMA extension statements for NMA. 28 29 Eligibility criteria 30 31 Population 32 Adults (older than 18 years) who have prediabetes will be eligible for inclusion. In this study, 33 prediabetic state involves separate IFG, separate IGT, separate elevated HbA1c or combinations 34 thereof. Diagnostic criteria for prediabetes should be established and described in eligible trials. 35 36 37 Intervention and comparator 38 This study will investigate comparisons of anti-diabetic drugs versus another anti-diabetic agent, 39 40 lifestyle interventions (diet, exercise, or both), placebo or no intervention. Anti-diabetic agents 41 include alpha-glucosidase inhibitors (e.g., acarbose and voglibose), sulphonylureas (e.g., glipizide 42 and glimepiride), meglitinide analogues (e.g., nateglinide), dipeptidyl-peptidase (DPP)-4 43 44 inhibitors (e.g., linagliptin and vildagliptin), glucagon-like peptide (GLP)-1 analogues (e.g., 45 exenatide and liraglutide), biguanides (e.g., metformin), thiazolidinediones (e.g., rosiglitazone and 46 pioglitazone), alone or in combination. In addition, studies using vitamins, traditional Chinese 47 medicines, or alternative/herbal supplements will be excluded. 48 49 50 Outcomes 51 The primary outcome will be the incidence of T2DM in patients with prediabetes at baseline. 52 53 Secondary outcomes will include the achievement of normoglycemia, all-cause mortality, 54 cardiovascular mortality, and hypoglycemic event. Classification and definition of T2DM could 55 be based on any recognized standard diagnosis criteria (e.g., the ADA guidelines). 56 57 58 Type of studies 59 All randomized controlled trials (RCTs) comparing anti-diabetic drugs with another anti-diabetic 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 12

1 2 3 agent, lifestyle interventions, placebo or no intervention for T2DM prevention in patients with 4 5 prediabetes will be included in this study. Duration of intervention has to be with a minimum of 6 12 weeks. 7 8 Search strategy 9 10 Several databases will be searched from inception to December 2019 for RCTs that investigated 11 anti-diabetic agents for prevention of diabetes among patients with prediabetes. The databases will 12 include PubMed, Embase, and the Cochrane Library. In addition, the language of publication will 13 14 be limited to English. Any potentially-relevant article will be retrieved for review. Details of 15 search strategy of PubMed database are shown in the supplemental material. The literature search 16 will be conducted using the following keywords: alpha-glucosidase inhibitors, sulphonylureas, 17 18 glinides, dipeptidyl-peptidaseFor peer (DPP)-4 inhibitors, review glucagon-like only peptide (GLP)-1 analogues, 19 biguanides, thiazolidinediones, diabetes, T2DM, prediabetes, prediabetic state, glucose intolerance, 20 impaired glucose, conversion, delay, and prevent. Moreover, all drug names in each drug class 21 will be included in key search terms, for instance, acarbose, voglibose, metformin, glipizide, 22 23 glimepiride, linagliptin, vildagliptin, nateglinide, liraglutide, exenatide, rosiglitazone, and 24 pioglitazone. To identify other eligible studies, reference lists of relevant publications (including 25 trials, reviews, and meta-analyses) will be reviewed for a manual search. 26 27 28 Selection of studies 29 In accordance with the prespecified inclusion criteria, two reviewers working independently will 30 31 evaluate all titles and abstracts to eliminate papers that were deemed irrelevant. The remaining 32 articles will be included in the further assessment. Reviewers will scrutinize full text for each 33 potentially-relevant article. The study identification and exclusion process will be depicted using 34 the PRISMA flow diagram. Discrepancies in study selection will be resolved by negotiation. 35 36 37 Data collection process 38 Two independent reviewers will use a standardized data form to extract trial information. All 39 40 disagreements will be settled via discussion with the third reviewer. The data extracted will be as 41 follows: 42 ► Patient characteristics (age, gender, race, and glycemic parameters). 43 44 ► Trial characteristics (author, year of publication, study design, number of participants, country 45 setting, and funding information). 46 ► Details of intervention and control (dosage, frequency, and treatment duration). 47 ► 48 Data on the outcomes mentioned above. 49 50 Assessment of methodological quality 51 The Cochrane risk of bias assessment tool will be used to assess risk of bias for individual studies. 52 53 This method includes the following seven domains: random sequence generation, allocation 54 concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete 55 outcome data, selective reporting, and other bias.20 Each item will be classified into one of three 56 57 categories as follows: unclear, high, or low risk. All discrepancies in quality assessment will be 58 resolved after mutual agreement and discussion. 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 12 BMJ Open

1 2 3 Data synthesis and statistical analysis 4 5 Initially, we will use a random-effects approach to pool effect estimates for all treatment 6 comparisons in conventional pairwise meta-analyses. For categorical outcomes, the pooled 7 estimates as risk ratios (RRs) with 95% confidence intervals (CIs) will be reported. When data is 8 available, to observe whether the effects of medications on diabetes prevention remain after 9 10 intervention withdrawn, the pooled RRs for diabetes of the intervention and wash-out or follow-up 11 periods, respectively, will be estimated. Continuous data will be reported as mean differences 12 (MDs) with their respective 95% CIs. Statistical heterogeneity across trials will be examined using 13 2 2 14 the I statistic. The I statistic of 75%, 50%, or 25% indicates high, moderate, or low heterogeneity, 15 separately.21 Then, a NMA will be conducted with a frequentist random-effects model. Local 16 inconsistency between direct and indirect evidence within each closed loop will be assessed using 17 22,23 18 a node-splitting test.For In addition,peer a “design-by-treatment” review modelonly will be applied to evaluate the 19 assumption of consistency in the whole network.22 We will generate the surface under the 20 cumulative ranking curve (SUCRA) to assess probabilities of interventions in superiority 21 regarding efficacy and safety outcomes, with higher SUCRA values indicating better effects or 22 23 safety.24 The level of significance will be set at an alpha of 0.05. All analyses will be performed 24 with Stata version 14.0 (StataCorp, College Station, TX). 25 26 27 Subgroup analyses 28 Where possible, analyses will be stratified by age (18-45 years and at least 45 years), gender, 29 ethnicity, and BMI (25-29.9 kg/m2 and ≥30 kg/m2). Moreover, we will also perform subgroup 30 31 analyses according to diagnostic criteria of prediabetes (IFG, IGT, and HbA1c). 32 33 Publication bias 34 We will use the comparison-adjusted funnel plot to assess small study effects including 35 36 publication bias at the network level.25 37 38 Quality of evidence 39 40 The quality of evidence of estimates derived from this study will be rated using the grading of 41 recommendations assessment, development and evaluation (GRADE) framework. The GRADE 42 approach characterises the quality of evidence according to publication bias, study limitations, 43 26 44 inconsistency, imprecision, and indirectness. Evidence of efficacy outcomes will be rated from 45 high quality to very low quality. 46 47 Patient and public involvement 48 49 No patients or public will participate in the study. 50 51 Ethics and dissemination 52 53 Since confidential patient data will not be involved in this study, formal ethics approval is not 54 required. The framework of the PRISMA statements for NMA will be applied to guide review 55 authors to perform this study. The results will be disseminated by a peer-reviewed publication. 56 57 58 DISCUSSION 59 This study is a comprehensive systematic review and NMA to compare and rank a variety of 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 12

