Br Heart J (Supplement) 1994; 72: 92-99

Inotropic agents for heart failure: what if digoxinincreases mortality?

W J Remme

In recent years the approach to heart failurehas changed from one that focused on thehaemodynamic changes at the basis of thesyndrome or occurring as a result of it to onethat recognises the significance of neuro-endocrine activation not only for theoccurrence of heart failure but also for itsclinical expression and ultimate prognosis.Moreover, the fact that heart failure dependsnot only on cardiac changes but to a largeextent on intrinsic peripheral changes has beengradually appreciated.

Consequently, the management of heartfailure has taken a different turn, focusing nowon modulation of neuroendocrine activationand other extracardiac abnormalities ratherthan trying to increase cardiac contractileforce through positive inotropic treatment.This attitude has been reinforced by currentdoubts about the clinical efficacy and safety ofpositive inotropic treatment. As other physio-logically more relevant treatments such asinhibition of angiotension converting enzymehave become available, it is not so much aquestion of when positive inotropic agentsshould be administered but whether there is infact a place and need for this form oftreatment.

In most patients, however, cardiaccontractile dysfunction is the primary disorderand the origin of the heart failure, sointuitively, pharmacological interventionsaimed at inotropic support have beenconsidered important in treating heart failurefor many years. Consequently, many inotropicagents have been developed. Positive inotropicdrugs are categorised as cyclic AMP (cAMP)dependent or cAMP independent.

SticaresCardiovascularResearch Foundation,Rotterdam, TheNetherlandsW J RemmeCorrespondence to:Dr W J Remme,Sticares Foundation, POBox 52006, 3007 LARotterdam, TheNetherlands.

Cardiac contractile force andcontractility: how to manipulate it?cAMP DEPENDENT MECHANISMS

Whether they act through cAMP or not,inotropic drugs increase cardiac contractileforce by altering the availability of calcium tothe contractile proteins and by altering theinteraction of calcium with the troponin-tropomyosin complex. In addition, cAMPdependent agents influence the velocity ofcontractile protein interaction and thesequestration of calcium from the cytosol aftercontraction.cAMP formation depends on the activity of

the adenylate cyclase system, which, in turn,depends on activation of stimulatoryadrenergic versus inhibitory muscarinic or

adenosine receptors and on the activity ofstimulatory versus inhibitory regulating

proteins, the G proteins (fig 1). Once formed,cAMP is degraded to 5-AMP under theinfluence of soluble or membrane boundcGMP inhibited, cAMP specific phospho-diesterases of subtype III.

In the heart an increase of this secondmessenger leads to cAMP dependent phos-phorylation of certain protein kinases, whichaffect the contraction-relaxation process inseveral ways. Firstly, the slow inward calciumcurrent is increased through voltagedependent L type channels in the sarcolemma.This calcium is instrumental in the subsequentrelease of larger amounts of calcium from thesarcoplasmic reticulum into the cytosol at thecontractile protein site. The degree of calciuminduced calcium release determines the forceof contraction.

Secondly, the concentration of cAMP ispivotal for myofibrillar ATPase activity andcrossbridging cycling rate, and, hence, forcontractility.

Finally, cAMP influences relaxation invarious ways. Phosphorylation of phospho-lamban enhances calcium reuptake by thesarcoplasmic reticulum, whereas phosphory-lation of troponin I reduces the calciumsensitivity of troponin C, both mechanismsenhancing cardiac relaxation.

cAMP INDEPENDENT MECHANISMSThere are several additional ways to interferewith cardiac contraction without affectingcAMP concentrations. Again, in these cAMPindependent mechanisms calcium plays acentral part. Experimental conditions whichincrease the intracellular calcium concen-tration directly or indirectly by introducingpaired stimulation enhance contractile force inparallel with the change in inward calciumcurrent.' For instance, cx adrenergic receptorstimulation increases inositol 1,4,5-tri-phosphate (IP3) concentrations, subsequentlyenhancing the inward calcium current andcontractile force, whereas sarcolemmalcalcium agonists may increase force by directlyopening the calcium channels-for example,Bay K 8644.2 However, potentially morerelevant mechanisms in terms of pharmaco-logical modulation of intracellular calciumrelate to either sarcolemmal sodium channelstimulation or to sarcolemmal sodium-potassium ATPase inhibition (digitalisglycosides), the latter indirectly raising theconcentration of intracellular calcium ionsthrough increased sodium influx withsubsequent sodium-calcium exchange.A different approach towards increasing

contractility is not to increase the

S 92

Inotropic agents for heart failure: what if digoxin increases mortality?

Figure 1 Schematic representation of cardiac cyclic AMP (cAMP) formation and itsinteraction with contraction and relaxation. cAMP, formed after activation of the ,3receptor(Rs) -Gs-adenylate cyclase axis, stimulates several protein kinases, subsequentlyenhancing sarcolemmal calcium fluxes, which release calcium from intracellular stores,such as the sarcoplasmatic reticulum. Binding of calcium with the troponin-tropomyosincomplex (Ti, Tc, Tm, Tt) leads to actin-myosin interaction and contraction. Subsequentsequestration of cytosolic calcium into the sarcoplasmic reticulum and relaxation isenhanced by cAMP induced phosphorylation of protein kinases that affect the activity ofphospholamban and Ti. cAMP is degraded to 5-AMP under the influence of cytosolicand particulate cyclic nucleotide phosphodiesterases. (M, muscarinic; P, purinergic; Ri,inhibitory receptor; Gi, inhibitory G protein.)

