Physiologically-based pharmacokinetic (PBPK) modeling for prediction of the optimal 2 dose regimens of quinine and phenobarbital co-administration in adult patients with 3 cerebral malaria and seizures

26 Background: Cerebral malaria is a fatal disease. Patients with cerebral malaria are at risk of 27 seizure development, therefore, the co-administration of antimalarial and antiepileptic drugs 28 are needed. Quinine and phenobarbital are standard drugs for the treatment of cerebral 29 malaria with seizures. However, there is no information on the optimal dosage regimens of 30 both drugs when used concomitantly.T he study applied physiologically-based 31 pharmacokinetic (PBPK) modeling for prediction of the optimal dose regimens of quinine 32 and phenobarbital when co-administered in patients with cerebral malaria and concurrent 33 seizures who carry wild type and polymorphic cytochrome P450 (CYP450) 2C9/2C19 . 34 Methods: The whole-body PBPK models for quinine and phenobarbital were constructed 35 based on the previously published information using Simbiology ® . One hundred virtual 36 population were simulated. Four published articles were used for model verification. 37 Sensitivity analysis was carried out to determine the effect of the changes in model 38 parameters on AUC 0–72h . Simulation of optimal dose regimens was based on standard drug- 39 drug interactions (DDIs), and actual clinical use study approaches. 40 Results: Dose adjustment of the standard regimen of phenobarbital is not required when co- 41 administered with quinine . The proposed optimal dose regimen for quinine, when co- 42 administered with phenobarbital for patients with a single or continuous seizure in all 43 malaria-endemic areas regardless of CYP2C9/CYP2C19 genotypes, is a loading dose of 1,500 44 mg IV infusion over 8 hours, followed by 1,200 mg infusion over 8 hours given three times 45 daily, or multiple doses of 1,400 mg IV infusion over 8 hours, given three times daily. In 46 areas with quinine resistance, the dose regimen should be increased as a loading dose of 47 2,000 mg IV infusion over 8 hours, followed by 1,750 mg infusion over 8 hours given three 48 times daily. 49 Conclusion: The developed PBPK models are reliable, and successfully predicted the 50 optimal doses regimens of quinine-phenobarbital co-administration with no requirement of 51 CYP2C9/CYP2C19 genotyping. 52


Quinine and phenobarbital dose regimens used in simulations 162
The standard regimen of quinine for severe malaria is the loading dose of 20 mg kg -1 (1,000 163 mg base total dose for the average 60 kg body weight) IV infusion over 4 hours, followed by 164 the maintenance dose of 10 mg kg -1 (500 mg base total dose for the average 60 kg body 165 weight) IV infusion over 4 hours, given three times daily (every 8 hours) [1] for 72 hours. 166 The simulation time used was based on the average time for patients with cerebral malaria to 167 regain consciousness (72 hours) [28]. 168 169

DDIs model simulation 170
Simulation based on standard DDIs study approach: For standard DDIs study approach, 171 plasma concentration-time profiles of phenobarbital and quinine following the co-172 administration of standard dose regimen of phenobarbital and quinine were simulated. 173 Phenobarbital is given at 2 mg kg -1 day -1 (1-3 mg.kg -1 .day -1 ) or 120 mg total dose for the 174 8 average 60 kg body weight) IV infusion over 30 minutes for 17 consecutive days. The 175 standard dose regimen of quinine for three days (described above) was given on day 14 of 176 phenobarbital administration when steady-state plasma concentration was achieved. 177 Simulation based on actual clinical use approach: For the PBPK simulation based on actual 178 clinical use study approach, two simulated scenarios were applied with the total simulation 179 time of 72 hours. The scenario-I applies for patients who have only a single seizure; 180 phenobarbital (15 mg kg -1 or 900 mg total dose for the average 60 kg body weight, IV 181 infusion over 30 minutes) is given as a single dose 6 hours after the first dose of quinine 182 (average time of occurrence of seizure after admission) [33]. The scenario-II applies for 183 patients who have continuous seizures; phenobarbital at a loading dose of 15 mg kg -1 or 900 184 mg total dose for the average 60 kg body weight IV infusion over 30 minutes, followed by 185 the maintenance dose of 1.5 mg kg -1 day -1 (1-3 mg.kg -1 day -1 ) or 90 mg total dose for average 186 60 kg body weight IV infusion over 30 minutes [33] is given every 24 hours, starting 6 hours 187 after the first dose of quinine until 72 hours. The time of simulation and seizure frequency 188 was based on the clinical report [34]. The predicted optimal dosage regimens were presented 189 as amount of quinine base. 190 191 Criteria for optimal dose regimens 192 The optimal dose regimens of quinine for adult patients with cerebral malaria with seizures 193 were proposed based on the therapeutic range of quinine, i.e., maximum plasma 194 concentration (C max )  20 mg.L -1 , and trough plasma concentration (C trough )  10 mg.L -1 [35]. 195 The optimal dosage of phenobarbital were proposed based on the therapeutic rage of 196