1 2 3 anti-diabetic agents for preventing the development of T2DM in patients with prediabetes. Our 4 5 study will provide a summary of available evidence concerning various anti-hyperglycemia agents 6 for T2DM prevention in patients with prediabetic state, benefiting for clinicians and 7 guideline-makers to generate high quality recommendations for these patients. Although a relevant 8 NMA27 published, the study was based on clinical trials before 2014. Additionally, included types 9 10 anti-diabetic agents in the study were limited, dipeptidyl-peptidase (DPP)-4 inhibitors, 11 glucagon-like peptide (GLP)-1 analogues, and some other glucose-lowering drugs that have been 12 tested by later trials clinically were not involved. It is essential to contain these commonly 13 14 prescribed medications in multiple comparisons of glucose-lowering agents for the prevention of 15 T2DM. Moreover, the definition of adults at high risk for T2DM was based on IFG and IGT, 16 excluding people identified by HbA1c. Importantly, HbA1c is a biomarker of long-term glycemic 17 18 control when comparedFor with peer IFG and IGT, review representing average only blood glucose levels during the 19 preceding two to three months.28 However, our network meta-analysis may have several possible 20 limitations. Firstly, the different frequencies, dosages, and routes of administration of 21 pharmacological therapies may result in considerable heterogeneity. Secondly, differences in the 22 23 inclusion criteria of participants and definition of the primary end-point events may influence the 24 quality of evidence. Finally, study level data will be used rather than data on individuals. 25 26 27 Author affiliations 28 1 The Second Clinical College of Guangzhou University of Chinese Medicine, Guangzhou, China 29 2 Shanghai University of Traditional Chinese Medicine, Shanghai, China 30 3 31 Shanghai University of Medicine & Health Sciences, Shanghai, China 32 4 The First Clinical College of Guangzhou University of Chinese Medicine, Guangzhou, China 33 5 Shenzhen Bao'an Traditional Chinese Medicine Hospital Group, Shenzhen, China 34 6 Department of Acupuncture and Moxibustion, Guangdong Provincial Hospital of Chinese Medicine, 35 36 Guangzhou, China 37 7 Shenzhen Bao'an Research Center for Acupuncture and Moxibustion, Shenzhen, China 38 8 Sanming Project of Medicine in Shenzhen, Shenzhen, China 39 40 41 Contributors HZ conceived the review. JRW, GXH, YLJ, and HZ wrote the first draft of this protocol. 42 WBF, ZJL, ML, and PZ were responsible for revision of the draft. HZ, ML and JRW contributed to 43 44 developing the search strategy and registering the protocol. WBF and PZ were the guarantors. All 45 authors scrutinized and approved the final manuscript. 46 Funding This research is supported by the Second Affiliated hospital of Guangzhou University of 47 Chinese Medicine (Guangdong Provincial Hospital of Chinese Medicine) (No. E43603 and E43703) 48 49 and the Sanming Project of Medicine in Shenzhen (No. SZSM201806077). 50 Competing interests None declared. 51 Patient consent Not required. 52 53 Provenance and peer review Not commissioned; externally peer review. 54 55 REFERENCES 56 57 1. American Diabetes Association. Classification and Diagnosis of Diabetes: Standards of 58 Medical Care in Diabetes-2019. Diabetes care 2019;42:S13-28. 59 2. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 12 BMJ Open

1 2 3 national incidence, prevalence, and years lived with disability for 354 diseases and injuries 4 5 for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of 6 Disease Study 2017. Lancet 2018;392:1789-858. 7 3. American Diabetes Association. Cardiovascular Disease and Risk Management: Standards of 8 9 Medical Care in Diabetes-2019. Diabetes care 2019;42:S103-23. 10 4. American Diabetes Association. Microvascular Complications and Foot Care: Standards of 11 Medical Care in Diabetes-2019. Diabetes care 2019;42:S124-38. 12 5. Callaghan BC, Xia R, Reynolds E, et al. Association Between Metabolic Syndrome Components 13 14 and Polyneuropathy in an Obese Population. JAMA neurology 2016;73:1468-76. 15 6. Moelands SV, Lucassen PL, Akkermans RP, et al. Alpha-glucosidase inhibitors for prevention 16 or delay of type 2 diabetes mellitus and its associated complications in people at increased 17 18 risk of developingFor typepeer 2 diabetes review mellitus. The Cochrane only database of systematic reviews 19 2018;12:Cd005061. 20 7. International Diabetes Federation. IDF Diabetes Atlas. 8th Edition. Brussels, Belgium: 21 22 International Diabetes Federation, 2017. 23 8. GBD 2017 Risk Factor Collaborators. Global, regional, and national comparative risk 24 assessment of 84 behavioural, environmental and occupational, and metabolic risks or 25 clusters of risks for 195 countries and territories, 1990-2017: a systematic analysis for the 26 27 Global Burden of Disease Study 2017. Lancet 2018;392:1923-94. 28 9. GBD 2017 Causes of Death Collaborators. Global, regional, and national age-sex-specific 29 mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic 30 31 analysis for the Global Burden of Disease Study 2017. Lancet 2018;392:1736-88. 32 10. Shrestha SS, Honeycutt AA, Yang W, et al. Economic Costs Attributable to Diabetes in Each 33 U.S. State. Diabetes care 2018;41:2526-34. 34 35 11. American Diabetes Association. Economic Costs of Diabetes in the U.S. in 2017. Diabetes care 36 2018;41:917-28. 37 12. Morris DH, Khunti K, Achana F, et al. Progression rates from HbA1c 6.0-6.4% and other 38 prediabetes definitions to type 2 diabetes: a meta-analysis. Diabetologia 2013;56:1489-93. 39 40 13. American Diabetes Association. Prevention or Delay of Type 2 Diabetes: Standards of 41 Medical Care in Diabetes-2019. Diabetes care 2019;42:S29-33. 42 14. American Diabetes Association. Pharmacologic Approaches to Glycemic Treatment: 43 44 Standards of Medical Care in Diabetes-2019. Diabetes care 2019;42:S90-102. 45 15. Hutton B, Salanti G, Caldwell DM, et al. The PRISMA extension statement for reporting of 46 systematic reviews incorporating network meta-analyses of health care interventions: 47 48 checklist and explanations. Annals of internal medicine 2015;162:777-84. 49 16. Rouse B, Cipriani A, Shi Q, et al. Network Meta-analysis for Clinical Practice Guidelines: A 50 Case Study on First-Line Medical Therapies for Primary Open-Angle Glaucoma. Annals of 51 internal medicine 2016;164:674-82. 52 53 17. Mills EJ, Thorlund K, Ioannidis JP. Demystifying trial networks and network meta-analysis. 54 BMJ (Clinical research ed) 2013;346:f2914. 55 18. Moher D, Shamseer L, Clarke M, et al. Preferred reporting items for systematic review and 56 57 meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews 2015;4:1. 58 19. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and 59 meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ (Clinical 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 12

