CABANA: Initial Thoughts on AF Ablation vs Drugs

John M. Mandrola, MD


May 14, 2018

The Catheter Ablation versus Antiarrhythmic Drug Therapy for Atrial Fibrillation (CABANA) Trial[1] dominated the Heart Rhythm Society (HRS) 2018 Scientific Sessions.

Ablation of atrial fibrillation (AF) is big business. Financial analysts estimate the AF ablation market has grown to about $4.5 billion in revenue. Exuberance for AF ablation has persisted despite little evidence the invasive procedure improves hard outcomes, such as mortality or stroke.[2]

The CABANA investigators asked a simple question: In symptomatic patients with risk factors for stroke, is AF ablation superior to rate or rhythm control medications for the reduction of major outcomes?

The prevailing positive views of ablation made CABANA a tough sell to symptomatic patients, many of whom were referred to expert centers for ablation. At HRS, numerous doctors who participated in CABANA's 120 centers worldwide told me it was hard to convince symptomatic patients to join the study.

Slow enrollment and lower than expected event rates prompted a monitoring committee, along with investigators, to expand the primary endpoint from overall mortality to a composite of overall mortality, disabling stroke, major bleeding, and cardiac arrest.

The Study

Dr Douglas Packer from the Mayo Clinic in Rochester, Minnesota, presented CABANA results to a packed ballroom and overflow areas.

The trial randomly assigned about 1100 patients (median age, 67 years) to each group. More than half of patients had persistent or long-standing persistent AF, and 37% were women. Baseline characteristics appeared equal in each group.

The most discussed issue concerned the high numbers of patients who did not receive the treatment they were assigned. In the ablation arm, 102 patients (9.2%) did not have an ablation. Packer said most of these were due to patient choice. In the drug arm, 301 patients (27.5%) crossed over and received ablation.

After 5 years of follow-up, there were no significant differences in the primary endpoint (hazard ratio [HR], 0.86; 95% CI, 0.65 - 1.15; P = .3). None of the components of the primary endpoint differed significantly.

These results were presented in the standard intention-to-treat (ITT) manner, which means patients who crossed over were still counted in their original group.

The controversial part of the presentation was when Packer showed the results based on treatment received. Given the large number of crossovers, 1307 patients received ablation and 897 received drugs. In this as-treated analysis, which was prespecified as a sensitivity analysis of the primary results, ablation resulted in a 3.9% absolute risk reduction in the primary endpoint (P = .006) and a 3.1% reduction in all-cause death (P = .005). 

In a similar sort of analysis, called per-protocol, ablation reduced the rate of the primary endpoint by 27% (HR, 0.73; 95% CI, 0.54 - 0.99; P = .046). 

Adverse events occurred in both groups. Nearly 4% of the ablation group had problems from catheter insertion; 3.4% of the ablation group had complications related to catheter manipulation in the heart, including 22 (2.2%) with pericardial effusion. No atrial-esophageal fistulas or deaths due to the procedure occurred.


CABANA has not been published. This is crucial as the manuscript will surely include details needed for a more thorough appraisal.

Here are my initial thoughts.

Science Deserves Respect

A core reason trials produce a higher level of evidence than observational studies is randomization. You can't make causal claims from observational data because the compared groups almost always have confounding factors. Randomization (usually) evens out all other factors besides the treatment.  

This is why readers of CABANA must acknowledge that any findings other than the ITT analysis are observational. ITT preserves randomization. Even though the as-treated comparison makes sense — as Packer put it, "you can't benefit from ablation unless you get ablated" — one must stick to the rules of science.

There are at least two reasons for this: First, the patients who crossed over to ablation may be different from the patients who remained in the drug arm. Second, because the trial was unblinded and conducted in ablation centers, the pro-ablation biases of the doctors can affect the choice to cross over as well as the care of patients after the procedure.

In his presentation, Packer called the ITT analysis honest. University of Michigan professor Rod Hayward equated use of as-treated and per-protocol analyses with scientific malfeasance.

