Drug-drug Interactions between and Commonly-used Antiretroviral 2 Treatment: An Individual Participant Data Population Pharmacokinetic Meta-Analysis.

Treating malaria in HIV co-infected individuals should consider potential drug-drug 53 interactions. Artemether-lumefantrine is the most widely recommended treatment for 54 uncomplicated malaria globally. Lumefantrine is metabolized by CYP3A4, an enzyme that 55 commonly-used antiretrovirals often induce or inhibit. A population pharmacokinetic meta- 56 analysis was conducted using individual participant data from ten studies, with 6,100 57 lumefantrine concentrations from 793 non-pregnant adult participants (41% HIV-malaria co- 58 infected, 36% malaria-infected, 20% HIV-infected, and 3% healthy volunteers). 59 Lumefantrine exposure increased 3.4-fold with co-administration of lopinavir/ritonavir-based 60 antiretroviral therapy (ART), while it decreased by 47% with efavirenz-based ART and by 61 59% in the patients with rifampicin-based anti-tuberculosis treatment. Nevirapine- or 62 dolutegravir-based ART and malaria- or HIV-infection were not associated with significant 63 effects. Monte Carlo simulations showed that those on concomitant efavirenz or rifampicin 64 have 49% and 80% probability of day-7 concentrations <200 ng/mL respectively, a threshold 65 associated with an increased risk of treatment failure. The risk of achieving sub-therapeutic 66 concentrations increases with larger body weight. An extended 5-day and 6-day artemether- 67 lumefantrine regimen is predicted to overcome these drug-drug interactions with efavirenz 68 and rifampicin respectively. 69

to the individual patient data (IPD) meta-analysis. Studies addressing the pharmacokinetics of 119 lumefantrine in pregnant women and children were not included as it was beyond the scope 120 of this meta-analysis. WWARN received and curated data from 11 artemether-lumefantrine 121 pharmacokinetic studies, 9 from Africa and 2 from the USA, but one USA study only  Table 1, while an overview of the studies included is provided in Table 2. 8 typical patient weighing 57 kg, the apparent clearance (CL) was 3.28 L/h [3.14 -3.46] ( Table   143 3). 144 Drug-drug interactions 145 Co-administration of lopinavir/ritonavir-based ART increased lumefantrine exposure 146 substantially; the area under the curve (AUC) was nearly 3.4-fold higher, due to 50.1 % 147 slower clearance (ΔOFV = 220, p<0.001) and 67.2% increased bioavailability (ΔOFV 40,148 p<0.001). Lopinavir/ritonavir-based ART was also found to slow down the rate of absorption 149 by 47.6% (ΔOFV =33, p<0.001). Efavirenz-based ART significantly increased the clearance 150 of lumefantrine by 89.9% (ΔOFV = 308, p<0.001), thus resulting in 47% lower AUC. 151 Rifampicin-based anti-tuberculosis treatment increased lumefantrine clearance by 142% 152 (ΔOFV =87, p<0.001) thus reducing AUC by 59%. A small number of patients (n=4) was 153 administered both rifampicin and efavirenz, and there was trend towards an even higher 154 clearance of lumefantrine, but this was not statistically significant and was not retained in the 155 final model. Dolutegravir-based ART didn't alter the lumefantrine exposure. Discordant 156 trends towards slightly higher or lower exposure in the nevirapine-based ART arms were significant consistent difference in pharmacokinetic parameters was found that could be 167 ascribed to malaria-infection. 168 Study and other covariate effects 169 Diurnal variation: After adjusting for the effects described above, significant differences in 170 drug concentrations remained between the studies and, when data were available, between 171 profiles collected after morning or evening doses. These differences were well captured in the 172 model using categorical covariate effects on relative bioavailability (i.e. separate values of 173 bioavailability on specific dosing occasions). The highest bioavailability was observed in the 174 InterACT and the SEACAT studies for the evening doses (with no significant difference 175 between these two studies) and this was chosen as the reference value (fixed to 1) to which 176 the bioavailability of other doses was compared. In the SEACAT studies, the relative 177 bioavailability was 48.6% lower for the first (morning) dose (ΔOFV = 63, p<0.001) and 74% 178 lower for the consecutive morning doses (ΔOFV =280, p<0.001). The value of relative 179 bioavailability in the Ugandan studies was similar and found to be 26.9% lower than the 180 reference (ΔOFV =31, p<0.001), while the value was 58.1% lower than the reference (ΔOFV 181 =18, p<0.001) for the 6 th (morning) AL dose in the Nigeria study 1. Lumefantrine 182 bioavailability in the US healthy volunteer study was not significantly different from the 183 reference group. 184 Matrix effect: Lumefantrine concentrations in both arms of Nigeria study 2, which was the 185 only study measuring concentrations from whole blood samples (as opposed to venous 186 plasma samples), were much higher than in all other studies. A scaling factor of 2.21-fold 187 was included to account for this matrix effect (ΔOFV =122, p<0.001), which is consistent Dosing time: The pre-dose (i.e. morning samples before the 6 th dose) concentrations in 190 Nigeria study 1 and the US healthy volunteer study were higher and inconsistent with the 191 profile collected after the observed 6 th dose. The actual dosing time of the previous (5 th ) dose 192 was not reported, so had been imputed to exactly 12 hours before the morning dose. We  is predicted to achieve satisfactory day-7 concentrations when treated with AL alone or 205 concomitantly with nevirapine or dolutegravir and largely exceed them if on 206 lopinavir/ritonavir. On the other hand, the same patient has 49% and 80% probability of not 207 achieving day-7 concentration above the target when co-treated with efavirenz or rifampicin, 208 respectively. Additionally, participants with larger body weight are predicted to have lower 209 exposures. The effect of body size on target attainment is modest when AL is used alone or 210 with nevirapine or dolutegravir, but it becomes critical for participants of larger weight co-211 treated with efavirenz or rifampicin. Our model predicts that the risk of day-7 concentrations 212 below the target increases to 62% and 87% for an 80-kg patient on efavirenz or rifampicin, respectively. The use of a 4-day regimen of AL is predicted to reduce the risk of sub-214 therapeutic day-7 concentration to 18% and 50% for a typical 57-kg patient on efavirenz or 215 rifampicin, respectively, and these probabilities drop further to 3% and 16% with a 5-day 216 regimen. Simulations showed that a 6-day regimen of AL was necessary to reduce the risk of 217 sub therapeutic day-7 concentration to 2% for a typical 57-kg patient on rifampicin. For an 218 80-kg patient, a longer 5-day and 6-day regimen for efavirenz or rifampicin co-treatment, 219 respectively, reduces the risk of sub-therapeutic concentrations to 5% and 2%.

