A detailed analysis of the SELECT trial examined how much of the reduction in major heart events could be attributed to changes in weight, inflammation, blood glucose, and other cardiovascular risk factors.
Study: Semaglutide and cardiovascular risk reduction: a mediation analysis of the SELECT trial. Image Credit: Edugrafo / Shutterstock
In a recent study published in the European Heart Journal, researchers investigated which measured cardiovascular risk factors might help explain semaglutide's cardioprotective effects.
The study comprised a pre-specified mediation analysis of the landmark SELECT trial, which evaluated 17,604 adults with overweight or obesity and established cardiovascular disease (CVD) without diabetes. Its primary objective was to estimate how much of the drug’s benefit could be explained by improvements in traditional risk factors.
The primary SELECT trial had previously shown that weekly semaglutide (target dose of 2.4 milligrams [mg]) significantly reduced major adverse cardiovascular events (MACE) by 20% compared with placebo.
In the multivariable model, conventional risk factors such as body weight, blood pressure, cholesterol, and blood glucose produced a joint mediation point estimate of 31.4%. Based on this estimate, about 68.6% of the treatment effect was not accounted for by the included measures, although the statistical uncertainty was too large to regard this proportion as precise.
The study did not establish which additional mechanisms accounted for the remaining effect, but the findings suggest that other unidentified pathways may contribute to semaglutide's cardiovascular protection. Proposed possibilities include vascular and anti-inflammatory effects, which the authors describe as speculative.
Background
Cardiovascular diseases (CVDs) remain a major cause of global mortality.
Obesity, commonly defined in adults as a body mass index (BMI) ≥ 30 kg/m2, is associated with increased cardiovascular risk, including atherosclerotic disease and stroke. Excess body weight is an important modifiable contributor to cardiovascular risk.
The development and introduction of glucagon-like peptide-1 receptor agonists (GLP-1 RAs) have expanded pharmacological options for weight management, with these drugs producing substantial weight loss in clinical trials.
GLP-1 RAs have already demonstrated substantial clinical efficacy in helping people with overweight or obesity achieve sustained weight reductions during treatment, while cardiovascular outcome trials have also reported benefits beyond weight loss alone.
For example, the primary SELECT trial demonstrated that once-weekly subcutaneous semaglutide (an approved GLP-1 RA) lowered MACE incidence (defined as “cardiovascular death, non-fatal myocardial infarction [MI], or non-fatal stroke”) from 8.0% (placebo) to 6.5% with semaglutide.
The original SELECT trial findings did not establish how much of this cardiovascular benefit could be explained by weight loss or changes in other measured risk factors, or whether additional pathways contributed.
About the study
The present study aimed to address this knowledge gap and inform future cardiovascular interventions by applying counterfactual mediation modeling, a statistical method that estimates how much of a treatment effect is explained by changes in specific measured factors, to the SELECT cohort dataset.
The study sample cohort comprised 17,604 participants aged 45 years or older with a BMI ≥ 27 kg/m2 and clinically validated pre-existing CVD. Participants with diabetes were excluded from the analyses, and the included participants were followed for a mean of 39.8 months.
The study’s analysis evaluated 12 candidate mediators, with repeated measurements used through at least 24 months and mediation estimated at 36 months. Measures included body weight, waist circumference, high-sensitivity C-reactive protein (hsCRP), glycated hemoglobin (HbA1c), blood pressure (SBP/DBP), lipid fractions (HDL, LDL, triglycerides), estimated glomerular filtration rate (eGFR), and urinary albumin-to-creatinine ratio (UACR).
The study’s statistical analyses primarily consisted of mediation analyses, quantified using the Vansteelandt repeated regression counterfactual method to assess individual and multivariable mediation, alongside time-dependent Cox proportional hazards models.
Study findings
Assignment to semaglutide (weekly target dose of 2.4 milligrams [mg]) was associated with significant improvements across all 12 candidate measures.
The individual counterfactual models produced the largest estimated mediation for change in waist circumference (64.0%), followed by hsCRP (42.1%), HbA1c (29.0%), and body weight (19.5%). However, substantial statistical uncertainty limited confidence in the size of these effects.
Triglycerides (18.6%), UACR (14.6%), SBP (14.3%), and LDL cholesterol (10.9%) also produced mediation point estimates. In the multivariable counterfactual model, selected measures produced a joint mediation estimate of 31.4%. This left about 68.6% of the treatment effect unaccounted for in the point estimate, although the uncertainty around the analysis prevented a precise division between explained and unexplained benefit.
Finally, sensitivity analyses raised the possibility that estimates involving body weight and waist circumference were affected by unintentional weight loss related to comorbid illness or frailty, particularly in placebo participants. The authors also noted that several mediation estimates were imprecise.
Conclusions
The present study provides the first mediation analysis of a GLP-1 receptor agonist cardiovascular outcome trial in people with established CVD but without diabetes at baseline. The analysis could not establish with certainty that changes in body weight, waist circumference, or the other measured cardiovascular risk factors fully accounted for semaglutide’s reduction in MACE.
These findings leave open the possibility that additional unmeasured biological pathways contribute to semaglutide's cardiovascular protection. Potential mechanisms discussed by the authors include effects on inflammation, atherosclerotic plaque biology, steatohepatitis, epicardial fat, and circulating proteins, but these explanations remain speculative.
The study was limited by uncertainty in the mediation estimates, assumptions about unmeasured confounding in mediation analyses, missing mediator data, and the frequency of measurements. The findings support further research into biological pathways beyond the measured risk factors to clarify how semaglutide reduces cardiovascular events.
As GLP-1 drugs become more complex, their effects on the heart are getting harder to predict