The Role of Digitalis Pharmacokinetics in Converting Atrial Fibrillation and Flutter to Sinus Rhythm

Abstract

This report examined the role of digitalis pharmacokinetics in helping to guide therapy with digitalis glycosides with regard to converting atrial fibrillation (AF) or flutter to regular sinus rhythm (RSR). Pharmacokinetic models of digitoxin and digoxin, containing a peripheral nonserum effect compartment, were used to analyze outcomes in a nonsystematic literature review of five clinical studies, using the computed concentrations of digitoxin and digoxin in the effect compartment of these models in an analysis of their outcomes. Four cases treated by the author were similarly examined.

Three literature studies showed results no different from placebo. Dosage regimens achieved ≤ 11 ng/gm in the model’s peripheral compartment. However, two other studies achieved significant conversion to RSR. Their peripheral concentrations were 9 to14 ng/gm.

In addition, four patients were treated by the author. Three converted using classical clinical titration with incremental doses, plus therapeutic drug monitoring and pharmacokinetic guidance from the models for maintenance dosage. They converted at peripheral concentrations of 9 to 18 ng/gm, similar to the two studies above. No toxicity was seen. Successful maintenance was achieved, using the models and their pharmacokinetic guidance, by giving somewhat larger than average recommended dosage regimens in order to maintain peripheral concentrations present at conversion. The fourth patient did not convert, but only reached peripheral concentrations of 6–7 ng/gm, similar to the studies in which conversion was no better than placebo.

Pharmacokinetic analysis and guidance play a highly significant role in converting AF to RSR. To the author’s knowledge, this has not been specifically described before. In my experience, conversion of AF or flutter to RSR does not occur until peripheral concentrations of 9 –18 ng/gm are reached. Results in the four cases correlated well with the literature findings. More work is needed to further evaluate these provocative findings.

1. Introduction

Managing patients with atrial fibrillation (AF) or flutter with digoxin has always been difficult. While control of ventricular rate is often achieved by titrating the patient with incremental doses of digoxin, it has been controversial whether or not digoxin is actually able to convert patients with AF to RSR successfully and to maintain them there. The transition from titration to maintenance dosage has usually not been accomplished using therapeutic drug monitoring, pharmacokinetic modeling, or any modern pharmacokinetic guidance. The general clinical impression the author has heard has been that digitalis is not useful in converting patients with atrial flutter, especially chronic, well-established atrial flutter, to RSR.

Therapeutic serum digoxin concentrations are said to range from about 0.5 to 2.0 ng/ml. Most patients with digoxin toxicity have serum concentrations above 2.0 ng/ml, and most clinicians have been loath to accept higher serum concentrations. The therapeutic range of serum concentrations of digitoxin is from about 10 to 35 ng/ml.

However, there is great variation in the clinical sensitivity of patients to digitalis. Doherty [1] showed that while most toxic patients have serum digoxin concentrations above 2.0 ng/ml, there are also many who tolerate quite high concentrations (up to 6.8 ng/ml), and that in his report, actually as many patients with serum concentrations of ≥ 3.0 ng/ml tolerated those high concentrations as were toxic. The author has seen one patient with AF who required 500 μg of oral digoxin 3 times daily to control his ventricular rate. Oral absorption was good. Serum digoxin concentration was 8.0 ng/ml. He was then changed to digitoxin, and required 400 μg of digitoxin daily for maintenance, with a serum digitoxin concentration of 115 ng/ml. A colleague in Albuquerque had a similar patient with AF who required a serum digoxin concentration of 6.0 ng/ml for adequate rate control (R Lueker, M.D., personal communication).

I believe it is highly likely that these unusual patients may have had genetically determined differences in the binding constants of digitoxin and digoxin to their Na-K ATPase. Such wide interindividual clinical variation in patient sensitivity has been largely forgotten since serum concentration guidelines have appeared. However, it clearly exists [1].

Population pharmacokinetic models of digitoxin [2] and digoxin [3] were developed, having an absorptive compartment for oral dosage, a central serum concentration compartment, and a peripheral, nonserum effect compartment, based on the original work of Reuning et al [4].

The literature was reviewed in a nonsystematic manner for references relating to digitalis, digitoxin, digoxin, and conversion of atrial fibrillation to RSR. Three studies showing failure of digoxin to convert AF to RSR compared to placebo [6–9] had computed peripheral compartment concentrations of ≤ 11 ng/gm, while two studies [10,11] showed significant conversion at concentrations of 9–18 ng/gm.

