A pooled analysis of the duration of chemoprophylaxis against malaria after treatment with 1 artesunate-amodiaquine and artemether-lumefantrine

50 Artemether-lumefantrine (AL) and artesunate-amodiaquine (AS-AQ) are the most commonly-used 51 treatments against Plasmodium falciparum malaria in Africa. The lumefantrine and amodiaquine 52 partner drugs may provide differing durations of post-treatment prophylaxis, an important additional 53 benefit to patients. Analyzing 4214 individuals from clinical trials in 12 sites, we estimated a mean 54 duration of post-treatment protection of 13.0 days (95% CI 10.7-15.7) for AL and 15.2 days (95% CI 55 12.8-18.4) for AS-AQ after allowing for transmission intensity. However, the duration varied 56 substantially between sites: where wild type pfmdr1 86 and pfcrt 76 parasite genotypes predominated, 57 AS-AQ provided ~2-fold longer protection than AL. Conversely, AL provided up to 1.5-fold longer 58 protection than AS-AQ where mutants were common. We estimate that choosing AL or AS-AQ as 59 first-line treatment according to local drug sensitivity could alter population-level clinical incidence of 60 malaria by up to 14% in under-five year olds where malaria transmission is high. 61


Introduction 64
There is evidence that the duration of protection after AS-AQ and AL treatment is affected by parasite 95 mutations associated with reduced drug sensitivity (9,11). These two drugs show collateral sensitivity, 96 such that the mutations 86Y and 1246Y in the pfmdr1 gene and 76T in the pfcrt gene are linked to 97 reduced sensitivity to AS-AQ but increased sensitivity to AL, which is thought to be due to differential 98 sensitivity to the amodiaquine and lumefantrine partner drugs rather than the artemisinin. Although the 99 overall efficacy of each drug remains high in Africa, a meta-analysis found that the N86 wild type 100 parasite was associated with a 4-fold increased risk of recrudescence after AL treatment (9,11). All 101 these mutations were also associated with a reduced time to reinfection after AS-AQ treatment, and an 102 increased time to reinfection after AL treatment, although the exact duration of protection was not 103 estimated since this also depends on the local rate of transmission and thus reinfection. 104

105
The duration of protection can be estimated from clinical trials where reinfection rates are monitored. 106 We previously estimated the mean protection provided by AL at 13.8 days, and DHA-piperaquine at 107 29.4 days (4). The duration of protection provided by amodiaquine is not well known, although there 108 are indications that it might confer longer protection than lumefantrine (23, 24). Here, we use a 109 statistical analysis of pooled clinical trial data from multiple sites in Africa, explicitly incorporating 110 local transmission intensity as well as drug effects into analyzing the time to reinfection, to estimate the 111 duration of post-treatment prophylaxis after AS-AQ and AL. We use these results in an epidemiological 112 transmission model to establish the differences in public health impact when AS-AQ versus AL is used 113 as first-line drug for P. falciparum case management. 114 115 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org /10.1101/19002741 doi: medRxiv preprint Results 116

