Reimbursement Review

Semaglutide (Wegovy)

Sponsor: Novo Nordisk Canada Inc.

Therapeutic area: Weight management

This multi-part report includes:

Clinical Review

Pharmacoeconomic Review

Clinical Review

Abbreviations

Executive Summary

An overview of the resubmission details for the drug under review is provided in Table 1.

Table 1: Background Information of Application Submitted for Review

BMI = body mass index; MI = myocardial infarction; NOC = Notice of Compliance; PAD = peripheral arterial disease.

^aThe NOC date is for the indication under review. Of note, this is distinct from the NOC issued in 2024 for Wegovy to reduce the risk of nonfatal MI in adults with established cardiovascular disease and BMI of 27 kg/m² or greater.

Sources: Product monograph for semaglutide injection (Wegovy).¹ Details included in the table are from the sponsor’s summary of clinical evidence.²

Submission History

Initial Submission

Semaglutide 2.4 mg was first reviewed by the Canadian Drug Expert Committee (CDEC) for weight management in 2022. CDEC issued a recommendation that semaglutide 2.4 mg not be reimbursed as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with an initial body mass index (BMI) of 30 kg/m² or greater (obesity) or 27 kg/m² or greater (overweight) in the presence of at least 1 weight-related comorbidity such as hypertension, type 2 diabetes mellitus, dyslipidemia, or obstructive sleep apnea.³

The 2022 reimbursement recommendation for semaglutide (Wegovy) and the Clinical Review report on semaglutide (Wegovy) that was used to inform the recommendation are both available on the Canada’s Drug Agency (CDA-AMC) project website.

Basis of Resubmission

The rationale for the 2022 recommendation included evidence from 4 placebo-controlled, double-blind, randomized controlled trials (RCTs) (STEP 1, STEP 2, STEP 3, and STEP 4). These trials demonstrated that treatment with semaglutide 2.4 mg injection reduced body weight in individuals with an initial BMI of 30 kg/m² or greater (obesity) or 27 kg/m² or greater (overweight) in the presence of at least 1 weight-related comorbidity but did not demonstrate improvement in or prevention of weight-related comorbidities. Comorbidities such as a major adverse cardiovascular event (MACE), osteoarthritis, and obstructive sleep apnea were not outcomes assessed in the STEP trials.³

CDEC noted there was an ongoing trial, the SELECT study,⁴ comparing semaglutide 2.4 mg injection with placebo for the prevention of MACE occurrences in patients with overweight or obesity who have established cardiovascular (CV) disease but not diabetes mellitus. CDEC concluded in 2022 that the results of that study, once completed, would address the evidence gap regarding the effects of semaglutide 2.4 mg once weekly on CV outcomes in the indicated population.³

Therefore, the objective of this report is to review and critically appraise the clinical evidence submitted by the sponsor on the beneficial and harmful effects of semaglutide 2.4 mg solution for subcutaneous injection in chronic weight management in patients with a BMI of 27 kg/m² or greater and established CV disease (myocardial infarction [MI], stroke, or peripheral arterial disease [PAD]).

Sponsor’s Clarifying Note on Resubmission

In 2024, Health Canada issued an indication for semaglutide injection (Wegovy) to reduce the risk of nonfatal MI in adults with established CV disease and a BMI of 27 kg/m² or greater. This indication was based on the evidence from the SELECT trial.

The sponsor indicated that the population for the revised reimbursement request, namely, adults with established CV disease and a BMI of 27 kg/m² or greater, is fully encompassed in the weight management indication and is the same population for which the CV indication was granted.

Introduction

Disease Background

Obesity is a chronic disease where excess body fat impairs health and increases the risk of long-term health complications.⁵ The presence of overweight (BMI of greater than 25 kg/m²) and obesity (BMI of greater than 30 kg/m²) are considered major contributors to CV disease progression through direct and indirect mechanisms.^6-15 In 2022, the estimated prevalence of obesity was 30% of individuals living in Canada aged 18 years or older.^16,17 In a study of patients with atherosclerotic cardiovascular disease (ASCVD), obesity was among the 30 most common comorbid conditions, with a prevalence of 38%.¹⁸ Worldwide, CV disease is among the leading causes of morbidity and mortality and is reported to be one of the leading causes of hospitalization in Canada, alongside stroke.^19,20

Standards of Therapy

The Canadian obesity guideline states there are 4 medications indicated for long-term obesity management as adjuncts to health-behaviour changes: liraglutide, naltrexone-bupropion in a combination tablet, orlistat, and semaglutide. The Canadian guideline recognizes all 4 medications as effective in producing clinically significant weight loss and health benefits relative to placebo over a period of at least 1 year.²¹ The clinical experts indicated that another medication, tirzepatide, is currently approved only for adjunct glycemic control in type 2 diabetes but is increasingly being prescribed off label for chronic weight management in Canada. Further, patient groups have indicated that semaglutide (Ozempic) has been prescribed off label for weight management as well. The clinical experts indicated that in clinical practice in Canada, semaglutide is being used in patients with diabetes for weight loss.

Perspectives of Patient, Clinicians, and Drug Programs

The information in this section is a summary of the input provided by the patient and clinician groups that responded to the call for input and from the clinical experts consulted for the purpose of this review.

Patient Input

Input for this review was submitted by 6 patient groups: GI Society, Obesity Canada, Obesity Matters, Fatty Liver Alliance, HeartLife Foundation, and Diabetes Canada. Input from the GI Society, Obesity Canada, and Diabetes Canada was based on surveys of individuals with obesity or diabetes conducted between 2021 and 2024. The GI Society also included input from an in-person focus group of individuals with obesity held in 2023. Input from the HeartLife Foundation was gathered through interviews and discussions with individuals with CV diseases and their health care providers. The Fatty Liver Alliance gathered physician insights on metabolic dysfunction–associated steatotic liver disease (MASLD) and metabolic dysfunction–associated steatohepatitis (MASH), supplemented by a 2022 to 2023 US survey of patients with nonalcoholic fatty liver disease or nonalcoholic steatohepatitis. Obesity Matters also provided feedback on the CDEC reimbursement recommendation issued for the initial 2022 review of semaglutide for weight management.

Patient groups highlighted the significant physical and mental health impact of obesity and related conditions, including CV diseases, diabetes, MASLD, and MASH. Common symptoms included mobility limitations, chronic pain, fatigue, swelling, shortness of breath, dizziness, and nausea, affecting daily life. Comorbidities were prevalent: in the GI Society’s 2021 survey, 91% of those who responded reported comorbidities such as arthritis (51%), hypertension (33%), sleep apnea (30%), gastroesophageal reflux disease (GERD) (29%), irritable bowel syndrome (29%), high cholesterol (25%), and diabetes (24%). Input from Diabetes Canada also noted high comorbidity rates for patients with diabetes, including high blood pressure, foot or eye problems, nerve damage, heart disease, and kidney-related comorbidities. Obesity not only contributes to these conditions but also complicates disease management and reduces quality of life. Mental health challenges, including anxiety, mood disorders, and social isolation due to stigma, especially in health care, were widely reported.

The patient groups highlighted weight loss as a key treatment outcome, along with reduced risk of comorbidities (e.g., CV disease), improvements in quality of life, and increased ability to perform everyday tasks. When choosing a new therapy, most patients primarily value efficacy (i.e., significant and sustained weight loss) and affordability.

Patient groups emphasized that traditional lifestyle modifications, such as diet and exercise, are often ineffective for long-term weight management. Few medication options exist, with semaglutide, liraglutide, and naltrexone-bupropion being commonly prescribed, but none have public or full private coverage, making cost the biggest barrier. Those who responded to the survey and who reported benefits of these treatments in managing obesity described them as “life-changing,” with most (94%) considering the side effects to be manageable. However, it was also noted that these medications lack long-term effectiveness, with many individuals gaining back the weight that was lost within 5 years. Bariatric surgery, while currently considered the gold standard for obesity treatment, carries risks of severe side effects, postsurgery weight gain, and long wait times, making it a last-resort option.

Patients with experience receiving semaglutide described several benefits, including substantial weight loss. It was highlighted that semaglutide also improved the management of comorbidities, such as CV disease, and improved quality of life. Thus, it was identified that semaglutide has the ability to address multiple health conditions simultaneously. Patients with experience with semaglutide also highlighted that its side effects — most commonly nausea, vomiting, and constipation — were typically temporary and manageable, and the once-weekly administration reduced treatment burden.

Obesity Matters provided additional feedback on the 2022 draft CDEC reimbursement recommendation for semaglutide (Wegovy). The feedback indicated that while the initial draft recommendation acknowledged the effect of semaglutide on body-weight reduction and the increased risk of comorbidities observed with obesity, it did not conclude that weight reduction can directly reduce comorbidities. The feedback also critiqued the rationale for a negative reimbursement recommendation, noting this overlooks the agency of individuals actively working with specialists to make lifestyle changes. The feedback also indicated that the previous recommendation did not adequately consider cost as a barrier to treatment, patient perspectives, the direct impact of treatment costs on quality of life, or how public reimbursement would reduce the long-term costs associated with obesity-related comorbidities and conditions, nor did it offer actionable solutions for integrating semaglutide into existing health care systems.

Clinician Input

Input From Clinical Experts Consulted for This Review

The clinical experts identified the following as limitations with weight management medications: barriers to public access, that not all patients experience weight loss with treatment, and that some patients experience adverse effects that necessitate stopping the drug. Specific to bariatric surgery, the clinical experts noted that in addition to barriers to access, not all patients are interested in pursuing this option because of the potential risks associated with surgery.

According to the clinical experts consulted for this review, semaglutide 2.4 mg has been used for weight management in Canada since it was approved by Health Canada in 2021. The clinical experts indicated that the anticipated place of semaglutide with regard to pharmacotherapy for chronic weight management is as a first-line treatment. The clinical experts advised that semaglutide would be combined with therapies that reduce cardiorenal risk, including combination treatment with renin-angiotensin-aldosterone system (RAAS) inhibitors (such as angiotensin-converting enzyme [ACE] inhibitors and angiotensin 2 receptor blockers) and sodium-glucose cotransporter-2 (SGLT2) inhibitor therapies as well as other standards of care to improve CV outcomes. The clinical experts indicated there is currently no high-level evidence to support combination treatment with other pharmacotherapies for weight management; however, they noted there is relevant evidence from small studies (e.g., a retrospective cohort study on the effect of a combined glucagon-like peptide-1 [GLP-1] receptor agonist [GLP-1 RA] and bupropion-naltrexone in weight loss²²). The clinical experts indicated that, in practice, specialists in obesity treat patients with combination therapy when needed and appropriate and when cost and insurance coverage are not limiting factors. The clinical experts indicated that the most common combination is a GLP-1 RA combined with bupropion-naltrexone.

The clinical experts deferred to the baseline characteristics of the SELECT trial population to identify patients who would most likely respond to treatment with semaglutide, where response is defined as reduced risk of future CV outcomes and weight loss. Of note, the clinical experts highlighted that the trial population included patients with a pre-existing history of CV disease, and 64.5% of trial participants had prediabetes (hemoglobin A1C > 5.7% per US criteria). Overall, the clinical experts advised that it would be appropriate to treat most patients with a BMI of 27 kg/m² or greater for the chronic weight management indication, and that patients with established CV disease (including heart failure [HF]), type 2 diabetes, multiple risk factors, severe disease, or a high BMI are most in need of an intervention for the secondary prevention of future events and to improve symptom burden.

When considering which patients would be least suitable for treatment with semaglutide, the clinical experts deferred to the product monograph for absolute contraindications and patients with a potential increased risk of adverse effects (e.g., history of pancreatitis, gallbladder disease, diabetic retinopathy, severe renal impairment, gastrointestinal conditions).

Although weight loss is an outcome that can be easily monitored in practice, the clinical experts advised that monitoring the recurrence of CV events and symptoms is equally as important. The clinical experts further noted that patient-reported outcome tools are used to measure response to treatment.

The clinical experts indicated that treatment with semaglutide would be long term, unless the following occurred: excessive weight loss or nutritional deficiencies (may also be considered for dose reduction), renal failure (however, treatment may be continued, depending on overall nutritional status and if tolerating otherwise), side effects (e.g., persistent severe nausea, vomiting, or diarrhea), development of a contraindication, pregnant or planning to become pregnant, and cost and/or loss of coverage. The experts indicated that the occurrence of a repeat CV event while receiving treatment with semaglutide would not be considered treatment failure, noting there would be too much uncertainty to determine whether treatment was associated with the delayed onset of a second CV event in an individual. Additionally, the clinical experts advised that pausing treatment for an elective surgery may be considered, given reports that semaglutide is associated with slow gastrointestinal transit, which can increase the risk of aspiration with endotracheal intubation.

The clinical experts advised that a specialist is not required to prescribe semaglutide because family physicians and primary care providers have been prescribing semaglutide (Ozempic) for diabetes and obesity and are familiar with its side effect profile and monitoring.

Clinician Group Input

A total of 4 clinician groups provided input for this resubmission: TotalCardiology Rehabilitation (TCR) (5 clinicians) and Cardiac Rehabilitation and Secondary Prevention Program (CRSP) (Western University Division of Cardiology, 5 clinicians), plus a joint input from Obesity Canada and the Canadian Association for Bariatric Physicians and Surgeons (CABPS) (total of 18 clinicians). TCR, a multidisciplinary group of family physicians, internists, and cardiologists, is the sole provider of cardiac rehabilitation in the city of Calgary and the surrounding area. The Western University CRSP program delivers comprehensive, multidisciplinary rehabilitation care, including secondary prevention through lifestyle and pharmacotherapeutic interventions. The CABPS is a group of specialists in Canada with experience treating obesity. Obesity Canada is a national registered charity that assisted with the coordination of the group clinician response from CABPS.

The clinician group input was largely consistent with that of the clinical experts. Both the clinician groups and experts highlighted the limitations of the currently available weight management treatment options and the barriers to patients accessing them. The CABPS noted that obesity management practice currently follows the 2020 Canadian Adult Obesity Clinical Practice Guidelines, which recommend lifestyle modification supported by psychological or behavioural therapy, pharmacotherapy, and bariatric surgery. The clinician groups further highlighted that lifestyle modifications alone are often insufficient to experience the weight loss needed to improve obesity-related complications such as type 2 diabetes and MASLD. The clinician groups noted that liraglutide, semaglutide, and bupropion-naltrexone are the most effective long-term pharmacologic treatments due to their impact on the neuroendocrine pathways associated with obesity. The clinician groups also outlined the role of medical bariatric centres in providing multidisciplinary care and differentiated them from nonmedical weight loss clinics. The CABPS expressed that obesity treatment should extend beyond weight loss to address dysfunctional adipose tissue driving adverse health outcomes. The clinician groups’ overview of therapies that reduce cardiorenal risk in patients with comorbid CV disease was consistent with the expert input. The clinician groups also added that patients with ASCVD become eligible for cardiac rehabilitation, which combines lifestyle and pharmacotherapeutic interventions.

Both the clinicians and experts agreed that a primary goal of CV disease treatment is to reduce the incidence of CV events, but the clinician groups also suggested additional goals for weight management and improved glycemic control, when applicable. Both the clinicians and experts highlighted the access challenges and potential risks associated with bariatric surgery. Both noted a need for additional treatment options that address both CV risk and weight reduction, noting that existing treatments for weight loss and CV disease can lack long-term effectiveness and fail to directly target weight management. The clinician groups noted this as being particularly true for patients with obesity who do not have diabetes, a group for which there are currently no targeted treatment options. The clinician groups noted that these significant treatment gaps contribute to a persisting high risk of morbidity and mortality in these populations.

In contrast to the clinical expert input, which noted that semaglutide would be combined with other therapies that reduce CV risk (i.e., RAAS inhibitors, SGLT2 inhibitors), the clinician groups indicated that semaglutide would be used for first-line treatment alongside lifestyle modifications. The clinician groups noted it is unnecessary to first try other treatments before semaglutide, given its effectiveness in achieving substantial weight loss and improving obesity-related health outcomes. The clinician groups anticipate that semaglutide could shift current treatment paradigms, noting its efficacy approaches that of bariatric surgery and could reduce the number of patients requiring surgery. The patient type that the clinician groups indicated would benefit most from semaglutide aligned exactly with the type indicated by the clinical experts. The clinician groups added that semaglutide might also benefit patients with diabetes who require additional CV protection, or those without diabetes who are at high risk of CV disease and have difficulty experiencing weight loss through lifestyle modifications alone. The CRSP clinicians estimated that 78% of their patients receiving cardiac rehabilitation do not have diabetes and that 45% of patients who complete the rehabilitation program could potentially benefit from an additional therapy such as semaglutide. The patients least suitable for treatment, identified by the clinician groups and experts, also aligned. According to the clinician groups, patients with obesity are identified through self-identification or by health care providers using diagnostic tools such as BMI, waist circumference, and staging systems such as the Edmonton Obesity Staging System, which also classifies the degree of associated conditions (e.g., type 2 diabetes, impaired mobility). However, the CABPS noted there are currently no methods for identifying which patients are likely to respond to semaglutide.

The physician group input was consistent with the expert insights on the use of weight loss to assess treatment response but emphasized additional markers of treatment benefit. The clinician groups highlighted improvements in prediabetes, lipids, blood pressure, mobility, and quality of life as equally important. Conversely, TCR indicated there is no definitive evidence that the absence of weight loss precludes other treatment benefits. The CABPS provided specific criteria for a meaningful treatment response, including a 5% reduction in total body weight after 3 months of treatment (aligning with the expert input) and improved laboratory markers, reduced osteoarthritis pain, and the ability to proceed with procedures like hip replacement. The clinician groups also suggested initial evaluations every 4 to 6 weeks, then every 3 months, recognizing variability based on clinician and patient preference.

Both the clinician groups and experts agreed that semaglutide treatment should be continued indefinitely, with the clinician groups likening the regimen to statin therapy or treatment with acetylsalicylic acid (ASA) after an MI. The clinician groups suggested the discontinuation of semaglutide be considered in the following scenarios: presence of treatment intolerance or intolerable side effects that do not improve over time with appropriate countermeasures, pancreatitis, pregnancy or planning pregnancy, lack of meaningful treatment response, lack of affordability, or a more effective treatment becomes available in the future that requires semaglutide discontinuation. The clinician groups noted that semaglutide treatment could be delivered in family medicine and primary care clinics, community-based obesity management or metabolic medicine clinics, and hospital-based medical and surgical centres. Like the experts, the clinician groups indicated that any health care provider with experience managing obesity could prescribe semaglutide, noting this role should not be restricted to specialists alone.

Drug Program Input

Input was obtained from the drug programs that participate in the CDA-AMC reimbursement review process. The following items were identified as key factors that could potentially impact the implementation of a recommendation to reimburse semaglutide 2.4 mg: relevant comparators, considerations for the continuation or renewal of therapy, considerations for the prescribing of therapy, and systemic and economic issues. The clinical experts consulted for this review provided advice on the potential implementation issues raised by the drug programs and are presented in Table 3.

Clinical Evidence

Systematic Review

Description of Study

The SELECT trial (N = 17,604) was a phase IIIb, multinational, multicentre, randomized, double-blind, parallel-group, placebo-controlled trial that evaluated the effect of semaglutide 2.4 mg subcutaneous injection compared with placebo in reducing the risk of MACE occurrences (CV death, nonfatal MI, or nonfatal stroke) in patients with established CV disease and overweight or obesity and without diabetes. The key secondary objective of the trial was to evaluate the effect of semaglutide in mortality compared with placebo. Patients were randomized in a 1:1 ratio to receive once-weekly treatment with either semaglutide 2.4 mg or placebo, both as an adjunct to the standard of care for CV disease.

A total of 8,803 patients were randomized to receive semaglutide 2.4 mg and 8,801 patients were randomized to receive placebo. The mean age of patients in the semaglutide 2.4 mg group was 61.6 years (standard deviation [SD] = 8.9 years) and 61.6 years (SD = 8.8 years) in the placebo group. The mean BMI of patients in the semaglutide 2.4 mg group was 33.30 kg/m² (SD = 5.03 kg/m²) and 33.37 kg/m² (SD = 5.04 kg/m²) in the placebo group. Most patients in the full analysis set (FAS) had a history of MI: 5,962 patients (67.7%) in the semaglutide 2.4 mg group and 5,944 patients (67.5%) in the placebo group. The mean duration of follow-up was 39.9 months (SD = 9.3 months) in the semaglutide 2.4 mg group and 39.7 months (SD = 9.5 months) in the placebo group. The mean duration of exposure to semaglutide 2.4 mg was 33.3 months (SD = 14.4 months) and to placebo was 35.1 months (SD = 13.0 months).

Efficacy Results

Key efficacy results are based on the in-trial observation period with a data cut-off date of July 18, 2023, and are presented in Table 2.