1 2 3 research ed) 2015;350:g7647. 4 5 20. Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration's tool for assessing risk 6 of bias in randomised trials. BMJ (Clinical research ed) 2011;343:d5928. 7 21. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 8 9 (Clinical research ed) 2003;327:557-60. 10 22. Higgins JP, Jackson D, Barrett JK, et al. Consistency and inconsistency in network 11 meta-analysis: concepts and models for multi-arm studies. Research synthesis methods 12 2012;3:98-110. 13 14 23. Dias S, Welton NJ, Caldwell DM, et al. Checking consistency in mixed treatment comparison 15 meta-analysis. Statistics in medicine 2010;29:932-44. 16 24. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical summaries for presenting 17 18 results fromFor multiple-treatment peer meta-analysis: review an overview only and tutorial. Journal of clinical 19 epidemiology 2011;64:163-71. 20 25. Chaimani A, Higgins JP, Mavridis D, et al. Graphical tools for network meta-analysis in STATA. 21 22 PLoS One 2013;8:e76654. 23 26. Puhan MA, Schunemann HJ, Murad MH, et al. A GRADE Working Group approach for rating 24 the quality of treatment effect estimates from network meta-analysis. BMJ (Clinical research 25 ed) 2014;349:g5630. 26 27 27. Stevens JW, Khunti K, Harvey R, et al. Preventing the progression to type 2 diabetes mellitus 28 in adults at high risk: a systematic review and network meta-analysis of lifestyle, 29 pharmacological and surgical interventions. Diabetes research and clinical practice 30 31 2015;107:320-31. 32 28. Inzucchi SE. Clinical practice. Diagnosis of diabetes. The New England journal of medicine 33 2012;367:542-50. 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 12 BMJ Open

1 2 3 4 5 6 Search strategy in PubMed. 7 Block 1: Prediabetes 8 9 #1 “Prediabetic state” [Mesh] 10 #2 “Glucose Intolerance” [Mesh] 11 #3 prediabet*[tiab] OR pre diabet*[tiab] 12 13 #4 (“impaired fasting”[tiab] AND glucose[tiab]) OR IFG[tiab] OR “impaired 14 FPG”[tiab] 15 #5 “glucose intolerance”[tiab] OR “impaired glucose”[tiab] AND (tolerance[tiab] OR 16 metabolism[tiab]) OR IGT[tiab] 17 18 #6 (risk[tiab]For OR progress*[tiab] peer review OR prevent*[tiab] only OR inciden*[tiab] OR 19 conversion[tiab] OR develop*[tiab] OR delay*[tiab]) AND (diabetes[tiab] OR 20 T2D*[tiab] OR NIDDM[tiab] OR “type 2”[tiab] OR “type II”[tiab]) 21 22 #7 #1 OR #2 OR #3 OR #4 OR #5 OR #6 23 24 Block 2: Alpha-glucosidase inhibitors 25 26 #8 “Acarbose” [Mesh] 27 #9 acarbos*[tiab] OR glucobay[tiab] OR precose[tiab] OR prandase[tiab] OR “bay g 28 5421”[tiab] OR BAYG5421 [tiab] 29 #10 voglibos*[tiab] OR glustat[tiab] OR basen[tiab] 30 31 #11 miglitol*[tiab] OR glyset[tiab] 32 #12 glucosidase*[tiab] AND inhibitor*[tiab] 33 #13 #8 OR #9 OR #10 OR #11 OR #12 34 35 36 Block 3: DPP-4 inhibitors 37 14# “Dipeptidyl-Peptidase IV Inhibitors” [Mesh] 38 39 15# “gliptin*”[tiab] OR “dipeptidyl peptidase 4” OR “DPP 4” OR DPP4[tiab] OR 40 “dipeptidyl peptidase IV”[tiab] OR “DPP IV”[tiab] 41 16# alogliptin[tiab] OR anagliptin[tiab] OR bisegliptin[tiab] OR carmegliptin[tiab] 42 OR denagliptin[tiab] OR dutogliptin[tiab] OR evogliptin[tiab] OR gemigliptin[tiab] 43 44 OR gosogliptin[tiab] OR linagliptin[tiab] OR melogliptin[tiab] OR omarigliptin[tiab] 45 OR sitagliptin[tiab] OR saxagliptin[tiab] OR teneligliptin[tiab] OR trelagliptin[tiab] 46 OR vildagliptin [tiab] 47 48 #17 #14 OR #15 OR #16 49 50 51 52 Block 4: GLP-1 analogue 53 #18 “Glucagon-Like Peptide 1” [Mesh] 54 #19 “glucagon like peptide*”[tiab] OR “GLP 1”[tiab] OR GLP1[tiab] 55 #20 exenatide[tiab] OR liraglutide[tiab] OR albiglutide[tiab] OR elsiglutide[tiab] OR 56 57 lixisenatide[tiab] OR dulaglutide[tiab] OR taspoglutide[tiab] OR semaglutide[tiab] 58 OR teduglutide[tiab] 59 #21 #18 OR #19 OR #20 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 12

