Volume 25, Issue 9 p. 2535-2544
ORIGINAL ARTICLE
Open Access

Clinical outcomes associated with drugs for obesity and overweight: A systematic review and network meta-analysis of randomized controlled trials

Andrea Iannone PhD

Corresponding Author

Andrea Iannone PhD

Section of Gastroenterology, Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy

Andrea Iannone and Patrizia Natale contributed equally to this work.

Correspondence

Andrea Iannone, PhD, Section of Gastroenterology, Department of Emergency and Organ Transplantation, University of Bari, Piazza Giulio Cesare 11, 70124, Bari, Italy.

Email: ianan@hotmail.it

Search for more papers by this author
Patrizia Natale PhD

Patrizia Natale PhD

Sydney School of Public Health, University of Sydney, Sydney, Australia

Department of Precision and Regenerative Medicine and Ionian Area (DIMEPRE-J), University of Bari Aldo Moro, Bari, Italy

Nephrology, Dialysis and Transplantation Unit, Department of Medical and Surgical Sciences, University of Foggia, Foggia, Italy

Andrea Iannone and Patrizia Natale contributed equally to this work.

Search for more papers by this author
Suetonia C. Palmer PhD

Suetonia C. Palmer PhD

Department of Medicine, University of Otago, Christchurch, New Zealand

Search for more papers by this author
Antonio Nicolucci PhD

Antonio Nicolucci PhD

Centre for Outcomes Research and Clinical Epidemiology (CORESEARCH), Pescara, Italy

Search for more papers by this author
Maria Rendina PhD

Maria Rendina PhD

Section of Gastroenterology, Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy

Search for more papers by this author
Francesco Giorgino PhD

Francesco Giorgino PhD

Section of Internal Medicine, Endocrinology, Andrology and Metabolic Diseases, Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy

Search for more papers by this author
Luigi Laviola PhD

Luigi Laviola PhD

Section of Internal Medicine, Endocrinology, Andrology and Metabolic Diseases, Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy

Search for more papers by this author
Alfredo Di Leo PhD

Alfredo Di Leo PhD

Section of Gastroenterology, Department of Emergency and Organ Transplantation, University of Bari, Bari, Italy

Search for more papers by this author
Giovanni F. M. Strippoli PhD

Giovanni F. M. Strippoli PhD

Sydney School of Public Health, University of Sydney, Sydney, Australia

Department of Precision and Regenerative Medicine and Ionian Area (DIMEPRE-J), University of Bari Aldo Moro, Bari, Italy

Search for more papers by this author
First published: 31 May 2023

Abstract

Aim

To compare the benefits and harms of drugs approved for weight management in adults with obesity or overweight.

Materials and Methods

We performed a systematic review of drugs approved for treating obesity and overweight. We searched MEDLINE, Embase and CENTRAL through 26 February 2023. Random-effects network meta-analysis was applied.

Results

A total of 168 trials (97 938 patients) were included. There was no evidence that drugs approved for weight management had different associations with cardiovascular death (69 trials, 59 037 participants). Naltrexone/bupropion was associated with lower cardiovascular mortality than placebo (odds ratio [OR], 0.62 [95% CI: 0.39, 0.99]; low certainty evidence). All drugs were associated with greater weight loss at 12 months than placebo (33 trials, 37 616 participants), mainly semaglutide (mean difference [MD], −9.02 kg [95% CI: −10.42, −7.63]; moderate certainty) and phentermine/topiramate (MD, −8.10 kg [95% CI: −10.14, −6.05]; high certainty); and with greater waist circumference reduction at 12 months than placebo (24 trials, 35 733 participants), mainly semaglutide (MD, −7.84 cm [95% CI: −9.34, −6.34]; moderate certainty) and phentermine/topiramate (MD, −6.20 cm [95% CI: −7.46, −4.94]; high certainty). Semaglutide and phentermine/topiramate were associated with lower or no difference in the odds of treatment withdrawal compared with all other drugs (87 trials, 70 860 participants).

