An Indirect Treatment Comparison of Semaglutide 2.0 mg vs Dulaglutide 3.0 mg and 4.5 mg Using Multilevel Network Meta-regression.

AIMS: Currently, no head-to-head data are available comparing semaglutide 2.0 mg with dulaglutide 3.0 mg or 4.5 mg. We conducted an indirect treatment comparison (ITC) of their effects on glycated hemoglobin (HbA1c) and body weight in patients with type 2 diabetes.
MATERIALS AND METHODS: Multilevel network meta-regression was conducted, based on a connected evidence network of published results from the A Study of the Efficacy and Safety of Dulaglutide (LY2189265) in Participants With Type 2 Diabetes 11 trial and individual patient data from the A Research Study to Compare Two Doses of Semaglutide Taken Once Weekly in People With Type 2 Diabetes (SUSTAIN) and SUSTAIN 7 trials.
RESULTS: Semaglutide 2.0 mg significantly reduced HbA1c vs dulaglutide 3.0 mg and 4.5 mg, with estimated treatment differences (ETDs) of -0.44% points (95% credible interval [CrI], -0.68 to -0.19) and -0.28% points (95% CrI, -0.52 to -0.03), respectively. Semaglutide 2.0 mg also significantly reduced body weight vs dulaglutide 3.0 mg and 4.5 mg with ETDs of -3.29 kg (95% CrI, -4.62 to -1.96) and -2.57 kg (95% CrI, -3.90 to -1.24), respectively. Odds of achieving HbA1c < 7.0% were significantly greater for semaglutide 2.0 vs dulaglutide 3.0 mg (odds ratio [OR]: 2.23 [95% CrI, 1.15-3.90]), whereas this did not reach significance for semaglutide 2.0 mg vs dulaglutide 4.5 mg (OR: 1.58 [95% CrI, 0.82-2.78]). Sensitivity analyses supported the main analysis findings.
CONCLUSIONS: This ITC demonstrated significantly greater reductions from baseline in HbA1c and body weight with semaglutide 2.0 mg vs dulaglutide 3.0 mg and 4.5 mg. The findings of this study provide important comparative effectiveness information until randomized head-to-head studies become available.

Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) provide an effective treatment option for people with type 2 diabetes (T2D), offering improved glycemic control alongside weight management and cardiovascular benefits (1). Current treatment guidelines recommend GLP-1 RAs across the treatment pathway for patients with T2D not adequately controlled on metformin alone and as a preferred option for those with established cardiovascular disease regardless of glycated hemoglobin (HbA1c) level (2). Semaglutide and dulaglutide are both long-acting GLP-1 RAs, which are available for once-weekly administration via subcutaneous injection. However, they differ in molecular structure, resulting in potential differences in their metabolic effects (3-5). Understanding the impact of these differences on clinical outcomes is important for clinicians.

Semaglutide and dulaglutide have both been investigated extensively as part of the A Research Study to Compare Two Doses of Semaglutide Taken Once Weekly in People With Type 2 Diabetes (SUSTAIN) and A Study of the Efficacy and Safety of Dulaglutide (LY2189265) in Participants With Type 2 Diabetes (AWARD) clinical trial programs, respectively. A head-to-head trial, SUSTAIN 7, compared the efficacy and safety of semaglutide and dulaglutide at lower doses (semaglutide 0.5 mg vs dulaglutide 0.75 mg) and higher doses (semaglutide 1.0 mg vs dulaglutide 1.5 mg) in patients with T2D inadequately controlled with metformin (4). Superiority of semaglutide compared with dulaglutide was demonstrated regarding reducing HbA1c and body weight at week 40 (4). The 2 treatments had similar safety profiles, with similar proportions of patients experiencing gastrointestinal adverse events and discontinuing treatment because of adverse events (4).

Semaglutide was initially approved for use at maintenance doses of 0.5 mg and 1.0 mg by the US Food and Drug Administration (FDA) in 2017 (6) and European Medicines Agency (EMA) in 2018 (7); and dulaglutide was initially approved at doses of 0.75 mg and 1.5 mg by the FDA and EMA in 2014 (8, 9). However, T2D is a progressive disease, and treatment intensification is often required for maintenance of glycemic control and minimization of complications (10). The results of SUSTAIN 7 and other trials in the SUSTAIN and AWARD programs indicate that higher doses of GLP-1 RAs are associated with greater reductions in HbA1c, although there is a need to balance treatment benefits with adverse events, particularly gastrointestinal adverse events (4). Therefore, higher maintenance doses of semaglutide and dulaglutide have recently been investigated to provide potential treatment options for patients already receiving GLP-1 RAs who require further treatment intensification. Dulaglutide 3.0 mg and 4.5 mg were investigated in AWARD-11 (11) and subsequently approved in 2020 and 2021 by the FDA and EMA, respectively (8, 12). Semaglutide 1.0 mg has previously been compared with dulaglutide 3.0 mg and 4.5 mg using a Bucher indirect treatment comparison (ITC) (13). Semaglutide 2.0 mg has also been investigated in SUSTAIN FORTE (14) and is currently under review by the FDA and EMA.

