Interpreting the DIGIT-HF trial: Time for a new look at digitalis?

How digoxin and digitoxin differ

Digoxin (C₄₁H₆₄O₁₄) and digitoxin (C₄₁H₆₄O₁₃) are nearly identical in their chemical structures, the difference being that digoxin has a hydroxyl (oxygen-hydrogen) group at a position where digitoxin has only a hydrogen atom.^11,12

Although the mechanism of action is the same, this subtle difference in structure makes digitoxin more lipophilic, thereby altering its pharmacokinetics (Table 1).^9–11,13 It has greater bioavailability, greater plasma protein binding, and a longer half-life, which contribute to more stable plasma levels. In addition, whereas digoxin is mostly excreted unchanged in the kidneys, digitoxin undergoes hepatic metabolism and significant biliary excretion, which may make it safer in patients with renal disease.¹³

Another relevant difference between the drugs is their interaction with P-glycoprotein (P-gp), an active membrane transport protein. Digoxin depends heavily on P-gp for its absorption and elimination; therefore, drugs that modify P-gp such as verapamil, amiodarone, and propafenone, which are commonly used in cardiology, can significantly alter its plasma concentration.^12,13 In contrast, digitoxin, which is more lipophilic, is less dependent on P-gp and relies more on passive diffusion to cross biological membranes.¹⁴

Inclusion and exclusion criteria and digitoxin therapy

Patients could be included if they were age 18 years or older, had heart failure, were receiving guideline-directed medical therapy, were in clinically stable condition for at least 6 months before randomization, and had either of the following:

• Left ventricular ejection fraction 40% or less and were in New York Heart Association functional class III or IV, or

• Left ventricular ejection fraction 30% or less and were in New York Heart Association functional class II.¹⁵

The exclusion criteria included conditions that could increase the risk of toxicity or interfere with the interpretation of efficacy and safety, including bradyarrhythmias, electrolyte imbalances, acute coronary syndrome, recent coronary revascularization, unstable heart failure, and severe noncardiac comorbidities that limited survival.¹⁵

Participants were randomized in a 1:1 ratio to receive digitoxin or placebo. Digitoxin was started at 0.07 mg/day, and the serum level was measured once after 6 weeks to adjust the dose and maintain the serum concentration in the range of 8 to 18 ng/mL.¹

End points and statistical analysis

The primary outcome was a composite of death from any cause or hospitalization for worsening heart failure, whichever occurred first. The key secondary outcome combined death from any cause and any hospitalization for heart failure during follow-up. Additional secondary outcomes included cardiovascular mortality, death from heart failure, sudden cardiac death, noncardiovascular death, and all hospitalizations, both cardiovascular and noncardiovascular.¹⁵

The trial was designed to detect an absolute difference of 5% between groups in the primary outcome, with an expected incidence of 26% in the digitoxin group and 31% in the control group. The planned sample size was 2,190 patients with 734 events over a 36-month recruitment period and up to 48 months of follow-up.¹

The primary analysis was by intention-to-treat. Two main hypotheses were tested:

• Digitoxin would be superior to placebo in terms of the primary outcome, and

• Digitoxin would not be inferior in terms of mortality (to rule out a possible harmful effect of digitoxin).¹

Population

Between May 2015 and September 2023, 1,240 patients were randomized, of whom 1,212 were included in the analysis, with 613 in the digitoxin group and 599 in the placebo group.¹

The mean age of the participants was 66 years, and 20.4% were women. Most had advanced heart failure: 70.4% were in New York Heart Association functional class III or IV, and their mean left ventricular ejection fraction was 29%. In addition, 27.2% had atrial fibrillation.

Regarding guideline-directed medical therapy, 95.7% of the patients were receiving a beta-blocker, 36.0% were receiving an angiotensin-converting enzyme inhibitor, 18.8% were receiving an angiotensin II receptor blocker, 39.5% were receiving an angiotensin receptor-neprilysin inhibitor, 76.2% were receiving a mineralocorticoid receptor antagonist, and 19.3% were receiving a sodium-glucose cotransporter 2 inhibitor. Regarding device therapies, 64.3% had an implantable cardioverter-defibrillator and 25.2% had received cardiac resynchronization therapy.¹⁶

Impact on end points

Over a median follow-up of 36 months, the primary outcome occurred in 39.5% of patients in the digitoxin group, compared with 44.1% in the placebo group (hazard ratio [HR] 0.82, 95% confidence interval 0.69–0.98, P = .03). Thus, 22 patients needed to be treated with digitoxin to prevent 1 primary outcome event. However, neither the all-cause mortality rate nor the rate of first hospitalizations (the components of the composite outcome) was statistically different between groups (Figure 1 and Table 2).¹

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Cardiovascular outcomes in the Digitoxin to Improve Outcomes in Patients With Advanced Chronic Heart Failure (DIGIT-HF) trial.

CI = confidence interval

From Bavendiek U, Großhennig A, Schwab J, et al. Digitoxin in patients with heart failure and reduced ejection fraction. N Engl J Med 2025; 393(12): 1155–1165. doi:10.1056/NEJMoa2415471. Copyright © 2025 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society. License No. 6139170247864.

