Background
Chemical insecticides are crucial tools in malaria control and elimination programmes. Introduced on walls inside houses as indoor residual sprays (IRS), or on/into the fibers of long-lasting insecticide-treated nets (LLINs), their main purpose is to reduce the risk of human malaria infections by killing the mosquito vectors [
1]. Based on World Health Organization (WHO) estimates, these tools have been extremely successful, their scale-up contributing to avert 1.1 million deaths between 2000 and 2012 [
2]. However, they also impose a strong selection pressure on the mosquito population for insecticide resistance; there is currently no malaria-affected country in Africa that is free of insecticide-resistant mosquitoes [
3]. The malaria community fears that this resistance will undermine effective mosquito control and elimination strategies, and erode the progress that has been made in reducing the number of malaria deaths to-date [
2‐
5].
It is not clear what level of resistance, as defined by current methods and guidelines, precipitates control failure in terms of human malaria cases in an area [
6]. For example, a recent systematic review and meta-analysis of the relationship between resistance and control concluded that ITNs remained more effective than untreated nets against resistant mosquitoes, in terms of entomological metrics such as mosquito mortality and blood feeding success [
6], though they found only one study that directly compared resistant and susceptible insects [
7]. Even with the uncertainty underlying the association, resistance monitoring remains a recommended priority of vector control programmes, so that action can be taken before failure occurs [
8,
9].
The National Malaria Control Programme (NMCP) of Mozambique, in collaboration with the WHO, Manhiça Foundation, the President’s Malaria Initiative, USAID and other partners, is currently designing and scaling up activities to eliminate malaria from southern Mozambique by 2020. Insecticidal vector control measures are an important component of the plans for the future, as they have been in the past and are today. In the district of Manhiça, for example, which experiences moderate levels of malaria transmission perennially [
10], IRS has been deployed annually since 2005. Mainly DDT and bendiocarb have been provided to the district, although lambda-cyhalothrin (2005 and possibly 2010) and deltamethrin (2009 and 2014) have also been used in IRS campaigns (Additional file
1). LLINs have been handed out to pregnant women at the local antenatal clinics as part of research studies before or, more recently, as national policy for pregnant women. In June–July 2014, a campaign ran in the Manhiça district to achieve universal LLIN coverage, defined as one net per 1.8 persons, and over 32,000 LLINs have been distributed.
Though insecticide use has been widespread, resistance monitoring of the local
Anopheles funestus population has not been conducted since 2009. An earlier study, in 2002 [
11], showed no resistance to permethrin or deltamethrin at the time. In 2009, however, resistance to deltamethrin (52.6% mosquito mortality) and lambda-cyhalothrin (33.3%) was observed [
12]. The intensified vector control underway in Manhiça and the surrounding districts will increase insecticidal selection pressure on local mosquito populations to develop insecticide resistance, if no resistance management strategy is in place. To design such strategies, continuous mosquito resistance monitoring and regular evaluations of frontline vector control tool efficacy against local vectors are crucial. To outline the current status of the
An. funestus population in Manhiça, this work presents (1) the March 2014 levels of resistance to three important different pyrethroids and (2) the impact of resistance on the efficacy of the two common LLIN types found in the district.
Methods
Mosquito collections
During February 2014, a total of 778 blood-fed anopheline mosquitoes were collected over a 3-week period from houses in the Ribangua area of Manhiça, close to the irrigated sugar cane fields of Maragra (Illovo) Sugar Mill. IRS teams from the NMCP visit this area annually (for history of IRS in Manhiça, see Additional file
1), but spraying had not yet been done. Females were caught indoors, using mouth aspirator and torch, from 6 am to 12 pm. Before a home was entered, the study was explained in the local language, any questions were answered, and the head of each household gave informed oral consent.
All field-collected females were given the opportunity to oviposit. Because live adults cannot be easily identified to species, a subset was examined by microscopy [
13] after eggs were laid.
Anopheles funestus
s.l. dominated the collection, with a few
Anopheles gambiae s.l. individuals. The species were easily separated as larvae, so the F
1 generation of each species could be used for bioassays. Larvae were reared to adults at ambient room temperature on a diet of TetraminBaby fish food (Tetra, Germany).
Resistance bioassays
Two-to-five day-old non-blood-fed female
An. funestus s.l. mosquitoes were exposed to 0.05% deltamethrin (2 exposures, each with 4 replicates), 0.05% lambda-cyhalothrin (1 exposure with 4 replicates), 0.75% permethrin (2 exposures, each with 4 replicates) for 1 h using WHO-standard insecticide-treated papers, exposure tubes, control papers (silicon oil) and exposure procedures [
14]. One-to-eight day-old non-blood-fed female
An. gambiae s.l. were exposed to 0.05% deltamethrin (two exposures, each with four replicates) in the same way (a larger age range was required to get a sufficient number of individuals). Sample sizes are listed in Table
1. After the exposure, mosquitoes had ad libitum access to sugar water. Mortality was scored 24 h post-exposure. All mosquitoes used in bioassays were identified to species by microscopy [
13] and stored individually on silica gel.