Although SUSTAIN 7 provides a head-to-head comparison between lower doses of semaglutide and dulaglutide (4), there are currently no head-to-head data comparing the higher doses of the 2 treatments. The objective of this study was to compare the efficacy of semaglutide 2.0 mg with dulaglutide 3.0 mg and 4.5 mg doses using an ITC based on aggregate and individual patient data (IPD) from SUSTAIN FORTE (14) and SUSTAIN 7 (4) and published aggregate data from AWARD-11 (11).

Methods

A multilevel network meta-regression (ML-NMR) was conducted in a Bayesian framework (15) to compare efficacy outcomes of dulaglutide 3.0 mg and 4.5 mg with semaglutide 2.0 mg using data from the SUSTAIN FORTE, SUSTAIN 7, and AWARD-11 randomized controlled trials. ML-NMR has been recently introduced as a methodology for conducting ITCs when IPD are available for some, but not all, included trials in a connected evidence network (15, 16). When implemented in a Bayesian framework, ML-NMR retains the flexibility and extensibility of Bayesian network meta-analysis, allowing prior information to be used, data types with differing likelihoods to be included, and the analysis to be embedded in a probabilistic cost-effectiveness framework (15). Unlike other methods for ITC that allow for population adjustment (eg, matching-adjusted indirect comparison [MAIC] and simulated treatment comparison methods), ML-NMR enables comparisons across multiple studies. It also allows comparison in any target population within a given covariate distribution, meaning that comparisons can be made in each of the included trial populations in a connected evidence network (15, 16). A shared effect modifier assumption (17, 18) was chosen to make the model estimable. Therefore, it was assumed that the potential effect modifier interaction coefficients are identical for all dulaglutide dose levels. This assumption was necessary because data for the high doses of dulaglutide (3.0 mg and 4.5 mg) were only available on aggregate level and for 1 trial (AWARD-11).

Data Sources

Published aggregate data were available from AWARD-11, and aggregate and IPD were available from SUSTAIN FORTE and SUSTAIN 7. SUSTAIN FORTE reported a comparison between semaglutide 1.0 mg and semaglutide 2.0 mg, SUSTAIN 7 compared semaglutide 1.0 mg with dulaglutide 1.5 mg, and AWARD-11 compared dulaglutide 3.0 mg and 4.5 mg with dulaglutide 1.5 mg. Therefore, the 3 trials formed the connected evidence network shown in Fig. 1, in which SUSTAIN 7 was used as an anchoring trial. Missing data in the studies for which IPD were available were imputed using multiple imputation and the multiply imputed data sets were incorporated into the ML-NMR model as described in Gelman et al (19).

Figure 1.

Network of evidence for semaglutide vs high-dose dulaglutide. Abbreviations: QW, once weekly.

Outcomes Assessed

Outcomes assessed were change from baseline in HbA1c and change from baseline in body weight, corresponding with the primary and secondary confirmatory outcomes in all included trials. These outcomes were assessed at 40 weeks in SUSTAIN FORTE and SUSTAIN 7 and at 36 weeks in AWARD-11. For these continuous outcomes, mean treatment differences with associated 95% credible intervals (CrIs) were calculated. Proportions of patients achieving HbA1c < 7.0% were also assessed and, for this dichotomous outcome, treatment odds ratios (ORs) with associated 95% CrIs were calculated. Statistical significance was determined as Crl not crossing 0.

Study and Patient Characteristics

Study characteristics and inclusion criteria of the SUSTAIN FORTE, SUSTAIN 7, and AWARD-11 trials used in the analysis are shown in Table 1. SUSTAIN FORTE and SUSTAIN 7 were multinational, phase 3b trials: SUSTAIN FORTE was a double-blind trial and SUSTAIN 7 was open-label. AWARD-11 was a double-blind, multinational, phase 3 trial. There were some differences in terms of trial design and inclusion criteria among the studies; notably, there were differences in HbA1c inclusion criteria among the trials.

