Safety and effectiveness of mass drug administration to accelerate elimination of artemisinin-resistant falciparum malaria: A pilot trial in four villages of Eastern Myanmar

Study design

This study was the pilot phase of a multicentre cluster-randomized control trial conducted in several sites in the GMS and therefore limited to a sample size of four villages. It is registered at ClinicalTrials.gov: NCT01872702.

The study was conducted in rural Eastern Kayin (Karen) state of Myanmar, a region endemic for P. falciparum and P. vivax malaria¹⁷. In early 2013, 12 villages were screened using a high volume ultra-sensitive qPCR assay method (uPCR) with the original intention of identifying villages with >30% malaria prevalence, of which >30% was P. falciparum (Supplementary File 1: Table S1). However, the P. falciparum prevalence was lower than expected, so after the survey the threshold proportion of P. falciparum for inclusion was lowered to 10%. Four villages fulfilling this revised entry criterion were selected after community engagement and agreement by the village leaders. A malaria post (MP) was set up at month 0 (M0) in all four villages and long lasting insecticidal nets (LLIN) were provided. Two villages (referred to as A1-KNH and A2-TOT) were randomized using a flipped coin to receive anti-malarial MDA for three consecutive months (M0 to M2) immediately. The other two villages (referred to as B1-TPN and B2-HKT) were monitored for a 9-month control period with MP and LLIN only. The control period was planned originally for 12 months, but MDA had to be expedited because of accessibility concerns during the rainy season, and was given from M9 to M11. All four villages were followed for 24 months in total.

Ethics statement

This study protocol was reviewed and approved by OxTREC (reference no. 1017–13 and 1015–13), by the Tak Community Advisory Board¹⁸, and by village committees. Participation in surveys and MDA activities was voluntary. In addition to group information during community engagement activities, participants received individual information in their language (Karen or Myanmar) and provided written informed consent before inclusion in a survey or MDA. Participants under 18 years provided an assent in addition to the consent of their parent or guardian.

Study participants

Participants were defined as individuals living in the village who provided information during home visits at any time point. Each participant was followed using a unique identification code to collect information on participation, adverse events, malaria carriage during surveys, clinical episodes of malaria, and mobility.

Interventions

Before and during the study, community engagement (CE) activities were conducted (detailed in 19). Briefly, the aim of community engagement was to present the project to the community and involve its members to create understanding and buy-in of the population, and result in a high coverage of the interventions²⁰. The different activities conducted were: workshops with authorities and local gatekeepers, meetings and activities with different population groups, health-related community incentives (ancillary care and health education; building of water catchment and distribution systems in each of the study villages)¹⁹. Through this mutual understanding, the study activities could be conducted while respecting the life of the community and ensuring maximal benefit for the participants.

A population census was conducted, individuals were linked to houses (with both individual and house identification codes), and geographic references (latitude and longitude) were recorded for each house. Population movement in and out of villages was monitored by census during the quarterly surveys and by home visits. Newcomers intending to stay in the village for more than two weeks were enrolled after consenting to participate in the follow-up surveys. All newcomers arriving after MDA were offered a single curative treatment course of DP-primaquine, irrespective of disease or infection status.

A MP was set up at M0 in all four villages to provide early detection and treatment of clinical cases⁷. Fever cases were tested with SD Bioline Pf/Pv rapid diagnostic test (RDT). Uncomplicated P. falciparum cases were given weight-adjusted doses of dihydroartemisinin-piperaquine (DP, Guilin Pharmaceutical Co, Guilin, PRC) for three days and one 0.25mg base/kg dose of primaquine (Governement Pharmaceutical Organization, Thailand). Uncomplicated P. vivax cases were given weight-adjusted doses of chloroquine (Governement Pharmaceutical Organization, Thailand) for three days. Severe illnesses were referred to the nearest clinic. LLINs were distributed to all households at M0.

Blood sampling surveys of the entire village population were conducted at months 0, 3, 6, 9, 12, 15, 18, and 24. Samples of 3 and 0.5 mL of blood were taken by venous puncture from survey participants over and under 5 years respectively. Malaria infection was detected in the laboratory by RDT, microscopy and by a uPCR method with an approximate limit of detection of 20 parasites/mL^8,21.

