Cardiovascular concentration–effect relationships of amodiaquine and its metabolite desethylamodiaquine: Clinical and preclinical studies

Aims Amodiaquine is a 4‐aminoquinoline used extensively for the treatment and prevention of malaria. Orally administered amodiaquine is largely converted to the active metabolite desethylamodiaquine. Amodiaquine can cause bradycardia, hypotension, and electrocardiograph QT interval prolongation, but the relationship of these changes to drug concentrations is not well characterized. Methods We conducted a secondary analysis of a pharmacokinetic study of the cardiac safety of amodiaquine (10 mg base/kg/day over 3 days) in 54 Kenyan adults (≥18 years) with uncomplicated malaria. Nonlinear mixed effects modelling was used to assess amodiaquine and desethylamodiaquine concentration–effect relationships for vital sign (pulse rate, blood pressure) and electrocardiograph interval (QT, QRS, PR) outcomes. We also measured the spontaneous beating heart rate after cumulative dosing of amodiaquine and desethylamodiaquine in isolated mouse atrial preparations. Results Amodiaquine and desethylamodiaquine caused concentration‐dependent mean decreases in pulse rate (1.9 beats/min per 100 nmol/L; 95% confidence interval: 1.5–2.4), supine systolic blood pressure (1.7 mmHg per 100 nmol/L; 1.2–2.1), erect systolic blood pressure (1.5 mmHg per 100 nmol/L; 1.0–2.0) and erect diastolic blood pressure (1.4 mmHg per 100 nmol/L; 1.0–1.7). The mean QT interval prolongation was 1.4 ms per 100 nmol/L irrespective of correction factor after adjustment for residual heart rate dependency. There was no significant effect of drug concentration on postural change in blood pressure or PR and QRS intervals. In mouse atria, the spontaneous beating rate was significantly reduced by amodiaquine (n = 6) and desethylamodiaquine (n = 8) at 3 μmol/L (amodiaquine: 10 ± 2%; desethylamodiaquine: 12 ± 3%) and 10 μmol/L (amodiaquine: 50 ± 7%; desethylamodiaquine: 46 ± 6%) concentrations with no significant difference in potency between the 2 compounds. Conclusion Amodiaquine and desethylamodiaquine have concentration‐dependent effects on heart rate, blood pressure, and ventricular repolarization.

Conclusion: Amodiaquine and desethylamodiaquine have concentration-dependent effects on heart rate, blood pressure, and ventricular repolarization. The cardiovascular effects of AQ have been recognized from the earliest studies in animal models. 1 During its development, pulsus bigeminus was noted in anaesthetized dogs receiving high doses of parenteral AQ. 1 Like chloroquine, AQ exhibits anti-arrhythmic properties, terminating experimental atrial arrhythmias in both decentralized and innervated canine hearts 6 but, unlike chloroquine, does not appear to protect against experimental ventricular arrhythmias. 7 Electrocardiograph QT interval prolongation, 7 bradycardia 8 , and hypotension 8 have also been observed after parenteral AQ administration

What is already known about this subject
• Amodiaquine is a 4-aminoquinoline antimalarial used extensively for the treatment and prevention of malaria with an excellent track record of cardiac safety.
• Bradycardia, hypotension and electrocardiograph QT interval prolongation have been observed after amodiaquine administration in humans and animals.
• A causal role of amodiaquine and its main active metabolite desethylamodiaquine for these cardiovascular effects has been proposed but has yet to be defined.

What this study adds
• We characterized the concentration dependency of the bradycardic, hypotensive and QT prolonging effects of amodiaquine and desethylamodiaquine in clinical and preclinical studies providing evidence of their causal role.
• In a nonlinear mixed effects modelling reanalysis of a cardiac safety study of amodiaquine for uncomplicated malaria, amodiaquine and desethylamodiaquine drug concentrations were associated with mean decreases in pulse rate and blood pressure along with a mean increase in electrocardiograph QT interval prolongation.
• In studies of spontaneously beating murine atrial preparations, amodiaquine and desethylamodiaquine had direct concentration-dependent bradycardic effects of similar potency.
to anaesthetized dogs and cats. We have reported recently that AQ prolongs the QT interval less, but is more bradycardic and hypotensive than chloroquine at current standard oral malaria treatment doses when given to adolescents 9 and adults. 10 The clinical significance of these cardiovascular effects with standard malaria treatment dosing is unclear 10 although bradycardia and hypotension may contribute to the higher incidence of mild asthenia and asthenia-like reactions after AQ compared to other antimalarials. [11][12][13] Direct multiple ion channel blockade of cardiac [14][15][16][17][18] and vascular 19 myocytes along with altered autonomic tone 20,21 may both be relevant. There are few data on the cardiovascular pharmacology of AQ, and information on its main metabolite desethylamodiaquine is especially limited, as the metabolite was only identified in the 1980s. 22 This was around the time reports of the fatal toxicity of AQ in chemoprophylaxis [23][24][25] led to its temporary withdrawal in 1990 26 from the list of WHO-recommended antimalarials.
We conducted a secondary analysis of a clinical pharmacokinetic study of ASAQ in adult malaria patients focusing on electrocardiographic interval (RR, QT, QRS and PR) and cardiovascular vital sign (pulse rate and blood pressure) outcomes. This analysis of the clinical study was complemented by laboratory assessment of the concentration-heart rate response in murine atrial preparations with intact sinoatrial node (SAN).  Exact sample times were recorded and used in the pharmacokinetic modelling.

