The genetics of pro-arrhythmic adverse drug reactions

Abstract

Ventricular arrhythmia induced by drugs (pro-arrythmia) is an uncommon event, whose occurrence is unpredictable but potentially fatal. The ability of a variety of medications to induce these arrhythmias is a significant problem facing the pharmaceutical industry. Genetic variants have been shown to play a role in adverse events and are also known to influence an individual's optimal drug dose. This review provides an overview of the current understanding of the role of genetic variants in modulating the risk of drug induced arrhythmias.

Does pro-arrhythmia have a genetic predisposition?

Drug-induced arrhythmia, also known as pro-arrhythmia, can be identified in up to 5% of patients receiving anti-arrhythmic drugs such as amiodarone, sotalol, flecainide and quinidine as well as a much smaller proportion who take non-cardiovascular medications such as antidepressants, antipsychotics, antibiotics and methadone (Table 1) 1. Clinical risk factors have been identified for drug-induced arrhythmia, and these include gender, pre-existent cardiac disease, bradyarrhythmias and liver disease (Table 2) 2–5.

Table 1.

Prominent examples of drugs that have been found to be associated with drug-induced long QT syndrome (adapted by: Behr & Roden 5; Ramirez et al. 30)

Drug group	Drugs associated with drug-induced arrhythmias
Anti-arrhythmics class I	Quinidine, procainamide, disopyramide, dihydroquinidine, bretyllium
Anti-arrhythmics class III	Sotalol, dofetilide, azimilide, Ibutilide, almokalant
Anti-anginals and vasodilators	Prenylamine, terodiline, lidoflazine, bepridil
Antihypertensives	Nicardipine, isradipine
Antihistamines	Terfenadine, astemizole
Serotonin agonists and antagonists	Cisapride, ketanserin
Antimicrobials
Macrolides	Erythromycin, spiramycin, azithromycin
Antivirals	Ganciclovir, foscarnet
Fluoroquinolones	Moxifloxacin, ciprofloxacin
Antifungals	Ketoconazole, fluconazole, voriconazole
Antimalarials	Halofantrine, quinine sulphate, chloroquine
Psychiatric medications
Tricyclic antidepressants	Amitriptyline, desipramine, imipramine, doxepin,
Antipsychotics	Thioridazine, chlorpromazine, haloperidol, droperidol
Serotonin re-uptake inhibitors	Citalopram, fluoxetine, sertraline, venlafaxine
Anti-emetic and gut motility	Metoclopramide, domperidone
Anticancer	Arsenic trioxide, vandetanibe, eribulin
Others	Probucol, methadone, pentamidine, sirolimus, lithium

Table 2.

Clinical risk factors that have been identified for drug-induced arrhythmias

Demographic factors

Female gender

Age (over 65 years)

Cardiac factors

Myocardial hypertrophy

Heart failure

Conversion of atrial fibrillation to sinus rhythm

Slow heart rhythms (bradycardia)

Undiagnosed congenital long QT syndrome

Pharmacological factors

Increased medication bioavailability

Inhibitory medications

Others

Electrolyte imbalance – hypokalaemia, hypomagnesaemia

Liver disease

Altered function of particular cytochrome P450 (CYP450) isoforms

Drug-induced arrhythmia is predominantly related to torsades de pointes (TdP) and is known as drug-induced long QT syndrome (DILQTS) or drug-induced TdP (DITdP). TdP is a characteristic form of polymorphic ventricular tachycardia that can lead to syncope and sudden cardiac death and is strongly associated with QT interval prolongation 5,6. The QT interval (Figure 1) measures the electrical depolarization, repolarization and plateau phases of the left and right ventricles of the heart on the electrocardiogram (ECG). QT prolongation is defined as the QT_c interval being greater than 440 ms in men and 460 ms in women, although arrhythmias are mainly associated with a QT_c of >500 ms. In the general population the change in the QT interval is associated with risk for sudden cardiac death 7–10. DITdP is similar to TdP seen in congenital LQTS (cLQTS) which is a result of mutations that disrupt cardiac sodium (Na⁺), calcium (Ca²⁺) and potassium (K⁺) ion channel function, such that repolarization of the heart is delayed 11.

