Exposure to antibacterial agents with QT liability in 14 European countries: trends over an 8-year period

Abstract

AIMS

(i) To classify antibacterial agents with QT liability on the basis of the available evidence, and (ii) to assess trends in their consumption over an 8-year period (1998–2005) in 14 European countries.

METHODS

Current published evidence on QT liability of antibiotics was retrieved through MEDLINE search and joined to official warnings from regulatory agencies. Each drug was classified according to an already proposed algorithm based on the strength of evidence: from group A (any evidence) to group E (clinical reports of torsades de pointes and warnings on QT liability). Consumption data were provided by the European Surveillance of Antibacterial Consumption (ESAC) project and were expressed as defined daily doses per 1000 inhabitants per day (DID).

RESULTS

Among 21 detected compounds, nine [six fluoroquinolones (FQs) and three macrolides (MACs)] belonged to group E. Use of group E drugs ranged from 1.3 (Sweden) to 4.1 DID (Italy) in 1998 and from 1.2 (Sweden) to 6.5 DID (Italy) in 2005. Significant exposure was observed in Italy and Spain (6.5 and 3.8 DID, respectively, in 2005). Only Denmark, Sweden and UK showed a slight decrease in use. Exposure to clarithromycin increased in 10 out of 14 countries, with a marked increment in Italy (3 DID in 2005).

CONCLUSIONS

Notwithstanding regulatory measures, in 2005 there was still significant exposure to antibacterials with strong evidence of QT liability and, in most countries, it was even increased. This warrants further investigation of appropriateness of use and suggests closer monitoring of group E drugs. Physicians should be aware when prescribing them to susceptible patients.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Several noncardiovascular drugs with QT liability are currently on the market.

• Previous epidemiological studies have shown significant exposure of the general population to drugs with QT liability with similar consumption in many European countries.

• Several regulatory measures have concerned medicinal products carrying a pro-arrhythmic risk in humans.

WHAT THIS STUDY ADDS

• The list of antibacterial agents with documented QT liability has grown over the last few years.

• Notwithstanding stringent regulatory measures, population exposure to antibiotics with QT liability is still significant in several countries.

• The magnitude of the problem is clearly heterogeneous, with remarkable diversity between Northern and Southern countries (lower and higher exposure, respectively).

Introduction

The explosion in the number of noncardiovascular drugs associated with ‘QT liability’ (i.e. capable to prolong the QT interval of the electrocardiogram) [1] is, at least in part, the result of the existing regulatory guidelines [International Conference on Harmonization (ICH) S7B [2] and ICH E14 [3], combining preclinical and clinical strategies to reveal the potential of drugs to delay ventricular repolarization, hence causing QT prolongation. This undesired effect is primarily caused by an intrinsic ability to block human ether a-go-go-related gene (hERG) K⁺ channels. Since QT prolongation may trigger the onset of potentially life-threatening arrhythmia, namely torsades de pointes (TdP) [4], the hERG K⁺ channel has become a primary antitarget in drug development [5–7].

Drug-induced TdP represents an important matter of concern for both clinicians and researchers, often culminating in regulatory interventions such as withdrawal (e.g. astemizole, grepafloxacin, cisapride, etc.) or restriction of use (e.g. terfenadine). A recent review [8] reported that QT prolongation (with or without TdP), together with hepatotoxicity, was responsible for >60% of drug withdrawals over the last 16 years. Moreover, estimates are that as many as 60% of new molecular entities are abandoned early during the drug development phase because they are positive on hERG assays, although their true torsadogenic potential is unknown. Therefore, QT liability of commonly prescribed drugs is a topic of particular interest, especially in patients susceptible to cardiac arrhythmias because of ‘reduced repolarization reserve’ (host- and drug-related risk factors) [9].