Duration of protection after AL and AS-AQ treatment in different trial sites 117
We analyzed 4214 individual participant data from randomized clinical trials in 12 sites obtained from 118 the WorldWide Antimalarial Resistance Network (WWARN) data platform (25) with the consent of 119 investigators or sponsors. The median age in the study population was 2.8 years (IQR 1.5-4.2). Patients 120 were followed up until at least day 28 and assessed for the presence of reinfection, using PCR to 121 distinguish reinfecting parasites from recrudescence of the original infection. The time to reinfection in 122 these trials is only in part determined by the duration of protection conferred by the drug. This is 123 because individuals do not immediately become reinfected after the protection ends, but rather enter an 124 "at-risk" state. Thereafter they are reinfected at a rate dependent on the incidence of blood-stage 125 infections in the population (which in turn depends on the entomological inoculation rate (EIR), the 126 number of infectious bites per person per year). We accounted for the differing incidence of infection in 127 the different trial sites using prior information on malaria transmission intensity from the Malaria Atlas 128 Project (26,27), estimated at the location and year in which each trial was carried out. We then 129 employed two statistical approaches: (1) Table 1). The proportion of patients reinfected in the AS-AQ trial arm was lower than the AL arm in 7 141 sites, while it was higher in the 5 other sites by the end of follow up. (Figure 2). This heterogeneity was 142 confirmed by the posterior estimates of the duration hyper-parameters, which suggested non-zero 143 variance of the random site effects. The heterogeneity existed despite the analysis taking into account 144 variation in EIR, which ranged from an estimated 2 to 117 infectious bites per person per year. While 145 there was, as expected, a reduced total time to reinfection with higher EIR, after accounting for EIR we 146 7 found no trend for duration of drug protection by EIR ( Figure S1). Overall the model was able to fit the 147 data well, with the model predicted values being within the 95% confidence intervals of the proportion 148 of individuals reinfected at each follow up time in almost all sites ( Figure 2). Posterior EIR values were 149 mostly in line with the MAP-based prior values but differed considerably for a small number of 150 locations ( Figure 3, Table 1). For sensitivity analysis, we tried including additional age-independent 151 variation in exposure to mosquito bites as in a previous analysis (see Methods), since this influences 152 the distribution of reinfection times within a cohort. Such additional variation represents factors such as 153 living close to a breeding site, housing quality, etc. This analysis found similar estimates of the duration 154 of protection after AS-AQ and AL as did the model without additional variation in exposure, at 16.5 155 days (95% CI 14.2-19.3) and 14.1 days (95% CI 11.7-16.9), respectively. Therefore, for parsimony we 156 did not include this factor in the final result. implemented for example in R as 1-pgamma(t, shape= r , scale=  ), where t is time in days, and r and  are the shape and scale parameters 163 of the gamma distribution, respectively. For AL, r = 93.5 and mean  =0.139. For AS-AQ, r =16.8 and mean  =0.906. The mean of each 164 gamma distribution r gives the duration of protection from each drug. The site-specific lines can be calculated using the median durations 165 of prophylaxis in Table 1

Factors affecting the duration of prophylaxis 183
To investigate which factors affect the duration of prophylaxis after AS-AQ and AL treatment and 184 might explain the heterogeneity between trial sites, the data were further analyzed by accelerated 185 failure time regression models. As expected, EIR (as estimated for each site by the HSMM analysis) 186 was strongly associated with time to reinfection (Table 2). We therefore adjusted for EIR before testing 187 the effect of any additional variables. Treatment arm had a small and significant effect on time to 188 reinfection overall, with AS-AQ being associated with a 1.09-fold increase in time to reinfection (95% 189 CI 1.05-1.13) compared to AL, after adjusting for log EIR. We explored the effect of molecular markers 190 associated with parasite sensitivity to AL and AS-AQ. These markers were not directly measured 191 during these trials. Instead, for each trial we sought studies close in space and time which measured the 192 prevalence of pfmdr1 86Y, pfmdr1 1246Y and pfcrt 76T mutations among infected individuals, using 193 recently completed systematic reviews. (63,64) We included matches when the study was conducted in 194 the same country, within 300km of the trial site and within 1 year of the trial start or end year. We 195 identified pfmdr1 matches to 11 trial sites, and pfcrt matches to 10 sites; however there were too few 196 matched surveys of pfmdr1 1246Y to analyze this third mutation further. Local prevalence of the 197 mutations pfmdr1 86Y and pfcrt 76T significantly altered the association between drug and time to 198 reinfection. AS-AQ was associated with a significant 1.37 (95% CI 1.28-1.47)-fold increase in time to 199 reinfection compared to AL when pfmdr1 86Y prevalence was 20% (the lowest level observed in the 200 trial sites), but a significantly shorter time to reinfection than AL when pfmdr1 86Y was 80% (ratio of 201 reinfection times AS-AQ vs AL= 0.89 95% CI 0.84-0.94). Similarly, AS-AQ was associated with a 1.54 202 (95% CI 1.38-1.71)-fold increase in time to reinfection compared to AL when pfcrt 76T prevalence was 203 20%, but a 1.06 (95% CI 1.03-1.10)-fold change when pfcrt 76T prevalence was 80%. Other factors 204 that were significantly associated with longer time to reinfection when adjusting each factor only for 205 log EIR were younger age and higher dose of lumefantrine (mg per kg) ( Table 2). Increasing age 206 amongst children was associated with a shorter time to reinfection in a non-linear manner, such that the 207 change in reinfection time with age was most rapid at younger ages, consistent with observed biting 208 patterns by age (65). There was a trend for shorter time to reinfection in underweight individuals and 209 when the loose non-fixed-dose combination (NFDC) formulation of AS-AQ was used, though the 210 association was not statistically significant after adjusting for log EIR. 211

