On the usefulness of AFR analysis and trending

The atrial fibrillatory rate (AFR) is an exciting topic that I’ve had the opportunity to research on over the last two decades, in my work with algorithms for signal processing of ECGs. Today we have a state-of-the-art algorithm for calculating atrial fibrillatory rate, which has important use cases in the prediction of spontaneous AF termination, or termination induced by DC conversion, anti-arrhythmic drugs and ablation procedures.

The seed to this success was planted twenty years ago, around 1995, were cardiologists in both Europe and the US started researching new and better ways to characterise and quantify the typical ECG pattern of the cardiac arrhythmia atrial fibrillation.

Atrial fibrillation (AF) is an arrhythmia that we can observe, but for which the exact mechanisms still remain uncertain. When measured on the surface of the body, i.e., in the ECG, atrial fibrillation manifests as an oscillating baseline, often expressed as a non-stationary oscillatory pattern both in terms of frequency and amplitude, and in terms of waveform morphology.

The cardiologists’ desire to quantify the disease state of the patient, and to investigate ways of predicting treatment outcome, lead them to discover a methodology involving estimation of the atrial fibrillatory rate (AFR). The AFR is a number that represents the dominant or average rate of the oscillating baseline in atrial fibrillation.

When I became involved, the focus was on devising algorithms for extracting a clean atrial signal from the ECG, and for second-to-second estimation of the AFR.

During this development we gained several new insights. For example, we learned that the atria of every patient is in fact oscillating around a certain average number of fibrillations per minute (typically 240-600 fpm), although with a rather large standard deviation between all second-to-second estimates (typically 60-120 fpm depending on the average rate). This implies that it is possible to discern between slower and faster atrial fibrillation cases.

AFR has since been validated and compared to invasive measurements. As expected, the activation rate on the body surface corresponds very well to intra-atrial activation rate especially in the right atrium being closest to the ECG leads on the chest.

AFR thresholds as guidance for treatment

In AFR validation studies, an important observation was made: Slow atrial fibrillation patterns in the ECG is linked to sharper and more stable waveform morphologies in ECG, while the opposite is true for faster atrial fibrillation patterns. The direct interpretation of this observation is that:

  • Slower activation patterns with shaper fibrillation waves in the ECG corresponds to larger and fewer simultaneous activation wavefronts in the atria, involving and synchronising a larger part of the tissue, while a
  • Faster and less organised ECG pattern corresponds to a larger number of less synchronised activation wavefronts in the atria. Thus, the higher rate may partly be a result of a higher possible activation rate for each cell and may partly be a result of the increased complexity of multiple activation wavefronts each activating smaller parts of the tissue.

From this interpretation, AFR is also a measure that express the level of atrial organisation. The higher AFR, the more disorganised, and less synchronised activation of the atrial tissue. A lower AFR approach the normal functioning of the heart, where activation is synchronised to a single wavefront in the atria.

Several smaller scientific studies have shown that patients with a low AFR are more likely to convert to normal sinus rhythm spontaneously, during DC conversion, during pharmaceutical treatment, as well as during catheter ablation.

In the future we may reach a position where we can state an AFR threshold value for which a lower AFR means a high likelihood for conversion without treatment, and also to state an AFR threshold value for which a higher AFR means that it is unlikely that any treatment will work.

Currently, however, the wide spectrum of possible mechanisms behind atrial fibrillation, which may result in heterogenous study groups, and the significance of the left atrium and the electrical connections between the two atria as drivers or substrate for atrial fibrillation, have resulted in rather inconsistent data, which makes it difficult to derive accurate thresholds for AFR.

Clinical and commercial application areas

The application area where AFR has been most successful, both clinically and commercially, is for drug response evaluation. In addition to being able to quantify the state of a disease at baseline, it is of great importance to be able to quantify the response to treatment or other interventions, which may reveal knowledge about the underlying mechanisms of the disease.

Numerous scientific studies shows how AFR is affected by different anti-arrhythmic drugs in the individual patient. As of now, AFR is the only practical way to see that an anti-arrhythmic drug actually works within a patient, unless the patient converts to normal sinus rhythm.

AFR has been used for evaluation of the effect of new anti-arrhythmic drugs in the individual patient before the drug reaches the market. A problem for the pharmaceutical companies developing anti-arrhythmic drugs is that they can only observe when patients converts to a normal rhythm. This is of course the main goal with the drugs, but some patients do not convert, and without AFR there is no way to know whether the patients were at all affected by the drug. Maybe they responded to the treatment with a decreasing AFR, but the decrease was not large enough to convert the patient to normal rhythm. Hence, dosage evaluation is an excellent application for AFR, where effects can be observed by trending.

In the clinical setting, it is also possible to monitor the effect of an anti-arrhythmic drug using a Holter device during the first days when the drug is administered. The effect of the drug intake on the atria becomes observable, and after a few days, it is possible to see the converged effect.

A number of AFR trend features have been used related to drug response evaluation including the:

  • Baseline AFR level before treatment,
  • Total AFR decrease,
  • AFR at conversion,
  • AFR decrease per minute directly after drug intake,
  • Lowest AFR after drug intake, and
  • AFR increase per minute after the time of the lowest AFR for non-converters.

This type of AFR trend analysis is particularly interesting in relation to cardiogenetics, where all these trend features can be seen as part of the AF phenotype and be compared to different genotypes.

Conclusion

I see a great potential in the use of AFR analysis and trending both in clinical research and in drug-development. The above examples with quantification of drug response are only the beginning of exploiting the usefulness of AFR to the benefit of AF patients everywhere.

If you want to know more about Cardiolund Research and the future of ECG analysis, don’t hesitate to contact us.