AF Ablation in End-Stage HF: Is CASTLE-HTx Too Good to Be True?

John M. Mandrola, MD 


August 30, 2023

Most patients with atrial fibrillation (AF) can be treated slowly with high doses of reassurance and education. There is, however, a small group of higher-risk patients who have both AF and heart failure that require prompt aggressive action.

The CASTLE HTx trial, presented at the European Society of Cardiology congress, studied catheter ablation of AF vs medical therapy in these sick patients with advanced heart failure. The results were stunningly positive for ablation.

The CASTLE-HTx Trial

CASTLE HTx is single-center trial done at a major referral center for heart transplantation in Bad Oeynhausen, Germany. The investigators screened just over 900 patients who were referred for assessment of eligibility for heart transplant or left ventricular assist device (LVAD) who also had symptomatic AF. About 20% (n = 194) of these patients were randomly assigned to either primary AF ablation or continued medical therapy.

The average age of these mostly male patients was 62 and 65 years, in the ablation vs medical therapy groups, respectively. The average left ventricular ejection fraction was 25% in the medical therapy group and 29% in the ablation arm. One third had New York Heart Association (NYHA) functional Class II heart failure while two thirds had Class III-IV heart failure. Baseline medical therapy was excellent in both groups.

The primary endpoint was a composite of death from any cause, implantation of an LVAD, or urgent heart transplantation.

Two experienced electrophysiologists performed the ablation procedures, 68% of which were pulmonary vein isolation (PVI) alone.

The CASTLE HTx Results

Of the 97 patients randomly assigned to ablation, 16 did not undergo the procedure. Of the 97 patients in the medical therapy arm, 16 crossed over to receive ablation.

Investigators planned for a 3-year follow up, but after 1 year, the data safety monitoring board recommended termination of the trial because of substantial benefits in the ablation arm.

After a median follow-up of 18 months, a primary endpoint occurred in 8% of patients in the ablation arm vs 30% in the medical therapy arm (hazard ratio, 0.24; 95% CI, 0.11 - 0.52; P <.001). There were no primary endpoint events in any of the 16 patients in the ablation group that did not undergo ablation.

Death from any cause occurred in 6% of the ablation group vs 20% of the medical therapy arm, (HR, 0.29; 95% CI, 0.12 - 0.72).

Additional endpoints favored ablation over medical therapy, including improvement of LV function, reduced AF burden, and less use of amiodarone. Procedure-related complications were minimal and included only groin access issues.


The first thing to note is that of the approximately 900 patients screened, just under 200 were enrolled. CASTLE HTx therefore involves a highly select group of patients with heart failure and AF.

Now to the hard part.

The effect size in the reduction of seriously bad outcomes, including death, is huge. There was much discussion at ESC and online about how to explain both the size of the benefit and the very early separation of the Kaplan-Meier survival curves.

Heart failure trialist Milton Packer, MD, of Baylor University Medical Center, wrote via email that an effect size of a near 80% reduction in major events, based on fewer than 30 events, is not a reproducible finding.

Cardiologist Sanjay Kaul, MD, of Cedars-Sinai Medical Center, an expert appraiser of medical evidence, agreed with Packer. Kaul also noted that the 16% crossover rate in the medical therapy arm dilutes the effect size, which would have been even higher in an on-treatment analysis. “This borders on incredulous,” he wrote.

I'd add another curious finding regarding the early separation of event curves. It is very different from CASTLE-AF, another trial comparing ablation vs medical therapy in patients with heart failure. In CASTLE-AF, the mortality benefit from AF ablation was not seen for 3 years — which, the authors explained, was the time it took for the reduction of AF to exert its effects.

On the early massive benefit, first author of CASTLE-HTx Christian Sohns, MD, explained by email that they anticipated that endpoints as well as treatment effects would occur early because these were severely sick patients. Some patients were prioritized in terms of time point of ablation due to severe symptoms. 

The ESC trial discussant, heart failure specialist Finn Gustafsson, from Copenhagen University Hospital, made the case that patients in the trial were perhaps not truly "end-stage." He noted that one third of them had NYHA Class II symptoms (mild impairment of function).

Packer concurred, noting in his email to me that 20% mortality at 18 months in the control arm of CASTLE HTx is far better than what is typically seen for end-stage heart failure. I went back to look at patient characteristics in the two CASTLE trials. The mean LVEF in CASTLE AF was 32%; in CASTLE HTx it was about 27%. And there is this quote from the CASTLE HTx New England Journal of Medicine paper: "at the time of randomization, all the patients had to be in a clinically stable condition."

The distinction regarding the degree of illness of the patients is important because it is the cited reason for the large effect size.

Trial discussant Gustafsson also noted other weaknesses of CASTLE-HTx: it was a single center, unblinded study with small sample sizes. And Sohns' comment that "some patients were prioritized in terms of time point of ablation," at least hints at the possibility of a non-nefarious performance bias among ablation proponents.

Curious, too, is that 16 patients in the ablation arm who did not have ablation also did not experience a primary outcome event. Herein lies the problem with small sample sizes: it's hard to know if this is noise or signal.

So far, I've been critical. Now let me add content expertise. I've seen these patients. A patient with established cardiomyopathy who develops AF can deteriorate rapidly. AF often recurs after cardioversion despite antiarrhythmic drugs. Permanent restoration of sinus rhythm with ablation can restore stability. If such a patient is not referred to an electrophysiologist, a bad outcome may follow. I add this anecdote because it underscores the highly selective nature of patients in the CASTLE trials; these results should not be translated to all patients with AF and heart failure.

Despite the limitations of CASTLE-HTx, my conclusion is that this evidence strongly suggests that electrophysiologists have a lot to offer the heart failure community.

When a patient with cardiomyopathy and advanced heart failure develops symptomatic AF, the evidence supports getting an electrophysiologist on the case. In selected patients, AF ablation may be the better course than continued medical therapy.

John Mandrola practices cardiac electrophysiology in Louisville, Kentucky, and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence.

Follow John Mandrola on Twitter

Follow | Medscape Cardiology on X (formerly Twitter)


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.