May 22 2026 This Week in Cardiology

This week, three more digoxin trials, yet another GLP drug on the horizon, vagal nerve stimulation, trial inside baseball, and more on the lipid guidelines. So I want to talk first about uh low dose digoxin and heart failure, the decision trial. And I almost missed this big digoxin trial.

At the ESC Hartweiler Congress in Vienna last weekend, Professor Dirk Jan van Veldhuissen from Groningen The Netherlands, and I'm sorry if I mispronounced all that, presented results of a decision trial of low dose digoxin in patients with hephra. The results were published simultaneously in Nature Medicine about ten days ago, and I missed them last week, even though I discussed a highly positive dig RHD trial of digioxin in patients with rheumatic disease.

The thing is the decision trial has been anticipated since the twenty twenty five publication of the positive but statistically fragile results of the digit HF trial. Decision randomized a thousand patients with symptomatic heart failure and a reduced ejection fraction, or at least ejection fraction less than fifty percent, to low dose ditch or placebo, with the target serum dig concentration of zero point five to zero point nine nanograms per milliliter.

The trial was carried out in forty three sites in the Netherlands. The mean age of patients seventy three years, twenty eight percent female, and almost a third had atrial fib, which is actually an important point. mean LVEF was thirty-three to thirty-five percent, and importantly, at baseline contemporary GDMT was excellent. So

The decision trial was a real test of add-on ditch to the contemporary therapy. The primary endpoint was a composite of total worsening heart failure, that's heart failure hospitalizations or total urgent heart failure visits. or C V death, and that's a little bit different than time to first event. The trial was powered to detect a twenty two percent reduction in the primary endpoint.

Now the results. The primary outcome instances were 3.6% lower in the dig arm. It was 15.7 versus 19.3 per 100 patient years. Hazard ratio zero point eight one confidence intervals zero. Sadly, 0.61 to 1.07. So the p-value was 0.13, which is above that 0.05 magical threshold. Now most of this non-significant reduction was due to worsening heart failure events. The heart failure event number was hazard ratio 0.76.

But again, confidence intervals above one, 0.54 to 1.05. CV death was about the same, 5.5 versus 5.8 per 100 patient years. And death from any cause was also similar seven point seven versus eight point four per one hundred patient years. Also, unlike many hardfire manuscripts, the decision authors tell us total hospitalizations, thank goodness, which were similar.

One of the reasons that a nineteen percent reduction in the primary endpoint did not reach the statistical threshold was a higher than expected dropout rate, potentially due to COVID per the author's discussion. Twenty-four percent in the dig arm and twenty-one percent in the placebo arm stopped the study drug for reasons other than death.

A sensitivity analysis using an as treated-only patients revealed a 34% reduction in the primary outcome. That hazard ratio was 0.66 with tight conference intervals, but of course, of course. We worry about as treated analyses as patients who remain on the drug may be different than those who discontinued.

Now treatment related adverse events were similar. So my comments technically, the decision trial delivered null results, the absolute risk reduction of three point six percent, NNT of twenty seven. is indeed clinically meaningful but did not reach statistical significance. So I suppose we would say that if you assume the null hypothesis of no difference, these results are not surprising enough. and could therefore be due to chance.

Yet if you take them into the context of the previous trials, that is, the DIG trial and the Digit H F trial, The results are directionally consistent, aren't they? There were not differences in serious adverse events, which is reassuring. Now, the problem the trialist had, I think, was not recruiting enough patients. So technically you would or could call this a negative trial, but I think that's too harsh a frame.

The majority of the ninety five percent conference interval was to the reduction side. What's more, if I inputted the positive results of the digit HF trial, albeit with digitoxin, not digoxin, And I do it into a Bayesian format, which I think is quite appropriate given the similar patient populations.

You get something along these lines and it makes total sense. You get a ninety-nine percent probability of any benefit of the primary endpoint over placebo. So anything greater than zero effect is highly likely. you get an 82% probability of more than a 10% benefit, but only a 24% probability of a 20% 20% benefit. I sent the numbers to Jay Brofrey at McGill University, another expert in Bayes, and he got similar estimates, as did Sanjay Call.

Strong probability that DIGE had a greater than nothing effect, but even using the enthusiastic priors from Digit HF, the probability of more than a twenty percent relative risk reduction was quite low at twenty-four percent. My take, therefore, is that it was a null trial and we should call it that, but there was no harm, and it was directly directionally consistent with Digit HF and DIG trial.