concentration of calcium but to enhance thesensitivity of the contractile apparatus-thatis, of troponin C-for the available calcium.Cardiac contraction is governed by threemechanisms: increase in intracellular calciumconcentration, the binding of calcium totroponin C, and subsequent alteration ofmolecular conformation of the myofibrils.Calcium occupancy of troponin C isdetermined by the ratio of the rates ofassociation and dissociation of calcium. As itis the latter which is affected by manoeuvres

that improve calcium sensitivity, it followsthat, as a consequence, relaxation may slowdown.3Calcium sensitisation is potentially an

interesting approach to positive inotropicsupport in heart failure. Apparently, it is moreeconomical and less energy consuming toincrease contractile force than increase theintracellular concentration of calcium,particularly by cAMP dependent mechan-isms-for example, with agents such as iso-prenaline and the predominant phospho-diesterase inhibitors, such as enoximone.' Thereasons are twofold. Firstly, the energyrequired for calcium recycling increases withhigher inward calcium currents. Secondly,myofibrillar ATPase activity is amplified withcAMP dependent inotropic mechanisms. Bycontrast, increasing cardiac force by enhancedcalcium sensitivity or enhanced maximumcalcium activated force avoids this extra

myocardial energy consumption.Moreover, calcium sensitisation may be

preserved in progressive heart failure,5 despitethe fact that ionic conditions such as acidosisand increased phosphate concentrations,6

prevalent in ischaemia and possibly also inischaemic cardiomyopathy and end stage heartfailure, negatively affect calcium sensitivity. Incontrast, cAMP dependent contractile effectsbecome progressively less during the laterstages of heart failure.7 8

Intrinsic cardiac changes during heartfailure that affect pharmacologicalinotropic interventionsSecondary to the enhanced sympatheticstimulation in heart failure, sarcolemmal 13receptor density decreases significantly as aresult of internalisation of the receptor and ofuncoupling.9 Consequently, adrenergic stimu-lation results in progressively less adenylatecyclase activity and cAMP formation.Moreover, there is enhanced expression of theinhibitor subunit of the guanine-nucleotide-binding regulatory proteins, Gj.,"1 1" whichreduces cAMP formation. This increase in thelevel of Gio is unrelated to muscarinic M2 or A,adenosine receptor activity.12 Furthermore,stimulation of adenylate cyclase by colforsinmay be reduced in heart failure.'3Consequently, not only is basal cAMPformation decreased in heart failure but alsothe potential to stimulate it pharmacologicallyis diminished.

Initially, a specific decrease in the PIsubtype was assumed, leading to a 50-70%downregulation of 1 receptors.'4 By contrast,12 receptors were not believed to be down-regulated, suggesting a potential therapeuticrole. Recent studies by Bristow et al indicate,however, that moderate (25-35%) uncouplingof 12 receptors occurs in heart failure,'5 albeitpredominantly in patients with ischaemiccardiomyopathy. These authors suggested thatthe regulatory behaviour of components of the1 receptor G protein-adenylate cyclase axisdiffered between idiopathic dilated andischaemic cardiomyopathy, with 13 receptordownregulation in the first and uncoupling of12 receptors in the latter patient group.9Moreover, the increase in Gia: may occur onlyin idiopathic cardiomyopathy.'6

Consequently, changes linked to the 1adrenergic receptor in heart failure haveimportant effects on cAMP formation in theheart. Deficient cAMP production un-doubtedly is an important cause of contractiledysfunction in heart failure and of the lack oftherapeutic efficacy of cAMP dependentpositive inotropic support.

Moreover, the high adrenergic drive duringheart failure which leads to these receptorchanges also causes marked depletion ofneurotransmitters in the heart. Both inidiopathic dilated cardiomyopathy and inischaemic cardiomyopathy, tissue noradrena-line, adrenaline, dopamine, and neuro-peptide Y concentrations are decreased.9Consequently, indirect acting 1 agonists, suchas dopexamine and dopamine, lose theirefficacy over time.By contrast, cAMP independent inotropic

mechanisms are possibly less affected byintrinsic cardiac changes. For instance,

S 93

S 94

myocardial ot adrenergic receptor density isunchanged in heart failure.'7 18 Although thiscould be important because cx adrenergicstimulation elicits a positive inotropic responsein humans, a diminished inotropic responsehas, nevertheless, been observed in heartfailure, either as a result of receptoruncoupling or secondary to a general defect inthe contractile potential in this condition.'9 Bycontrast, calcium sensitisation remains intacteven in advanced stages of failure, and there isno evidence for a diminished efficacy ofdigitalis in advanced heart failure, at least notin in vitro experiments. Similarly, there are nodata to indicate that sodium channelstimulation or other cAMP independent waysto increase intracellular calcium concen-trations lose their inotropic efficacy in endstage heart failure.

Clinical efficacy of cAMP dependentinotropic agentsSYMPATHOMIMETIC AGENTS (TABLE)Sympathomimetic agents are undoubtedly oneof the cornerstones of the treatment of acuteheart failure, particularly in low outputconditions and in cardiogenic shock.Parenterally administered 13 and 12 adrenergicagents, dopamine, and, possibly, colforsinhave clearly shown acute haemodynamic andclinical efficacy.The effect is, however, comparatively short

lived. Tolerance may occur after days,probably due to reduced 13 receptorresponsiveness secondary to further receptordownregulation, an increase in Gin,s andpossibly also inactivation of the adenylatecyclase system. Consequently, the capacity ofthe heart to produce cAMP diminishes.Although early studies suggested a sustainedclinical improvement lasting for several weeksafter a three day infusion of dobutamine, suchimprovement was observed only in a minorityof patients and then specifically in those witha fairly well preserved function at baseline.Subsequent attempts to obviate long term

Clinically evaluated positive inotropic agents

Sympathomimetic agents Phosphodiesterase inhibitors Agents affecting membranepumps or channels

Non-selective: Predominant phosphodiesterase Sodium channelNoradrenaline (a and ,3) inhibitors: stimulants:Adrenaline (a and (3) Amrinone DPI 201-106 (partialIsoprenaline (1B) Milrinone calcium sensitiser)

Enoximone PD 122 860Predominant (, agonists: PiroximoneDobutamine Adibendan VesnarinonePrenalterol (partial agonist) Imazodan (partial phospho-Xamoterol (partial agonist) diesterase inhibitor;

Partial phosphodiesterase affects outward andPredominant 2 agonists: inhibitors/calcium sensitisers: inward rectifyingDopexamine (partial Sulmazole potassium current)

dopaminergic) PimobendanSalbutamol Isomazole Sodium-potassiumPirbuterol UK-61 260 ATPase inhibitors:Terbutaline Org 30029 Digitalis glycosides

MCI-154Dopamine and (, agonists:DopamineLevodopaIbopamine*

Adenylate cyclase stimulant:Colforsin

*Possibly only dopaminergic at clinically relevant dosages (see text).

tolerance by intermittent dobutamineadministration provided some clinical efficacy,as measured by exercise capacity and clinicalimprovement, although baseline ventricularfunction remained unchanged.20 Again, theimprovement was observed in patients withless severe heart failure (New York HeartAssociation class III). In contrast, in patientswith more severe failure, for whom suchtreatment would be more relevant, clinicalimprovement has not been impressive andmortality may be increased despite initialhaemodynamic improvement.