The remainder of this paper considers first, the population models of digitoxin and digoxin, then a review of the literature with the use of these models, and finally, the use of these models in the management of four illustrative clinical cases.

2. The Population Pharmacokinetic Model of Digitoxin

This model was made as part of a bioavailability study for Lilly Laboratories [2]. The overall bioavailability of oral digitoxin tablets was found to be 75%.

While the percent of digitoxin eliminated metabolically is about 75%, with 25% excreted renally [5], this can lead to a problem when the model is fitted to serum concentration data in an anephric patient, who might have an elimination rate constant less than that the average found in such patients, due to the significant variability in the behavior of the drug between patients. This would result in a bad fit to the data, with a negative relationship between renal function and excretion, and an erroneous model for the patient. The estimated percent of drug eliminated by metabolism was therefore judgmentally reduced by the author from 75% to 55%, yielding a rate constant for metabolic metabolism of 0.018875 hr⁻¹. Renal elimination was therefore assumed to be 45%, with a rate constant of 0.000151 hr⁻¹ per unit of creatinine clearance (CR_CL), resulting in a rate constant for renal excretion of 0.0151 hr⁻¹ when CR_CL is 100 ml/min/1.73 M². This permits a safer model where it is unlikely that a patient will have an elimination rate constant less than that of the metabolic rate constant shown here. As a result, the relationship between renal function and excretion will almost always be positive, greater than zero.

Using this model to develop a loading and maintenance digitoxin dosage regimen designed to achieve and maintain a clinically selected target trough serum concentration goal of 15 ng/ml for a typical 65 year old man, 70 in (178 cm) tall, 70 kg body weight, with a serum creatinine of 1.0 mg/dL results in an ideal total loading dose of 750.88 μg, followed by 104.6 μg daily thereafter. This is easily and practically approximated by a 750 μg total loading dose, in two parts 6 hours apart, (checking for any toxicity) with 100 μg/day thereafter. Estimated peak serum concentration is 22 ng/ml with the first dose, and 18 ng/ml thereafter. All predicted trough serum concentrations are 15.0 ng/ml, as desired. Estimated peak concentrations in the peripheral effect compartment are 6.2 ng/gm, with trough concentrations of 5.4 ng/gm.

3. The Population Pharmacokinetic Model of Digoxin

This model of digoxin [3] was also based on the work of Reuning et al [4]. To make it more applicable for clinical use with the newer maximally precise technique of multiple model (MM) dosage design [12,13], its model parameter values (means and standard deviations - SD’s) were converted to discrete model parameter distributions employing a computer program developed for this purpose, using the method of maximum entropy [14]. In this way, the entire parameter distributions (not just single point estimates of mean values) permit development of maximally precise digoxin dosage regimens, individualized to an adult patient’s age, sex, body weight and renal function, to achieve selected target goals in either the central (serum) or peripheral (effect) compartment, using MM dosage design [12,13]. Much of the problem seen with anuric patients on digitoxin is avoided here because the nonparametric digoxin model parameters are complete parameter distributions, rather than just single point summary parameter values, as are usually obtained in parametric modeling approaches [12–14].

Using this model to achieve and maintain a target trough serum concentration of 0.9 ng/ml for a typical 65 year old man, 70 in (178 cms) tall, weighing 70 kg, with a serum creatinine of 1.0 mg/dL, the ideal dosage regimen is a total loading dose of 1027 μg, divided in three parts 6 hrs apart, checking for toxicity before each part, followed by 262 μg on day 2, tapering to 251 μg on day 8. Peak predicted serum concentrations are 2.4 ng/ml on day one, 1.49 ng/ml on day 2, and 1.47 ng/ml on day 8. All predicted trough serum concentrations are 0.9 ng/ml, as that is the target goal. Peak peripheral compartment concentrations are 6.36 ng/gm on day 1, gradually rising to 6.81 ng/gm on day 8. Trough peripheral concentrations are 5.48 ng/gm on day 1, rising to 5.72 ng.gm on day 8. This ideal regimen is easily approximated by a total loading dose of 1000 ug given as 500, 250, and 250 μg 6 hours apart, checking for toxicity, followed by 250 μg/day thereafter.

Comparing these two models, digitoxin and digoxin, approximately equal peripheral compartment concentrations are achieved with both drugs, suggesting that they are approximately equally potent. These findings are similar to those found with earlier more simple one compartment models [15,16].