Okay, so what do we do when we have three uh three trials? We meta-analyze them. So in JAMMA on May 10th, 12 days ago, a meta-analysis of the three-digit trials in Hefref. including DIG, Digit HF, and Decision, and the authors were mostly from the Decision and Digit HF trials.

And when they combined the three trials, it yielded a sample size of nine thousand patients randomized, the mean age sixty four, twenty two percent female, and the composite of C V death or worsening heart failure Occurred in forty-one percent in the dig arms versus forty-five percent in the placebo arm, that hazard ratio zero point eight five, conference interval zero point eight zero to zero point nine zero, and a highly significant p-value.

First worsening heart failure event occurred in 26% of patients in the dig arm versus 33% in the placebo arm hazard ratio 0.75. So a 25% reduction, highly statistically significant. CV death was twenty seven percent in both arms, all cause death was also thirty two and thirty three percent respectively. The the authors did a few sensitivity analyses and they found that dig effect was similar regardless of the background therapy.

So there was very little background therapy in the old DIG trial, and there was a lot of background therapy in decision and digit HF, and the results were similar. They were also similar using a fixed or random effects types of analyses. The authors state that the effect size for risk reduction of the composite primary was about fifteen percent, and it didn't matter the baseline event rate. The baseline event rate was high in the old ditch trial, a little bit lower in the newer trials.

The driver of the effect was obviously a reduction in heart failure events and not so much CV death. And regarding the lack of mortality effect, the authors rightly state the following quote. The discrepancy between the absence of a reduction in mortality and the reduction of worsening heart failure events associated with digitalis glycosides is not unique and was also observed with the sodium glucose co transporters. And I Vabradine and Pheneronone in the Fine Arts HF study.

Okay, the third study I wanted to mention is the decision withdrawal study. This is a really clever study. The European Heart Journal has published a super clever report from decision on what happens at the end of the study with the Joxin withdrawal. Now as background You recall the very old radiance study. This was Milton Packer led radiance trial where they randomized 178 patients with heart failure and low EF to continue receiving DIG or to take placebo.

And they found a six times higher rate of worsening heart failure events in the withdrawal group. Well, in decision, at the end of the trial, five hundred eighty seven patients were on active therapy, two hundred eighty-eight on Digge, two hundred ninety nine on placebo. And these patients underwent blinded withdrawal. During the one hundred days pre withdrawal, the incidence of C V death or worsening heart failure events were pretty close together.

They were a little lower for DIG, five point seven for dig, six point five uh per patient year one hundred patient years for placebo. Hazard ratio is zero point eight eight and conference intervals zero point two four to three point one zero. But following withdrawal, the incidence rate increased markedly in patients withdrawn from digoxin but not from placebo. It was 42.8 versus 5.9 events per 100 patient years. That's a uh highly significant p-value. It was 14 heart failure events versus two.

The relative risk increase seven point three seven with conference intervals going from one point five to thirty five. Now I know what you're thinking. You're saying those are small numbers of events, fourteen versus two. The conference intervals are wide, but consider mechanistically too, withdrawal of digge also led to a statistically significant increase in heart rate, reduction in blood pressure, and rise in NT ProBNP.

Now the authors tell us that not only is this similar to the old radiance study, but they also remind us of one of the most clever reanalyses ever done, and I want to mention this. Recall that the DIGE trial, the old DIG trial was NIH funded, which means the data eventually becomes available to other authors for further study. Well, fourteen years after publication, a group led by doctor Ali Ahmed and Michael Rich studied discontinuation of digioxin in the DIG trial. They did this by looking

at the subgroup of patients who entered the trial while taking digoxin and then randomized to either staying on digoxin or changing to placebo. And so if a patient was on digoxin chronically, And then we're randomized to placebo, that is a discontinuation. Their papers published in the American Journal of Cardiology, and there were seventy eight hundred participants in the DIG trial, thirty three hundred received digoxin before randomization.

So during a trial, about half of those were randomized to the dig arm and the thus conser cons thus continued to receive Their long term ditch therapy, however, half seventeen hundred patients were randomized to the placebo arm, and thus their long term digoxin therapy was discontinued during the trial. Serum digioxin concertations were made at random during the ditch trial and of the sixteen hundred patients who continued on ditch.

457 had low levels, that is 0.5 to 0.9, and 340 had high levels after one month of therapy. So now the the authors had a three way comparison. They had a discontinued dig in 1700 patients. They had a continued dig with low serum concentrations, 457, and discontinued dig with high serum concentrations, 340. There were some small differences in baseline characteristics. For instance, those with high serum digosin concentrations were slightly older and more often female.

