Digoxin

This chapter is relevant to Section G8(iv)  of the 2017 CICM Primary Syllabus, which asks the exam candidate to "understand the pharmacology of antiarrhythmic drugs".  Specifically, this is a monograph on digoxin, which for whatever reason has attracted a lot of attention in the CICM First Part Exam. An entire series of questions has been dedicated to the pharmacology of digoxin and its side-effects, mainly in the form of SAQs asking the candidates to compare it with another antiarrhythmic:

• Question 14 from the first paper of 2019 (digoxin vs. sotalol)
• Question 2 from the second paper of 2018 (amiodarone vs digoxin)
• Question 10 from the second paper of 2013 (digoxin vs. levosimendan)
• Question 22 from the second paper of 2010 (amiodarone vs digoxin)
• Question 5 from the second paper of 2009 (digoxin)

On some level, this makes sense, as digoxin is still quite popular and is widely spread through the community of chronic AF patients. On the other hand, its utility in the ICU is limited. This statement is not made with solid Level 1 evidence in hand, but with the understanding that a population of old-school intensivists is at the helm of the training program, and these people have a certain bias against this substance. Ancient behemoths of Australian critical care have historically dismissed it as useless. "Digoxin is used infrequently as there are other agents that have a superior inotropic effect, a greater ability to control and reverse supraventricular tachyarrhythmias, have a larger therapeutic window and are easier to regulate", they rumble in the distance.

In summary:

Eichhorn & Gheorghiade (2002) is probably the best single article for this topic, if you only needed to read one thing about digoxin. Unfortunately no single article gives a satisfactory account of all the electrophysiological effects and their relation to its positive and negative properties, making you dig through tons of articles from the sixties and seventies. The results of this archaeological expedition are presented below for the reader with infinite patience; the rest are directed to the Part One entry on digoxin, which presents it in an easily digestible form without snarky asides.

Chemical structure and chemical relatives of digoxin

Digoxin is a glycoside product of Digitalis sp., a figwort from the Scrophulariaceae family. It can really brighten up a vertical garden, and is variably known by gardeners as "foxglove",  "witch's gloves" or "dead man's bells", depending on how much of it they accidentally ingested. The foxglove plant, of course, has absolutely no interest in human welfare, and is not producing this substance for the pleasure of cardiologists. In fact, quite the opposite: it is supposed to be a deterrent. It appears that cardiac glycosides have no primary metabolic or structural function in plants, and in fact evolved convergently in numerous plants and animals as a defence mechanism against predation (Agrawal et al, 2011). Specifically, plants are being selected for higher and higher concentrations of cardiac glycosides by the activities of herbivores, at the same time as herbivores are evolving biochemical countermeasures to digoxin toxicity, in a slow arms race of predator-prey politics. Of course, this has absolutely no relevance to exams, and in fact the college comments for Question 10 from the second paper of 2013 chided candidates for wasting time on this pointless garbage.

Chemically, digoxin should be more properly called a "glycosylated steroid like drug". Its molecule consists of three sugars chained to a steroid ring structure. To be more precise, the sugar here is digitoxose, and the steroid belongs to a class known as cardenolides.

Digitoxose is a monosaccharide that mainly glycosylates the cardiac glycosides of Digitalis lanata,  the plant from which Lanoxin derives its lan. It is a sugar, which presumably means that it would be sweet to the tongue, and it appears to have some interesting anticancer properties. The cardenolide steroids are generally never seen in nature without their sugar tail, and aglucone (sugarless) versions are basically unknown, making it difficult to discuss their effects. 

The number of cardiac glycosides defies account. There are hundreds of recognised relatives, and probably thousands of molecular variations in the plant and animal kingdom which we are yet to catalogue. It appears to be pure accident that we landed on digoxin as the plant toxin of choice for poisoning cardiology patients. Apparently, Doctor William Withering discovered it among the ingredients of a folk remedy for dropsy, being peddled by some random gypsy from Shropshire (in his 1776 treatise, she gets absolutely no credit). Were events ever so slightly different, we might be feeding our patients some kind of "bufoxin", based on cane toad venom, or "coleopteroxin", derived from the poison glands of the Chrysolina beetle. The argument that digoxin is somehow optimal among naturally occurring cardioglycosides because it is the safest among them fails in confrontation with its known narrow therapeutic index and life-threatening toxicity. Inspecting carefully the rationale for its ongoing use, one arrives at the conclusion that it remains popular only by the inertia of complacent familiarity. Even though many of the other cardiac glycosides are pharmacologically quite similar to digoxin, none of them have the advantage of its long history, and would never pass modern safety standards on their own merits. In fact, even digoxin gets its name as a contraction of digitoxin, which was probably necessary from a marketing point of view (even though these two substances are distinct)

