Cardiovascular safety and population pharmacokinetic properties of piperaquine in African patients with uncomplicated falciparum malaria – a pooled multicentre analysis

Dihydroartemisinin-piperaquine has shown excellent efficacy and tolerability in malaria treatment. However, concerns have been raised of potentially harmful cardiotoxic effects associated with piperaquine. The population pharmacokinetics and cardiac effects of piperaquine were evaluated in 1,000 patients, mostly children enrolled in a multicentre trial from 10 sites in Africa. A linear relationship described the QTc-prolonging effect of piperaquine, estimating a 5.90ms mean QTc-prolongation per 100ng/mL increase in piperaquine concentration. The effect of piperaquine on absolute QTc-interval estimated a mean maximum QTc-interval of 456ms (EC50=209ng/mL). Simulations from the pharmacokinetic-pharmacodynamic models predicted 1.98-2.46% risk of having QTc-prolongation > 60ms in all treatment settings. Although piperaquine administration resulted in QTc-prolongation, no cardiovascular adverse events were found in these patients. Thus, the use of dihydroartemisinin-piperaquine should not be limited by this concern.


Introduction
patients was similar in all strata (approximately 50%). The baseline QTc SSB -interval at enrolment 162 was significantly different in each strata (p < 0.001), where patients with longer QTc SSB -intervals 163 also had higher baseline values. The median piperaquine C max was significantly different among 164 the groups (p < 0.0001), with a gradual increase with increasing QTc  covariates were significantly different between groups. The results of the statistical analysis of 166 clinical determinants of QTc-prolongation are shown in Table 3. 167 The possible drug-drug interaction of piperaquine and concomitant medications that prolong 168 QTc-interval was also investigated. There were twelve patients that received at least one 169 medication, listed on www. Crediblemeds.org (accessed: 2019-06-19) as drugs that prolong the 170 QTc-interval, during the study period. The concomitant medications were metronidazole, 171 ketoconazole, fluconazole, ciprofloxacin, furosemide, and metoclopramide. Among these twelve 172 patients, five patients had a ∆QTc SSB -interval ≤30ms, six patients had a ∆QTc SSB -interval of 31-173 60ms, and one patient had a ∆QTc SSB -interval >60ms. With respect to absolute QTc-interval, 174 seven patients had a QTc SSB -interval ≤450ms, four patients had a QTc SSB -interval of 451-480ms, 175 and one patient had a QTc SSB -interval of 481-500ms. 176 Relationship between piperaquine concentration and QTc-interval 177 To quantify the magnitude of absolute QTc-interval prolongation resulting from piperaquine 178 administration, a population pharmacokinetic-pharmacodynamic analysis was performed. The 179 absolute QT-intervals were corrected for heart rate using the study specific correction factor 180 (QTc SSB ,) and evaluated with nonlinear mixed-effects modelling. There was a 4.87 ms QTc-181 prolongation per 100 ng/ml increase in piperaquine plasma concentration when described by a  (full 3-day treatment course) and mass drug administration (full 3-day treatment course given 205 once a month for a total of 3 months). Two different dosing recommendations for 206 dihydroartemisinin-piperaquine were evaluated; old recommendation (2 nd edition) and new recommendation (3 rd edition) of the WHO guidelines for the treatment of malaria (Table 5) respectively. The simulated total probability of having a QTc SSB -interval above 500ms was 1.1%

216
(11 in 1,000 patients) and 1.2% (12 in 1,000 patients) in acute treatment of malaria using the old 217 and new piperaquine dosing regimen, respectively. Similarly, the simulated total probability of 218 having a QTc SSB -interval above 500ms was 1.2% (12 in 1,000 patients) and 1.3% (13 in 1,000 219 patients) in mass drug administration settings using the old and new piperaquine dosing regimen, 220 respectively. The probability of having QTc max > 500ms, stratified by body weight, is shown in 221 Figure S5. The simulations showed that predicted ∆QTc max values of more than 60ms were 222 infrequent (1.98-2.25%). Acute treatment resulted in a predicted median ∆QTc max of 16.8ms 223 (95% CI: 2.31-56.9ms) and 18.0ms (95% CI: 2.67-58.6ms) for the old and new piperaquine 224 dosing regimens, respectively. Similarly, the median predicted ∆QTc max was 17.6ms (95% CI: 225 2.58-57.9ms) and 18.5ms (95% CI: 2.90-59.3ms) for the old and new piperaquine dosing 226 regimens given in a mass drug administration scenario. Simulated maximum QTc-intervals 227 stratified by body weight are shown in Figure 4. The distribution of predicted QTc max and 228 ∆QTc max of each clinical scenario is shown in Figure 5. The probability of having QTc max and 229 ∆QTc max at different threshold levels is shown in Table S3.