Each MDA comprised a standard 3–day treatment course of DP (dihydroartemisinin (7mg/kg) and piperaquine (55 mg/kg) and one 0.25mg base/kg dose of primaquine given orally under supervision. Treatments were given three times at one-month intervals (M0 to M2 or M9 to M11).

Inclusion and exclusion criteria

Infants <6 months, individuals with a known allergy to any of the antimalarial drugs, and pregnant women in their first trimester were excluded from MDA participation. Pregnant women in their second or third trimester and breastfeeding mothers were eligible for DP, but excluded from primaquine treatment. Women of childbearing age were asked if they were or could be pregnant and a pregnancy test was offered if unsure. Infants <6 months were excluded from prevalence surveys.

Adverse event monitoring

Adverse events (AE) were monitored using a structured questionnaire on the second, third, and seventh day after the start of the MDA treatment course. During the three months of MDA, a mobile clinic staffed with a medical assistant was available in the village to provide free consultations to all villagers and to assess MDA participants presenting with AE. Serious adverse events were reported centrally.

Variable collection and definition

Participant follow-up. The individual follow-up time was defined as the number of days spent within the catchment area of the MP, i.e. the village and the surrounding farms.

Participation in MDA. Participation in MDA was measured as the total number of individuals completing one, two or three 3-day DP treatment courses divided by the total number of inhabitants present in the village at least once during the 3-month period, after excluding individuals who visited the village for two weeks or less. Individual participation in MDA was categorized in three groups: no MDA, one or two rounds (i.e. infection cured, but incomplete protection from reinfection by DP), and three rounds MDA (i.e. infection cured and 3-month protection).

Malaria infection prevalence. Prevalence of P. falciparum or P. vivax infection was defined as the proportion of individuals with a positive uPCR for each parasite divided by total number of individuals sampled. When the parasite species could not be determined (because of low DNA content) samples were attributed to species based on the falciparum/vivax ratio measured in clinical cases in the village during the previous period⁸.

Incidence of symptomatic malaria. Symptomatic malaria cases were defined as individuals with fever (temperature ≥37.5°C) or history of fever in the past two days and confirmed falciparum or vivax infection by RDT or microscopy. During the period between surveys, cases were detected passively at the MP and confirmed by RDT. During survey periods, individuals with fever or history of fever and a positive RDT or microscopy slide (with uPCR confirmation when available) were considered clinical cases. Incidence of clinical malaria episodes in the village was expressed in cases per 1000 persons per month (number of infections during the month/sum of individual follow-up time during the month). Patients residing in nearby villages and coming to the MP for diagnosis were excluded from the analysis. Mixed infections contributed to both P. falciparum and P. vivax incidence.

Detection of molecular markers of antimalarial resistance

Assessment of mutations in PfKelch13. Polymorphisms in the PfKelch gene were assessed by nested PCR amplification covering the full length of the gene (total 2181 bp)², and followed by sequencing of the gene by ABI Sequencer (Macrogen Inc, South Korea). Cross contamination was monitored by adding negative control samples in every run. Sequencing results were aligned against PfKelch13 of reference strain 3D7 (putative 9PF13_0238 NCBI Reference Sequence (3D7): XM_001350122.1), using Bioedit software (Abbott, CA, USA). Polymorphic patterns were assessed by two individuals blinded to the origin of the sample.

PfPlasmepsin2 gene amplification. Pfplasmepsin2 copy number was quantified using Relative-quantitative Real-time PCR based on Taqman probe on a Corbett Rotor-Gene^TM Q (Corbett Research, Australia). Primers and probes have been described previously²². Amplification was performed in triplicate on a total volume of 25 μL as multiplex PCR using Quantitec Multiplex PCR no ROX (QIAgen, Germany). Every amplification run contained 9 replicates of calibrators and triplicates without template as negative controls. β-tubulin served as an internal standard for the amount of sample DNA added to the reactions. Copy numbers were calculated using the formula: copy number= 2 ^-ΔΔCt with ΔΔ C_t denoting the difference between Δ C_t of the unknown sample and Δ C_t of the reference sample.

