Thoughts on CABANA — The biggest study in AF ablation in years

(The CABANA trial is slated for release in two days at the 2018 HRS Meeting. Here is a preview.)

Technology has made modern life easier. This is good for lots of things, but not for avoiding atrial fibrillation (AF).

As people in Western society grow older, heavier, less physically active, more distracted, and perhaps more anxious, the growing epidemic of AF should surprise no one.

AF is a serious condition that can impair quality of life and cause complications such as stroke and heart failure. AF drugs are woeful, risk factors are hard to modify; these realities have led many experts to embrace catheter ablation of AF. Favorable reimbursement for AF ablation has also not hurt its growth.

Increasing enthusiasm for AF ablation, however, has occurred without robust outcomes evidence.

• Does creating scar in the left atrium make people live longer or have fewer strokes?
• Is the risk of the procedure balanced by better outcomes in the future?
• Are AF episodes (recorded on monitors or smartphones) a good surrogate marker of trouble? We already know that AF ablation reduces AF episodes compared to drugs, but to what end–does this reduction in episodes translate to less stroke, longer life or or fewer  heart failure admissions?

This is why The Catheter Ablation vs Antiarrthymic Drug Therapy for Atrial Fibrillation (CABANA) trial[1] may be the most important trial in modern electrophysiology. Dr. Douglas Packer from the Mayo Clinic is set to present CABANA findings this week at the 2018 Heart Rhythm Society Scientific Sessions.

Here is a primer on what to expect:

We must first consider the type of patient included and excluded in the study. CABANA is an outcomes trial so the authors wanted patients with AF who had an increased risk of complications. They included patients similar to those in the famous rate- vs rhythm-control outcomes study called the AFFIRM trial[2]—those over the age of 65, or, if under 65, with at least one risk factor for stroke. Patients had to have documented AF and be eligible for both ablation and use of at least two drugs.

The authors excluded young patients with AF who had no risk factors, those who had already failed two or more drugs and those who had reversible causes of AF, such as post-op AF or alcohol-induced AF. Patients with hypertrophic cardiomyopathy and those with previous ablation were also excluded.

CABANA compared left atrial ablation (mandating pulmonary vein isolation) to medical therapy. Ablation could be done with various RF catheters or a cryoballoon.  Medical therapy could include either rate or rhythm-control drugs—which is important.

The distribution of rate vs rhythm medication used in the comparator arm deserves attention. It’s good that both rate- and rhythm-control strategies were allowed, because if only rhythm control drugs were allowed and then ablation did better, one could argue that ablation beat an inferior comparator. Letting doctors choose either rate-or rhythm-control mirrors regular practice and thus improves the external validity (generalizability) of the study.

Since CABANA is an outcomes trial, we also have to note any baseline differences in patients. Randomization usually sorts this out, but if more patients in one group received non-vitamin-K-oral antagonists, for example, that would be important.

CABANA recruited patients from more than 100 centers worldwide. That’s a lot of data and patients to keep track of. We must keep an eye out for dropout. One always looks at the rate of dropout vs the difference in outcomes. It hurts reliability if the trial has a 3% dropout rate and the absolute difference in outcomes was only 1%.

Clinical trials must choose an endpoint to measure. In CABANA, the original primary endpoint was mortality. That’s a good endpoint because it’s easy to count and not susceptible to bias.

The problem with death as an endpoint is that AF isn’t life-threatening in the short-term, and to show a difference in death rates requires a lot of patients. About halfway through the trial, CABANA leaders along with the data safety monitors met to address two problems–a lower than expected event rate, and slower than projected accrual of study subjects.

They did not have access to any treatment-specific outcomes but decided to change the endpoint to a composite of four outcomes–total mortality, disabling stroke, serious bleeding, or cardiac arrest.  Overall mortality was changed to one of many secondary endpoints.

Changing the primary endpoint allows for more events to accrue and makes finding differences between the treatment arms easier. But it also complicates understanding the trial because one has to see which of the primary endpoints “drove” the result.

One must also consider that two of four primary endpoints (disabling stroke and serious bleeding) are susceptible to ascertainment bias.

To learn more about adjudication of endpoints, I reached out to trialist Dr Milton Packer who explained in an email that in an unblinded trial, like CABANA, sorting out a disabling stroke from a non-disabling stroke could be tricky. All CABANA centers were ablation centers, so Packer asked, “Were investigators more likely to identify and report strokes that occurred in non-ablated patients? Remember, he added, the adjudication process only operates on reported strokes.”

Packer identified another tricky part of the CABANA results—the choice to analyze data using a time-to-first event approach. Here, a major bleed that occurs before a disabling stroke will mask the stroke. That means all strokes that occur in the trial may not be counted in the primary endpoint.

CABANA also includes 13 different secondary endpoints ranging from hard endpoints like death to softer endpoints like quality of life. One endpoint that you can expect to be better with ablation is freedom from recurrent AF episodes. But what if ablation dramatically reduces AF but there are no or minimal differences in outcomes? You see the problem: AF episodes might be a lousy surrogate measure. That would have great implications for digital AF screening.

Finally, the lack of a placebo controlled ablation arm will limit interpretation of subjective endpoints. I am sorry, but it’s true. Consider two key secondary endpoints, the decision to admit to the hospital for cardiovascular reasons and quality of life scores. Knowledge of the treatment arm surely affects these outcomes.

If CABANA finds no difference in mortality or stroke, and the only reason to ablate AF is to relieve symptoms, then a placebo-controlled AF ablation trial will be required.

Here are two posts I’ve written on AF ablation and placebo: Is AF ablation a big placebo?  and on theHeart.org | Medscape Cardiology: Could Ablation for AF Be an Elaborate Placebo?

References:

1.         Packer DL, Mark DB, Robb RA, et al. Catheter Ablation versus Antiarrhythmic Drug Therapy for Atrial Fibrillation (CABANA) Trial: Study Rationale and Design. American Heart Journal. 2018.
2.         A Comparison of Rate Control and Rhythm Control in Patients with Atrial Fibrillation. New England Journal of Medicine. 2002;347:1825-1833.