Figure 1.

The figure demonstrates an ECG rhythm strip of a heartbeat in which the QT interval is normal in contrast with an abnormal heartbeat in which the QT interval is prolonged. This is correlated with the ventricular and Purkinje myocyte action potential with associated prolongation of phases 2 and 3. The hashed line represents the prolonged QT interval

The depolarization phase of the heart depends mainly on the inward current movement of the Na⁺ channel, the repolarization phase of the heart depends on the outward current movement of the K⁺ channel and the plateau phase depends on the balance between the repolarizing outward K⁺ currents and depolarizing L-type inward currents of the Ca²⁺ channel. Anti-arrhythmic and non-anti-arrhythmic drugs have been demonstrated to block these ion channels leading to diminished currents which can sometimes result in drug-induced adverse reactions. These pro-arrhythmic adverse reactions are likely to be generated due to an association between the patient-specific and drug-specific clinical risk factors 12. Two such factors associated with patient-specific risk include: (i) pharmacokinetic factors which are responsible for regulating the drug concentration transferred to drug receptors and (ii) pharmacodynamic factors which control the response to doses of drugs 13.

One of the basic mechanisms playing an essential role in the repolarization of the human heart is the time-dependent outward rectifying K⁺ current (I_K) 14. Several studies since 1980 have demonstrated that this current has many subtypes of which the most important are: I_Kr and I_Ks (‘r’ stands for rapid and ‘s’ stands for slow) 15,16. Mutations that lead to loss of function in genes underlying even one of these two currents may be responsible for cLQTS. The inhibition of I_Kr is the main mechanism by which prolongation of the QT interval is caused by almost all culpable drugs 17. Pro-arrhythmia resulting from cardiovascular and/or non-cardiovascular drugs could therefore be a result of the inhibition of I_Kr with or without diminished I_K due to mutations in the underlying genes. This was conceptualized as the repolarization reserve by Roden who stated that ‘the complexity of repolarization includes some redundancy’ 18. Thus loss of one ionic current subtype such as I_Kr, will not necessarily result in complete failure of repolarization. As a result individuals with subclinical genetic lesions in different ionic currents such as I_Ks or even other ion currents such as Na⁺ or Ca²⁺, might not demonstrate any overt change in repolarization as suggested by Xiao et al. 19.

Strong evidence for a genetic basis to drug-induced arrhythmia was provided by Kannankeril et al. 20. They found that prolongation of the QT interval in first degree relatives of patients who suffered from DILQTS is greater than in first degree relatives of patients using an anti-arrhythmic drug for a long time without any sign of significant QT prolongation. A number of studies have since been carried out to study genetic variation in DILQTS patients, initially targeting candidate genes which cause cLQTS.

Genetic variation increases the risk of drug-induced arrhythmia

More than 700 mutations in at least 13 genes have been found to be associated with the cause of cLQTS. Six of these genes encode pore-forming ion channels (KCNQ1, KCNH2, SCN5A, KCNJ2, CACNA1C, and KCNJ5) and seven of these genes encode for ion channel subunits or regulatory proteins (KCNE1, KCNE2, ANKB, CAV3, SCN4B, AKAP9, and SNTA1). KCNQ1 and KCNE1 co-assemble to form I_Ks while KCNH2 and KCNE2 co-assemble to form I_Kr. A summary of the studies and genes investigated for association with drug-induced arrhythmias can be found in Table 3.

Table 3.

The genes and associated coding variants that have been found to be associated with drug-induced arrhythmias