Major Adverse CV Event (CV Death, Nonfatal MI, or Nonfatal Stroke)

At the data cut-off date, the percentages of patients (in the FAS) with their first event adjudication committee (EAC)–confirmed MACE (consisting of CV death, nonfatal MI, or nonfatal stroke) were 6.5% of 8,803 patients in the semaglutide 2.4 mg group and 8.0% of 8,801 patients in the placebo group. Semaglutide 2.4 mg was favoured over placebo (adjusted hazard ratio [HR] = 0.80; 95% confidence interval [CI], 0.72 to 0.90). A cumulative incidence plot of time from randomization to first EAC-confirmed MACE is presented in Figure 3 in Appendix 1. The absolute risk difference in time to first EAC-confirmed MACE between semaglutide 2.4 mg and placebo at week 156 was −1.1% (95% CI, −1.9% to −0.4%).

Consultation with the clinical experts did not identify any relevant potential treatment-effect modifiers to examine for this review. However, BMI was identified as the most relevant subgroup for the purpose of this review to inform expert committee deliberations and, as such, a summary of the subgroup analysis results by BMI for the primary end point is presented in Table 28 in Appendix 1.

CV Death

At the data cut-off date, the percentages of patients (in the FAS) with an EAC-confirmed CV death (including undetermined cause of death) were 2.5% in the semaglutide 2.4 mg group and 3.0% in the placebo group (HR = 0.85; 95% CI, 0.71 to 1.01). A cumulative incidence plot of time from randomization to EAC-confirmed CV death is presented in Figure 4 in Appendix 1. The absolute risk difference in time to EAC-confirmed CV death between semaglutide 2.4 mg and placebo at week 156 was 0% (95% CI, −0.5% to 0.4%).

HF Composite (CV Death or Hospitalization for HF or Urgent HF Visit)

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed composite HF outcome (comprising CV death, HF requiring hospitalization or urgent HF visit) were 3.4% in the semaglutide 2.4 mg group and 4.1% in the placebo group (HR = 0.82; 95% CI, 0.71 to 0.96). A cumulative incidence plot of time from randomization to first EAC-confirmed composite HF outcome is presented in Figure 5 in Appendix 1. The absolute risk difference in time to EAC-confirmed composite HF outcome between semaglutide 2.4 mg and placebo at week 156 was −0.2% (95% CI, −0.8% to 0.3%).

All-Cause Death

At the data cut-off date, the percentages of patients (in the FAS) with an EAC-confirmed all-cause death were 4.3% in the semaglutide 2.4 mg group and 5.2% in the placebo group (HR = 0.81; 95% CI, 0.71 to 0.93). A cumulative incidence plot of time from randomization to EAC-confirmed all-cause death is presented in Figure 6 in Appendix 1. The absolute risk difference in time to EAC-confirmed all-cause death between semaglutide 2.4 mg and placebo at week 156 was −0.5% (95% CI, −1.1% to 0.1%).

Nonfatal MI

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed nonfatal MI were 2.7% in the semaglutide 2.4 mg group and 3.7% in the placebo group (HR = 0.72; 95% CI, 0.61 to 0.85).

Nonfatal Stroke

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed nonfatal stroke were 1.7% in the semaglutide 2.4 mg group and 1.9% in the placebo group (HR = 0.93; 95% CI, 0.74 to 1.15).

Hemoglobin A1C of 6.5% or Greater

Classification of glycemic status was according to the American Diabetes Association Standards of Medical Care in Diabetes, published in 2018.²³ According to these standards, a patient with a hemoglobin A1C of 6.5% or greater has diabetes.²³

At the data cut-off date, 306 patients (3.5% of the FAS) in the semaglutide 2.4 mg group and 1,059 patients (12.0% of the FAS) in the placebo group experienced a first occurrence of a hemoglobin A1C of 6.5% or greater (HR = 0.27; 95% CI, 0.24 to 0.31).

Composite Nephropathy Event

The 5-component composite nephropathy end point consisted of onset of persistent macroalbuminuria (urinary albumin-to-creatinine ratio [UACR] > 300 mg/g), persistent 50% reduction in estimated glomerular filtration rate (eGFR) compared with baseline, onset of a persistent eGFR of less than 15 mL/min/1.73 m², initiation of chronic renal replacement therapy (i.e., dialysis or transplant), or renal death.

At the data cut-off date, 155 patients (1.8% of the FAS) in the semaglutide 2.4 mg group and 198 patients (2.2% of the FAS) in the placebo group had experienced a first composite nephropathy event (HR = 0.78; 95% CI, 0.63 to 0.96).

Body Weight

The mean treatment difference in the percent change from baseline in body weight at week 104 between semaglutide 2.4 mg and placebo was −8.51% (95% CI, −8.75% to −8.27%).

The analyses at weeks 156 and 208 were not performed as planned because more than 10% of patients in the FAS missed the relevant annual study visit due to trial termination. However, an analysis of the percent change from baseline in body weight was performed at week 208 and the estimated treatment difference (based on the in-trial period) between semaglutide 2.4 mg and placebo was −8.7% (95% CI, −9.42% to −7.88%). The on-treatment analysis, which defined treatment exposure as the observation period until first time off treatment for greater than 35 days, estimated a treatment difference of −10.2% (95% CI, −11.0% to −9.42%), which was similar to the result from the main analysis for change in body weight at week 208.²⁴

EQ-5D Index Score and EQ Visual Analogue Scale Score

The EQ-5D-5L questionnaire is a patient-reported outcome tool used to estimate health-related quality of life. The tool includes a descriptive system that provides a description of problems experienced by the patient who responded according to dimensions, a visual analogue scale (VAS) that provides a score indicating overall self-rated health (score ranges from 0 to 100), and an index score that ranges from 0 to 1. A higher score indicates better self-reported health status.

The treatment difference in the change from baseline in the EQ-5D index score at week 104 between semaglutide 2.4 mg and placebo was 0.01 (95% CI, 0.01 to 0.02).

The treatment difference in the change from baseline in the EQ VAS score at week 104 between semaglutide 2.4 mg and placebo was 1.60 (95% CI, 1.16 to 2.04).

Weight-Related Sign and Symptom Measure

The weight-related sign and symptom measure (WRSSM) is a patient-reported outcome tool (a self-rated VAS) used to assess the presence and bother associated with weight-related symptoms. Specifically, it is used to assess the impact of multifaceted aspects of obesity on symptom experience in individuals with overweight or obesity. The total score ranges from 0 to 4, with higher scores indicating worse symptomatology.

The mean change from baseline in WRSSM total score at week 104 was −0.26 (SD = 0.71) in the semaglutide 2.4 mg group and −0.12 (SD = 0.68) in the placebo group. The between-group difference was not estimated.

Cardiometabolic Risk Factors

Efficacy results on cardiometabolic risk factors (change from baseline in systolic blood pressure, change from baseline in total cholesterol, and change from baseline in high-density lipoprotein [HDL] cholesterol at week 104) were used to inform the accompanying pharmacoeconomic analysis and are presented in Table 29 in Appendix 1.

Harms Results

Harms results are based on the in-trial observation period with a data cut-off date of July 18, 2023, and are presented in Table 2.

Adverse Events

As described in the Clinical Study Report, nonserious adverse events (AEs) not fulfilling any of the prespecified criteria were not systematically collected. The focus of the safety evaluation per the Clinical Study Report for the SELECT trial was based on the reporting of serious adverse events (SAEs) and other systematically collected events (i.e., AEs of special interest).

Serious AEs

The proportion of patients with an SAE was 33.41% (2,941 of 8,803 patients) in the semaglutide 2.4 mg group and 36.40% (3,204 of 8,801 patients) in the placebo group. The most frequently reported SAEs (by preferred term) (frequency ≥ 2.0%) were coronary arterial stent insertion, acute MI, and unstable angina.

Withdrawals Due to AEs

The proportion of patients who permanently stopped treatment due to AEs was 16.60% (1,461 patients) in the semaglutide 2.4 mg group and 8.16% (718 patients) in the placebo group. The most frequently reported AEs (by preferred term) that led to permanent treatment discontinuation (frequency ≥ 2.0%) were nausea and diarrhea.

The proportion of patients who had their treatment interrupted or dose withdrawn due to AEs was 30.32% (2,669 patients) in the semaglutide 2.4 mg group and 16.00% (1,408 patients) in the placebo group. The most frequently reported AEs (by preferred term) that led to treatment interruption or dose withdrawal (frequency ≥ 2.0%) were nausea, diarrhea, vomiting, constipation, and decreased appetite.

Mortality

There were 375 all-cause deaths (4.26%) in the semaglutide 2.4 mg group and 458 all-cause deaths (5.20%) in the placebo group. There were 371 investigator-reported SAEs with fatal outcome (4.21%) in the semaglutide 2.4 mg group and 460 investigator-reported SAEs with fatal outcome (5.23%) in the placebo group.

Table 2: Summary of Key Results From the In-Trial Observation Period in the SELECT Trial (FAS)

AE = adverse event; ANCOVA = analysis of covariance; CDA-AMC = Canada’s Drug Agency; CI = confidence interval; CV = cardiovascular; EAC = event adjudication committee; eGFR = estimated glomerular filtration rate; FAS = full analysis set; HF = heart failure; HR = hazard ratio; MACE = major adverse cardiovascular event; MedDRA = Medical Dictionary for Regulatory Activities; MI = myocardial infarction; NSTEMI = non–ST elevation myocardial infarction; SAE = serious adverse event; SD = standard deviation; SE = standard error; STEMI = ST elevation myocardial infarction; VAS = visual analogue scale; vs. = versus; WRSSM = weight-related sign and symptom measure.

Note: Data cut-off date was July 18, 2023.

^aAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. The assumption of proportional hazards was evaluated by the standardized score process and was considered supported, per investigator. Patients without events of interest were censored at the end of their in-trial period. Based on the available number of events for analysis, the nominal significance level was updated to 0.02281 using the Lan-DeMets alpha spending function. Adjustment for group sequential design was done using likelihood ratio ordering.

^bAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. Patients without events of interest were censored at the end of their in-trial observation period.

^cBased on the available number of events for analysis, the nominal significance level was updated to 0.01148 using the alpha spending function described in the statistical analysis plan.

^dThe superiority of semaglutide 2.4 mg vs. placebo was confirmed for the primary end point of time to first EAC-confirmed MACE comprising CV death, nonfatal MI, and nonfatal stroke. Superiority of semaglutide 2.4 mg vs. placebo was not confirmed for the confirmatory secondary end point of time to EAC-confirmed CV death. As such, the superiority of semaglutide 2.4 mg vs. placebo was not tested for the confirmatory secondary end points that followed the order of the prespecified testing hierarchy (time to first EAC-confirmed composite HF outcome and time to EAC-confirmed all-cause death).

^eBased on the available number of events for analysis, the nominal significance level was updated to 0.01149 using the alpha spending function described in the statistical analysis plan.

^fBased on the available number of events for analysis, the nominal significance level was updated to 0.01213 using the alpha spending function described in the statistical analysis plan.

^gNot controlled for multiplicity.

^hAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. Patients without events of interest were censored at the end of their in-trial observation period. Patients randomized in error with a baseline hemoglobin A1C of ≥ 48 mmol/mol (6.5%) were excluded from this analysis.

ⁱAnalyzed using an ANCOVA with treatment as fixed factor and baseline value as covariate. Before analysis, missing data were multiple imputed. The imputation model (linear regression) was done separately for each treatment group and included baseline value as a covariate and was fitted to all patients with a measurement regardless of treatment adherence at week 104. The fitted model was used to impute values for patients without a measurement at week 104. Mean estimates were adjusted according to observed baseline distribution.

^jPresented descriptively only.

^kThe most frequently reported AEs by system organ class are summarized in the table, except for all-cause death and AEs of special interest.

^lCV AEs were collected systematically when serious. In addition, potential events of acute coronary syndrome, stroke, coronary artery revascularization, and HF for adjudication by the EAC were reported, irrespective of seriousness. A broad evaluation of CV safety, based on SAEs in the system organ class of cardiac disorders was performed. The evaluation of CV safety was supplemented by the assessment of blood pressure and heart rate.

^mAll AEs of gallbladder disease were collected systematically, irrespective of seriousness. Gallbladder-related disorders were evaluated based on a predefined MedDRA search on all AEs, supplemented with additional information collected from specific forms for recording events of gallbladder disease.

ⁿGastrointestinal AEs were collected systematically when serious. Gastrointestinal safety was evaluated based on SAEs in the system organ class of gastrointestinal disorders.

^oAll AEs of pancreatitis were collected systematically, irrespective of seriousness. The evaluation of pancreatitis was based on a predefined MedDRA search for pancreatitis among all AEs, which provided a broad evaluation of all types of investigator-reported pancreatitis, including search terms indicative of acute or chronic pancreatitis; additional information collected from a specific form for recording events of pancreatitis; and the outcome of the adjudication by the EAC, where the evaluation focused on events of acute pancreatitis.

^pA total of 69 patients (35 in the semaglutide 2.4 mg group and 34 in the placebo group) had a medical history of acute pancreatitis at screening. None of the patients with a history of acute pancreatitis had EAC-confirmed events of acute pancreatitis during the in-trial period.

Sources: Clinical Study Report version 1.0 for the SELECT trial²⁵ and the Novo Nordisk Canada Inc. responses to the January 15, 2025,²⁶ and January 23, 2025,²⁷ requests by CDA-AMC for additional information for the CDA-AMC review of Wegovy. Details included in the table are from the sponsor’s summary of clinical evidence.²

Notable Harms

The AEs of special interest identified for this review included CV AEs, cholelithiasis, nausea, vomiting, constipation, GERD, and pancreatitis. SAEs of cardiac disorders were reported in 11.45% (1,008 patients) in the semaglutide 2.4 mg group and 13.45% (1,184 patients) in the placebo group. All other AEs of special interest were reported in less than 2% of patients in each group.

Critical Appraisal

Internal Validity

The SELECT trial had a group sequential design, a type of adaptive clinical trial design that plans for interim analyses at predetermined points during follow-up and as data are collected in the trial.²⁸ Unlike traditional, fixed-period RCTs, this design offers structured opportunities to make earlier decisions when the intervention shows a significant benefit and improves efficiency in research by allowing adjustments to be made in the trial design. However, this design requires strong and appropriately conservative statistical plans due to the potential for unblinding, investigator influence, overestimation of treatment effect, and an increased chance of inflated type I error.^28-30 As will be described, the SELECT trial’s design, conduct, and analysis mitigate some of these potential limitations.

Overall, the baseline characteristics were similar between groups, suggesting randomization successfully distributed potential confounders. In consultation with the clinical experts, the baseline characteristics did not reflect any notable imbalances in the known prognostic factors for CV events and treatment-effect modifiers. As such, the review team concluded that the methods used to achieve allocation and allocation concealment for randomization were appropriate and that the risk of bias arising from the randomization process was low.

The percentage of patients who discontinued the study was relatively small and similar between the groups (approximately 3%). The reasons for discontinuing the study (patient withdrawal and lost to follow-up) were also similar between the groups and, therefore, the risk of attrition bias was likely low. The overall treatment discontinuation rates were similar between groups (approximately 30%). An imbalance in the discontinuation of treatment due to AEs was observed (16.3% in the semaglutide 2.4 mg group and 7.9% in the placebo group). This imbalance could have contributed to differential unblinding because those participants with treatment-emergent AEs — many of which were gastrointestinal in nature — might have inferred their assignment to semaglutide. Similarly, investigators might have formed expectations based on these observations, thereby introducing the potential for performance and reporting bias. While the observed difference between groups in treatment discontinuation due to lack of effect was relatively smaller, it was still differential (0.7% in the semaglutide 2.4 mg group and 2.8% in the placebo group) and, therefore, does not completely negate the potential impact on reporting and performance biases (i.e., relatively less concerning for biases). However, appropriate methods for adjudicating events in a robust and standardized way were employed and safeguards to prevent operational bias (in the context of a group sequential design) were reportedly in place; therefore, the review team concluded the probability for introducing bias into the results was low.

Supplementary analyses of absolute risk difference were conducted at week 156 of the trial to assess the robustness of various end points. A total of 27.3% of patients who were randomized in the trial did not have a visit at week 156 due to trial closure. The trial was event driven and would not be completed until the target number of primary end point events was reached. While the general assumption for censored observations was that the risk of experiencing an event was not changed by censoring, the potential for bias in the estimated cumulative incidence at week 156 would arise only if those patients who were censored due to what was termed “trial closure” in the Clinical Study Report had a different risk of MACE compared with those who remained in the study. However, the definition of trial closure in the Clinical Study Report is not entirely clear. Administrative censoring included trial completion, lost to follow-up, and patient withdrawal. The implications of administrative censoring for the robustness of absolute risk difference in MACE occurrences are difficult to determine based on the available information. The review team noted that the between-group distribution in administrative censoring, including “trial closure,” was similar and, therefore, if bias exists, it is not likely an important factor in the cumulative incidence estimates. Further, a total of 79.4% of patients who were randomized in the trial did not have a visit at week 208 due to trial closure. The review team considered this large percentage of missing data to seriously undermine the validity and reliability of the percent change from baseline in body weight at week 208 reported by Ryan et al.²⁴ There is a high chance that missing information was due to informative censoring and, in the absence of additional analyses evaluating the missing data patterns, it is not possible to draw firm conclusions on the results from the analyses on body weight at week 208.

The use of a single interim analysis in the SELECT trial reduces the risk of bias compared with multiple interim tests in a trial that uses a group sequential design. For the primary end point, the final analysis was adjusted for the group sequential design by using likelihood ratio ordering. The similarity between the unadjusted and adjusted estimates suggested that the single interim analysis had little influence on the final effect size and the reported treatment effect was unlikely influenced by type I error, which is associated with the group sequential design. Further, given the trial continued to its final analysis without stopping for efficacy, the concern for treatment-effect inflation due to early termination does not apply.

The proportional hazards assumption for the primary end point, which the investigators reported as being met, was evaluated graphically using Schoenfeld residuals and the standardized score process. However, an early deviation in the slope suggested a violation of the proportional hazards assumption in the first 12 months, although the stabilization afterward indicated the issue was not persistent over the full duration of the study. The statistical analysis plan for the SELECT trial did not describe methods for assessing the potential impact of this violation on the results. While a visual inspection of the score process suggested a time-dependent effect on the proportional hazards, there was no clear evidence that this influenced the confirmatory results in a meaningful way.

The superiority hypothesis was tested for each confirmatory secondary end point under multiplicity control using a stagewise hierarchical testing scheme according to the prespecified order. The statistical testing strategy for the confirmatory secondary end points used a separate alpha spending function to control the type I error rate at a 1-sided level of 2.5%, which aligned with the P value adjustment for the group sequential design. Importantly, the end points — the composite HF outcome and all-cause death — could not be interpreted formally for superiority because the prespecified hierarchical testing procedure failed to reach statistical significance with the analysis of CV deaths, meaning they should be interpreted as exploratory rather than confirmatory. The SELECT trial performed statistical comparisons for the nonconfirmatory secondary end points and some exploratory end points. However, the statistical comparisons were not included in the approach to adjust for multiple comparisons and, therefore, they increase the risk of type I error. The absence of a prespecified multiplicity control strategy for these end points limits the strength of the conclusions that can be drawn from the observed differences between groups.

The number of events contributing to the subgroup analyses of patients with a BMI of 40 kg/m² and greater was considered few (< 50 events in each of the 2 treatment groups), and the sample sizes for these 2 subgroups were considered small (< 10% of patients randomized in each group), thereby lowering the certainty in the consistency of the treatment effect for the primary end point in these subgroups.

The main analysis of the time-to-event end points assumed independent censoring of patients who had withdrawn from the trial or were lost to follow-up, while deaths from causes not included in the end point were handled as censored observations but not part of the independent censoring assumption. Sensitivity analyses (addressing the independent censoring assumption and assessing the potential impact of missing data for patients who had withdrawn from the trial or were lost to follow-up) and a supplementary analysis (assessing the influence of the competing risk of death from causes not included in the end point) were performed for the confirmatory end points. The results of the sensitivity analyses were consistent with the main analyses; the review team judged that the risk of bias due to the handling of these intercurrent events and missing outcomes data for the confirmatory end points was low. Of note, the tipping-point sensitivity analysis was conducted only for the primary analysis and not for the key secondary confirmatory analyses. The review team had concerns about the 2 additional sensitivity analyses that were described as applying to patients who withdrew consent or were lost to follow-up; however, these sensitivity analyses yielded results similar to those of the primary analysis (which was further supported by the tipping-point sensitivity analysis) and other confirmatory analyses, which suggests that the assumption of independent censoring was reasonable and that the trial’s conclusions are likely robust.

