Experimental hut trials
To assess the ability of the 20-wash method to simulate the end-of-life performance of ITNs, we performed an experimental hut trial comparing the entomological efficacy of ITNs washed 20 times to field-aged ITNs withdrawn after three years of operational use against wild, free-flying malaria vectors.
Study site and vector population
The experimental hut trials were performed at Centre de Recherche Entomologique de Cotonou and London School of Hygiene & Tropical Medicine (CREC/LSHTM) field site in the commune of Covè, Zou Department, southern Benin (7°14′N2°18′E). The huts are surrounded by rice paddies, which provide permanent and extensive breeding sites supporting a high year-round density of malaria vectors. An. coluzzii and An. gambiae s.s. occur sympatrically with the former predominating (21). Recent studies characterising the insecticide resistance profile of the Covè vector population using susceptibility bioassays have demonstrated a high frequency and intensity of pyrethroid resistance but continued susceptibility to other insecticides, including CFP and PPF (22, 23). Pre-exposure to PBO improves the mortality response to pyrethroids without restoring full susceptibility, demonstrating the partial contribution of cytochrome P450 monooxygenases (P450s) to pyrethroid resistance (23). This is corroborated by genotyping and gene expression studies, which reveal a high frequency of the knockdown resistance (kdr) mutation (89%) and four-fold overexpression of P450s, including CYP6P3 (21) – a validated marker for metabolic pyrethroid resistance (24). The experimental huts used were of standard West African design made of concrete bricks with cement-plastered walls enclosed with a corrugated iron roof and polyethene ceiling. Mosquitoes entered via four window slits on the front and side walls of the hut. Each hut had a wooden-framed veranda projecting from the rear wall to capture exiting mosquitoes and was surrounded by a water-filled moat to prevent entry of scavenging ants.
Experimental hut treatments
An untreated control net and four WHO-prequalified ITN products included under current WHO policy recommendations were used for the study. The specifications of the study ITNs are described below:
i. Interceptor® (BASF) is a standard pyrethroid-only ITN made of 100-denier polyester filaments coated with 5 g/kg (200 mg/m2) of alpha-cypermethrin. Its dimensions are 1.8 (L) x 1.8 (W) x 1.8 (H) m. The fabric weighs 40 g/m2 and has a minimum bursting strength of 405 kPa.
ii. PermaNet® 3.0 (Vestergaard Sàrl) is a pyrethroid-PBO ITN. It features a mosaic design consisting of 100-denier polyester side panels coated with 2.1 g/kg (84 mg/m2) of deltamethrin and a 100-denier polyethylene roof panel incorporated with a mixture of deltamethrin and PBO at 4 g/kg (120 mg/m2) and 25 g/kg (800 mg/m2) respectively. Its dimensions are 1.6 (L) x 1.8 (W) x 1.5 (H) m. The fabric weight and minimum bursting strength are 40 g/m2 and 350 kPa on the sides compared to 30 g/m2 and 400 kPa on the roof.
iii. Royal Guard® (Disease Control Technologies) is a pyrethroid-PPF ITN made of 120-denier high-density polyethylene filaments incorporated with a mixture of alpha-cypermethrin and PPF at target doses of 5.5 g/kg (208 mg/m2) each. Its dimensions are 1.8 (L) x 1.8 (W) x 1.6 (H) m, while its fabric weighs 38 g/m2 and has a minimum bursting strength of 350 kPa.
iv. Interceptor® G2 (BASF) is a pyrethroid-CFP ITN made of 100-denier polyester filaments coated with a mixture of alpha-cypermethrin and CFP at target doses of 2.4 g/kg (100 mg/m2) and 4.8 g/kg (200 mg/m2) respectively. Its dimensions are 1.8 (L) x 1.8 (W) x 1.8 (H) m, while its fabric weighs 40 g/m2 and has a minimum bursting strength of 405 kPa.