1 2 3 4 5 Block 5: Sulfonylurea 6 #22 “Sulfonylurea Compounds”[Mesh] 7 #23 glyburid*[tiab] OR gluborid*[tiab] OR glipizid*[tiab] OR glypidizine[tiab] OR 8 9 melizide[tiab] OR napizide[tiab] OR gliquidon*[tiab] OR glisoxepid*[tiab] OR 10 glimepirid*[tiab] 11 #24 #22 OR #23 12 13 14 Block 6: Glinide 15 #25 glinide[tiab] OR glinides[tiab] 16 #26 nateglinid*[tiab] OR senaglinid*[tiab] OR repaglinid*[tiab] OR mitiglinid*[tiab] 17 18 #27 #25 OR #26For peer review only 19 20 Block 7: Biguanides 21 22 #28 Biguanides[Mesh] 23 #29 buformi*[tiab] OR chloroguani*[tiab] OR metformi*[tiab] OR phenformi*[tiab] 24 #30 #28 OR #29 25 26 27 Block 8: Thiazolidinediones 28 #31 Thiazolidinediones[Mesh] 29 #32 thiazolidinedione[tiab] OR TZD[tiab] OR glitazone[tiab] OR glitazones[tiab] 30 31 #33 pioglit*[tiab] OR rosiglit*[tiab] OR troglit*[tiab] 32 #34 #31 OR #32 OR #33 33 34 35 Block 9: All medications 36 #35 37 #13 OR #17 OR #21 OR #24 OR #27 OR #30 OR #34 38 39 40 Block 10: RCT-filter 41 #36 “randomized controlled trial”[pt] OR “controlled clinical trial”[pt] OR 42 randomized[tiab] OR placebo[tiab] OR “drug therapy”[sh] OR randomly[tiab] OR 43 44 trial[tiab] OR groups[tiab] 45 46 Block 11: 47 48 #37 49 #7 AND #35 AND #36 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 12 BMJ Open

1 2 3 PRISMA-P (Preferred Reporting Items for Systematic review and Meta-Analysis Protocols) 2015 checklist: recommended items to 4 5 address in a systematic review protocol* 6 Section and topic Item Checklist item Reported on 7 No Page # 8 9 ADMINISTRATIVE INFORMATION 10 Title: 11 Identification 1a Identify the report as a protocol of a systematic review 1 12 Update 1b If the protocolFor is for an update peer of a previous systematic review review, identify as such only 13 14 Registration 2 If registered, provide the name of the registry (such as PROSPERO) and registration number 1 15 Authors: 16 Contact 3a Provide name, institutional affiliation, e-mail address of all protocol authors; provide physical mailing address of corresponding 1, 6 17 author 18 Contributions 3b Describe contributions of protocol authors and identify the guarantor of the review 6 19 Amendments 4 If the protocol represents an amendment of a previously completed or published protocol, identify as such and list changes; 20 otherwise, state plan for documenting important protocol amendments 21 Support: 22 23 Sources 5a Indicate sources of financial or other support for the review 6 24 Sponsor 5b Provide name for the review funder and/or sponsor 6 25 Role of sponsor 5c Describe roles of funder(s), sponsor(s), and/or institution(s), if any, in developing the protocol 6 26 or funder 27 28 INTRODUCTION 29 Rationale 6 Describe the rationale for the review in the context of what is already known 2 30 Objectives 7 Provide an explicit statement of the question(s) the review will address with reference to participants, interventions, 2, 3 31 comparators, and outcomes (PICO) 32 33 METHODS 34 Eligibility criteria 8 Specify the study characteristics (such as PICO, study design, setting, time frame) and report characteristics (such as years 3, 4 35 considered, language, publication status) to be used as criteria for eligibility for the review 36 Information sources 9 Describe all intended information sources (such as electronic databases, contact with study authors, trial registers or other grey 4 37 literature sources) with planned dates of coverage 38 Search strategy 10 Present draft of search strategy to be used for at least one electronic database, including planned limits, such that it could be 4 39 repeated 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 12 of 12

1 2 3 Study records: 4 Data 11a Describe the mechanism(s) that will be used to manage records and data throughout the review 4 5 management 6 7 Selection 11b State the process that will be used for selecting studies (such as two independent reviewers) through each phase of the review 4 process (that is, screening, eligibility and inclusion in meta-analysis) 8 9 Data collection 11c Describe planned method of extracting data from reports (such as piloting forms, done independently, in duplicate), any 4 10 process processes for obtaining and confirming data from investigators 11 Data items 12 List and define all variables for which data will be sought (such as PICO items, funding sources), any pre-planned data 4 12 assumptions andFor simplifications peer review only 13 Outcomes and 13 List and define all outcomes for which data will be sought, including prioritization of main and additional outcomes, with 3 14 prioritization rationale 15 Risk of bias in 14 Describe anticipated methods for assessing risk of bias of individual studies, including whether this will be done at the outcome 4, 5 16 individual studies or study level, or both; state how this information will be used in data synthesis 17 Data synthesis 15a Describe criteria under which study data will be quantitatively synthesised 5 18 15b If data are appropriate for quantitative synthesis, describe planned summary measures, methods of handling data and methods of 5 19 combining data from studies, including any planned exploration of consistency (such as I2, Kendall’s τ) 20 15c Describe any proposed additional analyses (such as sensitivity or subgroup analyses, meta-regression) 5 21 15d If quantitative synthesis is not appropriate, describe the type of summary planned 22 23 Meta-bias(es) 16 Specify any planned assessment of meta-bias(es) (such as publication bias across studies, selective reporting within studies) 5 24 Confidence in 17 Describe how the strength of the body of evidence will be assessed (such as GRADE) 5 25 cumulative evidence 26 * It is strongly recommended that this checklist be read in conjunction with the PRISMA-P Explanation and Elaboration (cite when available) for important 27 clarification on the items. Amendments to a review protocol should be tracked and dated. The copyright for PRISMA-P (including checklist) is held by the 28 PRISMA-P Group and is distributed under a Creative Commons Attribution Licence 4.0. 29 30 From: Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, Shekelle P, Stewart L, PRISMA-P Group. Preferred reporting items for systematic review and 31 meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015 Jan 2;349(jan02 1):g7647. 32 33 34 35 36 37 38 39 40 41 42 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open