Entomology

Entomological surveys to identify malaria vectors species and abundance were conducted monthly in each village for five nights per month, as described elsewhere^23,24. Briefly, mosquitos were collected using indoor and outdoor human landing catch (HLC) in five sites per village, yielding a total of 50 human.nights per survey. Infection of primary malaria vectors (Anopheles minimus s.l., An. maculatus s.l., and An. dirus s.l.) by plasmodium parasites was determined using a quantitative real-time PCR (qPCR) assay, confirmed with a second assay amplifying a different DNA target²⁵. The limit of detection of the whole procedure was 6 P. falciparum sporozoites per mosquito²⁵. Main entomological indices - human biting rate (HBR), sporozoite index (SI), and entomological inoculation rate (EIR) - were calculated.

Statistical analysis

Statistical analyses were performed using Stata 14.1 (StataCorp, USA). Binomial 95% confidence intervals (CIs) were calculated for prevalence and SI, and Poisson 95% CIs for incidence of clinical episodes, HBR, and EIR. Intervention and control group prevalence were compared using chi-square or Fisher’s exact test, as appropriate.

Pre-specified endpoints

Because the MDA in control villages was conducted earlier than originally planned (M9 instead of M12), the revised primary endpoints were defined as the prevalence of asymptomatic P. falciparum infection 3 and 9 months after MDA. The secondary endpoints were defined as 1) safety and 2) acceptability of MDA, as evaluated by questionnaires. Additional outcome measures reported here are the incidence of clinical malaria over the first 9 months, number of people participating in MDA, and proportion of artemisinin-resistant P. falciparum infections.

Before/after analysis

A post-hoc before/after analysis of the entire duration of follow-up was conducted to understand the factors influencing the impact of antimalarial MDA and the overall intervention. The effect of the intervention on P. falciparum carriage (recorded in prevalence surveys) was estimated using Generalised Estimating Equations (GEE) with a logit link to estimate Odds Ratios (ORs), an exchangeable correlation structure and robust standard errors to account for repeated measurements at individual level. The change in odds of P. falciparum carriage compared to baseline survey was estimated in the control period and in the after MDA period. To assess the duration of MDA effect, time-dependent changes in the odds of P. falciparum carriage were estimated separately for the control and after MDA periods. Demographic characteristics, village, age group (≤ & >10 years), sex and season, were included as covariates. The analysis was repeated to assess the effect of individual MDA participation. A direct comparison of control and after MDA periods was also conducted in which baseline P. falciparum infection status was added as a covariate in addition to the demographic characteristics.

Mass antimalarial drug administration (MDA)

MDA coverage. The target population was 2392 residents of the four villages at the time of MDA. Overall 78% (1866/2392) received at least one 3–day course of DP and 38% (911/2392) received the full course of three treatments. Coverage was higher in the two small villages compared to the two larger villages, for uptake of at least one treatment (A1–KNH: 93%; 360/386 and B1–TPN: 92%; 339/367 compared to A2-TOT: 74%; 487/662 and B2-HKT: 70%; 680/977) and uptake of three treatments (A1–KNH: 61%; 236/386 and B1–TPN: 57%; 210/367 compared to A2-TOT: 28%; 185/662 and B2-HKT:29%; 280/977) (Table 1; Figure 1).

Figure 1. Flow chart of the participation in antimalarial MDA in the four villages.

Safety and tolerability of MDA. MDA was well accepted and well tolerated. In 3719 of 3931 (94.6%) complete structured interviews, no AE were reported. Reported AE were in majority dizziness (N=192) and pruritus (N=17). There were three reports of black urine. One G6PD-deficient male reported black urine before primaquine and one G6PD-normal male reported black urine 48h after primaquine. Neither were anaemic and both remained well. The third case, a G6PD heterozygote woman with normal G6PD activity, became anaemic but concomitant HIV and TB were considered contributory, as reported previously²⁶. The percentage of participants reporting AE declined after each MDA round from 8.2% (114/1394) to 3.7% (49/1336). Among the 200 patients who presented at the mobile clinic over the 3-month MDA period, all AE were mild or moderate and most (188/200, 94%) were rated as unrelated or unlikely to be related to treatment (Table 2). During the 24 months of the study, there were 23 reported serious AE and 15 deaths; none were considered drug related (Supplementary File 1: Table S2). Haemoglobin decreases in G6PD-deficient individuals were minimal as already reported²⁶.

