Background
Following the renewed global call for eradication, the widespread rollout of insecticide-treated nets (ITNs), and increased usage of artemisinin-based combination therapy for first-line treatment, malaria burden has declined substantially over the last decade [
1]. Yet malaria remains recalcitrant in many areas, including high-burden countries where progress has stalled or reversed in the last few years [
2] as well as near-elimination areas where interruption of transmission continues to remain out of reach.
In areas where malaria persists despite high usage of ITNs, outdoor exposure can contribute a major share of residual transmission [
3]. Insecticide resistance is increasingly widespread, diminishing the community benefits of ITNs and potentially erasing much of the gains made in the last 20 years of malaria control [
4,
5]. Continued innovation along the path to eradication is needed to address both challenges.
One potential avenue for complementing the existing toolset is the use of systemic insecticides, known as endectocides when they also have antiparasitic properties, to reduce vector populations. Ivermectin has been distributed in mass drug administrations (MDA) for onchocerciasis and lymphatic filariasis [
6], and is lethal to mosquitoes upon ingestion during blood feeding on humans or animals [
6‐
9]. However, ivermectin has a short half-life of under 4 days in large mammals [
10] and 12–36 h in humans [
11]. While higher doses have been used to treat scabies, for example, at the currently approved standard doses given for treating non-malarial diseases, ivermectin concentrations in human blood can maintain mosquitocidal effects for approximately 48 h [
8,
12], which is predicted to be too short to have much impact on malaria burden [
13]. While the low concentrations of ivermectin observed 28 days post-dose continue to have a small impact on mosquito survival [
14], the epidemiological relevance is unknown. The duration of ivermectin efficacy can be extended via multiple doses over the course of weeks, but this could prove operationally challenging to deliver at scale. There has been recent interest in developing longer-lasting systemic insecticide formulations for malaria control and potential elimination, including slow release formulations [
8,
12] and high-dose ivermectin, which has been observed to reduce mosquito survival 28 days post-treatment in a study of outpatients at a hospital in Kenya [
15]. A recent study in Burkina Faso observed a 20% reduction in clinical incidence in an intervention group receiving ivermectin 6 times over 18 weeks at 3 week intervals, with young children, pregnant women, and women nursing newborns excluded from MDA eligibility [
16]. Another drug class, isoxazolines, shows promise to maintain mosquitocidal activity up to 50–90 days based on allometric scaling of preclinical pharmacokinetic data [
17], although safety concerns still need to be addressed [
18].
Previous mathematical modelling has suggested that high-dose ivermectin can improve the impact of anti-malarial MDAs [
15], a new long-lasting systemic insecticide with high efficacy lasting 30 days can substantially reduce burden when distributed with seasonal malaria chemoprevention (SMC) [
19], slow release formulations delivering a sustained therapeutic dose over 14 days can boost efficacy of mass drug administration with the goal of elimination [
12], and MDA with an isoxazoline lasting 90 days can also reduce burden in high-transmission areas [
17]. However, there has not been a systematic exploration of the potential impact of long-lasting systemic insecticides across multiple use cases within the same modelling framework. With the aim of describing the conditions under which safe and long-lasting systemic insecticides would be potentially beneficial and thus informing decision-making by funders of technology development, the ability of generic hypothetical systemic insecticides lasting 14 to 90 days to achieve the following goals is tested: additional burden reduction in the context of SMC, targeted elimination in the context of human travel to high-risk areas, and local elimination in the context of outdoor-biting vectors or insufficient coverage with traditional vector control interventions. Models predict that long-lasting systemic insecticides can complement other interventions to help reduce burden and achieve elimination. Systemic insecticides increase the impact of drug campaigns or alternatively can be used to achieve the same impact at lower, perhaps more feasible, campaign coverage. Unless systemic insecticides are extremely long-lasting, with mosquitocidal activity up to 90 days post dosage, improving their safety profile such that young children and women of childbearing age can safely participate in MDAs is essential.
Methods
Simulation framework
All simulations are performed with EMOD v2.15 [
20], an agent-based model of
Plasmodium falciparum malaria transmission that includes vector life cycle dynamics [
21], within-host parasite and immune dynamics calibrated to age- and season-stratified asexual and sexual stage parasite densities [
22], and drug pharmacokinetics and pharmacodynamics [
23].
Pharmacokinetics of systemic insecticides are modelled as constant efficacy of duration over 14, 30, 60, or 90 days, subsequently referred to as SI-14, SI-30, SI-60, and SI-90 respectively. Vectors feeding on a human with active systemic insecticide have 95% probability of death prior to the next feed, with 3 days between feeds. This corresponds to a hazard ratio of 4.4, which is in line with ivermectin [
19]. Systemic insecticides are assumed to have identical impact on all vector species and no impact of the systemic insecticide on parasite load. All models include surface area dependent biting, where children are less likely to be bitten by vectors due to their smaller size. This framework is able to capture effect sizes similar to those observed in the field with ivermectin that lasts around 3 days (Additional file
1) [
16].