Article Type: Protocol

Date Submitted by the 23-Jul-2019 Author:

Complete List of Authors: Wang, Xianzhe ; The Second Clinical College of Guangzhou University of Chinese Medicine Liu, Jiabin; Guangzhou University of Chinese Medicine, The Second Clinical College Huang, Lijin; Guangzhou University of Chinese Medicine, The Second Clinical College Zeng, Hai; The Second Clinical College of Guangzhou University of Chinese Medicine He, Guoxin; The First College of Guangzhou University of Chinese Medicine Chen, Ling; Guangdong Provincial Hospital of Chinese Medicine, Department of Acupuncture and Moxibustion Ma, Rui; Guangdong Provincial Hospital of Chinese Medicine, Department of Acupuncture and Moxibustion Ning, Baile; Guangdong Provincial Hospital of Chinese Medicine, Department of Acupuncture and Moxibustion Fu, Wenbin; Guangdong Provincial Hospital of Chinese Medicine

Primary Subject Diabetes and endocrinology Heading:

Secondary Subject Heading: Diabetes and endocrinology

diabetes, network meta-analysis, prevention, prediabetes, anti-diabetic Keywords: agents, systematic review

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 12 BMJ Open

1 2 3 4 Anti-diabetic agents for prevention of type 2 diabetes 5 6 7 mellitus in people with prediabetes: a systematic review and 8 9 network meta-analysis protocol 10 11 12 13 1 1 1 1 2 3 3 14 Xianzhe Wang, Jiabin Liu, Lijin Huang, Hai Zeng, Guoxin He, Ling Chen, Rui Ma, Baile 15 Ning,3 Wenbin Fu3,4,5* 16 17 18 Author affiliations:For peer review only 19 1 The Second Clinical College, Guangzhou University of Chinese Medicine, Guangzhou, China 20 2 The First Clinical College, Guangzhou University of Chinese Medicine, Guangzhou, China 21 3 Department of Acupuncture and Moxibustion, Guangdong Provincial Hospital of Chinese 22 23 Medicine, Guangzhou, China 24 4 Shenzhen Bao'an Research Center for Acupuncture and Moxibustion, Shenzhen, China 25 5 Sanming Project of Medicine in Shenzhen, Shenzhen, China 26 27 28 Correspondence to 29 Professor Wenbin Fu; [email protected] 30 31 32 ABSTRACT 33 Introduction Type 2 diabetes mellitus (T2DM) is a substantial health problem worldwide. 34 Prediabetic state is associated with increased risk for the development of diabetes. There are 35 36 various pharmacologic therapies with glucose-lowering activity for diabetes prevention. Of those, 37 most are being compared with placebo instead of active agents. The relative effects and safety of 38 different glucose-lowering drugs still remain uncertain. To address this gap, we will conduct a 39 40 systematic review and network meta-analysis (NMA) to evaluate comparative efficacy and safety 41 of glucose-lowering agents for T2DM prevention in patients with prediabetes. 42 Methods and analysis PubMed, the Cochrane library, and Embase will be searched from 43 44 inception to December 2019 for relevant randomized controlled trials (RCTs) that examined 45 anti-diabetic drugs for diabetes prevention in patients with prediabetes. Two reviewers working 46 independently will screen titles, abstracts, and full papers. Data extraction will also be completed 47 by two independent authors. The primary outcome will be the incidence of T2DM in patients with 48 49 prediabetes at baseline. Secondary outcomes will include the achievement of normoglycemia, 50 all-cause mortality, cardiovascular mortality, and hypoglycemic event. Pairwise meta-analysis and 51 NMA will be conducted for each outcome using a frequentist random-effects model. Additionally, 52 53 subgroup analyses will also be performed. The comparison-adjusted funnel plot will be used to 54 assess publication bias. The overall quality of evidence will be rated with the recommendations 55 assessment, development and evaluation (GRADE) framework. Data analysis will be conducted 56 57 using Stata V.14.0. 58 Ethics and dissemination Ethics approval is not required. We plan to submit the results of this 59 study to a peer-review journal. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 12

1 2 3 PROSPERO registration number CRD42019119157. 4 5 6 Strengths and limitations of this study 7 ► This is a comprehensive systematic review and network meta-analysis to evaluate the 8 effectiveness and safety of various glucose-lowering medications on diabetes prevention among 9 10 people with prediabetic state. 11 ► Where possible, a NMA will combine direct evidence with indirect evidence, allowing 12 comparisons of treatments without being compared to each other head-to-head in clinical trials. 13 14 ► This research will generate clinically useful evidence to benefit patients, clinicians, and 15 guideline-makers. 16 ► The different frequencies, dosages, and routes of administration of pharmacological therapies 17 18 may lead to considerableFor heterogeneity. peer review only 19 20 Keywords: anti-diabetic agents, diabetes, network meta-analysis, prediabetes, prevention, 21 systematic review 22 23 24 INTRODUCTION 25 Type 2 diabetes mellitus (T2DM) is a chronic and complex disease, related to insulin secretory 26 27 defects frequently on the background of insulin resistance; the progression of the disease is 28 associated with genetic factors, metabolic stress, and inflammation.1 The global prevalence of 29 T2DM was estimated to be 463 million people in 2017.2 People with T2DM are at elevated risk 30 31 for chronic kidney disease, heart failure, atherosclerotic cardiovascular disease, polyneuropathy, 32 cognitive impairment, anxiety disorder, and depression.3-5 The term prediabetes is used to describe 33 a blood glucose level higher than the normal range but below the cut-off value for T2DM.6 34 Different glycemic measurements to define the prediabetic stage exist, including impaired fasting 35 36 glucose (IFG), impaired glucose tolerance (IGT), and elevated glycosylated haemoglobin A1c 37 (HbA1c).1 The International Diabetes Federation (IDF) estimated that, in 2017, approximately 352 38 million persons globally had IGT, which is projected to exceed half a billion people before 2045.7 39 40 Hyperglycemia is a well described risk factor for all-cause mortality, total number of all-age 41 deaths attributable to high fasting plasma was 6.5 million people in 2017,8 with T2DM accounting 42 for 1 million deaths.9 Moreover, the economic burden of diabetes is large; in 2017, the American 43 44 Diabetes Association (ADA) estimated the total economic costs attributable to diabetes in the U.S. 45 to be $327 billion.10,11 Thus, there is an urgent need to address huge burden of this worldwide 46 disease with a growing number of suffers. Early interventions for preventing T2DM are 47 warranted.10 Persons diagnosed with prediabetes are thought to be at increased risk for developing 48 49 T2DM, the estimated incidence rate of diabetes among patients with prediabetes in the following 50 10 years exceeds one-third.12 These people are ideal candidates for diabetes prevention efforts. 51 To prevent the progression of prediabetes to T2DM, an intensive behavioral lifestyle 52 53 intervention program (e.g., medical nutrition therapy and physical activity) is recommended in the 54 ADA guidelines.13 Besides lifestyle modification, a variety of anti-diabetic agents (e.g., 55 glucagon-like peptide (GLP)-1 analogues, metformin, and thiazolidinediones) have been 56 57 investigated in clinical trials for diabetes prevention. These pharmacologic approaches with 58 intrinsic glucose-lowering activity (e.g., improve the insulin resistance and preserve pancreaticβ 59 -cell function) are recommended for glycemic treatment in patients with T2DM.14 Of these 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 12 BMJ Open