Investigators knew patients would cross over. In the power projections from the CABANA protocol paper,[1] they projected that 25% to 30% of the drug arm would cross over to ablation. They were spot on.

Crossovers reveal the challenges of a pure ablation vs drug comparison, but that does not give us permission to suspend the rules of scientific inquiry. The first line of the conclusion slide from the presentation stands alone: "Ablation did not produce a significant reduction in the primary endpoint and all-cause mortality."

Any electrophysiologist who uses these data to tout ablation as superior to drugs spreads disinformation and demeans our profession. Not only should we not make these false claims, we should vigorously oppose those who do.

Clinical Translation of CABANA Results

The problem with a strict scientific interpretation of CABANA is that it contradicts our clinical observations. After nearly 15 years of doing AF ablation, I believe it has a role because I have seen procedural suppression of AF correlate with improvement in health — hundreds of times.

Perhaps another way to interpret CABANA and preserve the ITT principle is to say the trial did not actually compare ablation vs drugs as was intended but rather it compared the strategy of ablation vs initial medical therapy with ablation for recurring symptoms.

In this scenario, there can be no claim that ablation is superior to drugs for reducing major outcomes, but we can preserve the possibility that ablation remains a reasonable option for selected patients with AF.

I intentionally hedged on calling ablation a reasonable option because of the lack of blinding in CABANA. (Note: Calling this out is not a criticism of the investigators; a placebo-controlled trial of AF ablation was impossible when the trial was conceived.)

Times have changed. We have witnessed the failure of many procedures to withstand the placebo-control test. Failures of pain relief with vertebroplasty,[3] blood pressure reduction with renal denervation,[4] and angina relief from percutaneous coronary intervention[5] underscore the power of the placebo effect. What's more, a doctor's knowledge of treatment assignment may affect basic clinical decisions that pertain to outcomes — the decision to admit to the hospital for cardiovascular reasons, for example.[6]

Permission to Test AF Ablation Against a Placebo

CABANA's failure to reject the null hypothesis will likely give rise to a placebo-controlled trial involving a sham procedure in which patients and doctors are blinded to left atrial ablation.

I know this because two separate investigators told me they are planning such a trial and one other said many leaders are not talking about it publically because they want to be first. No one wants to publish the second or third placebo-controlled trial of AF ablation.

If we ablate AF to relieve subjective endpoints, such as quality of life and cardiovascular admissions, then the only way not to fool ourselves is a placebo-controlled trial.

I will go on record to say that I believe the symptomatic response to AF ablation will be similar to that seen with cardiac resynchronization therapy.[7] Namely, left atrial ablation will cause a strong placebo effect, especially early after the procedure, but there will also be a real effect, most likely correlated with the reduction in AF burden. My degree of certainty, however, is modest. I would not be surprised if AF ablation is ORBITA-ized.


A reassuring finding of CABANA was that ablation did not do worse than drugs. But one of the messages I heard from HRS was that CABANA showed that AF ablation is safe. This is a problem.

The complications in the ablation arm were more serious and more numerous than those in the drug arm. We will have to wait for the published paper for formal comparisons.  CABANA likely represents a best-case scenario because it allowed only experienced operators and centers to be part of the trial. Many people undergo ablation by less experienced operators.

Another important safety issue is the asymmetry of procedural complications. When you talk privately with ablation doctors, many, perhaps most, relay the story of a tragic death of an otherwise healthy middle-aged adult from an atrial-esophageal fistula.

Emotion and Trial Results

The null results from CABANA caused many to feel sad. One colleague remarked to me that it must be deeply disappointing for the investigators.

Why are the results of trials so emotional? Why would we not celebrate the knowledge gained? CABANA studied more than 2200 patients from across the world. It was not able to show AF ablation reduces outcomes. This is important to know, as will be the information coming in future publications.

An experiment's worthiness should not turn on its showing benefit of an anointed drug or procedure. Clinical experiments and the people who run them deserve our respect regardless of the results.

Perhaps if there were less need to confirm our beliefs, there would be less temptation to spin results.


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.