No resistance to anti-malarials or to systemic insecticide is assumed. In each of the scenarios, coverage refers to the target demographic and not the entire population. For example, 50% coverage in a cohort excluding women of childbearing age and children under 5 would refer to 50% of males over 5 and females between the ages of 5 and 12 and over 51 years of age receiving the systemic insecticide.
Burden reduction scenarios
A simulation model of a well-mixed village is constructed based on northern Nigeria, with Sahelian seasonality and high-intensity transmission of annual entomological inoculation rate (EIR) 110 and mean annual all-age true prevalence of any infection of around 90% in the absence of any interventions. The vector population is modelled as Anopheles gambiae mosquitoes with 65% anthropophily and 90% indoor biting. Total human population is around 1000 individuals with birth and death rates of 45 per 1000 per year.
Drug campaigns are timed to start on June 28, just prior to the wet season, and the number of clinical cases averted is calculated for the year after the start of the campaign. Clinical cases are defined as malarial fevers of at least 38.5 °C occurring at least 14 days since the previous fever. SMC campaigns are carried out with dihydroartemisinin–piperaquine (DP), which has a 30-day period of prophylactic protection. SMC is distributed as 4 rounds separated by 1 month, with independent coverage between rounds, and targeted at children under 5 in standard SMC or children under 10 in expanded SMC. Systemic insecticide MDAs are given once, concurrently with the first round of SMC. Three systemic insecticide distribution scenarios are tested: where children under 5 and women of childbearing age (12–51 years) [
24] are ineligible to receive long-lasting systemic insecticides; where children under 5 are ineligible to receive long-lasting systemic insecticides but women of childbearing age are eligible; and where everyone is eligible to receive long-lasting systemic insecticides.
In all scenarios, the relative reduction in total clinical burden (number of clinical malaria cases in all age groups) is measured and compared with a standard SMC campaign in children under 5. Coverage of SMC and systemic insecticide MDA are assumed to be equivalent, although the denominators differ. Fifty stochastic realizations are run for each drug campaign and coverage combination. No other interventions are included in the simulations.
Targeted elimination scenarios
A simulation model of a transmission ecosystem of two villages that share a high-risk area is constructed (Fig.
3a) [
25]. Transmission is seasonal, peaking in December, and annual mean all-age prevalence of any infection is 29% in the absence of interventions. All villagers can travel between the villages, but only high-risk travellers visit the high-risk area. Each village is home to about 280 people and there is no importation from outside the modelled areas. Village vectors are
Anopheles minimus and vectors in the high-risk area are
Anopheles dirus, with 40% and 99% outdoor biting, respectively [
26]. Both vector species are modelled with 50% anthropophily and the systemic insecticides are assumed to have the same effect on both species.
High-risk travellers are 70% of people between ages 15 and 35. Migration happens between April and August with a mean stay duration of 30 days. During each month of high-risk travel, 50% of high-risk travellers who are currently at home in their village can make a trip to the high-risk area. Equal gender representation in the high-risk traveller population is assumed.
Treatment with artemether–lumefantrine (AL) is available in both villages for symptomatic cases. 80% of clinical malaria cases in children under 10 receive treatment, 70% of clinical cases in individuals over 10 receive treatment, and 95% of severe cases receive treatment. When treatment is sought, it is received within 3 days of symptom onset. High-risk travellers seek care at a lower (40%) and slower (within 5 days) rate. No treatment is available in the high-risk area and no other vector control is simulated.
All simulations last 3 years. The following drug combinations are considered: distribution of DP only, systemic insecticide only, and DP in combination with systemic insecticide. Children under 5 and women of childbearing age (12–51 years) are ineligible to receive systemic insecticides unless otherwise indicated but can still receive DP. MDA distributions occur only in year 1 and include three independent rounds separated by 30 days, beginning in June. MDA is never distributed in the high-risk area, and coverage refers to the fraction of eligible individuals residing in the village on the date of distribution who receive drugs. In scenarios where individuals receive drugs when they depart the village for the high-risk area, these trip-based drug distributions occur during all years. Coverage refers to the probability that drugs are taken for any given trip to the high-risk area by an eligible individual, and there are no restrictions on the number of times a person can take drugs.
Elimination is defined as zero infected individuals in the entire modelled area, including the high-risk area, and is assessed at the end of the third year. Each scenario is run for 100 stochastic realizations.
