Literature Search and Study Characteristics

The literature search of the MEDLINE, Embase, and ICHUSHI databases identified 21 publications (Fig. 1). Of these, 16 were excluded following primary screening. Among the five potentially relevant publications retained for secondary screening, two were excluded based on the study design and one based on geography (Supplementary Material Table S6). Finally, two studies meeting all the eligibility criteria were included in this SLR (Supplementary Material Table S7). No additional studies were identified from the gray literature search.

Fig. 1

PRISMA flow diagram depicting the study selection process. PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses

An overview of the two studies included in this SLR is presented in Table 2. The STEP 6 trial conducted by Kadowaki et al. was a double-blind, phase IIIa post hoc analysis that investigated the effect of semaglutide for weight management among adult East Asian (South Korea and Japan) PwOD with or without T2D [36]. Patients were randomized (4:1:2:1) to receive either once-weekly subcutaneous semaglutide 2.4 mg or a matching placebo, or semaglutide 1.7 mg or a matching placebo, for 68 weeks (via a dose-escalation regimen: initiation at 0.25 mg once-weekly for the first 4 weeks, followed by an escalation every 4 weeks to 0.5 mg, 1.0 mg, 1.7 mg, and 2.4 mg per week until the target dose is reached after 16 weeks), along with lifestyle intervention for D&E. Oral hypoglycemic medications were in concomitant use among patients at baseline.

Table 2 Overview of included studies

Kumahara et al. conducted a multicenter, double-blind study in Japan to assess the safety and efficacy of mazindol versus placebo among PwOD [37]. Patients randomly received a 0.5-mg mazindol tablet or an inert placebo for 12 weeks based on a fixed–flexible dosage schedule, with the final mean dose of mazindol of 2.11 mg.

Baseline Characteristics

Overall, patients in the STEP 6 trial had an average age range of 50–52 years, with a mean BMI of 31.6–32.0 kg/m2 (Supplementary Material Table S8) [36]. The semaglutide arms had a higher proportion of women (1.7 mg: 37%, n = 37; 2.4 mg: 43%, n = 85) than the placebo arm (26%, n = 26). The average systolic blood pressure (SBP) in the overall group was 133–135 mmHg. Lipid profiles (geometric mean) were similar across the three treatment arms (Supplementary Material Table S9). Total cholesterol ranged from 5.10 to 5.30 mmol/L, while high-density lipoprotein (HDL) cholesterol was 1.30 mmol/L, in all treatment arms and triglycerides were in the range of 1.4–1.6 mmol/L. The proportion of patients with prediabetes was similar across the three arms: 21% (n = 21), 22% (n = 44), and 25% (n = 25) of patients in the semaglutide 1.7 mg, 2.4 mg, and placebo arms, respectively. All treatment arms had the same mean HbA1c (6.4%) and the same proportion of patients with T2D (25%).

Kumahara et al. reported proportions of women of 73% (n = 83) and 78% (n = 89) in the mazindol and placebo arms, respectively (Supplementary Material Table S8) [37]. Other baseline characteristics were not reported for this trial.

Change in Body Weight from Baseline

The STEP 6 trial reported a higher reduction in mean body weight among patients in the semaglutide arms (2.4 mg: − 11.55 kg, 1.7 mg: − 8.53 kg) versus the placebo group (− 1.73 kg) (Table 3) [36]. From baseline to Week 68, the percentage change in body weight was − 13.20%, − 9.60%, and − 2.10% in the semaglutide 2.4 mg, semaglutide 1.7 mg, and placebo arms, respectively.

Table 3 Change in body weight, body weight %, total cholesterol, HDL cholesterol and SBP, and adverse events

Similarly, a significantly higher body weight reduction was noted at 12 weeks of mazindol therapy versus placebo (− 4.19 kg vs. − 1.20 kg; p < 0.001) (Table 3) [37]. The percentage body weight reduction was also higher in the mazindol group versus the placebo group (− 5.52% vs. − 1.50%).

Change in HbA 1c

Semaglutide therapy resulted in a higher reduction in mean HbA1c in the overall group (semaglutide 2.4 mg: − 0.96%, 1.7 mg: − 0.93%) versus placebo (− 0.03%) (Supplementary Material Table S10) [36]. This outcome was not reported for mazindol therapy.

Change in SBP, Total Cholesterol, and HDL Cholesterol

The STEP 6 trial reported a favorable change in total cholesterol from baseline to Week 68: − 17.20 mg/dL and − 16.90 mg/dL for semaglutide 2.4 mg and 1.7 mg, respectively, versus 0.80 mg/dL for placebo (Table 3) [36]. However, the change in HDL cholesterol was similar across all treatment arms, at 3.0–4.20 mg/dL. Semaglutide therapy resulted in a greater reduction in SBP (2.4 mg: − 10.92 mmHg, 1.7 mg: − 11.28 mmHg) versus placebo (− 5.26 mmHg).

Similarly, a significantly higher reduction in total cholesterol from baseline to 12 weeks was noted with mazindol therapy versus placebo (− 8.60 vs. − 5.80 mg/dL; p < 0.001); however, the mean change in HDL cholesterol was reported to be 0.30 mg/dL (p ≥ 0.10) for the mazindol group versus − 0.20 mg/dL for the placebo group (Table 3) [37]. The change in SBP from baseline to 12 weeks was − 1.50 mmHg in the mazindol arm versus − 3.50 mmHg in the placebo group.

Proportion of Patients Not Achieving Early Responder Status

A higher proportion of patients on placebo (79%, n = 79) failed to achieve early responder status (defined as ≥ 5% body weight reduction from baseline at Week 68) than in the semaglutide arms (17% [n = 33] and 28% [n = 27] of patients in the semaglutide 2.4 mg and 1.7 mg arms, respectively) (Supplementary Material Table S11) [36]. This outcome was not reported for mazindol therapy.