So the results, they they presented two main results. A result for all-cause mortality, and remember that was the primary endpoint of the ditch trial. And then they also present results for heart failure events. So all cause mortality. Get listen carefully to this. 38% in the discontinue arm, thirty-two percent in the continue arm with low serum concentrations, hazard ratio zero point seven three.

with conference intervals 0.61 to 0.87, but forty five percent in the continue with high serum ditch concentrations, so that hazard ratio is one point one seven and conference intervals zero point nine eight to one point three nine. Heart failure events, hospitalizations occurred in forty percent of patients in the discontinue group.

But twenty-nine percent of those in the low serum dig concentration, that's hazard ratio zero point six one, with conference intervals going to zero point five to zero point seven three. and thirty-four percent of those in the high serum digioxin concentration group, 0.78. And again, that was statistically significant as well.

So my conclusion and similar to the decision trial and the radiance trial, discontinuation of long term ditch therapy was associated With an increase in mortality compared to the low serum dig concentration group and worsening heart failure events in both. I would take from the withdrawal data a couple of lessons. One is that if a patient comes to you on digoxin and is doing well and you are titrating up the GDMT, do not stop the digoxin. Withdrawal does seem like it is associated with harm.

The second lesson is that the withdrawal data lends support for the findings of the DIG, digit HF and decision data, in my opinion. Dig skeptics may say the effect size of digioxin is modest, the effect is only on heart failure events, not CV death, and the drug requires careful usage and some periodic blood monitoring for levels. And my counter to that Is that its success its effect size is clinically meaningful? The reduction in heart failure is similar to that of SGLT2 inhibitors.

And two of the three trials, decision and digit HF, included patients with atrial fibrillation, and there's a lot of AF patients with heart failure. And remember, remember that the beta blocker data with heart failure is very much concentrated in patients with sinus rhythm. The Kolcheka meta-analysis and Lancet showed that when you meta-analyze all of the beta blocker heart failure trials, you find that the benefit of beta blockers are only in patients with sinus rhythm.

So when we look at the ditch trials, here we have uh at least two trials that include patients with atrial fibrillation and nurse positive effects. Now the third lesson I would take from this data is is more nuanced, and that is that the many of our foundational heart fire therapies, ACE inhibitors, carvatolol, m uh MRAs in the rails trial, spiralactone,

For instance, they were all shown to be effective in patients taking digoxin. When you look at baseline characteristics, Of those old trials, you'll find that a high percentage of patients were on digoxin. Now do you think that may have had an effect? For instance, Didioxin raises blood pressure. That may have been that may have benefited or made it more beneficial for these drugs that lower blood pressure.

Didioxin has a positive anotropic effect. That too could have modified the benefits of these foundational therapies. We start foundational therapies in 2026, rarely with dig on board, but when they were studied back in the day, these patients were on digoxin, and I think that's something to think about. A final conclusion.

Think about dig use and heart failure. If you if your patient has AF rather than pushing beta blockers higher, maybe add digoxin. For the patient with soft blood pressure who has limited ability to take good doses of ARNI or beta blocker. Considered digoxin. Of course, you always have to use it carefully. It's cleared via the kidney, so adjust the dose and check levels.

But Didge, I think, can help many of our patients. Now we have contemporary data that is directionally consistent with I think the main headwind for ditch use is that there will be no shiny happy farmer reps bringing free bagels in the morning and burritos for lunch. There will be no pictures with banners and cake celebrating the hundredth use of some dubious cardiac device. You just have to think about dig use on your own with no marketing help.

All right, next topic is yet another GLP one drug announced this week. Lily released results of the Triumph 1 trial for their new GLP drug called Ratatrutide. Ratatatrutide. Gosh, I wish they wouldn't make these drugs so hard to pronounce. This drug is actually a triple incrantin receptor agonist acting on the GLP1 receptor, the GIP, which is glucose-dependent insulotropic polypeptide, and the glucagon receptor.

Trial randomized twenty four hundred patients with obesity, one commodity, and no diabetes. The study evaluated multiple doses. Patients on the twelve milligram dose lost an average of seventy pounds or twenty eight percent of their body weight over eighty weeks, with forty five percent of patients achieving a greater than thirty percent weight loss.

