Efficacy of pyrethroid-pyriproxyfen and pyrethroid-chlorfenapyr nets on entomological indicators of malaria transmission: third year of a randomised controlled trial in Benin

The present cluster-randomised control trial aims to assess the entomological efficacy of pyrethroid-pyriproxyfen and pyrethroid-chlorfenapyr LLINs compared to the standard pyrethroid-only LLINs, in their third year of community usage. Adult mosquito collections were performed every 3 months, in 4 randomly selected houses in each of the 60 trial clusters, using human landing catches. Adult mosquitoes were morphologically identified and Anopheles vectors were molecularly speciated and screened for the presence of the L1014F kdr mutation using PCR. Plasmodium falciparum sporozoite infection was assessed using ELISA. A subset of An. gambiae s.l. was also dissected to examine parity and fertility rates across study arms. There was no evidence of a significant reduction in indoor vector density and entomological inoculation rate by the pyrethroid-pyriproxyfen [DR 0.94 (95% CI 0.46–1.88), p = 0.8527; and RR 1.10 (95% CI 0.44–2.72), p = 0.8380], and pyrethroid-chlorfenapyr [DR 0.74 (95% CI 0.37–1.48), p = 0.3946; and RR 1.00 (95% CI 0.40–2.50), p = 0.9957] LLINs, respectively. The same trend was observed outdoors. Frequencies of the L1014F kdr mutation, as well as parous and fertility rates, were similar between study arms. In the third year after net distribution, entomological indicators show that the two dual active-ingredients nets performed similarly to the standard pyrethroid-only LLIN. To maintain malaria gains, it is crucial that net distribution cycles fit with their operational lifespan.

developed over the past few years incorporate a pyrethroid plus a partner insecticide with a different mode of action, such as pyriproxyfen, a growth regulator which sterilizes mosquitoes, or chlorfenapyr, which affects the mitochondria and disrupts the production of cellular energy required for flight.
The findings of the two first years of the present cluster-randomized controlled trial (RCT) demonstrated that the pyrethroid-chlorfenapyr LLIN significantly reduced vector density and the entomological inoculation rate (EIR) by 56% and 66%, 53% and 70%, in the first and second year of the trial, respectively.However, the pyrethroid-pyriproxyfen LLIN significantly reduced the EIR by 58% only during the first year of the trial, but did not affect either of the two indicators (vector density, and EIR) in the second year of the trial 5 .Similarly, in participants of any age, malaria infection prevalence was significantly reduced by 40% by the pyrethroidchlorfenapyr LLIN at 18 months after LLIN distribution; a significant reduction in malaria prevalence was not observed with the pyrethroid-pyriproxyfen LLIN 5 .The same trend was observed in another RCT conducted in Tanzania 6 , which led the World Health Organization (WHO) to grant a full recommendation for the deployment of pyrethroid-chlorfenapyr LLINs in pyrethroid resistance areas, and a conditional recommendation for the use of pyrethroid-pyriproxyfen LLINs 7 .
Typically, net replacement in the community is conducted every three years through mass distributions led by National Malaria Control Programmes (NMCP).Based on previous data of field performance of standard pyrethroid-only, LLINs are expected to remain biologically efficacious over 3 years of community usage 8 .However, recent field studies indicate that net bio-efficacy and durability can vary substantially between net types and in different communities 9 .In the third year of the RCT conducted in Tanzania, the pyrethroid-chlorfenapyr LLIN was found to significantly reduce the vector density by 54% and the EIR by 68% compared to the pyrethroid-only LLIN while the pyrethroid-pyriproxyfen LLIN did not 10 .Similarly, malaria infection prevalence was significantly reduced by 43% with the pyrethroid-chlorfenapyr LLIN, while there was no evidence of a reduction of this indicator with the pyrethroid-pyriproxyfen LLIN 10 .
The present study reports the impact of pyrethroid-pyriproxyfen and pyrethroid-chlorfenapyr LLINs on entomological indicators in the third year of a RCT conducted in an area of high pyrethroid resistance in Benin.
The 123 villages of the study area were divided into 60 clusters, with 20 randomly allocated to each of the three followings arms 5 : (i) pyrethroid-chlorfenapyr LLIN: Interceptor® G2 (Intervention 1); (ii) pyrethroidpyriproxyfen LLIN: Royal Guard® (Intervention 2); and (iii) standard pyrethroid-only LLIN: Interceptor® (Control arm) (Fig. 1).Each cluster had a core area, and a buffer area of at least 1000 m for a minimum of 100 households.Entomological indicators were only measured in the core area, although the interventions were deployed in the whole cluster.

