Is CABANA a positive or negative trial?

Left to right: Eric Prystowsky, Peter Kowey and Munveer Thind

Several treatment options are available for atrial fibrillation (AF), including pharmacological rate or rhythm control, and catheter or surgical ablation. However, consensus as to the most effective treatment remains elusive. The Catheter Ablation versus Antiarrhythmic Drug Therapy for Atrial Fibrillation (CABANA) trial was designed to address this issue. Eric Prystowsky argues that CABANA is a positive trial, whereas Peter Kowey and Munveer Thind believe it is a negative one. Here, they put forward their respective arguments.

CABANA is a positive trial
The CABANA trial randomised 2,204 patients to ablation versus drug therapy.The primary endpoint was a composite of death, disabling stroke, serious bleeding or cardiac arrest. In the intention-to-treat (ITT) analysis, the primary endpoint was 8% for ablation and 9.2% for drug therapy patients, not significantly different. So, is this a negative trial as some “purists” of randomised trials would have us believe, who refuse to consider any other data from such a trial having relevance? I say no, for there are many other factors to consider from the data presented, including how the entire trial results may help patients and clinicians position ablation as a therapy of choice. Let us examine these other outcomes.

Using a treatment received analysis, the primary endpoint was 7% and 10.9% in the ablation and drug groups, respectively (p=0.006). Similarly, for the per-protocol analysis the primary endpoint favoured the ablation group. The secondary endpoint of all-cause mortality or cardiovascular hospitalisation by ITT analysis was significantly better in the ablation group. Atrial fibrillation recurrence rates by ITT were significantly reduced by ablation versus drug therapy. Last, in a different presentation, quality of life improved in both treatment groups, but more so in the ablation group.

Now, before all the PITON (Primary Intention to Treat Or Nothing) crowd send me nasty emails, I understand the problems of using such information when the primary ITT analysis shows no difference, but the other data are thought provoking.

For me, CABANA is a positive trial. My reasoning is as follows:

The primary endpoint is at worst Neutral, which is a major piece of new data on the issue of how to position ablation for AF. The guidelines assigned ablation as a first line treatment with a “dotted line” status and qualification. I say fill in that line. Multiple studies have shown ablation to be superior to drugs to reduce AF recurrences, and now CABANA has shown no significant difference in major endpoints between drugs and ablation therapy. Some will say that the electrophysiologists (EP) in the trial are more expert in the use of ablation and their results do not mirror the general EP community.  Well, the same can be said for their use of antiarrhythmic agents, which often are prescribed by general cardiologists who do not have the same knowledge and experience in their use. I would hope the PITON crowd would not use such reasoning and agree that the trial showed no difference in major outcomes.

When CABANA was presented by Douglas Packer at the May 2018 HRS annual Scientific Sessions, I was the discussant. I was surprised at the volume of negative comments about the trial results, and this background noise still exists. Some of the loudest voices were from those who have never accepted the value of catheter ablation as a treatment option for patients with AF—the “druggies”. These CABANA bashers remind me of what happened when the AFFIRM trial results were published.

AFFIRM compared rhythm control using antiarrhythmic drugs with rate control, and the primary endpoint was cumulative mortality. The trial showed no significant difference between groups. The message from this trial should have been that either treatment strategy is appropriate in patients who resemble the trial population. Yet, all too often the position taken was that rate control is easier and why bother with sinus rhythm? Many patients have suffered because of this, and I still have patients referred to me with symptoms from AF who have been in it for years because their doctor took the easy way out—after years of persistent AF it is very difficult to restore and maintain sinus rhythm even with ablation.

Let us learn from this lesson and use the CABANA data to respond appropriately to the following letter:

Dear Doctor,

I have atrial fibrillation and want treatment for it. I desire to feel better, minimise my chances of a cardiovascular hospitalisation or dying, and reduce AF recurrences after therapy. What do you recommend?

