Clarithromycin,QTc interval prolongation and torsades de pointes : the need to study case reports

Abstract

Background:

The manufacturers of clarithromycin sought a drug similar in efficacy to erythromycin but with a superior side-effect profile. They generally achieved this outcome, but postmarketing findings identified a series of reports linking clarithromycin to QTc interval prolongation and torsades de pointes (TdP) ultimately leading to a Black Box Warning. We sought to clarify risk factors associated with TdP among case reports of patients receiving clarithromycin linked to QTc interval prolongation and TdP.

Methods and results:

In a detailed literature search, we found 15 women, five men, and one boy meeting our search criteria. Among the 17 adults with reported clarithromycin dose and concurrent QTc interval measurement, we found no statistically significant relationship between clarithromycin dose and QTc interval duration. This did not change for the adults who developed TdP. Among adults, major risk factors were female sex (15), old age (11) and heart disease (17). A total of eight adult subjects had all three major risk factors and 14 of the 20 adults had at least two major risk factors. All adult subjects had at least two risk factors besides clarithromycin. A total of four of the 20 adults received cisapride and three received disopyramide. Three adults were considered to suffer from some aspect of the congenital long QT syndrome.

Conclusions:

We believe that the risk factor description for this drug should be refined to emphasize the major risk factors of (1) female sex, (2) old age and (3) heart disease.

Introduction

Clarithromycin is the most widely used macrolide antibiotic in the United Kingdom and is strongly recommended for patients with severe community-acquired pneumonia [Schembri et al. 2013]. The drug’s use among patients with chronic obstructive pulmonary disease or community-acquired pneumonia is associated with increased cardiovascular events over the course of a year among patients requiring hospitalization with acute coronary syndrome, decompensated heart failure, serious arrhythmias, or sudden cardiac death (SCD) [Schembri et al. 2013]. Because increased cardiovascular risk persisted over the next year after clarithromycin was discontinued [Schembri et al. 2013], the immediate effect of the drug did not explain the above findings even though clarithromycin has a proarrhythmia effect linked in some fashion to QTc interval prolongation and torsades de pointes (TdP) [Bril et al. 2010].

Bril and colleagues, in a study of antimicrobial agents linked to QTc interval prolongation, outlined risk factors using a flowchart [Bril et al. 2010]. Drug-related risk factors were separated into administration, pharmacokinetic interactions and pharmacodynamic interactions. Host-related risk factors included modifiable ones (internal environmental disturbances, heart disease, central nervous system and dysautonomy) and those that cannot be modified (older age, female sex, structural heart disease, long QT syndrome and ion channel polymorphisms). Based on our experience with a case report format [Vieweg et al. 2012, 2013a, 2013b; Kogut et al. 2013], we seek to organize risk factors to better help clinicians manage patients requiring treatment with clarithromycin.

Background

In the 1970s, the Japanese drug company Taisho Pharmaceutical developed the macrolide antibiotic clarithromycin aiming to improve on erythromycin’s spectrum of activities, frequency of administration and tolerability [Zuckerman, 2004]. A branded version was introduced into the Japanese market in 1991 and the US market later in the same year. Less than 6 years after its introduction, clarithromycin was linked to erythromycin’s adverse electrophysiological properties including drug-induced TdP [Kundu et al. 1997; Paar et al. 1997; Sekkarie, 1997]. This linkage is exacerbated by the common coprescription of QTc interval prolonging drugs in the outpatient setting [Curtis et al. 2003] and the coprescription of drugs blocking CYP3A4 [Sagir et al. 2003]. Clarithromycin went generic in Europe in 2004 and in the US in mid-2005.

Clarithromycin pharmacology

Pharmacokinetics and pharmacodynamics

The pharmacokinetics of clarithromycin are similar to those of erythromycin [McConnell and Amsden, 1999]. Clarithromycin is rapidly absorbed from the gastrointestinal tract. Clarithromycin has the greatest bioavailability of any of the macrolides [McConnell and Amsden, 1999]. Clarithromycin pharmacokinetics are basically linear and the steady-state concentration of this metabolite is reached within 3–4 days [Abbott Laboratories, 2013].

The postantibiotic effect of clarithromycin varies according to the pathogen [McConnell and Amsden, 1999]. However, the in vivo significance of this in vitro observation has not been determined. Clarithromycin penetrates infected tissue sites extensively. Efficacy is based on maintaining a drug concentration above the minimum inhibitory concentration and clarithromycin is dosed with a frequency to meet this requirement.