Orally active sympathomimetic agents havenot proved to be successful either. The P13agonist prenalterol, although resulting ininitial haemodynamic and clinical improve-ment, did not lead to sustained benefit, and itsuse has long been discarded. Also, long termresults with predominantly 12 agonists, such aspirbuterol, terbutaline, and salbutamol, whichin addition have some 13 agonist properties,have not been favourable. Whereas earlyhaemodynamic improvement was commonlypresent, this did not persist and long termclinical efficacy was inconsistent.2122 Higherdoses to improve efficacy have caused sideeffects, such as tremulousness or seriousventricular arrhythmias.23 As a result, orallyactive 12 agonists are currently not used totreat heart failure.

In contrast, the partial P13 agonist andantagonist xamoterol has, for some time atleast, been considered a useful addition to thetherapeutic armamentarium in mild heartfailure. Xamoterol probably does have avaluable role as an alternative to diuretics or inpatients with heart failure and atrialfibrillation. It may have a special role in elderlypatients with heart failure. Xamoterol mayalso be an excellent choice in patients withheart failure and angina. Indeed, ameliorationof ischaemic left ventricular dysfunction ratherthan heart failure itself may be its principalmode of action. A controlled study in severeheart failure failed to show clinicalimprovement but showed that xamoterolincreased mortality.24 Preliminary reports inpatients with milder heart failure suggest aneutral or slightly favourable effect onsurvival.

DOPAMINERGIC AGENTS

In acute heart failure dopamine is extensivelyused in low output conditions to improve renalflow at lower doses and to improve cardiacoutput and blood pressure at higher doses.This reflects the pharmacological profile of thedrug, which includes stimulation ofdopaminergic 1 and 2 receptors and, at higherdosages, activation of 13, a,, and a2 receptors.However, the major inotropic mechanism ofaction seems to be through cardiac adrenergicneurotransmittor release.25 As noradrenalinestores become depleted in the course of failurethe positive inotropic effect of dopaminediminishes.The orally active dopaminergic agents,

levodopa and ibopamine, have a similar profileand have positive inotropic effects as well as

Remme

Inotropic agents for heart failure: what if digoxin increases mortality?

1200 *

¢10-00.

°8-006.00-

L0

0 04*002.00

C 120 *iX 100

80,60-

0 4000220

480400 A320 o240

160

80

C 2400 mFu 2000R

1800.

160

>.1400.1200

co

> 0 110 20

Milrinone(gglminl

100 ml forearm

volume)

Figure 2 Regionalvasodilator effect of locallyadministered milrinone.

Forearm blood flowincreases and forearmvascular resistancedecreases parallel toincremental dosages ofmilrinone, which result indetectable milrinoneconcentrations in forearmvenous blood ('9.j butundetectable circulatingconcentrations ( ). 2Reproduced withpermission.

vasodilator actions on renal, cerebral,coronary, and visceral beds (postsynapticdopaminergic-I action), and modulatingeffects on noradrenaline release from nerveendings (presynaptic dopaminergic-2 action).Evidence in humans that these agents affectcontractility under standard pharmacologicalconcentrations is scarce. Recent data suggestthat ibopamine, the only orally activedopaminergic agent available for clinical use,

may be a (renal) vasodilator rather than a

positive inotropic agent.26

cAMP PHOSPHODIESTERASE INHIBITORS

By contrast to the lack of inotropic effects oforally active dopaminergic agents, cAMPphosphodiesterase (PDE) inhibitors clearlyimprove cardiac contractility in humans,shown when agents such as milrinone are

introduced into the coronary circulation.27PDE inhibitors are also potent vasodilators, as

may be observed after intra-arterialadministration in heart failure (fig 2).28Consequently, PDE inhibitors are betterclassified as inodilators than positive inotropicagents.29 Venodilating properties may pre-

dominate with these compounds, at least atlower doses,30 31 and, in patients with normalto low ventricular filling pressures as a result ofdiuretic treatment, PDE inhibition may

reduce cardiac output by reducing fillingpressures still further, an effect whichresembles that of nitrates under similarconditions (fig 3).32

In addition, tolerance to their vasodilatingproperties may occur. After long termtreatment with milrinone in patients withmoderate heart failure, the initial improve-ment in pump function disappears, whereasafter withdrawal, cardiac output decreases andsystemic resistance increases beyond pretreat-ment values.33 This rebound phenomenonmay reflect neurohormonal activation duringmilrinone treatment.Whether or not related to haemodynamic

tolerance, there is ample evidence that clinicalefficacy with predominant PDE inhibitors isnot sustained. In placebo controlled studies

OAx 3.0-

1 *0

CO~~~~~~~'

1.0~

lo 20 30

Pulmonary capillary wedge pressure(mm Hg)

Figure 3 Effect of oral milrinone (10 mg) on cardiac

pump function in nine patients with heart failure andraised left ventricular filling pressures (PCWP) (U;PCWP 3 18 mm Hg) and nine patients with heart failureand normal left ventricular filling pressures (0; PCWP<18 mm Hg. (Heart failure was class II-III according toclassification ofNew York Heart Association.) Cardiacindex improves in the first group but falls in the second.32Reproduced with permission.

with agents such as amrinone and enoximonesignificant improvements in exercise capacityor clinical wellbeing have not been sus-tained.34 3 One study that compared milrinone,digoxin, their combination, and placebo inpatients with class III heart failure (New YorkHeart Association's classification) suggested animprovement with the PDE inhibitor.36 How-ever, in this study, all patients were initiallytaking digoxin, which was withdrawn beforerandomisation. Although patients taking milri-none had better exercise capacity than thepatients who received placebo, when they werecompared with patients who continued to takedigoxin there was no such effect. Rather thansupporting the efficacy of milrinone, this studywould favour the use of digoxin in moderate tosevere heart failure.