In order to maintain constant environmental conditions during and after exposure, mosquitoes were kept in a polystyrene box [inner dimensions 31.5 × 35 × 22(h) cm] outfitted with heat cable (PT2012 Heat Cable, ExoTerra, USA) and a humidifier (RF-10E ReptiFogger, ZooMed, USA). Temperature and relative humidity (RH) were held at 26 ± 1°C and 90 ± 10% RH by a thermostat/humidistat (HT-10E Hygrotherm, ZooMed, USA) and were recorded at 5 min intervals by an automated logger (OM-62, Omega, Spain).
To authenticate the quality of the insecticide-treated papers, two-to-five day-old laboratory-reared susceptible Anopheles arabiensis females were exposed in a similar fashion. These experiments were performed at the insectaries of the National Institute of Health in Mozambique, at 27 ± 2°C and 80 ± 10% RH.
Long-lasting insecticide-treated net tests
The efficacy of two previously unopened Olyset (permethrin LLIN, manufactured 11/2010, obtained from the Manhiça Health Research Centre) and two previously unopened PermaNet 2.0 bed nets (deltamethrin LLIN, 05/2011, purchased in local store) was evaluated using two-to-five day-old, non-blood-fed, female
An. funestus s.l. in standard WHO cone bioassays [
9]. Groups of five mosquitoes were exposed for 3 min to pieces of netting (four pieces cut from sides of each net tested; Olyset: n = 97 mosquitoes, PermaNet 2.0: n = 96). A locally purchased untreated bed net (China Da Hua Shoes, LDA Moçambique), also previously unopened, served as control netting (n = 49). Mosquitoes were exposed to nets at 26 ± 2°C and 60 ± 10% RH, and then transferred to cups and maintained in the polystyrene box described above at 26 ± 1°C and 90 ± 10% RH for 24 h, after which mortality was recorded. To confirm the insecticidal activity of the LLINs, wild-caught blood-fed
Anopheles tenebrosus females (abundant in our study area) were tested in a similar fashion. All mosquitoes used in the bioassays were identified to species by microscopy [
13] and stored individually on silica gel.
In order to evaluate the level of resistance to LLINs, mosquitoes were also exposed to LLINs for longer durations than the 3 min advised in the WHO protocol (Experiment 1: 3 min, 1, 2, 4, 6 h; experiment 2: 3 min, 2, 4, 6, 8 h). In each experiment, for each time treatment, two replicates of five mosquitoes were exposed to untreated netting and four replicates of five mosquitoes to LLIN. Only the two PermaNet 2.0 LLINs were tested in time-exposure experiments, as mosquitoes could be caught beneath the larger mesh openings of the Olyset nets, making mortality rates for longer durations of exposure unreliable.
Mosquito identification
Forty-eight
An. gambiae s.l. and two hundred
An. funestus
s.l. F
1 females were identified to species by PCR using the methods of Scott et al. [
15] and Koekemoer et al. [
16], respectively. Extraction of DNA was done for the
An. funestus group, and a single leg was used directly as template for the
An. gambiae complex. Individuals identified as
An. arabiensis were screened for the presence of East- and West-knockdown resistance (
kdr) alleles using the methods of Bass et al. [
17].
Data analysis
‘Resistance’ to a given chemical or net was designated according to the current WHO criteria, based on the percent mortality observed 24 h after insecticide exposure: mosquito populations are classified as resistant if more than 2% of the exposed individuals survive [
18]. Following the classifications proffered by Strode et al. [
6], resistance to a given pyrethroid could be described as ‘low,’ if exposure led to mortality greater than 80%, ‘moderate’ if mortality was between 25 and 80%, and ‘high’ if less than 25% of females were killed by exposure.
Statistical analyses were performed in R v. 2.10.1 [
19]. Mortality data were analysed as Generalized Linear Models using a binomial error distribution and logit link function; in cases of overdispersion, the quasibinomial error distribution was used. For tube bioassay analyses, insecticide exposure (control or chemical) was the only independent variable. For the 3-min cone bioassay analyses, net (control, Olyset, or PermaNet 2.0) was the only independent variable. When investigating the strength of resistance to net exposure, duration of exposure (Experiment 1: 3 min, 1, 2, 4, 6 h; experiment 2: 3 min, 2, 4, 6, 8 h) was included as an independent variable as well as net (control or PermaNet 2.0). Pieces from two different PermaNets were used and the results from the two experiments were pooled, so net replicate and experiment were also included in the model as random effects. Maximal models were fitted with random effects and interaction terms first, and then non-significant terms were removed by backward-elimination.