Clarithromycin as a hERG channel inhibitor

A common feature of drugs associated with acquired long QT syndrome and TdP is their ability to produce pharmacological inhibition of the activity of the hERG (human Ether-a-go-go Related Gene) potassium ion channel and its native cardiac equivalent, the rapid delayed rectifier K ⁺ current, I_Kr [Finlayson et al. 2004; Hancox et al. 2008]. Macrolide antibiotics including clarithromycin have been reported to inhibit the hERG channel current (I_hERG) [Abbott et al. 1999; Volberg et al. 2002; Stanat et al. 2003; Duncan et al. 2006]. For example, using hERG channel expression in mammalian (HEK 293) cells, Volberg and colleagues demonstrated I_hERG inhibition by six macrolide drugs of which clarithromycin was the most potent (half maximal inhibitory concentration, IC₅₀, of 32.9 µM), whilst erythromycin exhibited intermediate potency (IC₅₀ of 72.2 µM) and oleandomycin was the least potent (IC₅₀ of 339.6 µM) [Volberg et al. 2002]. However, in a separate study using a similar hERG expression system, in a direct comparison erythromycin was reported to be more potent than clarithromycin (IC₅₀ values of 38.9 and 45.7 µM, respectively). In that study both drugs were also reported to inhibit I_hERG when this was elicited with an action potential shaped waveform [Stanat et al. 2003], consistent with an ability to reduce the contribution of I_hERG/I_Kr to action potential repolarization and, hence, the QT interval. Clarithromycin has been shown to prolong action potential duration (APD) in rabbit cardiac tissues in a concentration-dependent fashion (over concentrations ranging between 3 and 100 µM) and to reduce native I_Kr amplitude [Gluais et al. 2003]. For a given drug concentration, APD prolongation was found to be greater for Purkinje fibers than for ventricular myocytes [Gluais et al. 2003], presumably because the comparatively impoverished repolarization reserve of Purkinje fibers makes them particularly sensitive to hERG/I_Kr blocking drugs [Gintant, 2008; Hancox et al. 2008]. In experiments on intact perfused rabbit hearts, high concentrations of clarithromycin prolonged both monophasic action potential and QT interval duration [Milberg et al. 2002]. In the presence of simulated hypokalemia, clarithromycin could produce cellular arrhythmogenic events [early afterdepolarizations (EADs)] and TdP [Milberg et al. 2002]. Thus, the preclinical literature demonstrates that at some concentrations clarithromycin can inhibit I_hERG/I_Kr, prolong action potential repolarization and, under appropriate conditions, predispose towards arrhythmia.

An intriguing aspect of clarithromycin’s ability to inhibit hERG channels is the potential for this action of the drug to vary between individuals. In 1999, Abbott and colleagues discovered an accessory protein, MiRP1 (also referred to as KCNE2), that can co-assemble with hERG, modifying I_hERG properties and drug sensitivity [Abbott et al. 1999]. A MiRP1 mutation, Q9E-hMiRP1, which was identified in a patient treated who survived clarithromycin-induced TdP and ventricular fibrillation, was found in vitro to increase markedly hERG’s sensitivity to inhibition by clarithromycin [Abbott et al. 1999].

Analysis of static and dynamic complex variables

Multiple regression analysis evaluates the effects of more than one independent variable on a dependent variable. However, the assumption is that neither variable changes in the course of analysis. Risk factors associated with drug-induced QTc interval prolongation and TdP may change dynamically and the risk factors themselves may be approximations of more specific factors [Vieweg et al. 2012, 2013a, 2013b; Kogut et al. 2013].

Elements of nonlinear phenomena include chaos, fractals, cellular automata, genetic algorithms and fuzzy logic. They have been brought together by Willi-Hans Steeb in the 5th edition of his book The Nonlinear Workbook [Steeb, 2011]. Components likely best describe the interaction of drug-linked QTc interval prolongation and attendant risk factors to produce TdP [Vieweg et al. 2012, 2013a, 2013b; Kogut et al. 2013].