Table 1.

Overview of study characteristics and inclusion criteria of included trials

Study name (primary reference) NCT number	Study design	Inclusion criteria				Duration of follow-up	Randomized treatment	No. of patients randomized	Completion date
		HbA1c	BMI	T2D duration	Background medication
SUSTAIN FORTE(14) NCT03989232	Phase 3b, double-blind, RCT Multinational	8.0%-10.0%	No restriction	At least 6 mo	Metformin ≥ 1500 mg/d (or a maximal tolerated dose) ≥ 90 d with or without sulphonylurea (≥ half the maximum approved dose according to local label or maximum tolerated or effective dose)	40 wk	Semaglutide 2.0 mg	480	November 2020
							Semaglutide 1.0 mg	481
SUSTAIN 7(4)a NCT02648204	Phase 3b, open-label, RCT Multinational	7.0%-10.5%	No restriction	Not specified	Metformin ≥ 1500 mg/d (or a maximal tolerated dose) ≥ 90 d	40 wk	Semaglutide 1.0 mg	300	May 2017
							Dulaglutide 1.5 mg	299
AWARD-11(11) NCT03495102	Phase 3, double-blind, RCT Multinational	7.5%-11.0%	≥25 kg/m2	At least 6 mo	Metformin ≥ 1500 mg/d for ≥ 3 months	36 wk	Dulaglutide 3.0 mg	616	May 2019
							Dulaglutide 4.5 mg	614
							Dulaglutide 1.5 mg	612

Abbreviations: AWARD, A Study of the Efficacy and Safety of Dulaglutide (LY2189265) in Participants With Type 2 Diabetes; BMI, body mass index; HbA1c, glycated hemoglobin; RCT, randomised controlled trial; SUSTAIN, A Research Study to Compare Two Doses of Semaglutide Taken Once Weekly in People With Type 2 Diabetes; T2D, type 2 diabetes.

SUSTAIN 7 also included semaglutide 0.5 mg and dulaglutide 0.75 mg; however, these were not treatments of interest for the current analysis.

Although baseline characteristics were generally similar among the trials (Table 2), the analysis was conducted using a connected network in which SUSTAIN 7 was used as an anchoring trial, making robust analysis possible despite some differences. Although there is little evidence to suggest that differences in baseline characteristics (age, sex, diabetes duration, glycemic control, and body mass index [BMI]) result in effect modification for semaglutide vs dulaglutide (20), ML-NMR takes into account both potential effect modifiers and prognostic factors across trials, allowing adjustment for baseline covariates; therefore, in the main analysis, baseline HbA1c, BMI, and diabetes duration were adjusted for as potential effect modifiers. An unadjusted analysis was also conducted as a sensitivity analysis. ML-NMR also allows comparison in each of the included trial populations in a connected evidence network. For the main analysis, the patient population considered was that of AWARD-11, which is in line with the approach that would be used in other population-adjusted methods (eg, MAIC), in which IPD would be used to match baseline summary statistics to aggregate data (21). The impact of using different populations was explored in sensitivity analyses.

Table 2.

Characteristics of enrolled patients reported in SUSTAIN FORTE, SUSTAIN 7, and AWARD-11