Clinical efficacy of cAMP independentinotropesCALCIUM SENSITISERS

By contrast to predominant PDE inhibitors,late haemodynamic tolerance may not be aproblem with the few available calciumsensitisers, which all share some PDEinhibiting effects (table). Pimobendan, acalcium sensitiser and partial PDE inhibitorand the most extensively studied compound inthis category, retains its vasodilating capacityand haemodynamic efficacy for at least sixmonths in patients with moderate heartfailure.37 This indicates that long term haemo-dynamic effects may differ between pre-dominant PDE inhibitors and compoundswhich share calcium sensitising and PDEinhibiting properties. Also, long term clinicalefficacy may differ. Available controlledstudies indicate an improvement in exercisecapacity, peak oxygen consumption, and heartfailure status after treatment lasting betweenthree and six months.3SAO Whereas this mayalready be true for fairly high doses of pimo-bendan, it is interesting to note that clinicalefficacy is better with low to intermediatedoses.39 Whether this reflects relatively lessPDE inhibition at lower doses is as yet purelyspeculative.

Theoretically, calcium sensitisation mayfurther slow the process of myocardialrelaxation, which is already impaired in heartfailure. In myopathic tissue the capacity torestore normal concentrations of cytosoliccalcium during diastole is diminished. Conse-quently, both resting intracellular calciumconcentrations and diastolic developed forceare higher than in normal cardiac muscle. Asthis negatively affects contractile force, "pure"calcium sensitisers may further decreasecontractile reserve in addition to their effect onrelaxation.3 This is, however, less likely for thecalcium sensitisers currently available, whichall have additional PDE inhibiting properties.Pimobendan, administered intravenously topatients with heart failure (New York HeartAssociation class II-III) and ischaemic cardio-myopathy, impairs neither ventricularrelaxation nor contractility (fig 4). Instead,isovolumetric relaxation indices improve,

S 95

Remme

Contractility.> _

0 1600 * *

CL EEE1400- I

E ~~~~~~~~~~~~~~~~~~~~~~~Co

30- * 0

0(0~~~~~~~~~~~(D 25-I70-I Relaxation 0

l-E 60- L T

50 0C-

50-] * * * * *

_ c 40- 1 1*1EPimobendan30-' V

C 10 20 30 40 50 60Time (minutes)

Figure 4 Acute, but sustained improvement ofisovolumetric indices of contractility (LV dP/dt positiveand Vce40) and relaxation (T, and T2) after 10 mgintravenous pimobendan in patients with ischaemiccardiomyopathy and moderate heart failure. During themeasurements heart rate was fixed by way of atrialpacing. (T, tau (relaxation time constant).)

which, together with its acute vasodilatingeffects under these conditions, may representthe additional PDE inhibiting properties ofthis compound.4' Myocardial energetics arealso favourably affected with a decrease inmyocardial oxygen consumption. Thiscontrasts with predominant cAMP dependentinterventions-that is, xamoterol andenoximone-which under similar conditionsincrease oxygen consumption or leave itunaffected.42 43

SODIUM CHANNEL STIMULANTS

The few available sodium channel stimulants(table) have additional vasodilating properties,although they primarily induce positiveinotropic effects through sodium channelstimulation and subsequent sodium-calciumexchange.44 45 Like the available calciumsensitisers, this suggests that these agentsshould be considered inodilators rather thanpure positive inotropes.The most intensively studied sodium

channel stimulant in humans is OPC 8212(vesnarinone); it was recently shown toenhance survival in patients with severe heartfailure when given in low doses.46 Thisinteresting and unexpected finding for apositive inotrope may relate to the additionalpharmacological properties of this rathercomplex molecule. Besides sodium channelstimulating effects and mild, selective PDEinhibition, vesnarinone decreases the delayedoutward and inward rectifying potassiumcurrents. The latter are probably involved inthe antitachycardiac effects of vesnarinone,observed after sympathetic stimulation, and inthe prolongation of the action potential.47Whether this implies that vesnarinone shouldbe considered a class III antiarrhythmic agent

* Control A Denervated stripso Propranolol * Prazosin

0 10 20 30 40 50 60 70Time (minutes)

Figure 5 Biphasic pattern of constriction after ouabainhas been added to coronary vascular strips. The firstcontraction is the direct effect of the drug and depends onavailable calcium. The second is absent in denervatedstrips and reduced after a blockade with prazosin,indicating noradrenaline release by digitalis.49 Reproducedwith permission.

rather than an inotrope, and, consequently,the effects on mortality the result of an anti-arrhythmic effect, is uncertain. Firstly,Feldman et al found that the reduction inmortality applied not only to patients dyingsuddenly but also to patients dying ofprogressive failure.46 Moreover, vesnarinonereduced the risk of worsening heart failure andimproved quality of life. Secondly, availableantiarrhythmic agents have thus far notappreciably affected mortality in progressiveheart failure. The reasons for the unexpectedbeneficial effect of vesnarinone on survival andheart failure status are not clear, but they mayinclude the combination of cAMP indepen-dent inotropic effects, antiarrhythmicproperties, a bradycardiac effect, and,possibly, an inhibiting action on cytokineproduction.48There is some concern, though.

Vesnarinone induces agranulocytosis in 2-5%of patients. Moreover, as will be discussedbelow, it probably has a narrow therapeuticrange, like digitalis.46

DIGITALIS GLYCOSIDES

Digitalis glycosides inhibit the activity ofmembrane sodium-potassium ATPase. Thisresults in raised intracellular sodium, which isexchanged for calcium through the sodium-calcium exchanger. In addition, the rise inintracellular calcium concentration leads to anincrease in hydrogen ions, which areexchanged for sodium, in turn leading to moreintracellular calcium. This results in enhancedcardiac contractility and vasoconstriction.