212
We constructed multivariable models for each treatment arm separately. In the AL arm, EIR, age, 213 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint lumefantrine dose (mg per kg), local pfmdr1 86Y prevalence and pfcrt 76T prevalence remained at least 214 borderline significant predictors of time to reinfection (Tables 3 & S1). However, pfmdr1 86Y 215 prevalence and pfcrt 76T prevalence were so closely correlated ( Figure S2) that their effects could not 216 be distinguished from each other in the absence of haplotype data, and we built separate multivariable 217 models to look at each mutation. In the AL arm, both the pfmdr1 86Y and the pfcrt 76T mutations were 218 associated with a 1.04-fold increase in time to reinfection per 10% increase in their prevalence 219 (p=0.052 and p=0.005, respectively) after adjusting for EIR, age and lumefantrine dose. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint   was most pronounced in simulations with higher, very seasonal transmission. When slide-prevalence was 293 50% and transmission was seasonal, using AS-AQ rather than AL prevented 1.6 clinical episodes per child 294 over the 5 years ( Figure 5B) (14% of all clinical episodes; Figure 5C). When considering all age groups, an 295 estimated 10% of clinical episodes were prevented ( Figure S4). 296

297
In Nimba in Liberia (trial conducted 2008-2009) the local prevalence of pfmdr1 86Y and pfcrt 76T were 298 much higher at 69% and 95%, and the duration of prophylaxis provided by AS-AQ was estimated at only 299 11.6 days, while the AL prophylactic time was 17.9 days ( Figure 5D). Here, using AS-AQ rather than AL 300 increased the cumulative number of clinical episodes per 0-5 year old child by up to 1.1 over the 5 year 301 simulated period (an increase of 11%), with the largest difference between drugs again observed in the very 302 seasonal, high transmission scenario ( Figure 5E & 5F). When considering all age groups, clinical episodes 303 increased by up to 8% ( Figure S4). 304 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint It is uncertain whether there is any difference in human infectiousness after treatment with AL versus AS-306 AQ. We therefore ran the simulations twice, assuming firstly that patients are equally infectious after 307 treatment with either ACT, and secondly assuming that patients treated with AS-AQ are twice as infectious, 308 in accordance with previous studies (47,67). In both settings, there was minimal difference in impact on 309 clinical episodes (<1%) if we assumed that patients treated with AL were half as infectious as those treated 310 with AS-AQ, compared with the scenarios where infectiousness was assumed to be equal after each 311 treatment (results not shown). This is because even if there is some difference between treatments, both are 312 estimated to have a high impact on gametocytes. Therefore, at a population level, transmission to 313 mosquitoes is dominated by untreated infections which are thought to last on average about 6 months, 314 according to our model assumptions and parameters. (66,68,69) 315 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint is the

(which was not peer-reviewed)
The copyright holder for this preprint In this analysis of clinical trials from 12 sites in Africa, we initially estimated that AS-AQ provided a 330 slightly longer duration of post-treatment prophylaxis than AL (15.2 versus 13.0 days) when all data were 331 pooled together. However, the duration of protection varied considerably between trial sites. In some 332 locations AS-AQ provided up to an estimated 19 days of protection, ~2-fold longer than AL, while in other 333 trial sites the reverse was true, with AL providing up to 19 days of protection, which was up to 1.5-fold 334 longer than AS-AQ. This difference between sites appeared to be in part explained by the local prevalence 335 of pfmdr1 86Y and pfcrt 76T at the time of the trial, with AS-AQ providing better protection where wild 336 type parasites with N86 and K76 genotypes were predominant, and AL performing better where 86Y and 337 76T mutants were common. This is consistent with previous studies demonstrating the collateral sensitivity 338 of parasites with these different pfmdr1 and pfcrt genotypes to AL and AS-AQ. Our analysis extends 339 previous work (9,11,70) by explicitly estimating the duration of protection provided by each drug in sites 340 with different prevalence of 86Y and 76T mutants, also taking into account the different EIRs across the 341 trial sites so as to distinguish the effect of the drugs from that of the local transmission intensity on the time 342 to reinfection. 343 344 Our transmission modelling suggests that the difference in duration of protection between the two drugs in 345 areas with very low or very high 86Y and 76T prevalence can have a public health impact, especially where 346 malaria transmission is high and seasonal. We estimate that up to 14% of clinical episodes could be 347 prevented in 0-5 year old children by using the drug providing optimal protection in a given setting, due to 348 both individual protection from reinfection and population level reduction in transmission (when 80% of 349 clinical episodes receive treatment). Countries with low (<20%) or high (>80%) prevalence of 86Y and 76T 350 and intense transmission could consider the benefit of longer duration of protection if choosing between AL 351 and AS-AQ policies. Using a first line treatment with longer duration of protection is potentially a cost-352 effective way of reducing clinical cases and infections,(4) given the comparable price of AL and AS-AQ 353 (71). Compared to published estimates, both AL and AS-AQ provided a shorter duration of protection than 354 dihydroartemisinin-piperaquine (estimated at 29.4 days of >50% protection (4)), which is predicted to 355 prevent up to 15% more cases than AL (4,72). 356

