Diabetes Ther (2018) 9:973–986 https://doi.org/10.1007/s13300-018-0397-1 ORIGINAL RESEARCH A Network Meta-Analysis Comparing Semaglutide Once-Weekly with Other GLP-1 Receptor Agonists in Japanese Patients with Type 2 Diabetes . . . . Neil Webb Michelle Orme Michal Witkowski Rie Nakanishi Jakob Langer Received: January 19, 2018 / Published online: March 24, 2018 The Author(s) 2018 on diet and exercise, who have previously ABSTRACT received 0–1 oral antidiabetic drugs (OADs). Data at 52–56 weeks were extracted for the fol- Introduction: Semaglutide once-weekly (QW) lowing outcomes (feasible for analysis in an is a novel glucagon-like peptide-1 (GLP-1) ana- NMA): glycated hemoglobin (HbA ), fasting 1c logue administered at a 0.5 or 1.0 mg dose. In plasma glucose (FPG), weight, systolic blood the absence of head-to-head trials between pressure (SBP), and overall hypoglycemia. The semaglutide QW and other GLP-1 receptor data were synthesized using an NMA and a agonists (GLP-1 RAs) in a Japanese population, a Bayesian framework. network meta-analysis (NMA) was performed. Results: Four trials, identiﬁed from the SR and The objective was to assess the relative efﬁcacy Japanese-speciﬁc searches, were relevant for and safety of semaglutide QW vs GLP-1 RAs in inclusion in the NMA. When compared to Japanese patients with type 2 diabetes (T2DM), dulaglutide 0.75 mg QW, semaglutide 0.5 mg with a speciﬁc focus on the comparison QW was shown to provide signiﬁcant reduc- between semaglutide 0.5 mg QW and dulaglu- tions in HbA [- 0.61% (12.3 mmol/mol)], 1c tide 0.75 mg QW. weight (- 1.45 kg), SBP (- 5.03 mmHg), and Methods: A systematic review (SR) and supple- FPG (- 1.26 mmol/L). No signiﬁcant differences mentary Japanese searches were conducted to in the proportion of patients achieving a HbA 1c identify trials of GLP-1 RAs in Japanese patients level \ 7% (53 mmol/mol) or the risk of overall Enhanced content To view enhanced content for this hypoglycemia were observed between semaglu- article go to https://doi.org/10.6084/m9.ﬁgshare.59195 tide 0.5 mg QW and dulaglutide 0.75 mg QW. Conclusion: Overall, semaglutide 0.5 mg QW Electronic supplementary material The online was associated with signiﬁcant reductions from version of this article (https://doi.org/10.1007/s13300- baseline in HbA , weight, SBP, and FPG com- 1c 018-0397-1) contains supplementary material, which is pared with dulaglutide 0.75 mg QW in Japanese available to authorized users. patients with T2DM. These data may provide valuable evidence for clinical decision-making, N. Webb M. Witkowski DRG Abacus, Bicester, Oxfordshire, UK cost-effectiveness analyses, and health technol- ogy appraisal (HTA) requirements. M. Orme Funding: Novo Nordisk Pharma Ltd. ICERA Consulting Ltd., Swindon, UK R. Nakanishi J. Langer (&) Keywords: Dulaglutide; GLP-1 receptor Novo Nordisk Pharma Ltd., Tokyo, Japan agonist/analogue; Glycemic control; HbA ; e-mail: email@example.com 1c 974 Diabetes Ther (2018) 9:973–986 Indirect comparison; Japan/Japanese; Network preserving b-cell function, suppressing gluca- meta-analysis; Semaglutide; Type 2 diabetes; gon levels, and slowing gastric emptying Weight [12, 13]. Unlike other therapies in T2DM, treatment with GLP-1 RA is typically associated with weight loss [14–17] and may also reduce INTRODUCTION cardiovascular risk . In Japan, GLP-1 RAs are increasingly prescribed for the treatment of Type 2 diabetes mellitus (T2DM) is character- T2DM and several studies have demonstrated ized by increased insulin insensitivity coupled the efﬁcacy and safety of GLP-1 RAs in the with a progressive failure of pancreatic b-cells, Japanese population [19–23]. Studies have also resulting in a gradual loss of glycemic control shown that GLP-1 RAs are more effective in and hyperglycemia . If uncontrolled, hyper- Japanese/Asian populations when compared glycemia can lead to diabetic complications, with European populations [24, 25], and are including microvascular (e.g., retinopathy) and typically administered at a lower dose in the macrovascular (e.g., myocardial infarction) Japanese population . This is because T2DM complications . Globally, the prevalence of in Asian patients is primarily characterized by diabetes is increasing and it is estimated that higher b-cell dysfunction, rather than insulin 642 million people aged 20–79 years will have resistance , and GLP-1 RAs have a proven diabetes by 2040; of these people, 87–91% will ability to improve b-cell function . There- have T2DM . In Japan, a national survey has fore, GLP-1 RAs are a favorable choice of ther- shown that the prevalence of diabetes has apy for this population. Japanese patients with increased from 8.9 million in 2007 to 10 million T2DM are also generally less obese than their in 2016 . Speciﬁcally, a signiﬁcant increase in European counterparts and the exposure of a the prevalence of T2DM has also been recorded ﬁxed dose GLP-1 RA will be greater in lighter in a cohort study ; this study attributed the patients. increase in T2DM to higher rates of obesity and Semaglutide once-weekly (QW) is a novel reduced physical activity levels . GLP-1 analogue administered at a 0.5 or 1.0 mg Treatment for T2DM is focused on the dose. The efﬁcacy and safety of semaglutide QW management of hyperglycemia and reducing has been assessed across the Semaglutide Una- the levels of glycated hemoglobin (HbA ); 1c bated Sustainability in Treatment of Type 2 the reduction in HbA is associated with a 1c Diabetes (SUSTAIN) clinical trial program, reduction in the risk of diabetic