222
In this IPD meta-analysis of lumefantrine pharmacokinetics, we quantified the effect of 223 commonly prescribed ARTs on lumefantrine exposure. To the best of our knowledge, this is 224 the largest IPD meta-analysis of drug-drug interactions of lumefantrine with antiretrovirals to 225 date, combining 6,100 concentrations from 793 adults in 10 studies, 9 from sub-Saharan 226 Africa and 1 from North America. Although most included studies were from Africa, they 227 were carried out in different regions, and the continent is known for its genetic diversity, (24) 228 so we believe that significant genetic differences are represented in our pooled analysis. The 229 pooling of individual participant data allowed us to re-evaluate and characterize the various 230 drug-drug interactions and other covariate effects more robustly and reliably than in any 231 single study. This was accomplished thanks to the larger sample size, and to the flexibility of 232 population pharmacokinetic modeling, which is able to adjust for study-specific differences 233 and the known effects such as patient body size, and separately investigate the drug-drug exposure and, reassuringly, no effect was found.
The primary aim of this pooled analysis was to characterize the effect of lopinavir/ritonavir-,  Rifampicin-based tuberculosis treatment was found to decrease lumefantrine exposure by 257 59% which was expected, since rifampicin is known to be a potent inducer of CYP3A4 and 258 previous physiologically-based pharmacokinetic modelling had predicted this in silico (28).

259
Clinically, the interaction has been shown in a small study in HIV-infected malaria-260 uninfected adults (29), which in this analysis was pooled with data from patients on rifampicin co-treatment from InterACT, thus confirming and making the result more robust.

262
Of the 13 participants on rifampicin included in our analysis, four were co-treated with 263 efavirenz, while the rest were not on ART. There was a non-statistically significant trend 264 towards an even stronger effect on clearance when both efavirenz and rifampicin were 265 combined, but the limited sample size limited our ability to robustly characterise this   between the studies on lumefantrine exposure was necessary for better characterization of the drug-drug interactions. After adjusting for this biologically plausible effect, no effect of 287 nevirapine-based ART on lumefantrine exposure was found in the IPD meta-analysis. This 288 also points to the importance of standardizing food co-administration when lumefantrine 289 pharmacokinetics is investigated.