4. Review of the Literature, with pharmacokinetic analysis of that data

Digoxin is often said to be no better than placebo for converting patients with AF to RSR. The study by Falk et al [6] is widely cited. There were only 18 patients each in the study arm and placebo arm, making it difficult to detect any significant difference between them. They gave digoxin as a fixed protocol, rather than by classical titration. They did not report age, weight, or renal function. Eight of 18 placebo patients spontaneously converted to RSR within 24 hours, compared with 9 of 18 in the digoxin arm.

If one assumes a 65 year old man, 70 inches (178 cms) tall, weighing 70 kg, with a serum creatinine of 1.0 mg/dL, his estimated creatinine clearance [17] is 69 mL/min/1.73 m². If one uses the Bestdose software [18] with the digoxin model [3] and gives the above simulated patient the oral digoxin protocol described by Falk et al (0.6 mg orally, 0.4 mg at 4 hours, 0.2 mg at 8 hours, and finally 0.2 mg at 14 hours, the highest predicted serum concentration is 1.8 ng/ml, and the final serum concentration is 1.4 ng/ml, 10 hours after the last dose. More importantly, the average estimated peripheral compartment concentration reached was only 8.0 ng/gm, just slightly more than the approximately 7.0 ng/gm associated with reasonable therapy for patients in congestive failure with RSR.

The DAAF Trial Group [6] studied 239 patients in a randomized, double blinded, multicenter trial of digoxin versus placebo. Average age was 66.2 years, and weight was 78.2 kg. They received intravenous digoxin (duration of infusion not stated) at mean doses of 0.455 mg, 0.308 mg, and 0.318 mg at 0, 2 and 6 hours. Fifty six of 122 placebo patients (46%) converted to RSR, and 60 of 117 patients (51%) in the digoxin group.

If one assumes an average serum creatinine of 1.0 mg/dL, and gives the above regimen, the Bestdose software [18] predicts a peripheral peak concentration of 9.23 ng/gm at 9.25 hours into the regimen. This is only slightly higher than that predicted for the study of Falk above. The DAAF study found significant ventricular slowing, but no significant conversion to RSR.

The DAAF data was also used to make a population model of digoxin [8]. The authors found that a 2 compartment model fit the data better than a 1 compartment model, similar to our findings [3] and those of Reuning et al [4]. They also showed that the effect (reduction in ventricular rate) correlated well with computed concentrations in their peripheral compartment.

Jordaens et al [9] examined higher doses in 19 patients randomized to intravenous digoxin and 20 to placebo. Average body weight was 73 kg. They gave 0.75 mg at the start, over 10 min, then 0.25 mg over 5 min at 4 hours, and another 0.25 mg over 5 min at 8 hours into the regimen. Nine digoxin and 8 placebo patients converted to RSR. Digoxin slowed the ventricular rate significantly.

If one gives the regimen they gave to a simulated patient of their stated average age and weight, and assumes a height of 70 in (178 cms), male sex, and serum creatinine of 1.0 mg/dL, peak predicted peripheral concentrations were 11.25 ng/gm at 11.55 hours into the regimen. They gave more drug, and had somewhat higher estimated concentrations in the peripheral compartment.

In none of the above three studies was individualized clinical titration for each patient done, nor was renal function evaluated. While they often found a significant reduction in ventricular rate, no study found a significant difference between placebo and digoxin for conversion to RSR.

In contrast, Hou et al [11] compared intravenous digoxin to amiodarone in 50 randomly assigned patients. Digoxin was given as an average total dose of 910 μg/70 kg in three divided doses intravenously over 30 minutes every 2 hours. Average age was 70 years, weight was normalized to 70 kg, renal function and height were not stated. Assuming 70 in (178 cms) height and a serum creatinine of 1.0 mg/dL, our digoxin population model predicted an average measured serum concentration of 1.25 ng/ml at 24 hours after the start of the digoxin. When the model was fitted to the average serum concentration they found (1.02 ng/ml), the estimated peak peripheral concentration was 9.0 ng/gm. Digoxin achieved conversion to RSR in 17 of 24 patients (71%).

Weiner et al [10] studied 47 episodes of AF in 45 patients given rapid intravenous digitalization. Renal function, body weight, and duration of infusion were not stated. They gave 0.5 mg, another 0.5 mg at 4 hours, 0.25 mg at 8 hours, and 0.25 mg at 12 hours, for a total of 1.5 mg. In that study, 40 of 47 episodes of AF in 45 patients were converted to RSR. Assuming our typical 65 year old man, 70 kg, 70 in (178 cms) tall, with a serum creatinine of 1.0 mg/dL, average predicted peripheral dioxin concentrations were 13.9 ng/gm at 15.32 hours into the regimen.