complications which included two multicenter, head-to-head [2, 7]. In Japan, the main HbA target set by the 1c trials comparing semaglutide QW with exe- Japanese Diabetes Society is \ 7% natide QW or dulaglutide QW. In SUSTAIN 3, (53.0 mmol/mol), which was established from treatment with semaglutide QW provided sig- the perspective of preventing complications . niﬁcant reductions in HbA and weight when 1c However, despite the wide range of treatments compared with exenatide QW, both as an add- available, a recent survey has shown that 47% of on to 1–2 OADs [metformin and/or sulfonylurea patients with T2DM in Japan do not achieve the (SU) and thiazolidinedione (TZD)] . In recommended HbA goal . In addition, 1c SUSTAIN 7, semaglutide QW also provided sig- patients with a higher body mass index (BMI) niﬁcant reductions in HbA and weight when 1c were less likely to achieve the HbA target than 1c compared with dulaglutide QW, both as an add- patients with a lower BMI . Weight gain is on to metformin ; however, no Japanese also a known side effect of many oral antidia- patients were included in this trial. betic drugs (OADs) and insulins used to treat The efﬁcacy and safety of semaglutide QW in T2DM, and may be associated with an increased a Japanese population with T2DM were risk of cardiovascular disease [10, 11]. demonstrated in two recent trials. One trial Glucagon-like peptide-1 receptor agonists compared semaglutide QW (0.5 and 1.0 mg) as (GLP-1 RA) are a class of drugs that act by monotherapy with sitagliptin 100 mg once- increasing glucose-dependent insulin secretion, daily (QD) . After 30 weeks, both doses of Diabetes Ther (2018) 9:973–986 975 semaglutide QW demonstrated signiﬁcant Outcomes Research (ISPOR; 2014–2017), the reductions in HbA and weight compared with International Diabetes Federation (IDF; 2013 1c sitagliptin 100 mg QD; the safety proﬁle was and 2015), and the American Diabetes Associa- also comparable to other GLP-1 RAs. Semaglu- tion (ADA) Scientiﬁc Sessions (2014–2017). The tide QW (0.5 and 1.0 mg) was also assessed in a search results were then screened against the SR 56-week trial of patients who had poor glycemic eligibility criteria to generate a list of potential control on diet and exercise or one OAD . studies to include in the NMA (Table S1). As the The addition of semaglutide QW was associated eligibility criteria of the SR were restricted to with signiﬁcant improvements in HbA and studies published in English, a supplementary 1c weight when compared with the addition of search of four Japanese language databases (Ja- another OAD. pan Pharmaceutical Information Center, Japan Following the recent introduction of health Science and Technology Information Aggrega- technology appraisal (HTA) in Japan and the tor, J-GLOBAL, and Igaku Chuo Zasshi) was increasing amount of treatment options avail- conducted to ensure all potentially relevant able to patients with T2DM, decision-makers studies had been identiﬁed. need to assess the relative clinical beneﬁts and Citations of interest were identiﬁed by one risks of each treatment to make recommenda- reviewer and veriﬁed by a second independent tions on their use within a limited budget. The reviewer on the basis of title and abstract. Full availability of such comparative clinical data publications were obtained for all citations of can also be used for cost-effectiveness analyses interest and were assessed by one reviewer and in Japan in the context of HTA. In the absence veriﬁed by a second independent reviewer. Any of head-to-head trials between semaglutide QW uncertainties were resolved through discussion and other GLP-1 RAs in a Japanese population, a between reviewers. Data were then extracted network meta-analysis (NMA) was performed. into an Excel spreadsheet by one reviewer and The objective was to assess the relative efﬁcacy checked by a second reviewer. All included ref- and safety of semaglutide QW vs GLP-1 RAs in erences were assessed for risk of bias using a Japanese patients, with a speciﬁc focus on the seven-criteria checklist as approved by the comparison between semaglutide 0.5 mg QW National Institute of Health and Care Excel- and dulaglutide 0.75 mg QW; dulaglutide is the lence (NICE) . latest QW GLP-1 RA in Japan (only available as a 0.75 mg QW dose) and semaglutide 0.5 mg QW NMA Methodology is the expected maintenance dose for the Japa- nese population. An NMA was performed to compare the efﬁcacy and safety of GLP-1 RAs in Japanese patients, where the primary intervention of interest was METHODS semaglutide 0.5 mg QW and the comparators of interest were all other licensed doses of GLP-1 Systematic Review RAs in Japan—liraglutide QD, dulaglutide QW, exenatide twice-daily (BID), exenatide QW, and A systematic review (SR) was performed in lixisenatide QD. In order to reduce variability accordance with PRISMA guidelines to between populations in different trials, the identify trials of GLP-1 RAs in Japanese patients population of interest was aligned to the two with T2DM. Searches of MEDLINE , Embase, Japanese SUSTAIN trials of semaglutide QW and the Cochrane Library were performed via [NN9535-4091 trial (NCT02207374; now avail- Ovid on April 5, 2016, with updates occurring able as a full-text publication, Kaku et al. ) on October 3, 2016 and August 16, 2017. Sear- and NN9535-4092 (NCT02254291; now avail- ches of conference proceedings were also carried able as a full-text publication, Seino et al. )]. out for the European Association for the Study Therefore, while trials