Previous investigations [10] have found significant exposure to non-antiarrhythmic drugs with QT-prolonging potential in the community: in 1998, approximately 2–3% of all drug prescriptions in the UK and Italy involved medications that may unintentionally cause the long QT syndrome. Moreover, an international drug utilization study, carried out in seven countries, highlighted that the total amount of QT-liability agents dispensed through community pharmacies in 1998 ranged from 13.1 to 19.6 defined daily doses (DDD) per 1000 inhabitants per day [11]. Thus, the question arises whether we are dealing with a class effect (i.e. shared by all agents of a given pharmacological class) or a specific effect of a few agents within a pharmacological class [12–16].

Among drugs with recognized QT liability, antibacterial agents were recently involved in regulatory interventions because of this risk (e.g. withdrawal of grepafloxacin and sparfloxacin from both US and European markets). The pro-arrhythmic potential of antibacterial agents deserves further investigation to define the risk–benefit profile of each drug.

The aim of this study was twofold: (i) to classify antibacterials with QT liability on the basis of the available evidence, and (ii) to estimate population exposure to these agents assessing consumption over an 8-year period (1998–2005) in 14 European countries.

Methods

Organizing evidence on QT liability of antibacterial agents

A MEDLINE search was performed to collect any preclinical and clinical evidence on QT liability of antibacterial agents up to December 2007. Search terms included ‘Anti-Bacterial Agents (Pharmacological Action) (MeSH)’, ‘Long QT Syndrome/chemically induced (MeSH)’, ‘Torsades de Pointes/chemically induced (MeSH)’, ‘HERG*’, ‘QT*’, ‘arrhythmia’, combined with each ‘compound name’. Review article and related reference sections from significant studies were also examined. Moreover, relevant official warnings from regulatory agencies (Food and Drug Administration, USA; European Medicines Agency, Europe; the Italian Pharmaceutical Agency, Italy; and the Medicines and Healthcare products Regulatory Agency, UK) were retrieved.

Drugs were first evaluated according to previous criteria suggested by De Ponti et al.[17]. Briefly, these criteria are based on available clinical and preclinical data as well as official warnings from regulatory agencies concerning QT prolongation, TdP/ventricular arrhythmias occurrence or hERG K⁺ channel blockade. Compounds were then divided into five categories according to the strength of evidence, in increasing order of clinical relevance for TdP risk: from group A, encompassing all agents with any evidence of QT liability, to group E, including only those compounds with the strongest evidence (i.e. clinical reports of TdP and warnings on QT prolongation). Table 1 summarizes the proposed algorithm to assign that risk to each drug. For more details on the method used to organize available data, see De Ponti et al.[11].

Table 1.

Criteria, based on the evidence available as of December 2007, used to group drugs according to the strength of evidence on QT liability

Retrieval of consumption data

European consumption data of antibacterial agents were provided by the European Surveillance of Antibacterial Consumption (ESAC; http://www.esac.ua.ac.be) project, which is an international network of national surveillance systems officially launched during the European Conference on Antibiotic Use in Brussels in 2001. The start-up purpose was to collect harmonized and comparable data on antibacterial consumption (in and out of hospital setting) of the European community from publicly available sources, in order to assess the time trends in human exposure. The main long-term objectives of the ESAC project concern promotion of ‘Good Antibiotic Practice’. All European countries were invited to take part in this project and, in 2005, 34 of them participated, including all 25 EU countries, four applicant countries (Bulgaria, Croatia, Romania and Turkey) and three of the four members of the European Free Trade Association (Iceland, Norway and Switzerland). Before inclusion into the ESAC database, the validity of provided data was evaluated by means of a checklist including possible sources of bias (e.g. problems with population coverage, drug coverage and ambulatory care/hospital care mix).

As recommended by the World Health Organization (WHO), the Anatomical Therapeutic Chemical (ATC) code was assigned to each active substance (5th level of the ATC classification). The volume of consumption was expressed as DDD per 1000 inhabitants per day (DID). DDD represents the assumed average maintenance dose per day for a drug used in adults for its main indication (see also WHO Collaborating Centre for Drug Statistics Methodology Oslo, http://www.whocc.no/).