Genes	Associated phenotype	Variants	Functional assessment	Studies
SCN5A	LQT3	S1102Y	Yes	Splawski et al., 2002 23
		G1844A, C1852T, T3748C	Yes	Yang et al., 2002 25
KCNH2	LQT2	C2350T	Yes	Yang et al., 2002 25
		P347S	Yes	Paulussen et al., 2004 26
		R1033W	No	Ramirez et al., 2012 30
KCNQ1	LQT1	C1747T	Yes	Yang et al., 2002 25
		Y315C	Yes	Napolitano et al., 2000 24
KCNE1	LQT5	D85N	Yes	Paulussen et al., 2004 26 and Kaab et al., 2011 4
KCNE2	LQT6	T8A	Yes	Paulussen et al., 2004 26, Sesti et al., 2000 21, Abbott et al., 1999 22
NOS1AP	QT interval	rs10919035†	Yes	Jamshidi et al., 2012 42
CACNA1C	LQT8	A1733V	No	Ramirez et al., 2012 30
AKAP9	LQT11	Q3531E	No	Ramirez et al., 2012 30
SNTA1	LQT12	T147N	No	Ramirez et al., 2012 30
KCND3	Brugada syndrome	R566C	No	Ramirez et al., 2012 30
GPD1L	Brugada syndrome	V249M	No	Ramirez et al., 2012 30
RYR2	Catecholaminergic polymorphic ventricular tachycardia	L555V, L2607P, E4361Q	No	Ramirez et al., 2012 30
CACNB2	Short QT syndrome	M1V, I170V	No	Ramirez et al., 2012 30
KCNN3	Atrial fibrillation	F315L	No	Ramirez et al., 2012 30
PPP2R3A		F1000L	No	Ramirez et al., 2012 30
AKAP7		Q112R	No	Ramirez et al., 2012 30
APLP2		R504L	No	Ramirez et al., 2012 30
ATP2A2		S184C	No	Ramirez et al., 2012 30
AKAP6		V839A	No	Ramirez et al., 2012 30
ZFHX3		K3689E, T3640M, H3611Y, L741F, G117S	No	Ramirez et al., 2012 30
JPH3		R656W	No	Ramirez et al., 2012 30
CALR		D165G	No	Ramirez et al., 2012 30
JPH2		V345L, T286A	No	Ramirez et al., 2012 30

LQTX, subtype of the congenital long QT syndrome;

^†Non-coding variant associated with dLQTS.

Sesti et al. screened the KCNE2 gene in 98 cases of drug-induced arrhythmia and 200 control subjects 21. They identified three sporadic mutations (M54T, I57T and A116V) and one more common rare variant (T8A). T8A was found in only 16 of 1010 (1.6%) control individuals and 1 of 230 (0.4%) index patients with LQTS. Functional and in silico analyses demonstrated that although the mutations produced reductions in current density, none of the three sporadic mutations altered the susceptibility to drug inhibition. On the other hand, only a 15% decrease in current density was observed with the T8A variant, and patients with this genotype have a normal QT interval at baseline 22. Interestingly, on drug exposure this variant increased the inhibitory effects of antibiotic treatment and demonstrated that common sequence variations can be clinically silent before drug exposure yet increase the risk of DILQTS.

A non-synonymous variant, S1102Y in the SCN5A gene, was associated with an increased risk of sudden cardiac death (SCD) in African-American adults and children although it is rarer or even absent in other ethnic groups. Splawski et al. screened cases of cardiac arrhythmia (predominantly due to medication) and S1102Y heterozygosity was associated with an eight-fold higher risk of arrhythmia than in non-carriers of the variant 23. Functional and in silico studies of Y1102 and computational analysis of simulated action potentials generated by S1102/Y1102 channels demonstrated prolonged repolarization and early after depolarizations (EADs) associated with the Y1102 variant.

At the same time, an early candidate gene study was performed in 92 subjects who had experienced drug-induced arrhythmia and 157 healthy subjects in three genes firmly associated with cLQTS. The study identified mutations in five of the 92 subjects 25; three of them were in the SCN5A gene (G1844A, C1852T, T3748C), one in the KCNQ1 gene (C1747T) 24 and one in the KCNH2 gene (C2350T). Four polymorphisms were also identified but there was no difference in the frequency between the cases and the controls 25.

A candidate gene study by Paulussen et al. screened the five main cLQTS associated genes in 32 drug-induced arrhythmia patients with confirmed TdP and identified three missense variants: T8A in KCNE2, D85N in KCNE1 and P347S in KCNH2 26. T8A and D85N have been implicated with drug-induced arrhythmias in earlier studies as well as in patients suffering from QT prolongation 21,22. Three additional non-synonymous variations were found in both the controls and the cases with similar allele frequencies in the two populations in the KCNE1, KCNH2 and SCN5A genes. These variants have also been previously reported in earlier studies in patients suffering from cLQTS 25,27–29.