So, those are the chemical relatives of digoxin. Its pharmacological relatives (i.e. drugs which share its mechanism of action) are also few. Digoxin inhibits Na⁺/K⁺ ATPase, which is a pretty unique effect, and the only other drugs which do this are limited to experimental or laboratory use, i.e. they are sprinkled over cell cultures in order to inhibit that ATPase in order to see what happens in the absence of its activity. They include ouabain and strophantin, which are also plant-derived, as well as the inorganic ions of vanadium and the antifungal drug oligomycin. Beyond laboratory use, non-cardioglycoside inhibitors of Na⁺/K⁺ ATPase include the toxic fungal metabolites patulin and rubratoxin B, delta endotoxin from Bacillus thuringiensis,  the maitotoxin sequestered by tropical eels, palytoxin from corals and ptilodene from red sea algae (MacGregor & Walker, 1993). None of these agents has any clear pharmaceutical role outside of homicide.

Pharmacokinetics of digoxin

The ADME of digoxin forms a major part of answering a pharmacology question about it, and will probably be the source of a large proportion of marks. Ironically, the pharmacokinetics of digoxin seemed so boring that devoting a massive swath of text to them seems wasteful to the author, even as he squanders the reader's attention on discussions of toad venom. 

Routes of administration

Digoxin has reasonable oral bioavailability, around 70-80% (Marcus et al, 1976). The incomplete absorption is due to the effect of P-glycoprotein, an enterocyte efflux pump that ejects absorbed digoxin molecules back into gut lumen. The intravenous solution is usually presented with ethyl alcohol and propylene glycol (40% propylene glycol, 10% ethyl alcohol and 50% distilled water) but these are usually not a problem because the small (2ml) ampoule is diluted for intravenous administration.  Apparently, at least theoretically the IV formulation can also be given as an intramuscular dose, but in practice the excipients make this extremely painful. That hasn't ever stopped human experiments (Doherty et al, 1965), and if you did decide to administer it that way, you should expect a slightly longer half-life (38 vs 33 hours).

Distribution

Digoxin is basically insoluble in water (which is why so much alcohol is in the IV formulation). It is an acidic drug with a pKa of 7.15,  and is usually 25% protein bound. After administration it distributes widely, and is able to cross the blood-brain barrier; the volume of distribution is said to be high (5.1–7.4 L/kg). As the result, a loading dose needs to be given, in order for a therapeutic steady-state concentration to be achieved in the body fluids within a reasonable timeframe. Moreover, the poor water solubility and rapid distribution out of the circulation make digoxin entirely unsuitable for removal by dialysis.

Metabolism, clearance and half-life

There is very little hepatic metabolism of digoxin; only about 16% is broken down into various metabolites (Hinderling & Hartmann, 1991). The rest is excreted in the kidneys. As distribution of digoxin into tissues is very rapid, little remains in the circulation (where it would be available to the kidneys) and so the elimination half-life is relatively slow. Different textbooks give slightly different numbers, and the trainees are advised to pick one which sounds plausible. For a reference, Currie et al (2011) give 36-44 hours. This dependence on renal clearance produces accumulation-related toxicity in people with impaired renal function. 

Pharmacodynamics of digoxin

In short, digoxin inhibits the activity of Na⁺/K⁺ ATPase. This is a fundamentally important cell membrane ion transporter which is so vital to the function of eukaryotic cells that we multicellular organisms have not changed its basic form or function since our radially symmetrical days. From basic principles it is rather difficult to figure out how disabling this critical molecular engine could possibly be beneficial in any way. 

Mechanism and magnitude of inotropic effect

Digoxin is often said to be a positive inotrope. This mechanism takes a few steps to explain.

• As elaborated in the chapter on the normal processes of cardiac excitation, sodium is exchanged for calcium by the Na⁺/Ca²⁺ exchanger (INCX) during Phase 1 of the cardiac action potential.
• Ergo, anything that increases the availability of intracellular sodium should also increase the subsequent calcium entry.
• Increased calcium availability in the cardiac myocyte is the basis for increased cardiac contractility
• Sodium is generally pumped out of the cell by the activity of Na⁺/K⁺ ATPase, which means inhibiting it will increase the sodium availability, subsequent calcium entry, and therefore inotropy.