Using the Chi-squared goodness of fit test on the last two data sets, and assuming that a placebo response would result in equal numbers of patients in the AF and RSR groups, the data of Hou et al yields a chi-squared value of 4.17. With df = 1, p< 0.05, a significant difference from placebo. Further, the data of Weiner et al yielded a chi squared value of 23.17, df = 1, p≪ 0.005, a very significant result.

This literature review thus shows that conversion to RSR is significantly related to peripheral compartment digoxin concentrations.

Three other studies [19–21] examined the ability of digoxin to convert AF to RSR. However, the specifics of the dosage regimens were not described precisely enough to permit pharmacokinetic analysis.

4.1 Implications for dosing in acute clinical situations – Managing Atrial Fibrillation and Flutter

Patients with AF or flutter require titration to control ventricular rate and/or to convert them to RSR [22]. They usually require somewhat higher doses than those with RSR alone, and serum concentrations do not reflect a patient’s rapidly changing clinical behavior. The commonly accepted therapeutic range of serum digoxin concentrations is about 0.5 to 2.0 ng/ml. However, patients with AF who have good atrioventricular conduction require serum concentrations of about 2.0 ng/ml [23,24]. Others have described the inadequacy of therapeutic serum concentrations to control ventricular rate in patients with atrial fibrillation [25]. Patients with AF may therefore require a higher therapeutic range of serum concentrations, perhaps from 1.5 to about 2.4 ng/ml. Further, in managing patients in such acute clinical situations, since their clinical behavior at any instant does not correlate with serum concentrations, many clinicians believe that serum concentration monitoring is not useful. When one only looks at the raw data of the serum concentrations, they are quite right. However, as shown by these models, and especially by the work of Reuning et al [4], it is really the concentrations in the peripheral effect compartment that correlate with clinical response, especially in rapidly changing clinical situations, rather than the serum concentrations themselves. Nevertheless, the measured serum concentrations, coupled with models of this type, are the only way to compute and evaluate the complex and otherwise incomprehensible relationships between doses given, serum concentrations, effect compartment concentrations, and the patient’s clinical response, in order to develop the dosage regimens required to achieve and maintain desired target concentrations clinically chosen for each individual patient according to his/her perceived need for the drug, in either the central (serum) or the peripheral (effect) compartment. This point is made especially clear by the behavior of the patient described in Case 3 further on. While peripheral compartment peak target goals of 6 to 8 ng/gm appear appropriate for most patients with moderate congestive failure and RSR, higher target peak goals (9 to 18 ng/gm) appear required to achieve good rate control or conversion in patients with AF.

After one has titrated a patient with incremental doses and achieved a clinical goal such as good rate control or conversion to RSR, the difficult clinical problem has been to judge the correct maintenance regimen to maintain the patient in that state. Without guidance by models and software as described herein, this has usually been impossible.

5. Four Case Reports, with individual pharmacokinetic analyses

5.1 Case 1 – A Patient with Chronic Stable Atrial Flutter

RL was a 65 year old man, 69 in (175 cms) tall, weighing 75 kg. His serum creatinine was 1.0 mg/dL. He had had several small transient ischemic attacks and was admitted for physical therapy and rehabilitation. He had been in chronic atrial flutter for three years, with 2/1 AV block, and a ventricular rate of 140/min. He was well anticoagulated on coumadin both in the past and throughout his hospital course. He had been on oral digoxin 0.25 mg three times daily. Serum digoxin was 1.5 ng/ml, low for his high dose, suggesting an oral bioavailability problem.

It was thought that he might have intestinal organisms metabolizing digoxin to inactive dihydro forms. He was given tetracycline for a week, with no response. Digoxin was then stopped, and he was given digitoxin, as in the author’s opinion, it is the safer drug. Digoxin has been thought to be safer because its half time is shorter. However, drugs with shorter half lives require more frequent observation for equal safety. Just as a patient progresses rapidly from an initial state of having a short duration drug in his body to a final state without it in approximately five short drug half-lives, so also that patient progresses more rapidly into toxicity as renal function decreases without dosage adjustment, for example. Digitoxin is longer acting and therefore more clinically stable from day to day, and is much less sensitive to renal function [2,3]. It also has better bioavailability and more reproducible absorption from day to day [2,3]. Because of its longer half-life and resulting slower fluctuations from day to day variations in bioavailability, dosage, and renal function, toxicity is detected earlier in its slower onset following any change in dosing or renal function than is digoxin [2,3], when it is less life threatening. Further, because it is more lipid soluble, its first manifestation of toxicity is less likely to be a life-threatening arrhythmia and more likely to be anorexia, nausea, vomiting, or some other manifestation of central nervous system toxicity [26,27]. For any frequency at which a patient is seen, toxicity will be detected earlier in its development with a longer acting drug such as digitoxin, rather than with the shorter acting, more rapidly progressive, and more renally sensitive digoxin.