For all objectives, the primary estimand (intention to treat) was used to evaluate the treatment effect irrespective of adherence to treatment or changes to background medication. Although this approach to the handling of intercurrent events may be reflective of clinical practice, it can be a limitation for the interpretation of efficacy results, given that these intercurrent events also have an effect on CV risk. Further, the proportion of patients who discontinued treatment was relatively high (approximately 30%). A supplementary analysis based on the first on-treatment period was performed for the confirmatory end points to evaluate the effect in all patients who had remained on their randomized treatment. Because the results were consistent with the main analysis, the review team judged that the risk of bias due to treatment discontinuation was low for the confirmatory end points. The proportion of patients initiating a CV-related medication, an SGLT2 inhibitor, or metformin after randomization was, overall, slightly higher with placebo compared with semaglutide 2.4 mg. In consultation with the clinical experts, it was concluded that there is a potential for bias against semaglutide 2.4 mg in the assessment of CV outcomes due to this slight imbalance in the prescribing of these medications.

Continuous supportive secondary end points at week 104 were analyzed, with multiple imputation used for missing values under a missing-at-random assumption, which was likely not plausible because there were notable imbalances in the reasons for treatment discontinuation between groups. A sensitivity analysis that was planned to address this assumption for missing data were performed for the change from baseline in body weight at week 104. Because the results were consistent with the main analysis, the review team judged that the risk of bias due to missing outcomes data for this end point was low. No sensitivity analysis was planned for change from baseline in the EQ-5D, where missing outcome data were relatively high (approximately 16%) at week 104. As such, the review team concluded there is a potential for risk of bias due to missing outcomes data for these HRQoL end points.

WRSSM was an exploratory end point because the measure was not sensitive to weight loss in a psychometric evaluation, per the Clinical Study Report of the SELECT trial. The sponsor indicated that the WRSSM tool was under development at the time of the trial and, as such, no further evidence of its measurement properties in the population of interest was submitted by the sponsor.

External Validity

Evidence from the SELECT trial addressed the evidence gap on the effects of semaglutide 2.4 mg on CV outcomes, which was raised in the previous submission for the weight management indication, but only for a subset of patients in the indicated population.

There was overlap between the SELECT trial population and the population criteria specified by the sponsor in its reimbursement request: patients with a BMI of 27 kg/m² or greater and established CV disease (MI, stroke, and/or PAD). The clinical experts consulted for this review agreed that the inclusion criteria captured the target population with established CV disease (including HF) that is seen in practice and is in need of an intervention for the secondary prevention of future CV events. Notably, most patients in the trial had prediabetes, according to the hemoglobin A1C range defined in the American Diabetes Association Standards of Medical Care in Diabetes published in 2018 (i.e., 5.7% to < 6.5%).²³ However, it is important to note that the Diabetes Canada Clinical Practice Guidelines³¹ specify a slightly different hemoglobin A1C range (6.0% to < 6.5%) for diagnosing prediabetes. In consultation with the clinical experts, no major concern with the generalizability of the results due to this difference was identified.

Because treatment with semaglutide 2.4 mg in practice is not expected to be limited to patients who do not have diabetes, the review team found the trial’s exclusion of patients with a hemoglobin A1C level of 6.5% or greater to be a concern for the generalizability of the results to the target population. The investigator indicated that the trial population did not include patients with type 1 or type 2 diabetes, to remove any confounding that a diagnosis of diabetes could have on future CV risk. For more information on this evidence gap and how evidence from the SUSTAIN-6 trial addressed that gap, refer to the study addressing the gap in the systematic review evidence section in this document.

In consultation with the clinical experts, it was concluded that the comparison with placebo added to the standard of care for CV disease (i.e., antihypertensives, lipid-lowering drugs, anticoagulants, and ASA and other antiplatelet drugs) was appropriate, given that none of the current and accessible standard-of-care therapies to treat or prevent CV disease directly target weight loss. Regarding the standard of care for CV disease used in the trial, the clinical experts advised that, in practice, semaglutide would be combined with therapies that reduce cardiorenal risk, including combination treatment with ACE inhibitors, angiotensin 2 receptor blockers, statins, and SGLT2 inhibitors.

Long-Term Extension Study

The sponsor did not submit a long-term extension study for this review.

Indirect Evidence

The sponsor did not submit indirect evidence for this review.

Study Addressing a Gap in the Evidence From the Systematic Review

Study Description

Data from the SUSTAIN-6 trial (N = 3,297) was submitted to address the effects of semaglutide on CV outcomes in the target reimbursement population (patients with a BMI of 27 kg/m² or greater and established CV disease) who also have type 2 diabetes. SUSTAIN-6 (N = 3,297) was a long-term, multicentre, multinational, randomized, double-blind, parallel-group, controlled trial that evaluated the CV safety and long-term outcomes of semaglutide compared with placebo, when added to the standard of care, in patients with type 2 diabetes at high risk of CV events. Patients were randomized 1:1:1:1 to receive a once-weekly subcutaneous injection of either semaglutide 0.5 mg, semaglutide 1.0 mg, or volume-matched placebo. Treatment duration spanned 104 weeks and consisted of a 4-week to 8-week dose-escalation period followed by a 96-week to 100-week treatment maintenance period, and then a 5-week follow-up period.

From the SUSTAIN-6 trial data, a post hoc subgroup analysis was conducted that included patients with a BMI of 27 kg/m² or greater and established CV disease and type 2 diabetes. Post hoc subgroup analyses were conducted to determine whether the efficacy observed in this patient population was consistent with the SELECT trial and, therefore, the enrolment criteria for this subgroup replicated that of the SELECT study (i.e., limiting the subpopulation to patients with prior MI, stroke [excluding transient ischemic attack; TIA], and/or PAD and a BMI of ≥ 27 kg/m²). Consistent with the exclusion criteria for all SUSTAIN-6 trial participants, patients in the post hoc subgroup analyses did not have type 1 diabetes and had not used a GLP-1 RA or pramlintide within 90 days before screening. Although there were 2 maintenance doses used in the overall SUSTAIN-6 study (0.5 mg and 1.0 mg), only patients who had received 1.0 mg were included in the post hoc subgroup analyses because this dose was considered most comparable to that of the SELECT trial. However, the SUSTAIN-6 trial’s post hoc subgroup had still received a lower dose of semaglutide (1.0 mg) compared with participants in the SELECT trial (2.4 mg). Because the post hoc subgroup analyses were the most relevant for the purposes of this review, the results summarized in this section include only those pertaining to this subgroup.

The CV-related efficacy outcomes investigated in the SUSTAIN-6 study were consistent with those of the SELECT trial. The primary end point was time from randomization to the first occurrence of MACE, defined as CV death, nonfatal MI, or nonfatal stroke. The secondary end points identified by the clinical experts as most relevant to this review were time from randomization to first occurrence of all-cause death, nonfatal MI, or nonfatal stroke.

Demographics and Baseline Characteristics

Patients in the SUSTAIN-6 trial’s post hoc subgroup analyses had a mean age of ███████ (SD = ███), a mean type 2 diabetes disease duration of ██ years (SD = ███), and a mean BMI of 34 kg/m² (SD = ███). Most patients across both groups had triglyceride levels of 150 mg/dL or greater (███), with a mean triglyceride level of █████ mg/dL (SD = █████). The eGFR of patients across both groups were primarily between 60 mL/min/1.73 m² to 90 mL/min/1.73 m² (███), with a mean eGFR of 77.8 mL/min/1.73 m² (SD = ████). Most patients were taking antihypertensive medication (█████) and/or lipid-lowering drugs (█████). For CV history, only data for prior HF were provided, and ██ (█████) of patients in the semaglutide group had prior HF compared with ███ (███) of the placebo group. Data for other CV inclusion criteria (MI, stroke, and PAD) were not provided.

Results

The outcomes reported for the post hoc subgroup efficacy analyses align with the key outcomes identified for this review. Of all efficacy outcomes included in the post hoc subgroup analysis, only the key secondary outcome of time from randomization to first occurrence of an expanded composite CV outcome, defined as either MACE, revascularization (coronary or peripheral), unstable angina requiring hospitalization, or hospitalization for HF, demonstrated nominal statistical significance; however, this outcome was not among those of interest for the purposes of this review. The remaining CV-related outcomes demonstrated numerical improvement in the semaglutide group compared with the placebo group; however, they did not achieve statistical significance. The overall safety profile of semaglutide observed in the post hoc subgroup of the SUSTAIN-6 trial was consistent with that of the SELECT trial, with fewer CV SAEs with semaglutide 1.0 mg versus placebo.

Critical Appraisal

The post hoc nature of the subgroup analysis in the SUSTAIN-6 trial introduces important internal validity concerns and risk of bias. Because the subgroup was not prespecified, the analysis is exploratory rather than confirmatory, increasing the risk of spurious findings due to multiple comparisons. Additionally, patients in the SUSTAIN-6 study were not randomized within the subgroup of interest, meaning that unmeasured confounders could bias the treatment-effect estimate. Among all CV-related end points, only expanded MACE reached nominal significance, which may have been due to inadequate power. As a result, it is challenging to interpret the efficacy of semaglutide 1.0 mg in this patient subgroup.

The small sample size of the post hoc subgroup limits the generalizability of the findings, given that 80% to 90% of individuals in Canada with type 2 diabetes with overweight or obesity and 32% of people worldwide have CV disease. The high prevalence of potentially eligible individuals in Canada, combined with significant heterogeneity in CV disease presentation, driven by factors like age, diabetes duration, glycemic control, and comorbidities, limits the generalizability of the findings. A major limitation of the SUSTAIN-6 trial in the context of the current resubmission is that the dosages evaluated in the trial, specifically, semaglutide 0.5 mg and 1.0 mg once weekly, do not align with the proposed Health Canada–recommended therapeutic and maintenance dosage of semaglutide (Wegovy), which is 2.4 mg once weekly. The focus of the post hoc subgroup of the SUSTAIN-6 study was on the 1.0 mg dose because, according to the sponsor, it was most comparable to the dose in the SELECT trial; however, this remains a significant limitation to the generalizability of the evidence.

Conclusion

The results from the SELECT trial were submitted as new evidence for this resubmission to address the evidence gap on the effects of semaglutide 2.4 mg once weekly on weight-related comorbidities, such as CV outcomes, in the indicated weight management population. The SELECT trial demonstrated that semaglutide 2.4 mg once weekly, as an adjunct to the standard of care for CV disease, reduced the occurrence of MACE compared with placebo over a mean follow-up period of nearly 40 months in patients with established CV disease and overweight (BMI ≥ 27 kg/m²) or obesity (BMI ≥ 30 kg/m²) but without diabetes. The point estimates of the observed treatment effects for the individual MACE components — CV death, nonfatal MI, or nonfatal stroke — were directionally aligned with the overall reduction in MACE occurrences, although the certainty of the estimated effects varied across components. More specifically, the CIs for CV death and, notably, nonfatal stroke, included the possibility of no benefit with respect to these efficacy outcomes. Additionally, the trial suggested a reduction in the composite HF events and all-cause mortality with semaglutide 2.4 mg compared with placebo; however, these end points could not be interpreted formally for superiority because the prespecified hierarchical testing procedure failed to reach statistical significance.

No new harms with semaglutide 2.4 mg were identified within the SELECT trial. The harms that were observed were consistent with what has been observed clinically, as per the input from the clinical experts consulted for this review regarding experience with semaglutide in clinical practice, as well as what has been observed in previous studies. The AEs leading to the permanent or temporary discontinuation of treatment were notable and reported more frequently with semaglutide 2.4 mg compared with placebo. Throughout the trial and, in particular, during the dose-escalation phase, the most frequently reported AEs were gastrointestinal-related, suggesting that tolerability at the recommended maintenance dose may be an issue. The AEs of special interest for this review (cholelithiasis, nausea, vomiting, constipation, GERD, and pancreatitis) were reported infrequently.

Introduction

The objective of this report is to review and critically appraise the clinical evidence submitted by the sponsor on the beneficial and harmful effects of semaglutide 2.4 mg solution for subcutaneous injection in chronic weight management in patients with a BMI of 27 kg/m² or greater and established CV disease (MI, stroke, or PAD).

Submission History

Semaglutide 2.4 mg was first reviewed by CDEC for weight management in 2022. CDEC issued a recommendation that semaglutide 2.4 mg not be listed as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with an initial BMI of 30 kg/m² or greater (obesity) or 27 kg/m² or greater (overweight) in the presence of at least 1 weight-related comorbidity such as hypertension, type 2 diabetes mellitus, dyslipidemia, or obstructive sleep apnea.³

The 2022 reimbursement recommendation for semaglutide (Wegovy) and the Clinical Review report on semaglutide (Wegovy) that was used to inform that recommendation are both available on the CDA-AMC project website.

Basis of Resubmission

Rationale for the 2022 recommendation included evidence from 4 placebo-controlled, double-blind RCTs (STEP 1, STEP 2, STEP 3, and STEP 4) that demonstrated that treatment with semaglutide 2.4 mg injection reduced body weight in individuals with an initial BMI of 30 kg/m² or greater (obesity) or 27 kg/m² or greater (overweight) in the presence of at least 1 weight-related comorbidity but did not demonstrate improvement in or prevention of weight-related comorbidities. Comorbidities such as MACE, osteoarthritis, and obstructive sleep apnea were not outcomes assessed in the STEP trials.³

CDEC noted there was an ongoing trial, the SELECT study,⁴ comparing semaglutide 2.4 mg injection with placebo for the prevention of MACE occurrences in patients with overweight or obesity who have established CV disease but not diabetes mellitus. CDEC concluded in 2022 that the results of that study would address the evidence gap regarding the effects of semaglutide 2.4 mg once weekly on CV outcomes in the indicated population.³

Hence the focus of this resubmission for the weight management indication was on the sponsor’s revised reimbursement request: as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with a BMI of 27 kg/m² or greater and established CV disease (MI, stroke, or PAD).

Sponsor’s Clarifying Note on Resubmission

In 2024, Health Canada issued an indication for semaglutide injection (Wegovy) to reduce the risk of nonfatal MI in adults with established CV disease and a BMI of 27 kg/m² or greater. This indication was based on the evidence from the SELECT trial.

The sponsor indicated that the revised reimbursement request population, namely, adults with established CV disease and a BMI of 27 kg/m² or greater, is fully encompassed in the weight management indication and is the same population for which the CV indication was granted.

Disease Background

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following has been summarized and validated by the review team.

Obesity is a complex chronic disease where adiposity, or excess body fat, impairs health and increases the risk of long-term health complications and comorbidities.⁵ The presence of overweight (a BMI of greater than 25 kg/m²) and obesity (a BMI of greater than 30 kg/m²) are considered major contributors to CV disease through both direct and indirect mechanisms. Obesity can directly result in changes to cardiac loading, volume status, energy substrate use, tissue metabolism, and systemic inflammation, which are believed to promote CV disease progression.⁶ Additionally, obesity may contribute to CV disease indirectly through the development of exacerbating conditions that carry high CV mortality risk (e.g., sleep apnea, thromboembolic disease, or MI) and cardiometabolic diseases (e.g., hypertension, dyslipidemia, type 2 diabetes, nonalcoholic fatty liver disease, or steatohepatitis).^7-15 Obesity is also associated with an increased risk of a wide range of illnesses and long-term conditions, including type 2 diabetes, hypertension, gallstones, GERD, and cancer, as well as psychological and psychiatric morbidities.³² Individuals with obesity have been shown to have a higher risk of anxiety, psychosis spectrum disorders, eating disorders, and mood disorders, such as major depressive disorder, bipolar disorder, and dysthymia.^33-36 Psychiatric morbidities can be compounded by social stigma and discrimination toward individuals with obesity, which can result in social isolation and avoidance of seeking medical care.³⁷ Quality of life is further impacted by difficulty performing everyday activities, mobility challenges, physical discomfort, and pain.³⁸

BMI is the primary measure used to identify patients who may benefit from obesity management and is often employed alongside other indicators, such as waist circumference and presence of cardiometabolic complications.³⁹ Additional diagnostic testing, including blood pressure, blood lipid, liver and kidney panels, electrocardiogram and/or echocardiogram, angiography, Doppler ultrasound, or Holter and event monitoring, are broadly available and can be arranged by primary care providers to help identify metabolic problems and tailor therapy.³⁹ Based on epidemiological studies, overweight is currently classified as a BMI of 25 kg/m² to less than 30 kg/m² and obesity is classified as a BMI of 30 kg/m² or greater.⁵ Obesity can be further subclassified into class I (BMI of 30 kg/m² to 34.9 kg/m²), class II (BMI of 35 kg/m² to 39.9 kg/m²), class III (BMI of 40 kg/m² to 49.9 kg/m²), class IV (BMI of 50 kg/m² to 59.9 kg/m²), and class V (BMI of greater than 60 kg/m²).⁴⁰

In 2022, the Canadian Community Health Survey estimated that, in Canada, 30% of individuals aged 18 years or older were living with obesity.^16,17 Across Canada, the prevalence was higher than the national average in New Brunswick (43%), Newfoundland and Labrador (42%), Saskatchewan (38%), Prince Edward Island (36%), Nova Scotia (36%), and Manitoba (34%); lower than the national average in British Columbia (26%) and Quebec (29%); and similar in Ontario (30%) and Alberta (31%).^16,17 There are limited data on patients in Canada with overweight or obesity and established CV disease. Across CV diseases, the proportion of people with CV disease who are living with obesity tends to be higher than among those with normal weight, and CV disease rates appear to proportionally increase with BMI.⁴¹ For example, hypertension prevalence among those with a normal BMI is 45% compared with 67% for those with overweight, 79% for those with obesity class I and II, and up to 87% among those with obesity class III.⁴¹ In a study of patients with ASCVD, obesity was among the 30 most common comorbid conditions, with a prevalence of 38%.¹⁸ CV disease is among the leading causes of morbidity and mortality worldwide and is reported to be 1 of the leading causes of hospitalization in Canada, alongside stroke.^19,20

Standards of Therapy

Contents within this section have been informed by materials submitted by the sponsor and clinical expert input. The following has been summarized and validated by the review team.

Based on clinical expert input, current chronic weight management in patients with a BMI of 27 kg/m² or greater and established CV disease is informed in part by the Canadian Adult Obesity Clinical Practice Guidelines: Pharmacotherapy for Obesity Management, updated in 2022;²¹ the Canadian Cardiovascular Society Guideline for Use of GLP-1 Receptor Agonists and SGLT2 Inhibitors for Cardiorenal Risk Reduction in Adults, published in 2022;⁴² and the European Society of Cardiology Guidelines for the Management of Chronic Coronary Syndromes, published in 2024.⁴³

The Canadian obesity guideline states that there are 4 medications indicated for long-term obesity management as adjuncts to health-behaviour changes: liraglutide, naltrexone-bupropion in a combination tablet, orlistat, and semaglutide. The Canadian guideline recognizes all 4 medications as effective in producing clinically significant weight loss and health benefits relative to placebo over a period of at least 1 year.²¹ The clinical experts indicated that another medication, tirzepatide, is currently approved only for adjunct glycemic control in type 2 diabetes but is increasingly being prescribed off label for chronic weight management in Canada. Further, patient groups have indicated that semaglutide (Ozempic) has been prescribed off label for weight management as well. The clinical experts indicated that in clinical practice in Canada, semaglutide is being used in patients with diabetes for weight loss.

Based on clinical expert input, the most important goals of therapy for chronic weight management in patients with a BMI of 27 kg/m² or greater and established CV disease is treatment and prevention of ASCV and HF outcomes (fatal and nonfatal events). Other important goals of therapy include weight loss and improvement in quality of life and symptom burden. Additional goals of therapy include improvement in health parameters such as CV risk factors (e.g., blood pressure, lipid profile, and glycemic status) and mental health.

Drug Under Review

Semaglutide is a selective long-acting GLP-1 RA that is 94% similar to human GLP-1 that binds to and activates GLP-1 receptors in the same manner as the endogenous hormone incretin. GLP-1 is a physiological regulator of appetite and caloric intake and is present in several brain areas involved in appetite regulation. Compared with native GLP-1, semaglutide has a prolonged half-life of approximately 1 week due to albumin binding, which results in decreased renal clearance and protection from metabolic degradation.

Semaglutide is administered subcutaneously in the abdomen, thigh, or upper arm. The therapeutic and maintenance dose of 2.4 mg of semaglutide once weekly is reached by starting with a dose of 0.25 mg and then following a dose-escalation regimen, with dose increases every 4 weeks (to doses of 0.5 mg, 1 mg, 1.7 mg, and 2.4 mg per week) until the therapeutic-maintenance dose of 2.4 mg once weekly is reached after 16 weeks. This dose-escalation regimen is suggested to decrease the likelihood of gastrointestinal symptoms. If patients do not tolerate a dose during dose escalation, prescribers should consider delaying dose escalation for 4 weeks.