The experimental hut trial compared the entomological efficacy of new ITNs unwashed and after 20 standardised washes to field-aged ITNs withdrawn from households after three years of operational use. A total of 6 new unwashed and washed replicate nets and 78 field-aged replicate nets were tested and rotated within the treatments daily. A larger sample of field-aged nets was used to help control for variation in net care and use practices between users, which may affect their entomological efficacy and bias experimental hut results. The following 13 treatment arms were thus evaluated in experimental huts:
1. Untreated net (control) – 6 replicate nets
2. Interceptor® (new unwashed) – 6 replicate nets
3. Interceptor® (washed 20 times) – 6 replicate nets
4. Interceptor® (field-aged three years) – 78 replicate nets
5. PermaNet® 3.0 (new unwashed) – 6 replicates
6. PermaNet® 3.0 (washed 20 times) – 6 replicates
7. PermaNet® 3.0 (field-aged three years) – 78 replicates
8. Royal Guard® (new unwashed) – 6 replicates
9. Royal Guard® (washed 20 times) – 6 replicates
10. Royal Guard® (field-aged three years) – 78 replicates
11. Interceptor® G2 (new unwashed) – 6 replicates
12. Interceptor® G2 (washed 20 times) – 6 replicates
13. Interceptor® G2 (field-aged three years) – 78 replicates
Preparation of study nets
New nets were tested unwashed and washed 20 times to simulate end-of-life performance. Nets were washed according to WHO guidelines (18). The nets were submerged in an aluminium bowl containing 2 g of Savon de Marseille dissolved in 10 litres of water and washed for 10 mins. After washing, nets were rinsed twice in 10 litres of clean water following the same procedure and dried horizontally in the shade. The time interval applied between washes for each ITN product was selected based on the regeneration times outlined in WHO assessment reports as follows: 1 day for Interceptor®, PermaNet® 3.0 and Interceptor® G2, and three days for Royal Guard® (25–28). To simulate wear and tear from regular use, the untreated control nets and new unwashed and washed ITNs were given six holes measuring 4 x 4 cm – one on each short side panel and two on each long side panel. Field-aged nets withdrawn after three years were obtained from ongoing durability monitoring studies nested within a cluster-RCT designed to evaluate the epidemiological efficacy of next-generation ITNs compared to standard pyrethroid-only nets in Benin. The design of the durability monitoring study has been described previously (29). To support the interpretation of the experimental hut results, the hole index of field-aged nets was also calculated by counting and classifying the number of holes according to their location and size as per WHO guidelines (18). Nets were erected inside huts by tying the four corners of the roof panel to nails positioned in the top corners of the hut.
Experimental hut trial procedure
Consenting human volunteers slept in experimental huts from 21:00 to 06:00 to attract wild, free-flying mosquitoes. Treatments were rotated between huts weekly, while volunteers were rotated daily according to Latin square designs to mitigate variation due to differences in positional and host attractiveness to mosquitoes. Each morning, volunteers collected all mosquitoes from the different compartments of the hut (under the net, room and veranda) and deposited them in labelled holding cups using a torch and aspirator. Mosquito collections were transferred to the field laboratory for morphological identification and scoring of immediate mortality (live/dead) and blood-feeding (unfed/blood-fed). Live female An. gambiae s.l. were retained and provided access to cotton wool soaked in 10% (w/v) glucose solution. Delayed mortality was recorded every 24 h up to 72 h after collection, and % 72 h mortality was used as the primary mortality endpoint for all treatments to account for the delayed action of CFP (30). To evaluate the impact of Royal Guard® on fecundity, subsamples of surviving blood-fed An. gambiae s.l. were dissected to observe ovary development and score fertility as previously described. Mosquito collections were performed six days per week for one full treatment rotation (13 weeks, 78 nights) during the long rainy season between May and July 2023. Huts were cleaned and aired on the 7th day to prevent contamination before the next rotation cycle.
Experimental hut trial outcome measures
For each treatment, the total number of alive/dead, unfed/blood-fed, fertile/infertile mosquitoes collected in the different hut compartments were pooled to generate a variety of outcome measures (listed below) of entomological efficacy. The primary outcome measures used to compare the performance of ITNs against pyrethroid-resistant An. gambiae s.l. after artificial and operational ageing were:
i. Mortality (%) – the proportion of dead mosquitoes immediately after collection and every 24 h up to 72 h thereafter.
ii. Blood-feeding inhibition (%) – the reduction in the proportion of blood-fed mosquitoes in the treated hut relative to the untreated control hut. Calculated as follows:
$$\:Blood\:feeding\:inhibition\:\left(\%\right)=\frac{100\:\left(Bfu-Bft\right)}{Bfu}$$
ii. Reduction in fertility (%) – the reduction in the proportion of dissected mosquitoes scored as fertile for a given treatment compared to the control. Calculated as follows:
Where Fu is the proportion of fertile mosquitoes in the untreated control hut, and Ft is the proportion of fertile mosquitoes in the treated hut.
