Efficacy of Single-Dose Primaquine With Artemisinin Combination Therapy on Plasmodium falciparum Gametocytes and Transmission: An Individual Patient Meta-Analysis

Abstract Background Since the World Health Organization recommended single low-dose (0.25 mg/kg) primaquine (PQ) in combination with artemisinin-based combination therapies (ACTs) in areas of low transmission or artemisinin-resistant Plasmodium falciparum, several single-site studies have been conducted to assess efficacy. Methods An individual patient meta-analysis to assess gametocytocidal and transmission-blocking efficacy of PQ in combination with different ACTs was conducted. Random effects logistic regression was used to quantify PQ effect on (1) gametocyte carriage in the first 2 weeks post treatment; and (2) the probability of infecting at least 1 mosquito or of a mosquito becoming infected. Results In 2574 participants from 14 studies, PQ reduced PCR-determined gametocyte carriage on days 7 and 14, most apparently in patients presenting with gametocytemia on day 0 (odds ratio [OR], 0.22; 95% confidence interval [CI], .17–.28 and OR, 0.12; 95% CI, .08–.16, respectively). Rate of decline in gametocyte carriage was faster when PQ was combined with artemether-lumefantrine (AL) compared to dihydroartemisinin-piperaquine (DP) (P = .010 for day 7). Addition of 0.25 mg/kg PQ was associated with near complete prevention of transmission to mosquitoes. Conclusions Transmission blocking is achieved with 0.25 mg/kg PQ. Gametocyte persistence and infectivity are lower when PQ is combined with AL compared to DP.

gametocyte carriage can persist for several days and even weeks after ACT administration [3,6] and treated individuals can continue to be a source of mosquito infections [3,7,8]. As malaria control programs focus their efforts on regional elimination and global eradication and the necessity to contain drug-resistant parasites, targeting gametocytes as part of routine clinical care and community treatment campaigns is being recommended [9][10][11].
Primaquine (PQ), a drug that is used routinely for the radical cure of Plasmodium vivax and Plasmodium ovale infections, has been recast as a viable treatment strategy to reduce P. falciparum transmission. The ability of PQ and its predecessor plasmoquine to stop P. falciparum infectivity to malaria vectors has been known for many decades [12,13]. In 2012, the World Health Organization (WHO) recommended the use of PQ, in combination with ACTs, in areas approaching elimination and where artemisinin resistance was observed [10]. To mitigate concerns related to hemolysis in individuals with glucose-6-phosphate dehydrogenase (G6PD) deficiency and based on efficacy shown at low doses, a single low dose of 0.25 mg/kg of PQ was recommended for the gametocytocidal indication [10]. The safety of single low-dose PQ was confirmed in subsequent safety studies in individuals with G6PD deficiency [14,15]. Multiple efficacy studies have been conducted to determine the gametocytocidal and transmissionblocking activity of PQ at different doses and with different partner ACTs.
We conducted a systematic review and individual patient data (IPD) meta-analysis of clinical trials to quantify the ability of single-dose PQ given in combination with different ACTs to clear gametocytes and block transmission, and to compare efficacies of different combinations.

Data Pooling
Details of the systematic review (PROSPERO CRD42019126710) are provided in the Supplementary Material. Studies were eligible for the inclusion in this analysis if (1) IPD came from a clinical efficacy trial of patients with uncomplicated P. falciparum infection or asymptomatic parasite carriers containing at least 1 study arm with a combination of a blood schizonticide and a single dose of PQ; (2) patient demographics and information on dosing (mg/kg) of the blood schizonticide and PQ were available; (3) transmission potential was assessed by weekly gametocyte carriage (ie, prevalence) using molecular methods and/or by membrane feeding assay conducted on day 0 and any day post treatment; and (4) patients were followed up at least until day 14. In the eligible studies, non-ACT study arms, which were randomized to receive PQ or not, were also included in the analysis as they contributed to the overall estimate of PQ effect.

Ethics
All data included in this analysis were obtained in accordance with the laws and ethical approvals applicable to the countries in which the studies were conducted, and were obtained with the knowledge and consent of the individual to which they relate. Data were fully anonymized either before or during the process of uploading to the WorldWide Antimalarial Resistance Network repository. Use of existing data that are fully anonymized and that researchers cannot trace back to identifiable individuals does not require the review of the Ethics Committee under the guidelines of the Oxford University Research Ethics Committee.