investigating a broader of Diabetes (EASD; 2014–2016), the Interna- population were eligible for inclusion in the SR tional Society for Pharmacoeconomics and 976 Diabetes Ther (2018) 9:973–986 (Table S1), only trials investigating semaglutide of selected unknown parameters. Convergence QW or other licensed doses of GLP-1 RAs in for all models were assessed by analyzing his- Japan and in Japanese populations who have tory and density plots, and Brooks–Gel- previously received 0–1 OADs were considered man–Rubin diagnostic plots . In addition, for further analysis. autocorrelation plots were assessed to detect the The feasibility of performing an NMA at two presence of autocorrelation in the chains. Fol- time points (30 and 56 weeks), based on the lowing this, model convergence inferences were study durations of the two Japanese SUSTAIN made from data obtained by sampling for a trials, was examined. All trials identiﬁed in the further 20,000 iterations. SR were examined for data on at least one out- For continuous outcomes, a mean treatment come of interest at approximately 30 weeks effect with an associated 95% credible interval (duration of the NN9535-4092 trial ) and (CrI) is estimated and the treatment difference 56 weeks (duration of the NN9535-4091 trial (95% CrI) for semaglutide 0.5 mg QW vs com- ), and the ability to form a best-case con- parator is presented. For dichotomous out- nected network was assessed. The feasibility of comes, an odds ratio (OR) with an associated generating evidence networks for each of the 20 95% CrI is calculated for semaglutide 0.5 mg outcomes of interest (Table S1) was next exam- QW vs comparator. A difference between ined; the outcomes of interest included HbA semaglutide 0.5 mg QW and a comparator is 1c outcomes [e.g., change from baseline, propor- assumed to only exist when the 95% CrI does tion of patients achieving a level \ 7% not include the null value for treatment differ- (53 mmol/mol)], weight, BMI, systolic blood ences, or one for ORs. pressure (SBP), fasting plasma glucose (FPG), Finally, this article does not contain any new postprandial plasma glucose ,and safety out- studies with human or animal subjects per- comes (including the incidence of overall formed by any of the authors. hypoglycemia). The analysis of continuous outcomes (e.g., RESULTS change from baseline in HbA ) was performed 1c using a normal likelihood, identity link, single Identiﬁed Publications parameter model (based on arm-level data), or a shared parameter model, which allows for a A total of 2387 unique citations of potential single coherent synthesis when outcome data is reported at both the arm level and trial level. interest were identiﬁed in the electronic search and nine citations were identiﬁed in the sup- For dichotomous outcomes, a binomial likeli- hood, logit link model was used for efﬁcacy plementary searches. Data on two studies (NN9535-4091 and NN9535-4092) were pro- outcomes [e.g., proportion of patients achieving a HbA level \ 7% (53 mmol/mol)], while a vided by Novo Nordisk in the form of clinical 1c binomial likelihood, cloglog link model was trial summary reports (CSRs) ahead of publica- tion. Both studies are now publicly available as used for safety outcomes (e.g. ,incidence of overall hypoglycemia). All analyses were per- full-text publications [31, 32]; henceforth, the NN9535-4091 trial is also referred to as ‘‘Kaku formed using a ﬁxed effects (FE) model; the FE model provided a better model ﬁt compared 2018‘‘  and the NN9535-4092 trial is also referred to as ‘‘Seino 2018’’ . Of these, 33 with the random effects (RE) model in terms of deviance information criterion and residual publications relating to 29 unique randomized controlled trials (RCTs) were considered to meet deviance. The NMA models were implemented using the inclusion criteria for the SR (the overall ﬂow of studies across the SR and supplementary WinBUGS software (MRC Biostatistics Unit, searches is shown in a PRISMA diagram in Cambridge, UK)  and employed a Bayesian framework with the inclusion of vague prior Fig. S1). In total, ten trials were considered to be the most relevant for the NMA. distributions. Three Markov Monte Carlo chains were used, starting from different initial values Diabetes Ther (2018) 9:973–986 977 These ten trials were examined for time disease duration, which ranged from 6.6 to points for which data were available for at least 10.3 years across the trials. However, the trials one outcome of interest. Six trials reported on recruited patients of a similar level of disease an outcome of interest between 23 and severity as indicated by the baseline HbA 1c 30 weeks (approximately 6 months); however, levels; all trials enrolled patients with a HbA 1c an analysis at 30 weeks based on the duration of value of [ 7% (53.0 mmol/mol), which is sug- Seino 2018  was not feasible, as no con- gestive of inadequate control as per the ADA nected network between semaglutide and guidelines, and the mean baseline HbA ranged 1c dulaglutide could be formed. The analysis at from 8.09% to 8.82% (64.9–72.9 mmol/mol). 