Digitoxin, 0.2 mg/day, a not uncommon maintenance dose, was therefore begun. Figure 1A and B show his clinical response and serum concentrations, which were frequently monitored with peak (1.5 hr after a dose) and trough concentrations, and also with samples at 0.5 and 7 hours after a dose [22,28]. Throughout this period of slow and deliberate clinical titration, rate control was the only goal. Titration consisted of increasing the peripheral compartment target goal by 25 or 30% about 2 or 3 times a week, monitoring serum concentrations, analyzing his data, and computing the regimen to achieve the new goal. At 12 days (288 hrs in Figure 1A), he developed occasional 3/1 AV block, but no significant slowing of ventricular rate.

Figure 1.

Case 1, observed and estimated serum digitoxin concentrations in central and peripheral compartments during the first 32 days (a) and second 32 days (b). The upper lines represent the estimated peripheral compartment digitoxin concentrations, the lower lines represent estimated central compartment (serum) digitoxin concentrations. Crosses represent the observed serum digitoxin concentrations.

In the next 4 days, he developed occasional 4/1 AV block, but still no significant ventricular slowing. The peripheral target concentration was progressively increased, giving the required doses. Serum concentrations became ≥ 25 ng/ml. However, setting his target goal and developing the dosage to achieve it was always guided by his clinical behavior. Toxicity was never seen. Serum concentrations became about 40 ng/ml peak and 25 ng/ml trough. Ventricular rate was about 125/minute. The surprise came at 19 days (456 hrs in Figure 1A), when he converted to RSR at 110/minute. Peripheral compartment concentration at that time was 17 ng/gm.

Now the question was - what dose to give him to maintain his sinus rhythm? This was 0.7 mg/day. The other pleasant surprise was that his RSR persisted.

Everything continued stable until day 25 (600 hrs in Figure 1A), when an aspiration pneumonia developed. Chest x-ray showed infiltrates. Arterial oxygen saturation was low. Everything went back to atrial flutter, 2/1 AV block, and a ventricular rate of 140/minute again. A slightly higher target goal was set, extra small doses given, plus antibiotics. By day 28 (672 hrs), his pneumonia was better, arterial oxygen saturation improved, and he returned to RSR. After that, all digitoxin dosage was kept as it was. He remained in RSR but had two more episodes of aspiration pneumonia, with return to flutter and 2/1 block each time, showing the effect of his hypoxia against the effect of digitoxin. Digitoxin dosage was kept constant throughout. After each aspiration was treated and his arterial oxygen saturation improved, he returned to RSR each time. RSR persisted until he died after a total of about 6 weeks, more than three of which were spent in RSR. Digitalis toxicity was never seen. His serum concentrations and computed central and peripheral concentrations are shown in Figure 1A + B.

Without this software [18], this model [2], and the ability to understand the quantitative relationships between doses, serum concentrations, peripheral concentrations, and clinical behavior, it would have been impossible to develop the correct dosage for him, and to recognize how all this related to his overall clinical behavior and the effect of hypoxia with his episodes of aspiration pneumonia. In contrast to other software, the specific strengths of this software are that it uses nonparametric population models and multiple model (MM) dosage design, which are unique to it, coupled with Bayesian adaptive control, all using multiple model (MM) dosage design for maximum precision to best achieve the desired target therapeutic goal.

5.2 Case 2 – An elderly lady with end stage renal disease

AC was a 92 year old lady coming to the end of her life from an inoperable transitional cell carcinoma of the bladder which obstructed her ureters and produced severe renal failure. She was 66 in (167.5 cms) tall, weighed 75 kg, and serum creatinine was 8.0 mg/dL. Estimated creatinine clearance [17] was only 3.9 ml/min/1.73 M². She had stopped food or medicines by mouth.

