Known risk factors explain little of semaglutide's cardiovascular benefit in SELECT

Design

Post hoc mediation analysis of the SELECT randomised trial using a counterfactual regression method

Population

Adults with pre-existing cardiovascular disease and BMI of 27 or more, without diabetes

Primary outcome

Proportion of the reduction in major adverse cardiovascular events mediated by changes over 24 months in anthropometric, glycaemic, lipid, blood pressure, inflammatory and kidney measures

Effect

Waist circumference 64.0% (27.5 to 179.6), C-reactive protein 42.1% (17.9 to 110.5), all variables combined 31.4% (-30.1 to 143.6)

SELECT randomised adults with established cardiovascular disease and a BMI of 27 or more, without diabetes, to semaglutide 2.4 mg weekly or placebo, and semaglutide lowered major adverse cardiovascular events by 20%. This analysis asked which measured changes could explain that benefit.

All the candidates improved with semaglutide, including weight, waist circumference, C-reactive protein, HbA1c, lipids, blood pressure and kidney measures. Point estimates of mediation were largest for waist circumference (64%), C-reactive protein (42%), HbA1c (29%) and body weight (19.5%), but every confidence interval was wide and several crossed zero. Weight and waist estimates were judged unreliable. Combined, the variables explained 31.4% (95% CI -30.1 to 143.6).

The fair reading is that the mechanism is not pinned down. The finding does not suggest that benefit depends on weight loss, and it does not rule it out.

• Do not choose between agents on the basis of weight loss alone when the goal is cardiovascular protection
• Keep treating blood pressure and lipids separately; the trial did not show these drugs replace them
• Expect the cardiovascular benefit to apply to the SELECT population: established disease, overweight or obesity, no diabetes
• Treat C-reactive protein as an exploratory marker, not a treatment target

Why it matters

It questions the common view that the benefit of these drugs is mostly a weight effect.

Don't overread it

This is an exploratory mediation analysis; wide intervals mean it cannot show that any one pathway does or does not matter.

The statistics, in plain English

Percentage mediation estimates how much of a treatment effect travels through a given variable. Confidence intervals that run from below zero to above 100% mean the data cannot say, which is the case here for nearly all variables. Mediation analysis also relies on assumptions that randomisation does not protect.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.