Trial name	SUSTAIN FORTE	SUSTAIN 7a	AWARD-11
Treatment (pooled treatment arms)	Semaglutide 2.0 mg and semaglutide 1.0 mg	Semaglutide 1.0 mg and dulaglutide 1.5 mg	Dulaglutide 1.5 mg, dulaglutide 3.0 mg, and dulaglutide 4.5 mg
N	961	599	1842
Mean age (SD), y	58.0 (10.0)	55.6 (10.6)	57.1 (10.0)
Female, %	41.4	44.4	48.8
Race, n (%)
White	847 (88)	463 (77)	1580 (86)
Asian	69 (7)	93 (16)	45 (2.4)
Black or African American	43 (4)	36 (6)	82 (4.5)
American Indian or Alaska Native	1 (<1)	0	88 (4.8)
Native Hawaiian or Pacific Islander	0	0	5 (<1)
Other	1 (<1)	7 (1)	0
Multiple	0	0	42 (2.3)
Mean HbA1c (SD), %	8.9 (0.6)	8.2 (0.9)	8.6 (1.0)
Mean body weight (SD), kg	99.3 (23.5)	94.5 (21.4)	95.7 (20.3)
Mean BMI (SD), kg/m2	34.6 (7.0)	33.3 (6.5)	34.2 (6.3)
Mean T2D duration (SD), y	9.5 (6.2)	7.5 (5.7)	7.6 (5.7)
Mean fasting glucoseb (SD), mg/dL	194.4 (50.0)	174.8 (44.0)	184.1 (51.5)
Mean DBP (SD), mmHg	80.9 (9.4)	80.8 (8.9)	78.7 (9.0)
Mean SBP (SD), mmHg	134.4 (14.0)	132.6 (14.1)	131.8 (14.1)
Mean pulse rate (SD), beats/minute	75.9 (10.1)	75.1 (10.5)	75.5 (10.0)
Mean eGFR (SD), mL/min/1.73 m2	93.4 (16.9)	97.2 (16.0)	93.5 (18.1)
Mean lipid levels (SD), mg/dL
Total cholesterol	Not collected	183.2 (44.7)	177.5 (NR)c
HDL cholesterol	Not collected	45.0 (10.8)	45.6 (NR)c
LDL cholesterol	Not collected	103.8 (38.8)	93.4 (NR)c
VLDL cholesterol	Not collected	Not collected	34.5 (NR)c
Triglycerides	Not collected	183.3 (112.6)	202.4 (NR)c

Abbreviations: AWARD, A Study of the Efficacy and Safety of Dulaglutide (LY2189265) in Participants With Type 2 Diabetes; BMI, body mass index; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HbA1c, glycated hemoglobin; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NR, not reported; SUSTAIN, A Research Study to Compare Two Doses of Semaglutide Taken Once Weekly in People With Type 2 Diabetes; T2D, type 2 diabetes; VLDL, very low-density lipoprotein.

Note SUSTAIN 7 also included semaglutide 0.5 mg and dulaglutide 0.75 mg; however, these were not treatments of interest for the current analysis.

Fasting plasma glucose in SUSTAIN FORTE and SUSTAIN 7; fasting serum glucose in AWARD-11.

Pooled data are not presented in the AWARD-11 publication. Mean values have been calculated as weighted average.

Analyses were based on results for the trial product estimand in SUSTAIN FORTE and SUSTAIN 7, and the efficacy estimand in AWARD-11. These estimands were similar in that they were defined in all trials as including patients on treatment without rescue therapy. This means that they target the treatment effect when all patients continue to use the treatment for the planned duration of the trial without rescue medication, thus reflecting effect when treatment is used as intended (22).

Sensitivity Analyses

Three sensitivity analyses were conducted to assess the findings of the main analysis. For the first sensitivity analysis (sensitivity analysis 1), the impact of using the SUSTAIN 7 patient population as the target population was explored. In another analysis (sensitivity analysis 2), the impact of using the patient population from SUSTAIN FORTE was explored. To explore the impact of adjustment for potential effect modifiers, an unadjusted analysis (sensitivity analysis 3) using standard Bayesian network meta-analysis methodology was also conducted.

To account for the difference in timing of assessments between AWARD-11 (36 weeks) and the SUSTAIN trials (40 weeks), an additional sensitivity analysis was conducted using SUSTAIN 7 and SUSTAIN FORTE results corresponding to week 36 (obtained from linear interpolation between the week 28 and week 40 visits). Sensitivity analyses 1 and 2, exploring the impact of using the SUSTAIN 7 and SUSTAIN FORTE patient populations, respectively, as the target populations, were also conducted using the interpolated 36-week data from SUSTAIN 7 and SUSTAIN FORTE.

Results

In total, 1842 patients were randomly assigned to a dulaglutide dose in AWARD-11 (3.0 mg, n = 616; 4.5 mg, n = 614; 1.5 mg, n = 612). In SUSTAIN 7, 599 of 1199 patients were randomly assigned to the highest available doses of semaglutide or dulaglutide (semaglutide 1.0 mg, n = 300; dulaglutide 1.5 mg, n = 299). In SUSTAIN FORTE, 961 patients were randomly assigned to either dose of semaglutide (semaglutide 1.0 mg, n = 481; semaglutide 2.0 mg, n = 480).