56 weeks based on the duration of Kaku 2018  was possible as three additional trials Trial Data and Evidence Networks reported on an outcome of interest at 52 weeks (considered sufﬁciently close to 56 weeks to allow comparison) and formed a connected For the analysis at 52–56 weeks, outcome- network with Kaku 2018 . speciﬁc evidence networks were possible for 7 of Within the four trials included in the anal- the 20 outcomes of interest assessed for feasi- ysis at 52–56 weeks, the intervention of interest bility. It should be noted that it was not possible (semaglutide 0.5 mg QW) and two comparators to conduct NMAs on the majority of the safety of interest (dulaglutide QW and liraglutide QD) outcomes included in the SR eligibility criteria were assessed. Two trials included liraglutide (Table S1). This was due to a paucity of data, 0.9 mg QD, while one trial included dulaglutide which precluded the ability to generate con- 0.75 mg QW; the dosages of both GLP-1 RAs nected networks with semaglutide QW. All four represent the highest dose available in Japan trials reported data on HbA [change from 1c [23, 37]. A lower dose of liraglutide (liraglutide baseline and percentage achieving a HbA 1c 0.6 mg QD) was also reported in one trial. No level \ 7.0% (53 mmol/mol)] and FPG (change trials assessing exenatide (BID or QW) or from baseline; Table S2) at 52–56 weeks and lixisenatide QD were eligible for inclusion in were used to construct an evidence network to the NMA. compare these outcomes for semaglutide 0.5 mg The study design and patient characteristics QW, dulaglutide 0.75 mg QW, and liraglutide of the four trials included in the analysis at 0.6 and 0.9 mg QD (Fig. 1); in this network, an 52–56 weeks were generally similar and suit- additional OAD [metformin, SU, a-glucosidase able for comparison within an evidence net- inhibitor (a-GI), TZD, dipeptidyl peptidase-4 work (Table 1). All four studies were multicenter (DPP-4) inhibitor, or a glinide] and placebo are phase 3 trials conducted in Japan. The risk of essential secondary comparators. Although not bias (based on the NICE checklist) across the a primary intervention of interest, semaglutide four trials was considered low (Fig. S2); how- 1.0 mg QW was included in the network as it ever, Odawara 2016  and Seino 2011  was assessed alongside semaglutide 0.5 mg QW were both double-blind trials, while Kaku 2016 in Kaku 2018 ; the results for semaglutide and Kaku 2018 [22, 32] were both open-label 0.5 mg QW vs semaglutide 1.0 mg QW are pre- trials. In line with the population of interest, all sented to aid interpretation. Three trials repor- trials enrolled adult populations (C 20 years) ted data for the change from baseline in SBP and with T2DM who had not received any treat- weight, and the overall incidence of hypo- ments (n = 2) or had received one OAD (n =2). glycemia at 52–56 weeks (Table S2). It should be The mean age of trial populations ranged from noted that while the overall incidence of 57.6 to 59.7 years, and 18.5–36.0% of patients hypoglycemia was recorded in the NN9535- were female. The mean weight of the patients 4091 trial, these data are not published in Kaku was also similar across the trials and ranged 2018 ; for transparency, these data have from 65.8 to 71.6 kg, while the mean BMI ran- now been made available in Table S2. As Seino ged from 24.9 to 26.5 kg/m . The characteristic et al.  did not report any data for these with the greatest variability was the mean outcomes, liraglutide 0.6 mg QD is absent from 978 Diabetes Ther (2018) 9:973–986 Table 1 Overview of design and baseline characteristics of trials included in the NMA Trial Number of Randomized Background therapy Proportion of Female, Mean age, Mean Mean Mean baseline Mean patients treatment patients n (%) years (SD) baseline baseline duration of baseline randomized receiving one weight, kg HbA ,% diabetes, years BMI, kg/ 1c and exposed OAD (%) (SD) (SD) (SD) m (SD) Kaku 2016 240 LIRA starting dose Pretrial OAD 240 (100) 58 (24) 59.6 (11.6) 69.4 (14.2) 8.1 (0.8) 7.80 (5.77) 25.7 (4.2)  0.3 mg QD changing to 0.9 mg QD Open-label, parallel trial 120 LIRA starting dose Pretrial 120 (100) 40 (33) 59.2 (10.2) 68.2 (13.6) 8.1 (0.8) 8.47 (6.55) 25.5 (3.7) 0.3 mg QD changing OAD ? additional Japan to 0.9 mg OAD QD ? additional OAD Kaku 2018 239 SEMA 0.5 mg QW SU, glinide, a-GI, or NA 73 (30.5) 58.0 (10.6) 71.0 (15.4) 8.04 (0.89) 8.12 (6.01) 26.25 (4.80)  TZD Phase 3a, open- 241 SEMA 1.0 mg QW SU, glinide, a-GI, or NA 67 (27.8) 58.7 (10.2) 71.7 (15.9) 8.14 (0.96) 9.35 (6.50) 26.42 (4.71) label, active- TZD controlled, 120 Additional OAD SU glinide, a-GI, or All 31 (25.8) 59.2 (10.1) 72.2 (14.9) 8.10 (0.89) 9.27 (6.99) 26.73 (4.56) parallel trial (either of SU, TZD Japan glinide, a-GI, or TZD) Odawara 2016 280 DULA 0.75 mg QW None NA 52 (19) 57.2 (9.6) 71.3 (12.5) 8.15 (0.77) 6.8 (5.6) 25.6 (3.6)  137 LIRA starting dose None NA 24 (18) 57.9 (10.4) 70.2 (12.5) 8.08 (0.89) 6.3 (6.0) 25.5 (3.5) Phase 3, double 0.3 mg QD changing blind, parallel to 0.9 mg QD trial Japan Seino 2011 88 LIRA starting dose SU 88 (100) 29 (33) 61.3 (11) 64.5 (12.0) 8.61 (0.78) 11.6 (7.7) 24.4 (3.8)  0.3 mg QD changing to 0.9 mg QD Double-blind, parallel trial 88 LIRA starting dose SU 88 (100) 35 (40) 59.1 (10.3) 66.2 (12.0) 9.0 (0.91) 9.3 (5.8) 25.3 (3.6) 0.3 mg QD changing Japan to 0.6 mg QD 88 Placebo SU 88 (100) 31 (35) 58.6 (9.7) 66.8 (13.7) 8.85 (0.99) 10.1 (7.3) 24.9 (4.0) a-GI a-glucosidase inhibitor, BMI body mass index, DULA dulaglutide, HbA glycated hemoglobin, LIRA liraglutide, NA not applicable, OAD oral antidiabetic drug, QD once-daily, QW 1c once-weekly, SD standard deviation, SEMA semaglutide, SU sulfonylurea, TZD thiazolidinedione Diabetes Ther (2018) 9:973–986 979 Fig. 1 Evidence networks for a HbA (change from (overall incidence). HbA glycated hemoglobin, FPG 1c 1c baseline and percentage achieving a HbA level \ 7.0% fasting plasma glucose, OAD oral antidiabetic drug, QD 1c (53 mmol/mol) and FPG (change from