Discussion
This study represents the largest, “real-world” analysis of AF ablation outcomes in patients with continuous, pre- and post-ablation monitoring using a CIED showing how the measure and definition of recurrence following ablation can alter interpretation of procedural success. In short, the incidence of any perceived recurrence of AF following ablation varies substantially depending on definitions of the minimum duration of required AF used to define procedural failure and the surveillance method by which the outcome is assessed. Success rates quoted in the published literature vary substantially depending on the above factors as well as whether patients had paroxysmal or persistent AF at baseline, the ablation technique, the number of ablations and presence of antiarrhythmic drugs, with recurrences in the first year after the blanking period being most commonly reported between 60 to 75%20,21,22.
Detection of AF recurrence following ablation is significantly influenced by the method of monitoring, as is shown in the monitoring strategy simulation. The strategies of a single 24-hour, 48-hour or 7-day monitor had the poorest test characteristics, especially for AF events > 6 min where sensitivity and NPV were < 50%, worse than an unbiased coin flip. These results are likely due to patients having their first recurrences distributed over the year following the blanking period, and when these AF recurrences occur, they tend to cluster over consecutive days, as is shown in the heat maps (Supplement Figure 1 and 2). A 2018 meta-analysis of 66 studies (6941 patients) of AF ablation success in paroxysmal AF found that the method of AF detection following ablation was predictive of whether recurrence was detected rather than the specifics of the ablation procedure itself, with studies using long term CIEDs for monitoring being associated with higher perceived treatment failures than those using intermittent discrete monitoring methods13.
Our findings are concordant with the LINQ AF study, which enrolled 419 patients undergoing radiofrequency (RF) ablation to have a LINQ implantable loop recorder implanted following the procedure. They found success rate varied with the definition of AF recurrence from 46 – 79%, findings overall similar to our study, and the authors suggest that > 6 minutes of AF after a 90 days blanking and/or an AF burden >0.1% be used in future studies as measure of clinically significant success10.
While historically most clinical trials of AF ablation have used time to first recurrence as an endpoint3,4,23,24,25 due to relative ease of measurement, AF burden may be a more objective and clinically relevant endpoint2,12,26,11. In the CIRCA-DOSE trial, 346 paroxysmal AF patients were randomized 1:1:1 to RF pulmonary vein isolation (PVI), 2minute cryoballoon PVI or 4 minute cryoballoon PVI at least 30 days after implantable loop recorder implantation. At 12 months, despite only 51.7% - 53.9% of patients being free of AF, AF burden was reduced by 98.4 – 99.9%14. These results are similar to what was found in the present study, where in patients with paroxysmal AF, the survival estimates for the event AF >6 minutes at one year was 28.2% but had a reduction of 99.6% in the overall median percentage of time in AF. Similarly, in patients with persistent AF, despite the survival estimate at one year of only 24.9% for AF events >6 minutes, there was a 99.3% reduction in the overall median percentage of time in AF.
While improvement in symptoms is the major clinical endpoint of ablation for AF, AF duration and burden appear to correlate with risk of stroke15,16,18,27, raising the hypothesis that strategies that reduce AF burden could potentially mitigate stroke risk, even in the absence of completely eliminating recurrent AF. This hypothesis, along with the data that AF burden reduction may similarly reduce heart failure hospitalization and mortality was tested in a recent random-effect model meta-analysis of large databases, registries and a single randomized control trial (the CABANA trial1) with a total of 241,372 patients. It showed that catheter ablation was associated with a decreased risk of mortality (hazard ratio [HR] 0.62), stroke (HR 0.63) and hospitalization for heart failure (HR 0.64) when compared to medical therapy alone19. While the CABANA trial individually was negative for the primary outcome of death, disabling stroke, serious bleeding, or cardiac arrest, it was subject to substantial crossover and low event rates, and the per-protocol analysis findings are largely consistent with the findings of the aforementioned meta-analysis1.
This analysis has a number of important limitations. It is retrospective and observational in nature; thus, unmeasured confounding cannot be eliminated. The population studied were by definition required to have had prior Medtronic CIEDs implanted, thus introducing significant selection bias in this group, therefore the results may not be generalizable to patients without implanted CIEDs. Furthermore, we do not have accurate data on antiarrhythmic drug use post-ablation, so the actual success rates of ablation alone may be overestimated. Lastly, in order to reduce the false positive rate associated with short unadjudicated AF episodes found on CIEDs, the minimum duration of AF evaluated was 6 minutes, longer than the 30 second threshold recommended in guidelines.