Conclusions

Among adults with obesity or overweight, semaglutide and phentermine/topiramate were associated with greater body weight loss and waist circumference reduction at 12 months than all other drugs, and lower or no significant difference in risks of withdrawal. There was no evidence that drugs approved for weight management had different associations with cardiovascular death.

1 INTRODUCTION

Obesity and overweight represent a substantial burden on healthcare systems worldwide.1 The number of people with obesity is 650 million; more than 1.9 billion adults experience overweight,2 with a higher prevalence in low- and middle-income countries.3 Obesity and overweight increase the risks of non-communicable diseases, including type 2 diabetes, hypertension, cardiovascular disease, kidney failure and malignancy,4-7 and are associated with premature death and disability-affected life.8, 9 Diet, exercise or other lifestyle changes may improve clinical outcomes, although the lack of adherence with frequent occurrence of weight regain limits their efficacy in providing sustained weight loss.10

Several pharmacological interventions have been developed to manage obesity and overweight, including orlistat, liraglutide, semaglutide, phentermine/topiramate, naltrexone/bupropion and lorcaserin.11 Clinical practice guidelines suggest that tailored and multidisciplinary approaches, based on both lifestyle modifications and pharmacological interventions, might be effective in reducing weight loss and preventing cardiovascular complications.12, 13 However, current evidence is limited by few available head-to-head studies to assess the comparative benefits and harms of available drugs.14 Standard pairwise meta-analysis only allows the direct comparison of two drugs that have been directly evaluated in randomized trials. In a setting in which several pharmacological options for the treatment of obesity are available, a network meta-analysis enables comparisons of the different drugs simultaneously, to identify the most effective and safe therapies.

We conducted a systematic review with network meta-analysis of randomized controlled trials to compare the benefits and harms of weight-lowering drugs in adults with obesity or overweight.

2 RESEARCH DESIGN AND METHODS

2.1 Study design

We registered the protocol of this systematic review prior to conduct (PROSPERO CRD42021250646). The study protocol is available at https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=250646. The study was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines for systematic reviews with network meta-analysis.15

2.2 Search strategy and selection criteria

A highly sensitive search strategy was conducted in MEDLINE, Embase and the Cochrane Central Register of Controlled Trials (CENTRAL), from inception to 26 February 2023, without language restriction (Table S1).

Randomized controlled trials were eligible if they evaluated drugs to lower body weight in adults with obesity (body mass index [BMI] ≥ 30 kg/m2) or overweight (BMI 25-29.9 kg/m2).2 Comparisons of the following drugs against each other, placebo, or standard management were considered: orlistat, liraglutide, semaglutide, phentermine/topiramate, naltrexone/bupropion and lorcaserin. We only included trials in which liraglutide was administered at 3 mg/day, because lower doses have an indication only for diabetes. Only drugs specifically approved for the treatment of obesity were selected for inclusion in the review. Drugs approved only for the treatment of diabetes were excluded, even if a partial effect on body weight lowering has been shown for some of them. Crossover studies, trials investigating withdrawn drugs and those primarily estimating pharmacokinetics and pharmacodynamics, were excluded.

Two investigators (AI, PN) independently screened in duplicate the titles and abstracts of the retrieved search records to identify potentially eligible trials. Differences were resolved by consensus. Any potentially relevant citation was retrieved in full text and reviewed by the same two investigators to assess eligibility. Differences were resolved in discussion with a third investigator (SCP).

2.3 Data extraction

Two investigators (AI, PN) independently extracted data in duplicate from each eligible study using a standardized extraction form, including study characteristics (year of publication, country, duration, funding, trial registration number), participant characteristics (setting, sample size, age, sex, BMI, co-morbidities), description of intervention (type of drug, dose, route, frequency) and outcomes.

