Digoxin: ARISTOTLE Substudy Provides Clues to a Threshold for This Last-Resort Drug

Ileana L. Piña, MD, MPH; Gaetano M. De Ferrari, MD


April 10, 2017

This feature requires the newest version of Flash. You can download it here.

Ileana L. Piña, MD, MPH: Hello. I am Ileana Piña from the Albert Einstein College of Medicine at Montefiore Medical Center in the Bronx, New York. I am here at the 2017 American College of Cardiology Annual Scientific Session in beautiful Washington, DC.

It is my pleasure today to have Dr Gaetano De Ferrari, the chief of cardiology at the University Hospital in Pavia, who is going to talk to us about a late-breaking trial[1] on atrial fibrillation (AF).

Is Digoxin Effective? Safe? What Proof?

Dr Piña: I have discussed AF in this blog before. It is a growing problem. We used to say that AF was more common in older patients. We wonder whether we should give digoxin or a beta-blocker or go right to amiodarone. Practices are so different, and there is absolutely no consistency. The question about digoxin keeps coming up. This is a very interesting late-breaking trial. Gaetano, where did the idea to look at digoxin come from?

Gaetano M. De Ferrari, MD: We know that digoxin is over 200 years old; yet, we do not know how to use it. Despite the fact that there are no convincing trials indicating that it is really beneficial, approximately 30% of patients with AF worldwide still use it. There are guidelines[2,3] that, despite some differences, do suggest a benefit from digoxin. It is time to have randomized clinical trials; but, in the absence of this, the best we can do is to make very careful observational studies from large databases. With the collaboration of Duke, Duke Clinical Research Institute, and Uppsala Centre, we had the opportunity to analyze the database of the ARISTOTLE trial.

The ARISTOTLE trial compared apixaban, a direct oral anticoagulant, with warfarin on bleeding and embolic endpoints in patients with nonvalvular AF. Approximately 40% of the patients had heart failure (HF). A consistent number of patients, more than one third, were on digoxin. We took advantage of this to analyze the effect of digoxin on mortality.

Dr Piña: How many patients were in ARISTOTLE?

Dr De Ferrari: There were 17,000 patients. We used more than 3000 patients in our prevalence digoxin analysis.

Looking at Digoxin With Help From ARISTOTLE

Dr Piña: You used a subset of the ARISTOTLE population.

Dr De Ferrari: Correct. The goal of our study was to look at patients who were on digoxin. We used propensity score matching, which was very carefully based on 48 baseline variables. This was for the so-called prevalence analysis used to compare patients who were or were not on digoxin when they entered the trial. A major advantage of this analysis is that we also had incidence digoxin users (new users). In other words, we compared patients who started digoxin vs patients with exactly (at least we hope exactly) the same characteristics as those who did not start digoxin—we matched for a huge variety of variables.

Dr Piña: In the ARISTOTLE trial, physicians were allowed to start other antiarrhythmics if they wanted to. They captured that, and that is how you got the incidence users.

Dr De Ferrari: Correct, and also there was a very careful analysis at each visit of new drug treatment. I believe that this database will be superior to the previous one in helping us understand this issue.

Dr Piña: Yes, because it was collected prospectively as the trial was going on.

Dr De Ferrari: Additionally, the main advantage is that we have blood levels. Blood was stored so that we could go back and analyze it. We have to credit Professor Renato Lopes for finding the grant to sponsor this analysis, which was, of course, quite expensive.

Digoxin Levels Are Associated With Mortality

Dr Piña: What did you want to analyze in the blood?

For every 0.5-ng/mL increase in digoxin level, there is a 19% or 20% increase in mortality.

Dr De Ferrari: We thought that digoxin blood levels could be strictly related to mortality. And, as a matter of fact, we found that. There was an overall trend with a 10% increase in hazard ratio for mortality among all digoxin users. However, this increase was significant for patients with levels ≥1.2 ng/mL.

Dr Piña: You analyzed it categorically in groups?

Dr De Ferrari: Absolutely. We divided levels into three groups: <0.8 ng/mL, 0.8-1.1 ng/mL, and >1.1 ng/mL, which means ≥1.2 ng/mL. In this latter category, we found a marked increase—a highly significant increase—in overall all-cause mortality by almost 60%.

Dr Piña: Did you control for renal function? Of course, digoxin clearance is so related to renal function, and levels can go up quickly.

Dr De Ferrari: Absolutely. We did control for renal function—not only at baseline but also the last known renal function before digoxin initiation in the incidence analysis. The main message, perhaps, lies in the correlation between the digoxin level and mortality. We have a linear correlation that indicates that for every 0.5-ng/mL increase in digoxin level, there is a 19% or 20% increase in mortality, which I think is extremely important.