The secondary outcome measures used to express the efficacy of the experimental hut treatments against pyrethroid-resistant An. gambiae s.l. were:
i. Entry ( n ) – the number of mosquitoes collected inside the hut.
ii. Deterrence (%) – the reduction in entry in the treated hut relative to the untreated control hut. Calculated as follows:
$$\:Deterrence\:\left(\%\right)=\:\frac{100\:(Tu-Tt)}{Tu}$$
Where Tu is the number of mosquitoes entering the untreated control hut, and Tt is the number of mosquitoes entering the treated hut.
iii. Exophily (%) – exiting rates due to the potential irritant effect of treatments expressed as the proportion of mosquitoes collected in the veranda.
iv. Blood-feeding (%) – the proportion of blood-fed mosquitoes.
v. Fertility (%) – the proportion of dissected mosquitoes scored fertile.
Preparation of net pieces for supplementary bioassays and chemical analysis
Five net pieces (one from each panel) measuring 30 x 30 cm were cut from randomly selected nets per treatment arm at positions outlined in WHO guidelines (18). Given its mosaic design, an additional two net pieces were cut from the roof of PermaNet® 3.0 as per WHO specifications (31) to provide a total of 7 pieces and a representative sample of pieces incorporated with deltamethrin and PBO. After cutting, net pieces were labelled, wrapped in aluminium foil and stored in an incubator at 30±2°C before and between testing in laboratory bioassays.
Supplementary laboratory bioassays
To support the interpretation of the results, net pieces cut from whole unwashed, washed and field-aged nets were tested in supplementary laboratory bioassays to characterise the bioavailability of the key AIs on the surface of the study ITNs. Test methods and mosquito strains were selected to capture the intended biological effect of each AI. To characterise the rapid toxicity of alpha-cypermethrin on Interceptor® and the sterilising effects of PPF on Royal Guard®, cone bioassays were performed respectively with unfed mosquitoes of the susceptible Kisumu strain and blood-fed mosquitoes of the insecticide-resistant Akron strain. Meanwhile, given the well-documented unsuitability of cone bioassays for evaluating ITNs treated with certain insecticides (32), tunnel tests were performed with unfed mosquitoes of the insecticide-resistant Covè and Akron strains to characterise the killing effects of PBO on the roof of PermaNet® 3.0 and CFP on Interceptor® G2. The characteristics of these mosquito strains are described below:
- The An. gambiae s.s. Kisumu strain is a susceptible reference strain originating from Kisumu, western Kenya.
- The An. gambiae s.l. Covè strain is a pyrethroid-resistant strain, which is the first filial (F1) progeny of mosquitoes collected as larvae near the experimental hut site in Covè, Zou Department, southern Benin. The species composition and resistance profile have been described previously (21).
- The An. coluzzii Akron strain is an insecticide-resistant strain from Akron near Porto-Novo in southern Benin. It exhibits resistance to pyrethroids, organochlorines and carbamates mediated by the kdr 1014F and insensitive acetylcholinesterase (AChE (ace-1R) G119S mutations and elevated activity of P450s and esterases (33, 34).
During all tests, parallel exposures were performed with untreated net pieces as a negative control, and laboratory conditions were maintained at 27±2°C and 75%±10% humidity.
Cone bioassays
At least 20 mosquitoes aged three to five days were exposed in cone bioassays to each net piece for 3 mins in four replicates of approximately five mosquitoes. At the end of exposure, mosquitoes were transferred to holding cups, provided access to cotton wool soaked in 10% (w/v) glucose solution and scored for delayed mortality. Mortality after 24 h of the unfed Kisumu strain was selected as the primary endpoint to characterise the rapid action and bioavailability of the pyrethroid on Interceptor®. To assess the sterilising effects and bioavailability of PPF in Royal Guard®, blood-fed Akron mosquitoes surviving after 72 h were dissected to observe ovary development and score fertility as previously described. Reduction in fertility relative to control was used as the primary endpoint to characterise the sterilising effects of PPF on Royal Guard®.
Tunnel tests
Tunnel tests are an experimental chamber that simulates the natural behavioural interactions between free-flying mosquitoes and nets during host-seeking. Approximately 100 mosquitoes aged five to eight days were exposed to each net piece overnight in the presence of a guinea pig bait in one replicate test. Nets pieces were given 9 x 1 cm holes to facilitate entry into the baited chamber. In the morning, all mosquitoes were collected from the tunnel and scored for immediate mortality and blood-feeding. Surviving mosquitoes were provided access to 10% (w/v) glucose solution, and delayed mortality was recorded. Mortality after 24 h was used as the primary endpoint to assess the rapid toxicity of synergised pyrethroid on PermaNet® 3.0 (roof), while mortality after 72 h was selected to assess the delayed action of CFP on Interceptor® G2. After the bioassays, net pieces were transferred to a refrigerator and stored at 4±2°C before sending for chemical analysis.
