Background
Insecticide resistance among malaria vector species now occurs in 84 malaria-endemic countries and territories worldwide. Novel vector-control interventions, including long-lasting insecticidal nets (LLINs) that incorporate new active ingredients with distinct modes of action, are urgently needed to delay the evolution and spread of resistance and to alleviate reversals in malaria-control gains. We aimed to assess the longitudinal effect of two dual-active-ingredient LLINs on insecticide resistance during a cluster-randomised, controlled trial in Benin.

Methods
This 3-year, three-arm, cluster-randomised, controlled trial was conducted between Oct 17, 2019, and Oct 24, 2022, in three districts in southern Benin, to compare the effects of LLINs containing chlorfenapyr–pyrethroid or pyriproxyfen–pyrethroid with LLINs containing pyrethroid only. In 19 292 mosquitoes (Anopheles gambiae sensu lato) collected over 36 months—3 months of baseline followed by 3 years post-intervention—we measured longitudinal phenotypic insecticide resistance profiles using bioassays and genotypic resistance profiles using quantitative, real-time, reverse transcriptase PCR of metabolic resistance genes in two clusters per trial group. The trial was registered with ClinicalTrials.gov, NCT03931473.

Findings
In all three trial groups, a significant effect of LLINs on insecticide resistance selection was evident, with the median lethal dose (LD50) of α-cypermethrin approximately halving between baseline and 12 months post-LLIN distribution (pyrethroid-only LLIN cluster 21: LD50 78·78 μg/ml [95% CI 65·75–94·48] vs 35·93 [29·41–43.86] and cluster 31: 79·26 [65·40–96·44] vs 38·71 [30·88–48·53]; chlorfenapyr–pyrethroid LLIN cluster 43: 104·30 [82·97–133·58] vs 43·99 [35·30–54·86]; and pyriproxyfen–pyrethroid LLIN cluster 36: 63·76 [52·14–77·75] vs 37·96 [30·88–46·69] and cluster 53: 77·67 [57·63–104·56] vs 39·72 [29·26–53·97]). Over the subsequent 2 years, the LD50 of α-cypermethrin increased past baseline values in all three trial groups (year 3 pyrethroid-only LLIN cluster 21: 141·01 [111·70–181·90] and cluster 31: 115·15 [93·90–143·09]; chlorfenapyr–pyrethroid LLIN cluster 43: 97·00 [77·24–123·54] and cluster 55: 126·99 [102·34–161·26]; and pyriproxyfen–pyrethroid LLIN cluster 36: 142·29 [112·32–184·84] and cluster 53: 109·88 [79·31–157·70]). We observed minimal reductions in chlorfenapyr susceptibility and variable but significant reductions in fertility after pyriproxyfen exposure, with an overall trend of increasing susceptibility across trial years. Several metabolic genes were implicated in resistance selection, including CYP6P4 in the pyriproxyfen–pyrethroid LLIN group, which encodes an enzyme known to metabolise pyriproxyfen in vitro, and CYP6P3 and CYP9K1 in the chlorfenapyr–pyrethroid LLIN group, both of which encode enzymes that are involved in pro-insecticide activation.

Interpretation
After 24 months of use, chlorfenapyr–pyrethroid LLINs no longer mitigated pyrethroid resistance selection in this area of southern Benin, which has high malaria transmission dominated by highly resistant A gambiae sensu lato. This finding raises issues for current net-procurement schedules, which are based on an operational net lifespan of 3 years. Knowledge of the effects of next-generation LLINs on insecticide-resistance selection is crucial for the pragmatic design of prospective resistance-management strategies.

Funding
UNITAID and The Global Fund to Fight AIDS, Tuberculosis and Malaria.