WHO susceptibility tests using permethrin and alpha-cypermethrin treated papers were conducted against F1 progeny of mosquitoes collected from huts containing untreated nets before and during the trial. Mortality recorded using 0.75% permethrin papers was 46.7% for An. gambiae and 56.7% for An. funestus in the first trial (2015) and 43% and 52.6% respectively in the second (2016), indicating resistance to pyrethroids in both species. Mortality using 0.05% alpha-cypermethrin papers was 52.7% for An. gambiae during the first trial, and other alpha-cyano pyrethroids such as 0.05% deltamethrin and 0.05% lambdacyhalothrin gave a similar 73.8% and 50.6% mortality respectively during the second hut study (alpha-cypermethrin papers were not available at the time). Concurrent mortality using the same insecticide test papers against susceptible An. gambiae Kisumu was 100% in each case. Insecticide resistance intensity testing showed Zenet field An. gambiae to have over 30-fold resistance to pyrethroid (permethrin) compared to susceptible An. gambiae Kisumu.
Phase II - experimental hut trials
Mosquito entry into experimental huts
The average number of mosquitoes entering and exiting the hut are shown in Table 1. The geometric mean number of An. funestus collected during the first trial ranged from 0.6 to 1.5 per hut per night. During the second trial the geometric mean number of An. funestus ranged from 1.3 to 1.8 per hut per night. In both trials significantly fewer An. funestus were collected from the huts with chlorfenapyr CTN compared to the huts with the untreated nets. No consistent deterrent effect was observed with IG1 (alpha-cypermethin alone) or IG2 compared to untreated nets.
Mortality and overall killing effect
The overall percentage mortality by treatment arm is shown in Figure 1. Because chlorfenapyr shows the property of delayed mortality, which reaches a zenith 72h after mosquitoes enter into the huts with chlorfenapyr treated nets, both 24h and 72h mortality are presented in Table 2. Percentage mortality corrected for untreated net control is also shown.
In the first trial, control-corrected mortality of An. funestus after 24h was 5-6% in the huts with the unwashed IG1 and in the huts with the IG1 washed 20 times (Table 2). Mortality in these treatment arms was significantly different from the mortality in the huts with the unwashed IG2 (42%), the IG2 washed 20 times (44%) and the chlorfenapyr CTN (37%). After 72h, control corrected mortality was significantly higher than after 24h across most of these treatments (Table 2). Mortality was significantly higher in the huts with the IG2 unwashed and washed 20 times compared with the IG1 unwashed and washed 20 times treatments.
In the second trial, the trend was slightly different. Control corrected mortality significantly increased once again between 24h and 72h with the unwashed IG2 (from 22% to 46%), the IG2 washed 20 times (from 6% to 41%) and the chlorfenapyr CTN (from 18% to 36%) (Table 2). But unlike the first trial, control-corrected mortality showed no significant change between 24h and 72h with the unwashed IG1 (1.9% to 3.8%) and with the IG1washed 20 times (6.6% to 5.6%). Therefore, delayed mortality of An. funestus after 72h was significantly pronounced only in the huts with the chlorfenapyr CTN, the unwashed Interceptor G2 and the Interceptor G2 washed 20 times.
Natural mortality of An. funestus after 72h in the huts with the untreated nets in the first trial was significantly lower (21%) than the overall mortality in huts with the IG1 unwashed (37%) or IG1 washed 20 times (34%). In the second trial, natural mortality after 72h in huts with the untreated nets was lower (13%) than in the first trial (21%), but on this occasion the untreated nets showed no difference in mortality compared to IG1 unwashed (16%) or IG1 washed 20 times (18%) which also stayed low. A further difference between the two trials: in the first, both IG2 and IG1 showed significantly delayed mortality between 24h and 72h, in the second trial only IG2 showed significantly delayed mortality between 24h and 72h and not IG1.
Because the untreated net control showed 21% (trial 1) and 13% (trial 2) mortality after the 72h holding period, the observed mortality of IG2 presented in Fig 1 after 72h observation was considerably higher (range: 49-70%) than the control-corrected mortality (range: 41-62%) presented in Table 2.
The ‘overall killing effect’ by the IG1 and IG2 interventions were consistent with percentage mortality of the IG1 and IG2 treatments observed in the huts. In the first and second trials, IG1 killed up to 16% and 0% of An funestus respectively and IG2 killed up to 49% and 38% respectively.