Methods

We conducted a systematic review of case reports published up to and including 17 April 2013. Initially, we entered the following MeSH terms: ‘clarithromycin and QTc prolongation’ (13) and ‘clarithromycin and torsade’ (10) into Medline. We searched CredibleMeds (http://www.azcert.org/) for case reports of clarithromycin, QTc interval prolongation and torsades de pointes (19). This search was initiated via AZCERT: (“Clarithromycin”[MeSH] AND (“Long QT Syndrome”[MeSH] OR “Torsades de Pointes”[MeSH])) OR (((torsade[ti] OR torsadegenic[ti] OR torsades[ti] OR torsadogenesis[ti] OR torsadogenic[ti] OR torsadogenicity[ti]) OR qt[ti]) AND clarithromycin[ti]). We searched EMBASE (0) and Cochrane (0) only for case reports. Only human studies were included. We also reviewed reports from our files and reference lists yielding a total of 21 cases as shown in Table 1. (A more detailed case narrative appears in the appendix.) Titles and abstracts were independently reviewed by two investigators (WVRV and AB). Disagreement was resolved by consensus.

Table 1.

Risk factors for QTc interval prolongation and torsades de pointes (TdP) by case reports among patients receiving clarithromycin (CAM).

Case/arrhythmia	QTc (ms)	CAM daily dose	Female sex	Elderly	Heart disease	Hypo- K+	Hypo- Mg++	Bradycardia	CYP3A4 blockers	QTc prolonging drugs	Additional risk factors
1 [Lai et al. 1996] 70-year-old woman No arrhythmia documented	641	500	Yes	Yes	Yes	No	No	No	No	Sotalol	Acute medical illness, chronic medical illness
2 [Kundu et al. 1997] 27-year-old woman Ventricular tachycardia	468	500	Yes	No	No	No	No	No	No	No	Acute medical illness
3 [Paar et al. 1997] 74-year-old woman Ventricular fibrillation	625	500	Yes	Yes	Yes	Yes	No	No	Disopyramide	Disopyramide	Acute medical illness, chronic medical illness
4 [Sekkarie, 1997] 52-year-old woman TdP	630	1000	Yes	No	Yes	No	No	No	Cisapride	Cisapride	Acute medical illness, chronic medical illness
5 [Sekkarie, 1997] 83-year-old man TdP	Unknown	1000	No	Yes	Yes	No	No	No	Cisapride	Cisapride	Acute medical illness, chronic medical illness
6 [Gray, 1998] 53-year-old woman TdP	456	1000	Yes	No	Yes	No	No	No	Cisapride, fluoxetine	Cisapride, fluoxetine	Acute medical illness, chronic medical illness
7 [Lee et al. 1998] 40-year-old man TdP	670	1000	No	No	Yes	No	No	No	No	No	Acute medical illness, chronic medical illness, impaired liver function
8 [Lee et al. 1998] 25-year-old woman TdP	775	1000	Yes	No	Yes	No	No	No	No	No	Acute medical illness, chronic medical illness
9 [Hayashi et al. 1999] 76-year-old woman TdP	710	400	Yes	Yes	Yes	Yes	No	No	Disopyramide	Disopyramide	Acute medical illness
10 [Piquette, 1999] 77-year-old woman TdP	640	1000	Yes	Yes	Yes	Yes	No	No	Cisapride	Cisapride	Acute medical illness, chronic medical illness
11 [Kamochi et al. 1999] 78-year-old woman TdP	654	400	Yes	Yes	Yes	Yes	No	No	No	No	Acute medical illness
12 [Kamochi et al. 1999] 62-year-old man TdP	600	400	No	No	No	No	No	No	No	No	Acute medical illness, chronic liver disease
13 [Abbott et al. 1999] 76-year-old woman TdP	540	1000	Yes	Yes	Yes	Yes	No	No	No	Erythromycin	Mutation to MiRP1 protein, acute medical illness, chronic medical illness
14 [Choudhury et al. 1999] 76-year-old woman TdP	640	500	Yes	Yes	Yes	No	No	No	Disopyramide	Disopyramide	Acute medical illness, chronic medical illness, mild renal failure
15 [Wasmer et al. 1999] 29-year-old woman TdP	680	Unknown	Yes	No	Yes	No	No	No	No	No	Congenital long QT syndrome, acute medical illness
16 [Vallejo Camazon et al. 2002] 30-year-old man Monomorphic ventricular tachycardia	580	1000	No	No	Yes	No	No	Yes	Cotrimoxazole	Cotrimoxazole, methadone	Acute medical illness, chronic medical illness, chronic liver disease
17 [Diaz Garcia et al. 2005] 80-year-old man None	560	1000	No	Yes	Yes	No	No	No	No	No	Acute medical illness, chronic medical illness
18 [Hensey and Keane, 2008] 79-year-old woman TdP	542	500	Yes	Yes	Yes	No	No	No	No	Amiodarone	Acute medical illness, chronic medical illness, suspected congenital long QT syndrome
19 [Buchanan Keller and Lemberg, 2008] 50-year-old woman TdP	620	Unknown	Yes	No	No	Yes	No	No	No	Fluoxetine	Acute medical illness
20 [Alesso et al. 2009] 71-year-old woman Sustained ventricular tachycardia	570	1000	Yes	Yes	Yes	No	No	Yes	Ciprofloxacin	Amiodarone, Ciprofloxacin	Acute medical illness,
21 [Cetin et al. 2012] 6-year-old boy TdP	600	Unknown	No	No	No	No	No	No	No	No	Acute medical illness