The primary outcome was cardiovascular death. The secondary outcomes were all-cause death; non-fatal stroke; non-fatal myocardial infarction; onset of diabetes; body weight loss (kg), assessed at 6, 12, 24 and 60 months; waist circumference reduction (cm), assessed at 6, 12, 24 and 60 months; body weight loss (%), assessed at 6, 12, 24 and 60 months; withdrawal from treatment because of adverse events; serious gastrointestinal adverse events leading to treatment withdrawal, hospitalization or death; BMI (kg/m2), assessed at 6, 12, 24 and 60 months; systolic and diastolic blood pressure; lipid profile (total cholesterol, low-density lipoprotein [LDL] cholesterol, high-density lipoprotein [HDL] cholesterol; HbA1c for people with type 1 diabetes and type 2 diabetes; and health-related quality of life), assessed with any measure.

2.4 Risk of bias and quality assessment

Two investigators (AI, PN) assessed the risk of bias in the included studies using the Cochrane tool.16 Evidence certainty was assessed using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach for network meta-analysis to critically appraise the certainty of evidence for all evaluated outcomes.17 For this purpose, we used the Confidence in Network Meta-Analysis (CINeMA) web application.18 CINeMA considers six domains (i.e. within-study bias, reporting bias, indirectness, imprecision, heterogeneity and incoherence) and provides the contribution matrix, which shows how much information each study contributes to the results from network meta-analysis.18

2.5 Statistical analysis

Treatment effects were first estimated by random-effects pairwise meta-analysis. We reported treatment effects as an odds ratio (OR) for binary outcomes (cardiovascular death, all-cause death, non-fatal stroke, non-fatal myocardial infarction, onset of diabetes, withdrawal from treatment because of adverse events, serious gastrointestinal adverse events), mean difference (MD) for continuous outcomes reported on the same scale (body weight, waist circumference, BMI, systolic and diastolic blood pressure, HbA1c), or standardized mean difference (SMD) when continuous outcomes were reported on different scales (total cholesterol, LDL cholesterol, HDL cholesterol and quality of life), together with corresponding 95% confidence intervals (95% CIs).

Frequentist network meta-analysis was then performed to compare available treatment strategies within a single analytical framework.19, 20 For each network, the extent of heterogeneity was assessed using the maximum likelihood method. We examined the magnitude of a common heterogeneity variance for the network (tau squared [τ2]) as an indicator of the extent of heterogeneity among included studies in terms of the range of expected treatment estimates (risk ratios).21 We considered τ2 values less than 0.25 as low heterogeneity, values between 0.25 and 1.0 as moderate heterogeneity, and values greater than 1.0 as substantial heterogeneity.21-23 We evaluated the transitivity assumption of indirect comparisons by assessing the distribution of study and population characteristics across trial arms grouped by interventions.

To explore for evidence of within-network inconsistency, we used a loop-specific approach, which compared the estimated effects derived from direct and indirect evidence in all triangular and quadratic loops in a network.24 We assessed for evidence of small study effects by inspecting the symmetry of funnel plots for each outcome (comparison-adjusted). We evaluated for global network consistency using the ‘design-by-treatment’ interaction method.25

Sensitivity analyses were performed by follow-up duration (i.e. considering trials with a follow-up duration longer than 6 months), BMI category (i.e. considering trials including only obese people) and semaglutide dose (i.e. considering trials in which semaglutide was administered at 2.4 mg/week, the dose approved for the treatment of obesity). STATA version 16 software (StataCorp, College Station, TX) was used to conduct network meta-analysis using published routines.26, 27

3 RESULTS

The electronic search retrieved 6633 citations and five additional records were detected through hand searching (Figure 1). Overall, 168 trials involving 97 938 participants were eligible for inclusion (Table S2). Drugs were allocated in addition to lifestyle modification.