Malaria prevalence

There was a significant difference in baseline P. falciparum prevalence between intervention and control villages (18.7% (150/801) versus 6.8% (58/848); p<0.0001).

The majority of malaria infections detected by uPCR during the cross-sectional surveys were asymptomatic (86%; 1769/2059) and 70% were not detected by microscopy or RDT (1432/2059) (Supplementary File 1: Table S3). Overall, of 217 individuals positive for P. falciparum, 17 (8%) were still positive in the following survey 3 months later, three were positive after 6 months, and one after 9 months. DP was highly efficacious; no participant receiving at least one complete 3-day DP course was positive for P. falciparum in the survey following MDA (M3 or M12), irrespective of previous infection status (Supplementary File 1: Table S4 and Table S5). During the follow-up (excluding MDA intervention period), 69% (707/1018) of newcomers to the villages were included in the first survey after their arrival, of whom 1.6% (11/707) were positive for P. falciparum.

Efficacy of MDA

Pilot trial endpoints: MDA+MP versus MP only comparison

Prevalence. At M3 the prevalence of P. falciparum infection had fallen to 0.4% (2/552; 95% CI 0.04 to 1.3), in intervention villages A1-KNH and A2-TOT, and 2.7% (22/812; 1.7 to 4.1) in non-intervention control villages, B1-TPN and B2-HKT (p=0.001). Comparing M0 to M3, prevalence decreased by 98% in MDA villages compared to a 57% decrease in control villages. At M9, the difference in prevalence was no longer significant: 2.0% (11/564; 1.0-3.5) in MDA villages, and 0.9% (9/959; 0.04-1.8) in control villages (p=0.10). Comparing M0 to M9, MDA villages exhibited an 88% decrease in prevalence, compared to an 85% decrease in control villages.

Incidence. Between M0 and M9, falciparum malaria incidence was 1.0 case/1000 person-months (95%CI=0.4-1.9) in MDA villages (8/8352) compared with 1.3 case/1000 person-months (0.6-2.5) in control villages (10/7469, p=0.49). P. vivax malaria incidence was 8.6 cases/1000 person-months (6.8-10.9) in MDA villages (72/8352), compared with 5.6 cases/1000 person-months (4.1-7.6) in control villages (42/7469, p=0.03).

Before versus after MDA comparison

P. falciparum infection. The impact of MDA on P. falciparum carriage was different in the two intervention villages. In A1–KNH, there was a rapid >95% reduction in P. falciparum prevalence, which was sustained from M3 to M24 (Figure 2). In contrast, in village A2–TOT, prevalence fell between M0 and M3, to reach a peak at M15 during rainy season (Figure 2). In this village, prevalence was maintained around 5% among individuals who did not take MDA, from M3 to M24 (Supplementary File 1: Figure S4). Prevalence in the control villages decreased gradually over the control period, dropping at 0% at M12 immediately after MDA, and remained below after MDA (Figure 2).

Figure 2. Prevalence of P. falciparum and P. vivax carriage over 24 months measured by uPCR in quarterly cross sectional surveys conducted in the four study villages.

A steep decrease in prevalence immediately followed MDA in all 4 villages. The reduction in P. falciparum prevalence was sustained in 3 of 4 villages stabilizing to a <1% prevalence after MDA. In contrast P. vivax prevalence rebounded in 3 of 4 villages. In village A2-TOT, P. falciparum prevalence increased at M15 and M18, mostly from clinical cases occurring while vector densities and thus transmission was increased (Table 5). A decrease in P. falciparum prevalence was observed in the control period (M0–M9) in villages B1 and B2, possibly attributable to the effect of early diagnosis and treatment provided by MP.