In vitro, the vasocontrictor effects ofdigitalis are biphasic.49 Firstly, there is acontraction which depends on availablecalcium and reflects the direct constrictoreffect of the drug. This is followed by a secondvasoconstriction, which is lost in denervatedvascular strips and reduced by co-administration of the ot, blocker prazosin,probably by digitalis induced release ofnoradrenaline from nerve endings (fig 5). Invivo these direct vasoconstrictor effects of

S 96

Inotropic agents for heart failure: what if digoxin increases mortality?

digitalis are augmented through centralstimulation of sympathetic tone.50

In subjects with normal cardiac functiondigitalis has a vasoconstrictor effect, althoughthis depends on dose and rate ofadministration. Intravenous administration ofouabain for 10 seconds or two minutes causesa significant increase in systemic vascularresistance but not when a slower infusion rateis applied.5' In subjects without heart failurean intravenous bolus administration of digoxinresults in coronary vasoconstriction with asignificant reduction in the cross sectional areaof both normal segments and arteries withcoronary artery disease.52

In heart failure the acute vasoconstrictoraction of digitalis is counteracted by themodulating effects of digitalis on neuro-hormonal activation and by its sympatho-inhibitory activity. Thus, in patients withheart failure digitalis may induce vaso-dilatation through direct or indirectattenuation of neuroendocrine activation. Inaddition to earlier observations by Mason andBraunwald of the peripheral vasodilatingeffects of ouabain in patients with heart failureand of vasoconstriction in normal subjects,50Ferguson et al recently showed that short termtreatment with digitalis reduces sympatheticnerve activity, resulting in systemic vaso-dilatation in patients with heart failure but notin normal subjects.53 Glycosides increase baro-receptor sensitivity, both in normal subjectsand in patients with heart failure. Underlyingmechanisms may include direct receptorstimulation, an effect related to inhibition ofsodium-potassium ATPase, or a secondaryresponse to improved haemodynamics. Thefindings of Ferguson et al suggest a directeffect, as sympathoinhibitory effects occurredbefore haemodynamic changes wereapparent.53

Short term treatment with digitalis alsoleads to a reduction in circulating catechol-amine concentrations, plasma renin activity,and aldosterone concentrations. In addition,in vitro data indicate that ouabain maystimulate secretion of atrial natriuretic peptidefrom atrial cardiocytes.54 Recently the Dutchibopamine multicentre trial, a six monthcomparison of ibopamine, digoxin, andplacebo in patients with moderate heart failuretaking diuretics, indicated that digoxin alsosignificantly decreases circulating noradrena-line and renin concentrations.55 Thus, digitalisemerges as an interesting drug for managingheart failure, as it combines both cAMPindependent positive inotropic properties andneurohormonal modulating effects.But does it work in clinical practice? There

is little doubt about its efficacy in patients withheart failure and atrial fibrillation, partlybecause of its vagotonic effect. By contrast,despite its early recognition as a therapeuticagent for dropsy56 and the elaborate studies byWithering,57 its usefulness in heart failure andsinus rhythm has been and still is beingdebated. Whereas long term haemodynamicimprovement is clear, many studies that haveexplored the clinical effects of digoxin have

not shown convincing benefit.58-64 Fewadequate controlled studies of digoxin exist.Several included only patients in whomdigitalis could be withdrawn, henceintroducing bias. Enrolment criteria oftenwere not well defined-for example, thepresence of systolic dysfunction andbackground treatment were not controlled foror specified and sample size was ofteninadequate.A few studies have indicated that the

addition of digitalis to existing treatment or itswithdrawal under controlled conditions leadsto clinical improvement or deterioration,respectively.65-67 Patients with heart failure andsinus rhythm who improved tended to bethose with systolic dysfunction, large hearts,an S3 gallop, and heart failure (New YorkHeart Association class III) despite treatment.By contrast, the Dutch ibopamine multicentretrial suggested that digoxin exerted clinicalbenefits even in patients with moderate heartfailure who were treated only with diuretics.55Two recent studies in patients with mild tomoderate heart failure and sinus rhythmindicate that such patients also tend todeteriorate after abrupt withdrawal ofdigitalis.68 69 The RADIANCE study isparticularly noteworthy in this respect, as allpatients in this study, in contrast to previousinvestigations with digitalis, were taking ACEinhibitors69 (fig 6). Thus, when clinicalefficacy is concerned, it makes sense to givedigitalis to patients with comparatively severeheart failure and sinus rhythm despite theaccepted standard treatment of diuretics andACE inhibitors. Also, it makes sense tocontinue digitalis when such patients haveimproved to a milder stage of heart failure. Buthow safe is digoxin?

What if digoxin increases mortality?Despite centuries of use, doctors do not knowwhether digitalis has an effect on mortality andwhether this is favourable. Arrhythmias are awell recognised manifestation of digitalistoxicity. The therapeutic-toxic dose range isquite narrow under normal conditions; aging,abnormal renal haemodynamics, and thepropensity for hypokalemia in heart failuremay clearly enhance the potential for toxicity.

30 - Digoxin

-20Pacb

xa, 20n- (n=82)

E -301,,-0- cb \> |az 0

0 -40 -

0 2 4 6 8 10 12Weeks after randomisation

Figure 6 Worsening of exercise capacity in patients withmild heart failure previously taking digoxin, diuretics, andACE inhibitors in whom digoxin was replaced by placebo.The comparison group continued to take digoxin.69Reproduced with permission.

S 97

Remme

As suggested above, the clinical efficacy ofdigitalis is probably better in more severeconditions where pharmacokinetic aspectscould be more disadvantageous. Yet, severalfairly large comparative studies of digitalis inpatients with severe heart failure did not reporta significant increase in ventricular ectopicactivity.63 66 70 Moreover, a proarrhythmiceffect does not necessarily lead to increasedmortality in heart failure.