Details are in the caption following the image
Summary of study retrieval and identification for network meta-analysis. RCT, randomized controlled trial

Sixty-five (38.7%) trials were performed in the United States, 50 (29.8%) in Europe, 12 (7.1%) in Asia and four (2.4%) in Central or South America, while the remaining 37 (22.0%) enrolled participants from different continents. The median age of participants was 46.9 years, and the median proportion of males was 25.5%. At baseline, the median BMI of participants was 35.8 kg/m2, and 99 (58.9%) studies included participants with overweight and obesity. Forty-three (25.6%) trials included people with diabetes, and 41 (24.4%) patients with hypertension. The median trial follow-up duration was 12 months, and 35 (20.8%) studies had a follow-up lower than 6 months. The study and population characteristics were comparable across trials, with no evidence of relevant differences in the risk of cardiovascular outcomes. There was no evidence of transitivity assumption violation.

3.1 Risk of bias

Sixty (35.7%) of 168 trials were adjudicated as being at low risk of bias for random sequence generation and 51 (30.4%) were at low risk of bias for allocation concealment (Table S3 and Figure S1). Seventy-three studies (43.5%) reported blinding of participants and investigators, and 63 trials (37.5%) reported blinding of outcome assessment. Fifty-eight studies (34.5%) were judged to be at low risk of bias for incomplete outcome data, and 62 trials (36.9%) were at low risk of bias for selective outcome reporting.

3.2 Network consistency

The network for each outcome is shown in Figures S2–S19. There was evidence of global inconsistency only in the network of percentage body weight loss at 12 months (P = .03) (Table S4). Direct and indirect drug comparisons were generally consistent (Figure 1 and Tables S5–S21). There was evidence of loop-specific inconsistency between direct and indirect evidence only for the outcome of percentage body weight loss at 12 months (Figures S20–S37). Comparison-adjusted funnel plots showed no evidence of small-study effects (Figures S38–S55).

3.3 Treatment outcomes

3.3.1 Primary outcome

Sixty-nine trials, including 59 037 participants, reported the primary outcome of cardiovascular death in 346 (0.6%) patients (178 events using placebo and 168 using weight-lowering drugs). There was no significant difference in the associations between any drug with odds of cardiovascular death (Table 1). However, naltrexone/bupropion was associated with lower risks of cardiovascular death when compared with placebo (OR, 0.62 [95% CI: 0.39 to 0.99]) (Figure 2, Table 2). Rankings of drugs for cardiovascular mortality were uninformative (Figure 3).