Dr Piña: That is very significant.

Dr De Ferrari: Furthermore, we have the incidence data. If you imagine that the drug may be deleterious, you may imagine that patients who have survived for many years on it are likely to be somehow protected, making you underestimate the deleterious effect in your prevalence analysis. Within incidence analysis, we found a 59% increase in mortality among new digoxin users as compared with matched controls who did not start the drug.

Studying Digoxin in Everyday, Real-Life Clinical Practice

Dr Piña: Don't you wonder why the clinicians started digoxin in the first place? Were these patients sicker, or was it that their ventricular rate was harder to control? You found good background beta-blocker and amiodarone therapy.

Dr De Ferrari: Exactly. Eighty percent of patients were on beta-blockers. Patients were on amiodarone as well. Almost 30% were on calcium-channel blockers—but this may not be the best treatment.

Dr Piña: Not for HF.

Dr De Ferrari: It reflects what occurs in everyday clinical practice. The digoxin was "on top of" these meds and matched for all variables. We even matched for the setting in which digoxin was started. In other words, if it was started during an HF hospitalization, we had three types of controls with the same setting of a new HF hospitalization to account for potential selection bias. By this analysis, we found an almost 60% increase in mortality that appeared to be slightly greater among patients without HF. The increased mortality among incidence users appears to be slightly greater with a twofold increase in mortality among non-HF patients as compared with a 50% increase among HF patients.

Dr Piña: Both groups had increased mortality, but it was less so if the patient had HF.

Dr De Ferrari:. Although formally, the interaction with HF was not statistically significant. The indication suggests that it really is not a good idea, particularly for rate control among non-HF patients. That would be our suggestion.

Dr Piña: This is very important because the guidelines right now in HF[4] say that digoxin use is recommended if symptoms persist in spite of everything else that we give them: aldosterone II receptor blocker, angiotensin-converting enzyme inhibitor, angiotensin receptor/neprilysin inhibitor, beta-blocker. These results kind of go against that, even though we do not know the real reasons why patients were started on digoxin. Was it because of HF, or was it because of the heart rate?

Dr De Ferrari: Yes. We can go back and say exactly what the heart rate was in the beginning, but, again, it was matched. Despite the fact that the heart rate likely was higher among starters than nonstarters, they had exactly the same heart rate in the matched analysis.

Higher Digoxin Levels and Risk for Sudden Death

Dr De Ferrari: I think the final message—which also gives a mechanistic explanation—was that of sudden death. Sudden death among incidence cohort users was four times higher as compared with non-digoxin users. The increase in sudden death was very rapid. The curves diverged very early (within weeks), indicating that, indeed, it is likely that there is a proarrhythmic mechanism linked to digitalis.

Dr Piña: Maybe the digoxin levels were higher too.

Dr De Ferrari: Yes. Unfortunately for the incidence analysis, we largely did not have digoxin levels because blood samples were mainly taken at baseline.

Dr Piña: Right, so you do not have the follow-up.

Dr De Ferrari: Drug initiation could occur at any time during follow-up, but we will try to do some modeling to understand whether there are indirect indications that high digoxin levels could have contributed to this fourfold increase in sudden death.

Take-Home Messages

Dr Piña: If you had to tell clinicians about how to use digoxin and what to take home from this for their own patients, what would you tell them?

If you have to use it, please check blood levels and keep them ≤1.1 ng/mL.

Dr De Ferrari: I would tell them to be extremely cautious with digoxin, to give it as a last resort, and to try all other means to control rate and improve symptoms. I would state strongly to use it only if you have to. But if you have to use it, please check blood levels and keep them ≤1.1 ng/mL. This is a very practical take-home message from this study.

Dr Piña: I hope you take a look at the women separately. You had a good percentage of women. I congratulate the ARISTOTLE group for putting enough women in there.

Dr De Ferrari: Almost half of participants were women.

Dr Piña: I would like to really look at the digoxin levels in the women. Thank you for coming here and sharing this. This is a very important study for clinicians because digoxin really is on the borderline. There are some very well-known researchers who truly believe that it is time for another randomized controlled trial of AF with digoxin using a true control group, but it's not yet on the radar of anybody that I know. The clinical take-home points are very important. Take a look at the levels.

Women with higher digoxin levels tend to get into a lot of trouble and have a lot of side effects that are often not perceived because they are nondistinct and vague. "I do not feel quite well. My stomach does not feel quite right." Get that digoxin level, and check it out.

I want to thank everyone for joining me today. This is Ileana Piña signing off.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.