Digitalis could also beneficially affectsurvival by improving cardiac function, by itssympathoinhibitory and neurohormonalmodulating effects, and by its parasympatho-mimetic properties. Its inotropic effect mayprimarily occur at higher doses and its neuro-hormonal effect at lower doses.7' Althoughthis assumes less detrimental effects with lowdose digitalis-in line with the experienceswith vesnarinone and possibly enoximone-this is still speculative.As with other positive inotropes, a large trial

of digoxin is needed to assess its effects onsurvival at different stages of heart failure,preferably with monitoring of serum digoxinconcentrations and against the backgroundof currently accepted treatment for heartfailure. Such a study is presently beingconducted in over 8000 patients-the DIGtrial. If this trial shows a favourable effect ofdigoxin on survival then it is likely that allpatients with established heart failure will betreated with digoxin and an ACE inhibitor,diuretics retaining their role when fluidretention is a problem. If digoxin's effect onmortality is neutral then the knowledge thatdigoxin is safe will also lead to more wide-spread use. On the other hand, if digoxinincreases mortality use will decline. Most ofthe properties that have been imputed todigoxin at one time or another have alreadybeen proved with ACE inhibitors: sympto-matic improvement, delays in thedeterioration of heart failure, and increasedsurvival. However, digoxin seems to have arole in improving symptoms in addition to thatof an ACE inhibitor. Many physicians maywish to continue to use digoxin in severe heartfailure, believing that the relief of symptoms isa sufficient reward for the increase inmortality. This is similar to what recentlyhappened with flosequinan, a new agent forheart failure. The medicolegal position in thatcase was considered to be untenable, and thedrug was withdrawn. Will a similar situationdevelop with digoxin, the medicolegal threatnot being to a pharmaceutical company but toprescribing doctors?

1 Allen DG, Kurihara S. Calcium transients in mammalianventricular muscle. Eur Heart J 1980;1(suppl A):5-1 5.

2 Schramm M, Thomas G, Towart R, Franchowiak G. Noveldihydropyridines with positive inotropic action throughactivation of Ca2+ channels. Nature 1983;303:535-7.

3 Hajjar RJ, Gwathmey JK. Calcium-sensitizing inotropicagents in the treatment of heart failure: a critical view.Cardiovascular Drugs and Therapy 1991;5:961-6.

4 Holubarsch C, Hasenfuss G, Just H, Blanchard E, MulieriLA, Alpert NR. Influence of the positive inotropicsubstance pimobendan (UD-CG 115BS) on contractileeconomy of guinea pig papillary muscles. J CardiovascPhannacol 1989;14(suppl 2):S13-7.

5 Scheld HH, Fritsche R, Schlepper M, Van Meel JCA.Pimobendan increases calcium sensitivity of skinned

human papillary muscle fibres. J Clin Pharnacol1989;29:360-5.

6 Wilkie DR. Muscular fatigue: effects of hydrogen ions andinorganic phosphate. Fed Proc 1986;45:2921-3.

7 Silver PJ, Allen P, Etzler JH, Hamel LT, Bentley RG,Pagani ED. Cellular distribution and pharmacologicalsensitivity of low Km cyclic nucleotide phosphodiesteraseisozymes in human cardiac muscle from normal andcardiomyopathic subjects. Second-Messengers-Phospho-proteins 1990;13:13-25.

8 Bohm M, Diet F, Feiler G, et al. Subsensitivity of thefailing human heart to isoprenaline and milrinone isrelated to beta-adrenoceptor downregulation. JCardiovasc Pharnacol 1988;12:726-32.

9 Bristow MR, Anderson FL, Port JD, et al. Differences in 3adrenergic neuroeffector mechanisms in ischemic versusidiopathic dilated cardiomyopathy. Circulation 1991 ;84:1024-39.

10 Feldman AM, Cates AE, Veazey WB, et al. Increase in the40,000-mol wt pertussis toxin substrate (G-protein) inthe failing human heart. J7 Clin Invest 1988;82:189-97.

11 Eschenhagen T, Mende U, Nose M, et al. Increasedmessenger RNA level of the inhibitory G protein asubunit Gi-2 in human end-stage heart failure. Circ Res1992;70:688-96.

12 Bohm M, Ungerer M, Erdmann E. Beta adrenoceptors andm-cholinoceptors in myocardium of hearts with coronaryartery disease or idiopathic dilated cardiomyopathyremoved at cardiac transplantation. Ant J7 Cardiol1990;66:880-2.

13 Schnabel P, Bohm M, Gierschik P, Jakobs KH, ErdmannE. Improvement of cholera toxin-catalysed ADP-ribosylation by endogenous ADP-ribosylatin factor frombovine brain provides evidence for an unchanged amountof G., in failing human myocardium. Jf Mol Cell Cardiol1990;22:73-82.

14 Bristow MR, Ginsburg R, Umans V, et al. Beta, and beta2adrenergic receptor subpopulations in nonfailing andfailing human ventricular myocardium: coupling of bothreceptor subtypes to muscle contraction and selectivebeta, receptor down-regulation in heart failure. Circ Res1986;59:297-309.

15 Bristow MR, Hershberger RE, Port JD, Rasmussen R. ,Band 12 adrenergic receptor-mediated adenylate cyclasestimulation in nonfailing and failing human ventricularmyocardium. J Mol Pharmacol 1989;35:295-303.

16 Bohm M, Gierschik P, Jakobs K-H, et al. Increase of Gi,, inhuman hearts with dilated but not ischemiccardiomyopathy. Circulation 1990;82: 1249-65.

17 Bristow MR, Minobe W, Rasmussen R, Hershberger RE,Hoffman BB. Alpha- 1 adrenergic receptors in thenonfailing and failing human heart. J Pharmacol Exp Ther1988;247: 1039-45.

18 Bohm M, Diet F, Feiler G, Kemkes B, Erdmann E. aAdrenoceptors and a adrenoceptor-mediated positiveinotropic effects in failing human myocardium. JCardiovasc Pharmacol 1988;12:357-64.

19 Landzberg JS, Parker JD, Gauthier DF, Colucci WS.Effects of myocardial oa,-adrenergic receptor stimulationand blockade on contractility in humans. Circulation1991;84:1608-14.