Chemical analysis of net pieces
Net pieces cut from ITNs tested in experimental huts and laboratory bioassays were sent to Centre Walloon de Recherches Agronomiques, Belgium, for chemical analysis to determine within- and between net-variation of AI and retention of AI before and after artificial and operational ageing. The analytical methods used were based on those published by the Collaborative International Pesticides Analytical Council and have been described previously (19, 35).
Susceptibility bioassays
WHO tube tests and bottle bioassays were performed during the trial to assess the susceptibility of adult An. gambiae s.l. collected as larvae from breeding sites near the experimental huts to the AIs in the study ITNs and support interpretation of the results. Tube tests were conducted to assess the frequency and intensity of pyrethroid resistance by exposing mosquitoes to 1x, 5x, and 10x discriminating concentrations of alpha-cypermethrin and deltamethrin (0.05%). Mosquitoes were exposed to 1x, 5x and 10x the discriminating concentrations of alpha-cypermethrin and deltamethrin (0.05%) in tube tests to assess the frequency and intensity of pyrethroid resistance. PBO synergism and the contribution of P450s to pyrethroid resistance were also investigated by testing the discriminating concentrations of both pyrethroids with pre-exposure to PBO (4%). Susceptibility to CFP and PPF was assessed in bottle assays (100 µg/bottle for each AI). All tests followed WHO guidelines (36). Proportional mortality after 24 h and 72 h were used to assess susceptibility to the pyrethroids and CFP, respectively. PPF susceptibility, meanwhile, was assessed based on the proportional reduction in fertility among surviving blood-fed mosquitoes, determined using ovary dissection (37), compared to the untreated control.
Data analysis
For experimental hut trial data, the total number of alive/dead, blood-fed/unfed and fertile/infertile mosquitoes in the different hut compartments was summed for each treatment to calculate proportional means for each outcome with corresponding 95% confidence intervals (CIs). Differences between experimental hut treatments for these proportional outcomes were analysed using logistic regression, while differences in count outcomes (entry) were analysed using negative binomial regression. In addition to the primary explanatory variable of treatment, each model included hut, sleeper, hole index and day as fixed effects to control for variation associated with these factors. Model-derived p-values were used to assign compact letter displays denoting the statistical significance of all pairwise comparisons at the 5% level for all outcomes. Adjusted odds ratios (ORs) were also used to compare the impact of washed nets on mosquito mortality after 72 h to field-aged nets for each net type. The regression analyses were performed in Stata 18. Post-hoc simulation-based power analyses were performed using the ‘power_calculator_ITN’ function in R version 4.3.2. The estimated statistical power to detect a significant difference (i.e. p<0.05) in mosquito mortality after 72 h between PermaNet® 3.0 washed 20 times and field-aged PermaNet® 3.0 withdrawn from households three years post-distribution was 74.2%, 95% CIs: [71.4, 76.9].
Primary outcomes from supplementary laboratory bioassays (% mortality, % reduction in fertility) were plotted on graphs to visualise changes in AI surface bioavailability after washing and field-aging. For the chemical analysis data, we performed a one-way analysis of variance followed by a series of post-hoc Tukey’s Honest Significant Difference tests to compare total AI content between condition categories (new unwashed, washed 20 times, field-aged three years) for each net type. Proportional AI retention in washed and field-aged nets was also calculated relative to new unwashed nets. For next-generation nets containing two AIs, t-tests were conducted to assess whether there was a significant difference in the retention of the pyrethroid compared to the non-pyrethroid. This analysis was performed separately for washed nets and field-aged nets. The susceptibility of the Covè vector population was interpreted based on observed mortality and fertility rates in tube and bottle bioassays according to WHO guidelines (27). Control mortality was consistently low (<5%) in experimental hut trials, susceptibility bioassays and supplementary laboratory bioassays, and thus, adjustment of test mortality using Abbott’s formula was unnecessary.
Ethical considerations
Ethical approval for the experimental hut trials was granted by the institutional review boards of the Benin Ministry of Health (Ref: No. 50, 20/03/2023) and the London School of Hygiene & Tropical Medicine (Ref: 16237–1). Informed written consent was obtained from all human volunteers before their participation. All volunteers were offered a free course of chemoprophylaxis to mitigate the risk of malaria infection, and a stand-by nurse was available throughout to assess any volunteers presenting with febrile symptoms or adverse reactions. Approval for using guinea pigs for tunnel tests and blood-feeding of insectary-reared mosquito colonies was granted by the Animal Welfare Ethics Review Board (Ref: 2020-01B).