Individuals with a baseline BMI greater than or equal to thirty five who participated in a study extension continued to lose weight and achieved up to an average of eighty five pounds. of weight loss. At the four milligram dose reached with only a single escalation step, participants lost an average of forty seven pounds, nineteen percent at eighty weeks, with a lower observed discontinuation rate.

uh due to adverse events, at least versus placebo. Now side effects do seem slightly more common, which makes sense because I think it seems the more potent these drugs are, the more nausea, diarrhea, vomiting, and constipation there will be. it may be that terzepatide may have a better ratio of of efficacy to adverse effects.

The company says the Triumph one trial will be reported at an upcoming ADA meeting, and surely this drug will gain approval and yet there will be another GLP one like drug for patients with obesity. My friends, I don't think I'm wrong about this. I mean, maybe I am, but given the tremendous rise in obesity in the U.S. and other Western countries.

These drugs may end up being more of an advanced than statins. Of course, the main focus is on weight loss, but there's also reductions in blood pressure. an improvement in glucose handling and athrogenic lipids are lower. And of course we know that semaglutide has been shown to be a disease modifying drug in patients with established ASCVD. I hold little doubt that they will also have a primary prevention properties in ASCV and they're likely going to be AFib preventing drugs as well.

Now one place they're probably not going to be helpful is in patients with hefref as they have a substantial increase in heart rate. All right, I wanna tell you a big story in heart failure science regarding vagal nerve stimulation and hefref and this gets into inside baseball and it makes us think about how medical science is generated.

Uh Jack has published results of the Anthem Hefref trial of vagal stimulation. This trial was an FDA-informed pivotal trial examining efficacy of vagal nerve stimulation in patients with hefref who remain symptomatic on GDMT. Now VNS or vagal nerve stimulation proceeds via a proprietary system of right cervical vagus stimulation And from the protocol paper V and S is delivered using an atraumatic self sizing lead cuff and specifically designed lead and pulse generator, think pacemaker,

to electrically activate the vagus nerve in a manner that produces both afferent and efferent functional effects. The VNS intensity is uptitrated subsequently in the ambitory setting, and titration targets a small change in beat to beat distribution of R to R intervals. During uh fourteen second on time of the VNS duty cycle compared with preceding off time.

The randomization in Anthem was two to one, five hundred thirty-two patients. The primary endpoint was typical endpoint, C V da after hard fire hospitalization. And the secondary endpoints were also typical, things like quality of life, uh questionnaires, LVF. Sadly, sadly, after enrolling five hundred thirty two patients, the sponsor terminated enrollment

and this was unrelated to futility or efficacy. The primary efficacy endpoint was not met. Hazard ratio zero point eight four, conference interval zero point six two to one point one two, so the p value is zero point one one. Which is above that threshold. The primary safety endpoint was achieved: 97% freedom from procedure or device-related adverse effects. And the actual autonomic engagement persisted long term.

The still blinded investigators sought to extend data reporting while maintaining scientific integrity, and they had pre-specified a single exploratory outcomes symptoms win ratio. And of course this win ratio, uh which includes the questionnaire and quality of life issues, was favorable. although it's merely hypothesis generating given the primary endpoint being neutral and that there's multiplicity and subjective bias and marginal statistical significance.

The conclusion of the paper published in Jack was that early termination rendered the results uninformative. Now the controversy comes because the DSMB evaluation of the trial results led to a recommendation to continue enrolling patients. They apparently did not find criteria for futility. Now here comes the inside baseball part. And I think I have this correct. The sponsors, a company called Leva Nova, received a breakthrough device designation from FDA.

They also received flexibility to use a Bayesian adaptive design, which would uh allow consideration of a smaller sample size if evidence of benefit was identified at an interim analysis. And so based on these parameters, the Anthem trial included scheduled interim analyses starting at four hundred patients.

had been randomized and again after every additional one hundred patients to assess the futility of further accrual. Then, boom, in the fall of twenty twenty two, The second interim analysis was conducted at the TRILES DSMB found that quote, quote, the pre specified conditions for early FDA filing had not been met. And so the trialists wanted to keep going and randomize more patients. Basically we needed more data.

But the sponsor did not like this. Apparently they wanted early filing. They did their own analysis and a few months later, after not receiving this early filing criteria, said the trial was done. They were done. So later, as an editorial accompanying the manuscript noted, this decision was financial and motivation. Which

is like saying you get wet wet riding a bike in the rain. Of course the decision was financial. The thing is that five hundred thirty two patients had been experimented on. The trialist did the right thing in reporting the underpowered uncertain results. Their attempts to make more of the data with the win ratio was also laudable, but it's still uncertain data. One idea put forth by the editorialists.