20 Leier CV, Huss P, Lewis RP, Unverferth DV. Drug-induced conditioning in congestive heart failure.Circulation 1982;65: 1382-7.

21 Weber KT, Andrews V, Janicki JS, Likoff M, Reichek N.Pirbuterol, an oral beta-adrenergic receptor agonist, inthe treatment of chronic cardiac failure. Circulation1982;66: 1262-7.

22 Colucci WS, Alexander WR, Williams GH, et al. Decreasedlymphocyte beta-adrenergic-receptor density in patientswith heart failure and tolerance to the beta-adrenergicagonist pirbuterol. N EnglJ Med 198 1;305:185-90.

23 Mettayur B, Rouleau J-L, Burgess JH. Detrimentalarrhythmogenic and sustained beneficial hemodynamiceffects of oral salbutamol in patients with chroniccongestive heart failure. Anm Heart J 1985;109:840-7.

24 Xamoterol in Severe Heart Failure Study Group.Xamoterol in severe heart failure. Lancet 1990;336:1-6.

25 Anderson FL, Port JD, Reid BB, et al. Effect of therapeuticdopamine administration on myocardial catecholamineand neuropeptide Y concentrations in the failingventricles of patients with idiopathic cardiomyopathy. JCardiovasc Phannacol 1992;20:800-6.

26 Rousseau MF, Raigoso J, Van Eyll C, et aL Effects ofintravenous epinine administration on left ventricularsystolic performance, coronary hemodynamics andcirculating catecholamines in patients with heart failure.J Cardiovasc Pharnnacol 1992;19:155-62.

27 Ludmer PL, Wright RF, Arnold JMO, Ganz P, BraunwaldE, Colucci WS. Separation of the direct myocardial andvasodilator actions of milrinone administered by anintracoronary infusion technique. Circulation 1986;73:130-7.

28 Cody RJ, Muller FB, Kubo SH, Rutman H, Leonard D.Identification of the direct vasodilator effect of milrinonewith an isolated limb preparation in patients with chroniccongestive heart failure. Circulation 1986;73:124-9.

29 Remme WJ. Inodilator therapy for heart failure. Early, lateor not at all? Circulation 1993;87(suppl IV):IV-97-107.

30 Muir AL, Nolan J. Modulation of venous tone in heartfailure. Am HeartJ7 1991;121:1948-50.

31 Borow KM, Come PC, Neumann A, Baim DS, BraunwaldE, Grossman W. Physiologic assessment of the inotropic,vasodilator and afterload reducing effects of milrinone in

S 98

Inotropic agents for heart failure: what if digoxin increases mortality?

subjects without cardiac disease. Am _7 Cardiol1985;55: 1204-9.

32 Remme WJ, Van Hoogenhuyze DCA, Krauss XH,Kruyssen HACM, Pieper PG, Bruggeling WAJ. Preload-dependent hemodynamic effects of milrinone inmoderate heart failure. Cardiology 1992;80: 132-42.

33 Remme WJ. Vasodilator therapy without convertingenzyme inhibition in congestive heart failure-useful-ness and limitations. Cardiovasc Drugs Ther 1989;3:375-96.

34 Massie B, Bourassa M, DiBianco R, Hess M, Konstam M,Likoff M, et al for the Amrinone Multicentre StudyGroup. Long-term oral administration of amrinone forcongestive heart failure: lack of efficacy in a multicentercontrolled trial. Circulation 1985;71 :963-71.

35 Uretsky BF, Jessup M, Konstam MA for the EnoximoneMulticenter Trial Group. Multicenter trial of oralenoximone in patients with moderate to moderatelysevere congestive heart failure. Lack of benefit comparedwith placebo. Circulation 1990;82:774-80.

36 DiBianco R, Shabetai R, Kostuk W, Moran J, Schlant RC,Wright R. A comparison of oral milrinone, digoxin andtheir combination in the treatment of patients withchronic heart failure. N Engl J Med 1989;320:677-83.

37 Assmann I, Kassel P, Duck HG, et al. Akut- undLangzeiteffekte von Pimobendan (UD-CG 115) beiHerzinsuffizienz NYHA II und III. Ergebnisse einerrandomisierten multizentrischen Doppelblindstudie. ZKardiol 1991;80:687-94.

38 Katz SD, Kubo SH, Jessup M, et al. A multicenter,randomized, double-blind, placebo-controlled trial ofpimobendan, a new cardiotonic and vasodilator agent, inpatients with severe congestive heart failur. Ami Heart 71992;123:95-103.

39 Kubo SH, Gollub S, Bourge R, et al for the PimobendanMulticenter Research Group. Beneficial effects ofpimobendan on exercise tolerance and quality of life inpatients with heart failure. Results of a multicenter trial.Circulation 1992;85:942-9.

40 Remme WJ, Krayenbuhl HP, Baumann G, et al for thePimobendan-Enalapril Study Group. Long-term efficacyand safety of pimobendan in moderate heart failure. Adouble-blind parallel 6 month comparison with enalapril.Eur Heart 7 (in press).

41 Remme WJ, Wiesfeld ACP, Look MP, Kruyssen HACM.Hemodynamic effects of intravenous pimobendan inpatients with left ventricular dysfunction. _7 CardiovascPharmacol 1989;14(suppl 2):S41-4.

42 Hasenfuss G, Holubarsch C, Heiss HW, Rattert B, Just H.Influence of the calcium-sensitizer UDCG-1 15 onhaemodynamics and myocardial energetics in patientswith idiopathic dilated cardiomyopathy. Comparisonwith nitroprusside. Basic Res Cardiol 1989;84:225-33.

43 Hasenfuss G, Holubarsch C, Heiss W, et al. Myocardialenergetics in patients with dilated cardiomyopathy.Influence of nitroprusside and enoximone. Circulation1989;80:51-64.

44 Haleen SJ, Steffen RP, Sircar I, et al. PD 122,860: a noveldihydropyridine with sodium channel stimulating andcalcium blocking properties. _7 Phanracol Exp Ther1989;250:22-30.

