**1. Introduction**

Insecticide-treated nets (ITNs) have been critical in controlling malaria. However, widespread resistance to the pyrethroids, which have been the sole insecticide class used on all ITNs until recently, threatens the continued effectiveness of standard ITNs [1]. Therefore, there is a need for new ITNs that include insecticides from classes with new modes of action to combat pyrethroid-resistant vector populations [2]. Several ITNs have been pre-qualified by the World Health Organization (WHO) containing a pyrethroid plus a second active ingredient (AI), which may be another insecticide (chlorfenapyr, CFPR; pyriproxyfen, PPF) or the synergist piperonyl butoxide (PBO), hereafter referred to as dual-AI ITNs [3].

There is a need to test the bioefficacy of ITNs in the laboratory. Here we are using the term 'bioefficacy' to mean the ability of a net sample to kill mosquitoes in a bioassay, contrasted with the efficacy of an ITN, which describes the net's ability to meet its objective of offering personal and community protection against transmission of mosquito-borne disease. Prototypes may need to be compared during product development, and research may be conducted to explore how an ITN works. Before distributing an ITN, the national malaria control programmes (NMCPs), or funders, may want to test its efficacy against local mosquito populations. During randomised control trials to determine the efficacy of ITNs (for example [4,5]), and during post-deployment monitoring (for example [6]), use and attrition of ITNs are monitored, and samples of deployed nets are collected over time to monitor their physical durability, analyse insecticide content and measure the bioavailability of each AI, using agreed-upon and validated bioassay methodologies (i.e., WHO cone bioassay or tunnel tests) [7,8]. This testing may be done at the time of collection, or all samples may be accumulated for simultaneous testing at the end of the study. Existing methods for durability monitoring [9], are optimised to evaluate pyrethroid-only ITNs, but the bioassay component may need to be adapted to be suitable for dual-AI ITNs.

The ability of a dual-AI ITN to kill insecticide-susceptible mosquitoes can be measured using standard methodologies and a susceptible laboratory strain. If the entomological endpoint of the second AI is different to the rapid knockdown and kill achieved by a pyrethroid, it will be possible to separate the effect of the pyrethroid and the second AI. To monitor the persistence and additional efficacy of the second AI, a pyrethroidresistant strain must be used, the majority of which will survive contact with the pyrethroid so that the effect of the second AI can be measured. Traditionally, mortality caused by pyrethroid exposure is measured to 24 h, as this insecticide class is fast acting. To control for delayed mortality caused by the pyrethroid in a resistant strain, where the second AI causes delayed mortality, mortality could be measured to the same time point when characterising pyrethroid susceptibility. The nature of the resistant strain needs to be considered, as this will affect the interpretation of data from the durability monitoring testing. The Vector Control Advisory Group (VCAG) of the WHO proposed the following criteria, in 2014, for mosquito strains suitable for use in screening for cross-resistance between insecticidal products [10]: at least 3 strains, two of which have significant metabolic resistance, representing the broad spectrum of known resistance mechanisms, ideally from a provided list of standard strains, or a strain that is fully characterised, and has a resistance level greater than 10-fold that of a susceptible strain of the same species at the LC50, tested in parallel. Though not specified at the time, this LC50 would ideally be measured at the time point of interest for the second AI. This may form the basis of selecting suitable strains for bioefficacy testing or durability monitoring of dual-AI ITNs, but developments in the understanding of the molecular characteristics of mosquitoes have been made since these recommendations were released. New modes of action of insecticide are now being considered, meaning that there are limitations to these criteria, and practical challenges in meeting them, and they should, therefore, be reviewed.

The number of different resistance mechanisms that have now been identified, and would need to be screened to characterise a strain fully, is increasing over time and include overexpression of detoxifying enzymes [11,12], involvement of sensory appendage proteins [13] or the salivary protein gland [14,15], or cuticular thickening [16,17]. Insecticide resistant mosquito populations possess different combinations of mechanisms, and the relative contribution of these mechanisms to resistance differs between populations. These features have evolved to confer resistance to insecticides to which mosquitoes have been exposed, but some may also confer cross-resistance to new insecticides even with novel modes of action. Representing all known resistance mechanisms even in three strains, would be a major challenge, and, given that both the range of mechanisms expressed and our knowledge of these evolves over time, will always risk omitting resistance mechanisms that have not ye<sup>t</sup> been identified. This is of particular concern for entirely new AIs coming to market, resistance mechanisms for which have not been identified.

Even if a list of standard representative laboratory colonies was established, there would be no guarantee of expected results in testing between sites or across time. For example, a colony that is nominally from a common strain may differ from a colony of the same strain held at a different test facility, due to differences in establishment and/or stabilisation in new laboratories and related selection pressures, genetic drift, inbreeding and genetic bottlenecks [18], insecticide exposure to maintain resistance [19] or through contamination events, rearing conditions that may affect fitness [20–22] or microbiome characteristics [23,24]. Resistance may shift over time, particularly if a strain is transferred between facilities or if selection pressure is not maintained. In addition, mosquito strains show temporal variability in their physiological response to insecticides. Routine efforts to characterise resistance phenotypes in lab strains are commonly based on the use of discriminating concentrations (DCs) or resistance intensity assays, rather than dose-response assays, which would be needed to establish LC50 values and resistance ratios.