One day she became acutely dyspneic, with pulmonary edema, rales in both lung fields, and new onset AF with a ventricular rate of 170/minute. She received oxygen, diuretics, and an initial dose of 500 μg of digoxin intravenously over 5 minutes. After 3 hrs 20 min, her ventricular rate fell to 135/min, and 250 μg was given intravenously. Shortly after, however, she developed atrial flutter with 2/1 AV block, and her ventricular rate actually increased to 150/min. This raised the question of digoxin toxicity. She was watched and monitored very closely. After 7 hours, she reverted back to AF, again at about 135/min, but with larger fibrillatory waves on her ECG than before, as is commonly seen in titrations of patients with AF with digitalis. It seemed likely that she might have developed flutter as larger portions of atrial myocardium were participating in fewer random pathways of atrial depolarization. This was still evident now, with her larger fibrillatory waves. It thus seemed likely that slightly more digoxin might further facilitate this process. If so, it might bring about RSR if the entire atrial myocardium should participate in depolarization again. Because of this, and because she was again in AF with a still inadequately controlled ventricular rate, she was given another dose of 250 μg intravenously. After one hour and 15 minutes, she converted to RSR at a ventricular rate of 110/min.

Now the question was – what dosage regimen to give her to maintain her in RSR? Using the Bestdose software [18] and the digoxin model [3,22], her data, doses, and extremely low CR_CL were analyzed. Figure 2A and B show that the profile of her predicted serum concentrations was not helpful in understanding her situation, while her estimated peripheral concentrations correlated very well with the evolution of her clinical status. Ventricular rate dropped as peripheral compartment concentrations rose. She converted to RSR when her weighted average estimated peripheral concentration was 9.3 ng/gm. A target of 9.5 ng/gm was chosen, and a regimen developed to maintain that goal. The ideal intravenous regimen was 141 μg/day. The ideal weekly maintenance dose therefore was 141 times 7 = 987 μg/week. This was approximated by a weekly intravenous dose of 1000 μg/week, or 8 doses of 125 μg/week, each infused over 15 minutes. A daily dose of 125 μg was therefore given for 3 days, then a single dose of 250 μg, then 125 μg/day for the remainder of the week, and so on. She remained in RSR for the next two weeks of her life until she died of a pneumonia. Digitalis toxicity was never seen. In this lady, no serum concentrations were measured. She was managed using the digoxin population model alone.

Figure 2.

Case 2. Profile of estimated weighted average central compartment (serum) digoxin concentrations (a) and estimated weighted average peripheral compartment digoxin concentrations (b). The patient converted to RSR after dose 3, and remained in RSR for the remainder of her hospital course. Vertical bars at bottom represent the times of doses given. RSR regular sinus rythm

5.3 Case 3 – A patient already converted three times

A 58 year old man, 68 in (173 cms) tall, 75 kg, with serum creatinine = 0.8 mg/dL, had been in RSR on chronic oral digoxin 0.25 mg/day [21,28]. He missed a dose one day and developed new onset rapid AF. He was titrated with 4 intravenous doses of 0.25 mg each, given over 1 day, and converted to RSR.

His previous maintenance dose of 0.25 mg/day was resumed. He went back into AF within two days, at about 490 hours into the regimen, as 0.25 mg/day did not maintain the peripheral compartment concentration at the value associated with his successful conversion. A serum digoxin concentration drawn 11 hours and 15 minutes after the last dose was 1.0 ng/ml, and the patient was in AF.

He was again titrated with intravenous digoxin and converted to RSR after two intravenous doses of 0.25 mg three hours apart. A serum sample obtained 14 hours 20 minutes after the second dose above, when he had converted back to RSR again, was exactly the same – 1.0 ng/ml.

5.3.1 Three Questions

First, how can someone have AF at one time and RSR at another with exactly the same serum concentration? Many cardiologists feel that serum digoxin concentrations do not correlate with clinical behavior. This patient is a striking example.

Second, was the patient in a steady state before each serum sample was taken? Clearly, no. At the first sample (1.0 ng/ml), he had just reverted back to AF. At the second, (also 1.0 ng/ml) he had just converted to RSR. The third sample was 1.2 ng/ml, and he was again in RSR. So he was not at all in a steady state to permit the conventional interpretation of the relationship between serum digoxin and clinical behavior with any of those samples.

Third, were the serum samples taken at the same time after the dose? Again, no.

After his second conversion to RSR, he again received 0.25 mg/day. After three days, he reverted back to AF. At that time he had a concentration of only 7.0 ng/gm in his peripheral compartment, similar to the patients studied by Falk et al [6]. He was again titrated with five doses of 0.25 mg of intravenous digoxin over the next 36 hours and converted, for a third time, to RSR. He then had a concentration of 12.7 ng/gm in his peripheral compartment. Toxicity was never seen. A serum sample drawn 14 hours 45 minutes after the fourth dose was 1.2 ng/ml, and he was in RSR. There was no correlation between serum concentrations and clinical behavior. He was in AF at a serum digoxin concentration of 1.0, and in RSR at 1.0 and 1.2 ng/ml.