357
The pfmdr1 86Y and pfcrt 76T mutations, initially driven through the parasite population by the previous 358 widespread use of chloroquine, have been in decline in many parts of Africa. The decline has occurred 359 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint fastest in countries using AL, consistent with the expected direction of selection (64). The efficacy of AS-360 AQ appears to have improved in some countries and there is interest in increasing the use of the drug 361 regimen where 86Y and 76T prevalence have declined (55) individual patient data on genotypes post-treatment were available.(9, 11) We did not include such a wide 379 range of studies as the previous meta-analysis because our methods required that we estimate the EIR for 380 each included trial site, which is only possible when sufficient numbers of reinfections are observed per site 381 and we included only randomized trials. The advantage of our approach, however, is that we can obtain 382 estimates of prophylactic times after adjusting for the local transmission intensity. One limitation of our 383 study was that we did not have individual level data on genotypes pre and post-treatment, which were not 384 measured in the trials we included here. This might have allowed a more precise estimate of the effect of 385 mutations on prophylactic time, and ideally comparison of pfcrt and pfmdr1 haplotypes. Also, while we 386 matched trials to the closest possible measures of mutation prevalence, these may not reflect the prevalence 387 in the trial sites which can vary over space and time. We could not distinguish separate effects of 86Y and 388 76T in this analysis due to the close correlation of their prevalence. Other previous meta-analyses have 389 examined the effect of dosing and other covariates on the probability of recrudescence after AL (10) and 390 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint AS-AQ (12). The trends in our analysis looking at reinfection as the outcome rather than recrudescence 391 agree well with these previous studies; in particular the use of loose NFDC formulation of AS-AQ was 392 associated with reduced time to reinfection although it was not statistically significant after adjusting for 393 EIR. Of the three studies using loose NFDC, two of these showed a longer prophylactic time by AL, 394 compared to two out of the remaining 9 studies which used FDC. 395 396 Our estimate of the mean duration of prophylaxis after AL at 13.0 days is in good agreement with our 397 previous estimate of 13.8 days which was obtained from analysis of a completely different dataset of 398 clinical trials in six sites in Africa (4) (although the impact of 86Y and 76T was not previously 399 investigated). In the current analysis we found a more rapid decline of protection over time after AL 400 treatment than AS-AQ (Figure 1), and a similar rapid decline after AL was seen in our previous analysis. 401 The resolution of data informing this profile of post-treatment prophylaxis is not perfect, with most patients 402 observed only weekly after day 7. In 4 of the trial sites in the current analysis, no tests for reinfection were 403 done until day 14 (23). Nevertheless given the very low proportion of individuals reinfected at earlier times 404 in the other sites it is unlikely that many reinfections were missed. In most trials patients were followed up 405 until day 28, and differential reinfection rates may have been missed after this time. We lacked data from a 406 control arm to parameterize the proportion of individuals reinfected over time in the absence of treatment. If 407 our model underestimates the rate of increase in the proportion of individuals reinfected in the absence of 408 treatment, it could overestimate the rapid drop off in protection in the AL trial arms to compensate. There is 409 therefore some uncertainty in the shape of the prophylactic profile but if the rapid drop in protection is a 410 real finding, it has implications for the selection of partially resistant parasites to these partner drugs, with 411 lumefantrine potentially having a relatively short window of selection compared to amodiaquine. (76)  412   413 We also did not consider temporal changes in the EIR during the trial. However, these would affect both 414 trial arms equally and could therefore not reverse the relative order of duration of protection between the 415 drugs in one site. Variation between studies may occur due to other factors such as nutritional status, 416 dosage, the genetics of patients or variations in the accuracy of PCR in distinguishing reinfections from 417 recrudescence. For example, PCR accuracy depends on the number of molecular markers used, and in high 418 transmission areas multiple-clone infections can reduce accuracy (77,78). While none of the trials 419 distributed insecticide-treated nets as part of the study, trial areas probably varied in levels of vector control, 420 which is indirectly taken into account in our analysis since we use estimates of transmission intensity based 421 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint on the Malaria Atlas Project, who use data on prevalence trends and include vector control in their model. 422 In the randomized design of the studies, it is unlikely that one treatment arm would have had better vector 423 control than another, and therefore the comparison of the drugs would not be affected, although the overall 424 reinfection rate for the area could have been different from expected if the effect of vector control was not 425 well reflected in the MAP data and model. 426

