A Systematic Literature Review and Network Meta-Analysis Comparing Once-Weekly Semaglutide with Other GLP-1 Receptor Agonists in Patients with Type 2 Diabetes Previously Receiving 1–2 Oral Anti-Diabetic Drugs

An SLR was performed in accordance with PRISMA guidelines [29] to identify trials of GLP-1 RAs in patients with T2D who are inadequately controlled on 1–2 OADs. Searches of MEDLINE^®, Embase, and the Cochrane Library were performed via Ovid on April 5, 2016, with updates occurring on October 3, 2016 and August 16, 2017 (Table S1 in the Electronic supplementary material, ESM). Searches of conference proceedings were also carried out for the EASD (2014–2016), the International Society for Pharmacoeconomics and Outcomes Research (ISPOR; 2014–2017), the International Diabetes Federation (IDF; 2013 and 2015), and the ADA Scientific Sessions (2014–2017).

Following a study screening hierarchy for exclusion, all titles and abstracts identified through the literature searches were screened by two reviewers to assess whether they met the PICOS (population, interventions, comparators, outcomes, study design) selection criteria (Table S2 in the ESM). It should be noted that the PICOS criteria were slightly broader than what was required for the current analysis and included additional populations of interest. This is because the SLR was designed to support another NMA assessing the efficacy and safety of once-weekly semaglutide and other GLP-1 RAs in patients who are inadequately controlled on basal insulin. Once title and abstract screening were completed, the reviewers reconciled any existing discrepancies between their selections of studies. The same two reviewers independently screened full-text articles for all studies identified as included at the title and abstract screening phase. When a consensus could not be reached between the two reviewers during reconciliation processes, a senior reviewer provided arbitration. In addition, data from digital curves were extracted using digital extraction tools. Any discrepancies observed between the data extracted by the two analysts were adjudicated by a third reviewer.

A NMA was performed, in accordance with guidance from the National Institute for Health and Care Excellence (NICE), ISPOR, and the Cochrane Institute [30‐34], to compare the efficacy and safety of GLP-1 RAs in patients with T2D. In the analysis, the primary intervention of interest was once-weekly semaglutide (0.5 mg and 1.0 mg), and the primary comparators of interest were all other licensed doses of GLP-1 RAs—liraglutide once-daily (QD), dulaglutide QW, exenatide twice-daily (BID), exenatide QW, lixisenatide QD, and albiglutide QW. While albiglutide is soon to be withdrawn from the market, it remains a comparator of interest as the reason for withdrawal was not related to the safety of the medicine [35]. GLP-1 RAs were often taken with other background anti-diabetic medications in the trials. In order to reduce variability between populations across the different trials, the definition of the add-on to 1–2 OADs population was aligned as closely as possible to the relevant SUSTAIN trials of once-weekly semaglutide (the primary intervention of interest). Therefore, studies assessing GLP-1 RAs as an add-on to one OAD (defined as > 90% of patients inadequately controlled on metformin monotherapy, i.e., sufficiently similar to the patient population in SUSTAIN 2 or 7) or as an add-on to 1–2 OADs (defined as < 100% of patients inadequately controlled on two OADs, i.e., sufficiently similar to the patient population in SUSTAIN 3 and 4) were considered for inclusion. Trials involving patients inadequately controlled on one OAD that was not metformin were also excluded in order to reflect SoC and align with international guidelines [3].

All trials identified in the SLR were examined for data on at least one outcome of interest, and the ability to form a best-case connected network was assessed. The feasibility of generating evidence networks for each of the 20 outcomes of interest (Table S2 in the ESM) was examined next; the outcomes of interest included glycemic control outcomes (e.g., change from baseline in HbA_1c, proportion of patients achieving HbA_1c < 7% [53 mmol/mol] or ≤ 6.5% [48 mmol/mol]), weight outcomes (e.g., change from baseline in weight, body mass index [BMI], proportion of patients achieving weight loss of ≥ 5 or ≥ 10%), SBP, fasting plasma glucose (FPG), postprandial plasma glucose, proportion of patients achieving HbA_1c < 7% without weight gain and without hypoglycemia, and safety outcomes (including the incidence of discontinuations due to adverse events [AEs], nausea, vomiting, diarrhea, pancreatitis, and hypoglycemia [overall, severe, non-severe, nocturnal]).