TABLE 1. Direct and network estimated odds ratios (ORs) of treatments for obesity on cardiovascular mortality
Network estimate
DIRECT COMPARISON Semaglutide 1.47 (0.28, 7.74) 1.48 (0.81, 2.73) 1.02 (0.40, 2.63) 1.05 (0.41, 2.68) 0.89 (0.55, 1.46) 1.01 (0.15, 7.04) 0.92 (0.62, 1.37)
- PHEN/TPM 1.01 (0.19, 5.40) 0.70 (0.11, 4.32) 0.72 (0.12, 4.45) 0.61 (0.12, 3.12) 0.69 (0.06, 8.31) 0.63 (0.13, 3.15)
- - NB 0.69 (0.26, 1.82) 0.71 (0.27, 1.89) 0.60 (0.35, 1.04) 0.68 (0.10, 4.70) 0.62 (0.39, 0.99)
- - - Orlistat 1.03 (0.31, 3.39) 0.87 (0.35, 2.15) 0.99 (0.17, 5.85) 0.90 (0.38, 2.12)
0.38 (0.02, 6.12) - - 0.97 (0.02,49.32) Liraglutide 0.85 (0.34, 2.10) 0.96 (0.12, 7.66) 0.88 (0.37, 2.07)
- - - - - Lorcaserin 1.14 (0.17, 7.75) 1.03 (0.77, 1.39)
- - 0.58 (0.01, 29.64) 1.00 (0.14, 7.45) - - Standard care 0.91 (0.14, 6.07)
0.93 (0.63, 1.38) 0.63 (0.13, 3.15) 0.62 (0.39, 1.00) 0.90 (0.37, 2.16) 0.82 (0.34, 1.98) 1.03 (0.77, 1.39) - Placebo
  • Note: Values are given as OR (95% confidence interval). The grid should be read from left to right. The lower part of the grid reports estimates for direct comparison and risk estimate is for the column-defining intervention compared with the row-defining intervention. An OR < 1 indicates the column treatment is associated with lower odds of cardiovascular mortality than the row treatment. The upper part of the grid reports network estimates and the risk estimate is for the row-defining intervention compared with the column-defining intervention. An OR < 1 indicates the row treatment is associated with lower odds of cardiovascular mortality than the column treatment.
  • Note: There are 69 trials involving 59 037 participants in this network.
  • Note: Statistically significant estimates are in bold.
  • Abbreviations: NB, naltrexone/bupropion; PHEN/TPM, phentermine/topiramate.
Details are in the caption following the image
Forest plots of network estimates of the comparison between treatments for obesity and placebo on A, Cardiovascular mortality, B, Body weight loss at 12 months, and C, Waist circumference reduction at 12 months. CI, confidence interval; MD, mean difference; OR, odds ratio. Mean difference estimates for body weight loss at 12 months are expressed in kg; mean difference estimates for waist circumference reduction at 12 months are expressed in cm
TABLE 2. Summary of certainty in network intervention estimates for cardiovascular mortality, body weight loss at 12 months and waist circumference reduction at 12 months
Outcome and comparison Certainty in evidence Reasons for downgrading certainty in evidence Network estimatea
Cardiovascular mortality
Semaglutide versus placebo

Very low

●◯◯◯

Downgrade 3 levels for study limitations (−2) and imprecision (−1) 0.92 (0.62, 1.37)
PHEN/TPM versus placebo

Very low

●◯◯◯

Downgrade 3 levels for study limitations (−1) and imprecision (−2) 0.63 (0.13, 3.15)
NB versus placebo

Low

●●◯◯

Downgrade 2 levels for study limitations (−1) and inconsistency (−1) 0.62 (0.39, 0.99)
Orlistat versus placebo

Very low

●◯◯◯

Downgrade 3 levels for study limitations (−1) and imprecision (−2) 0.90 (0.38, 2.12)
Liraglutide versus placebo

Very low

●◯◯◯

Downgrade 3 levels for study limitations (−1) and imprecision (−2) 0.88 (0.37, 2.07)
Lorcaserin versus placebo

Low

●●◯◯

Downgrade 2 levels for study limitations (−1) and imprecision (−1) 1.03 (0.77, 1.39)
Standard care versus placebo

Very low

●◯◯◯

Downgrade 3 levels for study limitations (−1) and imprecision (−2) 0.91 (0.14, 6.07)
Body weight loss at 12 mo, kg
Semaglutide versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −9.02 (−10.42, −7.63)
PHEN/TPM versus placebo

High

●●●●

No reasons for downgrading −8.10 (−10.14, −6.05)
NB versus placebo

High

●●●●

No reasons for downgrading −4.53 (−5.92, −3.14)
Orlistat versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −2.36 (−3.03, −1.69)
Liraglutide versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −5.01 (−5.95, −4.07)
Lorcaserin versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −2.75 (−3.69, −1.81)
Standard care versus placebo

Low

●●◯◯

Downgrade 2 levels for study limitations (−1) and imprecision (−1) 1.07 (−0.98, 3.11)
Waist circumference reduction at 12 mo, cm
Semaglutide versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −7.84 (−9.34, −6.34)
PHEN/TPM versus placebo

High

●●●●

No reasons for downgrading −6.20 (−7.46, −4.94)
NB versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −3.50 (−4.45, −2.55)
Orlistat versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −1.98 (−2.98, −0.98)
Liraglutide versus placebo