| Pharmacokinetic methods
Venous plasma samples were analysed using liquid chromatographytandem mass spectrometry methods. AQ, desethylamodiaquine and the internal standard (AST-D4) were analysed by reversed-phase liquid chromatography (X Terra C18 MS -3.5 μm; 50 Â 3 mm id) and tandem mass spectrometry (Sciex API 3000) detection in the Turbo Ion Spray positive mode.

| Ethics
The trial protocol was approved by the Kenya Medical Research Institute (KEMRI) Ethical Review Committee. Additional ethical approval for this secondary analysis of fully anonymized individual patient data was not deemed necessary in keeping with University of Oxford Central University Research Ethics Committee guidance.

| Data analysis
Trial data were standardized and checked according to a specified data dictionary (Supporting Information) for analysis. Measurements from fixed and nonfixed dose ASAQ arms were pooled as these dose formulations are known to be bioequivalent for AQ and to not have an effect on the pharmacokinetic parameters of these drugs. 28,29 AS coadministration does not have a significant effect on the bioavailability of AQ. 29 It is also generally accepted that AS does not have a significant effect on the QT interval. 30 In view of the inverse relationship between the QT interval and heart rate, measured QT intervals were adjusted for heart rate with the widely used Bazett QTcB ¼ QT ffiffiffiffi A study-specific correction formula (QTcS ¼ QT RR 0:42 ) was also applied with the correction exponent derived from log-log linear regression (Supporting Information). The QT interval was analysed as adjusted with the study-specific (QTcS), Fridericia (QTcF) and Bazett (QTcB) heart rate correction formulae.
All statistical analyses and data visualization were done in R 31 Version 3.6.0, with linear mixed effects modelling conducted using the nlme 32 package. Model fit was assessed by visual inspection of residuals, whereas model discrimination was on the basis of likelihood ratio tests with P < 0.05 as the threshold for statistical significance.  38 The drug effect was evaluated using the total predicted concentration of AQ plus its metabolite desethylamodiaquine based on prior evidence that both AQ 6,7 and desethylamodiaquine 39 affect cardiovascular physiology as well as lack of evidence for any substantial difference in their activities from descriptive analyses. In the ECG interval models, the other fixed effects were body temperature change, age and sex, with the addition of RR interval change to adjust for residual heart rate changes. For cardiovascular vital signs measured at multiple time points after recovery from malaria, the malaria disease effect was incorporated as a binary categorical fixed effect variable present during Days 0-2 of treatment, with the addition of body temperature change and sex for the change in the pulse rate model only.  Table 1.

| Preclinical
Fifty-three completed the full treatment course. One patient in the fixed-dose ASAQ arm withdrew consent. Four patients were subsequently lost to follow-up and censored in analyses at the time of dropout. All patients recovered uneventfully. There were no serious cardiovascular events reported in any of the 54 patients.  (Table S1). Goodness-of-fit diagnostics and the prediction-corrected visual predictive checks ( Figures S7-S9) demonstrated that the model described the observed data adequately.

| Concentration-effect analyses: Cardiovascular vital signs
Plasma concentrations of AQ and desethylamodiaquine were summed. The population mean maximum plasma total concentration (C max ) of AQ and desethylamodiaquine was approximately 750 nmol/L (or 250 ng/mL; Table S1). This was associated with a mean decrease in pulse rate of 14.6 beats/min (95% confidence interval [CI]: 10.9-18.2). This effect was in addition to independent effects on pulse rate reduction following recovery from fever (5.7 beats/min per 1 C decrease; 95% CI: 3.7-7.6) and from acute malaria (3.0 beats/min; 95% CI: 0.5-5.5). Male sex was not associated with statistically significant effects on pulse rate compared with female sex at this sample size ( Table 2).
After adjusting for acute malaria effects, the total plasma concentration of AQ plus desethylamodiaquine at C max was associated with a  (Tables S2 and S3).

| Concentration-effect analyses: ECG intervals
After adjustment for change in body temperature, age, sex and change in RR interval, the mean-corrected QT interval prolongation resulting  .0985 a Mean maximum total plasma drug concentration (rounded) after a 3-day course of amodiaquine from pharmacokinetic analysis of same study.

