DANISH Details: ICD Use in Nonischemic HF

An Interview With Lars Køber

John M. Mandrola, MD; Lars V. Køber, MD, DSci


September 12, 2016

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

John M. Mandrola, MD: Hi, everyone. This is John Mandrola from on Medscape, and I'm in Rome, Italy, at the European Society of Cardiology meeting with Professor Køber of the University of Copenhagen, who is the primary investigator of the DANISH trial,[1] a trial of defibrillator therapy in patients with heart failure not due to ischemic heart disease. Professor Køber, welcome.

Lars V. Køber, MD, DSci: Thank you.

Dr Mandrola: Can you tell us about the background of the DANISH trial?

Guideline Skeptics

Dr Køber: Originally, we started back in 2008, 2007, because of the European guidelines[2] saying that every patient with heart failure and systolic dysfunction should receive an implantable cardioverter-defibrillator (ICD), and that was based on subgroup analysis of a larger trial and other, medium-to-small trials. We weren't entirely convinced that it was a good idea just to give it to everyone.

Dr Mandrola: You made a comment in the press conference about the Danish view of ICDs.

Dr Køber: Yes. At that time, the ESC guidelines were not endorsed by the Danish Society of Cardiology. There was this note that in nonischemic patients, ICD will not be offered routinely in Denmark. And in Denmark, the social system decides the rules; physicians cannot do as they like.

Dr Mandrola: Tell us about how you designed the trial. It was a long trial.

Dr Køber: We intended to include 1000 patients and we had information from previous trials on what the mortality rate would be in those 1000 patients. We calculated that we could include them over 3 years and follow them for about 3 years, and then we would be done. That turned out not to be the case; we actually had to include more than 1000 patients, because the event rate was lower than expected. We included 1100 patients and randomized them to stay on medical therapy—that therapy had to be as good as it could be so that everyone was on optimal medical therapy—or to get an ICD on top of optimal medical therapy.

Dr Mandrola: Denmark is a smaller country with a smaller numbers of centers. I read that every center recruited patients.

Dr Køber: When we started, there were five centers. Now we have six and everyone participated.

Dr Mandrola: The ejection fraction and B-type natriuretic peptide (BNP) measurements were made after the medical therapy was maximized.

Dr Køber: Yes. Quite often BNP is used as a way to enhance risk. That was not the idea here. We wanted to make sure that these were heart failure patients, so we said that the BNP just has to be a little above normal on optimal medical therapy.

Dr Mandrola: Another feature of this trial is that you included patients who had generator changes and were scheduled for an upgrade—say, CRT-P to a CRT-D [cardiac resynchronization therapy with a pacemaker (P) or defibrillator (D)].

Dr Køber: Yes. When we started, we had about 100 patients who already had a device—an ordinary bradycardia pacemaker or CRT-P—and they were then randomized to an upgrade or not.

Dr Mandrola: That probably increased the risk in the defibrillator arm.

Dr Køber: Those patients have not been included in previous trials, but we wanted to know what the adverse event rate would be if we put this treatment into everyone with guideline indications in Denmark. Another feature related to this is that a lot of patients were candidates for CRT. Those patients were randomized, not to an upgrade but to a CRT-P or a CRT-D.

Dr Mandrola: It's like a real-world application.

Dr Køber: It's really hard to get trial patients to be like real-world patients, but we did our best to get close.

Primary Endpoint and Age Subgroup Analysis

Dr Mandrola: Okay, let's go to the results. There are a couple of big top-line results.

Dr Køber: The primary outcome of the trial was all-cause mortality, and we were powered to show a reduction of 25%. We did not achieve that. We saw a nonsignificant 13% reduction in all-cause mortality. We had three secondary endpoints. One was cardiovascular death. The other one was sudden cardiac death, and the third one was quality of life, which we simply haven't had time to analyze. For cardiovascular mortality, we saw a nonsignificant reduction of 23%. The rate of sudden cardiac death was halved—a 50% reduction that achieved statistical significance (P < .01).