427
In summary, both AL and AS-AQ provide post-treatment prophylaxis which is important for reducing 428 reinfection rates in individuals in higher transmission settings, and may impact on the incidence of malaria 429 in the whole population when these regimens are used widely as first-line treatment. AS-AQ provides 430 longer protection than AL when most infections are by wild-type parasites, while AL provides longer 431 protection than AS-AQ in areas with higher prevalence of the pfmdr1 86Y and pfcrt 76T mutations. 432 Countries may wish to consider the prevalence of these mutations when deciding the first line treatment. In 433 future, it will be important to determine the role of other molecular markers in altering the post-treatment 434 protection provided by ACT partner drugs, such as increased copy number of pfmdr1, which is increasing in 435 prevalence in some parts of Africa. (63)  436 437 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint

Materials and Methods 438
Data 439 WWARN invited investigators to contribute individual-level patient data for this meta-analysis (79) if their 440 studies fulfilled the following criteria: randomized controlled trials of uncomplicated P. falciparum malaria; 441 AS-AQ and AL being compared; follow-up to at least day 28, with at least one follow-up visit at day 14 and 442 another before day 28; 100 or more participants per study site or more than 28 days follow-up; polymerase 443 chain reaction (PCR)-adjusted efficacy available; at least 95% PCR-adjusted treatment efficacy in both 444 study arms; PCR-unadjusted cure rates of <95% in at least one trial arm by day 28 (to indicate sufficient 445 number of reinfections to inform analysis on post-treatment prophylaxis); standard dose regimens of AL 446 and AS-AQ (we included studies regardless whether AS-AQ was given as a fixed-dose combination or not); 447 and known dosage taken for each patient. Individual patient data from eligible studies were shared, collated 448 and standardized using previously described methodology (80). 449