Kaab et al. confirmed the association of the KCNE1-D85N variant and drug-induced arrhythmia in 176 cases and 207 drug-exposed controls (individuals without QT prolongation when exposed to culpable medications) and 837 controls from the general population 4. The study included 1424 single nucleotide polymorphisms (SNPs) in 18 candidate genes (which included ion channels and other high priority candidate genes) for which the subjects were genotyped. The most statistically significant associated SNP was rs7295250 in the CACNA1C gene (odds ratio 1.88, P = 7.62 × 10⁻⁴). The strongest effect size within the study was detected for D85N (odds ratio 8.88, P = 1.95 × 10⁻⁵). The follow-up validation study failed to replicate the initial association although a trend for the D58N variant was observed (P = 0.58).

A more recent study by Ramirez et al. utilized a next generation sequencing panel of 79 genes associated with arrhythmia syndromes including 13 genes which have been previously associated with cLQTS, nine additional genes shown to be involved in congenital short QT syndrome (cSQTS) and Brugada syndrome and nine other genes related to familial arrhythmia syndromes 30. In 11/31 DILQTS (36%) subjects a novel missense mutation in genes known to be associated with congenital arrhythmia was found. In 6/26 (23%) DILQTS Caucasian subjects a conserved deleterious variation or an already identified congenital arrhythmia mutation was found. These variants were found in less than 2% of the individuals sequenced as part of the 1000 Genomes project. The conclusion from this study was that the rare variants in arrhythmia related genes, not just cLQTS genes, are responsible for a significant proportion of drug-induced arrhythmia cases. These conclusions may be limited, however, by the different platforms and depth of sequencing in the case and control groups. Overall, however, around 10% of cases may be associated with rare variation in the cLQTS genes 5.

Common variation in genes associated with QT prolongation has also been investigated as a potential cause of drug-induced arrhythmias. The NOS1AP (CAPON) gene has been repeatedly associated with QT prolongation in healthy individuals 31–33 and it has also been associated with severity in patients suffering from cLQTS 34. NOS1AP is a regulator of neuronal nitric oxide synthase (nNOS), which in turn regulates the intracellular levels of calcium and myocyte contraction in the heart 35–37. NOS1AP is thought to alter cardiac repolarization by interaction with nNOS by blocking L-type calcium channels 38–41. This may explain the association of NOS1AP gene variants with prolongation of the QT interval and TdP.

Jamshidi et al. demonstrated that a common variant (rs10919035, odds ratio 5.5, P = 3.0 × 10⁴) in the NOS1AP gene which has been previously associated with the QT interval prolongation plays an important role in the risk of amiodarone-induced drug-induced arrhythmia 42. The study included 86 cases, 192 controls and 68 drug-exposed individuals with no QT prolongation. The frequency of the common variant in cases was 27.8% and in controls 7.1%. This supports the theory of repolarization reserve and its role in DITDP.

Drug-induced arrhythmia associated with drug metabolism

Anti-arrhythmic drugs that are known to cause drug-induced arrhythmia as an adverse reaction include mainly class III anti-arrhythmic agents which prolong the QT interval by blocking the I_Kr current. Variability in the response to drug therapy can be observed in patients. This variability may result not only from end-organ effects but also from drug concentration differences 43. Prolonged QT intervals following drug therapy in some individuals can therefore potentially be caused by common variants that alter drug metabolism thereby increasing drug concentration 44,45.

