Introduction
Cardiac glycosides represent a valuable therapeutic option for patients with heart failure and a reduced ejection fraction (HFrEF), who remain symptomatic despite optimized guideline-directed pharmacotherapy and device therapy [
1]
. Further, current guidelines recommend cardiac glycosides for rate control in atrial fibrillation (AF) in patients with and without heart failure [
2,
3]. However, prescription rates of cardiac glycosides have been declining due to complementary pharmacologic therapies with beta-blocker, angiotensin-converting enzyme inhibitors (ACEi)/angiotensin-receptor-blocker (ARB), mineralocorticoid receptor antagonists (MRA), angiotensin receptor neprilysin inhibitors (ARNI), and sodium-glucose-cotransporter-2 inhibitors (SGLT2i) improving mortality and morbidity in patients with HFrEF [
4]. In addition, the relative narrow therapeutic range of cardiac glycoside blood concentrations caused concerns of overdosing and toxicity. Recommendations for simple and safe dosing of cardiac glycosides in clinical practice could enable safer use of cardiac glycosides in patients with HFrEF and/or AF.
Both digoxin and digitoxin are approved cardiac glycosides and most frequently used for the treatment of patients with HFrEF and/or AF. Information about the relationship of cardiac glycoside dose and blood concentration in HFrEF and/or AF based on clinical trial data is only available for digoxin [
3,
5], but not for digitoxin. However, digitoxin represents the pharmacokinetically more stable cardiac glycoside compared with digoxin, especially in patients with impaired renal function [
6,
7]. Therefore, digitoxin might even exceed potential beneficial effects observed with digoxin in patients with HFrEF and/or AF [
3,
5].
The DIGIT-HF trial is a randomized, double-blind placebo-controlled trial investigating the impact of digitoxin on mortality and hospitalizations for worsening heart failure in patients with advanced HFrEF [
8]. Importantly, patients with AF or advanced impairment of renal function were not excluded. Due to the pharmacologic properties of digitoxin, a simple standardized dose titration protocol was applied that aimed to achieve serum concentrations within the lower therapeutic range that is currently recommended for cardiac glycosides based on data from the DIG trial [
5]. The current analysis from the DIGIT-HF trial population sought to identify the factors influencing the relationship of digitoxin dose decisions on serum concentrations. Based on this analysis, a score was developed, which can easily be used in clinical practice to select initial digitoxin dosing to avoid overdosing.
Discussion
Although the cardiac glycoside digitoxin has been in clinical use for decades, little data are available enabling simple and safe dose titration of digitoxin, especially in patients with HFrEF. DIGIT-HF is the only sizeable randomized clinical trial investigating the effect of digitoxin within a prespecified range of serum concentrations on clinical outcomes in HFrEF [
8]. We analyzed dose titration data of a large sample randomized to digitoxin within DIGIT-HF, and identified dominant factors associated with the necessity of digitoxin dose reduction that allowed to derive and validate a digitoxin dosing score. These factors, age, sex, eGFR, and BMI, are readily available in clinical practice. Application of the score allowed to achieve the predefined desirable low therapeutic range of 10.5–23.6 nmol/l, a range equivalent to digoxin concentrations of 0.5–0.9 ng/ml that are currently recommended based on data from the DIG trial. Hence, application of the score is expected to enable safe initiation of digitoxin treatment in clinical routine avoiding overdosing.
The only larger data source available up to now demonstrating a relationship between serum concentrations of cardiac glycoside and outcomes in patients with heart failure is the DIG trial. DIG investigated the effect of digoxin vs. placebo on outcomes in patients in sinus rhythm and LVEF < 45% [
5]. Digoxin serum concentrations < 1.0 ng/ml were associated with reduced risks of mortality and heart failure hospitalizations, but concentrations > 1.2 ng/ml were associated with an increased mortality, mainly driven by the increased risk observed in women, but not in men [
11‐
13]. The recently published RATE-AF trial prospectively investigated effects of digoxin on patient-reported quality of life compared to bisoprolol in patients with permanent AF and heart failure symptoms. With a mean digoxin dose of 0.16 mg, mean digoxin serum concentrations were 0.78 ng/ml. Digoxin did not improve the primary quality of life endpoint (SF-36), but significantly improved the secondary outcomes European Heart Rhythm Association (EHRA)-class, New York Heart Association (NYHA) functional class and N-terminal pro-brain-natriuretic-peptide (NT-proBNP) levels, and was associated with fewer adverse events compared to bisoprolol treatment [
3].