Southern Africa elimination scenarios
A simulation model of a 10,000-person population centre is constructed as 332 separate 1 km2 grid cells. About half of the population is located in a dense central area comprised of about a dozen grid cells, while the rest of the population is distributed more sparsely in the outlying areas. Individuals have a daily probability to take overnight trips to other grid cells, according to a gravity model of migration. The gravity model is parametrized to human movement on scales of one to tens of kilometres observed in geotagged campaign data from a rural site in Zambia (unpublished data). A distance exponent of 1.1 and a population exponent of 0.95 are used, with an overall amplitude such that individuals take an average of 5 overnight trips per year; the results are not sensitive to these choices. There is no disease importation from outside the modelled area.
Transmission intensity and seasonality reflect southern Africa, with two vector species:
Anopheles arabiensis (80% of vectors) and
Anopheles funestus (20% of vectors) [
27,
28]. Both vector species are assumed to have 65% anthropophily, with 50% indoor biting for
An. arabiensis and 90% for
An. funestus. All-age parasite prevalence by rapid diagnostic test (RDT) fluctuates between 20% and 40% prior to any vector control or MDA. All simulations begin with identical pre-intervention transmission intensity. When presenting with a clinical episode, 60% of children under 5 and 40% of individuals over 5 receive treatment. 80% of severe clinical episodes are treated for all age groups. Cases that receive care are treated with AL.
Intervention scenarios are simulated for 4 years (Fig.
4a). ITN distributions occur on September 1st of years 1 and 3. ITN killing efficacy starts at 60% and exponentially decays with a time constant of 4 years, which corresponds to a half-life of roughly 2.7 years. ITN blocking efficacy starts at 90% and exponentially decays with a time constant of 2 years, which corresponds to a half-life of roughly 1.4 years. Individuals aged 5 to 20 years old are 10% less likely to use a net [
29]. Modelled ITN usage is seasonal, with a maximum of 100% of ITN owners using their net on January 1st, reaching its minimum of 50% in June. To capture incomplete net retention, 60% of the population is modeled as discarding their nets at a rate with exponential time constant 260 days and the remaining 40% of individuals modeled as keeping their nets for much longer, discarding with an exponential decay time constant of 2160 days [
30,
31]. Coverage across ITN distributions is uncorrelated.
IRS spraying campaigns occur on September 1st of years 2 and 3. IRS is modeled as an organophosphate insecticide with 90% initial killing rate that lasts for 7 months then rapidly decays [
32] (and unpublished observations from Mara Maquina). Coverage across IRS campaigns is uncorrelated.
MDA campaigns are implemented in both February and November of years 2 and 3, a total of 4 distributions. A small scatter in grid cell-level campaign dates such that each full MDA distribution completes over the course of a few weeks is incorporated. The following drug combinations are tested: DP-only MDA, systemic insecticide-only MDA, and MDA with both DP and systemic insecticide. Coverage across multiple MDA rounds is uncorrelated.
No interventions other than ongoing case management are implemented in year 4. Elimination is defined as zero infected individuals in the entire simulated area at the end of year 4. Each intervention scenario is simulated with 100 stochastic realizations.
Discussion
In this work, the efficacy of systemic insecticides as a malaria control tool was explored in three different transmission settings: a Sahelian setting with high annual EIR where the goal is burden reduction, a near-elimination setting with a group of adults who regularly visit high-risk areas, and a southern Africa context focusing on elimination as the desired endpoint through vector control and drug campaigns. These diverse settings were selected to identify opportunities where systemic insecticide could provide high value to a pre-existing drug delivery programme, and to explore some of the very different contexts in which additional malaria control tools might help accelerate burden reduction or elimination.
Across the three modelled settings, the use case for including a systemic insecticide is strongest when the systemic insecticide duration is greater than about 30 days in a near-elimination setting, or greater than 60 days in a high-burden setting. Distributing systemic insecticide that lasts only 14 days had little to no impact in all of the modelled settings, unless it was distributed frequently or at the optimum time. In a near-elimination context with robust vector control, good campaign coverage, and, importantly, a safety profile sufficient to be given to small children and women of childbearing age, a 14-day systemic insecticide can be beneficial.
The models predict that systemic insecticides offer the greatest benefit when layered on top of other interventions rather than administered in isolation. In the southern Africa context, elimination is highly unlikely without a foundation of robust vector control, and systemic insecticides add the most benefit when administered alongside an integrated system of vector control and DP drug campaigns; here, the systemic insecticide can substantially reduce the MDA campaign coverage needed to achieve elimination. Lowering the MDA coverage necessary to achieve elimination is an important operational consideration, partly because mobile, high-risk groups are often difficult to reach. However, if these high-risk groups can be targeted directly, for example by treating adults in Southeast Asian settings who enter the deep forest for work with systemic insecticide and DP before they set out, the interventions have an outsize impact on transmission. When close to elimination, trying to target the sources of imported cases may be more cost-effective than continuing to apply interventions in population centres.