1 2 3 medications, only metformin therapy for diabetes prevention is recommended as an option for 4 13 5 patients with prediabetes. However, to date, whether other glucose-lowering agents should be 6 considered in those patients or not has not yet to be clarified clearly, even though some findings of 7 recent studies have demonstrated that these pharmacological agents could also exert benefits to 8 prevent or delay the progression to T2DM. In addition, head-to-head comparisons of different 9 10 anti-diabetic agents have rarely been performed by previous clinical trials. A network 11 meta-analysis (NMA) method is able to combine direct and indirect evidence and assess 12 comparative efficacy and safety of various interventions.15-17 Therefore, we plan to conduct the 13 14 systematic review and NMA to assess comparative effects and safety of various anti-diabetic 15 medications in preventing T2DM in patients with prediabetes. 16 17 18 METHODS For peer review only 19 Study design and registration 20 This systematic review protocol is reported in line with the Preferred Reporting Items for 21 Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines.18,19 This study will be 22 23 performed in accordance with the PRISMA extension statements for NMA.15 24 25 Eligibility criteria 26 27 Population 28 Adults (older than 18 years) who have prediabetes will be eligible for inclusion. In this study, 29 prediabetic state involves separate IFG, separate IGT, or both. Diagnostic criteria for prediabetes 30 31 should be established and described in eligible trials. 32 33 Intervention and comparator 34 This study will investigate comparisons of anti-diabetic drugs versus another anti-diabetic agent, 35 36 lifestyle interventions (diet, exercise, or both), placebo or no intervention. Anti-diabetic agents 37 include alpha-glucosidase inhibitors (e.g., acarbose and voglibose), sulphonylureas (e.g., glipizide 38 and glimepiride), meglitinide analogues (e.g., nateglinide), dipeptidyl-peptidase (DPP)-4 39 40 inhibitors (e.g., linagliptin and vildagliptin), glucagon-like peptide (GLP)-1 analogues (e.g., 41 exenatide and liraglutide), biguanides (e.g., metformin), thiazolidinediones (e.g., rosiglitazone and 42 pioglitazone), alone or in combination. In addition, studies using vitamins, traditional Chinese 43 44 medicines, and alternative therapies will be excluded. 45 46 Outcomes 47 The primary outcome will be the incidence of T2DM in patients with prediabetes at baseline. 48 49 Secondary outcomes will include the achievement of normoglycemia, all-cause mortality, 50 cardiovascular mortality, and hypoglycemic event. Classification and definition of T2DM could 51 be based on any recognized standard diagnosis criteria (e.g., the ADA guidelines). 52 53 54 Type of studies 55 All randomized controlled trials (RCTs) comparing anti-diabetic drugs with another anti-diabetic 56 57 agent, lifestyle interventions, placebo or no intervention for T2DM prevention in patients with 58 prediabetes will be included in this study. Duration of intervention has to be with a minimum of 59 12 weeks. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 12

1 2 3 4 5 Search strategy 6 Several databases will be searched from inception to December 2019 for RCTs that investigated 7 anti-diabetic agents for prevention of diabetes among patients with prediabetes. The databases will 8 include PubMed, Embase, and the Cochrane Library. In addition, the language of publication will 9 10 be limited to English. Any potentially-relevant article will be retrieved for review. Details of 11 search strategy of PubMed database are shown in the supplemental material. The literature search 12 will be conducted using the following keywords: alpha-glucosidase inhibitors, sulphonylureas, 13 14 glinides, dipeptidyl-peptidase (DPP)-4 inhibitors, glucagon-like peptide (GLP)-1 analogues, 15 biguanides, thiazolidinediones, diabetes, T2DM, prediabetes, prediabetic state, glucose intolerance, 16 impaired glucose, conversion, delay, and prevent. Moreover, all drug names in each drug class 17 18 will be included inFor key search peer terms, for instance, review acarbose, voglibose, only metformin, glipizide, 19 glimepiride, linagliptin, vildagliptin, nateglinide, liraglutide, exenatide, rosiglitazone, and 20 pioglitazone. To identify other eligible studies, reference lists of relevant publications (including 21 trials, reviews, and meta-analyses) will be reviewed for a manual search. 22 23 24 Selection of studies 25 In accordance with the prespecified inclusion criteria, two reviewers working independently will 26 27 evaluate all titles and abstracts to eliminate papers that were deemed irrelevant. The remaining 28 articles will be included in the further assessment. Reviewers will scrutinize full text for each 29 potentially-relevant article. The study identification and exclusion process will be depicted using 30 31 the PRISMA flow diagram. Discrepancies in study selection will be resolved by negotiation. 32 33 Data collection process 34 Two independent reviewers will use a standardized data form to extract trial information. All 35 36 disagreements will be settled via discussion with the third reviewer. The data extracted will be as 37 follows: 38 ► Patient characteristics (age, gender, race, and glycemic parameters). 39 40 ► Trial characteristics (author, year of publication, study design, number of participants, country 41 setting, and funding information). 42 ► Details of intervention and control (dosage, frequency, and treatment duration). 43 44 ► Data on the outcomes mentioned above. 45 46 Assessment of methodological quality 47 The Cochrane risk of bias assessment tool will be used to assess risk of bias for individual studies. 48 49 This method includes the following seven domains: random sequence generation, allocation 50 concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete 51 outcome data, selective reporting, and other bias.20 Each item will be classified into one of three 52 53 categories as follows: unclear, high, or low risk. All discrepancies in quality assessment will be 54 resolved after mutual agreement and discussion. 55 56 57 Data synthesis and statistical analysis 58 Initially, we will use a random-effects approach to pool effect estimates for all treatment 59 comparisons in conventional pairwise meta-analyses. For categorical outcomes, the pooled 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 12 BMJ Open