All analyses of continuous outcomes were performed using a normal likelihood, identity link, shared parameter model, to account for both arm-level and trial-level data reported within the included studies. For the analysis of dichotomous outcomes, a binomial likelihood (assuming a normal distribution), logit link model was used. For each outcome, both fixed effects (FE) and random effects (RE) models were run, and the model with the better fit (based on the deviance information criterion [DIC] and average posterior residual deviance) was used. Exploratory meta-regression analyses were also performed to determine whether a covariate-adjusted model would provide a significantly better model fit. The NMA models were implemented using WinBUGS software (MRC Biostatistics Unit, Cambridge, UK) [36] and employed a Bayesian framework with the use of uninformative prior distributions. Three Markov Monte Carlo chains were used, starting from different initial values of selected unknown parameters. Convergence for all models was assessed using standard diagnostic methods for evaluating convergence [37]. In addition, autocorrelation plots were assessed to detect the presence of autocorrelation in the chains. Following this, model convergence inferences were made from data obtained by sampling for a further 20,000 iterations using all the samples. If models failed to converge, the feasibility of a Bucher indirect comparison was considered. Bucher indirect comparisons were calculated in STATA 13 (StataCorp., College Station, TX, USA), using the indirect command [38].

The results of the NMA are presented as mean treatment differences or odds ratios (ORs) and an associated 95% credible interval (CrI). For continuous outcomes of interest, a treatment which offers a greater mean reduction from baseline is favored—for example a reduction in HbA_1c (%), SBP (mmHg), and weight (kg). For dichotomous outcomes, a treatment which offers an increase in the OR is favored—for example, higher odds for achieving a HbA_1c level < 7%. Where reducing the probability of achieving an outcome is favorable (e.g., a reduction in discontinuations due to AEs), a treatment which offers a reduction in the OR is favored. In Bayesian statistics, it is considered that differences exist only where the CrI does not include the null value for treatment differences or one for ORs. In certain cases, once-weekly semaglutide may achieve a numerical reduction/increase against a comparator, but unless the CrI excludes the null value (for treatment differences) or one (for ORs), it is assumed that there is no difference.

In addition to the mean treatment difference or OR, the median ranks of each treatment (and associated 95% CrI) are provided, where a treatment with a median rank of 1 is considered the best. If two drugs share a ranking, then both are assigned one value lower (e.g., if two drugs are both ranked as the second highest, they will both be given the median rank score of 3). An additional ranking outcome is also presented—the surface under the cumulative ranking curve (SUCRA). The SUCRA is a numerical summary statistic of cumulative ranking probability plots (the probability a treatment is among the top n treatments [between the first and nth rank]) [39]. A higher SUCRA value indicates an increased possibility that a treatment is in the top rank. A treatment which is ‘certain to be the best’ will have a SUCRA value of 1 (i.e., 100%), and a treatment ‘certain to be the worst’ will have a value of 0 (i.e., 0%) [39]. This simplifies information about the effect of each treatment into a single value, allowing the complex results from NMA networks to be expressed with relatively few numbers. When the median rank and SUCRA values are in accordance (e.g., if a treatment has the highest median rank and the highest SUCRA score), this adds further weight to the interpretation.

NMAs combine all available evidence from clinical trials to estimate treatment effects. As this involves combining direct and indirect measures of effect, it is important to examine whether or not these two ‘sources’ of evidence are consistent with one another. Accordingly, all NMAs were formally assessed for inconsistency using Bucher’s method (as outlined by the NICE technical support document [TSD] 4) [33]. Informal assessments were also performed by comparing the results of the NMA with the data reported across the studies included in the evidence networks. In the event of evidence of a substantial inconsistency that may change the conclusions based on the analysis selected (direct, simple indirect, or NMA), the following three steps were taken [33]: (1) data were verified for accuracy; (2) if data were found to be correct, meta-regression or restricted analyses were performed to address the imbalance driving the issue; (3) if the second step failed, or was infeasible, further analyses were considered, within the limitations of avoiding bias.

Finally, this article does not contain any new studies with human subjects or animals performed by any of the authors.

A total of 1273 unique citations of potential interest were identified in the electronic searches. Of these, 85 citations were considered to meet the SLR PICOS criteria. A further 22 citations were identified through hand searching of conference abstracts or were provided by the sponsor in the form of clinical study reports (CSR). Therefore, a total of 107 publications reporting on 75 unique trials were included within the SLR (a list of these trials is provided in Table S3 of the ESM and a PRISMA flow diagram of the SLR is shown in Fig. S1 of the ESM). Of these 75 trials, 41 trials were considered to be relevant for inclusion in the current analysis [22‐24, 27, 40‐70]. The process of excluding the other 34 trials is detailed in Fig. S2 of the ESM.