Human landing catches (HLCs)
The present study took place between June 2022 and April 2023, with collections performed in all 60 clusters every three months, equating to a total of four collection rounds over the study period.In each cluster, 4 houses per round (1 randomly selected, and 3 others chosen by the field team, 15-20 m around the first one) were sampled from 07:00 p.m. to 06:00 a.m. using human landing catches (HLCs).In each house, two volunteers (one sitting indoors, and the second outdoors) collected all mosquitoes that landed on their lower limbs using haemolysis tubes and flashlights.For the entire study period, a total of 960 collections (4 indoor collectors × 60 clusters × 4 rounds) were performed indoors.The same number of collections were also performed outdoors.
Mosquitoes collected through HLCs were morphologically identified to species-level using the taxonomic keys of Gillies and Meillon 12 .Heads-thoraces of a subset (32.4%: 9356/28,915) of unfed specimens of An. gambiae s.l., randomly selected indoors and outdoors, and across collection hours in each cluster, were screened by ELISA-CSP to detect P. falciparum sporozoite infection 13 , and dissected to determine the parous rate 14 .Legs and wings of the same samples were used for molecular identification of sibling species 15 and presence/absence of the L1014F kdr mutation 16 .

Collection using mouth aspirators
To assess the impact of pyriproxyfen on the fertility of An. gambiae s.l., four clusters in each of the three study arms were surveyed over two rounds of collection (August 2022 & January 2023).At each round of collection, 10 houses randomly chosen per cluster were sampled between 07:00 to 09:00 a.m., by two entomology technicians and two local guides.Thus, a total of 240 houses (10 houses × 4 clusters × 3 arms × 2 rounds) were surveyed.In each house, all mosquitoes that rested on hung clothes, walls, furniture, roofs, earthen jars and others were collected using mouth aspirators.Thereafter, they were released into paper cups, provided with a 10% sweetened sugar solution, placed in cooler boxes, and carried to the field-laboratory where blood-fed An. gambiae s.l. were morphologically identified and released in cages for a three-day-resting time which allowed for blood-meal digestion, prior to their dissection and determination of Christopher's egg stage (I-V) of development 17 (Fig. 2).
Of note, the dissection was performed by delicately pulling the last two abdominal segments of each mosquito, which allowed to extract the ovaries.A mosquito was considered fertile when it had stage V eggs.

Data analysis
For both locations (indoor and outdoor), the nightly vector density was calculated at the household level by dividing the total number of Anopheles vectors (An.gambiae s.l.+ An. funetus s.l.+ An. nili s.l.) collected by the total number of collection nights for each round.The number of sporozoite positive vectors was divided by the total number of tested vectors to determine the sporozoite rate (SR).The indoor/outdoor nightly EIR was generated at the cluster level by multiplying the mean indoor/outdoor vector density by the indoor/outdoor SR.The parous rate in An. gambiae s.l. was calculated at the cluster level by dividing the number of parous mosquitoes by the total dissected.Similarly, the fertility rate in An. gambiae s.l. was estimated at the cluster level by dividing the number of fertile mosquitoes by the total dissected.The resistance mutation frequencies were calculated at Vector density and EIR were compared between study arms using a mixed-effect generalised linear model with a negative binomial distribution and with cluster as a random effect.To analyse SR, parous rate, and fertility rate between study arms, a mixed-effect logistic regression was used, with cluster as a random effect.All analyses were performed using Stata 15.0 (Stata Corp., College Station, TX).

Ethical considerations
Ethical clearance for the present trial was issued by Benin's National Ethics Committee for Health Research (N°30/MS/DC/SGM/DRFMT/CNERS/SA, Approval n°6 of 04/03/2019), and the Ethics Committee of the London School of Hygiene and Tropical Medicine (16237-1).Prior to their involvement, all mosquito collectors, heads of households, and community leaders gave their informed consent.Only collectors trained to capture mosquitoes prior to being bitten, were involved in the study.Before the start of the trial, they were vaccinated against yellow fever.Whenever they experienced symptoms similar to those of malaria, they were taken care of at the closest health facility.All study procedures were conducted following the relevant guidelines and regulations.
In the study area, Mansonia spp. was the most collected mosquito both indoors and outdoors in all study arms as it accounted for half of all collected mosquitoes, while Culex spp.accounted for one-fifth.Aedes spp (2.1%) and other mosquitoes (0.003%) were also found, but at very low frequencies (Fig. 3).

Allelic frequency of the L1014F kdr mutation in An. coluzzii and An. gambiae s.s.
Irrespective of location, the allelic frequencies of the L1014F kdr mutation were overall similar between the two intervention arms and the control arm for each molecular species (An.coluzzii and An.gambiae s.s.) (Table 4).