The all-cause mortality rate was 27.2% in the digitoxin group and 29.5% in the placebo group (HR 0.86, 95% confidence interval 0.69–1.07), satisfying the prespecified noninferiority condition (HR < 1.303). No significant differences were observed in the other secondary outcomes.¹

Subgroup analysis of the primary outcome showed positive trends in nearly all subgroups, but the results were statistically significantly better with digitoxin only in the following subgroups:

• Ejection fraction less than 30%

• Nonischemic heart failure

• Heart rate 75 beats per minute or higher

• Systolic blood pressure 120 mm Hg or lower (Those with higher systolic pressure actually had a slightly higher rate with digitoxin, but it was not statistically significant.)

• Body mass index 30 kg/m² or lower

• No hypertension

• No diabetes

• Estimated glomerular filtration rate 60 mL/min/1.73 m² or lower

• Without advanced devices.

Safety

At least 1 serious adverse event was recorded in 29 patients (4.7%) in the digitoxin group and in 17 (2.8%) in the placebo group, including cardiac disorders in 3.4% and 1.8%, respectively.

Digitoxin or placebo was discontinued due to an adverse event in 56 patients (9.1%) in the digitoxin group and 61 (10.2%) in the control group.¹

Novelty and design

DIGIT-HF is the first trial to evaluate a digitalis drug since the DIG trial.^1,8 The use of digitoxin, whose benefit was uncertain, stands out as its main novelty. It was not sponsored by a pharmaceutical corporation, and it evaluated a drug that costs much less than most recent therapies for heart failure.¹⁵

Meaningful population

DIGIT-HF enrolled a cohort of patients with advanced HFrEF and a high symptom burden. This group was underrepresented in other contemporary clinical trials in heart failure.^17–20 However, in a population with advanced heart failure, we would have expected more events and more deaths.

Simple protocol

The authors used a simple digitoxin dosing protocol, with a single level measurement at 6 weeks to adjust the dose.¹⁵ This contrasts with the complex management of digoxin, which requires frequent monitoring of blood levels.¹³

After measurement, 58% of patients continued on the same dose, and 34% required a dose reduction.¹ Dose reduction was mainly associated with female sex, age 75 years and older, estimated glomerular filtration rate less than 50 mL/min/1.73 m2, and body mass index less than 27 kg/m2; in such patients, an initial dose of 0.05 mg/day may be reasonable.²¹

Clinical applicability and impact

Patient selection was based exclusively on clinical characteristics, without the need for biomarkers or complex tests, making it applicable in real-world practice.¹⁵ Even in patients with advanced disease on guideline-directed medical therapy and with a high rate of device therapy, in whom the incremental benefit of a new therapy may be marginal, digitoxin offers a relevant additive effect, with a number needed to treat of 22, comparable to that in trials of newer drugs (Table 3).^1,16,21

Efficacy in relevant subgroups

Subgroup analysis suggests the benefit of digitoxin is maintained in patients with elevated heart rates, low systolic blood pressure, and renal disease, which are highly prevalent in patients with advanced heart failure who often cannot tolerate guideline-directed medical therapy.¹

Safety

The rate of serious adverse events was low, and discontinuation of treatment due to these events was similar across all groups. No signs of unexpected toxicity were detected, nor was there any increase in mortality attributable to digitoxin.¹ These findings contrast with historical concerns about the safety of digitalis compounds.^7,8,22 Consistent with the post hoc analyses of DIG, DIGIT-HF suggests that digitalis compounds are beneficial at low concentrations.^22–24

Population and underrepresented subgroups

The trial was conducted in an exclusively European population, with fewer women (20%) and fewer patients with atrial fibrillation than in comparable trials.¹⁶ These characteristics partially restrict the generalizability of the results.

Unblinding and performance bias

Unblinding occurred in a limited number of cases (54 in the digitoxin group, 28 in the placebo group).¹ However, the relatively low number limits its overall impact and bias risk.

Lower recruitment and rate of events than expected

DIGIT-HF, with 1,212 patients and 506 events, did not meet the planned sample size or event rate.¹ This reduced the study’s statistical power and limited its ability to detect differences with treatment. The inclusion rate was 2.4 patients per center per year, which was well below usual practice. Recruitment had to be extended from 36 to 84 months, the planned interim analysis could not be performed, and the protocol had to be amended, all of which may have introduced additional biases and warrant cautious interpretation.²⁵

These limitations should be viewed in the context of contemporary guideline-directed medical therapy, particularly following the approval of sodium-glucose cotransporter 2 inhibitors, which significantly reduce heart failure hospitalizations.^18,19 Compared with previous trials, DIGIT-HF included one of the largest populations receiving guideline-directed medical therapy.

Significant composite outcome, nonsignificant individual components

The reduction in statistical power was probably part of the reason that although the primary composite outcome reached statistical significance, none of its components did when analyzed separately. However, effects for the components were directionally consistent with the overall result. Secondary outcomes were also not statistically significant.¹

Discontinuation rate and losses

Treatment was discontinued for reasons other than death in 58.9% of the digitoxin group and 55.1% with placebo.¹ This high discontinuation rate can dilute treatment effects. Although the intention-to-treat analysis partially mitigates this problem, the actual benefit may be underestimated in patients who stay on the drug. The authors do not specify clear reasons for these high discontinuation rates.

Absence of functional parameters and quality of life

Mortality and readmissions are not the only essential factors in heart failure. Other parameters, such as quality-of-life tests, the 6-minute walk test, or the N-terminal probrain natriuretic peptide (NT-proBNP) level, could strengthen the potential benefits of digitoxin and would have allowed us to compare DIGIT-HF with the most recent heart failure trials.