Moderate

●●●◯

Downgrade 1 level for study limitations (−1) −3.71 (−4.46, −2.96)
Lorcaserin versus placebo

High

●●●●

No reasons for downgrading −2.26 (−3.08, −1.44)
Standard care versus placebo

Low

●●◯◯

Downgrade 2 levels for study limitations (−1) and imprecision (−1) 1.64 (−1.90, 5.19)
  • Note: Statistically significant estimates are in bold.
  • Abbreviations: NB, naltrexone/bupropion; PHEN/TPM, phentermine/topiramate.
  • a Expressed as odds ratio (95% CI) for cardiovascular mortality, and as mean difference (95% CI) for body weight loss at 12 mo and waist circumference reduction at 12 mo.
Details are in the caption following the image
Rankings of treatments for obesity for cardiovascular mortality, body weight loss (kg) at 12 months, waist circumference reduction (cm) at 12 months and treatment withdrawal because of adverse events. The lines show the probability of the treatment ranking between best (= 1) and worst (= 8) for each outcome, and the peak indicates the ranking with the highest probability for the corresponding treatment

3.3.2 Secondary outcomes

There was no significant difference in the associations between any drug for all-cause death (73 trials, 60 066 participants, 703 events), non-fatal stroke (28 trials, 38 207 participants, 313 events) or non-fatal myocardial infarction (29 trials, 39 691 participants, 722 events) (Tables S5–S7).

Liraglutide (OR, 0.32 [95% CI: 0.20 to 0.51]), phentermine/topiramate (OR, 0.57 [95% CI: 0.33 to 0.96]), orlistat (OR, 0.65 [95% CI: 0.51 to 0.84]) and lorcaserin (OR, 0.78 [95% CI: 0.63 to 0.96]) were associated with lower risks of diabetes onset compared with placebo (Table S8). Among drug comparisons, lorcaserin (OR, 2.42 [95% CI: 1.47 to 4.01]) and orlistat (OR, 2.02 [95% CI: 1.20 to 3.40]) showed higher risks of diabetes onset than liraglutide.

The analyses for body weight and waist circumference are reported at 12 months. The results of the 6-month analyses were consistent with those at 12 months but were reported in a significantly smaller number of studies, while there were only a few or no trials assessing these outcomes at other prespecified time points (i.e. 24 and 60 months). All drugs were associated with greater body weight loss compared with placebo at 12 months, excepting standard care: semaglutide (MD, −9.02 kg [95% CI: −10.42 to −7.63]), phentermine/topiramate (MD, −8.10 kg [95% CI: −10.14 to −6.05]), liraglutide (MD, −5.01 kg [95% CI: −5.95 to −4.07]), naltrexone/bupropion (MD, −4.53 kg [95% CI: −5.92 to −3.14]), lorcaserin (MD, −2.75 kg [95% CI: −3.69 to −1.81]) and orlistat (MD −2.36 kg [95% CI: −3.03 to −1.69]) (Figure 2, Table 2). The analysis of percentage body weight loss confirmed the results of absolute body weight loss at 12 months (Table S10). Semaglutide and phentermine/topiramate were associated with greater body weight loss at 12 months than all other drugs (Table S9). All drugs were associated with greater waist circumference reduction compared with placebo at 12 months, excepting standard care: semaglutide (MD, −7.84 cm [95% CI: −9.34 to −6.34]), phentermine/topiramate (MD, −6.20 cm [95% CI: −7.46 to −4.94]), liraglutide (MD, −3.71 cm [95% CI: −4.46 to −2.96]), naltrexone/bupropion (MD, −3.50 cm [95% CI: −4.45 to −2.55]), lorcaserin (MD, −2.26 cm [95% CI: −3.08 to −1.44]) and orlistat (MD, −1.98 cm [95% CI: −2.98 to −0.98]) (Figure 2, Table 2). Semaglutide and phentermine/topiramate were associated with greater waist circumference reduction at 12 months compared with all other drugs (Table S11). Semaglutide was ranked highest for lowering body weight and reducing waist circumference at 12 months (Figure 3). Semaglutide was associated with a larger BMI reduction at 12 months compared with other interventions (Table S12). Semaglutide, phentermine/topiramate and liraglutide were associated with greater systolic and diastolic blood pressure reduction compared with placebo. Semaglutide was ranked highest for lowering systolic and diastolic blood pressure (Tables S13 and S14). Orlistat was associated with greater total (SMD, −0.75 [95% CI: −1.10 to −0.40]) and LDL (SMD, −1.05 [95% CI: −1.64 to −0.45]) cholesterol reduction compared with placebo. There was no significant difference in the associations between any drug class and lipid levels (Tables S15–S17). Semaglutide (MD, −0.58% [95% CI: −0.81 to −0.34]), liraglutide (MD, −0.43% [95% CI: −0.59 to −0.26]) and orlistat (MD, −0.31% [95% CI: −0.51 to −0.12]) were associated with a greater HbA1c reduction compared with placebo (Table S18). Naltrexone/bupropion was associated with a significantly higher health-related quality of life compared with placebo (SMD, 2.06 [95% CI: 0.55 to 3.58]). There was no significant difference in the associations between any drug class and quality of life (Table S19).