Adverse Events

A higher occurrence of adverse events (AEs) was associated with semaglutide (2.4 mg: 86%, 1.7 mg: 82%) versus placebo (79%) (Table 3) [36]. Similarly, mazindol use (41.2%) resulted in a higher number of AEs than placebo (16.7%) (Table 3) [37].

Treatment Discontinuation

Overall, 20 patients (~ 7%) in the semaglutide arms of the STEP 6 trial (2.4 mg: 13 patients, 1.7 mg: 7 patients) and 3 patients (~ 3%) receiving placebo discontinued treatment (Supplementary Material Table S12) [36]. Trial product discontinuation because of AEs was low for all treatment arms. However, when compared with placebo, the semaglutide arms showed a slightly higher discontinuation due to AEs: about 3% of patients (n = 5 for 2.4 mg, n = 3 for 1.7 mg semaglutide) on semaglutide therapy discontinued treatment owing to AEs versus only one patient (1%) in the placebo arm.

Similarly, Kumahara et al. reported an overall discontinuation rate of 14.9% (n = 17) for mazindol versus 14.0% (n = 16) for placebo (Supplementary Material Table S12) [37]. The nonresponder discontinuation rate was lower in the mazindol group (0.9%, n = 1) versus the placebo group (2.6%, n = 3).

Feasibility Assessment for Indirect Treatment Comparisons

The feasibility of conducting a population-adjusted ITC between two active treatments of semaglutide and mazindol was assessed by investigating two key randomized placebo-controlled trials: STEP 6 for semaglutide and Kumahara et al. for mazindol [36, 37]. In this framework, the semaglutide 2.4 mg group was anchored to mazindol 0.5 mg by placebo (Fig. 2). IPD was available for the former trial, but aggregate data were used for the latter trial in the absence of IPD.

Fig. 2

Network of the population-adjusted ITC employed in the current study. ITC indirect treatment comparison, STC simulated treatment comparison, STEP Semaglutide Treatment Effect in People with Obesity

The following efficacy outcomes were reported in both STEP 6 (at Weeks 4, 8, 12, 16, 20, 28, 36, 44, 52, 60, and 68) and Kumahara et al. (at Week 12): 1) change in body weight; 2) change in SBP; and 3) change in lipid levels (LDL, HDL, total cholesterol, and triglycerides) [36, 37].

Potential Effect Modifiers

According to a clinical expert, age, sex, and baseline body weight were identified as potential effect modifiers for all the efficacy outcomes, while presence/absence of diabetes, hypertension, and lipid levels at baseline were effect modifiers for changes in body weight, SBP, and lipid levels, respectively.

Other Risks of Bias

Regarding patient eligibility criteria, obesity was defined based on BMI in the STEP 6 trial and percentage higher weight than standard body weight in the study by Kumahara et al. [36, 37]. This body weight criterion introduced in the mazindol trial was comparable with a BMI ≥ 27 kg/m2 in STEP 6 provided the height of the patients in Kumahara et al. was > 180 cm (Fig. 3). However, it is possible that several patients with BMI < 27 kg/m2 were enrolled in Kumahara et al., as 99% (n = 113/114) of patients were < 180 cm in height and 90% (n = 102/114) weighed < 90 kg.

Fig. 3

Relationship between eligibility criteria for the semaglutide (STEP 6, Kadowaki et al. 2022) and mazindol (Kumahara et al. 1985) trials. BMI body mass index, STEP Semaglutide Treatment Effect in People with Obesity

Regarding population overlap, the demographics were substantially different between STEP 6 and Kumahara et al. with proportions of men of 63.4% and 27.2%, respectively [36, 37]. Additionally, as mentioned above, a few patients in Kumahara et al. were likely to have a lower BMI than the minimum BMI imposed in STEP 6.

Furthermore, the definition of “placebo” was different between STEP 6 and Kumahara et al. In STEP 6, D&E was concurrent with treatment with semaglutide or placebo, while no relevant intervention was reported during treatment with mazindol or placebo in Kumahara et al. [36, 37]. Nonetheless, both trials were restricted to eligible patients who had ≥ 1 unsuccessful dietary attempt to lose body weight (including those in whom diet therapy was not available in Kumahara et al.). Additionally, the STEP 6 trial was designed in a manner in which participants continued in the trial after treatment discontinuation with all data being included in the analysis; this was not the case for Kumahara et al. and this bias in trial design should also be considered.

Other Sources of Bias in the Analysis, from the Public Health Perspective

From a public health perspective, a substantially higher criterion for body weight in the approved indication of mazindol in Japan (≥ + 70% of degree of obesity, where obesity degree in % = (actual weight – standard weight) / standard weight × 100, or BMI ≥ 35 kg/m2) compared with Kumahara et al. would make indirect comparison in a Japanese public healthcare setting difficult (Supplementary Material Table S13) [37, 38].

Naïve Indirect Comparisons

As a population-adjusted ITC was not feasible, relative treatment effects between semaglutide and mazindol were estimated by naïve indirect comparison using the Bucher method [39]. Outcomes at Weeks 8 and 12 after the 16-week escalation dose of semaglutide or placebo in STEP 6 (i.e., at Weeks 20 and 28) as base case and scenario, respectively, were compared with the outcomes at Week 12 in Kumahara et al.

Semaglutide was associated with significantly greater body weight reduction compared with mazindol (at a 5% significance level) in both the base case and the scenario (Supplementary Material Tables S14 and S15). A greater reduction in SBP was noted with semaglutide versus mazindol, although this was not statistically significant in either the base case or the scenario.