Dr Mandrola: There was a very important subgroup finding.

Dr Køber: We had prespecified a number of subgroups before we did any analysis; most of them are standard: age, gender, previous diabetes, etc. We had one significant interaction, which was age. I always note that you should be very careful with subgroup analyses and how to interpret them, but we had an interaction P value of .009. In the paper we break it into age tertiles; the lower your age, the more likely you are to have a significant reduction in all-cause mortality.

Dr Mandrola: In regard to age subgroup analysis, how do you approach criticisms of looking at a subgroup when the overall trial or primary endpoint is negative?

Dr Køber: I would say that we cannot be certain of the subgroup finding. There is a signal and we have to find out whether we believe it. Does it require more trials? You can always do more trials, but you could also ask how this finding fits with other data. That is generally what the guideline writers do. They consider all the information. Does it make sense that the younger patients without comorbidity, those who are least likely to die from noncardiovascular causes, whose only risk is that they could die from progressive heart failure or arrhythmia—are they the ones most likely to benefit? Does that sound plausible? I think that is what the guideline committee is going to do. You can believe it or not, but I think there is a signal here.

Dr Mandrola: The trial showed a major reduction in sudden death but no overall difference in mortality. Could that have been due to the high percentage of noncardiovascular death?

Dr Køber: Yes. It reflects the length of the trial. We followed people, on average, for more than 5 years (up to over 8 years), and nearly 40% of the patients died from noncardiovascular causes. Of course, an ICD can't really help them. Over time, more and more of the events are noncardiovascular.

Dr Mandrola: That was one of the comments you made in the discussion section of the New England Journal of Medicine paper, about the divergence and then convergence of the curves. In some ways, it's not that much different from the SCD-HeFT trial.[3]

Dr Køber: I think it's a reflection of this being a longer-term trial. If we had done a trial twice as big and stopped after 4 years, I think we likely would have had a positive trial overall. It's interesting, because we're the first trial to include battery replacement, which increased the rate of adverse events because patients are having more than one surgery. We had a prespecified endpoint of infection because we wanted to be certain that we captured all infections. We reported a higher rate of infections than seen in a previous trial, but that was mostly because we included upgrades, battery changes, and generator replacement.

Dr Mandrola: One of the big features of this trial is the percentage of patients on CRT. Do you feel that the results inform the decision of CRT-P versus CRT-D, because in the United States we favor CRT-D?

Dr Køber: I've heard the argument that if you're going to give a device (a CRT), why not just give them the biggest or what you consider the best? My guess before the study results was that patients with CRT probably don't need an ICD, while patients without CRT would need it. But there was no interaction. The results were the same whether or not they had CRT. They saw the reduction in mortality in the beginning and then the curves come back together.

Dr Mandrola: There's some speculation that because the background medical therapy for heart failure has improved, and CRT improves heart failure, there's less likely to be a benefit for ICD therapy.

Dr Køber: I remember when heart failure was a 5-year disease; everyone died after 5 years. Now, approximately 50% die over 5 years. In our trial, it was even less, so that's the good part. We have very effective treatments in beta-blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), spironolactone, CRT. We had a low event rate.

Dr Mandrola: One final question. In the accompanying editorial,[4] there was a lot of talk about risk stratification and opening up a debate about how to best pick patients who are most likely to benefit from defibrillator implants. What do you think about that?

Dr Køber: That is based on our subgroup analysis of age. Older age probably also reflects frailty. There are very old people who are vigorous with no comorbidity, and there are younger patients who are not ICD candidates. Maybe we should try to figure out how to find the right patients. Maybe we should open up that discussion a little more instead of saying that this is for all patients.

Dr Mandrola: Thank you for joining us.

Dr Køber: It has really been a pleasure.

Dr Mandrola: That's it for the first day of the European Society of Cardiology Congress in Rome, Italy. This is John Mandrola from on Medscape.


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.
Post as: