CMR-guided defibrillator implantation did not beat monitoring in mildly reduced LVEF

A scar-based strategy for implantable cardioverter-defibrillator placement failed to deliver a clear advantage in patients with left ventricular ejection fraction between 36% and 50%, even when late gadolinium enhancement on cardiovascular magnetic resonance identified myocardial scar. In the CMR GUIDE randomized clinical trial, the composite outcome of sudden cardiac death or hemodynamically significant ventricular arrhythmia occurred at similar rates in the ICD and implantable loop recorder groups, despite a biologically plausible rationale for treating scar as an arrhythmic substrate.

The result sits in a difficult clinical zone. Current guideline thresholds for primary prevention ICD therapy largely stop at an LVEF of 35% or less, yet a substantial share of sudden cardiac deaths occur in the adjacent range of mildly to moderately reduced systolic function. This trial asked a direct question: if scar can be seen, and if scar is where malignant ventricular rhythms often begin, can imaging identify a group that benefits from prophylactic defibrillator therapy before the ejection fraction falls below the conventional cutoff?

Trial design and the population actually studied

The study enrolled adults with ischemic or nonischemic cardiomyopathy, LVEF of 36% to 50%, documented myocardial scar on CMR, and background guideline-directed medical therapy. Recruitment took place at 18 sites in Australia, Germany, and the UK between 2015 and 2022, with follow-up completed in 2026. A total of 353 patients were randomized: 180 to ICD implantation and 173 to implantable loop recorder monitoring.

The cohort was predominantly older men. Median age was 65 years, 18% were female, and 72% had ischemic cardiomyopathy. Seventy percent had an LVEF of 40% or greater, which matters clinically because this was not a population hovering just below the classic ICD threshold; many participants had function that would ordinarily place them outside standard device eligibility. The median follow-up of 6.3 years provided enough time for arrhythmic and competing nonarrhythmic outcomes to emerge, which makes the neutral composite result harder to dismiss as a short-duration artifact.

The composite endpoint stayed neutral while the components diverged

The primary composite outcome occurred in 14 patients in the ICD group and 16 patients in the loop recorder group, corresponding to 7.8% versus 9.2% and a hazard ratio of 0.76. The confidence interval crossed unity widely, leaving no statistically secure evidence that the device strategy reduced the composite endpoint.

One sentence captures the tension in the data: the trial did not show a composite benefit, even though sudden cardiac death alone was lower in the ICD arm.

Sudden cardiac death occurred in 3 patients assigned to ICDs and 10 patients assigned to monitoring alone, a difference that favored the device strategy and yielded a hazard ratio of 0.26. By contrast, hemodynamically significant ventricular arrhythmia occurred more often in the ICD arm, 12 patients versus 6 patients, with a hazard ratio of 1.77. The imbalance in these components is not easily read as a simple win or loss for the defibrillator. ICDs can terminate malignant rhythms, but they also record and treat arrhythmias that may never have reached a fatal endpoint, while loop recorders detect events that may reflect a different burden of electrical instability or differing thresholds for adjudication.

Why scar did not translate into a broader treatment signal

Myocardial scar has a strong mechanistic link to ventricular reentry. Fibrotic tissue creates conduction heterogeneity, slow pathways, and electrical discontinuity, all of which can support sustained ventricular arrhythmias. On first principles, that biology favors a scar-guided device strategy in patients whose ejection fraction is only modestly reduced. Yet biology and outcome data do not always align once competing risks enter the frame.

In this cohort, all-cause mortality, cardiovascular mortality, and heart failure-related hospitalization were similar in both groups. Those neutral secondary outcomes suggest that many patients were more likely to experience nonarrhythmic progression, comorbidity burden, or death from causes not preventable by a defibrillator than a sudden rhythm event that the device could avert. The finding is especially relevant in older patients, where competing mortality can dilute any incremental advantage of arrhythmic protection.

The subgroup analysis points in the same direction. Among participants younger than 70 years, the primary outcome occurred in 3.3% of the ICD group versus 10.0% of the monitoring group, with a hazard ratio of 0.28. In those 70 years or older, the event rate was 16.9% with ICD therapy and 7.5% with monitoring, yielding a hazard ratio of 2.33, and the interaction was statistically significant. Age is an imperfect surrogate, but in practice it often reflects substrate complexity, competing noncardiac death, frailty, and differential tolerance of device therapy. The younger subgroup may still harbor a window where arrhythmic prevention carries value, while older patients may carry enough nonarrhythmic risk that the defibrillator cannot move the overall needle.

What this means for imaging-guided device selection

The trial does not support routine ICD implantation for every patient with LVEF of 36% to 50% and CMR-defined scar. A scar signal alone was not enough to produce a statistically persuasive reduction in the composite endpoint, and the absence of benefit on mortality and heart failure hospitalization weakens the case for broad expansion of device therapy beyond current guideline thresholds.

The results also caution against treating imaging markers as stand-alone surrogates for clinical benefit. Late gadolinium enhancement marks arrhythmic substrate, but arrhythmic substrate is only one piece of the sudden death equation. Event trigger, scar burden, autonomic tone, ischemic burden, electrolyte instability, and the competing risk of pump failure all shape whether a patient will actually benefit from prophylactic defibrillation.

For bedside decision-making, the study supports restraint rather than escalation. Patients with mildly to moderately reduced LVEF and scar should not automatically be routed to ICD implantation on imaging alone. The more difficult question now is how to refine risk selection within this intermediate ejection fraction range without over-treating older patients or missing the subgroup most likely to benefit.

Trial registration and interpretive limits

The study was registered at ClinicalTrials.gov under NCT01918215. Its open-label design reflects the practical realities of device trials, but the long follow-up and randomized structure still provide an important test of a clinically attractive hypothesis.

The most consequential limitation is not statistical noise alone; it is the mismatch between a compelling biologic marker and a clinical endpoint that refused to separate cleanly. ICD therapy lowered sudden cardiac death, yet the composite outcome remained neutral because hemodynamically significant ventricular arrhythmia did not move in the same direction and overall mortality stayed unchanged. Future work will need to define whether scar burden, age, ischemic versus nonischemic etiology, or another imaging and clinical combination can identify a narrower group in whom the benefit becomes large enough to justify device implantation.


Reporting by Dr. Aishwarya N., MBBS for Medicine World Foundation, with additional data provided by JAMA (Journal of the American Medical Association).
PRIMARY CLINICAL EVIDENCE & REGISTRY DATA