Sensitivity analysis 209
Sensitivity coefficient analysis values for AUC 0-72h of quinine, phenobarbital, and quinine 210 and phenobarbital co-administration ranged from -0.72 to +0.14, -0.02 to +0.10, and -0.48 to 211 +0.12, respectively. All values were less than one, indicating no significant impact of the 212 model parameters on model construction.  Table 4. 234 Simulation based on standard DDIs study approach: Standard dose regimen of quinine 235 provided inappropriate plasma drug concentrations when co-administered with phenobarbital 236 (C trough < 10 mg.L -1 ) ( Table 4). The initially proposed quinine (regimen-1: a loading dose of 237 2,200 mg IV infusion over 4 hours, followed by maintenance doses of 1,200 mg IV infusion 238 over 4 hours given three times daily) in patients with wild type and polymorphic 239 CYP2C9/CYP2C19 provided 2.2-fold lower plasma drug concentrations compared with 240 standard quinine regimen, with AUCR ranging from 0.42 to 0.45. This 2.2-fold increase of 241 quinine standard dose provided the C max exceeding 20 mg.L -1 , but the C trough lower than 10 242 mg.L -1 both in patients with wild type and polymorphic CYP2C9/CYP2C19 (Fig. 1, and Fig.  243 2 for wild type and polymorphic CYP2C9/CYP2C19, respectively) ( Table 5). The time to 244 reach therapeutic concentration ranged from 2 to 3 hours. The three subsequent dose 245 regimens were therefore simulated, e.g., a loading dose of 2,200 mg IV infusion, followed by 246 2,000 mg IV infusion (regimen-2), a loading dose of 2,000 mg IV infusion, followed by 247 1,500 mg IV infusion (regimen-3), and a loading dose of 1,750 mg IV infusion, followed by 248 1,500 mg IV infusion (regimen-4). All regimens were given as IV infusion over 8 hours, and 249 11 the maintenance doses were given three times daily. The average C max and C trough (95%CI) for 250 each regimen (-2, -3, and -4) are shown in Fig. 1  Simulation based on actual clinical use study approach: Standard dose regimen of quinine 254 provided inappropriate plasma drug concentrations when co-administered with phenobarbital 255 (C trough < 10 mg.L -1 ) ( Table 4). Dose adjustment in both clinical scenarios based on standard 256 dosage regimen of quinine provided the AUCR of quinine when co-administered with 257 phenobarbital in patients with wild type genotype, and polymorphic CYP2C9/CYP2C19 258 ranging from 0.47 to 0.53 (Table 4). The 2-fold increase of quinine standard dose (regimen-259 5: a loading dose of 2,000 mg IV infusion over 4 hours, followed by maintenance doses of 260 1,000 mg IV infusion over 4 hours given three times daily) provided the concentrations out 261 of therapeutic range (C max > 20 mg.L -1 , and C trough < 10 mg.L -1 ) both in patients with wild 262 type and polymorphic CYP2C9/CYP2C19 ( Fig. 3 (scenario I and II), Fig. 4 (scenario I), and 5 263 (scenario II) for wild type and polymorphic CYP2C9/CYP2C19, respectively) ( Table 6 and 7  264 for scenario I and scenario II, respectively). The three subsequent dose regimens for the 265 scenario I (single seizure) and II (multiple seizures) were therefore simulated, i.e.., a loading 266 dose of 2,000 mg IV infusion, followed by 1,750 mg IV infusion three times daily (regimen-267 6), a loading dose of 1,500 mg IV infusion, followed by 1,200 mg IV infusion three times 268 daily (regimen-7), and multiple doses of 1,400 mg IV infusion three times daily (regimen-8). 269 The IV infusion duration in all regimens was 8 hours, and the maintenance doses were given 270 three times daily. The average C max , and C trough (95%CI) of quinine for wild type genotype in 271 each simulation are shown in Fig. 3, and those for polymorphic CYP2C9/CYP2C19 for the 272 scenario I and scenario II are presented in Fig. 4 and Fig. 5, respectively. Table 5, 6, and 7 273 summarize the parameters predicted based on standard DDI and actual clinical use (single 274 12 and continuous seizures) study approach, respectively. Time to reach therapeutic quinine 275 levels ranged from 4 to 6 hours. Plasma quinine concentrations following regimen-7 and -8 276 except regimen-5 and -6 were within the therapeutic range for both scenarios I and scenario 277 II. The C max of quinine following regimen-6 in both scenarios ranged from 21-23 mg.L -1 . 278 279

Discussion 280
Results of the current study based on PBPK modeling and simulation raise a concern about 281 the potential DDIs between quinine and phenobarbital in patients with cerebral malaria who Simulation of the optimal phenobarbital dose regimens in patients with seizures who carry 292 polymorphic CYP2C9/CYP2C19 was investigated using dose regimens based on the two 293 seizures (a loading dose of 15 mg. kg -1 .day -1 , followed by 1.5 mg.kg -1 .day -1 once-daily) 307 regardless of patients' CYP2C9/CYP219 genotypes. Genotyping is therefore not necessary, 308 which is practical both in developed and developing countries. Besides, the advantage of 309 using phenobarbital over other anticonvulsants is its relatively low cost [2].