Meta-analysis of mortality
In the meta-analyses of mortality between the two trials, the comparison of relative risk between the unwashed IG2 and the untreated net was 3.36 (CI 2.3, 4.9) (P=0.001). The comparison of mortality relative risk between the chlorfenapyr CTN and untreated net, 3.24 (CI 2.4, 4.2) (P=0.001), was therefore quite similar to that of the unwashed IG2 and untreated net. The comparison of relative risk between the unwashed IG1 and the untreated net was rather less (1.60, CI 1.1-2.3) (P=0.01), indicating a smaller effect size of alpha-cypermethrin on mortality. The effect of the comparison between IG2 and IG1 was 2.27 (1.1, 4.6) (P=0.012), confirming the greater contribution of chlorfenapyr than of alphacypermethrin to IG2 mortality. This was further confirmed by the comparison of chlorfenapyr CTN to IG2: the risk ratio was a non-significant 0.96 (0.7, 1.23) (P=0.231) implying that chlorfenapyr was making most of the contribution to mortality in IG2 and not alpha-cypermethrin. The similarity of relative risk between unwashed IG2 and IG2 after 20 washes (1.04, CI 0.8-1.3) (P=0.73) indicated no loss of mortality effect in IG2 between 0 and 20 washes (Fig 3a).
Blood feeding rates and personal protection
In the first trial, blood-feeding rate of An funestus was significantly higher in the huts with the untreated net than in the huts with IG1. Blood-feeding rate of An funestus was highest in the huts with the chlorfenapyr CTN. There were no significant differences in blood-feeding rates between the huts with the IG1 or the IG2, with or without washing (Table 3).
In the second trial, while blood-feeding rate may have seemed higher in the huts with the untreated net than in the huts with the unwashed IG1 or IG1 washed 20 times the differences were not significant. Once again, no significant differences were evident between any of the IG1 and IG2 treatments. And on this occasion, the difference between the untreated net and the chlorfenapyr CTN was also non-significant. Seven of the eight treatments that did show some degree of blood-feeding inhibition contained an alpha-cypermethrin component whether in IG1 or when twinned with chlorfenapyr in IG2.
In the first trial, personal protection in huts with IG1 and IG2 was significantly higher than in huts with the untreated nets. In the second trial, while the numbers of An funestus that were blood fed was highest in huts with the untreated net, neither the IG1 nor the IG2 treatments showed any reduction in the proportion blood-feeding nor personal protection that was statistically significant compared to huts with the untreated net.
Meta-analysis of percentage blood feeding
In the meta-analyses of blood-feeding between the two trials, the comparison of relative risk between the unwashed IG2 versus the untreated net was 0.70 (CI 0.44, 1.11) (P=0.133). The comparison of relative risk between the unwashed IG1 versus the untreated net was also quite similar (0.61, CI 0.39, 0.97) (P=0.035) to that of IG2 above (Fig 4b). The comparison of relative risk between the chlorfenapyr CTN versus the untreated net was 0.97 (CI 0.39-2.44) (P=0.95). Considering these results in reverse order: chlorfenapyr treatment had no effect on blood-feeding inhibition compared to no treatment, alpha-cypermethrin was the sole AI contributing to blood-feeding inhibition in the comparison of IG1 to untreated net. The inference is, the contributing active ingredient to blood-feeding inhibition in IG2 versus untreated net must be the alpha-cypermethrin rather than the chlorfenapyr. Further, the meta-analysis of relative risk of the comparison of IG2 versus chlorfenapyr CTN was 0.74 (0.3-2.0) (P=0.67). This, being in the same direction as the relative risk between IG1 versus untreated net (0.61, 0.39-0.97), supports the interpretation that the chlorfenapyr has no role in blood-feeding inhibition of IG2 nor antagonises the positive effect alpha-cypermethrin has on blood feeding inhibition (Fig 3b).
In the first trial, mosquito exiting rates were significantly higher in the huts with IG1, IG2 and chlorfenapyr CTN treatments compared to the huts with untreated nets (Table 1). In the second trial the exiting rates from huts with IG1, 1G2 and chlorfenapyr CTN were not significantly different from exiting rates from huts with the untreated net nor from one another (Table 1).
Meta-analysis of enhanced exiting
In the meta-analysis these differences between the first and second trials led to heterogeneity in several of the comparisons of relative risk for exiting rates between treatments. No comparison between IG2 and any other treatment (untreated net, alpha-cypermethrin net, chlorfenapyr net) was significantly different from unity (Fig 3c).
Anopheles gambiae s.l.
Abundance of An. gambiae was very low in trial 1 with only 42 mosquitoes collected from the six treatments over 54 nights. However, differences in mortality were observed at 72 h with significantly higher mortality observed in huts with unwashed IG2 and IG2 washed 20 times (14/16) compared to IG1 (4/11) or untreated nets (1/10) (Table 4), which is consistent with the An. funestus dataset trends. Insufficient An. gambiae were collected during trial 2 for formal analysis.