All drugs were associated with higher odds of withdrawal from treatment because of adverse events compared with placebo, excepting standard care (Figure 3). Naltrexone/bupropion was associated with higher odds of withdrawal than other interventions (Table S20). Naltrexone/bupropion was associated with higher risks of serious gastrointestinal adverse events than other pharmacological interventions (Table S21).

3.4 Sensitivity analyses

The results of the sensitivity analyses by BMI category (i.e. considering trials including only obese people) are reported in Tables S36–S51. The sensitivity analysis for cardiovascular mortality showed no significant difference between naltrexone/bupropion and placebo in obese people (Table S36). The sensitivity analysis for body weight loss showed no significant difference between lorcaserin and placebo in obese people (Table S41).

No evidence was found that treatment effects differed according to follow-up duration (i.e. considering trials with a follow-up duration longer than 6 months) (Tables S22–S35) or semaglutide dose (i.e. considering trials in which semaglutide was administered at 2.4 mg/week) (Tables S52–S69).

4 DISCUSSION

In adults with obesity or overweight, there is no evidence that drugs approved for weight management have different effects on cardiovascular death, all-cause death, non-fatal myocardial infarction or non-fatal stroke. Combined treatment with naltrexone/bupropion is associated with lower cardiovascular mortality when compared with placebo. Semaglutide and phentermine/topiramate were associated with 8-9 kg of body weight loss and 6-7 cm of waist circumference reduction at 12 months compared with placebo and were more effective than all other drugs, with similar or lower risks of withdrawal from treatment because of adverse events. At 12 months, semaglutide was ranked highest for lowering body weight and reducing waist circumference and blood pressure. Taken together, in adults with obesity or overweight, semaglutide appears to show favourable associations with lowering body weight, as well as reducing waist circumference and blood pressure, without excess withdrawal from adverse events. Based on current trials, there is no evidence that semaglutide or other weight-lowering drugs have different associations with cardiovascular complications or death, and suggest that clinicians and patients may select drug management based on their efficacy related to body weight loss and waist circumference reductions.

Most trials evaluating approved treatments for weight management in people with overweight and obesity provide weight management and patient outcome data during 12 months of treatment. It is possible that the existing trial evidence cannot infer whether body weight reductions translate to the lowering of cardiovascular death and complications until a substantially longer follow-up (over several years) in a sufficiently large group of participants is available. Notably, trials of sodium-glucose cotransporter-2 (SGLT-2) inhibitors and glucagon-like peptide 1 (GLP-1) receptor agonists have shown substantial and consistent improvements in survival and prevention of cardiovascular complications among people with type 2 diabetes.28 These findings suggest that the same benefits might be obtained in people with obesity. Several cardiovascular outcome trials for weight management are scheduled for completion in 2023 and 2024, including SELECT, SURPASS CVOT, STEP-HFpEF, SUMMIT and LOSE-AF.