baseline), b SBP, once-daily, QW once-weekly, SBP systolic blood pressure weight (both change from baseline), and hypoglycemia the evidence network (Fig. 1). An additional - 2.16 mmol/L). The treatment differences evidence network was also possible for these compared with liraglutide 0.6 and 0.9 mg QD three trials for the incidence of nocturnal were also signiﬁcant in favor of semaglutide hypoglycemia; however, the model was unsta- 0.5 mg QW. An additional analysis of glycemic ble and failed to converge because of the low control showed that the likelihood of achieving number of events reported in the trials. a HbA level \ 7.0% (53.0 mmol/mol) was 1c comparable between semaglutide 0.5 mg QW and dulaglutide 0.75 mg QW [OR 2.01 (95% CrI NMA Results 0.91, 4.49), Table 2]; the likelihood of achieving this outcome was 78.4% and 64.3% for The results of the NMA for the change from semaglutide 0.5 mg QW and dulaglutide baseline in HbA and FPG are presented as 1c 0.75 mg QW, respectively (Table S3). However, treatment differences in Fig. 2 (the absolute patients were found to be signiﬁcantly more differences from baseline are shown in likely to achieve a HbA level \ 7.0% 1c Table S3). This analysis demonstrated that when (53.0 mmol/mol) with semaglutide 0.5 mg QW compared with dulaglutide 0.75 mg QW, compared with liraglutide 0.6 mg and 0.9 mg semaglutide 0.5 mg QW is associated with a QD (Table 2). Overall, semaglutide 1.0 mg QW signiﬁcant 0.61% (95% CrI - 0.92, - 0.30) was associated with the greatest improvements reduction in HbA (absolute reduction 1c in both HbA and FPG. 1c - 2.00% vs - 1.39%) and a signiﬁcant The NMA results for the change from base- 1.26 mmol/L (95% CrI - 1.78, - 0.74) reduc- line in weight (Fig. 3) demonstrated that tion in FPG (absolute reduction - 3.42 vs semaglutide 0.5 mg QW is associated with a 980 Diabetes Ther (2018) 9:973–986 Fig. 2 Forest plots of the NMA results (semaglutide credible interval, FPG fasting plasma glucose, HbA 1c 0.5 mg QW vs comparator) for the change from baseline glycated hemoglobin, NMA network meta-analysis, QD in HbA and FPG. Treatment differences are considered once-daily, QW once-weekly 1c signiﬁcant when the 95% CrI excludes the null value. CrI signiﬁcant 1.45 kg (95% CrI - 2.65, - 0.25) However, a signiﬁcant reduction in SBP of reduction in weight compared with dulaglutide 5.03 mmHg (95% CrI - 9.29, - 0.79) with 0.75 mg QW (absolute reduction - 1.62 vs semaglutide 0.5 mg QW compared with - 0.17 kg; Table S3). The treatment difference dulaglutide 0.75 mg QW was still observed (ab- compared with liraglutide 0.9 mg QD was also solute difference from baseline - 3.60 vs signiﬁcant in favor of semaglutide 0.5 mg QW. 1.45 mmHg; Table S3). Semaglutide 1.0 mg QW The analysis of the change from baseline in SBP demonstrated the greatest improvements in (Fig. 3) was associated with a higher level of weight and SBP. uncertainty as shown by the wide 95% CrIs. Diabetes Ther (2018) 9:973–986 981 Table 2 NMA results (semaglutide 0.5 mg QW vs com- DISCUSSION parator) for percentage of patients achieving a HbA level 1c of \ 7% (53 mmol/mol) and the incidence of overall The objective of this study was to assess the hypoglycemia relative efﬁcacy and safety of semaglutide 0.5 mg QW vs GLP-1 RAs in Japanese patients, Comparator Odds ratio (95% CrI)— with a speciﬁc focus on the comparison semaglutide 0.5 mg QW vs comparator between semaglutide 0.5 mg QW and dulaglu- tide 0.75 mg QW. Overall, semaglutide 0.5 mg Percentage of patients achieving a HbA level of \ 7% 1c QW was associated with signiﬁcant reductions (53 mmol/mol) from baseline in HbA , weight, SBP, and FPG, 1c compared with dulaglutide 0.75 mg QW. In Dulaglutide 2.01 (0.91, 4.49) addition, semaglutide 0.5 mg QW achieved 0.75 mg QW signiﬁcant reductions from baseline in HbA 1c Liraglutide 10.64 (3.96, 29.36) and FPG vs both liraglutide 0.6 and 0.9 mg QW, 0.6 mg QD and a signiﬁcant reduction in weight vs liraglutide 0.9 mg QW (liraglutide 0.6 mg QD Liraglutide 2.85 (1.45, 5.62) was not available for comparison). Furthermore, 0.9 mg QD semaglutide 0.5 mg QW was associated with Semaglutide 0.49 (0.27, 0.85) signiﬁcantly higher odds of achieving a HbA 1c 1.0 mg QW level \ 7% (53.0 mmol/mol) compared with liraglutide 0.6 mg and 0.9 mg QD. Incidence of overall hypoglycemia To our knowledge, this is the ﬁrst NMA Dulaglutide 2.99 (0.21, 40.09) comparing semaglutide QW with other GLP-1 0.75 mg QW RAs in a Japanese population. However, the results of this NMA can be compared with the Liraglutide 3.11 (0.31, 30.35) recent head-to-head, multicenter (Asia, Europe, 0.9 mg QD and the USA), clinical trial SUSTAIN 7, which Semaglutide 0.75 (0.47, 1.18) assessed semaglutide QW and dulaglutide QW, 1.0 mg QW both as an add-on to metformin . The direction of effect for HbA and weight 1c Bold values indicate odds ratios where the associated 95% demonstrated between semaglutide 0.5 mg QW CrI excludes 1 and dulaglutide 0.75 mg QW in our analysis is CrI credible interval, HbA glycated hemoglobin, NMA 1c broadly consistent with the data reported in network meta-analysis, QD once-daily, QW once-weekly SUSTAIN 7. In SUSTAIN 7 (which did not include any Japanese patients), semaglutide Generally, the incidence of overall hypo- 0.5 mg QW demonstrated a statistically signiﬁ- glycemia was low across the three trials inclu- cant reduction in HbA (- 1.5% vs - 1.1%) 1c ded in the NMA of this outcome (Table S2). and weight (- 4.6 vs - 2.3 kg) compared with Overall, the results of the NMA did not reveal dulaglutide 