Statistical Analysis
Statistical analyses were carried out according to an a priori statistical analysis plan [16]. The prevalence of gametocytemia on days 7 and 14 after first administration of any treatment (day 0) was determined separately for patients without and with gametocytes on enrollment. Logistic regression models for gametocyte prevalence (0/1), as measured by molecular methods (quantitative reverse-transcriptase-polymerase chain reaction [qRT-PCR] or quantitative nucleic acid sequence-based amplification [QT-NASBA]), on each day were fitted with random intercepts for study site.
Data from membrane feeding experiments were analyzed using logistic regression to identify predictors of (1) probability of a participant infecting at least 1 mosquito, and (2) probability of a feeding mosquito being infected. Random intercepts were included to account for multiple measurements per patient (1) or clustering within a membrane feeding experiment (2).
Additional details such as predictors considered in each of the regression models and assessment of risk of bias analysis are given in Supplementary Material.

RESULTS
The systematic review identified 13 studies eligible for inclusion and 2 additional studies were identified in response to the call for data (Supplementary Figure 1). IPD from 14 studies were shared; their details are presented in Supplementary Table 1. Five studies used QT-NASBA (including 2 where quantification was not performed), 8 used qRT-PCR, and 1 study used both. The target transcripts in these molecular assays included Pfs25, Pfs230p, and Pfg377 mRNA. In addition to sexual-stage specific parasite detection, 3 of these studies also included data from membrane feeding experiments, where infectiousness was directly quantified by feeding mosquitoes on infected blood and assessing oocyst development 1 week later. G6PD deficiency was assessed using fluorescence spot test in 4 studies, rapid diagnostic test in 5 studies, or genotyping in 3 studies. All studies, except 1 from Colombia, were conducted in Africa at sites with varying transmission intensities. Administration of PQ was randomized and compared to a no-PQ arm in all studies except for 1 in which the dose of PQ was increased sequentially (study 8 Abbreviations: AL, artemether-lumefantrine; ASSP, artesunate and sulfadoxine-pyrimethamine; DP, dihydroartemisinin-piperaquine; G6PD, glucose-6-phosphate dehydrogenase; Hb, hemoglobin; N, number of patients evaluated; n, number of patients in that category; QT-NASBA, quantitative nucleic acid sequence-based amplification; RT-PCR, reverse transcription polymerase chain reaction; SPAQ, sulfadoxine-pyrimethamine and amodiaquine; WAZ, weight-for-age score. a Includes 20 patients who received DP and methylene blue and only contributed baseline data from membrane feeding experiments. 12.8% (239/1860) with fever, and 5.8% (139/2392) had more than 100 000 parasites/µL (Table 1); 12.2% (59/484) of the children <5 years of age were underweight (weight-for-age z-score < −2). The proportion of participants with fever at enrolment was lower in the group of individuals receiving PQ compared to the group that did not receive PQ (9.9% vs 18.2%, respectively); however, the difference was not significant after adjusting for study site (P = .966). Six studies' protocols excluded individuals with G6PD deficiency (Supplementary Table 1).  Table 2). The effect of PQ on gametocyte appearance was similar (P = .308) between day 7 (OR, 0.58; 95% CI, .33-1.01; P = .053) and day 14 (OR, 0.30; 95% CI, .14-.63; P = .002).  Table 2 and Figure 1). In multivariable mixed effects models, gametocyte positivity on day 7 was associated significantly with gametocyte and asexual parasite densities and hemoglobin concentration at baseline (  Figure 2). Whilst addition of PQ reduced gametocyte carriage for both ACTs, the rate of decline in gametocyte carriage associated with PQ dose differed between patients treated with AL and DP (test for interaction, P = .010 for day 7 and P < .001 for day 14). Among individuals treated with AL, most of the reduction in gametocyte carriage probability was achieved with the recommended 0.25-mg/kg PQ dose, whereas in individuals treated with DP higher doses of PQ were associated with additional substantial reductions in gametocyte carriage. Administration of a PQ dose of 0.25 mg/kg in patients treated with AL reduced risk of gametocytemia on day 7 to 26.0% (95% CI, 18.7%-34.9%) and on day 14 to 7.6% (95% CI, 4.3%-13.2%) compared to 37.1% (95% CI, 27.6%-47.8%) and 18.2% (95% CI, 11.4%-27.9%) in patients treated with DP, respectively.

