Objective A prolonged QTc (LQT) is a surrogate for the risk of torsade de pointes (TdP). QTc interval duration is influenced by sex hormones: oestradiol prolongs and testosterone shortens QTc. Drugs used in the treatment of breast cancer have divergent effects on hormonal status.
Methods We performed a disproportionality analysis using the European database of suspected adverse drug reaction (ADR) reports to evaluate the reporting OR (ROR χ2) of LQT, TdP and ventricular arrhythmias associated with selective oestrogen receptor modulators (SERMs: tamoxifen and toremifene) as opposed to aromatase inhibitors (AIs: anastrozole, exemestane and letrozole). When the proportion of an ADR is greater in patients exposed to a drug (SERMs) compared with patients exposed to control drug (AIs), this suggests an association between the specific drug and the reaction and is a potential signal for safety. Clinical and demographic characterisation of patients with SERMs-induced LQT and ventricular arrhythmias was performed.
Results SERMs were associated with higher proportion of LQT reports versus AIs (26/8318 vs 11/14851, ROR: 4.2 (2.11–8.55), p<0.001). SERMs were also associated with higher proportion of TdP and ventricular arrhythmia reports versus AIs (6/8318 vs 2/14851, ROR: 5.4 (1.29–26.15), p:0.02; 16/8318 vs 12/14851, ROR: 2.38 (1.15–4.94), p:0.02, respectively). Mortality was 38% in patients presenting ventricular arrhythmias associated with SERMs.
Conclusions SERMs are associated with more reports of drug-induced LQT, TdP and ventricular arrhythmias compared with AIs. This finding is consistent with oestradiol-like properties of SERMs on the heart as opposed to effects of oestrogen deprivation and testosterone increase induced by AIs.
Trial registration number NCT03259711.
- Long QT syndrome
- torsade de pointes
- ventricular arrhythmias
- selective estrogen receptor modulators
- aromatase inhibitors
- sex steroid hormones
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Long QT syndrome
- torsade de pointes
- ventricular arrhythmias
- selective estrogen receptor modulators
- aromatase inhibitors
- sex steroid hormones
Cardiac repolarisation, measured by QT interval duration corrected for heart rate (QTc) on surface ECG,1 is influenced by sex steroid hormones.2–4 Long QT (LQT) is a surrogate for the risk of torsade de pointes (TdP).2 In women, oestradiol is associated with QTc increase while testosterone is associated with QTc shortening.2 It has been shown that contraceptive pills might influence QTc depending on their antiandrogenic properties.5 Selective oestrogen receptor modulators (SERMs) and aromatase inhibitors (AIs) are cornerstone endocrine drugs used for hormone sensitive breast cancer treatment.6 SERMs used in clinical practice are mainly tamoxifen and toremifene, to a lesser degree.6 7 SERMs have variable agonists/antagonists effects depending on the target tissue, being antagonists for breast and agonists for metabolism and cardiovascular organs.6 8 AIs block peripheral conversion of testosterone into oestradiol, leading to decreased oestradiol and increased testosterone levels.6 9 AIs, that is, anastrozole, letrozole and exemestane are used in postmenopausal women first line or when there is a disease progression following tamoxifen therapy.6 9 Therefore, due to their peculiar pharmacology, SERMs are expected to increase LQT and TdP compared with AIs.
We performed a disproportionality analysis using the European-database of suspected adverse drug reaction (ADR) reports to evaluate the reporting OR (ROR) of LQT, TdP and ventricular arrhythmias as a function of SERMs or AIs use. European-database of suspected ADR reports is a publicly accessible portal designed to search data on suspected ADR for authorised medicinal products in the European Economic Area (http://www.adrreports.eu/fr/search_subst.html). Case/non-case analyses were performed for SERMs (tamoxifen and toremifene) versus AIs (anastrozole, exemestane and letrozole). This method compares the proportion of specific ADR reported for these two groups. Reactions are based on the medical dictionary for regulatory activities classification dictionary of terms for side effects. Disproportionality was estimated by calculating the ROR χ2 (table 1). If the proportion of an ADR is greater in patients exposed to a drug (SERMs) than in patients not exposed to this drug (AIs), this suggests an association between the specific drug and the reaction and is a potential signal for safety.10–12 We searched for ADR related to suspected drug-induced LQT, TdP and ventricular arrhythmias. Selected medical dictionary for regulatory activities classification terms used for definition of ventricular arrhythmias were: ventricular arrhythmias, ventricular tachycardia, ventricular fibrillation and TdP. Selected medical dictionary for regulatory activities classification terms used for definition of LQT were: LQT syndrome, ECG QT prolonged or ECG QT abnormal.