0.75 mg at 40 weeks. This equates to any signiﬁcant difference in the incidence of a treatment difference of - 0.4% for HbA and 1c overall hypoglycemia between semaglutide - 2.3 kg for weight (both in favor of semaglu- 0.5 mg QW and dulaglutide 0.75 mg QW [OR tide 0.5 mg QW), which are similar to the 2.99 (95% CrI 0.21, 40.09)] or liraglutide 0.9 mg treatment differences estimated in our NMA at QD [OR 3.11 (95% CrI 0.31, 30.35), Table 2]; the 52–56 weeks [- 0.61% (95% CrI - 0.92, - 0.30) incidence was 5.2%, 1.8%, and 1.7% for and - 1.45 kg (95% CrI - 2.65, - 0.25), semaglutide 0.5 mg QW, dulaglutide 0.75 mg respectively]. SUSTAIN 7 also demonstrated that QW, and liraglutide 0.9 mg QD, respectively more patients achieved a HbA level B 7% 1c (Table S3). The risk of overall hypoglycemia was (53.0 mmol/mol) with semaglutide 0.5 mg QW also similar between semaglutide 0.5 mg QW compared with dulaglutide 0.75 mg QW (69% and semaglutide 1.0 mg QW. 982 Diabetes Ther (2018) 9:973–986 Fig. 3 Forest plots of the NMA results (semaglutide credible interval, FPG fasting plasma glucose, HbA 1c 0.5 mg QW vs comparator) for the change from baseline glycated hemoglobin, NMA network meta-analysis, QD in weight and SBP. Treatment differences are considered once-daily, QW once-weekly, SBP systolic blood pressure signiﬁcant when the 95% CrI excludes the null value. CrI vs 52%). In our analysis, semaglutide 0.5 mg GLP-1 RAs. However, the two recent Japanese QW was associated with an OR of 2.01 vs trials of semaglutide QW have demonstrated dulaglutide 0.75 mg QW for the proportion of that its safety proﬁle is comparable to other patients achieving a HbA level \ 7% GLP-1 RAs [31, 32]. Our analysis of the inci- 1c (53.0 mmol/mol), thereby corroborating what dence of overall hypoglycemia is in agreement was observed in SUSTAIN 7; however, as the with these ﬁndings, as no signiﬁcant differences 95% CrI included 1, statistical signiﬁcance for semaglutide 0.5 mg QW vs dulaglutide could not be concluded. 0.75 mg QW or liraglutide 0.9 mg QD were Overall, our analysis provides a robust detected. The NMAs were also subject to addi- assessment of the efﬁcacy of semaglutide QW tional limitations. Although an SR and supple- across ﬁve outcomes compared with dulaglutide mentary Japanese-speciﬁc searches were QW and liraglutide QD. In contrast, it was only conducted to identify relevant trials, only possible to provide estimates for one safety four trials were available for the analysis at outcome (incidence of overall hypoglycemia), 52–56 weeks. Consequently, the direct compar- which limits the ability to make any deﬁnitive isons within the evidence network were sup- conclusions from our study on the relative ported by data from only one trial each. In safety of semaglutide QW compared with other addition, while the overall risk of bias across the Diabetes Ther (2018) 9:973–986 983 trials included in the analyses was low, the efﬁcacy of semaglutide QW. Following the highest risk was associated with study blinding recent introduction of HTA in Japan, and the due to the presence of two open-label trials. The increasing amount of treatment options avail- absence of double-blinding may be considered a able for T2DM, these data provide valuable limitation as this can potentially introduce evidence for decision-making and cost-effec- performance bias into the NMA . tiveness analyses. Analysis of a few secondary outcomes asses- sed may have also been limited by poor reporting or low event rates. The overall inci- ACKNOWLEDGEMENTS dence of hypoglycemia was low across the trials, which may have contributed to greater uncer- tainty in the analysis. Hypoglycemic episodes Funding. This study, medical writing, and are often underreported by patients, which may article processing charges were funded by Novo be due to reasons such as fear of being judged, Nordisk Pharma Ltd. losing their job or driving license, and unawareness of nocturnal events [40, 41]. In Editorial Assistance. Editorial assistance in addition, studies often report the number of the preparation of this manuscript was provided patients with hypoglycemic events, rather than by Alan Weids of DRG Abacus. Support for this the actual frequency of events . Further- assistance was funded by Novo Nordisk Pharma more, non-severe events are not always reported Ltd. in trials. Patients may not consider such mild events as important enough to mention to their Authorship. All named authors meet the doctor [41, 42]. Greater uncertainty was also International Committee of Medical Journal observed in the analysis of SBP, when compared Editors (ICMJE) criteria for authorship for this to the analysis of HbA , FPG, and weight. The manuscript, take responsibility for the integrity 1c of the work as a whole, and have given ﬁnal measurement of SBP is known to be variable and values can differ for the same patient through- approval for the version to be published. All out the day. In addition, there are often patient- authors had full access to all of the data in this , device-, and procedure-related inaccuracies in study and take complete responsibility for the the way that blood pressure is monitored . integrity of the data and accuracy of the data Together, these may have contributed to the analysis. uncertainty present in our analysis of SBP. Disclosures. Jakob Langer is an employee of Despite this, a signiﬁcant difference in the Novo Nordisk Pharma Ltd. and shareholder of reduction of SBP from baseline with semaglu- Novo Nordisk A/S. Rie Nakanishi is an employee tide 0.5 