But if FDA does give a sponsor special regulatory designation for a breakthrough device, there ought to be also an explicit condition on which trial termination can take place. Here the sponsor overrode the DSMB slow solely because they calculated the device was not going to make it or not gonna make enough money. But it might have made a positive result and we would have learned something.

As I have said many times on this podcast, these are the downsides of dependence on so much industry sponsored science. When the confluence of interest between a profit motive and learning move away from each other even a little bit, industry science sort of fails, doesn't it? I mean, we like industry. They help us. They help make better devices. They iterate and patient care gets better and they make more money. But the nanosecond that that confluence of interest comes apart.

Then it's a problem. I think a better way for medical science would be that the government funding was responsible for the final step, these phase three trials, an independent body studying the phase three studies. That way a neutral party adjudicates efficacy, and we don't end up with early termination of trials like Anthem, and we don't end up with horribly biased endpoints, such as those that we saw in Option or Champion AF.

Alright, final topic today is about lipid guidelines. As many of you know, my most recent column on Medscape chronicled four major concerns I had with the new lipid guidelines. I worried about a the discordance between evidence and the class of recommendations, B changing thresholds for treatment with the prevent equation, C

undue complexity of the guidelines and finally conflicts of interest. I was especially put off by the use of CAC, lip lipoprotein little A, and emphasis on such early and repeated screening. Now, it's had some good attention. Uh the piece has 846 likes, 33 comments, and it's the most emailed piece on Medscape right now. So that's nice. Well, coming soon is a rebuttal from the authors of the guidelines, led by Doctor Richard Blumenthal.

Now I'm not gonna comment further on their thoughts. I've read their piece. Uh rather I would encourage you to go to the website and read both columns. Read me and read your and read their response and see what you think. Finally, on the matter of lipid treatment, there's also one more good piece to talk about, and that's the Jack editor-in-chief Harlan Krumholz has a very provocative editorial on the nocebo effect of statins. He wrote this in his editor's column.

He was commenting on a recent commentary where a doctor described a patient with a coronary calcium score of two hundred eighty. an LDL cholesterol of one sixty eight, a father who died of a heart attack at fifty eight, and this patient had stopped her statin because she felt foggy and achy.

her decision reinforced by what she had read online. At the same time she was injecting B P C one hundred fifty seven, the synthetic peptide, with little evidence of safety or efficacy, into her thigh three times a week for a knee injury on the recommendation of a podcast. HMK went on to describe the problem with statin nocebos, and I call it that because it surely is a nocebo effect, and the evidence for this is as strong as it gets.

Evidence A. Blinded randomized trials of statins find that discontinuation of statin is similar to placebo. That is, if patients don't know what they're on, there's no more uh adverse effect than a placebo. And finally, Francis Wood and James Howard's glorious Samson N of 1 trial in statin intolerant patients clearly showed that statin pills cause the same ill health as statin placebo.

I will link to the Samson trial because I think it one of the maybe one of the most clever studies in all of modern medicine, bar none. The thing that Dr. uh Crumholz highlights is that there is so much bad information out there on statins, and it's true. And I don't know what to do about it except to promote and celebrate the good science that underpins the statins effect.

But I also think we need to be humble and graceful about patient values. For instance, I'm now of the age where cardiovascular disease prevention is relevant. I look at my numbers and risk, it's borderline, and when uh considering a statin, I see a small risk reduction from taking a pill every day for ten years. That's three thousand six hundred and fifty tablets to reduce non fatal events by a few percent.

And different feet people will feel differently about these things. Maximizers will maximize. Minimizers will minimize. And our role as trusted expert advisors is to advise. It's also reasonable to consider in a decision to take three thousand plus tablets in the next decade the skills of our colleagues in the Cath lab. They are truly amazing. And as a treatment for a disease gets better and better, the value in prevention or early detection decreases.

Finally, I ask, but I don't know the answer to this question, what role will GLP1 drugs have? What if a person who's overweight takes a GLP drug and the blood pressure, hemoglobin A one C, triglycerides and LDL plummets? Now a tenure risk might go to six percent. Then what? Maybe GLP1 drugs will reduce the benefit of statins. Let me know what you think. So before I close I just want to say that

Next week, this week in cardiology, we'll transmit from Nyborg, Denmark. I will be at the Danish Cardiac Society meeting talking about evidence and a value, and I'm looking forward to seeing my Danish colleagues. Remember, if you like this podcast, please take the time to give us a rating, write us a review, that helps others find us. Until next week, this is John Mandrola from the Heart.org Medscape Cardiology.