Clinical and demographic characterisation of the cohort of patients with suspected SERMs-induced LQT, TdP and ventricular arrhythmias was performed. Details were extracted from VigiBase,13 where data from European database of suspected ADR reports are usually transferred electronically on a daily basis. VigiBase is the international pharmacovigilance database managed by the WHO’s Uppsala Monitoring Center. Complementary information and relevant unreported cases to European database of suspected ADR reports or VigiBase were searched in PubMed, last accessed December 2017.14
Analysis of European database of suspected ADR reports from 12/2001 to 08/2017 revealed 23 169 individual safety case reports (ISCRs) of suspected ADR for SERMs (n=8318) and AIs (n=14851). These ISCRs mainly involved adult (18–64 years old: 50.3%, 65–85 years old: 44.7%, >85 years old: 4.4%, <18 years old: 0.6%) women (97.7%). Adult women on SERMs were younger compared with women on AIs (18–64 years old: 54.3% vs 48.3%, p<0.0001). For SERMs, most of these ISCR were associated with tamoxifen (95.2%). For AIs, repartition of ISCRs by molecule was: letrozole (44.4%), anastrozole (33.1%) and exemestane (22.5%). For ISCRs reported on SERMs, 26 were LQT (24 on tamoxifen), 6 were TdP (5 on tamoxifen), and 16 were ventricular arrhythmias (14 on tamoxifen). For ISCRs reported on AIs, 11 were LQT (9 on letrozole and 2 on exemestane), 2 were TdP (1 on exemestane and 1 on anastrozole) and 12 were ventricular arrhythmias (6 on letrozole, 5 on anastrozole and 1 on exemestane).
SERMs were associated with higher proportion of LQT compared with AIs (26/8318 vs11/14851, ROR: 4.2 (2.11–8.55), p<0.001; figure 1). SERMs were also associated with higher proportion of TdP and ventricular arrhythmias compared with AIs (6/8318 vs 2/14851, ROR:5.4 (1.29–26.15), p:0.02; 16/8318 vs 12/14851, ROR:2.38 (1.15–4.94), p:0.02; respectively; figure 1). More details concerning specific numbers of ISCR as a function of medical dictionary for regulatory activities classification terms and molecules are detailed in table 2.
Clinicodemographic details concerning ISCRs of LQT (n=30) and ventricular arrhythmias (n=21) associated to SERMs extracted from the international pharmacovigilance database (where information from European Medicines Agency are transferred) and PubMed (through December 2017) are detailed in tables 3 and 4. LQT and ventricular arrhythmias overwhelmingly affected women with breast cancer on tamoxifen taking standard dose (10–30 mg/day, oral) with no other culprit drugs or liable conditions (tables 3 and 4). Time to onset between SERM intake and LQT apparition could be as fast as 24 hours, and the median time was of a week. Information concerning LQT evolution after SERM withdrawal was available in 10 patients. In these latter, QTc duration normalised after stopping SERM. Recurrence of LQT after SERM rechallenge was reported in one women. In two women with breast cancer, SERM was switched to an AI with subsequent normalisation of QT duration. In one women, genetic study of enzymes involved in metabolism of tamoxifen identified a poor metaboliser status for CYP2D6, CYP3A, UGT2B7 and UGT1A1. In two other women, it was hypothesised that tamoxifen-induced LQT resulted from a pharmacokinetic interaction leading to CYP3A4 inhibition by an interacting drug (acitretin, norfloxacin). Of note, SERM-induced ventricular arrhythmias lead to life-threatening situations with recovery in 24% and death in 38% of patients.
The main finding of this study is that SERMs were associated with higher proportion of ADR reports related to LQT, TdP and ventricular arrhythmias compared with AIs in European database of suspected ADR reports (figure 2). Almost all these reports on SERMs were observed with standard oral dose of tamoxifen which is much more widely used than toremifene.6 7 However, the overall number of events were small but in an era where treatment for hormone-positive breast cancer is moving towards combination therapy where SERMs and AIs are combined with other therapies (such as kinase inhibitors) which have their own effects of QT, these results can be more significant for the cancer population.
Tamoxifen was approved before the systematic requirement for dedicated thorough QTc studies, and there is also no information on QTc measurements from large clinical trials evaluating the efficacy of tamoxifen in breast cancer.7 To date, tamoxifen is often considered by tertiary drug information sources as a drug with intermediate effects on QTc interval. US Food and Drug Administration label does not mention any warning, precaution of use in population at risk or particular ECG monitoring while using tamoxifen. Our finding reveals the potential for tamoxifen to induce ventricular arrhythmias, particularly TdP. TdP induced by tamoxifen has rarely been reported in the literature. Our finding is in line with three case reports of acquired LQT on tamoxifen used at standard dose (20 mg/day).14 Two phase I trials investigating use of high-dose tamoxifen (80–680 mg/m²/day) in advanced epithelial tumours (n=53) and paediatric malignant gliomas (n=14) have shown that up to 40% of patients had a significant QT prolongation (≥10%) and about 10% had very marked QTc increase (prolongation ≥20%).15 16 In these trials, one patient had sudden death associated with QTc prolongation and premature ventricular contractions.15 16 Development of tamoxifen in these indications was not pursued. A thorough QT study conducted in men showed that toremifene (differing from tamoxifen for one chlorine atom) prolongs the QTc interval in a dose-related and concentration-related manner.7
The influence of tamoxifen and/or its active metabolite (endoxifen) on ventricular repolarisation has been studied in several animal models.17–21 Data mainly showed a rapidly activating delayed rectifier potassium channel (IKr) inhibition, a mechanism also involved in oestradiol-induced2 22 and in most drug-induced LQTs (figure 2).2 23 In human embryonic kidney cells expressing IKr currents, tamoxifen and endoxifen inhibited IKr currents by direct channel blockage and by disruption of channel trafficking to the plasma membrane in a concentration-dependent manner.19 IKr blockage induced by tamoxifen was also found in Xenopus laevis oocytes expressing IKr channels.20 However, it has also been shown in ventricular myocytes that tamoxifen inhibited sodium (in rat)17 and L-type calcium (canine18 and rabbit21) currents, beyond potassium currents inhibition. This multichannel blockade induced by tamoxifen might counterbalance effects of isolated IKr inhibition17 18 21 and, therefore, might result in a mild ventricular repolarisation prolongation and explain a modest QTc prolongation in clinical settings when there is no other liable conditions potentiating LQT.