A sample of 14 countries (Austria, Belgium, Denmark, Finland, France, Germany, Hungary, Italy, Luxembourg, the Netherlands, Slovenia, Spain, Sweden, UK) with consistent and bias-free data was chosen for a careful estimation of the European population exposure. Since Italian providers′ data were not able to share the ESAC project for 1998, we retrieved original data from the IMS Health database (their original source) and then converted them into DID in order to have comparable data. Results were presented both as absolute values and as fractions of overall antibacterial use. The drug utilization analysis of the 8-year period was performed according to the current published literature (as of December 2007).

A more comprehensive and thorough description of data providers, details on methodological approach and in-depth discussion are available in previous publications [18, 19].

Results

Literature data mining

Twenty-one antibacterial agents fulfilled at least one criterion (Table 2); nine of them were labelled to have the strongest evidence on QT liability (group E): six FQs (cipro-, gati-, grepa-, levo-, moxi- and sparfloxacin) and three MACs (azi-, clari- and erythromycin).

Table 2.

Antibacterial agents with documented QT liability (the references refer to information available through December 2007)

	Inclusion criteria						Group
Drug	Ia	Ib	IIa	IIb	IIIa	IIIb	2001	2007
Azitromycin	[34–37]	[38, 39]	[40]a	[40, 41]a	SPC1, PDR2, BNF3		/	E
Ciprofloxacin	[42–44]	[45–49]a	•	[50–52]a	PDR	SPC, BNF	/	E
Clarithromycin	•	•	•	•	•		E	E
	[53, 54]	[55, 56]	[21, 57]	[41]	SPC, PDR, BNF
Clindamycin	•						D	D
Erythromycin	•	•	[62–65][21, 57, 66]	•	•	•	E	E
	[58–61]			[41, 62, 63, 67–69]	BNF, PDR	SPC
Josamycin			[21]				/	B
Gatifloxacin	[70–72]	[73]	•	•	•		B	E
			[74, 75]	PDR
Gemifloxacin		[76]	[77]				/	C
Grepafloxacin	•	•	•	•	•		E	E
Levofloxacin	•	[46, 48, 49, 84–86]	•	[50, 74]	SPC, PDR	BNF	D	E
	[72, 78–83]
Lomefloxacin			[87]			SPC	/	B
Metronidazole	[88]						/	D
Moxifloxacinb	[59, 79, 89, 90]	•	•	•	•		C	E
		[46, 48, 91, 92]	[40, 62, 63, 93–95]	[40, 50, 63, 94–97]	PDR, SPC, BNF
Norfloxacin			[87]		PDR		/	B
Ofloxacin	[72]			[50]		SPC, BNF	/	D
Prulifloxacin			[95]	[74, 95]	SPC		/	B
Roxithromycin	[98–100]	•	[21]	•			C	D
Sparfloxacin	[101, 102]	•	•	•	•		C	E
		[62, 103]	[68, 74, 75, 104, 105]
Spiramycin	•						D	D
Cotrimoxazole	•	•	•			BNF	D	D
Telavancinc		[106]a
Telithromycin		[107–110]	[62, 63]	[63]	EMEA, SPC, PDR, BNF		/	C

• Criteria met up to 2001[17].

^aNegative studies (i.e. reporting no effect) have been added for the sake of completeness.

^bThorough QT Studies (TQTSs) submitted for regulatory purposes have not been included in the table.

^cTelavancin was not included in the analysis since no effect was reported for QT liability. Cases of TdP/QT prolongation caused by interactions have been included despite the causal association being doubtful.

¹Italian Summary of the Product Characteristics.

²Physician Desk Reference or Dear Doctor Letters from the Food and Drug Administration (resulting in labelling changes).

³British National Formulary 2004 edition.

