PracticeUpdate: Cardiology - Winter 2018

EXPERT OPINION 21

mortality by intention to treat was almost parallel. The hazard ratio was 0.85, with ablation slightly lower, but the P-value was 0.3. Clearly, from a statistical purity stand- point, an intention-to-treat analysis is what’s looked at first to define a potential therapeu- tic benefit in a trial arm. However, looking at this type of analysis only is simplistic. In a trial in which there is an active therapy and you want to understand the potential bene- fit or risk of that therapy, it is helpful to look at those people who received the therapy independent of their initial randomization. For trials like CABANA, the per-protocol analysis becomes critical to understand- ing the intervention, particularly when you consider that 27% of the patients who were assigned to drug therapy were ultimately crossed over to ablation. Dr. Zipes: The purity of a prospective randomized trial is based on an intention- to-treat analysis, and the statisticians give us lots of reasons why that’s important; but, even so, as a practicing clinician you really want to know what happened with the actual therapy that a particular patient got, and, as Doug Packer said at the meet- ing, if the patient doesn’t get ablation then you can’t evaluate the impact of ablation. So, what did the on-treatment analysis show? Dr. Bunch: Clearly the on-treatment analysis is the pragmatic outcome of the trial. Those results are what clinicians want to know when discussing therapy choices with patients. Those results are what patients want to know as they become personal for them. To me, what is interesting with the on-treatment analysis is that you see a profound benefit of ablation compared with drug therapy. Among the patients who underwent an ablation for the combined endpoint, there was a 33% reduction in mortality, stroke, serious bleeding, or cardiac arrest. In regard to just mortality alone, there was a 34% reduction with ablation compared with drug therapy. In additional subgroup comparisons, we really start to understand the potential benefit, or groups who may benefit, from early ablative intervention compared with exposure to drugs long term. Dr. Zipes: These results, when you look at the on-treatment analysis, are pretty profound and certainly separate ablation from drug therapy, but the purists will argue that intention to treat is really the way to go. Also important, and I think you need to stress this, was that the compli- cation rate from ablation was pretty low, albeit these were highly experienced electrophysiologists.

" I think we have to have some hesitancy before suggesting that ablation itself can

Dr. Bunch: I agree, and I don’t think that the intention-to-treat results should be dimin- ished; however, I think all who read the study need to look at the results through the different lenses of discrete statistical approaches. I think one of the most intrigu- ing aspects of the trial was the overall low rate of complications associated with the procedure itself. There were no esopha- geal fistulas, no tamponades that resulted in periprocedural death. These results were obtained from high-volume centers that can perform ablation safely. I think also what is critical in understanding this trial is that these patients, in a randomized design, really did better than patients in the com- munity. The risk of stroke was exceptionally low in the trial in both treatment arms. Heart failure hospitalization was also low; so, these were patients who were engaged in an aggressive medical treat- ment approach, which likely resulted in higher adherence to anticoagulants, a higher adherence to therapies for heart failure and ischemic disease, etc. The event rates seen were much lower than would have been projected from popula- tion studies or even randomized trials such as AFFIRM, which did not necessarily have all the therapies that we currently consider standard of care. Dr. Zipes: The trial included all forms of atrial fibrillation, not just relatively simple paroxysmal, but persistent and long- standing persistent. Dr. Bunch: Yes. It was almost a 50/50 split between paroxysmal and persistent in the trial, and there were a few longstand- ing persistent patients. I think one of the things that was lost in the meeting and in the commentary that followed was even an intention-to-treat ablation was markedly superior to drugs relative to arrhythmia recurrences. An ablation reduced the risk of atrial fibrilla- tion recurrence by 45%, and that separation in the survival analysis was early and per- sisted over 4 years. In our center, often we interrupt the natural history of atrial fibrillation and its comorbidities such as heart failure, mortality, and cardiac mortality. The trial doesn’t show that completely. "

regard to mortality, stroke, serious bleed- ing, and cardiac arrest, 9.2% of patients in the drug therapy group versus 8.6%, expe- rienced the endpoint; so, it was completely non-significant by intention to treat. How- ever, unlike the AFFIRM trial, the aggressive rhythm control option did not trend toward an increased hazard, but it clearly didn’t improve outcomes by intention to treat. I have heard many people say across social media and mainstem media venues that CABANA was neutral or negative. That is not the case. Ablation significantly lowered the risk of arrhythmia recurrence and car- diovascular hospitalizations even in the intention-to-treat analysis. Although abla- tion failed to lower the primary endpoint, it is stretching it to say that the trial was neg- ative and catheter ablation failed. Dr. Zipes: For our readers, an intention- to-treat analysis means that, once an individual is randomized to treatment A versus treatment B, that person remains in that category A versus B with his ini- tial randomization even though he may switch from A to B. With that in mind what were the crossovers? Dr. Bunch: That was one of the challenges. The other area I forgot to mention was that

continued on page 22

VOL. 3 • NO. 3 • 2018