Semaglutide (2.4 mg) was approved by Health Canada in November 2021 as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with an initial BMI of 30 kg/m² or greater (obesity) or 27 kg/m² or greater (overweight) in the presence of at least 1 weight-related comorbidity such as hypertension, type 2 diabetes mellitus, dyslipidemia, or obstructive sleep apnea. Semaglutide (2.4 mg), which was previously reviewed by CDA-AMC in 2022, received a “do not reimburse” recommendation for the sponsor’s request that semaglutide be reimbursed as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with an initial BMI of 35 kg/m² or greater and prediabetes.

For the current resubmission review, the sponsor has requested that semaglutide (2.4 mg) be reimbursed as an adjunct to a reduced calorie diet and increased physical activity for chronic weight management in adult patients with an initial BMI of 27 kg/m² or greater and established CV disease (MI, stroke, or PAD).

Perspectives of Patients, Clinicians, and Drug Programs

The full patient and clinician group submissions received are available on the CDA-AMC project website for this review in the consolidated patient and clinician group input document.

Patient Group Input

This section was prepared by the review team based on the input provided by patient groups.

Input for this review was submitted by 6 patient groups: GI Society, Obesity Canada, Obesity Matters, Fatty Liver Alliance, HeartLife Foundation, and Diabetes Canada. GI Society input was based on 2 international surveys for individuals with obesity from 2021 (1,550 individuals who responded, 62% from Canada) and 2024 (1,050 individuals who responded, 96% from Canada), as well as insights from an in-person focus group of individuals with obesity held in 2023. Obesity Canada input was based on 1 Canadian survey (170 individuals who responded) conducted between September and October 2024 of individuals with obesity. Survey responses were primarily from Ontario (42%), British Columbia (17%), and Alberta (13%), with additional representation from Saskatchewan, New Brunswick, Nova Scotia, and Prince Edward Island; most individuals who responded were female (92%) and between the ages of 45 and 64 years (58%). Input from the HeartLife Foundation was gathered through interviews and discussions with individuals with CV disease and their health care providers. The Fatty Liver Alliance gathered physician insights on MASLD and MASH, supplemented by a 2022 to 2023 US survey of 1,518 patients with nonalcoholic fatty liver disease or nonalcoholic steatohepatitis. Input from Diabetes Canada was based on a Canadian survey (October to November 2024) with 170 individuals who responded, including individuals with diabetes (59% with type 2 and 22% with type 1 diabetes, 12% with prediabetes) and their caregivers. Obesity Matters collected input in October 2024 through a Canadian obesity survey (186 individuals who responded) and a patient testimonial. Obesity Matters also provided feedback on the CDEC reimbursement recommendation issued for the initial 2022 review of semaglutide for weight management.

The patient groups highlighted the significant physical and mental health impact of obesity and related conditions, including CV diseases, diabetes, MASLD, and MASH. Common symptoms included mobility limitations, chronic pain, fatigue, swelling, shortness of breath, dizziness, and nausea, affecting daily life. Comorbidities were prevalent: in the GI Society’s 2021 survey, 91% of individuals who responded reported comorbidities such as arthritis (51%), hypertension (33%), sleep apnea (30%), GERD (29%), irritable bowel syndrome (29%), high cholesterol (25%), and diabetes (24%). Input from Diabetes Canada also noted high comorbidity rates for patients with diabetes, including high blood pressure, foot or eye problems, nerve damage, heart disease, and kidney-related comorbidities. Obesity not only contributes to these conditions but also complicates disease management and reduces quality of life. Mental health challenges, including anxiety, mood disorders, and social isolation due to stigma, especially in health care, were widely reported.

The patient groups highlighted weight loss as a key treatment outcome, along with a reduced risk of comorbidities (e.g., CV disease), improvements in quality of life, and increased ability to perform everyday tasks. When choosing a new therapy, most patients primarily value efficacy (i.e., significant and sustained weight loss) and affordability.

The patient groups emphasized that traditional lifestyle modifications, such as diet and exercise, are often ineffective for long-term weight management. Few medication options exist, with semaglutide, liraglutide, and naltrexone-bupropion being commonly prescribed, but none have public or full private coverage, making cost the biggest barrier. Those who responded to the survey and who reported the benefits of these treatments in managing obesity described them as “life-changing,” with most (94%) considering the side effects to be manageable. However, it was also noted that these medications lack long-term effectiveness, with many individuals gaining back the weight that was lost within 5 years. Bariatric surgery, while currently considered the gold standard for obesity treatment, carries risks of severe side effects, postsurgery weight gain, and long wait times, making it a last-resort option.

Patients with experience taking semaglutide described several benefits, including substantial weight loss. It was highlighted that semaglutide also improved the management of comorbidities, such as CV disease, and improved quality of life. Thus, it was identified that semaglutide has the ability to address multiple health conditions simultaneously. Patients with experience with semaglutide also highlighted that its side effects — most commonly nausea, vomiting, and constipation — were typically temporary and manageable, and the once-weekly administration reduced treatment burden.

Obesity Matters provided additional feedback on the 2022 draft of the CDEC reimbursement recommendation for semaglutide (Wegovy). The feedback indicated that while the initial draft recommendation acknowledged the effect of semaglutide on body-weight reduction and the increased risk of comorbidities observed with obesity, it did not conclude that weight reduction can directly reduce comorbidities. The feedback also critiqued the inclusion of the lack of structured weight programs available in Canada as part of the rationale for a negative reimbursement recommendation, noting this overlooks the agency of individuals actively working with specialists to make lifestyle changes. Obesity Matters indicated that the previous recommendation did not adequately consider cost as a barrier to treatment, patient perspectives, or the direct impact of treatment costs on quality of life. The feedback indicated that certain access barriers and implementation issues did not provide necessary guidance, required further elaboration on relevance, and were not specific to semaglutide, noting these challenges are routinely addressed for other treatments. Obesity Matters felt that the recommendation did not adequately consider how public reimbursement would reduce long-term costs associated with obesity-related comorbidities and conditions, nor did it offer actionable solutions for integrating semaglutide into existing health care systems.

Clinician Input

Input From the Clinical Experts Consulted for This Review

All CDA-AMC review teams include at least 1 clinical specialist with expertise in the diagnosis and management of the condition for which the drug is indicated. Clinical experts are a critical part of the review team and are involved in all phases of the review process (e.g., assisting in the critical appraisal of clinical evidence, interpreting the clinical relevance of the results, and providing guidance on the potential place in therapy). The following input was provided by 2 clinical specialists with expertise in the diagnosis and management of overweight and obesity and CV disease.

Unmet Needs

The clinical experts identified the following as limitations with weight management medications: barriers to public access, that not all patients experience weight loss with treatment, and that some patients experience adverse effects that necessitate stopping the drug. Specific to bariatric surgery, the clinical experts noted that in addition to barriers to access, not all patients are interested in pursuing this option because of the potential risks associated with surgery.

Place in Therapy

According to the clinical experts consulted for this review, semaglutide 2.4 mg has been used for weight management in Canada since it was approved by Health Canada in 2021. The clinical experts indicated that the anticipated place of semaglutide with regard to pharmacotherapy for chronic weight management is as a first-line treatment. Of note, the clinical experts anticipated that clinicians will increasingly prescribe medications with public reimbursement that specifically target obesity and have evidence for reduced risk of future CV events.

The clinical experts advised that semaglutide would be combined with therapies that reduce cardiorenal risk, including combination treatment with RAAS inhibitors (such as ACE inhibitors and angiotensin 2 receptor blockers) and SGLT2 inhibitor therapies as well as other standards of care to improve CV outcomes. The clinical experts indicated there is currently no high-level evidence to support combination treatment with other pharmacotherapies for weight management; however, they noted there is relevant evidence from small studies (e.g., a retrospective cohort study on the effect of combining a GLP-1 RA with bupropion-naltrexone in weight loss²²). The clinical experts indicated that, in practice, specialists in obesity treat patients with combination therapy when needed and appropriate and when cost and insurance coverage are not limiting factors. The clinical experts indicated that the most common combination is a GLP-1 RA combined with bupropion-naltrexone.

The clinical experts advised that semaglutide would not be combined with another GLP-1 RA (such as tirzepatide, which is increasingly being prescribed off label for weight management).

Patient Population

To identify patients who would most likely respond to treatment with semaglutide, the clinical experts deferred to the baseline characteristics of the SELECT trial population in which response was defined as a reduced risk of future CV outcomes and weight loss. Of note, the clinical experts highlighted that the trial population included patients with a pre-existing history of CV disease, and 64.5% of trial participants had prediabetes (hemoglobin A1C > 5.7% per US criteria). The clinical experts advised caution when prescribing semaglutide for patients with nutritional deficiencies or very low appetite at baseline.

The clinical experts noted that the benefits of semaglutide for primary prevention of CV disease are less established and less clear in patients without established CV disease or type 2 diabetes.

Overall, the clinical experts advised that it would be appropriate to treat most patients with a BMI of 27 kg/m² or greater for the chronic weight management indication, and that patients with established CV disease (including HF), type 2 diabetes, multiple risk factors, severe disease, or a high BMI are most in need of an intervention for the secondary prevention of future events and to improve symptom burden.

The clinical experts indicated that patients with a history of CV disease best suited for treatment with semaglutide can be identified through their medical records, with supporting documentation on their admission and/or procedure, or results from a diagnostic test such as echocardiogram. BMI is typically measured with a scale and self-reported height per clinical expert input. Of note, the clinical experts indicated that CV disease in the presence of overweight and obesity are underdiagnosed in practice if the clinician does not gather a standard history, perform a physical, and document the BMI in the chart.

When considering which patients would be least suitable for treatment with semaglutide, the clinical experts deferred to the product monograph for absolute contraindications and patients with a potential increased risk of adverse effects (e.g., history of pancreatitis, gallbladder disease, diabetic retinopathy, severe renal impairment, or gastrointestinal conditions).

Assessing Response to Treatment

Although weight loss is an outcome that can be easily monitored in practice, monitoring the recurrence of CV events and symptoms is equally as important.

When considering a clinically meaningful response to treatment in a patient-reported outcome, the clinical experts advised on the following:

• A weight loss of 5% or more is often considered clinically meaningful.

• A change of 5 points or more in the Kansas City Cardiomyopathy Questionnaire is considered clinically meaningful; some studies have estimated a change of 20 points or more to be clinically meaningful for HF.

• An increase of at least 30 m in the 6-minute walk distance is commonly considered clinically meaningful for HF.

Discontinuing Treatment

The clinical experts indicated that treatment with semaglutide would be long term, unless the following occurred (factors for consideration in discontinuing treatment with semaglutide):

• excessive weight loss or nutritional deficiencies (may also be considered for dose reduction)

• renal failure (however, treatment may be continued, depending on overall nutritional status and if tolerating otherwise)

• side effects, such as persistent severe nausea, vomiting, or diarrhea

• development of a contraindication

• pregnant or planning to become pregnant

• cost and/or loss of coverage.

The experts indicated that the occurrence of a repeat CV event while receiving treatment with semaglutide would not be considered treatment failure, noting there would be too much uncertainty to determine whether treatment was associated with the delayed onset of a second CV event in an individual. Additionally, the clinical experts advised that pausing treatment for an elective surgery may be considered, given reports that semaglutide is associated with slow gastrointestinal transit, which can increase the risk of aspiration with endotracheal intubation.

Prescribing Considerations

The clinical experts advised that a specialist is not required to prescribe semaglutide because family physicians and primary care providers have been prescribing semaglutide (Ozempic) for diabetes and obesity and are familiar with its side effect profile and monitoring.

Clinician Group Input

This section was prepared by the review team based on the input provided by clinician groups.

A total of 4 clinician groups provided input for this resubmission: TCR (5 clinicians) and the CRSP at the Western University Division of Cardiology (5 clinicians), plus a joint input from Obesity Canada and the CABPS (total of 18 clinicians). TCR, a multidisciplinary group of family physicians, internists, and cardiologists, is the sole provider of cardiac rehabilitation in the city of Calgary and surrounding area. The Western University CRSP program delivers comprehensive, multidisciplinary rehabilitation care, including secondary prevention through lifestyle and pharmacotherapeutic interventions. The CABPS is a group of specialists in Canada with experience treating obesity. Obesity Canada is a national registered charity that assisted with the coordination of the group clinician response from CABPS.

The clinician group input was largely consistent with that of the clinical experts. Both the clinician groups and experts highlighted the limitations of the currently available weight management treatment options and the barriers to patients accessing them. The CABPS noted that obesity management practice currently follows the 2020 Canadian Adult Obesity Clinical Practice Guidelines, which recommend lifestyle modification supported by psychological or behavioural therapy, pharmacotherapy, and bariatric surgery. The clinician groups further highlighted that lifestyle modifications alone are often insufficient to experience the weight loss needed to improve obesity-related complications such as type 2 diabetes and MASLD. The clinician groups noted that liraglutide, semaglutide, and bupropion-naltrexone are the most effective long-term pharmacologic treatments due to their impact on the neuroendocrine pathways associated with obesity. The clinician groups also outlined the role of medical bariatric centres in providing multidisciplinary care and differentiated them from nonmedical weight loss clinics. The CABPS expressed that obesity treatment should extend beyond weight loss to address dysfunctional adipose tissue driving adverse health outcomes. The clinician groups’ overview of therapies that reduce cardiorenal risk in patients with comorbid CV disease was consistent with the expert input. The clinician groups also added that patients with ASCVD become eligible for cardiac rehabilitation, which combines lifestyle and pharmacotherapeutic interventions.

Both the clinicians and experts agreed that a primary goal of CV disease treatment is to reduce the incidence of CV events, but the clinician groups also suggested additional goals for weight management and improved glycemic control, when applicable. Both the clinicians and experts highlighted the access challenges and potential risks associated with bariatric surgery. Both noted a need for additional treatment options that address both CV risk and weight reduction, noting that existing treatments for weight loss and CV disease can lack long-term effectiveness and fail to directly target weight management. The clinician groups noted this as being particularly true for patients with obesity who do not have diabetes, a group for which there are currently no targeted treatment options. The clinician groups noted that these significant treatment gaps contribute to a persisting high risk of morbidity and mortality in these populations.

In contrast to the clinical expert input, which noted that semaglutide would be combined with other therapies that reduce CV risk (i.e., RAAS inhibitors, SGLT2 inhibitors), the clinician groups indicated that semaglutide would be used for first-line treatment alongside lifestyle modifications. The clinician groups noted it is unnecessary to first try other treatments before semaglutide, given its effectiveness in achieving substantial weight loss and improving obesity-related health outcomes. The clinician groups anticipate that semaglutide could shift current treatment paradigms, noting its efficacy approaches that of bariatric surgery and could reduce the number of patients requiring surgery. The patient type that clinician groups indicated would benefit most from semaglutide aligned exactly with the type indicated by the clinical experts. The clinician groups added that semaglutide might also benefit patients with diabetes who require additional CV protection, or those without diabetes who are at high risk of CV disease and have difficulty experiencing weight loss through lifestyle modifications alone. The CRSP clinicians estimated that 78% of their patients receiving cardiac rehabilitation do not have diabetes and that 45% of patients who complete the rehabilitation program could potentially benefit from an additional therapy such as semaglutide. The patients least suitable for treatment, identified by the clinician groups and experts, also aligned. According to the clinician groups, patients with obesity are identified through self-identification or by health care providers using diagnostic tools such as BMI, waist circumference, and staging systems such as the Edmonton Obesity Staging System, which also classifies the degree of associated conditions (e.g., type 2 diabetes, impaired mobility). However, the CABPS noted there are currently no methods for identifying which patients are likely to respond to semaglutide.

The physician group input was consistent with the expert insights on the use of weight loss to assess treatment response but emphasized additional markers of treatment benefit. The clinician groups highlighted improvements in prediabetes, lipids, blood pressure, mobility, and quality of life as equally important. Conversely, TCR indicated there is no definitive evidence that the absence of weight loss precludes other treatment benefits. The CABPS provided specific criteria for a meaningful treatment response, including a 5% reduction in total body weight after 3 months of treatment (aligning with the expert input) and improved laboratory markers, reduced osteoarthritis pain, and the ability to proceed with procedures like hip replacement. The clinician groups also suggested initial evaluations every 4 to 6 weeks, then every 3 months, recognizing variability based on clinician and patient preference.

Both the clinician groups and experts agreed that semaglutide treatment should be continued indefinitely, with the clinician groups likening the regimen to statin therapy or treatment with ASA after an MI. The clinician groups suggested the discontinuation of semaglutide be considered in the following scenarios: presence of treatment intolerance or intolerable side effects that do not improve over time with appropriate countermeasures, pancreatitis, pregnancy or planning pregnancy, lack of meaningful treatment response, lack of affordability, or a more effective treatment becomes available in the future that requires semaglutide discontinuation. The clinician groups noted that semaglutide treatment could be delivered in family medicine and primary care clinics, community-based obesity management or metabolic medicine clinics, and hospital-based medical and surgical centres. Like the experts, the clinician groups indicated that any health care provider with experience managing obesity could prescribe semaglutide, noting this role should not be restricted to specialists alone.

Drug Program Input

The drug programs provide input on each drug being reviewed through the CDA-AMC reimbursement review processes by identifying issues that may impact their ability to implement a reimbursement recommendation. The implementation questions and corresponding responses from the clinical experts consulted for this review are summarized in Table 3.

Semaglutide 2.4 mg was first reviewed by CDEC for the weight management indication in 2022. The implementation questions and corresponding responses from the clinical experts consulted for the past review were summarized in the 2022 Clinical Review report on semaglutide (Wegovy) used to inform the reimbursement recommendation for semaglutide (Wegovy). Both the Clinical Review report and the recommendation document are available on the project website.

Table 3: Summary of Drug Plan Input and Clinical Expert Response

CDA-AMC = Canada’s Drug Agency; CDEC = Canadian Drug Expert Committee; CV = cardiovascular.

Clinical Evidence

The objective of this Clinical Review report is to review and critically appraise the clinical evidence submitted by the sponsor on the beneficial and harmful effects of semaglutide 2.4 mg solution for subcutaneous injection in chronic weight management in patients with a BMI of 27 kg/m² or greater and established CV disease (MI, stroke, or PAD). The focus will be placed on comparing semaglutide 2.4 mg with relevant comparators and identifying gaps in the current evidence.

A summary of the clinical evidence included by the sponsor in the review of semaglutide injection is presented in 2 sections, with the review team’s critical appraisal of the evidence included at the end of the section. The first section, the systematic review, includes pivotal studies and RCTs that were selected according to the sponsor’s systematic review protocol. The second section includes additional studies that were considered by the sponsor to address important gaps in the systematic review evidence.

Included Studies

Clinical evidence from the following is included in the review and appraised in this document:

• 1 RCT identified in the systematic review

• 1 additional study addressing a sponsor-identified gap in the evidence.

Systematic Review

Contents within this section have been informed by materials submitted by the sponsor. The following has been summarized and validated by the review team.

Description of the Study

Characteristics of the included study, the SELECT trial, are summarized in Table 4.

The SELECT trial was a phase IIIb, multinational, multicentre, randomized, double-blind, parallel-group, placebo-controlled trial that evaluated the effect of semaglutide 2.4 mg subcutaneous injection administered once weekly compared with placebo, both of which were added to the standard of care, in reducing the risk of MACE occurrences (CV death, nonfatal MI, or nonfatal stroke) in patients with established CV disease and overweight or obesity and without diabetes (N = 17,604). The key secondary objective of the trial was to evaluate the effect of semaglutide in mortality compared with placebo.

Table 4: Details of the Study Included in the Systematic Review

ABI = ankle-brachial index; BMI = body mass index; BP = blood pressure; CV = cardiovascular; EAC = event adjudication committee; eGFR = estimated glomerular filtration rate; GLP-1 RA = glucagon-like peptide-1 receptor agonist; HDL = high-density lipoprotein; HF = heart failure; hsCRP = high-sensitivity C-reactive protein; LDL = low-density lipoprotein; MACE = major adverse cardiovascular event; MI = myocardial infarction; NYHA = New York Heart Association; PAD = peripheral arterial disease; T1D = type 1 diabetes; T2D = type 2 diabetes; UACR = urinary albumin-to-creatinine ratio; UAP = unstable angina pectoris; VAS = visual analogue scale; WRSSM = weight-related sign and symptom measure.