The number of antimicrobials carrying a documented pro-arrhythmic potential has increased over recent years, as indicated by Table 2, where a synopsis of information available up to December 2001 and December 2007 is provided [17]. Notably, group A now includes 10 additional compounds, mainly due to novel published data for drugs already on the market (e.g. azithromycin) as well as marketing approval of new molecules (i.e. moxifloxacin, telithromycin, prulifloxacin, which is currently marketed only in Italy, and gemifloxacin, which received only a US market authorization in 2003). Moreover, for agents already known to affect cardiac repolarization (e.g. levofloxacin) new evidence on QT liability has become available. Thus, as shown in Table 2, several compounds (nine drugs in 2007) now belong to group E because of regulatory measures on the labelling or publication of new data.

Drug utilization data mining

Over the period of interest, the overall European consumption of antibacterials was substantially stable (266 DID in 1998 and 264 in 2005), with some notable exceptions: an increment was observed in Italy (6 DID), Denmark and Austria (2 DID, each). The overall use of antibiotics having any published evidence on QT liability ranged from 2.2 (Sweden) to 7.1 DID (Italy) in 1998 and from 2.0 (Sweden) to 8.4 DID (Italy) in 2005. Consumption increased over the years in half of countries: percentage increase ranged from 5% (Denmark) to 33% (Hungary). The remaining countries (Belgium, France, Luxembourg, Slovenia, Spain, Sweden and the UK) showed a decrease in use, with a maximum of 15% in Spain.

Focusing on drugs belonging to group E, their consumption increased during the period of interest (1998–2005) in most of countries (except for Denmark, Sweden and the UK). Specifically, their use ranged from 1.3 (Sweden and the Netherlands) to 4.1 DID (Italy) in 1998 and from 1.2 (Sweden) to 6.5 DID (Italy). Italy displayed the highest increase: in the 8-year period, community exposure increased to 1.6-fold (an increase of 2.4 DID). Significant exposure was observed also in Spain, Luxembourg (3.8 DID each), Hungary and Belgium (3.7 DID each). When the same data were expressed as a fraction of overall antibacterial use, most countries (except Denmark, Sweden and the UK) showed a positive time trend, with the highest increase in Hungary: from 11% (1998) to 19% (2005) (Figure 1).

Figure 1.

Consumption of antibacterial agents grouped by the strength of evidence on QT liability: comparison between 1998 and 2005 based on information available as of December 2007. Most countries showed an increase of population exposure both in terms of absolute value and as a fraction of overall antibacterial use. Countries were ranked by magnitude of total antibacterial consumption in 1998. Total consumption of antibacterials without evidence on QT-liability (□); Total consumption of antibacterials with QT-liability except for compounds listed in group E (); Group E, strongest level of evidence on QT-liability (▪). Percentage: antibacterial agents labeled as group E/total consumption of antibacterials. In parenthesis changes in use of group E compounds (absolute values); + = increment in consumption; − = decrease in consumption

All compounds with DID ≥0.1 were analysed in detail (Figure 2): FQs and MACs represented the most used drugs with documented QT liability within the antibacterial class. The use of FQs carrying a documented risk of TdP increased in all countries except in Slovenia (in which the decrease could probably be related to norfloxacin). This result may be partially explained by the current tendency to prescribe new-generation antibiotics (e.g. levofloxacin and moxifloxacin), albeit they are not first-line agents to treat outpatients with respiratory tract infections. As a matter of fact, their consumption increased in all countries except for levofloxacin in Denmark and moxifloxacin in the UK. By contrast, the overall use of MACs associated with QT-prolonging potential increased only in seven countries despite marketing introduction of telithromycin, which had significant use in France (0.4 DID in 2005), Belgium and Luxembourg (0.3 DID each in 2005).

Figure 2.

