Most sudden cardiac deaths occur in people whose ejection fraction is above the 35% threshold that triggers a primary prevention defibrillator. CMR GUIDE tested whether late gadolinium enhancement could identify who among them to implant. It randomised 353 patients with ischaemic or non-ischaemic cardiomyopathy, an ejection fraction of 36% to 50%, myocardial scar on cardiac MRI and guideline-directed medical therapy, to a defibrillator or to an implantable loop recorder. Median follow-up was 6.3 years.
The primary composite of sudden cardiac death or haemodynamically significant ventricular arrhythmia occurred in 14 patients (7.8%) with a defibrillator and 16 (9.2%) with a loop recorder, hazard ratio 0.76 (95% CI 0.37 to 1.58). The trial missed.
The components pull in opposite directions, which is the interesting part. Sudden cardiac death alone fell — 3 patients versus 10, hazard ratio 0.26 (95% CI 0.07 to 0.95) — while haemodynamically significant ventricular arrhythmia was numerically higher in the defibrillator arm, 12 versus 6. That is an artefact of ascertainment as much as biology: a device that treats an arrhythmia also detects and records it, and a loop recorder patient may die suddenly without the arrhythmia being counted. A composite that combines a fatal event with the event that causes it will behave oddly when one arm prevents the death.
A prespecified subgroup analysis found benefit under age 70 (3.3% versus 10.0%, HR 0.28, 95% CI 0.09 to 0.89) and none over 70 (16.9% versus 7.5%, HR 2.33, interaction p=0.01). That is hypothesis-generating and no more. The practical answer today is that the 35% threshold stands, and scar imaging does not yet move it.
- Do not implant a primary prevention defibrillator on the basis of cardiac MRI scar alone at an ejection fraction above 35%
- Continue to optimise guideline-directed medical therapy and reassess ejection fraction before considering a device
- Read the composite endpoint carefully: sudden cardiac death fell, but arrhythmia detection is not blind to having a device
- Treat the age interaction as a signal for future trials, not as a selection rule for practice
- Cardiac MRI remains valuable for aetiology and prognosis — this trial questions its use as a device-selection tool, not the imaging itself
The statistics, in plain English
A hazard ratio of 0.76 with a confidence interval from 0.37 to 1.58 spans a 63% reduction to a 58% increase — the trial is compatible with meaningful benefit and with meaningful harm, which is what an underpowered result looks like. Only 30 primary events occurred in total across 353 patients over 6.3 years, so this is a question of sample size rather than a demonstration of no effect. The subgroup interaction, p=0.01 across six prespecified subgroups, is exactly the kind of finding that fails to replicate: with six comparisons, one significant interaction is close to what chance produces.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for cardiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free