^aEAC-confirmed event.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

The trial design is presented in Figure 1. The trial included a 16-week dose-escalation period, a maintenance period of up to 231 weeks (up to 53 months), and a 5-week follow-up period. The trial was conducted in 41 countries and regions, and patients were randomized at a total of 804 sites, 30 of which were in Canada. Patients were randomized in a 1:1 ratio, using an interactive web response system, to receive once-weekly treatment with either semaglutide 2.4 mg or placebo, both as an adjunct to the standard of care for CV disease. The trial was event driven; trial closure was planned when 1,225 primary end point events (EAC-confirmed MACEs) were reached. The trial used a group sequential design with 1 interim testing planned for superiority.

Based on its evaluation of the results from the interim analysis, on July 19, 2022, the data-monitoring committee advised to continue the trial. The final results presented in this document are based on a data cut-off date of July 18, 2023.

Population

Inclusion and Exclusion Criteria

The trial population was patients aged 45 years and older with established CV disease (prior MI, prior stroke, and/or symptomatic PAD) and a BMI of 27 kg/m² or greater (overweight or obesity).

The trial population did not include patients with any of the following in the past 60 days before screening: MI, stroke, hospitalization for unstable angina pectoris, or TIA. The trial population also did not include patients with planned coronary, carotid, or peripheral artery revascularization known on the day of screening. Patients with New York Heart Association class IV HF were also excluded from the trial.

The trial population did not include patients with type 1 and type 2 diabetes at randomization; however, patients with a history of gestational diabetes were eligible for enrolment in the trial. The trial population also did not include patients with a hemoglobin A1C of 6.5% or greater at screening. Patients who were treated with any glucose-lowering drug and/or any GLP-1 RA in the past 90 days before screening were also excluded from the trial.

Figure 1: The SELECT Trial Design

s.c = subcutaneous.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

Interventions

The intervention was semaglutide 2.4 mg subcutaneous injection once weekly as an adjunct to the standard of care for CV disease. Dose escalation was every fourth week until a maintenance dose of 2.4 mg was reached; the dose escalations were 0.24 mg, 0.5 mg, 1.0 mg, and 1.7 mg.

The comparator was placebo subcutaneous injection once weekly as an adjunct to the standard of care for CV disease. Dose escalation was every fourth week (same dosing schedule as the intervention).

An extension of dose-escalation intervals and treatment pauses could be considered if treatment was associated with unacceptable AEs or due to other circumstances. If a patient did not tolerate the target maintenance dose, then they may have stayed at the lower dose, preferably 1.7 mg once weekly. Re-escalating to the target dose or restarting treatment after treatment discontinuation was advised if it was considered safe to continue the intervention, per the investigator’s discretion.

Semaglutide 2.4 mg and placebo were supplied as prefilled pen injectors for self-administration in the thigh, abdomen, or upper arm at any time of the day, irrespective of meals. Training in the use of prefilled pen injectors was provided at randomization and at weeks 4, 8, 12, and 16, and at annual visits thereafter, as needed.

Concomitant Therapy

The standard of care (medical treatment) for comorbidities and CV risk factors could be initiated or adjusted (optimized) throughout the trial based on treatment guidelines and local clinical practice, and at the investigator’s discretion. Individualized healthy lifestyle counselling, including diet and physical activity, was offered at every study visit.

Only drugs received at screening or received during the trial for the following indications were recorded:

• to treat or prevent CV disease (e.g., antihypertensives, anticoagulants, lipid-lowering drugs, and ASA and other antiplatelet drugs)

• to treat overweight or obesity

• to treat diabetes and diabetes complications in patients who developed diabetes during the trial

• to treat an SAE or a drug that provided further information on the SAE (i.e., alternative etiology)

• in relation to a clinical trial for COVID-19 prevention or treatment

• in relation to an approved COVID-19 vaccine.

Note that simultaneous participation in a trial with the primary objective of evaluating an approved or nonapproved investigational medicinal product for the prevention or treatment of COVID-19 disease or postinfectious conditions was generally permitted at the investigator’s discretion, without discontinuing trial treatment.

Prohibited Therapy

Starting another GLP-1 RA was prohibited during the trial.

Outcomes

A list of efficacy end points assessed in this Clinical Review report is provided in Table 5, followed by descriptions of the outcome measures. The summarized end points are based on outcomes included in the sponsor’s summary of clinical evidence as well as any outcomes identified as important to this review according to the clinical experts consulted for this review and the input from the patient and clinician groups and public drug plans. Using the same considerations, the review team selected end points that were considered to be the most relevant to inform expert committee deliberations and finalized this list of end points in consultation with members of the expert committee.

Table 5: Outcomes Summarized From the SELECT Trial

AE = adverse event; CV = cardiovascular; eGFR = estimated glomerular filtration rate; GERD = gastroesophageal reflux disease; HDL = high-density lipoprotein; HF = heart failure; HHF = hospitalization for heart failure; MACE = major adverse cardiovascular event; MI = myocardial infarction; SAE = serious adverse event; UACR = urinary albumin-to-creatinine ratio; VAS = visual analogue scale; WDAE = withdrawal due to adverse event; WRSSM = weight-related sign and symptom measure.

^aAll time-to-event end points were assessed within the interval from time of randomization to the follow-up visit (up to 59 months or longer) and reported in months. If the event rate was lower than anticipated, then visits and related assessments were repeated every 13 weeks beyond week 234 (the visit scheduled before the end-of-treatment visit) until the necessary number of primary end point events were accrued.

^bStatistical testing for these end points was adjusted for multiple comparisons (hierarchal testing).

^cCardiometabolic risk factors (systolic blood pressure, total cholesterol, and HDL cholesterol) were included because they were used to inform the accompanying pharmacoeconomic analysis.

^dFor analyses of binary and continuous end points at visits after week 52, an analysis was performed only if less than 10% of patients from the full analysis set missed the relevant visit due to trial termination earlier than planned.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

CV Outcomes

The primary end point was time from randomization to the first occurrence of MACE, a composite end point consisting of CV death, nonfatal MI, or nonfatal stroke. Fatal MI was defined as an MI that occurred in 30 days before a CV death that was classified with MI as the cause of death. All other MIs were defined as nonfatal MIs. Similar definitions were applied to fatal and nonfatal stroke. Undetermined cause of death was presumed to be CV death, which was consistent with guidelines, according to the investigator.^46,47

The confirmatory secondary end points were time from randomization to CV death, time from randomization to the first occurrence of a composite HF end point consisting of CV death or HF hospitalization or urgent HF visit, and time from randomization to all-cause death.

Supportive secondary end points included time from randomization to the first occurrence of nonfatal MI and time from randomization to the first occurrence of nonfatal stroke.

The evaluation of CV end points comprised the events summarized in Table 6. The descriptions of events were used to guide the investigator in the reporting of AEs, while definitions of events for adjudication were outlined in the charter for the EAC.

Table 6: Events for Adjudication in the SELECT Trial

AE = adverse event; CABG = coronary artery bypass surgery; CV = cardiovascular; EAC = event adjudication committee; HF = heart failure; MI = myocardial infarction; PCI = percutaneous coronary intervention; UAP = unstable angina pectoris.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

An independent external EAC was established to perform ongoing adjudication, based on blinded review, of predefined event types according to definitions and guidelines outlined in the committee’s charter. All relevant events reported from the day of randomization until database lock were adjudicated. Events for adjudication were either directly reported by the investigator, identified through a standardized screening for potential events for adjudication, or identified by the EAC while reviewing source documents for another event for adjudication. The assessments made by both the EAC and the investigator were included in the clinical trial report.

All time-to-event end points were assessed within the interval from time of randomization to the follow-up visit (up to 59 months or longer) and reported in months.

Glucose Metabolism

Supportive secondary end points included time from randomization to the first occurrence of a hemoglobin A1C of 48 mmol/mol (6.5%) or greater.

Classification of glycemic status was according to the American Diabetes Association Standards of Medical Care in Diabetes published in 2018²³ and summarized in Table 7. Evaluation of glycemic status could involve any method (i.e., fast plasma glucose, plasma glucose, or hemoglobin A1C reflecting current health status) in addition to the investigator’s evaluation of all available relevant information, including prescribed glucose-lowering medications.

Table 7: Classification of Glycemic Status From the SELECT Trial

FPG = fasting plasma glucose; PG = plasma glucose.

Note: Glycemic status classifications are based on the 2018 American Diabetes Association Standards of Medical Care in Diabetes.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

Central laboratory assessments of glucose metabolism (hemoglobin A1C parameter) were performed at screening; weeks 20, 52, 104, 156, and 208; and the end-of-treatment visit.

Renal Outcomes

Supportive secondary end points included time from randomization to the first occurrence of a 5-component composite nephropathy end point consisting of the onset of persistent macroalbuminuria (UACR > 300 mg/g), persistent 50% reduction in eGFR compared with baseline, onset of a persistent eGFR of less than 15 mL/min/1.73 m², initiation of chronic renal replacement therapy (i.e., dialysis or transplant), or renal death.

A persistent change in eGFR was defined as having 2 consecutive samples (at least 4 weeks apart) meeting the criteria. A persistent change in macroalbuminuria was defined as having 2 out of 3 consecutive postbaseline samples (at least 4 weeks apart) above the limit for macroalbuminuria.

The evaluation of kidney function comprised central laboratory assessments of UACR and eGFR, and outcomes from event adjudication by the EAC (events of renal death and nephropathy, per Table 6).

Central laboratory assessments of urine samples to measure UACR (urinalysis) and blood samples to measure creatinine (eGFR was calculated based on the creatinine value using the Chronic Kidney Disease Epidemiology Collaboration equation) were performed at randomization; weeks 20, 52, 104, 156, and 208; and the end-of-treatment visit.

Cardiometabolic Risk Factors

Supportive secondary end points included change from randomization in systolic blood pressure and change from randomization in lipids (total cholesterol, HDL cholesterol, low-density lipoprotein [LDL] cholesterol, triglycerides) at week 104.

Blood pressure measurements were assessed at every study visit with an automated device; manual techniques were used only if an automated device was not available.

Central laboratory assessments of nonfasting lipids (total cholesterol, HDL cholesterol, LDL cholesterol, and triglycerides) were performed at randomization; weeks 20, 52, 104, 156, and 208; and the end-of-treatment visit.

Anthropometric Outcome

Supportive secondary end points included change from randomization in body weight at week 104.

Body weight was measured on a calibrated digital scale and recorded in kilograms or pounds at every visit during the trial.

Health-Related Quality of Life

Supportive secondary end points included change from baseline in the EQ-5D index score and EQ VAS at week 104.

The EQ-5D-5L questionnaire is a patient-reported outcome tool used to estimate health-related quality of life. The tool includes a descriptive system that provides a description of problems experienced by the patient who responded according to dimensions, a VAS that provides a score indicating overall self-rated health (score ranges from 0 to 100), and an index score that ranges from 0 to 1. A higher score indicates better self-reported health status.

The EQ-5D-5L questionnaire was administered at randomization; weeks 20, 52, 104, 156, and 208; and the end-of-treatment visit.

Symptom Burden

Exploratory end points included change from baseline in WRSSM total score at week 104.

The sponsor indicated the WRSSM tool was under development at the time of the trial. The WRSSM is a patient-reported outcome tool (a self-rated VAS) used to assess the presence and bother associated with weight-related symptoms. Specifically, it is used to assess the impact of multifaceted aspects of obesity on symptom experience in individuals with overweight or obesity. The total score ranges from 0 to 4, with higher scores indicating worse symptomatology.

The WRSSM questionnaire was administered at randomization; at weeks 20, 52, 104, 156, and 208; and the end-of-treatment visit.

Safety Assessments

Only the following types of AEs were recorded:

• all SAEs

• AEs leading to discontinuation of trial product, irrespective of seriousness

• AEs of COVID-19, irrespective of seriousness

• AEs requiring additional data collection (e.g., pancreatitis, acute renal failure, gallbladder disease, malignant neoplasm) and events for adjudication (e.g., events to evaluate CV end points), irrespective of seriousness.

Nonserious AEs not fulfilling any of the prespecified criteria were not systematically collected. The focus of the safety evaluation per the Clinical Study Report for the SELECT trial was based on the reporting of SAEs and other systematically collected events (i.e., AEs of special interest).

The prespecified AEs were collected at every study visit from the day of randomization until the end-of-trial visit. AEs were coded and reported in the Medical Dictionary for Regulatory Activities Version 26.0.

In consultation with the clinical experts, the notable harms for this review were identified to be CV AEs, cholelithiasis, nausea, vomiting, constipation, GERD, and pancreatitis.

The sponsor established an internal safety committee to endorse appropriate actions in case of a safety signal by conducting integrated blinded assessments of safety data. If the internal safety committee advised to unblind any data for further analyses, then an independent ad hoc group was established to maintain blinding of trial personnel.

Potential withdrawal or rebound effects were investigated for semaglutide 2.4 mg for weight management in the Wegovy phase III program. Except for increased appetite and hunger, no withdrawal effects were observed upon discontinuation of semaglutide treatment, per the investigator. The investigator concluded that withdrawal and rebound effects have therefore not been further investigated in the SELECT trial.

Statistical Analysis

An overview of the statistical analyses of efficacy end points from the SELECT trial is presented in Table 10.

Sample Size and Power Calculation

The trial was designed with 90% power to confirm superiority of the primary end point (i.e., reject the null hypothesis of an HR ≥ 1.0 against the 1-sided alternative of HR < 1.0, where HR was the HR of semaglutide 2.4 mg versus placebo).

The assumed true HR of 0.83 was based on a conservative assessment of the point estimate for the HR in the SUSTAIN-6 trial,⁴⁸ which was 0.74 (95% CI, 0.58 to 0.95) for a similar definition of MACE. Based on a randomization ratio of 1:1 and under the assumed true HR of 0.83, a total of 1,225 primary end point events were required.

The following assumptions were made to calculate the number of patients to be randomized:

• uniform recruitment in 28 months (2,500 patients every 4 months based on the DEVOTE trial⁴⁹)

• annual primary end point rate of 1.8% in the semaglutide 2.4 mg group and 2.2% in the placebo group, based on the SUSTAIN-6, LEADER, FOURIER, SCOUT, and IRIS trials^48,50-53 and adjusted to the inclusion and exclusion criteria of the SELECT trial

• annual lost to follow-up rate of 1% in both treatment groups.

With a sample size of 17,500 patients and under the alternative hypothesis, the 1,225 primary end point events were expected to be accrued between months 58 and 59 (approximately 5 years after randomization of the first patient), with a mean observation time of 44 months.

Interim Analysis (Group Sequential Design)

One interim test for superiority of the primary end point was introduced through a protocol amendment dated March 7, 2019. The interim test was performed when two-thirds of the planned total number of primary end point events was accrued (817 of 1,225 events). Patients without an EAC-confirmed primary end point event before the interim analysis data cut-off date were considered censored. The censoring date was defined as the end date of the first in-trial observation period or interim analysis data cut-off date. Under the design assumptions and sample size calculation, the timing of the interim test was expected to be at month 44, with a mean observation time of 30 months.

The Lan-DeMets alpha spending function, approximating the O’Brien-Fleming stopping boundaries, was used to test superiority at a studywise 1-sided type I error rate of 2.5%. Testing for futility was not included. The same Cox model described for the primary analysis of the primary end point was used for the interim analysis, addressing the primary estimand and using the fixed-sample 1-sided lower P value from the score test. Only a fixed-sample P value below the boundary specified by the error spending function enabled the data-monitoring committee to advise early trial termination for superiority. The actual stopping boundaries were based on the exact number of events available at the interim analysis.

An independent external data-monitoring committee was established to assess the benefit-risk balance by evaluating accumulated unblinded data at predefined time points and ad hoc. The data-monitoring committee advised on the continuation, modification, or early termination of the trial based on its evaluation of the result from the interim test for superiority of the primary end point. The interim analysis was performed by an external independent statistical service provider. Throughout the trial, the data-monitoring committee and the statistician were the unblinded parties, while the sponsor, investigators, event adjudication vendor, and the EAC were the blinded parties. Unblinded data were discussed at closed session meetings at which only the unblinded parties were present, while blinded data were discussed at open session meetings where all parties were present.

Based on its evaluation of the results from the interim analysis, on July 19, 2022, the data-monitoring committee advised to continue the trial. Per the Clinical Study Report, the interim analysis was found to have no impact on the conduct of the trial or on the interpretation of results; hence, the results from the interim analysis were not submitted by the sponsor.

Definitions of Estimands and Observation Periods

For all objectives, the primary estimand (intention to treat) was used to evaluate the treatment effect irrespective of adherence to treatment or changes to background medication. This estimand was addressed using the FAS and the in-trial observation period, defined in Table 8.

For the confirmatory end points, a secondary estimand was used to evaluate the treatment effect in all patients who were randomized who had remained on their randomized treatment for the entire trial. This estimand was addressed using the FAS and the first on-treatment observation period, defined in Table 8.

Table 8: Observation Periods in the SELECT Trial

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

General Considerations for Time-to-Event End Points

Vital status was systematically ascertained throughout the trial until database lock, but other types of events could not be systematically collected after withdrawal, lost to follow-up, or after the end-of-trial visit. As such, any event occurring after the in-trial observation period was not included in the analyses, unless otherwise stated.

In general, time-to-event end points were time-to-first-event end points. If a patient experienced an event of interest during the in-trial observation period, then the observation of the time-to-event end point was the time from randomization to the date of the event.

The observation of the time-to-event end point was censored if the event of interest did not occur during the in-trial observation period and if the patient was alive at the end of the observation period. The general assumption for censored observations was that the risk of experiencing an event was not changed by censoring (i.e., an assumption of independent censoring).

The observation of the time-to-event end point was terminated if the event of interest did not occur before death, unless death was part of the end point. Terminating events (competing risks) were potentially present for all time-to-event end points except for all-cause death. For the primary end point, non-CV death was a competing risk that could terminate the observation of the event of interest (MACE). Terminated observations due to competing risks from non-CV death were handled as censored observations in the Cox analysis but were not part of the independent censoring assumption.

Statistical Test

The primary analysis addressed the primary estimand (intention to treat). The HR (and the 2-sided 95% CI and 1-sided fixed design P value for hypothesis testing) for comparing semaglutide 2.4 mg versus placebo was estimated from a Cox proportional hazards model with treatment group as a fixed factor under the assumption of independent censoring. The score test from the Cox model was used for testing. The assumption of proportional hazards was investigated by residuals. Tied event times were handled using the exact method and CIs were based on the profile likelihood.

The population-level summary measure for time-to-event end points was the HR for semaglutide 2.4 mg versus placebo. Cumulative incidence functions for time-to-event end points were estimated with the Aalen-Johansen estimator, which accounted for competing risks.

The nominal significance level was calculated using the alpha spending function and based on the actual observed number of events at the analysis. Final inference on termination was adjusted for the group sequential design by using likelihood ratio ordering.

The handling of intercurrent events for the primary estimand and primary end point is summarized in Table 9. Intercurrent events were reported using descriptive statistics.

Table 9: Handling of Intercurrent Events for the Primary Estimand and Primary End Point in the SELECT Trial

CV = cardiovascular; MACE = major adverse cardiovascular event.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

Missing Data

Unobserved data pertaining to patients who withdrew consent or were lost to follow-up and were not censored at the time point in question were considered missing data, irrespective of any vital status data collected at the end of the trial.

Sensitivity Analyses

Because the primary analysis assumed independent censoring for patients who withdrew consent or were lost to follow-up, a 2-way tipping-point analysis based on the approach described by Zhao et al.⁵⁴ was performed. In this sensitivity analysis, imputation of event times for these patients was based on the conditional event distribution with a penalty (i.e., the risk of MACE changed after censoring compared with while under observation). Multiple imputed datasets were analyzed with separate Cox regressions and results were combined using the Rubin rule. The tipping points were defined as the combination of changes in penalties in each treatment group needed to turn around the superiority conclusion.

Two additional sensitivity analyses were performed by multiple imputation of event times for patients who withdrew consent or were lost to follow-up:

• The first sensitivity analysis used an estimated annual event rate from patients who permanently discontinued treatment but remained in the trial (i.e., the event rate was based on events and time while these patients were permanently off treatment; this period was defined as > 5 weeks since any dose of trial product was administered, and the patient remained off treatment for the remainder of the trial).

• The second sensitivity analysis used an estimated annual event rate from patients who discontinued treatment at any point in the trial, which avoids conditioning on the future (i.e., the event rate was based on events that occurred from the first time they were off treatment and until the end of the in-trial observation period; therefore, this can include time periods when patients went back on trial treatment).