Consumption of each antibacterial drug with documented QT liability country by country: comparison between 1998 and 2005 [abscissa: defined daily doses per 1000 inhabitants per day (DID)]. Antibacterial agents with DID ≤0.1 were not included. Relevant consumption (≥2 DID) of some compounds have been outlined. Mtd, metronidazole; Mox, moxifloxacin; Prx, prulifloxacin; Lvx, levofloxacin; Lmx, lomefloxacin; Gpx, grepafloxacin; Nox, norfloxacin; Cpx, ciprofloxacin; Ofx, ofloxacin; Clim, clindamycin; Try, telithromycin; Azy, azithromycin; Cly, clarithromycin; Joy, josamycin; Rxy, roxithromycin; Ery, erythromycin; Cox, cotrimoxazole. 2005 (▪); 1998 ()

Among MACs, exposure to clarithromycin increased in 10 out of 14 countries, with the highest increment in Hungary (up to twofold), where in 2005 it accounted for 2 DID (54% of total group E). Moreover, a peak of 3 DID was observed in Italy. By contrast, in Sweden it reached only 0.1 DID in 2005 (6% of total group E). Erythromycin use showed a decrease in all countries; however, it represented the most used drug in the UK (1.7 DID in 2005).

Discussion

Following an apparent ‘pharmaco-epidemic’ of antimicrobials with QT liability culminating in several regulatory interventions (e.g. withdrawals and warnings on FQs), the list of antibacterial agents with recognized risk of TdP onset has grown rapidly: 21 compounds were identified through literature data mining.

It should be pointed out that cardiac safety is not a novel topic of interest for antimicrobial agents, since most of them (e.g. MACs, ketolides, FQs, azoles, etc.) are known to affect in a significant way cardiac repolarization because of hERG blockade [15, 20, 21]. Moreover, existing guidelines recommended an integrated strategy based on clinical and preclinical studies to reveal the pro-arrhythmic potential of a new compound. Based on results of this integrated risk assessment, a molecule may obtain marketing approval despite being tested positive when assayed on QT liability because of a positive benefit–risk profile. This is the case with moxifloxacin, currently used as a positive control in thorough QT studies to check the sensitivity of clinical trials submitted for regulatory purpose [22, 23].

In clinical practice, TdP liability of individual drugs should be kept in mind, particularly in case of patients at risk because of ‘reduced repolarization reserve’[9] (e.g. inherited long QT syndrome), when the superimposition of a hERG-blocking drug may precipitate arrhythmia. Thus, this potentially fatal event, albeit rare, shared by drugs within the same therapeutic class makes choice of therapy very hard for physicians, especially in case of drugs used worldwide in general practice, such as antibacterial agents.

The present study has shown that, notwithstanding several regulatory measures, in 2005 there was still considerable use of antimicrobials associated with the strongest evidence on QT liability: 10 out of 14 countries showed an increase in consumption over the 8-year period. By contrast, Northern countries (i.e. Sweden, Denmark and Finland) displayed stable, low consumption. The use of group E agents, expressed as percentage of overall antimicrobial use, decrease only in Denmark and the UK in an 8-year period.

The overall trend towards increased consumption was of some concern in Italy, where in 2005 a peak of 6.5 DID was observed for compounds included in group E (an increase of 2.4 DID with respect to 1998). Exposure to clarithromycin accounted for 3 DID in Italy, with the highest increase in Hungary (1 DID) in the 8-year period (Figure 2). The reasons for this phenomenon are at present unclear: the fact that clarithromycin is used in several infective diseases (e.g. community-acquired pneumonia by atypical pathogens and eradication of Helicobacter pylori) may have contributed [24]. This remarkable use of clarithromycin should not be overlooked, not only because of its hERG-blocking properties, but also for its inhibitory effect on cytochrome P4503A4. Metabolic inhibition may indeed contribute to mixed pharmacokinetic–pharmacodynamic interactions with concomitant QT-prolonging agents: these interactions may lead to TdP, especially in patients with reduced repolarization reserve [25].