Results

We found 15 women, five men and one boy (Table 1). Among the 17 adults with reported clarithromycin dose and concurrent QTc interval measurement, we found no statistically significant relationship between clarithromycin dose and QTc interval duration (Pearson’s r = −0.086, p = 0.744; Kendall’s τ_B r = −0.185, p = 0.359; Spearman’s rho ρ = −0.223, p = 0.390). These findings did not change for the adults who developed TdP (Pearson’s r = −0.093, p = 0.785; Kendall’s τ_B r = −0.159, p = 0.541; Spearman’s rho ρ = −0.161, p = 0.636). Among adults, major risk factors were female sex (15), old age (11) and heart disease (17). A total of eight adult subjects had all three major risk factors and 14 of the 20 adults had at least two major risk factors. All adult subjects had at least two risk factors besides clarithromycin. A total of four of the 20 adults received cisapride and three received disopyramide. Three adults were considered to suffer from some aspect of the congenital long QT syndrome.

Discussion

In healthy volunteers, the administration of clarithromycin did not lengthen the QTc interval [Van Haarst et al. 1998]. However, plasma clarithromycin levels were not reported in this study. In our study, there was neither parametric nor nonparametric statistical association between drug dose and QTc interval measurement.

The finding of congenital long QT syndrome in three (ages 29, 76 and 79 years) of our 21 patients (14.3%) is quite surprising because the prevalence in the general population is approximately 1:2000 [Schwartz et al. 2009]. Our finding argues strongly for inquiry into a family history of cardiac arrhythmia, presyncope, syncope and SCD before administering clarithromycin.

Cisapride and disopyramide

Among our 20 adult patients (Table 1), four received cisapride (Cases 4, 5, 6 and 10) and all four had at least two major risk factors (female sex, old age and heart disease). That is, clinicians may have to take into consideration other risk factors even when clarithromycin and cisapride are co-administered. Three of our patients (Cases 3, 9 and 14, ages 74, 76 and 76 years, respectively) received disopyramide and all three were elderly and had heart disease. Again, clinicians may have to look beyond the co-administration of clarithromycin and disopyramide to more completely explain drug-related QTc interval prolongation and TdP. These observations highlight the point that these drugs when combined with clarithromycin act synergistically to increase QTc interval duration and increase the risk of subsequent TdP, but may not be the only risk factors in operation.

Risk factors discussed in the literature

In 2001, Bednar and colleagues identified 16 risk factors for QTc interval prolongation and TdP [Bednar et al. 2001]. In 2003, Witchel and coworkers described high-risk patients [Witchel et al. 2003]. In 2003, Viskin and colleagues identified 12 risk factors [Viskin et al. 2003]. In a study of 249 patients, Zeltser and coworkers made the point most patients at increased risk for QTc interval prolongation and TdP have readily identifiable risk factors [Zeltser et al. 2003]. The majority of their 249 patients had at least one risk factor that could be easily identified and 71% had ≥2 easily identifiable risk factors for QTc interval prolongation and TdP. In addition, most women (female sex being the most common identifiable risk factor) had additional risk factors.