450
For the present analyses, we used data on PCR-confirmed reinfections as well as the proportion of patients 451 who were not reinfected during follow up, to estimate the duration of chemoprophylaxis. Time of 452 reinfection is included in the analysis so that different follow up times between studies are accounted for 453 (see also below). Patients who experienced PCR-confirmed recrudescence were excluded. In two studies (in 454 Tororo, Uganda & Sikasso, Mali, see Table 1) the patients were followed up longitudinally across several 455 episodes, and consequently treated multiple times within short intervals. We only used the first treatment 456 episode and follow-up data collected before the next episode from these studies in order to avoid 457 confounding of our results by residual drug levels from a previous treatment. 458 459 For each trial, we sought surveys of the prevalence of pfmdr1 86Y, pfmdr1 1246Y and pfcrt 76T mutations 460 in the same country, within 300km of the trial site and within 1 year of the trial start or end year (see also 461 Results). For sites with many matching molecular marker surveys, we applied a stricter distance criterion of 462 100km of the trial site. When more than one matching survey was found, we took a weighted average of the 463 mutant prevalence. We did not include molecular marker studies on post-treatment samples. 464 465 One included study did not have available data on the individual ages of participants, but provided body 466 weight (55), and another study recorded age but not body weight (49). We imputed the missing values in 467 order to be able to include these studies. To impute missing age, we randomly sampled ages of participants 468 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint of the same gender from all other studies who had bodyweights within 0.5kg of the observed participants' 469 weights; to impute missing body weight we sampled weights of individuals of the same gender within 0. host i once at risk, and δ represents the time required for a blood-stage infection to become patent after 494 hepatocyte rupture (assumed 3.5 days (81)). P and I were parameterized as random variables as follows: 495 Where the drug-specific scale parameter λ and shape parameter r are to be estimated, and 499 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Ii ~ exp(1/φi) 501 502 with φi being the force of infection to which individual i was exposed during the trial follow-up. Individual-503 specific EIR values εi were determined, taking into account that young children are bitten less often due to 504 their smaller body size, according to the formula 505 506 εi = εadult (1-ρ exp(-ai/a0)) 507 508 where εadult is the estimated site-specific EIR experienced by fully grown individuals, a is age and 509 parameters a0=2920 days and ρ=0.85 control the shape of the relationship (82). Pre-erythrocytic immunity , 510 i.e. an immune response that reduces the proportion of infectious bites resulting in successful blood stage 511 infections was computed for each individual according to their age, prior exposure and local EIR , using the 512 same mathematical model referenced above (66). For sensitivity analysis, we also tried assuming additional 513 age-independent variation in exposure to mosquito bites, with the distribution of relative biting rates across 514 people following a log-normal distribution. We used informative priors on the lognormal distribution of 515 bites of mean=1 and variance=1.76 as previously estimated (66). 516 517 A number of HSMM variants were fitted via MCMC (Markov-Chain Monte Carlo), using the JAGS ("Just 518 Another Gibbs Sampler") software for Bayesian inference in conjunction with the "rjags" package using R 519 statistical software (83). The likelihood calculation took into account the interval-and right-censoring of 520 observations in the data. EIR values εadult for each site were estimated simultaneously with the other 521 parameters, with moderately informative gamma priors with median as predicted by MAP (27) (Table 1) 522 and a shape parameter of 1.56. Using this prior information on EIR was essential, otherwise a slow 523 reinfection rate could be explained equally well by either a low EIR or a long drug prophylactic time. After 524 examining the posterior distributions of several candidate models, we included heterogeneity among trial 525 sites in the mean duration of chemoprophylaxis, which was modeled as a gamma-distributed random effect. 526 A weakly informative, empirical-Bayes gamma prior was used for the shape parameter r, with 527 hyperparameters (parameters of the prior distribution) determined using a fit of the HSMM with non-528 informative priors. This improved MCMC convergence. Non-informative gamma priors were chosen for all 529 remaining estimated parameters. We ran the MCMC procedure for 1.25 million iterations, retaining 100,000 530 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint samples of the posterior after discarding 4000 adaptation steps, 4000 burn-in steps and thinning. 531 532 Accelerated failure time models: In order to identify which factors influence the time until a reinfection is 533 detected, we used accelerated failure time models, as implemented in the "survival" package in R (84). We 534 explored lognormal and log-logistic distributions of time to reinfection, which allow the hazard of 535 reinfection to vary over time, and selected lognormal which produced lowest Akaike Information Criterion 536 (AIC). Several covariates were compared with respect to their ability to predict time to reinfection. Since 537 EIR is such a critical predictor of the time to reinfection, we adjusted for this variable in all models, initially 538 in bivariate models with each other covariate, using the log posterior mean EIR estimates from the HSMM 539 analysis for each site. When analyzing age as a covariate, we explored polynomial relationships with 540 reinfection time. The small proportion of individuals in the analysis over 20 years of age (294/3840 with 541 available age data) were grouped together, since model convergence problems were created by lack of data 542 at older ages and because age-dependent exposure to mosquito bites (related to body surface area)(65), as 543 well as development of immunity (66), tends to plateau by 20 years of age. Otherwise, linear relationships 544 were assumed for continuous variables. We tested for interactions between AL or AS-AQ treatment, 545 prevalence of the pfmdr1 86Y mutant versus N86 wild type parasites and pfcrt 76T mutant versus K76 wild 546 type parasites, since there is evidence of differential effects of each drug on these parasite genotypes (9,11). 547 We tested for an effect of different formulations of AS-AQ, i.e. fixed dose combination (from Sanofi), 548 blister pack or loose dose (see also Table 1 for dose information). For AL, all included studies used the 549 same fixed dose combination from Novartis. We calculated weight-for-age Z scores for patients under 5 550 years old according to the WHO age and gender specific reference values, using the WHO Anthro software 551 in R (85). Individuals were classified as underweight if they had a Z score of less than -2. We investigated 552 being underweight in the children under five years because this was a factor associated with recrudescence 553 after AL in a previous analysis.(10) We calculated mg per kg dose of lumefantrine or amodiaquine for each 554 patient according to their dose and weight. Goodness of fit of the models was assessed by Akaike's 555 Information Criterion (AIC). We used stepwise regression, with both forward selection and backward 556 elimination to ensure all covariates of interest were identified. The best-fitting model was identified using 557 AIC and covariates significantly improving the prediction (LR-test) were kept. 558 559 Epidemiological Simulations: An existing mathematical model of Plasmodium falciparum epidemiology 560 (66) was used to assess the impact of first-line antimalarial treatment on malaria transmission outcomes. 561 . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint The probability of a mosquito becoming infected when feeding on individuals treated with AL relative to 562 untreated individuals was assumed to be 0.051 (66). It is uncertain whether there is any difference in human 563 infectiousness after treatment with AL versus AS-AQ. We therefore ran the simulations twice, assuming 564 firstly that patients are equally infectious after treatment with either ACT, and secondly assuming that 565 patients treated with AS-AQ are twice as infectious, in approximate accordance with the ratio of areas under 566 curves of post-treatment gametocyte prevalence in Schramm et al (47) which is consistent with a meta-567 analysis showing reduced gametocytemia after treatment with AL compared with AS-AQ (67). The model 568 was first run to equilibrium in the absence of interventions, then we simulated first line treatment with AS-569 AQ or AL, assuming that 80% of clinical episodes are treated with an antimalarial, that both drugs are 95% 570 efficacious at clearing parasites and that the switch is instantaneous and complete. Prior to introducing ACT, 571 we assume SP was in use, also at 80% coverage but only 60% efficacy. We simulated a population of 572 600,000 individuals to smooth stochastic variation. We adjusted mosquito densities to represent low, 573 medium and high transmission areas (pre-intervention slide prevalence in 2-10 year olds=5%, 15% and 574 50%, respectively in the non-seasonal settings). In the simulations with seasonal variation, we adjusted 575 mosquito densities to achieve the same annual EIR as in each respective low, medium or high transmission 576  . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Tables   Table S1. Risk factors for reinfection: multivariable analysis with pfcrt 76T.