Such puzzling clinical behavior is seen when patients are being monitored with occasional serum digoxin concentrations, looking for clinical correlations. They are not there. Many clinicians feel that monitoring digoxin serum concentrations is not useful. They are correct, if one looks only at the raw data. However, Figure 3B shows that fitting the model to the serum concentrations [18] yields good correlations with this patient’s clinical course.

Figure 3.

Case 3. Profile of estimated weighted average serum concentrations (a) and estimated weighted average peripheral compartment concentrations (b) of digoxin. The patient missed his usual daily dose of 250 μg of digoxin at 400 hours, and went into rapid AF. He converted to RSR at 450 hrs, went back into AF at 475 hrs, back into RSR at 490 hrs, back into AF at 570 hrs, and back into RSR at 600 hrs. The target peripheral compartment peak concentration goal of 11.5 ng/gm was closely approximated by the regimen given thereafter, and he remained in RSR thereafter. Note the good correlation between his clinical behavior and the peripheral compartment concentrations, and the lack of such correlation in Figure 3a between his behavior and his central (serum) concentrations. Vertical bars at bottom reresent times of doses given. AF atrial fibrillation, RSR regular sinus rythm

In contrast to his serum compartment data, the relationship between his clinical behavior and his computed peripheral compartment concentrations was easy to understand. For some reason, not clinically clear, his requirements for digoxin had changed, and he now remained in RSR only with computed peripheral compartment concentrations between 10 and 13 ng/gm, not 5 ng/gm, as they had been before he missed his dose and developed AF, or 7.0 ng/gm as when he reverted back to AF after his second conversion to RSR.

Based on this analysis and his clinical behavior, a target peak goal in the peripheral compartment of 11.5 ng/gm at 7 hours (the usual model peak time after an oral dose) was selected. An ideal oral dosage regimen of 468 μg, followed by 578 μg, and then by 572 μg/day was computed [18]. This was judgmentally revised to a first dose of 250 μg, and then (since 572 μg is about halfway between 500 and 625 μg), to 625 and 500 μg on alternate days, or an average of 562.5 μg/day. His previous maintenance dose of 250 μg/day had no longer been able to maintain him in RSR. As shown in Figure 3, estimated trough serum concentrations back when he had previously been in RSR were only about 0.5 ng/ml, and his estimated peak peripheral concentrations were only about 5.0 ng/gm. On his new dosage regimen, estimated trough serum concentrations ranged from 0.88 ng/ml the first day to 0.92 after one week, and the target peak peripheral compartment goal of 11.5 ng/gm was predicted to be closely approximated. He was given the above revised regimen. The individualized, clinically selected, target peripheral compartment goal was closely approximated, as shown in Figure 3B.

On that regimen, he remained in RSR, and was able to leave the hospital in RSR, whereas the previous full week of therapy had been unsuccessful. Two weeks later, still on the above regimen, he was still in RSR when seen in the follow-up clinic. Unfortunately, no serum sample was obtained then.

5.4 Case 4 – A very large, heavy patient

DD, a 41 year old man, 71 in (180 cms) tall, weighing 300 pounds (118 kg), was in chronic AF. His serum creatinine began at 0.6 mg/dL, rose to 1.4 mg/dL, and then fell to 0.8 mg/dL. His estimated creatinine clearance began with a high value of 205 ml/min/1.73 m². It then fell to 75, and rose again to 155 ml/min.1.73m² [16]. His ventriclar rate was about 130/min throughout, with little change during his therapy.

Oral digoxin dosage began at 250 μg/day, but rose to 500 and 875 μg/day in divided doses, checking clinically before each next dose, and to 1250 μg/day one day, followed by 1000 μg/day, checking before each dose. At this point others involved in his care did not wish to keep up these doses, even though 1000 μg/day for a 300 lb man is equivalent to 500 μg/day for a 150 lb person.

He had 10 serum concentrations measured, ranging from 0.9 to 2.7 ng/ml after a dose. When his data was fitted [18], a hybrid Bayesian procedure [29] was used which can reach out beyond the stated parameter ranges of the population model, which was developed for patients not so heavy. A reasonable fit was obtained, as shown in Figure 4A. Peripheral compartment concentrations reached only about 6 ng/gm, as shown in Figure 4B. This very heavy patient’s clinical behavior correlated well with other patients in the literature whose clinical response concerning conversion to RSR was not significantly different from placebo [6,7,9].

Figure 4.