How potent is this effect? Not as potent as the "real" inotropes, but clearly palpable nonetheless. Cohn et al, in 1969, infused a whole range of cardiac glycosides into patients with cardiogenic shock and found that systolic blood pressure increased almost immediately following IV infusion of the drug, by about 13 mmHg on average.  As one can plainly see, it did not have much positive impact on outcomes, but this was the 1960s and cardiogenic shock management was somewhat rudimentary.

Cardiac output also clearly improves with digoxin, but it is not clear by how much it increases, or in whom. A slightly later study with slightly less critical patients by Marchionni et al (1985) found that the cardiac index increased by about 20%, but not in every group of patients. Gheorghiade et al (1987) looked at patients with heart failure and also found that there were "responders" and "non-responders", with the former enjoying an impressive increase in their cardiac index (from 2.1 to 3.1). 

Mechanism of antiarrhythmic effect

Digoxin is often listed as an antiarrhythmic, usually grouped with "misc" agents as it defies standard Vaughan Williams classification. Other authors have tried to shoehorn it into various categories (Class V, etc), with limited acceptance in the scientific community. 

In short, digoxin acts as an AV node blocker. Functionally, it resembles the effect of beta-blockers  calcium channel blockers and amiodarone, in the sense that it delays the conduction of atrial impulses to the ventricles, which can control the ventricular rate during AF. Przybyla et al (1974) determined that it accomplishes this by prolonging the refractory period of the AV node, which manifested as a slowed conduction time (from 120 msec to 270 msec following 750 mcg of digoxin). Conduction through the Bundle of His remained unchanged, which means your QRS complexes will remain nice and narrow.

This appears to be a vagally mediated phenomenon, in the sense that it is completely abolished by atropine. Rossen et al (1975) continued this line of thought by experimenting on patients with denervated hearts (i.e. following cardiac transplant) and demonstrating that digoxin had basically zero effect on AV nodal conduction in these people, a finding they attributed to the severed vagus. 

A large number of physiological mechanisms have been proposed to explain this vagotonic effect of digoxin, which suggests that it is probably poorly understood. Watanabe et al (1985) describe a series of plausible-sounding mechanisms:

• Activation of baroreceptors: digoxin appears to directly stimulate baroreceptors. This was demonstrated with stark simplicity by Quest & Gillis (1971), who injected ouabain directly into the carotid sinuses of cats, and observed that this manoeuvre had the tendency to "greatly augment the spontaneous electrical firing of the carotid sinus nerve"
• Sensitisation of parasympathetic ganglia (i.e. increased effects from the same amount of released acetylcholine)
• Sensitisation of the myocardium to acetylcholine

The net effect of digoxin, therefore, should be similar to the effects of performing a vagal manoeuvre. However, in addition to these vagotonic effects, it does have a measurable effect on the action potential. 

Effect of cardiac glycosides on the cardiac action potential

The effects of digoxin on the cardiac action potential are illustrated by this figure in the authoritative paper by Worthley & Holt (1999). 

Worthley & Holt do not mention where this figure comes from, nor is it possible to easily dig up the  reference for their description of how digoxin affects the action potential, which is:

"By inhibiting conducting tissue NaK-ATPase, digoxin alters the action potential by, a) prolonging phase 3 and shortening phase 2 (i.e. shortens the QTc interval), b) increasing the slope of phase 4 (i.e. enhances the automaticity of atrial, junctional and ventricular tissue), and c) slows phase 0 (i.e. reduces the conduction velocity)."

The attentive reader will note that the above statement completely contradicts everything they know about digoxin. Is it not expected to slow the rate? How would that make any sense, if the automaticity of all these tissues was increased?  One might be tempted to disregard the opinion in this article,  if it were not for the fact that L.I.G Worthley has authored a vast number of CICM exam questions, and was an influential figure in the training program. For the trainee, it would be disastrous to be more correct than Worthley.

Rather than looking at stylised and unreferenced diagrams, we might ask - what does the digitalised action potential really look like? Katzung's Basic and Clinical Pharmacology (14th ed) gives probably the best account, supported by some unpublished action potential recordings from Hess & Wier. These recordings are excellent and are reproduced here with minimal modification:

Now, it's not mentioned exactly which tissue Hess & Wier were torturing with ouabain, but it looks like a Purkinje fibre, and it also looks like the recordings published by Ruch et al (2003). None of these sources specifically comment on the specific phases, but it is clear that the action potential duration is definitely shortened. In fact, with larger doses, Phases 2 and 3 disappear entirely (Woodbury & Hecht, 1952). This latter article is perhaps the most informative for the purpose of answering this question, as it clearly spells out the investigators' experience with action potential phases in frog cardiac myocytes.