mg QW compared with dulaglutide of Novo Nordisk Pharma Ltd. Neil Webb is an 0.75 mg QW was still observed. employee of DRG Abacus. Michal Witkowski is an employee of DRG Abacus. Michelle Orme CONCLUSION received funding from Novo Nordisk Pharma Ltd. to conduct the analysis. Our analysis suggests that semaglutide 0.5 mg QW provides greater efﬁcacy compared with Compliance with Ethics Guidelines. This dulaglutide 0.75 mg QW in the treatment of article does not contain any new studies with Japanese patients with T2DM. Statistically sig- human or animal subjects performed by any of niﬁcant improvements for semaglutide 0.5 mg the authors. QW vs dulaglutide 0.75 mg QW were found for Data Availability. The datasets generated HbA , weight, SBP, and FPG. In the absence of 1c and/or analyzed during the current study are head-to-head trials between semaglutide QW available from the corresponding author on and other GLP-1 RAs in a Japanese population, reasonable request. this NMA provides an estimate of the relative 984 Diabetes Ther (2018) 9:973–986 9. Yokoyama H, Oishi M, Takamura H, et al. Large- Open Access. This article is distributed scale survey of rates of achieving targets for blood under the terms of the Creative Commons glucose, blood pressure, and lipids and prevalence Attribution-NonCommercial 4.0 International of complications in type 2 diabetes (JDDM 40). BMJ License (http://creativecommons.org/licenses/ Open Diabetes Res Care. 2016;4(1):e000294. by-nc/4.0/), which permits any noncommer- 10. Russell-Jones D, Khan R. Insulin-associated weight cial use, distribution, and reproduction in any gain in diabetes–causes, effects and coping strate- medium, provided you give appropriate credit gies. Diabetes Obes Metab. 2007;9(6):799–812. to the original author(s) and the source, provide 11. Carver C. Insulin treatment and the problem of a link to the Creative Commons license, and weight gain in type 2 diabetes. Diabetes Educ. indicate if changes were made. 2006;32(6):910–7. 12. Prasad-Reddy L, Isaacs D. A clinical review of GLP-1 receptor agonists: efﬁcacy and safety in diabetes and beyond. Drugs Context. 2015;4:212283. REFERENCES 13. Baggio LL, Drucker DJ. Biology of incretins: GLP-1 1. Saini V. Molecular mechanisms of insulin resistance and GIP. Gastroenterology. 2007;132(6):2131–57. in type 2 diabetes mellitus. World J Diabetes. 2010;1(3):68–75. 14. Kalra S, Baruah MP, Sahay RK, Unnikrishnan AG, Uppal S, Adetunji O. Glucagon-like peptide-1 2. Stratton IM, Adler AI, Neil HA, et al. Association of receptor agonists in the treatment of type 2 dia- glycaemia with macrovascular and microvascular betes: past, present, and future. Indian J Endocrinol complications of type 2 diabetes (UKPDS 35): Metab. 2016;20(2):254–67. prospective observational study. BMJ. 2000;321(7258):405–12. 15. Pozzilli P, Norwood P, Jodar E, et al. Improved glycemic control and weight loss with once-weekly 3. Ogurtsova K, da Rocha Fernandes JD, Huang Y, dulaglutide vs placebo, both added to titrated daily et al. IDF Diabetes Atlas: global estimates for the insulin glargine, in type 2 diabetes patients prevalence of diabetes for 2015 and 2040. Diabetes (AWARD-9). In: American Diabetes Association’s Res Clin Pract. 2017;128:40–50. 76th scientiﬁc sessions, New Orleans, LA, 12 June 4. Ministry of Health, Labour and Welfare. Results of Heisei 28 ‘‘National Health and Nutrition Survey’’ 16. Pozzilli P, Norwood P, Jodar E, et al. Placebo-con- [in Japanese]. 2017. http://www.mhlw.go.jp/stf/ trolled, randomized trial of the addition of once- houdou/0000177189.html. Accessed Dec 2017. weekly glucagon-like peptide-1 receptor agonist dulaglutide to titrated daily insulin glargine in 5. Mukai N, Doi Y, Ninomiya T, et al. Trends in the patients with type 2 diabetes (AWARD-9). Diabetes prevalence of type 2 diabetes and prediabetes in Obes Metab. 2017;19(7):1024–31. community-dwelling Japanese subjects: the Hisayama study. J Diabetes Investig. 17. Russell-Jones D, Vaag A, Schmitz O, et al. Liraglu- 2014;5(2):162–9. tide vs insulin glargine and placebo in combination with metformin and sulfonylurea therapy in type 2 6. Inzucchi SE, Bergenstal RM, Buse JB, et al. Man- diabetes mellitus (LEAD-5 met ? SU): a randomised agement of hyperglycaemia in type 2 diabetes, controlled trial. Diabetologia. 2015: a patient-centred approach. Update to a 2009;52(10):2046–55. position statement of the American Diabetes Asso- ciation and the European Association for the Study 18. Mundil D, Cameron-Vendrig A, Husain M. GLP-1 of Diabetes. Diabetologia. 2015;58(3):429–42. receptor agonists: a clinical perspective on cardio- vascular effects. Diabetes Vasc Dis Res. 7. Tanaka S, Tanaka S, Iimuro S, et al. Predicting 2012;9(2):95–108. macro- and microvascular complications in type 2 diabetes: the Japan Diabetes Complications Study/ 19. Araki E, Inagaki N, Tanizawa Y, Oura T, Takeuchi M, the Japanese Elderly Diabetes Intervention Trial risk Imaoka T. Efﬁcacy and safety of once-weekly engine. Diabetes Care. 2013;36(5):1193–9. dulaglutide in combination with sulphonylurea and/or biguanide compared with once-daily insulin 8. Araki E, Haneda M, Kasuga M, et al. New glycemic glargine in Japanese patients with type 2 diabetes: a targets for patients with diabetes from the Japan randomized, open-label, phase III, non-inferiority Diabetes Society. J Diabetes Investig. study. Diabetes Obes Metab. 2015;17(10):994–1002. 