Change in HbA1c

In the main analysis, semaglutide 2.0 mg was significantly more effective at reducing HbA1c from baseline compared with dulaglutide 3.0 mg and 4.5 mg with estimated treatment differences of –0.44% points (95% CrI, –0.68 to –0.19) and –0.28% points (95% CrI, –0.52 to –0.03), respectively (Fig. 2A). Sensitivity analyses adjusting for different target populations (sensitivity analyses 1 and 2) and without adjustment for potential effect modification (sensitivity analysis 3) supported the findings of the main analysis (Fig. 2A), although in sensitivity analysis 1, the difference between treatment did not reach statistical significance. Sensitivity analyses conducted using SUSTAIN 7 and SUSTAIN FORTE results corresponding to week 36 also supported the findings of the main analysis (Fig. 3A).

Figure 2.

Efficacy outcomes of semaglutide 2.0 mg vs dulaglutide 3.0 mg and 4.5 mg. Change in HbA1c from baseline (A); change in bodyweight (kg) from baseline (B); proportion of patients achieving target HbA1c (C). Main analysis, sensitivity analyses 1-3. Abbreviations: Crl, credible interval; ETD, estimated treatment difference; HbA1c, glycated hemoglobin; OR, odds ratio.

Figure 3.

Efficacy outcomes of semaglutide 2.0 mg vs dulaglutide 3.0 mg and 4.5 mg using SUSTAIN 7 results corresponding to week 36. Change in HbA1c from baseline (A); change in bodyweight (kg) from baseline (B); proportion of patients achieving target HbA1c (C). Main analysis, sensitivity analyses 1 and 2. Abbreviations: Crl, credible interval; ETD, estimated treatment difference; HbA1c, glycated hemoglobin; OR, odds ratio.

Change in Body Weight

In the main analysis, semaglutide 2.0 mg was significantly more effective at reducing body weight from baseline compared with dulaglutide 3.0 mg and 4.5 mg with ETDs of –3.29 kg (95% CrI, –4.62 to −1.96) and –2.57 kg (95% CrI, –3.90 to –1.24), respectively (Fig. 2B). Sensitivity analyses adjusting for different target populations (sensitivity analyses 1 and 2) and without adjustment for potential effect modification (sensitivity analysis 3) supported the findings of the main analysis (Fig. 2B). Sensitivity analyses conducted using SUSTAIN 7 and SUSTAIN FORTE results corresponding to week 36 also supported the findings of the main analysis (Fig. 3B).

Patients Achieving HbA1c < 7.0%

In the main analysis, the proportion of patients achieving HbA1c < 7.0% was significantly greater for semaglutide 2.0 compared with dulaglutide 3.0 mg, with an OR of 2.23 (95% CrI, 1.16-3.90), whereas this did not reach significance for semaglutide 2.0 mg vs dulaglutide 4.5 mg (OR: 1.58 [95% CrI, 0.82-2.78]) (Fig. 2C). Sensitivity analyses adjusting for different target populations (sensitivity analyses 1 and 2) and without adjustment for potential effect modification (sensitivity analysis 3) supported the findings of the main analysis (Fig. 2C). Sensitivity analyses conducted using SUSTAIN 7 and SUSTAIN FORTE results corresponding to week 36 also supported the findings of the main analysis (Fig. 3C).

Discussion

This study indirectly compared the effect of 2 high-dose GLP-1 RAs (semaglutide 2.0 mg vs dulaglutide 3.0 mg and 4.5 mg), on reducing HbA1c and body weight from baseline in patients with T2D using ML-NMR. In this study, semaglutide 2.0 mg provided significantly greater reductions from baseline in HbA1c and body weight vs dulaglutide 3.0 mg and 4.5 mg. In addition, significantly greater proportions of patients achieving HbA1c < 7.0% with semaglutide 2.0 mg vs dulaglutide 3.0 mg, and comparable proportions of patients achieved HbA1c < 7.0% with semaglutide 2.0 mg and dulaglutide 4.5 mg. Results of the main analysis were supported by sensitivity analyses and consistent regardless of the target population used in the ML-NMR.

Semaglutide 1.0 mg has previously been compared with dulaglutide 3.0 mg and 4.5 mg using a Bucher ITC. In the analysis, significantly greater reductions in HbA1c were demonstrated with semaglutide 1.0 mg vs dulaglutide 3.0 mg, and comparable reductions in HbA1c were shown for semaglutide 1.0 mg and dulaglutide 4.5 mg. Significantly greater reductions in body weight were also demonstrated with semaglutide 1.0 mg vs both doses of dulaglutide (13). However, in the absence of a head-to-head trial comparing semaglutide 2.0 mg with dulaglutide 3.0 mg and 4.5 mg, this is the first comparison of these doses of semaglutide and dulaglutide. The results provide comparison of 2 high-dose GLP-1 RAs, which have been shown in clinical trials to provide greater reductions in HbA1c and body weight than their lower dose equivalents (11), thereby providing additional options for patients who require treatment intensification.

