**4. Discussion**

Resistance to insecticides used to control mosquito vectors of disease is widespread, strengthening and evolving in the face of selection pressure from a limited number of chemistries available for use in public health [75]. New generations of insecticide-treated nets (ITNs) are now available based on novel mode of action chemistries, and other novel insecticide-based tools are in development to address this challenge. Dual-AI ITNs, including those containing two insecticidal compounds and a single insecticide paired with a synergist, promise greater effectiveness against pyrethroid-resistant mosquitoes. It is intended that the partner AI will have an effective lifespan of three years to match that of the pyrethroids currently in use, so that the new ITNs will fit into the existing campaign framework and contribute to resistance management. The dual-AI ITNs do, however, present a challenge in measuring their bio-efficacy in a laboratory setting, which is required to monitor their effective life through durability monitoring studies. Existing methods designed for ITNs containing only pyrethroids may not be suitable for those containing different modes of action insecticides or synergists. There is a need to test them against pyrethroid-resistant strains to quantify the entomological impact of the second AI.

Bioassays are an important proxy test for the surface availability of AI, and for demonstrating the efficacy of ITNs in killing mosquitoes under standardised conditions. As our understanding of resistance mechanisms increases, so does the complexity in determining relative contributions and how they affect the bio-efficacy of different chemistries and formulated products. These dynamics may alter with changing parameters, such as surface concentrations of AI declining over the lifespan of an ITN. There is, thus, the potential for the introduction of grea<sup>t</sup> variability into the results of bioassays designed for pyrethroid susceptible subjects, when considering the specific characteristics of the pyrethroid-resistant strain used, as well as methodological issues related to the mode of action of the ITN. To help minimise the noise in bioassay results due to these various factors, it is imperative that we clearly define or describe material inputs into these studies. A key aspect of this is to standardise or characterise the mosquito strains being used in these assays as far as possible, to provide interpretable data for analysis and to allow the comparison of results over time, between products and between testing centres. In an operational setting there is inevitably a need to balance improved characterisation or standardisation of inputs with the availability of suitable controls and logistical and financial constraints.

This paper describes a collaborative effort by researchers and implementers interested in insecticide resistance and evaluation of ITNs to agree on an approach to characterise mosquito strains to evaluate dual-AI ITNs and a set of specific criteria for the phenotype a suitable strain should have. Such an approach to method development, while somewhat time-consuming, does allow those implementing these activities to agree on a standardised method. This approach could be applied to other sources of potential variation in vector control efficacy and/or durability studies. For example, current guidelines for monitoring durability of ITNs were developed for pyrethroid-based nets but have been adapted and updated for dual-AI ITNs through a similar consensus approach by Innovation to Impact (I2I) [30].

Care was taken in designing a methodology for strain characterisation to ensure a comprehensive, robust approach, feasible in the context of the level of effort needed from those facilities carrying out this work. The standard operating procedure (SOP) decided upon and presented here (Supplementary Information) identifies some key parameters for characterisation, presents criteria for a suitable strain, and provides guidance on the rearing and quality control of the mosquitoes used in testing. Components are separated into those which are critical and those which are desirable and should be included where resources and logistics allow. Although these recommendations may have cost and time implications, these are balanced by promise of greater interpretability of the data produced in notoriously difficult studies to analyse and compare. The SOP will be made freely available through Innovation to Impact (I2I), to be trialled. Future studies with dual-AI ITNs, such as durability monitoring activities currently underway [76,77], allow its suitability to be reviewed and the methodology to be refined based on the experience of operators.

The consensus recommendation of the group of experts was to use a laboratory strain of mosquitoes for durability monitoring of dual-AI ITNs, to allow controlled rearing, quality control and characterisation to maintain and monitor the consistency of material over time. Regular and thorough characterisation of laboratory strains used for longitudinal bioefficacy testing is critical to ensure data validity and reliable interpretation of findings. There was some discussion favouring using locally relevant mosquitoes, and a desire to determine the operational significance of strain characterisation of bioefficacy data generated in the laboratory. Although the goal of durability monitoring is separate from efficacy testing, if the latter is a key question, there may be a preference for testing nets against local strains or against multiple recently-colonised resistant strains, which may express different, but locally-relevant, resistance mechanisms and give additional information about how the nets perform in situ. However, durability monitoring aims to determine that over multiple geographical locations ITNs continue to remain physically and biologically active for the duration of their expected lives. This is particularly difficult for dual-AI ITNs that must

be tested against resistant strains to ensure the non-pyrethroid component is still biologically active. For longitudinal experiments, such as durability monitoring or RCT trials, it is appropriate to use a well-characterised and consistent strain of mosquitoes. When a dual-AI ITN is being tested, it is critical to know the pyrethroid resistance phenotype of the mosquitoes being used. Positive control nets are a useful benchmark for interpreting changes in relative bioefficacy of dual-AI ITNs through time against a background of slight fluctuations in resistance phenotype of laboratory test strains.

The scope of this consultative exercise was the efficacy testing of the second AI in a dual-AI ITN combining a pyrethroid with a second insecticide or synergist. The 2022 Product Review Report from the WHO PQT/VCP team on insecticide treated nets formulated with a pyrethroid and either PBO or a second AI [78] recommended the development of 'improved guidance regarding the selection of mosquito strains to be used in bioassay and efficacy testing' including characterisation of resistance. The proposed strain characterisation approach addresses this need, and could be used in any situation where a pyrethroid-resistant strain is used in research. The general approach of characterising the biological material used in research and reporting results of the characterisation alongside the experimental data to aid interpretation is recommended as good practice. For example the WHO PQT/VCP Product Review Report [78] recommends the characterisation of the local vector population at the sites of experimental hut trials. This method establishes a solid framework that could be used with minor modifications to adapt to ITNs with unique AIs as they develop and become available. For example, specific additional or alternative considerations may apply when characterising a strain used to test ITNs containing two non-pyrethroid AIs. In this case, there will likely not be populations of mosquitoes available that are resistant to either AI. An alternative method would be needed to separate and measure the activity of each AI; for example, based on their differential speed of action. However, the requirements for maximising consistency and characterising the mosquito strain used to test the durability of these nets would be the same. The approach could also be readily adapted for characterisation of strains for evaluation of dual-AI products beyond nets, such as IRS formulations.

The development of this consensus methodology is part of a wider effort spearheaded by I2I to identify and address sources of variability in entomological data related to vector control product evaluation. To produce robust data, consistent across time and between operators, and to interpret results in a meaningful way, it is important to standardise or characterise material inputs into studies. The proposed method for characterising pyrethroid-resistant mosquitoes is the first of what is hoped to be a collection of supporting SOPs generated by, and made available to, the vector control community to help improve the generation and interpretation of entomological data for decision making.