Discussion
In the last 5 years, pyrethroid resistance in the
An. funestus population of Manhiça has increased to extremely high levels, and this area in southern Mozambique is clearly a pyrethroid-resistance hotspot (Fig.
1; [
12]). Over 90% of mosquitoes survived a deltamethrin or lambda-cyhalothrin exposure, and almost three-quarters of mosquitoes survived an exposure to permethrin (Table
1). Mosquitoes were equally as likely to die from 3-min exposure to a net that was not treated with insecticide as from exposure to a pyrethroid-treated Olyset (permethrin) or PermaNet 2.0 (deltamethrin) (Fig.
2). In addition, the deltamethrin LLIN was unable to kill more than half of the mosquitoes in 6 h. Given that mosquitoes are unlikely to rest on a bed net, undisturbed, for 6 h, these results indicate that the current pyrethroid-based vector control tools are unlikely to effectively kill the major malaria vector in this part of Mozambique.
Due to limited numbers, the local An. gambiae s.l. was only screened for resistance to deltamethrin. These mosquitoes, a mix of ninety percent An. arabiensis and ten percent An. merus, were resistant, with ten percent surviving deltamethrin exposure (note that this ten percent includes both species). The level of resistance appears to be much lower than in An. funestus, in which around ninety percent of mosquitoes survived. This difference is interesting, given that the field-caught females that generated these experimental populations were captured in ostensibly identical locations, indicating some overlap in their behaviour and ecology and, therefore, some shared selection pressures. For resistance management strategies to have any hope of being successful, it is critical that the major selection pressures at work in a population be identified. In a transmission setting with multiple vector species, differences in the ecology and the biology of those species (such as differences in biting time, resting habits, or larval habitats) could prove to be important.
Even in the face of insecticide resistance, LLINs will continue to play an important role in malaria control programmes. Well-maintained LLINs still provide a physical barrier between mosquitoes and human hosts and can prevent bites regardless of insecticidal activity [
20]. When pyrethroids are no longer as deadly to resistant vectors, they can still act as repellents and/or irritants [
6,
21]. These effects could interfere with disease transmission even when no mosquitoes are killed by an insecticide (described in [
1]). However, to increase the impact of vector control strategies, alternative LLINs, such as the Olyset Plus and PermaNet
® 3.0 should be tested against the local
An. funestus populations. In addition to a pyrethroid, both of these nets include the pyrethroid-synergist piperonyl-butoxide (PBO), which inhibits the activity of metabolic detoxification enzymes [
22]. These nets have demonstrated efficacy against pyrethroid-resistant mosquitoes [
23,
24], and two studies indicate that such LLINs may be effective in the Manhiça area [
12,
25].
While these alternative LLINs could improve the mosquito control generated by bed nets, they still can only be treated with pyrethroid insecticides [
26,
27] and, thus, will not remove the selection pressure for pyrethroid resistance. Control of the highly-pyrethroid-resistant mosquitoes in Manhiça, therefore, will rely on the use of non-pyrethroid chemicals for IRS. Unfortunately, high levels of resistance to the carbamate bendiocarb ([
12] and Additional file
2) have also been detected in the area, which limits our chemical arsenal to DDT ([
12] and Additional file
2) or pirimiphos-methyl (Additional file
1), and perhaps malathion [
12], but additional, systematic evaluations are needed.
Together, the high levels of pyrethroid resistance detected in standard bioassays and the duration of LLIN exposure necessary to kill
An. funestus from Manhiça seen here indicate that vector control plans for southern Mozambique will have to carefully integrate multiple control tools in order to achieve effective malaria control in this pyrethroid resistance ‘hot spot’ [
5]. The NMCP of Mozambique is aware of the pressing issue of pyrethroid resistance, is currently improving its resistance-monitoring programme and is drafting a resistance management strategy to counteract the rising problem of insecticide resistance. As mosquito control, resistance-monitoring and –management programmes proceed in this area, it will be important to consistently evaluate their impact, in order to maintain effective control and provide useful information to the other countries pushing toward malaria elimination in the coming decades.
Authors’ contributions
KDG and KPP conceived and planned the experiments, collected and reared the mosquitoes and performed the bioassays. KDG, KPP and APA performed the bioassays with the susceptible mosquito colony. AEG performed the molecular analysis. KDG and SH analysed the data. QB, HB, MNM, EM and PA provided institutional support and guidance in logistics. KDG and KPP wrote the paper. All authors read and approved the final manuscript.