231
Dihydroartemisinin-piperaquine was well tolerated and highly effective in the treatment of 232 uncomplicated malaria as reported previously in this and earlier studies. (2,17,18). This analysis 233 focused on electrocardiographic QT-prolongation, a risk factor for ventricular tachyarrhythmias 234 (TdP). Piperaquine, like several other aminoquinolines and structurally related antimalarial 235 drugs, prolongs the QT-interval. Halofantrine was an antimalarial drug which caused marked 236 QT-prolongation and was associated with sudden death. It has now been discontinued, but 237 concern over a "class effect" remains. This study sought to characterize the relationship between 238 piperaquine plasma concentrations and QT-prolongation, and thereby gauge the potential risk of  modelling was also attempted using pre-treatment data. However, only one triplicate ECG 290 measurement during the pre-treatment period was available for each patient, which was 291 insufficient to estimate an individual correction factor precisely. Study specific correction factors 292 or individual correction factor should be used when possible, and if data allow these to be 293 estimated reliably. However, the Bazett correction can be used when data or study design do not 294 allow for these study specific or individual correction factors to be estimated.

295
Clinical determinants of QTc-prolongation 296 The statistical analysis performed in the current study evaluated and identified biological factors 297 which might influence the QTc-interval, especially in the subpopulations presenting with 298 extreme QTc-prolongation. The analysis revealed that in patients who had ∆QTc SSB -interval >60ms also had a significantly shorter baseline QTc SSB -interval compared to the other two 300 groups. Moreover, they had the highest median body temperature at enrolment (37.6 °C, IQR: 301 36.8-38.6). Fever has been identified as a factor associated with QTc-prolongation in patients 302 with congenital long QT syndrome (24, 25). However, in the general population, fever has been 303 reported as a factor associated with 26,27). This might partly explain the 304 shorter baseline QTc SSB -interval in this group of patients, and therefore also the apparent large 305 QTc SSB -interval prolongations as patients recover from malaria. The highest QTc SSB -306 prolongation in most of the patients was observed on day 3 after the last dose of piperaquine 307 administration. This occurred approximately at the same time as patients recovered from fever, 308 and might therefore reflected an additional QTc-prolongation on top of the drug effect. Previous 309 studies report a mean QTc-prolongation of 11-18ms (comparing baseline and day 3 values with 310 varying heart rate correction methods) in patients receiving antimalarial treatments unlikely to 311 increase the QTc-interval (i.e. mefloquine and sulfadoxine-pyrimethamine). The QTc-312 prolongation reported in these studies was explained preliminary by the resolution of fever 313 associated with the recovery from the malaria disease (23, 28, 29).

314
The analysis on the absolute QTc-interval found a total of 27 (2.7%) patients with an observed 315 maximum QTc SSB -interval >500ms and 19 patients (70.4%) in this group also had ∆QTc SSB -316 interval >60ms. These patients also had longer median baseline QTc SSB -interval (435ms, IQR: 317 425ms-450ms) compared to other groups. The longer baseline QTc SSB -interval might have 318 resulted in a longer absolute QTc SSB -interval at a given piperaquine concentration, compared to 319 patients with lower baseline values. Furthermore, the median piperaquine C max was significantly 320 higher in patients with a maximum QTc SSB -intervals >500ms compared to those with QTc SSB -321 intervals ≤450ms, but not different from patients with QTc SSB -intervals 451-480ms. Thus, piperaquine concentrations alone did not put patients at high risk of QTc SSB -intervals >500ms. estimate precisely the fluctuation of the QT-interval through the whole 24-hour period.

346
Maximum daily fluctuations of the QT-interval have been reported to be 6.75-7.80 ms in healthy 347 adult subjects (34, 36). However, this value could not be incorporated a priori in the model since 348 the circadian pattern in malaria patients has not been well characterized. A thorough QT study, 349 with measurements over a 24-hour period, in malaria patients receiving antimalarial drugs that  Re-dosing was performed once and rescue treatment was given for unsuccessful re-dosing.  were implemented in PsN.

539
All QT-intervals were corrected for heart rate before further analyses were conducted. Four

845
The graphs shows the probability density distribution of maximum QTc-intervals and maximum 846 ∆QTc-intervals based on a total 480,000 simulated patients after receiving the old (grey solid 847 lines) and new (red solid lines) dosing regimen for (A, B) acute malaria treatment (3-day 848 regimen) and for (C, D) mass drug administration (monthly 3-day regimen).