All 41 trials considered in the NMA formed a connected network, which allowed for the comparison of once-weekly semaglutide with dulaglutide QW (0.75 mg and 1.5 mg), liraglutide QD (1.2 mg and 1.8 mg), exenatide QW (2 mg dose), exenatide QD (5 µg and 10 µg), albiglutide 30 mg, and lixisenatide 20 µg. Placebo, sitagliptin, insulin glargine, and the fixed combination of insulin glargine and lixisenatide are secondary comparators that connect primary comparators of interest. All other secondary comparators were removed from the network, including comparators from the DUAL-1 trial (insulin degludec and the fixed combination of insulin degludec and liraglutide), as they do not contribute to any connections between primary comparators of interest in the network. This resulted in a total of 34 trials [22‐24, 27, 40‐69] remaining within the network.

Within this network, there are multiple arms of insulin glargine and lixisenatide. While the insulin glargine arms were assessed as similar enough to combine into a single treatment node, important differences in the titration strategies between the lixisenatide arms were observed. Accordingly, lixisenatide treatment arms in the network were pooled into one-step (10 µg QD starting dose followed by escalation to 20 µg QD after 15 days) or two-step (10 µg QD starting dose for 1 week, followed by 15 µg QD for 1 week and then 20 µg maintenance dose) arms. Note that the GET GOAL-M trial is included in the network but the lixisenatide arms were not pooled because the primary objective of this trial was to investigate morning vs evening administration.

The 34 trials within the network were next examined for time points for which data were available for at least one outcome (Fig. S3 of the ESM). All 34 trials (100%) reported on at least one outcome of interest between 20 and 28 weeks. Based on this, it was decided to analyze each outcome at 24 ± 4 weeks (approximately 6 months) of treatment—it was assumed that the level of response to treatment within 4 weeks of the target week was unlikely to vary considerably. Overall, the majority of trials (82%) reported at either 24 or 26 weeks (range: week 23–28) in the analysis at 24 ± 4 weeks.

The study design and patient characteristics of the 34 trials are presented in Table S4. Overall, the risk of bias across the 34 studies was considered to be low; however, the highest risk of bias across the studies was associated with elements of study blinding and dropouts. Generally, most trials were deemed sufficiently homogeneous to combine for analysis; however, eight studies were considered to be potential outliers due to study design and patient characteristics: Derosa [44] enrolled treatment-naive patients who were instructed to take metformin for 8 months prior to treatment randomization; Van Gaal [68] specifically enrolled young (mean age of 43 years) and obese (mean BMI of 36.8 kg/m²) patients; and five studies (Araki [41], GetGoal-M Asia [52], Inagaki [59], Ji [60], Lew [63], and Zang [69]) enrolled only Asian patients. It was therefore decided to exclude these studies from the base-case analysis, yielding a total of 26 studies; however, the impact of excluding these trials was explored in sensitivity analyses.

For the analysis, outcome-specific evidence networks were possible for 8 of the 20 outcomes of interest assessed for feasibility. It should be noted that, due to a paucity of data, it was not possible to conduct NMAs on the majority of the safety outcomes included in the SLR PICOS criteria (Table S2 in the ESM). Twenty-six trials were considered in the base-case analysis, and the evidence network is shown in Fig. 1.

In the base-case analysis of the change from baseline in HbA_1c and FPG, and the proportion of patients achieving a HbA_1c < 7% or ≤ 6.5%, the RE model provided a better fit than the FE model in terms of DIC and the average posterior residual deviance (Table S5 of the ESM). For the base-case analyses assessing the change from baseline in SBP and weight as well as the proportion of patients discontinuing due to AEs, the FE model was preferred; no important differences between the FE and RE models were observed in terms of DIC and average posterior residual deviance (Table S5 of the ESM).

Exploratory meta-regression analyses were performed on four outcomes (change from baseline in HbA_1c, FPG, weight, and SBP) to validate the use of unadjusted models in the base case. Overall, meta-regression analysis did not result in a significant improvement in model fit or a reduction in the between-study SD (for the RE model). Therefore, regression may have resulted in overfitting.

The results of the NMA are presented as treatment differences or ORs (once-weekly semaglutide vs comparator) in Fig. 2a–g (the full matrix of relative treatment effect results are shown in Tables S14–S21 in the ESM). The associated treatment ranks (SUCRA and median rank) are also presented in Tables 1 and 2.