Gametocytemia After Treatment in Participants With
The risk for gametocyte carriage was significantly higher on day 7 in patients treated with PQ on day 2 or 3 compared to patients treated with PQ on day 0 (AOR, 2.28; 95% CI, 1.66-3.69; P < .001, adjusted for covariates in the main analysis; Table 2). However, this difference was not statistically significant by day 14 (AOR, 1.74; 95% CI, .80-3.81; P = .164, adjusted for covariates in the main analysis; Table 2).
Administration of PQ also reduced gametocyte density in those positive on days 7 or 14. Expressed as a proportion of the baseline gametocyte density, gametocyte densities reached median values of 2.0% (interquartile range [IQR], 0.3%-10.2%) relative to baseline by day 7 in PQ-treated individuals compared to 29.8% (IQR, 8.1%-77.4%) in individuals who did not receive PQ (P < .001 Wald test, adjusted for ACT and study). The corresponding values on day 14 were 0.5% (IQR, 0.1%-5.6%) in PQ-treated individuals and 9.6% (IQR, 1.5%-36.0%) in individuals who did not receive PQ (P < .001, Wald test adjusted for ACT and study).

Mosquito Feeding Assays
In the 3 studies undertaking mosquito feeding experiments (Supplementary Table 1 and Supplementary Table 3), participants were treated with either AL (1 study), DP (2 studies), or SPAQ (1 study) and a PQ dose of 0.25 mg/kg was compared to ACT alone in all studies. In 1 of these studies, the 0.40-mg/ kg dose was tested, and in another study, PQ doses of 0.0625, 0.125, and 0.50 mg/kg were also administered. These data are presented in Supplementary Table 4.
Among 316 feeding experiments conducted prior to treatment on participants with baseline gametocytemia, 186 (58.9%) infected at least 1 mosquito, with a median of 13.9% (range, 1.2%-96.5%) of mosquitoes infected ( Figure 3 and Supplementary  proportion of the infected mosquitoes (in infectious feeds) was similar between the 3 studies (P = .369), the number of noninfectious feeds ranged from 37.8% to 67.9% (P < .001) between studies, with the lowest proportion observed in study 6 (AL/AL + PQ). This study had the lowest baseline gametocytes levels; 79.0% of patients had fewer than 50 gametocytes/µL compared to 24.7% and 42.5% in the other 2 studies.
In patients with confirmed gametocytemia at baseline and at the time of sampling post treatment, 13.2% of feeds (36/272) of those treated with PQ infected at least 1 mosquito, compared to 35.6% (63/177) of non-PQ-treated patients sampled at the same timepoints ( Figure 3 and Supplementary   In studies where only gametocyte positivity was determined by a molecular method, density measures by microscopy were included. For patients with positive samples by molecular method and zero microscopy count (n = 230 on day 7 and n = 180 on day 14), density was assumed to be 8 (half of the detection limit by microscopy assuming microscopic quantification against 500 white blood cells or 1/16th of a microliter).   Abbreviations: AL, artemether-lumefantrine; AOR, adjusted odds ratio; CI, confidence interval; DP, dihydroartemisinin-piperaquine; PQ, primaquine; SPAQ, sulfadoxine-pyrimethamine and amodiaquine. a Estimates also adjusted for study included as a covariate The risk of a participant infecting at least 1 mosquito and the risk of a feeding mosquito becoming infected were strongly associated with gametocyte density at the time of mosquito feeding (AOR, 8.33; 95% CI, 3.91-17.78 and AOR, 6.58; 95% CI, 4.16-10.40 for 10-fold increases in gametocyte density, respectively) and significantly decreased following PQ treatment ( Table 3). The reduction in odds of mosquito infectivity over time associated with PQ dose of 0.25 mg/kg was significantly higher compared to lower doses (0.0625-0.125 mg/kg) (ratio of AORs per day, 17.84; 95% CI, 4.93-64.52; P < .001 for a participant infecting at least 1 mosquito and 10.36; 95% CI, 4.67-22.98; P < .001 for a mosquito becoming infected) and not statistically different from higher doses (0.4-0.5 mg/kg) (P = .433 and P = .070, respectively). With the exception of those treated with AL, the odds did not decrease significantly over time for any of the schizontocidal drugs. A PQ dose of 0.25 mg/kg decreased the risk of infecting at least 1 mosquito practically to zero by day 3 (Figure 4 and Supplementary Figure 2).