  For each sensitivity analysis, the off-treatment event time data were fitted within treatment groups to an exponential distribution using Bayesian analysis and accounting for censoring. A noninformative prior distribution was used for the rate parameter in each treatment group. A total of 500 replicates of the 2 off-treatment event rates were randomly sampled from the posterior distribution. A total of 500 copies of the original dataset were created and linked to the corresponding replicate of the off-treatment event rates. For each patient who was censored due to withdrawal or lost to follow-up, the event time was imputed by adding a random variable to the original censoring date. The random variables were generated from an exponential distribution using the off-treatment event rate for the corresponding replicate and treatment group, rounded up to whole days. If the imputed event time occurred after the patient’s planned end-of-trial time, then the patient was censored at the planned end-of-trial time. Each complete dataset was analyzed using the same Cox regression used in the primary analysis to provide the estimated log HR and associated standard error, which were pooled using the Rubin rule to obtain a single point estimate, CI, and P value.

Supplementary Analyses

The following supplementary analyses were performed:

• absolute risk difference estimate and 95% CI between semaglutide 2.4 mg and placebo at week 156 based on the cumulative incidence function for each treatment group

• an analysis using a Cox proportional hazards model with the first on-treatment period to address the secondary estimand

• an analysis of non-CV death using a Cox proportional hazards model to assess the influence of the competing risk of non-CV death on the results from the primary analysis

• various supplementary analyses to assess the impact of the COVID-19 pandemic as well as the impact of coparticipation in COVID-19 trials:

  • Supplementary analyses were done to assess the impact of the COVID-19 pandemic on the primary end point. The analyses addressed 2 scenarios: 1 in which MACEs have been impacted by an increased MACE rate and potentially a different treatment effect for events occurring concurrently with COVID-19 infection, and a second scenario in which there was a reduced MACE rate due to concurrent COVID-19 infection leading to fewer CV deaths because the patient died (prematurely) due to COVID-19 infection and not their underlying atherosclerotic disease.

Subgroup Analyses

To explore the consistency in the treatment effect for the primary end point, subgroup analyses based on the following baseline characteristics were performed:

• sex (female, male)

• age (< 55 years, 55 to 65 years, 65 to 75 years, ≥ 75 years)

• race (white, Black or African American, Asian, other)

• ethnicity (Hispanic or Latino, not Hispanic or Latino)

• region (Europe, North America, Asia, other)

• hemoglobin A1C less than 5.7% (yes, no)

• BMI (< 30 kg/m², 30 kg/m² to < 35 kg/m², 35 kg/m² to < 40 kg/m², 40 kg/m² to < 45 kg/m², ≥ 45 kg/m²)

• CV disease (MI alone, stroke alone, PAD alone, any combination)

• eGFR of 60 mL/min/1.73 m² or less per the Chronic Kidney Disease Epidemiology Collaboration equation (yes, no)

• chronic HF (yes, no).

The subgroup analyses used Cox proportional hazards models with an interaction between the treatment groups and the specific subgroup as factor.

Of the preceding subgroups listed, BMI was identified as the most relevant for the purpose of this review to inform expert committee deliberations. Consultation with the clinical experts did not identify any relevant potential treatment-effect modifiers to examine for this review.

Testing Hierarchy

One major (classification per investigator) change to the statistical analyses before trial unblinding was implemented through a protocol amendment dated February 9, 2022. The composite HF end point, consisting of EAC-confirmed CV death or HF hospitalization or urgent HF visit, was moved from being a supportive secondary end point to a confirmatory secondary end point. This change was reportedly introduced to add to the observed overall data indicating a beneficial effect of GLP-1 RA in preventing HF events.

A second major (classification per investigator) change to the statistical analyses before trial unblinding was implemented through a protocol amendment dated January 4, 2021. The statistical testing strategy for the confirmatory secondary end points was updated with a separate alpha spending function, as described by Glimm et al.,⁵⁵ to control the type I error rate at a 1-sided level of 2.5%.

If superiority of the primary end point was confirmed, then the superiority hypothesis was tested for each confirmatory secondary end point under multiplicity control using a stagewise hierarchical testing scheme according to the following order:

• time from randomization to EAC-confirmed CV death

• time from randomization to first occurrence of a composite HF end point consisting of EAC-confirmed CV death or HF requiring hospitalization or urgent HF visit

• time from randomization to EAC-confirmed all-cause death.

The testing procedure was stopped the first time an analysis failed to confirm superiority of the end point in question using the nominal significance level. No adjustment to the results for the group sequential design was made for the confirmatory secondary end points.

The nominal significance levels for the confirmatory secondary end points were derived from the alpha spending function and were based on the actual observed number of events at the analysis.

Secondary End Points

For confirmatory end points controlled for multiplicity, estimated treatment effects were presented together with 2-sided 95% CIs and 1-sided P values for superiority tests. For the reporting of results, the HR and 95% CI were presented with the 2-sided P value.

For nonconfirmatory end points, estimated treatment effects were reported with 2-sided 95% CI and 2-sided P values.

Time-to-Event Confirmatory Secondary End Points

The confirmatory secondary end points were analyzed using a Cox proportional hazards model as described for the primary end point and addressed the primary estimand.

The 2 additional sensitivity analyses with imputation from patients off treatment were also performed for the confirmatory secondary end points. Supplementary analyses (absolute risk difference at week 156 and first on-treatment observation period) described for the primary end point were also performed for the confirmatory secondary end points.

For the composite HF end point, an additional supplementary analysis was performed by replacing the CV death component with all-cause death.

For the end point of all-cause death, an additional supplementary analysis was performed using the FAS and an extended in-trial observation period that included the follow-up for vital status for patients who withdrew consent or were lost to follow-up.

Time-to-Event Supportive Secondary End Points

The supportive secondary end points were analyzed using a Cox proportional hazards model as described for the primary end point and addressed the primary estimand.

For nonfatal MI and nonfatal stroke, supplementary analyses were performed for end points defined as time from randomization to first MI (fatal or nonfatal) and time from randomization to first stroke (fatal or nonfatal).

Continuous Supportive Secondary End Points

The population-level summary measure for continuous end points was the mean difference for semaglutide 2.4 mg versus placebo. End points on lipid parameters were logarithmic transformed and the mean difference on the logarithmic scale was back transformed to the original scale and reported as geometric mean ratios.

The continuous supportive secondary end points (change from baseline at week 104) were analyzed by an analysis of covariance (ANCOVA) with multiple imputation for missing values under a missing-at-random assumption. The imputation model (linear regression) was performed separately for each treatment group and included baseline value as a covariate and fitted to patients with an observed data point, irrespective of adherence to randomized treatment. The fitted model was used to impute values for all patients with missing data at week 104 to create complete datasets. Patients without a baseline measurement were not part of the model. The Rubin rule was used to combine the results. The complete datasets were analyzed by ANCOVA with treatment as fixed factor and baseline value as covariate.

Analyses for body weight were carried out at weeks 52, 156, and 208 if less than 10% of patients from the FAS missed the corresponding visit due to trial termination.

Table 10: Overview of Statistical Analyses of Efficacy End Points From the SELECT Trial

ANCOVA = analysis of covariance; BP = blood pressure; CV = cardiovascular; eGFR = estimated glomerular filtration rate; HDL = high-density lipoprotein; HF = heart failure; HHF = hospitalization for heart failure; MACE = major adverse cardiovascular event; MI = myocardial infarction; UACR = urinary albumin-to-creatinine ratio; VAS = visual analogue scale; WRSSM = weight-related sign and symptom measure.

^aFor composite end points, the ordering defined the hierarchy of components when reporting events contributing to a composite end point when there was a tie in the event dates. Tied event times were handled using the exact method.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

A sensitivity analysis using a jump to reference multiple imputation approach was performed for change from baseline in body weight at week 104 (and weeks 52, 156, and 208 if less than 10% of patients from the FAS missed the corresponding visit due to trial termination). This analysis used a multiple imputation approach in which missing values from both treatment groups were imputed using an imputation model for the placebo group only, irrespective of adherence to randomized treatment at week 104.

Exploratory End Point

Change from baseline in WRSSM total score at week 104 was presented descriptively.

Analysis Population

The FAS was defined as all patients who were randomized; all efficacy and safety analyses were performed according to the treatment assigned at randomization.

For patients who were randomized more than once, only the patient identification number and treatment corresponding to the first randomization were included in the FAS. Any additional identification numbers of patients who were randomized were excluded from the FAS.

Results

Patient Disposition

A summary of patient disposition from the SELECT trial is presented in Table 11.

A total of 21,089 patients were screened, of which 3,480 patients (16.5%) were not eligible for randomization because they did not meet all eligibility criteria, did not return for randomization, or withdrew consent before randomization. The most frequently reported reason for screening failure was due to a hemoglobin A1C level of 6.5% or greater in 1,724 patients (49.5%).

Initially, a total of 8,805 patients were randomized to receive semaglutide 2.4 mg and 8,804 patients were randomized to receive placebo. However, 5 patient identifiers were removed because they were randomized more than once. Thus, the FAS comprised 8,803 patients in the semaglutide 2.4 mg group and 8,801 patients in the placebo group.

Of the patients who were randomized and included in the FAS, a total of 259 patients (2.9%) in the semaglutide 2.4 mg group and 284 patients (3.2%) in the placebo group discontinued from the study due to either withdrawal by patient or lost to follow-up. A total of █████ patients (█████) in the semaglutide 2.4 mg group and █████ patients (█████) in the placebo group discontinued treatment. Notably, 1,434 patients (16.3%) in the semaglutide 2.4 mg group and 696 patients (7.9%) in the placebo group discontinued treatment due to AEs, while 64 patients (0.7%) in the semaglutide 2.4 mg group and 244 patients (2.8%) in the placebo group discontinued treatment due to lack of effect.

Of the patients who were randomized and included in the FAS, a total of 4 patients (< 0.1%) did not have a study visit at week 104 due to trial closure. Notably, 4,814 patients (27.3%) did not have a study visit at week 156 and 13,977 patients (79.4%) did not have a study visit at week 208 due to trial closure.

Table 11: Summary of Patient Disposition From the SELECT Trial

FAS = full analysis set.

^aFrequency ≥ 5.0%.

^bRescreening was permitted for all inclusion and exclusion criteria; a new patient identification number was assigned in the interactive web response system. For patients randomized more than once, only the patient identification number and treatment corresponding to the first randomization were included in the FAS. The additional identification numbers of patients who were randomized were excluded from the FAS.

^cProportion of patients in the FAS who missed the annual visit due to trial termination. If 10% or more of patients from the FAS missed the annual visit due to trial termination, then the analysis for the supportive secondary end points, body weight and hemoglobin A1C, was not performed for the corresponding week.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Protocol Deviations and Serious Breaches

A summary of important protocol deviations from the SELECT trial is presented in Table 12. No patient in the FAS was excluded from the analyses due to protocol deviations.

Several site-level and patient-level important protocol deviations regarding quality and good clinical practice issues were reported at 2 sites, resulting in serious breaches. A total of 11 serious breaches of good clinical practice were reported in the trial. An assessment of the potential impact on patient safety and rights and data integrity was performed by the investigator; only those breaches that were considered to have a potential impact on data integrity are summarized in Table 13.

There were 3 cases of unintentional unblinding before database lock involving unintentional unblinding of clinical study group, treatment, and trial product batch numbers.

There was a total of 724 cases of intentional unblinding before database lock to handle suspected trial product–related unexpected serious adverse reactions as part of the regulatory reporting requirements for SAEs. Breaking the randomization code for a patient before database lock in the case of a suspected unexpected serious adverse reaction was prespecified. It was reported that these cases of intentional unblinding were handled according to internal procedures, and it was concluded that they did not affect trial data integrity.

Table 12: Summary of Important Protocol Deviations From the SELECT Trial (Patient Level)

AE = adverse event; SAE = serious adverse event.

Notes: The data cut-off date was July 18, 2023.

Protocol deviations deemed important were defined as deviations that could significantly impact the completeness, accuracy, and/or reliability of the trial data or that could significantly affect the patient’s rights, safety, or well-being.

Percentages were of the total number of patient-level protocol deviations within each category.

Only a frequency of ≥ 10.0% by subcategory in any treatment group was summarized in the preceding table.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

Table 13: Summary of Serious Breaches With Potential Impact on Data Integrity From the SELECT Trial

AE = adverse event; EDC = electronic data capture; GCP = good clinical practice; HDL = high-density lipoprotein; LDL = low-density lipoprotein; SAE = serious adverse event; VLDL = very-low-density lipoprotein.

Note: A total of 11 serious breaches of GCP were reported in the trial. An assessment of the potential impact on patient safety and rights and data integrity was performed by the investigator; only those breaches that were considered for potential impact on data integrity were summarized in the preceding table.

Source: Clinical Study Report version 1.0 for the SELECT trial.²⁵

Intercurrent Events

A summary of intercurrent events from the SELECT trial is presented in Table 14.

Notably, █████ of 8,803 patients (█████) in the semaglutide 2.4 mg group and █████ of 8,801 patients (█████) in the placebo group discontinued treatment, and █████ patients (█████) in the semaglutide 2.4 mg group and █████ patients (█████) in the placebo group started or changed medication that modifies CV risk. Medication modifying CV risk comprised glucose-lowering medication, weight-lowering medication, and CV-related medication.

Table 14: Redacted

AE = adverse event; CDA-AMC = Canada’s Drug Agency; CV = cardiovascular; SAE = serious adverse event.

^aMore than 99% of patients were on medications modifying CV risk. An initiation may be a change in medication and not a reflection of a true modification. Medication modifying CV risk comprised glucose-lowering medication, weight-lowering medication, and CV-related medication.

^bOne of the AEs was reported as an SAE. Because bariatric surgery is a surgical procedure, only those procedures related to SAEs were systematically collected, per protocol. Non-SAE hospitalizations were collected, but with less detailed information.

Source: Novo Nordisk Canada Inc. response to the January 15, 2025, CDA-AMC request for additional information regarding the CDA-AMC review of Wegovy.²⁶

Baseline Characteristics

The baseline characteristics outlined in Table 15 are limited to those that are most relevant to this review or were felt to affect the outcomes or interpretation of the study results.

The mean age of patients in the semaglutide 2.4 mg group was 61.6 years (SD = 8.9 years) and 61.6 years (SD = 8.8 years) in the placebo group. The mean BMI of patients in the semaglutide 2.4 mg group was 33.30 kg/m² (SD = 5.03 kg/m²) and 33.37 kg/m² (SD = 5.04 kg/m²) in the placebo group. Most patients in the FAS had a BMI between 30 kg/m² and 35 kg/m², 3,693 of 8,803 patients (42.0%) in the semaglutide 2.4 mg group and 3,781 of 8,801 patients (43.0%) in the placebo group.

The mean hemoglobin A1C was 5.78% (SD = 0.34%) in the semaglutide 2.4 mg group and 5.78% (SD = 0.33%) in the placebo group. Most patients in the FAS had prediabetes, 5,701 patients (64.8%) in the semaglutide 2.4 mg group and 5,661 patients (64.3%) in the placebo group, according to the classification of glycemic status published in 2018 in the American Diabetes Association Standards of Medical Care in Diabetes.²³

Most patients in the FAS had a history of MI, 5,962 patients (67.7%) in the semaglutide 2.4 mg group and 5,944 patients (67.5%) in the placebo group. A total of 2,155 patients (24.5%) in the semaglutide 2.4 mg group and 2,131 patients (24.2%) in the placebo group had chronic HF, of which 1,174 patients (13.3%) in the semaglutide 2.4 mg group and 1,099 patients (12.5%) in the placebo group had HF with preserved ejection fraction.

Table 15: Summary of Baseline Characteristics From the SELECT Trial (FAS)

BMI = body mass index; CV = cardiovascular; DBP = diastolic blood pressure; eGFR = estimated glomerular filtration rate; FAS = full analysis set; HF = heart failure; hsCRP = high-sensitivity C-reactive protein; MI = myocardial infarction; NYHA = New York Heart Association; PAD = peripheral arterial disease; SBP = systolic blood pressure; SD = standard deviation.

^aCategories are as reported in the study.

^bA total of 7 patients (3 in the semaglutide 2.4 mg group and 4 in the placebo group) had hemoglobin A1C ≥ 6.5% at screening and were randomized in error (captured in protocol deviations). These 7 patients were categorized by the investigator as being either normoglycemic or having prediabetes at baseline.

^cRenal function category was based on the eGFR calculated by the central laboratory based on the creatinine value using the Chronic Kidney Disease Epidemiology Collaboration equation.

^dIncluded patients who were not known to have fulfilled only 1 or more than 1 CV inclusion criteria (i.e., a criterion was selected in the electronic case report form but the other criteria were left blank, and it was not assumed that missing responses meant criteria not met), and patients who were randomized in error and did not fulfill any criteria (these cases were categorized as protocol deviations).

^eThreshold indicative of inflammation, per investigator.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Exposure to Study Treatments

A summary of patient exposure from the SELECT trial is presented in Table 16.

After the initial 16 weeks of dose escalation, approximately 75% of patients who were receiving treatment at that time point in the semaglutide group were receiving the planned maintenance dose of 2.4 mg. The sponsor noted that 77.12% of patients were receiving the semaglutide 2.4 mg dose at any point in the study, indicating that the patients not receiving the target dose at week 16 were not just extending the dose-escalation phase. For patients who did not reach the semaglutide 2.4 mg target dose at any time in the study, the maximum semaglutide dose was 1.7 mg for 7.97%, 1.0 mg for 7.22%, 0.5 mg for 4.79%, and 0.24 mg for 2.78% of patients.⁵⁶

A total of 9 patients (0.1%) in the semaglutide 2.4 mg group and 19 patients (0.2%) in the placebo group in the FAS were not exposed to the trial product; no patient in the FAS was excluded from the analyses.

The mean duration of follow-up was 39.9 months (SD = 9.3 months) in the semaglutide 2.4 mg group and 39.7 months (SD = 9.5 months) in the placebo group. The mean duration of exposure to semaglutide 2.4 mg was 33.3 months (SD = 14.4 months) and to placebo was 35.1 months (SD = 13.0 months).

Patients in the semaglutide 2.4 mg group and the placebo group received their assigned treatment for 82.5% (SD = 30.8%) and 87.7% (SD = 25.2%) of the potential treatment time, respectively.

Table 16: Summary of Patient Exposure From the SELECT Trial (FAS)

FAS = full analysis set; SD = standard deviation.

^aThe on-treatment period was defined as all days from the date of the first dose to the end of the in-trial period during which a dose was administered within 5 weeks (35 days). The period could include nonconsecutive time intervals.

^bThe first on-treatment period was defined as the date of the first dose until the first time that no dose was administered within 5 weeks (35 days) or the end of the in-trial period, whichever occurred first.

^cThe adherence on-treatment period was defined as all days from the date of the first dose to the end of the in-trial period during which a dose was administered within 1 week (7 days). The period could include nonconsecutive time intervals. The dose did not have to be the target dose.

^dThe planned on-treatment period was defined as the date of randomization until the end-of-treatment visit plus 6 days or the end of the in-trial period, whichever occurred first.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Concomitant Treatments

A summary of concomitant CV-management and weight management–related treatments ongoing at randomization from the SELECT trial is presented in Table 17. Patients receiving glucose-lowering medication or any GLP-1 RA in the past 90 days before screening were not eligible to enrol in the trial. Notably, the majority of patients (> 85% of patients in the FAS) were receiving CV drugs, lipid-lowering drugs, and platelet aggregation inhibitors at randomization, while few patients (< 0.1% of patients in the FAS) were receiving weight management–related medications.

A summary of concomitant CV- and weight management–related and glucose-lowering treatments initiated in the in-trial observation period is presented in Table 18. Glucose-lowering medications (except GLP-1 RA) could be initiated during the trial if the patient developed diabetes. The initiation of a medication could be due to adding a new drug or switching from 1 product to another, which was perceived as a change in medication by the investigator. Notably, the initiation of any CV-related, weight management–related, or glucose-lowering medication after randomization was slightly higher in the placebo group than observed in the semaglutide 2.4 mg group.

Table 17: Summary of Concomitant CV-Management and Weight Management–Related Treatments Ongoing at Randomization From the SELECT Trial (FAS)

ACE = angiotensin-converting enzyme; ARNI = angiotensin receptor–neprilysin inhibitor; ASA = acetylsalicylic acid; CV = cardiovascular; DOAC = direct oral anticoagulant; FAS = full analysis set.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Table 18: Summary of Concomitant CV-Management and Weight Management–Related and Glucose-Lowering Treatments Initiated in the In-Trial Observation Period From the SELECT Trial (FAS)

ACE = angiotensin-converting enzyme; ARNI = angiotensin receptor–neprilysin inhibitor; ASA = acetylsalicylic acid; CV = cardiovascular; DOAC = direct oral anticoagulant; DPP-4 = dipeptidyl peptidase-4; FAS = full analysis set; GLP-1 RA = glucagon-like peptide-1 receptor agonist; SGLT2 = sodium-glucose cotransporter-2.

Note: Initiation of a medication could be due to adding a new medication or switching from one product to another which was perceived as a change in medication by the investigator.

^aGlucose-lowering medications (except GLP-1 RA) could be initiated during the trial if the patient developed diabetes.