Adjusting for age group, sex, village and season, MDA was associated with an overall 12-fold decrease in odds of P. falciparum carriage over the full observation period (21 month for intervention and 15 months for controls) compared to baseline (OR [95%CI]=0.08 [0.05–0.11]) (Table 3). The control period before MDA was also associated with a significant decrease in odds of P. falciparum carriage (0.4 [0.3–0.7]) (Table 3). This translated into a significant 5-fold odds reduction after MDA compared to control period (0.2 [0.1–0.3]).

Compared to baseline, the decrease in odds of P. falciparum carriage was 30-fold immediately after MDA (0.03 [0.02–0.06]). Each additional month after MDA was associated with a minimal increase in odds of P. falciparum carriage (1.08 [1.03–1.12] for one additional month). For example, 12 months after MDA, a 15-fold reduction persisted (0.06 [0.04–0.09]). Similar results were obtained when estimating odds for each quarter instead of assuming a monthly linear trend (Supplementary File 1: Figure S5).

During the pre-MDA control period, individuals with baseline P. falciparum infections had a significantly increased odds of subsequent P. falciparum carriage compared to baseline uninfected individuals (17.2 [6.5–45.6]) (Supplementary File 1: Table S8). Irrespective of their baseline infection status, individuals taking one or two rounds and individuals completing the three rounds of MDA had a significantly decreased odds of being infected with P. falciparum compared to individuals in the control period (0.2 [0.1–0.6] and 0.1 [0.03–0.4] respectively). In contrast, individuals who did not receive MDA in villages allocated to it had no decrease in the odds of P. falciparum carriage (0.4 [0.1–1.9]) compared to the control period (Supplementary File 1: Table S8).

P. falciparum clinical malaria. A total of 113 P. falciparum symptomatic cases were detected among study participants, of which 91 were detected passively by the MP and 22 were detected during surveys. An additional 19 cases were detected among non-participants (individuals who reported living in the village but were not recorded in the study, e.g. visitors).

In 3 out of 4 villages, the combined intervention of early diagnosis and treatment through the MP and MDA resulted in a dramatic and sustained decrease in the incidence of P. falciparum clinical cases, which was close to zero after 12 months (Figure 3). Most of the clinical episodes (75/113, 66%) occurred during a 7–month period in village A2–TOT between July 2014 and January 2015 (M14 to M20), corresponding to the rainy and following cool seasons.

Figure 3. Changes in the incidence of P. falciparum and P. vivax clinical cases.

Clinical falciparum malaria cases persisted up to 9 months in villages achieving stable <1% P. falciparum prevalence after MDA (A1, B1 and B2), before zero incidence was achieved. In village A2, where the intervention with MDA was not considered successful, a significant increase in incidence was observed during the rainy and cold seasons from July 2014 to January 2015.* P. vivax incidence rate in B2-HKT for June 2015 was 100 cases/1000 person-month (95%CI=81-122).

P. vivax infection and clinical malaria. In 3 out of 4 villages, P. vivax prevalence was reduced only transiently by MDA (Figure 2). Even when MDA participation was high, P. vivax prevalence rebounded quickly to reach initial levels within 3 to 6 months. In village B1–TPN, P. vivax prevalence did not rebound as rapidly. P. vivax symptomatic malaria incidence was unaffected beyond the 3-month MDA period in three villages, with seasonal peaks of clinical cases during transmission seasons both before and after MDA (Figure 3). In village B1–TPN, the incidence did not return to pre-MDA levels.

Prevalence of molecular markers of antimalarial drug resistance

Data on the artemisinin resistance marker Pfkelch13 (PfK13) was available for 107 of 247 detected P. falciparum infections. The prevalence of PfK13 mutations was 77.6% (83/107) (Table 4). The most frequent mutations were C580Y (37.4%; 31/83) and G538V (24.1%; 20/83). The prevalence of PfK13 mutants was 85.6% (95%CI=75.9–92.6) before, and 56.7% (37.4–74.5) after MDA.

No plasmepsin 2 amplification (the molecular marker of piperaquine resistance) was found in the 69 samples that could be analysed (53 before MDA and 16 after) (Table 4).