Risk of Bias
Methodological factors potentially contributing to the risk bias and attrition bias are presented in Supplementary Table  5. Measurement of gametocyte carriage using molecular methods is automated, minimizing the risk of observer bias; laboratory personnel performing molecular assays or dissecting mosquitoes were blinded in all studies. Sensitivity analyses showed that exclusion of any of the studies did not change the main conclusions of the analysis. The effect of PQ dose on gametocyte positivity was estimated as median AOR 0.69 (range, 0.65-0.70) on day 7 and 0.58 (range, 0.54-0.62) on day 14 for a 0.1-mg/kg increase.
The only eligible study for which data were not available for this meta-analysis [8] presented similar findings to results of this analysis. In this study, the addition of a single dose of 45 mg of PQ to DP treatment was associated with increased clearance of gametocytes (measured by PCR) on day 7 and day 14. In the PQ arm, of 24 patients with gametocytes on enrolment, 22 cleared gametocytemia by day 7 and all by day 14, compared to 11 (day 7) and 16 (day 14) of the 22 patients in the DP only arm. In their membrane feeding experiments, no mosquito infections occurred in the PQ arm 1 and 2 weeks post treatment, while in the no-PQ arm 6.9% of feeding mosquitoes were infected on day 7 and 5.0% on day 14.