^bInitiating another GLP-1 RA was not permitted during the entire trial, otherwise, treatment with the trial product was discontinued.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Efficacy

A summary of key efficacy results based on the in-trial observation period and a data cut-off date of July 18, 2023, is presented in Table 19.

CV Outcomes

MACE (CV Death, Nonfatal MI, or Nonfatal Stroke)

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed MACE (consisting of CV death, nonfatal MI, or nonfatal stroke) were 6.5% of 8,803 patients in the semaglutide 2.4 mg group and 8.0% of 8,801 patients in the placebo group. Semaglutide 2.4 mg was favoured over placebo (adjusted HR = 0.80; 95% CI, 0.72 to 0.90). A cumulative incidence plot of time from randomization to first EAC-confirmed MACE is presented in Figure 3 in Appendix 1. A summary of absolute risk difference from the supplementary analysis of confirmatory end points at week 156 is presented in Table 27 in Appendix 1. The absolute risk difference in time to first EAC-confirmed MACE between semaglutide 2.4 mg and placebo at week 156 was −0.011 (95% CI, −0.019 to −0.004) (for interpretation of results, transformed to −1.1%; 95% CI, −1.9% to −0.4%).

The 2-way tipping-point sensitivity analysis suggested that the results from the primary analysis were robust (i.e., the observed treatment effect of semaglutide 2.4 mg compared with placebo was unlikely affected by missing data).

The 2 additional sensitivity analyses, which used multiple imputation to impute events for patients who withdrew consent or were lost to follow-up, estimated the HRs for MACE, which were the same as the result from the primary analysis, and estimated the associated 95% CIs, which were consistent with the result from the primary analysis. For the sensitivity analysis using imputation of event times based on patients who permanently discontinued trial treatment, the HR was 0.80 (95% CI, 0.72 to 0.90). For the sensitivity analysis using imputation of event times based on patients not receiving treatment at some point during the trial, the HR was 0.80 (95% CI, 0.72 to 0.89).

The on-treatment supplementary analysis — which defined treatment exposure as the period from the date of the first dose until the first time during which no dose had been administered in the past 5 weeks or the end of the in-trial period, whichever occurred first — estimated the HR and associated 95% CI for MACE (HR = 0.76; 95% CI, 0.67 to 0.87), and the result was consistent with the result from the primary analysis.

The competing risk supplementary analysis on EAC-confirmed non-CV deaths (141 in the semaglutide 2.4 mg group and 176 in the placebo group) estimated the HR and associated 95% CI, which were similar to the result from the primary analysis (EAC-confirmed non-CV death HR of 0.77; 95% CI, 0.62 to 0.95).

Consultation with the clinical experts did not identify any relevant potential treatment-effect modifiers to examine for this review. However, BMI was identified as the most relevant subgroup for the purpose of this review to inform expert committee deliberations and, as such, a summary of the subgroup analysis results by BMI for the primary end point is presented in Table 28 in Appendix 1.

EAC-confirmed events of MACE with onset after the in-trial observation period comprised 6 events of undetermined cause of death in the placebo group: 1 event of non-CV, nonrenal death in the placebo group; 3 events of acute MI (1 event in the semaglutide 2.4 mg group and 2 events in the placebo group); and 2 events of stroke in the placebo group. Note that events of MI and stroke were not systematically collected outside the in-trial period. For patients who did not complete the study, vital status was collected at the end of trial to the extent possible.

CV Death

At the data cut-off date, the percentages of patients (in the FAS) with an EAC-confirmed CV death (including undetermined cause of death) were 2.5% in the semaglutide 2.4 mg group and 3.0% in the placebo group (HR = 0.85; 95% CI, 0.71 to 1.01). A cumulative incidence plot of time from randomization to EAC-confirmed CV death is presented in Figure 4 in Appendix 1. The absolute risk difference in time to EAC-confirmed CV death between semaglutide 2.4 mg and placebo at week 156 was 0% (95% CI, −0.5% to 0.4%).

As with the primary analysis, the 2 sensitivity analyses using multiple imputation to impute events for those who were lost to follow-up or withdrew consent estimated HRs for CV death and associated 95% CIs (for both sensitivity analyses: HR = 0.84; 95% CI, 0.70 to 1.00) that were consistent with the result from the confirmatory secondary analysis.

The on-treatment supplementary analysis estimated the HR for CV death and associated 95% CI (HR = 0.85; 95% CI, 0.67 to 1.07) that were consistent with the result from the confirmatory secondary analysis.

HF Composite (CV Death or Hospitalization for HF or Urgent HF Visit)

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed composite HF outcome (comprising CV death or HF requiring hospitalization or urgent HF visit) were 3.4% in the semaglutide 2.4 mg group and 4.1% in the placebo group (HR = 0.82; 95% CI, 0.71 to 0.96). A cumulative incidence plot of time from randomization to first EAC-confirmed composite HF outcome is presented in Figure 5 in Appendix 1. The absolute risk difference in time to EAC-confirmed composite HF outcome between semaglutide 2.4 mg and placebo at week 156 was −0.2% (95% CI, −0.8% to 0.3%).

The 2 sensitivity analyses using multiple imputation to impute events for those who were lost to follow-up or withdrew consent estimated HRs for the HF composite outcome (and associated 95% CIs) that were consistent with the result from the confirmatory secondary analysis. For the sensitivity analysis using imputation of event times based on patients who permanently discontinued trial treatment, the HR was 0.82 (95% CI, 0.71 to 0.96). For the sensitivity analysis using imputation of event times based on patients not receiving treatment at some point during the trial, the HR was 0.82 (95% CI, 0.70 to 0.96).

The on-treatment supplementary analysis estimated the HR for the HF composite outcome and associated 95% CI (HR = 0.76; 95% CI, 0.63 to 0.93) that were consistent with the result from the confirmatory secondary analysis.

The supplementary analysis replacing the CV death component with all-cause death in the HF composite outcome estimated the HR and associated 95% CI (HR = 0.80; 95% CI, 0.70 to 0.91), which were consistent with the result from the confirmatory secondary analysis.

All-Cause Death

At the data cut-off date, the percentages of patients (in the FAS) with an EAC-confirmed all-cause death were 4.3% in the semaglutide 2.4 mg group and 5.2% in the placebo group (HR = 0.81; 95% CI, 0.71 to 0.93). A cumulative incidence plot of time from randomization to EAC-confirmed all-cause death is presented in Figure 6 in Appendix 1. The absolute risk difference in time to EAC-confirmed all-cause death between semaglutide 2.4 mg and placebo at week 156 was −0.5% (95% CI, −1.1% to 0.1%).

The 2 sensitivity analyses using multiple imputation to impute events for those who were lost to follow-up or withdrew consent estimated HRs for all-cause death and associated 95% CIs that were consistent with the result from the confirmatory secondary analysis. For the sensitivity analysis using imputation of event times based on patients who permanently discontinued trial treatment, the HR was 0.81 (95% CI, 0.71 to 0.93). For the sensitivity analysis using imputation of event times based on patients not receiving treatment at some point during the trial, the HR was 0.81 (95% CI, 0.70 to 0.93).

The on-treatment supplementary analysis estimated the HR for all-cause death and associated 95% CI (HR = 0.74; 95% CI, 0.61 to 0.89) that were consistent with the result from the confirmatory secondary analysis.

The supplementary analysis based on an extended in-trial observation period that included follow-up for vital status in patients who withdrew consent or were lost to follow-up estimated the HR for all-cause death and associated 95% CI (HR = 0.81; 95% CI, 0.71 to 0.92) that were consistent with the result from the confirmatory secondary analysis.

Nonfatal MI

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed nonfatal MI were 2.7% in the semaglutide 2.4 mg group and 3.7% in the placebo group (HR = 0.72; 95% CI, 0.61 to 0.85).

The supplementary analysis on first EAC-confirmed fatal or nonfatal MI estimated the HR and associated 95% CI (HR = 0.72; 95% CI, 0.61 to 0.85), which were consistent with the result from the supportive secondary analysis.

Nonfatal Stroke

At the data cut-off date, the percentages of patients (in the FAS) with their first EAC-confirmed nonfatal stroke were 1.7% in the semaglutide 2.4 mg group and 1.9% in the placebo group (HR = 0.93; 95% CI, 0.74 to 1.15).

The supplementary analysis on first EAC-confirmed fatal or nonfatal stroke estimated the HR and associated 95% CI (HR = 0.89; 95% CI, 0.72 to 1.11), which were consistent with the result from the supportive secondary analysis.

Glucose Metabolism

At the data cut-off date, 306 patients (3.5% of the FAS) in the semaglutide 2.4 mg group and 1,059 patients (12.0% of the FAS) in the placebo group experienced a first occurrence of a hemoglobin A1C of 6.5% or greater (HR = 0.27; 95% CI, 0.24 to 0.31).

Renal Outcomes

At the data cut-off date, 155 patients (1.8% of the FAS) in the semaglutide 2.4 mg group and 198 patients (2.2% of the FAS) in the placebo group had experienced a first composite nephropathy event (HR = 0.78; 95% CI, 0.63 to 0.96).

Anthropometric Outcomes

The mean treatment difference in the percent change from baseline in body weight at week 104 between semaglutide 2.4 mg and placebo was −8.51% (95% CI, −8.75% to −8.27%).

The sensitivity analysis that imputed missing values in both groups using an imputation model for the placebo group only, irrespective of adherence to randomized treatment at week 104, estimated the treatment difference (mean treatment difference of −7.39%; 95% CI, −7.63% to −7.16%), which was similar to the result from the main analysis for change in body weight.

The analyses at weeks 156 and 208 were not performed as planned because greater than 10% of patients in the FAS missed the relevant annual study visit due to trial termination. However, an analysis of the percent change from baseline in body weight was performed at week 208 and the estimated treatment difference (based on the in-trial period) between semaglutide 2.4 mg and placebo was −8.7% (95% CI, −9.42% to −7.88%). The on-treatment analysis, which defined treatment exposure as the observation period until first time off treatment for greater than 35 days, estimated the treatment difference (−10.2%; 95% CI, −11.0% to −9.42%), which was similar to the result from the main analysis for change in body weight at week 208.²⁴

Health-Related Quality of Life

The treatment difference in the change from baseline in EQ-5D index score at week 104 between semaglutide 2.4 mg and placebo was 0.01 (95% CI, 0.01 to 0.02).

The treatment difference in the change from baseline in EQ VAS score at week 104 between semaglutide 2.4 mg and placebo was 1.60 (95% CI, 1.16 to 2.04).

Table 19: Summary of Key Efficacy Results From the In-Trial Observation Period in the SELECT Trial (FAS)

ANCOVA = analysis of covariance; CDA-AMC = Canada’s Drug Agency; CI = confidence interval; CV = cardiovascular; EAC = event adjudication committee; eGFR = estimated glomerular filtration rate; FAS = full analysis set; HF = heart failure; HR = hazard ratio; MACE = major adverse cardiovascular event; MI = myocardial infarction; NSTEMI = non–ST elevation myocardial infarction; SD = standard deviation; SE = standard error; STEMI = ST elevation myocardial infarction; VAS = visual analogue scale; vs. = versus; WRSSM = weight-related sign and symptom measure.

Note: The data cut-off date was July 18, 2023.

^aAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. The assumption of proportional hazards was evaluated by the standardized score process and was considered supported, per investigator. Patients without events of interest were censored at the end of their in-trial period. Based on the available number of events for analysis, the nominal significance level was updated to 0.02281 using the Lan-DeMets alpha spending function. Adjustment for group sequential design was done using likelihood ratio ordering.

^bAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. Patients without events of interest were censored at the end of their in-trial observation period.

^cBased on the available number of events for analysis, the nominal significance level was updated to 0.01148 using the alpha spending function described in the statistical analysis plan.

^dThe superiority of semaglutide 2.4 mg versus placebo was confirmed for the primary end point of time to first EAC-confirmed MACE comprising CV death, nonfatal MI, and nonfatal stroke. Superiority of semaglutide 2.4 mg versus placebo was not confirmed for the confirmatory secondary end point of time to EAC-confirmed CV death. As such, superiority of semaglutide 2.4 mg versus placebo was not tested for the confirmatory secondary end points that followed the order of the prespecified testing hierarchy (time to first EAC-confirmed composite HF outcome and time to EAC-confirmed all-cause death).

^eBased on the available number of events for analysis, the nominal significance level was updated to 0.01149 using the alpha spending function described in the statistical analysis plan.

^fBased on the available number of events for analysis, the nominal significance level was updated to 0.01213 using the alpha spending function described in the statistical analysis plan.

^gNot controlled for multiplicity.

^hAnalyzed using a Cox proportional hazards model with treatment as categorical fixed factor. Patients without events of interest were censored at the end of their in-trial observation period. Patients randomized in error with a baseline hemoglobin A1C ≥ 48 mmol/mol (6.5%) were excluded from this analysis.

^IAnalyzed using an ANCOVA with treatment as fixed factor and baseline value as covariate. Before analysis, missing data were multiple imputed. The imputation model (linear regression) was done separately for each treatment group and included baseline value as a covariate and was fitted to all patients with a measurement regardless of treatment adherence at week 104. The fitted model was used to impute values for patients without a measurement at week 104. Mean estimates were adjusted according to observed baseline distribution.

^jPresented descriptively only.

Sources: Clinical Study Report version 1.0 for the SELECT trial²⁵ and the Novo Nordisk Canada Inc. responses to the January 15, 2025,²⁶ and January 23, 2025,²⁷ requests by CDA-AMC for additional information for the CDA-AMC review of Wegovy. Details included in the table are from the sponsor’s summary of clinical evidence.²

Symptom Burden

The mean change from baseline in WRSSM total score at week 104 was −0.26 (SD = 0.71) in the semaglutide 2.4 mg group and −0.12 (SD = 0.68) in the placebo group. The between-group difference was not estimated.

Cardiometabolic Risk Factors

Efficacy results on cardiometabolic risk factors (change from baseline in systolic blood pressure, change from baseline in total cholesterol, and change from baseline in HDL cholesterol at week 104) were used to inform the accompanying pharmacoeconomic analysis and are presented in Table 29 in Appendix 1.

Harms

A summary of harms results based on the in-trial observation period is presented in Table 20.

Adverse Events

As described in the Clinical Study Report, nonserious AEs not fulfilling any of the prespecified criteria were not systematically collected. The focus of the safety evaluation per the Clinical Study Report for the SELECT trial was based on the reporting of SAEs and other systematically collected events (i.e., AEs of special interest).

Serious AEs

The proportion of patients with an SAE was 33.41% (2,941 of 8,803 patients) in the semaglutide 2.4 mg group and 36.40% (3,204 of 8,801 patients) in the placebo group. The most frequently reported SAEs (by preferred term) (frequency ≥ 2.0%) were:

• coronary arterial stent insertion, reported in 2.78% (245 patients) in the semaglutide 2.4 mg group and 3.17% (279 patients) in the placebo group

• acute MI, reported in 2.37% (209 patients) in the semaglutide 2.4 mg group and 3.33% (293 patients) in the placebo group

• unstable angina, reported in 2.07% (182 patients) in the semaglutide 2.4 mg group and 2.37% (209 patients) in the placebo group.

Withdrawals Due to AEs

The proportion of patients who permanently stopped treatment due to AEs was 16.60% (1,461 patients) in the semaglutide 2.4 mg group and 8.16% (718 patients) in the placebo group. The most frequently reported AEs (by preferred term) that led to permanent treatment discontinuation (frequency ≥ 2.0%) were:

• nausea, reported in 4.29% (378 patients) in the semaglutide 2.4 mg group and 0.40% (35 patients) in the placebo group

• diarrhea, reported in 2.35% (207 patients) in the semaglutide 2.4 mg group and 0.36% (32 patients) in the placebo group.

The proportion of patients who had their treatment interrupted or withdrawn due to AEs was 30.32% (2,669 patients) in the semaglutide 2.4 mg group and 16.00% (1,408 patients) in the placebo group. The most frequently reported AEs (by preferred term) that led to dose interruption or withdrawal (frequency ≥ 2.0%) were:

• nausea, reported in 8.07% (710 patients) in the semaglutide 2.4 mg group and 0.75% (66 patients) in the placebo group

• diarrhea, reported in 5.12% (451 patients) in the semaglutide 2.4 mg group and 0.75% (66 patients) in the placebo group

• vomiting, reported in 3.62% (319 patients) in the semaglutide 2.4 mg group and 0.20% (18 patients) in the placebo group

• constipation, reported in 2.20% (194 patients) in the semaglutide 2.4 mg group and 0.37% (33 patients) in the placebo group

• decreased appetite, reported in 2.15% (189 patients) in the semaglutide 2.4 mg group and 0.22% (19 patients) in the placebo group.

Mortality

There were 375 all-cause deaths (4.26%) in the semaglutide 2.4 mg group and 458 all-cause deaths (5.20%) in the placebo group. There were 371 investigator-reported SAEs with fatal outcome (4.21%) in the semaglutide 2.4 mg group and 460 investigator-reported SAEs with fatal outcome (5.23%) in the placebo group.

Notable Harms

The AEs of special interest identified for this review included CV AEs, cholelithiasis, nausea, vomiting, constipation, GERD, and pancreatitis. SAEs of cardiac disorders were reported in 11.45% (1,008 patients) in the semaglutide 2.4 mg group and 13.45% (1,184 patients) in the placebo group. All other AEs of special interest were reported in less than 2% of patients in each group.

Table 20: Summary of Safety Results From the In-Trial Observation Period in the SELECT Trial (FAS)

AE = adverse event; CV = cardiovascular; EAC = event adjudication committee; FAS = full analysis set; MedDRA = Medical Dictionary for Regulatory Activities; MI = myocardial infarction; NEC = not elsewhere classified; SAE = serious adverse event.

Note: The data cut-off date was July 18, 2023.

The preceding table was sorted in descending order by system organ class, high-level group term, and preferred term based on the proportion of patients with at least 1 event in either group. Only frequencies of ≥ 2.0% are summarized in the table, except for all-cause deaths and AEs of special interest.

^aCV AEs were collected systematically when serious. In addition, potential events of acute coronary syndrome, stroke, coronary artery revascularization, and heart failure for adjudication by the EAC were reported, irrespective of seriousness. A broad evaluation of CV safety, based on SAEs in the system organ class of cardiac disorders was performed. The evaluation of CV safety was supplemented by the assessment of blood pressure and heart rate.

^bAll AEs of gallbladder disease were collected systematically, irrespective of seriousness. Gallbladder-related disorders were evaluated based on a predefined MedDRA search on all AEs, supplemented with additional information collected from specific event forms for events of gallbladder disease.

^cGastrointestinal AEs were collected systematically when serious. Gastrointestinal safety was evaluated based on SAEs in the system organ class of gastrointestinal disorders.

^dAll AEs of pancreatitis were collected systematically, irrespective of seriousness. The evaluation of pancreatitis was based on a predefined MedDRA search for pancreatitis among all AEs, which provided a broad evaluation of all types of investigator-reported pancreatitis, including search terms indicative of acute or chronic pancreatitis; additional information was collected from a specific event form for pancreatitis, and the outcome of the adjudication by the EAC, the evaluation of which focused on events of acute pancreatitis.

^eA total of 69 patients (35 in the semaglutide 2.4 mg group and 34 in the placebo group) had a medical history of acute pancreatitis at screening. None of the patients with a history of acute pancreatitis had EAC-confirmed events of acute pancreatitis during the in-trial period.

Sources: Clinical Study Report version 1.0 for the SELECT trial.²⁵ Details included in the table are from the sponsor’s summary of clinical evidence.²

Critical Appraisal

Internal Validity

The SELECT trial had a group sequential design, a type of adaptive clinical trial design that plans for interim analyses at predetermined points during follow-up and as data are collected throughout the trial.²⁸ This trial design offers structured opportunities for making decisions earlier when the intervention shows a significant benefit (i.e., allows early stopping for efficacy, futility, or safety concerns). It can also improve efficiency in research by allowing adjustments to be made in the trial design, such as reducing the required sample size and duration, compared with traditional fixed-period RCTs. However, this trial design requires strong and appropriately conservative statistical plans because of the potential for unblinding, investigator influence, overestimation of treatment effect, and an increased chance of inflated type I error.^28-30 These points, combined with the need for a follow-up time that is sufficiently long to observe deaths and CV events, are especially important for evaluating the treatment effect of semaglutide 2.4 mg in preventing further CV morbidity and mortality. As will be described, the SELECT trial’s design, conduct, and analysis mitigate some of the potential limitations.