1 2 3 estimates as risk ratios (RRs) with 95% confidence intervals (CIs) will be reported. When data is 4 5 available, to observe whether the effects of medications on diabetes prevention remain after 6 intervention withdrawn, the pooled RRs for diabetes of the intervention and wash-out or follow-up 7 periods, respectively, will be estimated. Continuous data will be reported as mean differences 8 (MDs) with their respective 95% CIs. Statistical heterogeneity across trials will be examined using 9 10 the I2 statistic. The I2 statistic of 75%, 50%, or 25% indicates high, moderate, or low heterogeneity, 11 separately.21 Then, a NMA will be conducted with a frequentist random-effects model. Local 12 inconsistency between direct and indirect evidence within each closed loop will be assessed using 13 22,23 14 a node-splitting test. In addition, a “design-by-treatment” model will be applied to evaluate the 15 assumption of consistency in the whole network.22 We will generate the surface under the 16 cumulative ranking curve (SUCRA) to assess probabilities of interventions in superiority 17 18 regarding efficacyFor and safety peer outcomes, withreview higher SUCRA only values indicating better effects or 19 safety.24 The level of significance will be set at an alpha of 0.05. All analyses will be performed 20 with Stata version 14.0 (StataCorp, College Station, TX). 21 22 23 Subgroup analyses 24 Where possible, analyses will be stratified by age (18-45 years and at least 45 years), gender, 25 ethnicity, and BMI (25-29.9 kg/m2 and ≥30 kg/m2). Moreover, we will also perform subgroup 26 27 analyses according to diagnostic criteria of prediabetes (IFG and IGT). 28 29 Publication bias 30 31 We will use the comparison-adjusted funnel plot to assess small study effects including 32 publication bias at the network level.25 33 34 Quality of evidence 35 36 The quality of evidence of estimates derived from this study will be rated using the grading of 37 recommendations assessment, development and evaluation (GRADE) framework. The GRADE 38 approach characterises the quality of evidence according to publication bias, study limitations, 39 26 40 inconsistency, imprecision, and indirectness. Evidence of efficacy outcomes will be rated from 41 high quality to very low quality. 42 43 44 Patient and public involvement 45 No patients or public will participate in the study. 46 47 Ethics and dissemination 48 49 Since confidential patient data will not be involved in this study, formal ethics approval is not 50 required. The framework of the PRISMA statements for NMA will be applied to guide review 51 authors to perform this study. The results will be disseminated by a peer-reviewed publication. 52 53 54 DISCUSSION 55 This study is a comprehensive systematic review and NMA to compare a variety of anti-diabetic 56 57 agents for preventing the development of T2DM in patients with prediabetes. Our study will 58 provide a summary of available evidence concerning various anti-hyperglycemia agents for 59 T2DM prevention in patients with prediabetic state, benefiting for clinicians and guideline-makers. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 12

1 2 3 Previous relevant reviews and meta-analyses27-29 only included clinical trials published before 4 30,31 5 2015. Importantly, recent large-scale RCTs (e.g., the ACE and IRIS trials) have provided 6 substantial data with respect to this topic. Additionally, dipeptidyl-peptidase (DPP)-4 inhibitors 7 and glucagon-like peptide (GLP)-1 analogues are not involved in previous studies. It is essential to 8 contain these commonly prescribed medications in multiple comparisons of glucose-lowering 9 10 agents for the prevention of T2DM. Moreover, the influence of different diagnostic criteria for 11 prediabetes (IFG and IGT) on the prevention efficacy of anti-diabetic agents remains uncertain.28 12 Thus, we plan to conduct this study to investigate various anti-diabetic agents for diabetes 13 14 prevention. The findings of our study will generate high quality recommendations regarding the 15 optimal anti-diabetic agent to reduce risk of diabetes for patients with prediabetes. This study will 16 combine data of all glucose-lowering drugs that have been tested for diabetes prevention by 17 18 clinical trials. To developFor better peer individualized review strategies for diabetes only prevention, intervention 19 efficacy according to diagnostic criteria for prediabetes (IFG and IGT) will also be explored. 20 However, our study may have several possible limitations. Firstly, the different frequencies, 21 dosages, and routes of administration of pharmacological therapies may result in considerable 22 23 heterogeneity. Secondly, differences in the inclusion criteria of participants and definition of the 24 primary end-point events may influence the quality of evidence. Finally, study level data will be 25 used rather than data on individuals. 26 27 28 29 Contributors: HZ and XZW conceived the review. BLN, GXH, JBL, and XZW wrote the first draft of 30 31 this protocol. WBF, LC, LJH, and RM were responsible for revision of the draft. HZ, LJH, and JBL 32 contributed to developing the search strategy and registering the protocol. WBF and HZ were the 33 guarantors. All authors scrutinized and approved the final manuscript. 34 35 36 Funding: This research was supported by Guangzhou University of Chinese Medicine, Guangdong 37 Provincial Hospital of Chinese Medicine (No. E43603 and E43703), and the Sanming Project of 38 Medicine in Shenzhen (No. SZSM201806077). 39 40 41 Competing interests None declared. 42 43 44 Patient consent Not required. 45 46 Provenance and peer review Not commissioned; externally peer review. 47 48 49 50 51 REFERENCES 52 53 54 55 1. American Diabetes Association. Classification and Diagnosis of Diabetes: Standards of 56 57 Medical Care in Diabetes-2019. Diabetes care 2019;42:S13-28. 58 2. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and 59 national incidence, prevalence, and years lived with disability for 354 diseases and injuries 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 12 BMJ Open