DISCUSSION
This IPD meta-analysis estimated the effect of PQ as a single dose (ranging from 0.0625 to 0.75 mg/kg) on the transmission potential of falciparum malaria infections, when coadministered with schizonticidal drugs. Our findings confirm the gametocyte clearing and sterilizing effects of single-dose PQ and indicate that both the PQ and the schizonticidal partner drug are important determinants of gametocyte clearance and transmission potential. Regardless of the schizonticidal partner drug, mosquito infections were rarely observed 1 week after administration of PQ; however, only 3 of the 14 studies contributed data to this analysis. Among currently licensed antimalarials for P. falciparum, PQ is unique in its ability to clear mature gametocytes persisting after ACT treatment. Because the impact of ACTs is largely restricted to immature, developing gametocytes [17], only a small proportion of infections develop gametocytes after ACTs whilst gametocytes that are present prior to treatment may persist [6]. In the current analysis, more than 20% of individuals who were gametocyte negative at enrolment became gametocyte positive by molecular gametocyte detection methods shortly after treatment. Given that gametocytes first appear 8.5-12 days after their asexual progenitors [18] and transcripts specific to mature gametocytes are first observed on day 3 based on the current data, it is likely that this reflects density fluctuations of mature gametocytes already present prior to treatment [19], rather than de novo gametocyte production. In line with this, PQ administration prior to first detection of gametocytes reduced the proportion of patients with gametocytes during follow-up.
Gametocyte kinetics in patients who presented with peripheral gametocytemia were strongly dependent on the schizontocidal treatment administered. Non-ACTs leave gametocytes largely unaffected, with gametocyte kinetics resembling a natural decay, while ACTs are only effective against early gametocytes [2,20]. Also, ACTs differ markedly in their impact on gametocyte carriage [6,7,21], potentially due to the effects of the nonartemisinin partner drugs. Whilst lumefantrine affects gametocytes and their infectivity [22], piperaquine has limited effect on either developing or mature gametocytes [23]. Furthermore, the artemisinin derivative dose recommended by the manufacturer is significantly higher for AL than for DP. In the current pooled analysis, individuals receiving AL were considerably less likely to have gametocytemia on day 14 compared to DP (AOR, 0.18; 95% CI, .08-.44) and considerably less likely to infect mosquitoes. The addition of PQ significantly reduced gametocyte carriage in all treatment groups [24] and did so in a dose-dependent manner [25]. When given in combination with AL, the 0.25-mg/kg WHO-recommended dose reduced gametocyte prevalence 7 days after treatment initiation to 22%, and this reduction is similar to that observed for higher PQ doses (16%, P = .202). For individuals receiving DP, the average gametocyte prevalence reduction for 0.25 mg/kg PQ was only to 39% on day 7 post treatment but higher PQ doses accelerated gametocyte clearance (to 15%, P = .002), and a 0.40-mg/kg PQ dose coadministered with DP achieves a similar effect to a 0.25mg/kg dose coadministered with AL.
However, gametocyte sterilization may precede gametocyte clearance [26,27]. In 3 studies included where mosquito infection was used as an endpoint, the effect of PQ on preventing mosquito infection was apparent before gametocytes were fully cleared. Whilst the gametocyte clearing effect of PQ only became apparent on day 7 post initiation of treatment, mosquito infections were already very rare on day 3 following treatment with 0.25 mg/kg PQ. PQ doses below 0.25 mg/kg were associated with higher mosquito infection rates on day 3 whilst doses higher than 0.25 mg/kg did not augment or accelerate the transmission-blocking properties of PQ.
Use scenarios for single-dose PQ include elimination settings and areas threatened by drug resistance [10]. The findings from this meta-analysis, of increased gametocyte clearance and near absence of mosquito infections after administration (only 10/220 individuals who received at least 0.25 mg/kg PQ infected mosquitoes in feeding assays), support PQ deployment in these scenarios. PQ has been coadministered with schizonticides in community-wide treatment campaigns [9,28,29], on the assumption that asymptomatic infections constitute a substantial proportion of the human infectious reservoir for malaria in low-endemic settings [30,31]. However, concerns have been raised regarding the risk to benefit ratio in these settings. A proportion of these populations are likely to be G6PD deficient with a concern that they may be at an increased risk of PQ-induced hemolysis. The WHO-recommended single low dose of PQ has shown no significant risk in recent studies specifically designed to assess safety in this population [14,15], nor in recent studies primarily designed to determine PQ efficacy [32][33][34]. Results of an IPD meta-analysis of all available safety data will be published separately (PROSPERO CRD42019128185).
While CYP2D6 activity is essential for the generation of metabolites implicated in hypnozoite-clearance in P. vivax [35,36], less is known about its potential impact on gametocytocidal or transmission-blocking properties of PQ. Whilst PQ's gametocytocidal activity may in part be unrelated to cytochrome CYP2D6 activity [36], gametocytes may persist longer after PQ treatment in individuals with low-moderate CYP2D6 activity [37]. A shortcoming of our meta-analysis is that we could not incorporate these possible effects of CYP2D6 metabolizer status on post-PQ gametocyte carriage or transmission. In general, the added value of gametocytocidal drugs in community treatment campaigns continues to be a matter of debate. Mathematical simulations indicate that the fraction of the asymptomatic population that is successfully treated with ACTs is considerably more important for the impact of treatment campaigns than the addition of PQ to ACTs and that impact will depend on transmission intensity [38][39][40].
This study also highlights SPAQ's poor ability to clear gametocytes with a considerably higher gametocyte prevalence on day 7 post initiation of treatment compared to DP or AL [41]. Seasonal malaria chemoprevention (SMC) using SPAQ is widely deployed across the Sahel region of Africa to reduce malaria morbidity in children younger than 5 years and consists of giving all children SPAQ 3 to 4 times monthly during the transmission season. In scenarios where SMC campaigns are considered in wider age groups, SMC may impact gametocyte carriage [42] and malaria transmission. For such scenarios, our findings suggest that either adding single low-dose PQ to SPAQ or changing to an artemisinin-based combination of drugs may increase SMC impact [3].

CONCLUSIONS
Our analysis, based on IPD from clinical trials that were primarily conducted in Africa, supports the use of PQ as a potent gametocytocide and transmission-blocking tool for P. falciparum malaria. Gametocyte carriage and transmission after PQ treatment depend on the schizonticidal drug that PQ is combined with, and PQ doses higher than 0.25 mg/kg may accelerate gametocyte clearance. However, this WHO-recommended dose effectively achieves near-complete reductions in mosquito infections regardless of ACT. Additional clinical trials are necessary to quantify the effect of PQ use at community level; that is, to determine whether the effect of PQ observed in mosquito feeding assays leads to detectable changes in community-wide transmission levels when the drug is systematically used in clusters of transmission.