Figure 4a. Case 4. Profile of estimated weighted average serum concentrations (a) and estimated weighted average peripheral compartment concentrations (b) of digoxin. Note the failure of this patient to convert at peripheral concentrations of 6 ng/gm, similar to the findings of Falk et al. [6]. The numbered dots represent the various measured serum concentrations obtained and the vertical bars at bottom represent times of doses given.

6. Discussion

Target digoxin goals must be clinically individualized for each patient. General guidelines set down by committees, such as a therapeutic range of trough serum digoxin concentrations from 0.5 to 2.0 ng/ml, are useful for many patients, but not for all. As shown by Doherty [1], sensitivity to digoxin varies widely. It thus becomes each individual physician’s responsibility to consider carefully just how much each individual patient needs the drug, how dangerous the patient’s clinical situation is without it, and how great a risk of toxicity appears acceptable to obtain the anticipated effect of the drug. If a patient does not need the drug very much, only a low risk of toxicity is acceptable. One selects a low target goal and a gentle regimen. However, if the patient is in significant danger and needs the drug “pushed”, one selects a higher target goal and develops the more aggressive regimen to achieve it. Doing this with maximal precision, using MM dosage design [12,13], permits maximum precision and safety. Even in patients with congestive failure and RSR, Hoeschen and Cuddy [30] showed that one can often improve many dosage regimens by carefully selecting a higher target and giving a somewhat higher regimen to achieve it. Such approaches have been shown to result in improved myocardial function [30]. Their average serum concentration in patients receiving maintenance dosage of 0.25 mg/day was 0.56 ng/ml, while 0.5 mg/day yielded 1.2 ng/ml. Their optimal dosage was 0.43 mg/day. They showed a significant inverse correlation between left ventricular ejection time and serum digoxin concentration. Their findings indicate that patients may well benefit from careful dosage escalation, guided, for example, by modeling, pharmacokinetic guidance in dosage design, and the serum concentration monitoring strategies described herein.

The old saying that “the right dose of digitalis is enough” has never been more true. In today’s world of serum concentrations and fear of using one’s individual judgment concerning each patient’s probable clinical sensitivity to digoxin, using guidelines to avoid guilt rather than accepting one’s real clinical responsibility as a thoughtful advocate for each patient’s individually optimized therapy, digoxin is too often thought of as a last-line drug. Pharmacokinetically designed approaches to drug therapy badly need to be taught to medical students in a clinically meaningful way. Almost no physicians have been trained in the modern techniques of drug therapy that can be achieved with pharmacokinetic approaches such as those described herein.

Lastly, digitalis toxicity is considered as something that is diagnosed by its clinical manifestations. Serum concentrations are often used to aid in its recognition. In fact, in diagnosing digitalis toxicity, there is actually nothing to diagnose. The patient’s clinical behavior, whether nausea, vomiting, an arrhythmia, or a psychosis, has a well-known list of causes, such as a high serum digoxin concentration, hypoxia, hypokalemia, hypercalcemia, or hypomagnesemia, and of course, heart disease, for example. One can easily go through such a list and optimize it for the patient. The cell responds to the total environment which surrounds it. Case 1 is a good illustration of the combined factors in the intra and extra cellular environment and the development of arrhythmias, for example.

7. Conclusion

Based on the review of the literature, and on the cases reported here, digitalis glycosides can convert patients with AF, and even chronic, well-established atrial flutter, to RSR, and can maintain them there. The key is the use of pharmacokinetic models which describe the behavior of these drugs, and which, coupled with thoughtful measurement of serum concentrations, permit more informed, pharmacokinetically guided approaches to dosing. It is essential to compare each individual patient’s clinical behavior with that of his/her pharmacokinetic model, setting each patient’s individualized target goal based on that comparison, and development of the required dosage regimen to achieve it. Nonparametric population pharmacokinetic models [3], Bayesian adaptive control [18,31], and maximally precise MM dosing design [12,13] provide the structure for this approach.

Further work appears indicated to further evaluate these interesting and provocative findings, and perhaps to re-examine the role digitalis glycosides may be able to play in a more informed and optimal pharmacological approach to patents with AF.

Key points.

1. The important relationships between dosage given, serum concentrations, concentrations in the peripheral effect compartment, and clinical behavior in individual patents cannot be seen without the help and guidance provided by pharmacokinetic models of drug behavior.

2. Use of these models, along with strategies of Bayesian adaptive control of dosage regimens, permits improved management of patients with atrial fibrillation and flutter, and often achieves and maintains conversion of the arrhythmia to sinus rhythm.