Experimental hut trial
Experimental hut site
Experimental huts are small standardized human habitations approved by the WHO for the controlled evaluation of indoor vector control tools against wild free-flying mosquitoes. Mosquitoes enter the huts freely at night to interact with the human host and the vector control intervention and in the morning on each day of the trial, they are collected from the different compartments of each hut and scored for entomological outcomes. The experimental hut study was performed at the CREC/LSHTM experimental hut station situated in a large rice growing area in Cové, Southern Benin, where the local mosquito population has been shown to be resistant to pyrethroids [20]. The rice paddies provide extensive breeding sites for Anopheles gambiae throughout the year. The huts are built on concrete plinths surrounded by water-filled moats to prevent entry of scavenging ants and have veranda traps to capture the exiting mosquitoes. The walls are made of brick plastered with cement on the inside, with a corrugated iron roof. The huts have a ceiling of palm thatch and four window slits (1 cm gap) on the walls through which mosquitoes enter. The local vector population in Cove is resistant to pyrethroids and DDT and consists of a mixture of Anopheles coluzzii and Anopheles gambiae sensu stricto (s.s.), with the latter occurring at lower proportions (23%) and only in the dry season [20]. Molecular analysis revealed a L1014F kdr allele frequency of 89%. Microarray studies also found CYP6P3, a P450 validated as an efficient metabolizer of pyrethroids [21], to be overexpressed in Cove [20].
Insecticide resistance bioassays
To assess the frequency of pyrethroid resistance and presence of mixed function oxidases in the Cové vector population during the trial, adult mosquitoes that emerged from larvae collected from breeding sites close to experimental huts were tested in WHO cylinder bioassays with and without pre-exposure to PBO. A total of ~ 100 mosquitoes of the pyrethroid resistant An. gambiae s.l. Cove strain and the pyrethroid susceptible An. gambiae Kisumu strain were exposed to treated filter papers in WHO cylinder bioassays in batches of 25. Tests were performed with papers treated with permethrin 0.75%, alpha-cypermethrin 0.05% and deltamethrin 0.05%. To assess presence of MFO, some mosquitoes were also pre-exposed to papers treated with 4% PBO prior to exposure to insecticide-treated papers. Exposure to PBO and to insecticides lasted 1 h, knockdown was recorded after 60 min and mortality after 24 h.
Experimental hut treatments
Olyset Plus and PermaNet 3.0 were compared in the experimental huts when unwashed and after 20 standardized washes. A WHO-recommended pyrethroid-only long-lasting net (Olyset Net) was included to demonstrate the added effect of PBO on the insecticide-resistant local vector species. Nets were washed using savon de Marseilles and rinsed twice following WHO procedures for washing nets for experimental hut studies [22].
The following seven (7) treatments were thus tested in seven experimental huts:
-
1.
Untreated polyethylene net
-
2.
Olyset Net unwashed (permethrin only)
-
3.
Olyset Net washed 20 times.
-
4.
PermaNet 3.0 unwashed (Roof: deltamethrin plus PBO; sides: deltamethrin only)
-
5.
PermaNet 3.0 washed 20 times.
-
6.
Olyset Plus unwashed (permethrin plus PBO on all panels)
-
7.
Olyset Plus washed 20 times.
Hut trial procedure
Treatments were allocated to the experimental huts on a weekly basis using a randomized Latin square design to adjust for any variation in hut attractiveness and minimize any carry over effect between treatments. Three replicate nets of each type were prepared, and these were rotated every 2 days on each week (6 days) of the trial. To simulate wear and tear, each net was intentionally holed with six 16 cm2 holes (two holes on each side and one on each end).
The trial ran for 42 nights between February and April of 2017. Consenting human volunteer sleepers slept in the huts from 9:00 p.m. to 5:00 a.m. each night and were rotated daily through the huts to account for individual attractiveness to mosquitoes. At dawn, the volunteer sleepers collected mosquitoes in the room of the hut and under the bed nets and the veranda using torches and aspirators. The mosquitoes were then transferred to the laboratory for processing where they were identified and scored for their blood feeding status, mortality and hut position. Mosquitoes were held at 27 ± 2 °C during the observations.
The following outcome measures were used to assess the efficacy of each treatment in the experimental huts:
-
1.
Deterrence—the proportional reduction in number of mosquitoes entering huts with treated nets.
-
2.
Exiting rates estimated from the proportions of mosquitoes collected from the verandas of treatment and control huts.
-
3.
Mortality—the proportion of mosquitoes killed (immediate plus delayed) relative to the total collected.
-
4.
Blood-feeding—the proportion of blood-fed mosquitoes relative to the total collected.
-
5.
Blood-feeding inhibition—the proportional reduction in blood feeding in huts with insecticide treated nets relative to controls with untreated nets.
-
6.
Personal protection—the proportional reduction in mosquito biting by insecticide treated nets relative to untreated nets.