There are many genetic variants responsible for altered biological activity of enzymes belonging to the hepatic cytochrome P450 enzymes such as CYP3A4, CYP3A5 and CYP3A7. CYP2D6 activity is absent in 5–10% of Caucasian and African individuals due to underlying genetic variants in the CYP2D6 gene. This enzyme is responsible for metabolizing thioridazine, an anti-psychotic drug that has been discontinued in the UK due to its risk of TdP. It is therefore possible that the dysfunctional metabolic pathway may lead to QT prolongation and perhaps drug-induced arrhythmia 46. Paulussen et al. reported an individual with already compromised repolarization reserve due to a KCNH2-P347S mutation 26. This subject received a combination of drugs that both prolonged the QT interval 47: cisapride and clarithromycin. In addition, both of these drugs are metabolized by the same enzyme which is inhibited by clarithromycin thereby slowing the metabolism of cisapride. This exacerbates I_Kr blockade by increasing the concentrations of circulating cisapride in plasma 48 and offers a model for how genetic variability in the capacity for drug metabolism may work. As yet, however, there are no genetic data to support metabolic predisposition as a cause for TdP.

Genome-wide association studies (GWAS) and drug-induced arrhythmia

More recently the GWAS approach has been employed to investigate the association of the genetic risk factors across the genome with disease. GWASs that focus on common disorders have the advantage of including thousands of cases, whereas GWASs that focus on rarer disorders such as drug-induced adverse reactions do not involve many cases. Nonetheless, there are phenotypes where the GWAS approach has been successful with less than one hundred individuals 49,50.

A recent meta-analysis of 10 cohorts including 33 781 participants of European descent was performed to examine whether common genetic variants modify the effects of exposure to drugs on QT interval. The study examined four drug classes previously associated with QT prolongation, with the estimated prevalence of drug exposure highest for diuretics (13.6%). The study by Avery et al. found no genome-wide significant interactions for any of the four drug classes 51. An analysis limited to common variants identified in previously reported GWAS of QT interval yielded similarly null results, as did one restricted to recent pharmacogenomic studies of QT and drug-induced QT prolongation. Statistical power simulations suggested that at least for the diuretics class, if the genetic variation was common, the study had 80% power to detect an effect of 3.25 ms. However, this size of genetic effect is typically outside the range usually observed for QT interval. Furthermore, the study did not address the potential for bias related to duration of drug use, particularly as participants taking the drugs for years or decades are those least likely to have experienced side effects. Nevertheless, the results do suggest that increasing the number of measures per participant in longitudinal modelling may help to increase statistical power.

A more recent effort to investigate the role of common genetic variation on risk of DITdP in a GWAS concluded that common genetic variation with a strong effect-size did not play a role in the risk of drug-induced arrhythmias across all drug groups 6. The study included 264 Caucasian cases selected from two groups, the Trans-Atlantic Alliance against Sudden Death (the Fondation Leducq) and the Drug-induced Arrhythmia Risk Evaluation (DARE) study and 5350 controls, some of whom received medications showing no considerable QT interval prolongation and others general population controls. There were no genome-wide associations between common SNPs or genes and the risk of DITdP.

Conclusions

DILQTS remains a significant cause of morbidity and mortality, and is the major reason for the withdrawal of drugs from the market. Knowledge of rare and common variants that may modulate patients' susceptibility to adverse drug reactions such as DILQTS may allow clinicians to identify prospectively subpopulations at risk in the future. A number of genetic studies have determined that DILQTS is predisposed to by ethnically specific common variants found in genes modulating the length of the QT interval (SCN5A S1102Y, KCNE1 D85N and NOS1AP rs10919035). Rare variation in arrhythmia genes also appears to have a role in approximately 10% of cases. Attempts to investigate the role of common variants in a genome wide analysis has not been successful and will most likely require recruitment of substantially larger number of cases, with drug-specific analyses 6. However, the use of whole exome sequencing in subjects with DILQTS might be worth considering for future studies as it would reveal novel and low frequency alleles that may carry a genetic risk for a disease, whereas GWAS can only interrogate risk markers that are more common in the population. Recent breakthroughs in stem cell biology have led to a number of papers showing the possibility of using patient-derived induced pluripotent stem cells to further our understanding of the underlying mechanisms of pro-arrhythmia 52–54. These studies may identify novel target genes involved in drug sensitivity and pro-arrhythmia which may warrant investigation for underlying genetic variants that can modulate their expression.

Competing Interests

Research funds were received from the international Serious Adverse Events Consortium.

YJ is supported by the British Heart Foundation (Project Grant 29615).