2017;8(1):123–5. Diabetes Ther (2018) 9:973–986 985 20. Inagaki N, Atsumi Y, Oura T, Saito H, Imaoka T. 29. Hramiak I, Ahmann A, Capethorn M, et al. Efﬁcacy Efﬁcacy and safety proﬁle of exenatide once weekly and safety of once-weekly semaglutide vs. exenatide compared with insulin once daily in Japanese ER after 56 weeks in subjects with type 2 diabetes patients with type 2 diabetes treated with oral (SUSTAIN 3). Can J Diabetes. 2016;40(5):S41. antidiabetes drug(s): results from a 26-week, ran- domized, open-label, parallel-group, multicenter, 30. Pratley R, Aroda V, Lingvay I, et al. Semaglutide noninferiority study. Clin Ther. versus dulaglutide once weekly in patients with 2012;34(9):1892.e1–1908.e1. type 2 diabetes (SUSTAIN 7): a randomised, open- label, phase 3b trial. Lancet Diabetes Endocrinol. 21. Inagaki N, Ueki K, Yamamura A, Saito H, Imaoka T. 2018. https://doi.org/10.1016/S2213-8587(18)300 Long-term safety and efﬁcacy of exenatide twice 24-X. daily in Japanese patients with suboptimally con- trolled type 2 diabetes. J Diabetes Investig. 31. Seino Y, Terauchi Y, Osonoi T, et al. Safety and 2011;2(6):448–56. efﬁcacy of semaglutide once weekly vs sitagliptin once daily, both as monotherapy in Japanese peo- 22. Kaku K, Kiyosue A, Ono Y, et al. Liraglutide is ple with type 2 diabetes. Diabetes Obes Metab. effective and well tolerated in combination with an 2018;20(2):378–388. oral antidiabetic drug in Japanese patients with type 2 diabetes: a randomized, 52-week, open-label, 32. Kaku K, Yamada Y, et al. Safety and efﬁcacy of once- parallel-group trial. J Diabetes Investig. weekly semaglutide versus additional oral antidia- 2016;7(1):76–84. betic drugs, in Japanese people with inadequately controlled type 2 diabetes: a randomized trial. 23. Odawara M, Miyagawa J, Iwamoto N, Takita Y, Diabetes Obes Metab. 2018. https://doi.org/10. Imaoka T, Takamura T. Once-weekly glucagon-like 1111/dom.13218. peptide-1 receptor agonist dulaglutide signiﬁcantly decreases glycated haemoglobin compared with 33. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA once-daily liraglutide in Japanese patients with type Group. Preferred reporting items for systematic 2 diabetes: 52 weeks of treatment in a randomized reviews and meta-analyses: the PRISMA statement. phase III study. Diabetes Obes Metab. BMJ. 2009;339:b2535. 2016;18(3):249–57. 34. National Institute for Health and Care Excellence. 24. Kim YG, Hahn S, Oh TJ, Park KS, Cho YM. Differ- Single technology appraisal: user guide for com- ences in the HbA1c-lowering efﬁcacy of glucagon- pany evidence submission template. 2015. https:// like peptide-1 analogues between Asians and non- www.nice.org.uk/process/pmg24/resources/single- Asians: a systematic review and meta-analysis. technology-appraisal-user-guide-for-company-evid Diabetes Obes Metab. 2014;16(10):900–9. ence-submission-template-pdf-72286715419333. Accessed Mar 2018. 25. Terauchi Y, Satoi Y, Takeuchi M, Imaoka T. Monotherapy with the once weekly GLP-1 receptor 35. Lunn D, Jackson C, Best N, Thomas A, Spiegelhalter agonist dulaglutide for 12 weeks in Japanese D. The BUGS book a practical introduction to patients with type 2 diabetes: dose-dependent Bayesian analysis. Boca Raton: CRC; 2013. effects on glycaemic control in a randomised, double-blind, placebo-controlled study. Endocr J. 36. Welton NJ, Sutton AJ, Cooper NJ, Abrams KR, Ades 2014;61(10):949–59. AE. Evidence synthesis for decision making in healthcare. 1st ed. New York: Wiley; 2012. 26. Seino Y, Takami A, Boka G, Niemoeller E, Raccah D. Pharmacodynamics of the glucagon-like pep- 37. Suzuki S, Oura T, Takeuchi M, Boye KS. Evaluation tide-1 receptor agonist lixisenatide in Japanese and of the impact of once weekly dulaglutide on Caucasian patients with type 2 diabetes mellitus patient-reported outcomes in Japanese patients poorly controlled on sulphonylureas with/without with type 2 diabetes: comparisons with liraglutide, metformin. Diabetes Obes Metab. insulin glargine, and placebo in two randomized 2014;16(8):739–47. studies. Health Qual Life Outcomes. 2017;15(1):123. 27. Yabe D, Seino Y, Fukushima M, Seino S. Beta cell dysfunction versus insulin resistance in the patho- 38. Seino Y, Rasmussen MF, Nishida T, Kaku K. Gluca- genesis of type 2 diabetes in East Asians. Curr Diab gon-like peptide-1 analog liraglutide in combina- Rep. 2015;15(6):602. tion with sulfonylurea safely improves blood glucose measures vs sulfonylurea monotherapy in 28. Seino Y, Kuwata H, Yabe D. Incretin-based drugs for Japanese patients with type 2 diabetes: results of a type 2 diabetes: focus on East Asian perspectives. 52-week, randomized, multicenter trial. J Diabetes J Diabetes Investig. 2016;7(Suppl 1):102–9. Investig. 2011;2(4):280–6. 986 Diabetes Ther (2018) 9:973–986 39. Turner RM, Spiegelhalter DJ, Smith GC, Thompson 42. Ostenson CG, Geelhoed-Duijvestijn P, Lahtela J, SG. Bias modelling in evidence synthesis. J R Stat Weitgasser R, Markert Jensen M, Pedersen-Bjergaard Soc Ser A Stat Soc. 2009;172(1):21–47. U. Self-reported non-severe hypoglycaemic events in Europe. Diabet Med. 2014;31(1):92–101. 40. Graveling AJ, Frier BM. Hypoglycaemia: an over- view. Prim Care Diabetes. 2009;3(3):131–9. 43. Kallioinen N, Hill A, Horswill MS, Ward HE, Watson MO. Sources of inaccuracy in the measurement of 41. Elliott L, Fidler C, Ditchﬁeld A, Stissing T. Hypo- adult patients’ resting blood pressure in clinical glycemia event rates: a comparison between real- settings: a systematic review. J Hypertens. world data and randomized controlled trial popu- 2017;35(3):421–41. lations in insulin-treated diabetes. Diabetes Ther. 2016;7(1):45–60.
– Springer Journals
Published: Mar 24, 2018