Strengths of this study include the comparison method used. As discussed in a recent report by the National Institute for Heath and Care Excellence Decision Support Unit, ML-NMR provides a means of conducting a population-adjusted analysis using a connected evidence network, as opposed to only 2 trials, in situations where IPD are available for some, but not all included trials (15, 16). The report recommends development of a new Technical Support Document recommending ML-NMR as the preferred population-adjusted approach for anchored comparisons (16). Although there were some differences in baseline characteristics of patients enrolled across the three included trials, based on subgroup analyses of SUSTAIN 7, there is little evidence to suggest that differences in baseline characteristics (age, sex, diabetes duration, glycemic control, and BMI) result in effect modification for semaglutide vs dulaglutide (20). The main analysis adjusted for baseline HbA1c, BMI, and diabetes duration as potential effect modifiers, thereby accounting for differences in baseline characteristics among trials. However, a sensitivity analysis conducted without adjustment for potential effect modifiers supported the findings of the main analysis, confirming no appreciable impact of effect modification on the results of the analysis. The ML-NMR approach allows comparison in any target population within a given covariate distribution, meaning that comparisons can be made in each of the included trial populations in a connected evidence network (15, 16). Comparisons were conducted using the AWARD-11 population in the main analysis. Because only published aggregate data were available for AWARD-11, this is in line with the approach that would be used in other population-adjusted methods, such as MAIC, in which IPD would be used to match baseline summary statistics to aggregate data (21). This is also aligned with the MAIC conducted as a supplementary analysis in the recent publication indirectly comparing semaglutide 1.0 mg with dulaglutide 3.0 mg and 4.5 mg (13). The impact of using the SUSTAIN FORTE and SUSTAIN 7 populations was explored in supplementary analyses, all of which supported the findings of the main analysis.

The current study focused only on efficacy outcomes and did not evaluate differences between treatments in terms of safety outcomes, which are highly relevant to clinical decision making. However, a quantitative ITC of safety outcomes is intrinsically more difficult because of cross-trial differences in the assessment of adverse events. In all included trials, a similar safety profile was observed across treatment arms; adverse event rates were comparable between semaglutide 1.0 mg and dulaglutide 1.5 mg treatment arms in SUSTAIN 7 (4); dulaglutide 1.5 mg, 3.0 mg, and 4.5 mg treatment arms in AWARD-11 (11); and semaglutide 1.0 mg and 2.0 mg treatment arms in SUSTAIN FORTE (14). Furthermore, the current study did not evaluate differences in cardiovascular outcomes between treatments. Although cardiovascular outcomes trials have demonstrated cardiovascular benefits of semaglutide 0.5 mg and 1.0 mg (23) and dulaglutide 1.5 mg (24) among people with T2D at high cardiovascular risk, cardiovascular outcomes for semaglutide 2.0 mg and dulaglutide 3.0 mg and 4.5 mg are yet to be fully established.

A limitation of this study is comparison of estimands across trials because there are subtle differences in the handling of intercurrent events, with different criteria for initiation of rescue medication. Furthermore, it was not possible to conduct an analysis based on the results of the treatment policy estimand in SUSTAIN FORTE (14) and SUSTAIN 7 (4), similar to the treatment-regimen estimand in AWARD-11 (11), which included efficacy data for patients regardless of treatment discontinuation or rescue medication. This was because confidence intervals and SEs for estimates on treatment level for the treatment-regimen estimand from AWARD-11 have not been published. However, the trial product estimand may be seen as the more relevant estimand for comparisons between treatments, as it provides estimates of treatment effects without potential confounding because of intercurrent events.

Despite the limitations, the ML-NMR approach allowed for a robust ITC of changes in HbA1c and body weight associated with semaglutide 2.0 mg compared with dulaglutide 3.0 mg and 4.5 mg, and showed consistent results regardless of the target population used in analyses. As for any ITC, a head-to-head trial would be required to fully validate the findings.

Conclusion

This ITC demonstrated significantly greater reductions from baseline in HbA1c and body weight with semaglutide 2.0 mg vs dulaglutide 3.0 mg and 4.5 mg. The findings of this study, and particularly the results provided in different trial populations, provide important comparative effectiveness information until randomized head-to-head studies become available.