John Mandrola, MD


August 29, 2017

Few problems in cardiology are more common and vexing than the combination of heart failure and atrial fibrillation. The relationship goes both ways. Primary heart failure can induce atrial structural changes that cause AF. And AF, with its loss of AV synchrony and rapid ventricular rates, can cause or worsen heart failure.

Treatment of these patients is tough. The reflex is to get the patient out of AF—in other words, use rhythm control. Before the European Society of Cardiology (ESC) 2017 Congress, there was no evidence that this intuitive strategy improved outcomes. In fact, the use of antiarrthymic drugs in this scenario does not reduce stroke or death but does increase hospitalizations.[1]

Ablation of AF is another (possibly better) rhythm-control strategy. Multiple studies have shown that ablation in patients with heart failure and AF reduces AF burden, improves quality of life, and increases ejection fraction when compared with a rate-control strategy.[2] And a small trial of AF ablation vs amiodarone in patients with heart failure and AF hinted at a mortality reduction (8 vs 18%; P=0.037), albeit as a secondary end point.[3]

The Study

The multicenter CASTLE-AF trial,[4] presented in the first hotline session of the ESC meeting, set out to definitively test the ability of AF ablation to improve hard outcomes in patients with systolic heart failure and AF. Definitive because the investigators chose a composite end point of all-cause mortality or heart-failure admissions.

The trial took place in 31 sites in nine countries. Investigators screened slightly more than 3000 and ended up with 363 patients randomized to either an AF ablation arm (n=179) or conventional guideline-directed therapy arm (n=184).

Dr Nassir Marrouche (Comprehensive Arrhythmia Research and Management [CARMA] Center, University of Utah Hospital, Salt Lake City) delivered the strikingly positive results. AF ablation resulted in a 38% reduction in the primary end point (HR 0.62; 95% CI 0.43–0.87). Lower rates of all-cause mortality (HR 0.53; 95% CI 0.32–0.86) and heart-failure admissions (HR 0.56; 95% CI 0.37–0.83) drove the primary end point. The Kaplan-Meier curves for all-cause mortality took 3 years to separate, while the curves for heart-failure admissions separated at 6 months.

Cardiovascular death and cardiovascular hospitalization were also lower in the ablation arm. And similar to prior studies, ablation resulted in lower AF burden (patients had ICDs, so this finding is quite strong) and improved ejection fraction. Subgroup analyses suggested patients with class 3-4 heart failure did not garner benefit from ablation.

Response From the EP Community

The response from the ablation community has been celebratory. Discussant Carina Blomstrom-Lundqvist (Uppsala University, Sweden) called the trial and its result relevant and novel.

By email, Dr Hugh Calkins (Johns Hopkins, Baltimore, MD), who is lead author of the recent expert consensus statement on AF ablation, was impressed and summarized the positive response to CASTLE-AF:

"The results of this study were predictable given the huge number of prior CHF studies, all of which are cited in the consensus document. . . . The strengths of the study include the fact that AF burden was tracked with the device."

Calkins also noted that most patients were class 2 and that amiodarone was used in a considerable number of patients in both arms. "But [the results of CASTLE-AF] are consistent with my experience. I think AF ablation really works—and improves hard end points. So add me to the list of enthusiasts. But we need to see the details in the paper," he added.

My Thoughts

The study and its results are impressive, and I, like Dr Calkins, have patients similar to those enrolled in CASTLE-AF who have done beautifully after ablation. This is the first adequately powered study to show mortality reductions with AF ablation. That is a big deal.

But I have concerns about clinical translation of this study—which is not yet published.

It took screening of 3000 patients to get 363 into the trial. Between CASTLE-AF–induced enthusiasm and pay-for-procedure reimbursement model in the US, choosing patients who are not appropriate for ablation will require keen judgment and stewardship.

For instance, CASTLE-AF enrolled relatively young (average age 64 years) symptomatic patients, most of whom (70%) had class 1-2 heart failure. The results do not apply to those with diastolic heart failure and probably do not apply to patients who are asymptomatic, elderly, or have advanced heart failure.

Another concern is plausibility. An all-cause mortality reduction of 47% far exceeds any proven heart-failure therapy. Relative mortality reductions with ACE inhibitors, ARBs, angiotensin receptor-neprilysin inhibitors, mineralocorticoid receptor antagonists, beta-blockers, CRT-Ps, ICDs, and CRT-Ds range from 15% to 35%.[5] More than two-thirds of patients in CASTLE-AF had persistent AF, and we know ablation results in these patients are poor. How can a procedure with modest success rates, serious complications, and partial elimination of AF translate to this much success?

A higher 5-year mortality in the comparator arm of CASTLE-AF may explain the dramatic results. If you plug a typical patient enrolled in CASTLE-AF, a 64-year-old man with class 2-3 heart failure, who is on ACE inhibitors and beta-blockers, into the Seattle Heart Failure Model—which does not include AF—you get a 5-year predicted mortality of 24%. But a review of the Kaplan-Meier curves from slides presented during the session showed a 35% to 40% estimated probability of 5-year survival in the conventional arm. Did ablation do better because of the procedure or because the conventional arm did worse? Of note, one in three patients in the conventional arm received antiarrthymic drugs—which have no proven benefit and likely harm. (Surely, many in the ablation arm came off amiodarone.)

Another concern is procedural complications. I tallied 19 significant complications from the procedure (10%). Given that CASTLE-AF enrolled patients with heart failure, a 10% complication rate—even from centers of excellence—is not surprising. The problem, which is inherent to all ablation trials, is whether the results can be reached in the real world where most ablation is done by low-volume operators. I doubt it.

Finally, with all unblinded interventional trials, it's vital to consider how the investigators dealt with performance bias, reporting bias, and allocation concealment. For this, we will need to see the full paper.

Stay tuned for more commentary on this important study when it is published.

Editor's note: An earlier version of this column incorrectly identified the 5-year mortality in the conventional arm as 60%. The error was due to the authors' extrapolation of the right side of a Kaplan-Meier plot. The image of the slide had been distorted during conversion to a PDF file.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.