Modes of action of insecticides used in dual-AI ITNs currently under evaluation mean that bioassays and protocols designed to measure bioefficacy of a pyrethroid may not be suitable. When considering chlorfenapyr, for example, the metabolic status of a mosquito is believed to affect metabolism of the pro-insecticide to the active form, and subsequently mortality rate [25]. Metabolic rate may be affected by extrinsic factors, such as temperature, time of day [26], and intrinsic factors, such as the nutritional status of the mosquito [27–29]. Bioassay choice and design may play a part, affecting, for example, how much mosquitoes move, or whether mosquitoes blood feed. As well as having higher rates of expression of detoxifying enzymes, different strains may have different metabolic rates, which may be related to the resistance mechanisms they express.

Finally, different criteria may apply when selecting a single strain or multiple strains to monitor the bioefficacy of dual-AI ITNs for durability monitoring, which is the focus of the present study. For example, to monitor the durability of ITNs it is not critical that the target species is used for bioefficacy testing, as long as the sensitivity of the species used is such that the bioefficacy of each AI can be detected across a relevant concentration range.

When conducting a durability study on ITNs, testing all samples at a similar time at the end of the study may help to minimise any temporal rearing effects on mosquito strains. Alternatively, testing ITN samples as they are collected avoids the logistical resources needed to store nets or net samples, and will minimise net degradation and/or loss of insecticide bioavailability before the bioassays. However, correct storage, according to manufacturers' instructions, should minimise degradation. Regardless of the approach, the large number of samples means it will not be possible for all net samples to be assayed by the same facility, at the same time, using the same cohort of resistant mosquitoes. In order to be able to compile and compare results of testing across a study, between facilities and time points, and to compare results between studies, there is a requirement for either (a) standardisation, such that the method and inputs are the same in all cases to minimise differences between results, or (b) characterisation of the inputs, so that results can be interpreted and, where differences are seen, any disparities between the inputs can be taken into account, or (c) a combination of the two. Depending on the specific questions of a study it might not be interesting to separate out the bioefficacy of each AI in a dual-AI ITN, and it may be sufficient to test the relevant endpoints in the mosquito population of interest or to compare results of bioefficacy testing with chemical analysis results on the same net samples. For the purpose of this consultation, we were interested in being able to separate out the effects of each AI, which is particularly relevant for randomised control trials (RCTs) of new types of ITN, where the durability of the second AI may not be known but is important to understand the added benefit over a pyrethroid-only ITN. In this case, the pyrethroid may be tested using a standard susceptible strain of the target species in the case where the second AI has an effect other than the rapid knock down and mortality caused by pyrethroids, but to test the additional benefit of the second AI (chlorfenapyr (CFPR), pyriproxyfen (PPF) or piperonyl butoxide (PBO)) mosquitoes must be pyrethroid-resistant, and assessed over the timescale of action relevant to the second AI. Inputs to the protocol for durability monitoring of dual-AI ITNs, therefore, include the pyrethroid-resistant mosquitoes used to test the second AI.

Selection, characterisation and standardisation of resistant mosquitoes are complex. The consultation process described in this paper aimed to produce a guide to the use of resistant strains for laboratory bioefficacy testing of dual-AI ITNs. Bioefficacy testing of ITNs tests for the presence of sufficient quantities of bioavailable compound to induce the desired endpoint in mosquitoes, usually mortality, and repeated testing over time can be used to measure durability of an ITN, for example, during an RCT trial. The purpose of this consultation was to develop, by consensus, a protocol for ensuring that the use of pyrethroid-resistant mosquitoes can be sufficiently characterised or standardised to allow compilation, comparison and interpretation of bioefficacy data across studies designed to monitor durability. A standard operating procedure (SOP) was produced which can be used by project teams, and is a supporting document to consensus SOPs recently developed for durability monitoring of new net types [30]. The same SOP may be useful in characterising a pyrethroid-resistant strain of mosquitoes for other purposes, or it could be adapted to meet different specific needs. This project forms part of a package of work to improve entomological methods in vector control, and is supported by Innovation to Impact (I2I) at the Liverpool School of Tropical Medicine (LSTM).

#### **2. Materials and Methods**

A group of experts was assembled, based on attendance at a preliminary discussion during a sidebar meeting at the American Society of Tropical Medicine and Hygiene (ASTMH) conference in November 2019, with additional invitees identified by the initial group, based on research interest in insecticide resistance, experience in the testing of new net types, or involvement in current or recent community scale trials of ITNs. Four virtual meetings of these stakeholders were held between April 2020 and August 2021, during which the need for a means to standardise or characterise resistant mosquitoes for the purpose of efficacy testing of dual-AI ITNs was agreed upon, possible approaches proposed and advantages and challenges of each discussed. Based on these discussions a protocol was drafted and iteratively refined by the group, who all then approved the final consensus SOP. A summary of these discussions, the final protocol, and the justifications for arriving at the proposed approach are presented here, and a detailed SOP is included as Supplementary Information.