A network meta-analysis previously indicated that phentermine/topiramate and liraglutide were the most effective drugs in reducing body weight, although they were associated with higher odds of adverse event-related treatment discontinuation compared with placebo, liraglutide and naltrexone-bupropion.29 More recently, a systematic review and network meta-analysis of 143 trials and 49 810 participants assessed the benefits and harms of weight-lowering drugs, also including drugs without any specific indication for the treatment of obesity, such as metformin, SGLT-2 inhibitors or levocarnitine.30 The study documented that phentermine/topiramate was the most effective in lowering weight, followed by GLP-1 receptor agonists. In a post hoc analysis, semaglutide showed substantially larger benefits than other drugs for both probable body weight loss of 5% or greater and percentage body weight change. The risk of adverse events with semaglutide was similar to that with other drugs.

Our systematic review of 168 trials involving 97 938 participants provides additional information, expanding the assessment of the comparative efficacy and safety of drugs approved for weight management to patient-centred priority outcomes including all-cause death, major cardiovascular events, waist circumference, progression to diabetes, health-related quality of life and treatment withdrawal. These additional aspects provide a comprehensive picture of the comparative efficacy and safety/tolerability of the different drugs, and show that semaglutide ranked highest, not only for lowering body weight and reducing waist circumference at 12 months, but also for reducing blood pressure. Semaglutide was also associated with a significantly lower HbA1c than other interventions. The ongoing SELECT trial will further elucidate whether these benefits will translate into a reduction in major cardiovascular events.31

The strength of this systematic review includes the comparison of approved pharmacological approaches for weight management in adults with obesity or overweight. The systematic review had limitations that should be considered when interpreting the results. Many randomized trials included in this review had a high methodological risk of bias that may reduce the certainty we can have in the resulting treatment estimates. Primary trials were designed to estimate the effects of surrogate metabolic and weight over several months. Cardiovascular death was included as the primary outcome in this review, although studies were not powered to evaluate this endpoint, and most included few or no events. Tirzepatide has been included recently as a pharmacological treatment for weight management,32 but was beyond the scope of this review.

In conclusion, semaglutide and phentermine/topiramate were associated with greater body weight loss and waist circumference reduction at 12 months than all other drugs, and lower or no significant difference in risks of withdrawal. Trials evaluating cardiovascular outcomes of pharmacological interventions for weight management in people with obesity or overweight are awaited, as current evidence for these endpoints is limited because of short-term follow-up in the existing trials.

AUTHOR CONTRIBUTIONS

A.I.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, software, validation, visualization, writing: original draft, review and editing; P.N.: data curation, formal analysis, investigation, methodology, project administration, software, validation, visualization, writing: original draft, review and editing; S.C.P.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing: original draft, review and editing; A.N.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing: original draft, review and editing; M.R.: validation, visualization, writing: original draft, review and editing; F.G.: validation, visualization, writing: original draft, review and editing; L.L.: validation, visualization, writing: original draft, review and editing; A.D.L.: validation, visualization, writing: original draft, review and editing; G.F.M.S.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing: original draft, review and editing. A.I. and P.N. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

ACKNOWLEDGEMENTS

We thank Ema Avdic (Faculty of Medicine, University of Sarajevo, Sarajevo, Bosnia and Herzegovina) for providing translations for foreign studies. Open access publishing facilitated by The University of Sydney, as part of the Wiley - The University of Sydney agreement via the Council of Australian University Librarians.

    CONFLICT OF INTEREST

    All authors declare no conflicts of interest.

    PEER REVIEW

    The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.15138.

    DATA AVAILABILITY STATEMENT

    The data that support the findings of this study are available on request from the corresponding author.