Supplementary Figures and
Data from AS-AQ and AL trial arms were analyzed separately using accelerated failure-time analysis.
Regression coefficients are the ratio of time to reinfection, such that a coefficient>1 indicates a longer time to reinfection. Covariates significantly associated with reinfection time after adjusting for EIR (Table 3, main text) were included in the final model. The prevalence of pfmdr1 86Y also had a significant effect in a multivariable model with the same covariates (Table 3, main text) but could not be included in the same model with pfcrt 76T due to strong correlation between the two variables.
Models assume a log-normal time to reinfection and random site effects. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Figure S1. The duration of post-treatment prophylaxis at different trial locations in order of increasing estimated EIR.
Posterior estimates of the duration of protection provided by AL or AS-AQ are shown. The study sites are shown in order of increasing transmission intensity left to right according to posterior EIR estimates.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Figure S2. Correlation between pfcrt 76T prevalence and pfmdr 86Y prevalence, in the surveys matched to the trial sites according to year and geographic distance. When more than one molecular marker survey was matched to a trial site, a weighted average prevalence was taken. In some cases, these two molecular markers were assessed in the same matched survey(s), but in other cases matches from different surveys were found.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Figure S3 Simulated annual seasonal variation in EIR assumed in the analysis of potential impact of AL and AS-AQ on population level transmission ( Figure 5, main text). The EIR shown is for the simulated seasonal medium transmission setting (slide prevalence = 15%), but the relative EIR variation across the year was the same in the seasonal low and high simulated transmission settings.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint Figure S4. As Figure 5 in the main text, except panels B,C,E and F show impact on clinical incidence in the whole population (rather than 0-5 year old children only).
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19002741 doi: medRxiv preprint