45 Holck M, Osterrieder W. Interaction of the cardiotonicagent DPI 201-106 with cardiac Ca2 channels. .Cardiovasc Pharmacol 1988; 11:478-82.

46 Feldman AM, Bristow MR, Parmley WW, et al for theVesnarinone Study Group. Effects of vesnarinone ormorbidity and mortality in patients with heart failure. NEnigl_7 Med 1993;329:149-55.

47 Yanagisawa T, Ishi K, Taira N. Antitachycardiac effect ofOPC-8212, a novel cardiotonic agent, on tachycardiacresponses of guinea pig isolated right atria toisoproterenol and histamine. .7 Cardiovasc Pharmacol1987;10:47-54.

48 Busch FW, Tillmann A, Becker EW, Owsianowski M, BergPA. The inhibitory effects of a positive inotropicquinolinone derivative, 3,4-dihydro-6-[4-(3,4-dimeth-oxybenzoyl)- 1 -piperazynyl] 2(IH)-quinolinone (OPC-8212), on bone marrow progenitor cells and peripherallymphocytes. EurJ7 Clin Pharmiacol 1992;42:629-33.

49 Cooke JP, Shepherd JT, Vanhoutte PM. Vasoconstrictioninduced by ouabain in the canine coronary artery:contribution of adrenergic and nonadrenergic responses.Cardiovascular Drugs and Therapy 1988;2:255-63.

50 Mason DT, Braunwald E. Studies on digitalis. X. Effectsof ouabain on forearm vascular resistance and venoustone in normal subjects and in patients in heart failure..7 Clin Invest 1964;43:532-43.

51 DeMots H, Rahimtoola SH, McAnulty JH, Porter GA.Effects of ouabain on coronary and systemic vascularresistance and myocardial oxygen consumption inpatients without heart failure. Amin 7 Cardiol1978;41 :88-93.

52 Indolfi C, Piscione F, Russolillo E, et al. Digoxin-inducedvasoconstriction of normal and atherosclerotic epicardialcoronary arteries. Am.-7 Cardiol 1991 ;68: 1274-8.

53 Ferguson DW, Berg WJ, Sanders JS, Roach PJ, Kempf JS,Kienzle MG. Sympathoinhibitory responses to digitalisglycosides in heart failure patients. Circulation1989;80:65-77.

54 Bloch KD, Zmir N, Lichtstein D, Seidman CE, SeidmanJG. Ouabain induces secretion of pro-atrial natriureticfactor in rat atrial cardiocytes. Am _7 Phvsiol1988;255:E383-7.

55 Van Veldhuisen DJ, Man in 't Veld AJ, Dunselman PHJM,et al. Double-blind placebo-controlled study ofibopamine and digoxin in patients with mild to moderateheart failure: results of the Dutch ibopamine multicentertrial (DIMT). _J Ami Coll Cardiol 1993;22:1564-73.

56 Jacobs M. A history of digitalis therapy. Annals of MedicalTherapeutics 1936;492.

57 Withering W. An account of the foxglove and somle of itsmiedical uses-practical remnarks on dropsy and other diseases.Birmingham: M Swinney, 1785.

58 Fleg L, Gottlieb SH, Lakatta EG. Is digoxin reallyimportant in compensated heart failure? Ami . Med1982;73:244-50.

59 Taggart AJ, Johnston GD, McDevitt DG. Digoxinwithdrawal after cardiac failure in patients with sinusrhythm. _7 Cardiovasc Pharmtacol 1983;5:229-34.

60 Guyatt GH, Sullivan MJJ, Fallen EL, et al. A controlledtrial of digoxin in congestive heart failure. Anm _7 Cardiol1988;61:371-5.

61 Aronow WS, Starling L, Etienne F. Lack of efficacy ofdigoxin in treatment of compensated congestive heartfailure with third heart sound and sinus rhythm in elderlypatients receiving diuretic therapy. Amn _7 Cardiol1986;58: 168-9.

62 German and Austrian Xamoterol Study Group. Double-blind placebo-controlled comparison of digoxin andxamoterol in chronic heart failure. Lancet 1988;i:489-93.

63 Captopril-Digoxin Multicenter Research Group.Comparative effects of therapy with captopril anddigoxin in patients with mild to moderate heart failure.JAMA 1988;259:539-44.

64 Davies RF, Beanlands DS, Nadeau C, et al. Enalaprilversus digoxin in patients with congestive heart failure: amulticenter study. _7 Ami Coll Cardiol 199 1;18:1602-9.

65 Lee DC-S, Johnson RA, Bingham JB, et al. Heart failure inoutpatients: a randomized trial of digoxin versus placebo.N Engl _J Med 1982;306:699-705.

66 DiBianco R, Shabetai R, Kostuk W, et al. A comparison oforal milrinone, digoxin, and their combination in thetreatment of patients with chronic heart failure. N Engl_7 Med 1989;20:677-83.

67 Fleg JL, Rothfeld B, Gottlieb SH. Effect of maintenancedigoxin therapy on aerobic performance and exercise leftventricular function in mild to moderate heart failure dueto coronary artery disease: a randomized, placebo-controlled crossover trial. .7 Amn Coll Cardiol 1991;17:743-51.

68 Uretsky BF, Young JB, Shahidi FE, Yellen LG, HarrisonMC, Jolly KM. Randomized study assessing the effect ofdigoxin withdrawal in patients with mild to moderatechronic congestive heart failure: results of the PROVEDtrial. .7 Amti Coll Cardiol 1993;22:955-62.

69 Packer M, Gheorghiade M, Young JB, et al. Withdrawal ofdigoxin from patients with chronic heart failure treatedwith angiotensin-converting enzyme inhibitors. N Engl _7Med 1993;329:1-7.

70 Lown B, Graboys TB, Podrid PJ, Cohen BH, StockmanMS, Gaugham LE. Effect of digitalis drug on ventricularpremature beats. N Etngl _J Med 1977;296:301-6.

71 Packer M. The development of positive inotropic agents forchronic heart failure: how have we gone astray?7Ani CollCardiol 1993;22(suppl A):I 19-26A.

S 99