1 2 3 for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of 4 5 Disease Study 2017. Lancet 2018;392:1789-858. 6 3. American Diabetes Association. Cardiovascular Disease and Risk Management: Standards of 7 Medical Care in Diabetes-2019. Diabetes care 2019;42:S103-23. 8 9 4. American Diabetes Association. Microvascular Complications and Foot Care: Standards of 10 Medical Care in Diabetes-2019. Diabetes care 2019;42:S124-38. 11 5. Callaghan BC, Xia R, Reynolds E, et al. Association Between Metabolic Syndrome Components 12 and Polyneuropathy in an Obese Population. JAMA neurology 2016;73:1468-76. 13 14 6. Moelands SV, Lucassen PL, Akkermans RP, et al. Alpha-glucosidase inhibitors for prevention 15 or delay of type 2 diabetes mellitus and its associated complications in people at increased 16 risk of developing type 2 diabetes mellitus. The Cochrane database of systematic reviews 17 18 2018;12:Cd005061.For peer review only 19 7. International Diabetes Federation. IDF Diabetes Atlas. 8th Edition. Brussels, Belgium: 20 International Diabetes Federation, 2017. 21 22 8. GBD 2017 Risk Factor Collaborators. Global, regional, and national comparative risk 23 assessment of 84 behavioural, environmental and occupational, and metabolic risks or 24 clusters of risks for 195 countries and territories, 1990-2017: a systematic analysis for the 25 Global Burden of Disease Study 2017. Lancet 2018;392:1923-94. 26 27 9. GBD 2017 Causes of Death Collaborators. Global, regional, and national age-sex-specific 28 mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic 29 analysis for the Global Burden of Disease Study 2017. Lancet 2018;392:1736-88. 30 31 10. Shrestha SS, Honeycutt AA, Yang W, et al. Economic Costs Attributable to Diabetes in Each 32 U.S. State. Diabetes care 2018;41:2526-34. 33 11. American Diabetes Association. Economic Costs of Diabetes in the U.S. in 2017. Diabetes care 34 35 2018;41:917-28. 36 12. Morris DH, Khunti K, Achana F, et al. Progression rates from HbA1c 6.0-6.4% and other 37 prediabetes definitions to type 2 diabetes: a meta-analysis. Diabetologia 2013;56:1489-93. 38 13. American Diabetes Association. Prevention or Delay of Type 2 Diabetes: Standards of 39 40 Medical Care in Diabetes-2019. Diabetes care 2019;42:S29-33. 41 14. American Diabetes Association. Pharmacologic Approaches to Glycemic Treatment: 42 Standards of Medical Care in Diabetes-2019. Diabetes care 2019;42:S90-102. 43 44 15. Hutton B, Salanti G, Caldwell DM, et al. The PRISMA extension statement for reporting of 45 systematic reviews incorporating network meta-analyses of health care interventions: 46 checklist and explanations. Annals of internal medicine. 2015;162:777-84. 47 48 16. Rouse B, Cipriani A, Shi Q, et al. Network Meta-analysis for Clinical Practice Guidelines: A 49 Case Study on First-Line Medical Therapies for Primary Open-Angle Glaucoma. Annals of 50 internal medicine 2016;164:674-82. 51 17. Mills EJ, Thorlund K, Ioannidis JP. Demystifying trial networks and network meta-analysis. 52 53 BMJ (Clinical research ed) 2013;346:f2914. 54 18. Moher D, Shamseer L, Clarke M, et al. Preferred reporting items for systematic review and 55 meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews 2015;4:1. 56 57 19. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and 58 meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ (Clinical 59 research ed) 2015;350:g7647. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 12

1 2 3 20. Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration's tool for assessing risk 4 5 of bias in randomised trials. BMJ (Clinical research ed) 2011;343:d5928. 6 21. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. 7 BMJ (Clinical research ed) 2003;327:557-60. 8 9 22. Higgins JP, Jackson D, Barrett JK, Lu G, Ades AE, White IR. Consistency and inconsistency in 10 network meta-analysis: concepts and models for multi-arm studies. Research synthesis 11 methods. 2012;3:98-110. 12 23. Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in mixed treatment 13 14 comparison meta-analysis. Statistics in medicine 2010;29:932-44. 15 24. Salanti G, Ades AE, Ioannidis JP. Graphical methods and numerical summaries for presenting 16 results from multiple-treatment meta-analysis: an overview and tutorial. Journal of clinical 17 18 epidemiologyFor 2011;64:163-71. peer review only 19 25. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network 20 meta-analysis in STATA. PLoS One 2013;8:e76654. 21 22 26. Puhan MA, Schunemann HJ, Murad MH, et al. A GRADE Working Group approach for rating 23 the quality of treatment effect estimates from network meta-analysis. BMJ (Clinical research 24 ed) 2014;349:g5630. 25 27. Stevens JW, Khunti K, Harvey R, et al. Preventing the progression to type 2 diabetes mellitus 26 27 in adults at high risk: a systematic review and network meta-analysis of lifestyle, 28 pharmacological and surgical interventions. Diabetes research and clinical practice 29 2015;107:320-31. 30 31 28. Haw JS, Galaviz KI, Straus AN, et al. Long-term Sustainability of Diabetes Prevention 32 Approaches: A Systematic Review and Meta-analysis of Randomized Clinical Trials. JAMA 33 internal medicine 2017;177:1808-17. 34 35 29. Galaviz KI, Weber MB, Straus A, et al. Global Diabetes Prevention Interventions: A Systematic 36 Review and Network Meta-analysis of the Real-World Impact on Incidence, Weight, and 37 Glucose. Diabetes care 2018;41:1526-34. 38 30. Holman RR, Coleman RL, Chan JCN, et al. Effects of acarbose on cardiovascular and diabetes 39 40 outcomes in patients with coronary heart disease and impaired glucose tolerance (ACE): a 41 randomised, double-blind, placebo-controlled trial. The lancet Diabetes & endocrinology 42 2017;5:877-86. 43 44 31. Spence JD, Viscoli CM, Inzucchi SE, et al. Pioglitazone Therapy in Patients With Stroke and 45 Prediabetes: A Post Hoc Analysis of the IRIS Randomized Clinical Trial. JAMA neurology 46 2019;76:526-35. 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 12 BMJ Open

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 12 BMJ Open