Multiple-hit hypothesis

Owens and Ambrose in their study of TdP associated with fluoroquinolones describe the ‘multiple-hit hypothesis’ in which several factors influence TdP development [Owens and Ambrose, 2002]. Many prescribed drugs including antibiotics and antipsychotics unfavorably affect I_Kr kinetics. However, this impact in-and-of itself rarely explains substantial QTc interval prolongation and resultant TdP [Zareba and Lin, 2003]. Combining the list of risk factors produced by several authors produces the following items [Zareba and Lin, 2003; Sauer and Newton-Cheh, 2012]: (1) prolonged QTc interval, (2) female sex, (3) advanced age, (4) bradycardia, (5) hypokalemia, (6) hypomagnesemia, (7) structural heart disease, (8) cardiac arrhythmias, (9) drug combinations including ion channel blockers and CYP450 inhibitors, (10) reduced repolarization reserve and (11) genetic polymorphisms of gene coding cardiac ion channels or enzymes in liver metabolizing drugs.

Pathogenic theories include (1) triangulation (short-term variability of action potentials preceding TdP [Michael et al. 2007]), (2) reverse-use dependence (action potential prolongation at slower heart rates [Hondeghem and Snyders, 1990]), (3) short-term viability (instability [Sauer et al. 2012]), (4) repolarization dispersion [Sauer and Newton-Cheh, 2012]—these four known together as TRIaD [Sauer and Newton-Cheh, 2012]—and (5) EADs [Sauer and Newton-Cheh, 2012].

We need to use all available information about risk factors for QTc interval prolongation and TdP

As pointed out by Owens and Ambrose [Owens and Ambrose, 2002], information from spontaneous reporting rates (case reports) is not synonymous with incidence rates (all cases occurring over a specific interval such as a year). However, we believe that if regulatory agencies more vigorously emphasize (perhaps through a reward system) the need for pharmaceutical companies and clinicians to report all major drug-related adverse cardiac events, spontaneous reporting rates for drug-related QTc interval prolongation and TdP would better approximate incidence rates. If more information unfolds supporting the hypothesis that drug safety related to QTc interval prolongation and TdP more depends on clinical assessment (such as case reports) than preclinical studies such as hERG analysis, clinicians may have better guidance when prescribing drugs and more drugs may reach the level of phase II and phase III studies ultimately providing a better selection of possible treatments for patients with serious illnesses.

However desirable it may be to have (1) baseline EKGs, (2) serum concentrations of nonclarithromycin drugs and (3) genetic information identifying subjects with the congenital long QT syndrome, the clinician is left to decide on the merits and demerits of clarithromycin administration based on the presence or absence of more readily available risk factors for QTc interval prolongation and TdP. We propose that sufficient information based on case report analysis exists to presently identify the triad of (1) female sex, (2) old age and (3) heart disease as major risk factors for QTc interval prolongation and TdP among patients about to receive clarithromycin. We believe that the FDA and manufacturer of clarithromycin should provide this information in the clarithromycin Black Box Warning. The closer the patient comes to having all three major risk factors, the more likely clarithromycin administration will link to QTc interval prolongation and TdP.

Limitations

The topic discussed in this manuscript is not new [Raschi et al. 2013], but we believe merits discussion from a clinical point of view. Owing to the low incidence of TdP, QTc interval prolongation is used as a surrogate marker of drug-induced TdP risk [Shah, 2005]. Large cohort or population studies and ‘thorough QT’ investigations on healthy volunteers can index the likelihood and extent of QTc interval prolongation with individual drugs, but nonetheless usually provide information on a surrogate marker for TdP rather than TdP itself. By contrast, case reports provide direct clinical information in the setting of TdP occurrence. Consequently, case reports provide valuable insight into risk factors concomitant with drug administration. However, the low incidence of TdP means that the numbers of case reports with individual drugs tend to be limited, as is the case here for clarithromycin, and results from our 21 case reports (20 adults and one child) must only be extrapolated to larger groups with great caution. Another potential limitation with case reports is selection bias as fatalities may not always be reported. Thus, both large sample/population data and case report information need to be considered together when attempting to gain an overall picture of TdP risk of particular drugs.

Conclusion

The FDA has taken the position that current information identifies clarithromycin as an at-risk drug for QTc interval prolongation and TdP. In addition to clarithromycin administration, risk factors include (1) uncorrected hypokalemia or hypomagnesemia, (2) clinically significant bradycardia, (3) patients receiving Class IA or Class III antiarrhythmic drugs and (4) elderly patients. On the basis of case report analysis, elderly female patients with heart disease may be particularly at risk.