Extrapolating preclinical models to humans is tricky, in part due to potential differences in the composition of cardiac ionic channels among the different species.24 In addition, given the large number of women with breast cancer requiring SERMs for a period ranging from 5 to 10 years,5 it can be expected that these patients will take a number of other drugs at risk for TdP.23
Our findings are further supported by previous data showing that AIs increase testosterone and decrease oestradiol levels resulting in ventricular repolarisation shortening (figure 2).2 22 Accordingly, Kurokawa et al recently showed that ablation of circulating levels of oestradiol in aromatase knockout mice blunted the QT prolonging effect of E-4031, an IKr blocker.22 Moreover, it has been shown that high-dose anastrozole leads to QT shortening in dogs.25
There are limitations to our work based on EMA and VigiBase reports. The sources are non-homogeneous, and there is limited possibility of verification of the rhythm abnormalities by means of ECG tracings or completeness of reporting for comorbid conditions and concomitant drugs. The exact denominator and clinical characteristics of patients exposed to SERMs or AIs cannot be evaluated, but in the present situation, it was mainly women patients with breast cancer. Total number of ISCRs for each drug is used as the denominator for this kind of disproportionality analysis in pharmacovigilance databases for signal detection.10–12 The volume of reports for a particular medicinal product may be influenced by the extent of use of the product, publicity, the nature of the reactions and other factors such as competition bias. Even though magnitude of disproportionality (ROR) for LQT and TdP have been associated to extent of IKr drug-induced blockade using VigiBase11 or to identification of over-reporting for LQT and TdP on antiandrogenics versus testosterone using EMA,4 there is still a risk that comparisons of disproportionality between medicinal products in pharmacovigilance databases may be misleading. Older age is expected to contribute to higher proarrhythmia risk.2 26 Therefore, the age difference found between AIs versus SERMs users is unlikely to bias our results, but status for other risk factors for TdP between AIs versus SERMs users are unknown.
In conclusion, LQT and TdP were more likely to be reported on SERMs as compared with AIs. Further investigation may be needed to better understand risk of LQT or TdP in high-risk patients such as carrier of congenital LQT syndrome or when SERMs are used in combination with other QT prolonging drugs, such as new anticancer drugs developed for breast cancer. In addition, AIs appear to be a therapeutic alternative in patients with breast cancer developing LQT and TdP on SERMs. ECG monitoring should be considered in patients on SERMs with a risk of TdP.
What is already known on this subject?
Cardiac repolarisation assessed by QTc interval duration is influenced by sex steroid hormones: oestradiol prolongs QTc and testosterone shortens it.
Drugs used in the treatment of breast cancer have divergent effects on hormonal status.
What might this study add?
The effects of selective oestrogen receptor modulators (SERMs) compared with aromatase inhibitors (AIs) on QT duration and torsade de pointes (TdP) risk are unknown.
We showed that SERMs are associated with more reports of drug-induced long QT (reporting OR: 4.2, p<0.001), TdP (reporting OR: 5.4, p:0.02) and ventricular arrhythmias (reporting OR: 2.38, p:0.02) compared with AIs.
This finding is consistent with oestradiol-like properties of SERMs on the heart as opposed to effects of oestrogen deprivation and testosterone increase induced by AIs.
How might this impact on clinical practice?
AIs might be a better alternative to SERMs in patients with breast cancer at risk for long QT and TdP.
This issue will be more critical as endocrine therapy is combined with other therapies for the treatment of breast cancer.
ECG monitoring should be considered in patients on SERMs with a risk of TdP.
Funding This study was supported by Philippe Foundation and Fondation Recherche Medicale “SPE20170336816”.
Disclaimer The information does not represent the opinion of the European Medicines Agency or World Health Organization.
Competing interests JJM (Consultant: Novartis,Pfizer, Bristol Myers Squibb, Takeda). The other authors have nothing to disclose.
Patient consent Not required.
Ethics approval The use of confidential electronically processed patient data was approved by the French National Commission for Data Protection and Liberties (Commission Nationale de l’Informatique et des Libertés; reference:1922081).
Provenance and peer review Not commissioned; externally peer reviewed.