Discussion
Current LLIN procurement and deployment strategies are based on an anticipated operational longevity of three years.However, there is a growing body of evidence from the field that net bio-efficacy, physical durability and attrition rates can vary substantially between countries and populations 9 .As novel vector control interventions are developed, including dual active-ingredient (AI) LLINs, it is crucial to understand their impact throughout their lifespan, to develop best practices for their use in pragmatic insecticide resistance management schemes 18,19 .
In this study, during the third year of community usage, pyrethroid-chlorfenapyr and pyrethroid-pyriproxyfen LLINs showed no evidence of a significant reduction of vector density, SR and EIR.Similarly, parous and fertility rates did not significantly decrease in the two intervention arms compared to the control one.These results align with the lack of epidemiological impact seen between arms in the third year 20 and contrast with the results of the first two years of the trial, where pyrethroid-chlorfenapyr LLINs had a greater impact on epidemiological and entomological indicators than the standard pyrethroid-only LLINs.Moreover, no significant increase in the L1014F kdr frequency was induced by the two dual AI LLINs compared to the standard pyrethroid-only LLIN.
A contributing factor which may explain the reductions in efficacy of pyrethroid-chlorfenapyr LLINs may include decreases in community usage over time (from 83% at 9 months post-distribution to 52% at 36 months post-distribution) 20 .In addition, reduction in biological efficacy of the partner insecticide (Chlorfenapyr) may www.nature.com/scientificreports/also explained the lack of efficacy observed on entomological outcomes in the third year of the RCT.Indeed, after 2 years of community usage of LLINs in the present trial, there was a 75% reduction of chlorfenapyr concentration (52 mgAI/sqm at 24 months versus 208 mgAI/sqm when new) in pyrethroid-chlorfenapyr LLINs 20 , and after 3 years of use nearly no chlorfenapyr were left in the same type of nets evaluated in another RCT conducted in Tanzania 10 .In an associated study conducted in experimental huts in Tanzania, the decline in chlorfenapyr concentration was associated with reduced vector mortality following exposure to pyrethroid-chlorfenapyr LLINs that had been deployed in the field for two years or longer 19 .Findings of a trial conducted in Papua New Guinea also showed that the decreased bio-efficacy of long-lasting insecticidal nets was followed by the loss of their effectiveness on malaria transmission 21 .Study findings contrast with those from the RCT in Tanzania, where pyrethroid-chlorfenapyr LLINs continued to significantly reduce indoor vector density by 54% and EIR by 68% in the third year post-net distribution 10 , though the net usage in Tanzania (23%) was considerably lower compared to Benin (52%).Additionally, despite indications of a waning insecticidal effect of Chlorfenapyr in the nets after one year 19 , the Tanzanian RCT demonstrated ongoing effectiveness over a 3 years period.One possible explanation is the much (5-10 times) lower vector density in Tanzania compared to Benin.In Tanzania, the gradual build-up of the mosquito population would inherently take a longer period than in Benin.Therefore, the differences in entomological indicators between pyrethroid-chlorfenapyr LLINs and standard pyrethroid-only LLINs reported in the third year in Tanzania may be a carryover effect from the previous year rather than a true effect of the pyrethroid-chlorfenapyr LLINs.
In the study area, An. gambiae s.l.constituted the predominant Anopheles mosquitoes species, primarily composed of An. coluzzii, followed by An. gambiae s.s.This trend remained consistent with observation made before the net distribution and during the two first years post-intervention of the trial 11,22 , indicating that net deployment did not significantly change Anopheles species composition over time.This contrasts with findings in Burkina-Faso by Sanou et al. 23 , where the authors reported a significant shift in Anopheles species composition from An. coluzzii to An. gambiae s.s. as the most prevalent species following the introduction of LLINs.
In An. gambiae s.l., parous rates were similar between the pyrethroid-chlorfenapyr LLIN and standard pyrethroid-only LLIN arms in the present trial.The same observation was made over the two first years of the trial 22 .This might be due to highly variable confounding factors such as temperature and relative humidity that strongly influence parity 24 .
A three-minute exposure of mosquito tarsi to pyriproxyfen should suffice to inhibit their fertility 25 , therefore would expect a higher fertility rate in both the pyrethroid standard-only LLINs and the pyrethroid-chlorfenapyr LLIN arms as compared to the pyrethroid-pyriproxyfen LLIN arm, which did not occur in the present trial.This unexpected trend might be due to the decay of the pyriproxyfen incorporated into the nets as observed in the present study (90 mgAI/sqm at 24 months versus 289 mgAI/sqm when new: 69% reduction in pyriproxyfen concentration over 2 years).
The similar frequencies of the L1014F kdr mutation across study arms could be due to the mutation originally close to fixation in the whole study area due to successive mass LLIN distribution campaigns, and uncontrolled use of pyrethroids for agricultural purposes.
The present trial emphasizes the need to assess the long-term efficacy of any mosquito nets in the community, as this may vary over time, and from place to place.Indeed, findings of the present trial suggest that the two dual ai LLINs had an effective lifespan of two rather than three years, which corroborates previous findings by Gnanguenon et al. 26 .This needs to be taken into consideration by NMCPs, decision-makers, and LLINs procurement and distribution agencies which so far assume a three-year LLINs lifespan for any LLINs.Thus, to achieve a good efficacy of LLINs in the community, the replacement policy should be aligned with their operational lifespan.Moreover, manufacturers should develop LLINs with better materials, and their distribution should be followed by community sensitization on correct usage to avoid early appearance of holes, and repair of torn/ holed nets.

Conclusion
Findings from the present trial revealed that pyrethroid-chlorfenapyr LLINs were no longer better to standard nets against key entomological indicators of malaria transmission in their third year of community usage.There is a need for ongoing monitoring and evaluation of LLINs with dual active ingredients during their deployment to understand their effective lifespan in diverse epidemiological settings.This information is essential for malaria control programs to formulate appropriate net deployment strategies and guide manufacturers in developing longer-lasting nets to sustain control effort.