Children under the age of 5 and women of childbearing age comprise about 38% of the population in SE Asia and 44% of the population in Sub-Saharan Africa. A systemic insecticide that cannot be safely given to these large portions of the population is more limited in its potential impact, although the models have identified cases where such a restricted systemic insecticide may still be beneficial. To minimize the complexity and cost of MDA programmes, no pregnancy testing was assumed, although if pregnancy testing is an option, many more people would be able to take the systemic insecticide. If a systemic insecticide is unsafe to give to vulnerable subgroups, then to have the same efficacy, either the campaign coverage or the systemic insecticide duration must be increased. For example, in the high-transmission Sahelian context, an unrestricted systemic insecticide of 60-day duration has a similar impact on clinical burden as a restricted systemic insecticide of 90-day duration. In the targeted elimination scenario, systemic insecticide safety restrictions can more than double the systemic insecticide duration necessary to achieve elimination. Prior to approval for new long-lasting systemic insecticides in children and pregnant women, these groups could instead receive regular dose ivermectin [
35]. While the safety of ivermectin use in children under 15 kg and pregnant women is inconclusive, current onchocerciasis and lymphatic filariasis programmes have expanded the use of ivermectin and only exclude the severely ill, children under 15 kg, and visibly pregnant women [
36].
The pharmacokinetic and pharmacodynamic properties of systemic insecticides were simplified by assuming a constant killing efficacy over the course of the insecticide’s duration, allowing us to isolate the impact of insecticide duration on epidemiological outcomes; however, this assumption could slightly underestimate the impact of systemic insecticides. The implications for more complex within-host insecticide dynamics can be interpolated from the results presented here. For example, a 300 μg/kg ivermectin dose can be compared with the modelled hypothetical SI-14 [
15]. The model also does not incorporate possible non-fatal vector outcomes of the systemic insecticide such as reduced fecundity, nor did were any parasite killing effects modeled since these outcomes are not well understood and the effects are likely secondary to killing the mosquito [
37].
The scenarios presented in this work do not directly incorporate effects of drug or insecticide resistance, the exception being in assuming a low killing efficacy of ITNs in the Southern Africa scenario due to pyrethroid resistance, which is widespread [
38]. In settings where resistance is present, systemic insecticides of a shorter duration than 30 or 60 days could still be an important operational tool. Against the backdrop of growing insecticide resistance, systemic insecticides have added utility since they operate by a novel mechanism not shared by current ITN and IRS insecticide compounds, and there is potential for synergy between the insecticides’ killing mechanisms. Against drug resistance, systemic insecticides layered on top of an anti-malarial could also be beneficial, since even if parasites have some protection against the anti-malarial, the systemic insecticide may be able to kill most of the vectors carrying this resistant strain. If systemic insecticide replaces DP entirely for mass administrations, selection pressure could be partially transferred from the parasite onto the vector, making further development of anti-malarial drug resistance less likely.
Anthropophily, the fraction of bites that are taken on a human host, is a critical vector bionomic parameter that is often poorly understood in the local vector context. For a fixed mosquito population size, lowering anthropophily lowers the number of bites on humans and thus reduces the chance of a mosquito transmitting malaria. However, a lower anthropophily also makes the mosquito population more resilient to the impact of vector control methods such as ITN, IRS, or systemic insecticide where vectors encounter vector control while seeking human hosts. The models predict that longer-lasting systemic insecticides are less sensitive to uncertainty in anthropophily than DP or systemic insecticides of duration 30 days or less, and that probability of elimination decreases as anthropophily increases regardless of the drug deployed (Additional file
3). Distribution of systemic insecticides to livestock was not considered, although this strategy may be fruitful where vectors are fairly zoophilic and humans and livestock live in close proximity [
39].
The WHO’s preferred product characteristics for endectocides recommends at least 20% reduction in clinical incidence for one month following administration, and does not make recommendations for product characteristics for elimination purposes [
40]. The simulations predict that even SI-14 is likely to meet this target if coverage is at least 50% (Fig.
2d). If distribution of SI-14 is restricted for safety concerns, then the minimum coverage will be higher; however, systemic insecticides with duration at least 30 days are very likely to meet the WHO target even if vulnerable populations are ineligible for administration. Previous work on developing a target product profile for endectocides recommended maintaining hazard ratio above 4 for 30 days post-dose [
19], analogous to the modeled SI-30. This results from this work agree that SI-30 would be highly beneficial if administered with all SMC rounds, and furthermore would be effective in an elimination context when targeted to high-risk travellers.
While this work finds that long-lasting systemic insecticides can yield large benefits in terms of burden reduction or hitting elimination targets, given a good safety profile, co-administration with anti-malarials, and a foundation of robust traditional vector control, the operational challenges and costs of administering MDA should not be ignored when considering this intervention. Other new tools for control of outdoor-biting vectors are also under development, and the cost-effectiveness of systemic insecticides should be compared with other available options.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.