All 26 trials reported data on the change from baseline in HbA_1c (Table S6 in the ESM). The evidence network for the change from baseline in HbA_1c is shown in Fig. 1. The results (Fig. 2a, Table S14 of the ESM) showed that once-weekly semaglutide 1.0 mg was associated with a significantly greater reduction in HbA_1c vs all GLP-1 RA comparators. The analyses also suggested that once-weekly semaglutide 0.5 mg can provide significantly greater reductions in HbA_1c vs the majority of GLP-1 RA comparators (11 out of 13). It should be noted that inconsistency (p = 0.006) was detected in one loop of evidence (insulin glargine, exenatide 2 mg QW, and once-weekly semaglutide 1.0 mg) within the network for the change from baseline in HbA_1c. When considering the treatment comparison between once–weekly semaglutide 1.0 mg and exenatide 2 mg QW, the direction of the treatment effect is the same for the direct effect (− 0.36; 95% CI − 0.52, − 0.20), indirect effect (− 0.74; 95% CI − 0.96, − 0.52), and NMA (− 0.51; 95% CI − 0.72, − 0.30). Furthermore, while the magnitude of effect differs, the treatment difference between once-weekly semaglutide 1.0 mg and exenatide 2 mg QW (in favor of once-weekly semaglutide 1.0 mg) remains significant in each analysis. In accordance with the protocol outlined to address inconsistency, restricted analyses were performed to assess the impact that the presence of this inconsistency may have. This involved removing trials (DURATION-3) and arms (the exenatide 2 mg QW arm of SUSTAIN 3 or the insulin glargine arm of SUSTAIN 4) within the loop of evidence. Together, these analyses demonstrate that this inconsistency is unlikely to have a large impact on the wider network, as excluding these trials/arms (one at a time) had a minimal effect on the overall results of the NMA.

An additional sensitivity analysis was performed to validate the legitimacy of excluding eight outlier trials [41, 44, 52, 59, 60, 63, 68, 69] (on the basis of heterogeneity) from the base-case analyses of the change from baseline in HbA_1c. When compared with the base-case analysis, the inclusion of the outlier trials resulted in an additional treatment node for lixisenatide (lixisenatide 10–20 μg) in the network (Fig. S5 in the ESM). The same statistical approach as used in the base-case analyses was taken for this analysis. The inclusion of these outlier trials had little impact on the results and overall interpretation (Table S22 of the ESM).

NMAs were also feasible for three other outcomes: HbA_1c level < 7%, HbA_1c level ≤ 6.5%, and FPG. The evidence networks and data supporting these analyses are shown in Fig. S4 and Tables S7–S9 in the ESM. The analyses of the proportion of patients achieving a HbA_1c level < 7% or ≤ 6.5% suggested that the improved HbA_1c reduction observed with once-weekly semaglutide 1.0 mg vs other GLP-1 RAs increases the probability of achieving these recommended glycemic targets. In line with the results from the analysis of the change from baseline in HbA_1c, once-weekly semaglutide 1.0 mg was associated with significantly higher odds of achieving a HbA_1c level < 7 or ≤ 6.5% vs all GLP-1 RA comparators (Fig. 2b, c, Tables S15 and S16 in the ESM). Furthermore, once-weekly semaglutide 0.5 mg was also associated with higher odds of achieving these HbA_1c targets vs most GLP-1 RA comparators (8 out of 12 for HbA_1c level < 7%; 7 out of 12 for HbA_1c level ≤ 6.5%). The analysis of the change from baseline FPG showed some significant differences in favor of once-weekly semaglutide 1.0 mg vs exenatide (2 mg QW and 10 μg BID) and lixisenatide (20 μg and 20 μg in two steps), while once-weekly semaglutide 0.5 mg was associated with a comparable reduction in FPG vs all GLP-1 RA comparators (Fig. 2d, Table S17 in the ESM).

Across these analyses, once-weekly semaglutide 1.0 mg was the highest-ranked GLP-1 RA, achieving median ranks of 1–2 and SUCRA scores of 94–100% (Tables 1, 2). Together, these treatment ranks showed that once-weekly semaglutide 1.0 mg is likely to be the most clinically efficacious treatment within these networks. Once-weekly semaglutide 0.5 mg was also ranked highly within the networks, and achieved the second-highest SUCRA score and median rank for these four outcomes.