Randomization used appropriate methods (a 1:1 ratio using an interactive web response system) to achieve allocation and allocation concealment. Overall, the baseline characteristics were similar between groups, suggesting randomization successfully distributed potential confounders. In consultation with the clinical experts, the baseline characteristics did not reflect any notable imbalances in the known prognostic factors for CV events and treatment-effect modifiers. As such, the review team concluded that the methods used for randomization were appropriate and that the risk of bias arising from the randomization process was low.

The percentage of patients who discontinued the study was relatively small (approximately 3%) and similar between the groups. The reasons for discontinuing the study (patient withdrawal and lost to follow-up) were also similar between the groups and, therefore, the risk of attrition bias was likely low. The overall treatment discontinuation rates were similar between groups (approximately 30%). An imbalance in the discontinuation of treatment due to AEs was observed (16.3% in the semaglutide 2.4 mg group and 7.9% in the placebo group). This imbalance could have contributed to differential unblinding because those participants with treatment-emergent AEs — many of which were gastrointestinal in nature — might have inferred their assignment to semaglutide. Similarly, investigators might have formed expectations based on these observations, thereby introducing the potential for performance and reporting bias. While the observed difference between groups in treatment discontinuation due to lack of effect was relatively small, it was still differential (0.7% in the semaglutide 2.4 mg group and 2.8% in the placebo group) and, therefore, does not completely negate the potential impact on reporting and performance biases (i.e., relatively less concerning for biases). In consideration of the group sequential design, an additional key risk of bias is operational bias, in which the knowledge of treatment allocation potentially influences subsequent trial conduct, such as changes in how end points are assessed or reported near interim time points and recruitment patterns. However, an independent external EAC performed ongoing adjudication, based on a blinded review, of predefined event types according to definitions and guidelines outlined in its charter. In addition to reports by the investigator, events for adjudication were identified by the EAC and through standardized screening. Further, it was reported that the 2008 FDA guidance for evaluating CV safety in glucose-lowering therapies was adhered to for best practice on event adjudication and safety evaluation.⁵⁷ Unblinded data were reviewed exclusively in closed sessions, attended only by the independent external data-monitoring committee and the independent statistical analysts responsible for interim analyses, while blinded data were discussed in open sessions that included all relevant trial members, including sponsor representatives. Overall, the review team concluded the probability for introducing bias into the results was low.

Supplementary analyses of absolute risk difference were conducted at week 156 of the trial to assess the robustness of various end points. A total of 27.3% of patients who were randomized in the trial did not have a visit at week 156 due to trial closure; however, the distribution by treatment group was not reported. The trial was event driven and would not be completed until the target number of primary end point events was reached. While the general assumption for censored observations was that the risk of experiencing an event was not changed by censoring, the potential for bias in the estimated cumulative incidence at week 156 would arise only if those patients who were censored due to what was termed “trial closure” in the Clinical Study Report had a different risk of MACE compared with those who remained in the study. However, the definition of trial closure in the Clinical Study Report is not entirely clear. Administrative censoring included trial completion, lost to follow-up, and patient withdrawal. The implications of administrative censoring for the robustness of absolute risk difference in MACE occurrences are difficult to determine based on the available information. The review team noted that the between-group distribution in administrative censoring, including “trial closure,” was similar and, therefore, if bias exists, it is not likely an important factor in the cumulative incidence estimates. Further, a total of 79.4% of patients who were randomized in the trial did not have a visit at week 208 due to trial closure. The review team considered this large percentage of missing data to seriously undermine the validity and reliability of the percent change from baseline in body weight at week 208 reported by Ryan et al.²⁴ There is a high chance that missing information was due to informative censoring and, in the absence of additional analyses evaluating the missing data patterns, it is not possible to draw firm conclusions on the results from the analyses on body weight at week 208.

Several important site-level and patient-level protocol deviations in quality and good clinical practice were reported at 2 sites that resulted in serious breaches. Other serious breaches (electronic data capture system shutdown, flooding at 1 study site that resulted in the loss of trial data, potential external access to trial data due to hacking at 1 study site, and lack of principal investigator oversight at 1 study site) were concluded by the investigator to have a potential impact on data integrity (validity, reliability, and robustness of the trial data). Therefore, the review team considered these to be serious breaches potentially undermining the validity and reliability of the trial data, particularly when fraudulent data were entered for patient questionnaires (i.e., patient-reported outcomes) by a study nurse for missed visits to enable the questionnaire for the upcoming visit to become activated in the system. However, the overall impact on the trial data is likely low, given the small proportion of patients potentially impacted by these serious breaches.

Important potential limitations of the group sequential design include early stopping rules that may overestimate treatment effect and multiple interim statistical testing points that can introduce bias if the stopping threshold is met at a random peak in treatment effect. While the use of a single interim analysis in the SELECT trial reduces the risk of bias compared with multiple interim tests, the absence of a futility boundary means that the trial was not designed to stop early if semaglutide 2.4 mg showed little or no benefit versus placebo. However, given the trial continued to its final analysis without stopping for efficacy, and a reasonable approach to controlling for multiplicity was used (Lan-DeMets alpha spending function approximating O’Brien-Fleming boundaries) for the primary end point, the concern of treatment-effect inflation due to early termination does not apply.

For the primary end point, the final analysis was adjusted for the group sequential design by using likelihood ratio ordering. This method accounts for the adaptive nature of interim analyses by incorporating the sequential monitoring plan into the estimation of treatment effects, aiming to control the type I error and avoid overestimation of efficacy due to early stopping. In this case, the unadjusted HR was 0.80 (95% CI, 0.72 to 0.89) and the adjusted HR was 0.80 (95% CI, 0.72 to 0.90), indicating minimal impact of the adjustment. The similarity between the unadjusted and adjusted estimates suggested that the single interim analysis conducted in the trial had little influence on the final effect size, likely because the trial was not stopped early for efficacy. This additional information supports the robustness of the reported treatment effect in terms of it being unlikely that the estimated treatment effect was influenced by type I error potentially associated with the group sequential design.

The proportional hazards assumption for the primary end point, which the study investigators reported as being met, was evaluated graphically using Schoenfeld residuals and the standardized score process. However, the standardized score plot showed a strong downward slope from 0 in the first 12 months, returning to 0 at approximately month 21. Thereafter, it fluctuated near 0 until the end of the follow-up period. The early deviation suggested a violation of the proportional hazards assumption in the first 12 months, although the stabilization afterward indicated the issue was not persistent over the full duration of the study. The statistical analysis plan and Clinical Study Report for the SELECT trial did not describe any methods for assessing the potential impact of this violation on the results, such as adding time-dependent covariates to the Cox models or conducting sensitivity analyses with models that do not rely on the proportional hazards assumption. While visual inspection of the score process suggested a time-dependent effect on the proportional hazards, there was no clear evidence that this influenced the confirmatory results in a meaningful way.

The superiority hypothesis was tested for each confirmatory secondary end point under multiplicity control using a stagewise hierarchical testing scheme according to the prespecified order. The statistical testing strategy for the confirmatory secondary end points used a separate alpha spending function to control the type I error rate at a 1-sided level of 2.5%, which aligned with the P value adjustment for the group sequential design. Importantly, the end points — the composite HF outcome and all-cause death — could not be interpreted formally for superiority because the prespecified hierarchical testing procedure failed to reach statistical significance with the analysis of CV deaths, meaning they should be interpreted as exploratory rather than confirmatory.

The SELECT trial performed statistical comparisons for the nonconfirmatory secondary end points and some exploratory end points. However, the statistical comparisons were not included in the approach to adjust for multiple comparisons and, therefore, they increase the risk of type I error (false-positive findings). The absence of a prespecified multiplicity control strategy for these end points limits the strength of the conclusions that can be drawn from the observed differences between groups.

The number of events contributing to the subgroup analyses of patients with a BMI of 40 kg/m² and greater was considered few (< 50 events in each of the 2 treatment groups), and the sample sizes for these 2 subgroups were considered small (< 10% of patients randomized in each group), thereby lowering the certainty in the consistency of the treatment effect for the primary end point in these subgroups.

The main analysis of the time-to-event end points assumed independent censoring of patients who had withdrawn from the trial or were lost to follow-up, while deaths from causes not included in the end point were handled as censored observations but not part of the independent censoring assumption. Sensitivity analyses (which were planned to address the independent censoring assumption and assess the potential impact of missing data for patients who had withdrawn from the trial or were lost to follow-up) and a supplementary analysis (which was planned to assess the influence of the competing risk of death from causes not included in the end point) were performed for the confirmatory end points. The results of the sensitivity analyses were consistent with the main analyses; the review team judged that the risk of bias due to the handling of these intercurrent events and missing outcomes data for the confirmatory end points was low. Of note, the tipping-point sensitivity analysis was conducted only for the primary analysis and not for the key secondary confirmatory analyses. The review team had concerns about the 2 additional sensitivity analyses that were described as applying to patients who withdrew consent or were lost to follow-up (i.e., permanently discontinued the trial). Both analyses derived event rates from patients who stopped treatment but remained in the trial. As a result, the imputed events for those who permanently discontinued the study may not fully reflect their true risk. These patients might have different characteristics (e.g., worse health status, higher event risk), which could lead to the potential underestimation or overestimation of the true event rate. If the assumption underlying these sensitivity analyses — that off-treatment event rates from retained participants apply equally to those who permanently discontinued the study — does not hold, then the imputed results could introduce bias rather than correct for it. Nonetheless, these sensitivity analyses yielded results similar to those of the primary analysis (which was further supported by the tipping-point sensitivity analysis) and other confirmatory analyses, which suggests that the assumption of independent censoring was reasonable and that the trial’s conclusions are likely robust.

For all objectives, the primary estimand (intention to treat) was used to evaluate the treatment effect irrespective of adherence to treatment or changes to background medication. Although this approach to the handling of intercurrent events may be reflective of clinical practice, it can be a limitation for the interpretation of efficacy results, given that these intercurrent events also have an effect on CV risk. Further, the proportion of patients who discontinued treatment was relatively high (approximately 30%). A supplementary analysis based on the first on-treatment period, which was planned to address the secondary estimand (used to evaluate the effect in all patients who were randomized who had remained on their randomized treatment), was performed for the confirmatory end points. Because the results were consistent with the main analysis, the review team judged that the risk of bias due to treatment discontinuation was low for the confirmatory end points. The proportion of patients initiating a CV-related medication, an SGLT2 inhibitor, or metformin after randomization was, overall, slightly higher with placebo compared with semaglutide 2.4 mg. In consultation with the clinical experts, it was concluded there is a potential for bias against semaglutide 2.4 mg in the assessment of CV outcomes due to this slight imbalance in the prescribing of these medications.

Continuous supportive secondary end points at week 104 were analyzed, with multiple imputation used for missing values under a missing-at-random assumption, which was likely not plausible because there were notable imbalances in the reasons for treatment discontinuation between groups. A sensitivity analysis that was planned to address this assumption for missing data were performed for the change from baseline in body weight at week 104. Because the results were consistent with the main analysis, the review team judged that the risk of bias due to missing outcomes data for this end point was low. In contrast, no sensitivity analysis was planned for change from baseline in the EQ-5D, where missing outcome data were relatively high (approximately 16%) at week 104. As such, the review team concluded there is a potential for risk of bias due to missing outcomes data for these HRQoL end points.

In the protocol amendment dated January 4, 2021, the WRSSM assessment was removed from the supportive secondary end points and included in the exploratory end points because the measure was not sensitive to weight loss in a psychometric evaluation. According to the sponsor, the WRSSM tool was under development at the time of the trial and was used to gather preliminary insights as an exploratory outcome. In this context, no evidence of its measurement properties, including validity, in patients with established CV disease and obesity or overweight was submitted by the sponsor.

External Validity

Evidence from the SELECT trial addressed the evidence gap on the effects of semaglutide 2.4 mg on CV outcomes, which was raised in the previous submission for the weight management indication, but only for a subset of patients in the indicated population.

There was overlap between the SELECT trial population and the population criteria specified by the sponsor in its reimbursement request: patients with a BMI of 27 kg/m² or greater and established CV disease (MI, stroke, and/or PAD). The clinical experts consulted for this review agreed that the inclusion criteria captured the target population with established CV disease (including HF) that is seen in practice and is in need of an intervention for the secondary prevention of future CV events. Notably, most patients in the trial had prediabetes according to the hemoglobin A1C range defined in the American Diabetes Association Standards of Medical Care in Diabetes published in 2018 (i.e., 5.7% to < 6.5%).²³ However, it is important to note that the Diabetes Canada Clinical Practice Guidelines³¹ specify a slightly different hemoglobin A1C range (6.0% to < 6.5%) for diagnosing prediabetes. In consultation with the clinical experts, no major concern with the generalizability of the results due to this difference was identified.

Because treatment with semaglutide 2.4 mg in practice is not expected to be limited to patients who do not have diabetes, the review team found the trial’s exclusion of patients with a hemoglobin A1C level of 6.5% or greater to be a concern for the generalizability of the results to the target population. The investigator indicated that the trial population did not include patients with type 1 or type 2 diabetes, to remove any confounding that a diagnosis of diabetes could have on future CV risk. For more information on this evidence gap and how evidence from the SUSTAIN-6 trial addressed that gap, refer to the study addressing the gap in the systematic review evidence section in this document.

One exclusion criterion for the SELECT trial was patients who had experienced an MI, stroke, hospitalization for unstable angina pectoris, or TIA in the past 60 days before screening. In clinical practice, the clinical experts indicated that the typical treatment paradigm is to start drugs known to reduce the risk of future CV events at the time of the triggering CV event and continue these drugs long-term.

In consultation with the clinical experts, it was concluded that the comparison with placebo added to the standard of care for CV disease (antihypertensives, lipid-lowering drugs, anticoagulants, and ASA and other antiplatelet drugs) was appropriate, given that none of the current and accessible standard-of-care therapies to treat or prevent CV disease directly target weight loss. Regarding the standard of care for CV disease used in the trial, the clinical experts advised that, in practice, semaglutide would be combined with therapies that reduce cardiorenal risk, including combination treatment with ACE inhibitors, angiotensin 2 receptor blockers, statins, and SGLT2 inhibitors.

Long-Term Extension Study

The sponsor did not submit a long-term extension study for this review.

Indirect Evidence

The sponsor did not submit indirect evidence for this review.

Study Addressing the Gap in the Systematic Review Evidence

Contents within this section have been informed by materials submitted by the sponsor that addresses a gap in the systematic review evidence (Table 21). The following has been summarized and validated by the review team.

Table 21: Summary of Gap in the Systematic Review Evidence

BMI = body mass index; CV = cardiovascular; CVD = cardiovascular disease; T2D = type 2 diabetes.

Sources: SUSTAIN-6 post hoc subgroup analysis Clinical Study Report.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.²

Description of Study

SUSTAIN-6 was a long-term, multicentre, multinational, randomized, double-blind, parallel-group, controlled trial that evaluated the CV safety and long-term outcomes of semaglutide compared with placebo, when added to the standard of care, in patients with type 2 diabetes at high risk of CV events. A total of 3,297 patients were randomized (1:1:1:1) to receive a once-weekly subcutaneous injection of either semaglutide 0.5 mg (826 patients), semaglutide 1.0 mg (822 patients), placebo 0.5 mg (824 patients), or placebo 1.0 mg (825 patients). Refer to Table 22 for a summary of the SUSTAIN-6 trial.

From the SUSTAIN-6 trial data, a post hoc subgroup analysis was conducted that included patients with a BMI of 27 kg/m² or greater and established CV disease and type 2 diabetes. Post hoc subgroup analyses were conducted to determine whether the efficacy observed in this patient population was consistent with the SELECT trial and, therefore, the enrolment criteria for this subgroup replicated that of the SELECT study (i.e., limiting the subpopulation to patients with prior MI, stroke [excluding TIA], and/or PAD) and a BMI of ≥ 27 kg/m²). Additionally, although there were 2 maintenance doses used in the overall SUSTAIN-6 study (0.5 mg and 1.0 mg), only patients who had received 1.0 mg were included in the post hoc subgroup analyses because this dose was considered most comparable with that of the SELECT trial. However, the SUSTAIN-6 post hoc subgroup had still received a lower dose of semaglutide (1.0 mg) compared with participants in the SELECT trial (2.4 mg). Because the post hoc subgroup analyses were the most relevant for the purposes of this review, the results summarized in this section include only those pertaining to this subgroup.

Refer to Figure 2 for an overview of the details of the SUSTAIN-6 study.

Figure 2: Overview of the SUSTAIN-6 Trial Design

Sources: SUSTAIN-6 post hoc subgroup analysis Clinical Study Report.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.²

Table 22: Details of the SUSTAIN-6 Study

BMI = body mass index; CHD = coronary heart disease; CV = cardiovascular; HF = heart failure; GLP-1 RA = glucagon-like peptide-1 receptor agonist; MACE = major adverse cardiovascular events; MI = myocardial infarction; NYHA = New York Heart Association; T1D = type 1 diabetes; T2D = type 2 diabetes; TIA = transient ischemic attack.

^aBoth dosages in the SUSTAIN-6 study were lower than the recommended dose for the indication under review, aligning more closely with Ozempic than Wegovy.

Sources: SUSTAIN-6 post hoc subgroup analysis Clinical Study Report.⁵⁸ Details included in the table are from the sponsor’s summary of clinical evidence.²

Population

The post hoc analysis of patients with established CV disease and type 2 diabetes was designed to replicate the enrolment criteria of the SELECT trial, comparing semaglutide 1.0 mg versus placebo 1.0 mg. Patients in the post hoc subgroup analyses were aged 50 years or older with a BMI of 27 kg/m² or greater, type 2 diabetes, and clinical evidence of CV disease defined by prior MI, stroke (excluding TIA), and/or PAD. Consistent with the exclusion criteria for all SUSTAIN-6 participants, patients in the post hoc subgroup analyses did not have type 1 diabetes and had not used a GLP-1 RA within 90 days before screening.

Interventions

In the post hoc subgroup analyses, patients received either semaglutide 1.0 mg or volume-matched placebo. Although there were 2 maintenance doses in the overall SUSTAIN-6 study, 0.5 mg and 1.0 mg, only patients who had received the 1.0 mg dose were included in the subgroup analyses because this dose was considered by the sponsor as most comparable to that of the SELECT trial. The maintenance dose of 1.0 mg was reached after 4 doses (4 weeks) of 0.25 mg, followed by 4 doses (4 weeks) of 0.5 mg. After the maintenance dose was reached, the dose was not to be changed during the remainder of the trial. The total treatment duration for each patient was 104 weeks.

Outcomes

Refer to Table 22 for a summary of all outcomes included in the SUSTAIN-6 study, which were consistent with the SELECT trial’s CV-related end points. The primary end point was time from randomization to first occurrence of MACE, defined as CV death, nonfatal MI, or nonfatal stroke. The secondary end points identified by the clinical experts as most relevant to this review were time from randomization to first occurrence of all-cause death, nonfatal MI, or nonfatal stroke.

While the expanded composite CV end points (defined as either MACE, revascularization [coronary or peripheral], unstable angina requiring hospitalization, or hospitalization for HF) were also included as secondary outcomes in the SUSTAIN-6 trial, the clinical experts indicated that the 3-component MACE end points were most important to this review because these are the typical CV outcome benchmarks used in clinical trials for CV outcomes and the most clinically important for patients.

Statistical Analysis

An unplanned post hoc analysis of patients with established CV disease and type 2 diabetes replicated the enrolment criteria of the SELECT trial, comparing semaglutide 1.0 mg versus placebo 1.0 mg. Efficacy analyses were conducted using a stratified Cox proportional hazards model.

Results

Patient Disposition

In the post hoc subgroup, approximately ███ of patients in the semaglutide 1.0 mg group and ███ of patients in the placebo group discontinued treatment prematurely. The sponsor did not provide the reasons for discontinuation in this subgroup, but in the full SUSTAIN-6 study population, discontinuation was largely due to gastrointestinal tolerability issues in the semaglutide group (0.5 mg, ████; 1.0 mg, ████) compared with placebo (0.5 mg, ████; 1.0 mg, ████). Treatment discontinuations were reported for 2 different study time frames with different numbers of total participants (N): from 0 to 6 months (patients who were randomized) and from 6 to 24 months (patients who remained and who did not discontinue). In the post hoc subgroup, the highest proportion of premature discontinuations among patients in the semaglutide group took place at between 0 to 6 months (█████ of patients who were randomized) compared with 6 to 24 months (███ of the patients who remained), while placebo-group discontinuations were greatest between 6 and 24 months (█████ of the patients who remained) compared with 0 to 6 months (████ of patients who were randomized) (refer to Table 23 for details). In the larger SUSTAIN-6 study population, common reasons for premature discontinuation in the placebo group were trial fatigue, suspicion of placebo, and introduction of disallowed medication.

Table 23: Summary of Treatment Discontinuations in the SUSTAIN-6 Trial Post Hoc Subgroup