Simulations based on standard DDIs approach:
The three proposed quinine dosage 319 regimens, i.e., regimen-2, -3 and -4) provided adequate C trough above 10 mg.L -1 in both 320 wildtype genotype (Fig 1B, C, D for regimen-2, 3, and -4, respectively) and polymorphic 321 CYP2C9/CYP2C19 (Fig 2B, C, D for regimen-2, -3, and -4, respectively). Nevertheless, 322 regimen-2 (a loading dose of 2,200 mg IV infusion over 8 hours, followed by 2,000 mg IV 323 infusion over 8 hours given three times daily) is not appropriate for clinical use due to high 324 14 drug concentrations above 20 mg.L 1 in individuals with both wild type and polymorphic 325 CYP2C9/CYP2C19 and thus, the possibility of toxicity. Since the reported minimum 326 inhibitory concentration (MIC) of quinine in quinine-resistant areas has been rising to over 10 327 mg.L -1 [39], the most optimal dose regimen would be regimen-3 (a loading dose of 2,000 mg 328 IV infusion over 8 hours, followed by 1,500 mg IV infusion over 8 hours given three times 329 daily). In low quinine resistant malaria-endemic areas, regimen-4 (a loading dose of 1,750 330 mg IV infusion over 8 hours, followed by 1,500 mg IV infusion three times daily) could be 331 an alternative regimen. It is noted, however, that the infusion duration of 8 hours might result 332 in the delay of time to reach therapeutic level compared with the recommended standard 333 regimens (4-6 and 2-3 hours for the recommended and standard regimens, respectively). 334 Since the critical clinical period for treatment of patients with cerebral malaria is during the 335 first 24 hours [27], such delay is unlikely to pose the patients at risk of complicated 336 manifestation or death. These quinine regimens can be co-administered with phenobarbital 337 without consideration of CYP2C9/CYP2C19 genotypes. 338 Simulations based on actual clinical use study approach: Optimal C max and C trough of 339 quinine were achieved with the two proposed quinine dose regimens (regimen-7 and -8) 340 when co-administered with phenobarbital in both clinical scenarios (scenario I for a single 341 seizure and scenario II for continuous seizures) using PBPK modeling and simulation, but not 342 the AUCR, resulted in adequate plasma concentrations (Fig. 3, Fig. 4, and Fig. 5 for wild 343 type, the scenario I in polymorphic CYP2C9/CYP2C19, and scenario II in polymorphic 344 CYP2C9/CYP2C19, respectively). With the administration time of quinine 6 hours prior to 345 phenobarbital in actual clinical study approach, the optimal quinine dosage regimens can be 346 reduced as a loading dose of 2,000 mg IV infusion over 8 hours, followed by 1,750 mg IV 347 infusion three times daily (regimen-6), or alternatively, 1,500 mg IV infusion, followed by 348 1,200 mg IV infusion (regimen-7), or 1,400 mg IV infusion over 8 hours given three times 349 15 daily (regimen-8). These quinine regimens can be co-administered with phenobarbital 350 without consideration of CYP2C9/CYP2C19 genotypes since plasma quinine concentrations 351 in patients with wild type, and polymorphic CYP2C9/CYP2C19 were comparable. There is 352 no influence of CYP2C9/CYP2C19 genotypes on the inducing effect of quinine metabolism 353 since the steady-state drug concentrations are not achieved with a short duration of 354 phenobarbital dosing. The possibility of dose reduction could be due to the lack of CYP450 355 enzyme-inducing effect of phenobarbital during this period (6 hours). In the case of quinine 356 resistant malaria, particularly with the contribution of large interindividual variability in 357 quinine clearance, quinine regimen-6 would be the best option. The higher C max of 1-3 mg.L -1 358 above the therapeutic range is unlikely to cause toxicity due to high plasma protein binding 359 during the acute phase of malaria infection. It is noted that the recommended optimal dose 360 regimens of quinine and phenobarbital co-administration apply for cerebral malaria patients 361 with seizures who have normal hepatic function but not in those with impaired function. 