Entomological findings

The main anopheles vectors in the villages belonged to the Minimus Complex (41% of total anopheles), Maculatus Group (11%) and Dirus Complex (1%). Overall, the SI of primary malaria vectors was 2.2/1000 for P. vivax and 0.4/1000 for P. falciparum (Table 5). During the period before MDA, the HBR was between 200 and 300 bites per human per month in all four villages. After MDA, the HBR was unchanged in A1-KNH and B2-HKT villages, decreased in B1-TPN, and increased in A2-TOT. In spite of these variations in abundance of vectors, SI and EIR decreased after MDA in A2-TOT. Before MDA, P. falciparum SI was 0.9/1000 (95%CI=0.5-1.6) compared to 0.2 (95%CI=0.05-0.4) after (p=0.0001). Likewise, before MDA, P. falciparum EIR was 0.25 infective bites/human/month (95%CI=0.13-0.43) compared to 0.06 (95%CI=0.02-0.15) after (p=0.005).

Limitations of the study

The main limitation of this study is that, being a pilot investigation, it is underpowered to draw definitive conclusions on the overall benefit of this approach. The comparison between intervention and control is indeed strongly influenced by individual village characteristics: although the four villages had similar malaria prevalence in the pre-survey (Table 1, Supplementary File 1: Table S1), they were significantly different in the baseline exhaustive survey with a higher prevalence of asymptomatic malaria in the villages receiving MDA first.

A second limitation was that the beneficial effects of MDA could not be evaluated independently from the continuing effects of the malaria post and the LLINs. Their benefits are illustrated in the control villages where there was three-fold reduction in P. falciparum prevalence over a nine month observation period until MDA was conducted. As the two control villages had a relatively low baseline P. falciparum prevalence, the impact in a higher prevalence setting of instituting a malaria post and distributing LLIN without providing MDA cannot be predicted from these observations. Nevertheless providing MDA in these control villages after nine months reduced prevalence by 100% immediately after MDA, showing an additional effect of MDA to enhanced control measures, as shown recently in Zambia¹⁵.

Third, those who refused MDA were less likely to participate in surveys (as in village A2-TOT). This led to underestimation of the true prevalence of malaria, which was close to zero in the general survey population but 5% in the group of survey participants who did not take MDA. Individual participation in MDA was therefore included in the statistical models of impact to take this into account.

Interpretation

The sustained reduction in P. falciparum prevalence in village A1-KNH, where MDA coverage was high, suggests a reduction of malaria transmission. This is supported by finding reductions of similar magnitude in P. falciparum SI and EIR, while HBR remained stable. By contrast, in A2–TOT, P. falciparum prevalence and incidence increased as vector abundance rose during the rainy and cold seasons of 2014, 14 to 20 months after MDA (Table 5). In this community broken by years of conflict, participation with MDA was low and the reservoir of sub-microscopic malaria parasite carriage was reduced only partially. The non-MDA participants may have contributed as a reservoir for P. falciparum, as indicated by a continuously high prevalence of asymptomatic P. falciparum carriage (5%) in this group. In addition, the MP was not functioning properly during the first year, with frequent interruptions of activity and the lowest rate of consultations among the four villages (Table 1 and Supplementary File 1: Figure S6). These observations illustrate the ‘real-life’ challenges that would confront a programme aiming for rapid elimination, and confirm the well described necessity of ensuring high coverage in any MDA campaign. They also underline the essential importance of providing diagnosis and effective treatment, both immediately after MDA and in an outbreak context as experienced in A2-TOT between M14 and 20. The low prevalence observed at M24 (2%) is likely the result of the adequate treatment of clinical episodes by the MP during the period M12 to M24, contrasting with the dysfunctions observed between M0 and M12.

Importantly this intervention does not appear to have modified the resistance profile of P. falciparum in the area. The prevalence of mutations in the artemisinin resistance marker PfK13 did not increase and there was no indication that piperaquine resistance was selected, although numbers were small. No markers of piperaquine resistance were found either in >200 samples from clinics covering the same area and period, where DP was used for routine treatment of clinical cases (Miotto O, unpublished data).