So, in summary,  what do we write in an answer to the very reasonable exam question, "describe the effects of digoxin on the action potential"?

• The whole action potential duration decreases (Ruch et al 2003)
• Phase 0 can be either shortened or lengthened (Hordof et al, 1978); most textbooks seem to think it is lengthened.
• Phase 1 is shortened (Woodbury & Hecht, 1952).
• Phase 2 is shortened the most (Woodbury & Hecht, 1952).
• Phase 3 is lengthened (Woodbury & Hecht, 1952).
• Phase 4 slope is increased in pacemaker tissues (Hordof et al, 1978) and in Purkinje fibres (Rosen et al, 1975), increasing their automaticity.

Finally, the question "how come digoxin doesn't cause tachycardia if it increases automaticity" can be answered by combining its electrophysiological and vagal effects. Or, at least, that appears to be the upshot of everything that has been written in the aforelisted references, though it is not spelled out clearly in any of them. Because of the vagotonic effects on the AV node and other conducting tissues, the increase in automaticity does not manifest clinically. Remember that the effects of the parasympathetic nervous system are to depress the rate of Phase 4 self-depolarisation in nodal tissue. In other words, sure digoxin increases the automaticity of various random pacemaker cells, but at the same time its vagal effects slow the firing rate of these pacemakers and slow the conduction of action potentials, which leads to an overall decrease in heart rate.

Clinical use of digoxin

At risk of becoming too clinically relevant and trespassing beyond First Part Exam territory, it is possible to make some brief comments about the currently accepted indications for the use of digoxin. These were extracted from the excellent paper by van der Meer (2019), comparing the contemporary AHA and ESC guidelines for the management of heart failure. 

• Second line agent for AF: Digoxin is indicated for the management of AF in patients with heart failure as a second-line agent, i.e. added to a beta-blocker (ACC). In Europe, digoxin would be first-line for severely volume-overloaded or haemodynamically unstable patients, but in combination with amiodarone.
• A possible addition for heart failure: Digoxin is "may be considered" in symptomatic patients in sinus rhythm to reduce all-cause and heart failure hospitalisations, but it will have no effect on mortality.

Utility of digoxin in the ICU as an inotrope with rate control properties

Returning to venerated CICM examiners, Worthley & Holt (1999) have published an opinion on the ICU uses of digoxin which (from brief informal interactions with Australian intensivists) appears to enjoy releatively broad support in the professional community. That opinion is that digoxin is crap. Or, to rephrase it more scientifically, 

"...There are other agents available that have a superior inotropic effect, a greater ability to control (and reverse) supraventricular tachyarrhythmias, have a larger therapeutic window and are easier to regulate."

So, everything digoxin can do, something else can do better. Want more inotropy? Use a proper inotrope. Want rate or rhythm control? Choose from the vast armamentarium of classifiable Vaughan Williams agents. So, where does digoxin fit in? 

For acute control of rapid AF, digoxin again appears to be a second-tier therapy. Several examples are available. For instance, Shojaee et al (2017) ran a trial of digoxin vs amiodarone for controlling AF in a group of patients in whom first-line agents (beta-blockers and calcium channel blockers) were contraindicated. These were not exactly critically ill (in fact haemodynamic instability was an exclusion criterion). The results were strongly supportive of amiodarone; the digoxin group had about 60% rate of treatment failure (defined as persistent rapid AF)

Toxicology of digoxin

ECG findings

Characteristic ECG findings of digoxin are well-described and may appear attractive to examiners. ECG interpretation is not strictly speaking a part of the First Part, but it would be reasonable to expect a question that asks you to dumbly list them. In that case, there would be no better resource than the LITFL page on digoxin toxicity. Without restating what is already said well enough by Ed Burns, one may repeat the list of features here to simplify revision:

• Prolonged PR interval (due to AV node block by the vagal effects)
• Downsloping "sagging" ST-segment changes
• Flattened, inverted, or biphasic T waves.
• Prominent U waves
• Peaking of the terminal portion of the T waves.
• J point depression
• Shortened QT interval (due to the shortened repolarisation)

Drug interactions

Digoxin interacts with a surprisingly large number of drugs, for a substance which is not really dependent on clearance by metabolism. The reason for this is its reliance on active secretion at the renal tubule, and its tendency to be lost back into the gut lumen by the actions of P-glycoprotein, an efflux pump susceptible to being clogged by other competing substrates. From Brown et a (1980): 