T A B L E 3 Multivariable linear mixed effects regression analysis of the corrected QT interval in malaria following treatment with amodiaquine
QTcS-Study-specific correction (ms)

| DISCUSSION
The 4-aminoquinoline antimalarial AQ is a prodrug converted rapidly after oral administration to its active metabolite desethylamodiaquine by cytochrome P450 isozyme 2C8 (CYP2C8). 40

| Cardiovascular vital signs
Bradycardia is a common cardiovascular effect after antimalarial treatment with AQ and mefloquine. 44 It is observed more in adolescents and adults than in children 10,45 for reasons which are not fully understood, although modulation of cardiac ion currents by sex hormones may play a role. 46 Our murine atrial studies show that both AQ and desethylamodiaquine have direct concentration-dependent bradycardic effects of similar potency. These effects are greater than that of hydroxychloroquine at the concentrations evaluated, measured using the same method in our previous study. 15 The AQ-induced bradycardia in both F I G U R E 1 Change (%) in atrial beating rate during cumulative doses of 4-aminoquinoline antimalarials compared to controls. DMSO, dimethyl sulfoxide; *P < 0.05 innervated (malaria patients, 9 anaesthetized dogs 8 ) and decentralized (mouse) hearts supports a direct pharmacological effect on cardiac myocyte ion channels through modulation of the pacemaker I f current at the SAN, as observed previously with hydroxychloroquine. 15 Autonomic tone may also be relevant as both AQ and mefloquine are associated with reversible inhibition of human acetylcholinesterase, with AQ having a much higher potency than mefloquine. 20,21 The quinoline antimalarials quinine 47 and chloroquine 48 are known to cause lethal hypotension when injected rapidly but can be used safely with rate-controlled continuous intravenous infusion. As proposed for AQ, 19 these hypotensive effects are likely due to vasodilation and negative inotropy from multiple ion channel blockade. 49 Orthostatic hypotension is a feature of acute malaria, which is exacerbated by the quinoline antimalarials quinine and mefloquine. 50 However, in this study of uncomplicated malaria infections, there was no significant effect of the total plasma concentration of AQ and desethylamodiaquine on postural changes in systolic or diastolic blood pressure after adjustment for malaria recovery.
AQ is the most bradycardic of the front-line antimalarials. 10 Although AQ and desethylamodiaquine cause concentrationdependent bradycardia and hypotension in adult malaria patients, the clinical impact of these effects appears to be mild, 27 although they may contribute to the commonly reported asthenia.

| ECG intervals
Drug-induced QT interval prolongation is the most widely-used surrogate marker of the risk of development of torsades de pointes (TdP), a polymorphic ventricular tachycardia that can degenerate in some cases into ventricular fibrillation and cause sudden cardiac death. 51 Despite their QT-prolonging potential, the front-line quinoline and structurally related antimalarials recommended currently by the WHO all have excellent track records of cardiac safety. They have not been associated with increased risk of sudden cardiac death or cases of TdP in their extensive use at standard doses for the treatment or prevention of malaria over 7 decades. 10,44,52 The QT interval lengthens as heart rate decreases. Correction formulae modelling this inverse and nonlinear relationship are used to attempt to minimize the heart rate dependency of measured QT intervals to allow for comparisons across different heart rates.
However, the commonly used Bazett and Fridericia corrections are both known to retain significant heart rate dependency, 51,53 particularly in malaria where there is further confounding from disease recovery that occurs as antimalarial concentrations peak. 10,37 As before, 10 a study-specific correction provided the best heart rate correction of the QT interval (QTcS) in our analysis, although use of the Fridericia (QTcF) and Bazett (QTcB) corrections, respectively, overestimated and underestimated AQ-related QT prolongation in malaria patients. Once any residual heart rate dependency had been adjusted for the drug-attributable QT prolongation from AQ and desethylamodiaquine was comparable regardless of correction factor used and consistent with QT prolongation similar to that observed with piperaquine 54,55 but less than with chloroquine 56 at standard malaria doses.
In contrast, we found no significant effect of total AQ and desethylamodiaquine concentration on PR or QRS intervals after adjustment for demographic factors (age, sex) and malaria recovery (change in body temperature, change in heart rate). Small increases in unadjusted PR and QRS intervals after AQ for malaria have been previously reported, 57,58 which may predominantly reflect recovery from malaria rather than a direct drug effect. This differs from chloroquine, 56 which prolongs both the PR and QRS intervals with >1/4 of the QT prolongation following chloroquine resulting from QRS widening.

| Conclusion
The