Adverse effects and toxicity of cardiac glycosides are very rare at serum concentrations in the lower therapeutic range (for digoxin: 0.5–0.9 ng/ml or 0.65–1.15 mmol/L; for digitoxin 8–18 ng/ml or 10.5–23.6 mmol/l), but have been described at serum concentrations > 2.0 ng/ml of digoxin and > 30 ng/ml of digitoxin [
14]. Based on the DIG and RATE-AF data, treatment with digoxin within serum concentrations below 1.0 ng/ml seems to be safe in patients with heart failure and/or AF [
3,
5]. In the DIG trial, digoxin doses for each patient were chosen with a nomogram taking into account age, sex, body weight, and renal function [
15,
16]. However, no data are available in patients with HFrEF demonstrating the relationship between digitoxin dose and blood concentrations.
Because of their different pharmacokinetic properties, digitoxin represents the pharmacologically more stable cardiac glycoside compared to digoxin, especially in patients with impaired renal function [
6,
7]. Lipophilic digitoxin shows almost complete enteral absorption with a high bioavailability after oral administration (95–100%) and a high plasma protein binding (90–97%). Skeletal muscle, myocardium, kidney, and liver represent the main compartments of distribution. In the elderly, the distribution volume is considerably lower due to reduced skeletal muscle mass (up to 40%), which represents the major tissue compartment. In contrast to digoxin, which is primarily eliminated renally by passive glomerular filtration and tubular secretion, impaired renal function does not influence elimination and half-life of digitoxin, because the reduced renal clearance is entirely compensated by extrarenal (entero-hepatic) clearance keeping the total clearance constant. Only in patients with advanced liver and renal dysfunction, total clearance of digitoxin is impaired with relevant elevation of digitoxin serum concentrations. Overall, the pharmacokinetic properties of digitoxin ensure stable blood concentrations after reaching a steady state at a given dose even in patients with impaired renal function. In contrast, digoxin blood concentrations are less stable, especially at renal dysfunction, with the need of regular controls of blood concentrations.
In the current analysis, we identified eGFR, BMI, and in particular female sex as factors associated with the necessity of a digitoxin dose reduction in the derivation cohort chosen from the DIGIT-HF population. The strong association with female sex in multivariable regression models might reflect the lower skeletal muscle mass of women compared to men, associated with a reduced compartment and overall digitoxin distribution volume resulting in higher digitoxin serum concentrations in women compared to men for a defined digitoxin dose [
6]. Cytochrome P450 IIIA isoenzymes are the main metabolizing enzymes of digitoxin mainly expressed in the liver [
17]. As expression of cytochrome P450 IIIA isoenzymes is reduced in women compared to men [
17]
, this may result in a lower digitoxin metabolization and potentially higher serum concentrations in women [
18]
.
The next most significant factor associated with the necessity of a digitoxin dose reduction was BMI. BMI, in general, positively correlates with total skeletal muscle mass. Therefore, the observed negative association between BMI and necessity for digitoxin dose reduction reflects, similar as for female sex, the relationship between the compartment and overall volume of digitoxin distribution, which is higher in patients with a higher BMI resulting in the need for higher digitoxin doses to achieve therapeutic serum concentrations.
Unexpectedly, we also identified a significant negative association between eGFR and necessity of digitoxin dose reduction. The causal relationship is not clear, because digitoxin elimination is entirely compensated by extrarenal (entero-hepatic) clearance even in patients with impaired renal function and digitoxin serum concentrations are constant even if renal function is markedly impaired [
6]. Although skeletal muscle mass, the main compartment of digitoxin distribution, is reduced by up to 40% in old age [
19], the effect for older patients (age ≥ 75 years) was modest and not statistically significant in the multivariable model, most likely due to collinearity between age and eGFR.
Specifically, the correlation between age and eGFR is -0.55 in our data set, which is substantial. Guided by our aim to maximize the probability of detecting all patients who need a dose reduction, we added age to the score and detected in the analysis and the validation set each three more patients for a dose reduction, which increased score sensitivity (data not shown). Therefore, and taking into account knowledge for the need of digitoxin dose reduction because of reduced muscle mass and, therefore, smaller volume of digitoxin distribution in the elderly [
6], age still was considered for dosing score development.