Differences among European countries in term of drug utilization may be due to several factors such as healthcare system, regulatory practice and prescription habits, as well as the number of marketed products, as recently suggested by Monnet [26]. The large variability in use of antibacterials across Europe has been discussed in recent papers [19, 27]. Based on 2005 consumption data of group E, we could divide countries into two main categories: Denmark, Finland, the Netherlands and Sweden can be considered low-exposure countries (<2 DID) and showed also the lowest pattern of overall consumption of antibacterials. The second group included countries with a higher exposure to group E drugs: Belgium, Luxemburg, Italy and France showed also high overall antibacterial consumption (>20 DID), whereas Germany, Austria, Slovenia, Spain and Hungary used lower amounts of antibacterials.

When considering community exposure, we should take into account both absolute and fractional values of antibacterial use: for example, in 2005 France and Austria had comparable use of antibiotics associated with the strongest evidence of QT liability (3.2 and 3.6 DID, respectively), although the same data represented 11 and 25%, respectively, of the overall use of antibacterials (29 and 14 DID). In order to decrease exposure to drugs with QT liability, one could tentatively consider shifting prescription habits from QT-prolonging drugs to safer agents. To this end, we should not overlook that current European guidelines do not recommend FQs and MACs as first-line therapy of respiratory infectious diseases; β-lactams are indeed preferred to treat nonsevere community-acquired pneumonia and have also no documented evidence on QT liability. By contrast, France may further consider reducing the overall use and proportionally those of QT-prolonging drugs. In fact, in order to decrease microbial resistance, France recently started the process of controlling antibacterial misuse and overuse in the community [28].

The clinical implications of this study should be viewed in the light of some intrinsic limitations of our approach, as already discussed elsewhere [11, 18]. Briefly, apart from potential bias affecting any method based on literature search, consumption data are only an indirect measure of population exposure and, more importantly, actual risk of TdP. To this end, pharmacovigilance databases may help to detect very rare adverse drug reactions such as TdP [29], although it should be acknowledged that over-consumption of an antibacterial with QT liability may not be associated with over-reporting of such a rare event, because of the known bias of spontaneous reporting systems [30–33]. For this reason, we carried this work to identify those antibacterials that warrant further investigation into pharmacovigilance databases because of their known QT liability and large population exposure.

Concerning the criteria that we adopted to classify drugs with QT liability, we acknowledge that they are not officially validated from a regulatory point of view, although they have already been used in previous publications [11, 17]. In any case, all antibacterials labelled as group E in our analysis are listed at http://www.torsades.org, a website where the University of Arizona Center for Education and Research on Therapeutics maintains an updated list of drugs with at least a conditional risk of TdP.

Furthermore, our approach labels as ‘alarming drugs’ (i.e. compounds classified in group E) many molecules regardless of large differences in terms of hERG blockade potency [20, 21] or consistency of clinical evidence. This is mainly caused by the regulatory concern often culminating in warnings, precautions and contraindications for drugs having doubtful evidence of QT liability (i.e. levofloxacin, norfloxacin and lomefloxacin; see Table 2). Thus, QT liability should not be viewed as a class effect of antibacterial agents because there seems to be large intraclass variability, as recently postulated [15, 16]. Therefore, additional studies are necessary to ascribe the individual risk of TdP to each compound: one resource can be data mining of spontaneous reporting systems.

In conclusion: (i) several antibacterial agents mostly belonging to macrolide and fluoroquinolone classes share torsadogenic potential; (ii) significant exposure is still present in the community, and in most countries consumption of antibacterial agents associated with the strongest evidence on QT liability has even increased, particularly in Italy and Hungary; and (iii) among European countries, remarkable differences emerged in the pattern of antimicrobial use carrying a documented risk of TdP. From a regulatory standpoint, this high antibacterial consumption in Southern European countries warrants further investigation of appropriateness of use and suggests closer monitoring of those agents associated with the strongest evidence on QT liability (group E drugs). From a clinical standpoint, these results should prompt careful risk–benefit assessment of each agent within antibacterial class, especially when a drug included in the list is prescribed to patients with known risk factors for TdP occurrence.

Competing interests

None to declare.