362 Therapeutic drug monitoring for quinine in those patients is recommended. 363 The limitations of the study include the exclusion of the contribution of P-364 glycoprotein transporter on quinine disposition (due to lack of information on in vitro studies), 365 as well as the inhibitory effect of 3-hydroxyquinine metabolite on CYP3A4 activity. 366 Nevertheless, the significant impacts of these two factors on quinine disposition are unlikely 367 In summary, PBPK models are a potential tool for dose optimization of quinine in 369 patients with cerebral malaria in resource-limited countries. The developed PBPK models for 370 phenobarbital and quinine-phenobarbital co-administration are reliable, and successfully 371 predicted the optimal doses regimens of phenobarbital in cerebral malaria patients with single 372 or continuous seizures with no requirement of CYP2C9/CYP2C19 genotyping. Dose 373 adjustment based on PBPK modeling but not AUCR provided desirable plasma quinine 374 16 concentrations. Dose adjustment of the standard regimen of phenobarbital is not required 375 when co-administered with quinine. The proposed optimal dose regimen for quinine when 376 co-administered with phenobarbital for patients with a single seizure (scenario I), and 377 continuous seizures (scenario II) in all malaria-endemic areas regardless of 378 CYP2C9/CYP2C19 genotypes is a loading dose of 1,500 mg IV infusion over 8 hours, 379 followed by 1,200 mg IV infusion over 8 hours given three times daily (regimen-7), or 380 multiple doses of 1,400 mg IV infusion over 8 hours, given three times daily (regimen-8). In 381 areas with quinine resistance, the dose regimen should be increased as a loading dose of 382 2,000 mg IV infusion over 8 hours, followed by 1,750 mg IV infusion over 8 hours given 383 three times daily (regimen-6). are C max  20 mg.L -1 , and C trough  10 mg.L -1 (therapeutic range of quinine). 589 590 Tables 591 Table 1 Model verification of quinine and phenobarbital in cerebral malaria, severe malaria, 592 and healthy adults. Comparisons of quinine and phenobarbital between predicted results and 593 published data in patients with cerebral malaria, severe malaria, and healthy adults. 594 Table 2 Simulated standard dose regimens of phenobarbital based on standard DDI study 595 approach. Simulation of standard dose regimens of phenobarbital in patients with seizures 596 with wild tyle and polymorphic CYP2C9/CYP2C19 based on standard DDI study approach. 597 Table 3 Simulated standard dose regimens of phenobarbital based on actual clinical use 598 DDI study approach. Simulation of standard dose regimen of phenobarbital in patients with 599 seizures with wild type and polymorphic CYP2C9/CYP2C19 based on actual clinical use 600 study approach 601 Table 4 Simulated standard dose regimens of quinine when co-administered with 602 phenobarbital based on actual clinical use approach. Simulation of the standard dose of 603 quinine when co-administered with phenobarbital in cerebral malaria patients with 604 concomitant seizures with wild type and polymorphic CYP2C9/CYP2C19. 605 Table 5 Prediction of quinine dose regimens when co-administered with phenobarbital 606 based on standard DDI study approach. Prediction of quinine dosage regimens when co-607 administered with phenobarbital in cerebral malaria patients with concomitant seizures and 608 polymorphic CYP2C9/CYP2C19 based on standard DDI study approach. 609 Table 6 Prediction of quinine dose regimens when co-administered with phenobarbital based 610 on scenario-I. Prediction of quinine dosage regimen when co-administered with 611 phenobarbital in cerebral malaria patients with concomitant seizures with wild type and 612 polymorphic CYP2C9/CYP2C19 based on actual clinical study approach (scenario I: single 613 seizure). 614 Table 7 Prediction of quinine dose regimens when co-administered with phenobarbital based 615 on scenario-II. Prediction of quinine dosage regimen when co-administered with 616