The intention of this analysis was to develop a score that faciliatates the dosing decision when treatment with digitoxin needs to be started in HFrEF patients. The components of the score are readily available and with high validity in clinical routine. Female sex, high age, low BMI, and impaired renal function fulfilled these criteria. From the clinical perspective and patient safety, a dosing score should, in particular, prevent overdosing of digitoxin, because digitoxin serum concentrations exceeding the therapeutic range might cause grave side effects or even toxicity. Also, the long half-life of digitoxin leads to prolonged elimination times rendering overdose a significant clinical problem of digitoxin. Therefore, a respective dosing score should have high sensitivity to detect the need of a dose,reduction from 0.07.mg to 0.05 mg o.d. in the present population, which would encourage to start with the low dose of 0.05 mg digitoxin o.d. in clinical practice. Specificity of the score on the other hand is negligible, because we would recommend measurement of digitoxin serum concentration after 6 weeks to check whether a dose of 0.05 mg digitoxin o.d. was sufficient to reach serum concentrations within the therapeutic range. Based on the digitoxin dosing score, all women, and all patients ≥ 75 years or with a BMI < 27 kg/m2 or a eGFR < 50 ml/min/1.73m2 should start with 0.05 mg digitoxin o.d.. This approach appears easy to implement and may be expected to improve clinical care. Importantly, the digitoxin dosing score was developed based on randomized, placebo-controlled, double-blinded clinical trial data. However, if this score that we propose here will prove useful in clinicial practise will have to be determined, ideally prospectively in a patient population within standard clinical care.
Our prospective trial data support a retrospective analysis of adverse drug reactions in Germany, indicating adverse drug reactions because of overdosing with digitoxin particular in patients < 70 kg, > 80 years, and women [
20]. Despite that, another retrospective analysis reported a lower risk for toxicity in the elderly for digitoxin than for digoxin [
21], which may presumably be due to the advantageous pharmacokinetic properties of digitoxin compared to digoxin as described above. Our analysis identified eGFR as an important parameter for digitoxin dosing, which was not expected because of the pharmacokinetic properties of digitoxin and not evident from retrospective data. This underlines the importance of prospective, randomized, and blinded clinical studies and is of particular relevance in clinical practice because a significant number of heart failure patients have concomitant impairment of renal function.
Based on recent analyses, cardiac glycosides still may be an important option for the treatment of patients suffering from HFrEF [
22‐
24]. Cardiac glycosides may be, in particular, valuable in advanced heart failure with highly symptomatic patients despite exploited modern pharmacotherapy and device therapy or if applicability of these therapies is limited due to relevant comorbidities, e.g., hypotension and impaired renal function. DIGIT-HF will provide important evidence, whether the cardiac glycoside digitoxin improves prognosis and reduces hospital admissions for worsening heart failure in advanced chronic HFrEF, which will significantly be supported by the digitoxin dosing recommendations based on the DIGIT-HF trial data presented in this manuscript. Because DIGIT-HF included a substantial number of patients with atrial fibrillation, the digitoxin dosing score might also work in atrial fibrillation to achieve safe and effective rate control irrespective of left ventricular ejection fraction.
Declarations
Conflicts of interest
U.B. and J.B. represent the study heads of the DIGIT-HF-study and applied for funding of DIGIT-HF described above. U.B., J.B., A.K. H. v. d. L., C. V., M. B. and S.S. are members of the DIGIT-HF trial steering committee. U.B. received travel support and honoraria for lectures/consulting from Alnylam Pharmaceutical, Amgen, Astra Zeneca, Bayer Vital, Novartis, and Pfizer and institutional research support from Alnylam Pharmaceuticals, all not related to the current manuscript. J.B. has received honoraria for lectures/consulting from Amgen, AstraZeneca, Bayer, BMS, Boehringer Ingelheim, Cardior, Corvia, CVRx, Novartis, Pfizer, Vifor and institutional research support from Zoll, CVRx, Abiomed, all not related to the current manuscript. MB is supported by the Deutsche Forschungsgemeinschaft (German Research Foundation; TTR 219, project number 322900939) and reports personal fees from Abbott, Amgen, Astra Zeneca, Bayer, Boehringer Ingelheim, Cytokinetics, Edwards, Medtronic, Novartis, ReCor, Servier and Vifor during the conduct of the study. A.J.R. received honoraria for lectures from AstraZeneca and Bayer, and travel support from Johnson&Johnson and Servier, all not related to the current manuscript. C.V. received honoraria for lectures/consulting from Abbott, Bayer, Biotronik, BMS, Boehringer Ingelheim, CVRx and Medtronic, all not related to the current manuscript. H.v.L is serving as medical director of Orgenesis, Inc, unrelated to the current manuscript. L.S.M. has no conflicts of interest with respect to the drug studied in this trial. MH received speakerhonoraria, grants and advisory honoraria form Roche Diagnostics, Boehringer Ingelheim, Astra Zeneca, Novartis, Vifor, Biopeutics, Sandoz, Baxter. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.