• Pharmacokinetic interactions:
  • Decreased absorption: antacids
  • Increased absorption:
    • amiodarone, verapamil, rifampicin (by blocking the P-glycoprotein enterocyte efflux pump)
    • Tetracyclines, macrolides (by decreasing metabolism of digoxin by gut bacteria)
  • Decreased renal clearance: spironolactone, quinidine
• Pharmacodynamic interactions:
  • Diuretics: as they decrease magnesium and potassium, they predispose you to digoxin toxicity
  • Suxamethonium: by some unknown mechanism, it causes what Brown et al describe as "ventricular irritability", which might actually mean VF arrest.
  • Beta-blockers and calcium channel blockers: additive AV node blockade and bradycardia 

Adverse effects of digoxin

Digoxin has a large number of adverse effects, some of which are listed in this excellent table from Worthley & Holt:

It would, of course, be quite unexpected to find this level of detail in a CICM exam answer, and so the trainees should probably prepare a shorter list from the "not uncommon" column (otherwise known as "common"), as follows:

• Bigeminy and ectopics
• Bradycardia and heart block
• Nausea, vomiting
• Visual disturbances
• Delirium and agitation

Massive, ridiculous overdose

For a drug that causes nausea and bradycardia, it would be reasonable to expect that overdose might bring more nauseating nausea and slower bradycardia. That does happen, and the bradycardia apparently responds well to atropine; but also digoxin of course has to be different and surprising. Specifically, the shortened repolarisation period and increased automaticity till a fertile soil for nasty tachyarrhythmias. This can happen even at therapeutic levels, but the risk increases dramatically as the serum levels rise above 2.0 ng/ml (2.6 nmol/L). On top of that are the neurological effects (as this lipophilic drug suffuses every fatty gram of brain) and the unsurprisingly serious systemic consequences of disabling a critically important electrolyte pump.

• Heart block and asystole,  as the natural extension of rate control, is not unexpected or surprising for a drug that disables nodal tissues. As this is a phenomenon mediated by excessive vagal activity, anticholinergic drugs are the logical solution.
• Tachyarrhythmias: "Almost any arrhythmias can occur", the literature specifies helpfully. Fortunately, the excess vagotonic effects in overdose should be disabling your AV node even more than usual, which means you should be safe from atrial arrhythmias. Unfortunately, that leaves ventricular ones, which are more rapidly fatal. Some hidden interplay of vagally increased refractory period and electrophysiologically reduced refractory period creates a balance that can easily tip into a territory where ventricular myocytes take it upon themselves to generate spontaneous action potentials and annoy their neighbours with delayed afterdepolarisations. This manifests initially as copious ventricular ectopic complexes and bigeminy, and then degenerates into VT or VF. The VT can be a "bidirectional ventricular tachycardia" in which every second beat has an opposite axis.
• Vasoconstriction: As digoxin can increase intracellular calcium levels in myocytes, it can do the same in smooth muscle, and promote vasoconstriction. The normal sympathetic response to a decreased cardiac output also helps. Realistically, the positive effect of this on the arterial blood pressure may be completely lost in the cardiovascular collapse associated with severe overdose.
• CNS toxicity: Digoxin can produce a decreased level of consciousness, and toxicity can present as encephalopathy (Greenaway et al, 1996). The authors helpfully produce this box listing some other toxic CNS effects:
  
  The crippling nausea appears to be a direct effect of the drug on the medullary emetic chemoreceptor trigger zone (Gaitonde et al, 1965), and is completely abolished by the selective ablation of this medullary region, though most people would just probably give some antiemetics.
• Hyperkalemia: as Na⁺/K⁺ ATPase mediates cellular sodium efflux and potassium influx, it stands to reason that blocking this pump may lead to some extra potassium hanging around in the extracellular fluid. 

Prior to the introduction of digoxin antibodies, the mortality from severe digoxin overdose approached 20%. Interestingly, children appear to be resistant to the worst of it, and truly absurd serum levels (48 ng/ml) have been measured in relatively intact survivors who did not require much treatment. Adults, by comparison, do poorly. We are fortunate that the use of this drug has been declining in recent years, and truly massive overdose of that magnitude is now quite rare. Historical accounts are available of the heroic measures taken in vain by rescuers trying to manage these people in the pre-ECMO era, such as this horrifying narrative by Nicholls (1977): 

...She quickly developed ventricular tachycardia and fibrillation despite lignocaine 100 mg
intravenously. DC defibrillation produced asystole, and there was no response to isoprenaline 100 μg or potassium chloride 25 mmol intravenously. The heart was exposed via a left-sided thoracotomy, but there was no response to internal cardiac massage or to suturing the pacing wires directly to the myocardium. Death was presumed some 2 hr after admission."