Dear Patient,

There was a major scientific study recently published, CABANA, that addresses your questions. Your best treatment option is catheter ablation. However, antiarrhythmic drugs are still an option.

CABANA is a negative trial
About halfway into the CABANA trial, the death/stroke/bleeding/cardiac arrest composite secondary endpoint was elevated to become the primary endpoint when it became clear that the initial enrolment target of 3,000 patients would not be achieved and the number of deaths would not be large enough to draw reliable conclusions. The original primary endpoint, all-cause mortality, became a secondary endpoint.

By the prespecified intention to treat (ITT) analysis, there was no significant difference between drug therapy and catheter ablation with regard to the primary composite endpoint or all-cause mortality. However, there was a reduction in the composite outcome of death or cardiovascular (CV) hospitalisation and a reduction in AF recurrence in the ablation arm. Also prespecified were per protocol (PP) and treatment received (TR) analyses, which both showed a significant reduction in the primary composite endpoint and in death in the ablation group.

There are many who have examined the CABANA results who hypothesise that there may have been a real benefit of ablation compared to drug therapy but that the ITT analysis failed to demonstrate statistical significance due to a lower than expected mortality, and a higher than expected crossover and loss to follow-up, diluting the power of the study. They also suggest that the significance achieved in the PP and TR analyses, which curtail these post-randomisation biases, further supports this theory. However, these factors were known and acknowledged at the inception of the study, but were not deemed important enough to abandon the most powerful way of interpreting this important outcomes study, ITT.

Intention to treat is the only analysis that maintains the integrity of the randomisation process and is the most powerful tool we have to mitigate bias. This is why it was chosen as the principal analysis method when the statistical analysis plan was constructed. While the negative results may be due to low event rates, crossover, and loss to follow-up, it may also be due to unknown confounders that bias the result of the PP/TR analyses. As such, the statistically significant reduction in mortality in patients treated with ablation based on the PP/TR analyses, while interesting and hypothesis generating, should not be used to make a significant change in clinical practice.

As mentioned, there were several secondary endpoints in CABANA that were considered clinically important. Curiously, the original statistical analysis plan did not specify how alpha was to be apportioned to these secondary endpoints. Without such a contingency, secondary endpoints, such as CV hospitalisation and AF recurrence lack statistical reliability, and as such, must be regarded as “exploratory” and “hypothesis-generating.” The use of an alternative analysis decided upon mid-way through the recruitment process also has the potential to introduce bias.

Finally, many within the ablation community have become enamored of the idea of offering catheter ablation to older individuals with persistent AF, including those with congestive heart failure. They have made the argument that these subgroups in CABANA exhibit benefit akin to what was seen in CASTLE-AF, a seriously flawed study of patients with heart failure. What they fail to consider however, is that by ITT, and in most cases even by PP analyses, none of these sub-groups showed statistical benefit using standard interactional analyses.

If one were to pursue a “glass half full” philosophy, we learned that catheter ablation is as safe as antiarrhythmic drug therapy. In fact the adverse events seen in CABANA were remarkably low, a testimony to the experience and skill of the centres selected to participate. Nevertheless, electrophysiologists, like all clinical investigators, need to adhere to the rules for interpretation of clinical trials. To do anything less will diminish our reputation in the scientific community, and cause our colleagues to question our judgement and integrity. For the sake of our patients, this simply cannot happen.

Eric Prystowsky is a cardiologist with St Vincent Medical Group and director of the Cardiac Arrhythmia Service, St Vincent Hospital, Indianapolis, USA. He is also a consulting professor of Medicine at Duke University Medical Center, Durham, USA.

Peter Kowey is a professor of Medicine and Clinical Pharmacology at Thomas Jefferson University in Philadelphia, USA, and the William Wikoff Smith Chair in Cardiovascular Research.

Munveer Thind is a fellow in cardiovascular medicine at Lankenau Heart Institute, Wynnewood, USA.


Please enter your comment!
Please enter your name here