Supplementary laboratory bioassays
To help further explain the results obtained in the experimental huts, WHO cone bioassays and tunnel tests were performed on samples of netting (30 × 30 cm) obtained from Olyset Net and Olyset Plus when unwashed and after 10 and 20 washes. Washing was performed in the laboratory following WHO guidelines [22]. PermaNet 3.0 was not tested in the laboratory bioassays owing to the restricted application of PBO to the roof of the net preventing a realistic direct comparison with Olyset Plus in bioassays especially tunnel tests. Net samples from each ITN type and each wash point were tested against the following strains:
-
1.
An. gambiae sensu lato (s.l.) strains from Cove, Benin (Cove strain) which is highly pyrethroid resistant. It originates from the experimental hut station in Cove and has shown > 200-fold resistance compared to the susceptible Kisumu strain in susceptibility bioassays. Resistance is mediated by elevated levels of P450s and high frequencies of kdr [20].
-
2.
An. gambiae VKPer strain, which originated from the Kou Valley in Burkina Faso. VKPer has moderate levels of pyrethroid resistance mediated only by high frequencies of kdr.
-
3.
An. gambiae s.s. Kisumu strain, a reference susceptible strain which originated from Kisumu Kenya.
Approximately two hundred 2–5 days old mosquitoes of each strain were exposed for 3 min in cone bioassays to four net samples of each net type in cohorts of 5 mosquitoes per cone. Knock down in cone bioassays was recorded after 1 h and mortality after 24 h.
Two to three hundred 5–8 days old mosquitoes of each strain were also exposed to each net type in tunnel tests in replicates of 50 mosquitoes per net sample. The tunnel test is a laboratory assay designed to simulate natural host-seeking behaviour of mosquitoes at night in the presence of a net. It consists of a square glass cylinder (25 cm high, 25 cm wide, 60 cm in length) divided into two sections by means of a netting frame fitted into a slot across the tunnel. An anesthetized guinea pig was housed unconstrained in a small cage in one section, and mosquitoes were released in the other section at dusk and left overnight. The net samples were holed with nine 1-cm diameter holes to allow host-seeking mosquitoes to penetrate the baited chamber; an untreated net sample served as the control. The tunnels were kept overnight in a dark room at 25–29 °C and 75–85% RH. The next morning, the numbers found alive or dead, fed, or unfed, in each section were recorded. Live mosquitoes were provided with sugar solution and delayed mortality recorded after 24 h. The guinea pigs used in this study were kept in accordance with institutional guidelines for animal care.
Chemical analysis
At the end of the experimental hut trial, five pieces of netting (25 × 25 cm) obtained from the panels of replicate nets of each net type (before and after washing) used in the huts were assessed for deltamethrin, permethrin and PBO content using HPLC. Insecticide was extracted from each net piece with an area of 48 sq cm collected from the five net samples (25 × 25 cm) obtained from each whole net. The insecticide content of each sample was determined by injecting ten μl aliquots of the extract on a reverse-phase Hypersil GOLD C18 column (75 Å, 250 × 4.6 mm, 5-μm particle size; Thermo Scientific) at room temperature. A mobile phase of 70% acetonitrile in water was used at a flow rate of 1 ml min−1 to separate the target analyte. Chromatographic peaks of the insecticides and internal standard were detected at a wavelength of 232 nm with the Ultimate 3000 UV detector and analysed with Dionex Chromeleon™ 6.8 Chromatography Data System software. Quantities of insecticide were calculated from standard curves established by known concentrations of the insecticide authenticated standards and corrected by internal standard readings in each sample relative to control.
Data from chemical analysis was used to calculate the percentage retention of each active ingredient after 20 washes relative to the unwashed net and the wash retention index. Wash-resistance index was calculated according to WHO guidelines [22] as indicated below:
$${\text{Wash}}\;{\text{resistance}}\;{\text{index}} = 100 \times {\text{n}}\sqrt {\left( {{\text{t}}{\mathbf{n}}/{\text{t}}{\mathbf{0}}} \right)} \;\left( {{\text{free}}\;{\text{migration}}\;{\text{stage}}\;{\text{behaviour}}} \right)$$
where tn = total active ingredient content after n washing cycles, t0 = total active ingredient content before washing, n = number of washes.
Data analysis
Proportional outcomes (blood-feeding, exiting and mortality) related to each experimental hut treatment (unwashed and washed 20 times) were assessed using binomial generalized linear mixed models (GLMMs) with a logit link function, fitted using the ‘lme4’ package for R (version 3.5.3). A separate model was fitted for each outcome. In addition to the fixed effect of each treatment, each model included random effects to account for the following sources of variation: between the huts; between the sleepers; between the weeks of the trial; and finally, an observation-level random effect to account for variation not explained by the other terms in the model (over dispersion).
Ethical considerations
This study received ethical approval from the Ministry of Health in Benin and from the Ethics Review Committee of the London School of Hygiene & Tropical Medicine. Informed consent was obtained from each human volunteer sleeper who slept in the huts to attract mosquitoes prior to their participation. Sleepers were also offered chemoprophylaxis. Through the course of the study, they were examined regularly for signs of fever by a stand-by nurse; any sleepers testing positive for malaria were withdrawn from the study and treated properly.