The mean alpha-cypermethrin content in unwashed IG2 for trial 2 (the WHO trial) was 2.81 g/kg (Table 5). The nets complied with the target dose of 2.4 g/kg ± 25% for 100 denier yarn. The mean chlorfenapyr content in unwashed IG2 for trial 2 was 5.22 g/kg. The nets complied with the target dose of 4.8 g/kg ± 25%. The within-net variation showed an acceptable homogeneity of active ingredient within the nets. After 20 washes the IG2 alpha-cypermethrin content for trial 2 was 1.65 g/kg, corresponding to an overall alpha-cypermethrin retention of 59%. The chlorfenapyr content was 1.66 g/kg after 20 washes, corresponding to an overall chlorfenapyr retention of 32% for trial 2. Netting samples were not kept back pre-washing in trial 1 for chemical analysis and therefore retention of chlorfenapyr and alpha-cypermethrin in IG2 after washing could not be accurately estimated. However, chemical analyses were conducted after the nets had been washed and tested in the huts and data was consistent with trial 2 post-trial retention estimates (see Table 5). The mean alpha-cypermethrin content in unwashed IG1from trial 2 was 5.55 g/kg. The alpha-cypermethrin content after twenty washes was 1.59 g/kg, corresponding to alpha-cypermethrin retention of 30% in IG1.
Supporting bioassay tests on Interceptor and Interceptor G2 nets used in the hut trials
The purpose of the supplementary bioassays was to sample netting from the IG2 and IG1 used in the experimental hut trials to 1) test bioefficacy against pyrethroid resistant (Zenet) and susceptible (Kisumu) strains in mosquito bioassay, 2) confirm the bioefficacy of alpha-cypermethrin and chlorfenapyr components after multiple washing, 3) examine the capacity of tunnel tests to predict the performance IG2 netting under simulated hut conditions to control An gambiae s.s.
Standard WHO Cone bioassay tests on nets with 3 min exposure of the susceptible strain and a 72h holding period, induced mortality of 96% and 100% on the unwashed IG1 and the IG1 washed 20 times. With the chlorfenapyr CTN, mortality was 90%, 95% and 100% after 24h, 48h and 72h. For the unwashed IG2, mortality was 100% after 24h exposure. For IG2 washed 20 times mortality was 62%, 72% and 86% after 24h, 48h and 72h intervals (Fig 2a).
In further supplementary 3 min cone tests using the susceptible strain, mortality was 100% on unwashed IG1 and IG2, and on the IG1 and IG2 washed 20 times mortality was reduced to 82% and 86% respectively. With the Zenet pyrethroid resistant strain, cone mortality was reduced to 16% and 40% with unwashed IG1 and IG2 respectively, and 16% and 20% respectively after 20 washes (Fig 2b).
Supplementary tunnel tests were conducted using susceptible and resistant strains tested on unwashed IG2 and IG2 washed 20 times (Fig 2c). With untreated netting, 100% of the susceptible and 86% of the resistant mosquitoes penetrated the holes into the baited chamber, 100% of the susceptible and 69% of the resistant mosquitoes blood-fed, and 2% of the susceptible and 2% of the resistant mosquitoes died. With unwashed IG2 netting fewer of the susceptible (89%) and resistant (38%) mosquitoes penetrated the holes, and even fewer susceptible (70%) and resistant (0%) mosquitoes blood-fed. However, 98% of the susceptible and 36% of the resistant mosquitoes were killed by the unwashed IG2 their attempts to feed. With the IG2 washed 20 times, a smaller percentage of the susceptible (65%) and resistant (20%) mosquitoes penetrated the holes (surprisingly), fewer susceptible (18%) and resistant (0%) blood-fed, and yet 100% of susceptible and 26% of resistant mosquitoes were killed by the IG2 washed 20 times. The newcomer Zenet strain was evidently less well adapted to the tunnel test, penetrating holed netting and responding/feeding on guinea pigs less well than did the long-established Kisumu.
Comparison of supplementary bioassay tests with hut trial results
Comparing the laboratory cone and tunnel bioassay results against the pyrethroid resistant An gambiae s.s. strain and the experimental hut results against the wild pyrethroid resistant An funestus population, both types of bioassay predicted the response in the hut to the pyrethroid-only IG1: mortality was 16% in the cone and 13% in the hut against the unwashed IG1, and 16% in the cone and 11% in the hut against the IG1 20 times washed (averaged control-corrected mortality). When tested against the unwashed IG2, mortality was 40% in the cone, 36% in the tunnel and 51% in the hut; when tested with the 20 times washed IG2 mortality was 20% in the cone, 26% in the tunnel and 46% in the hut.
With the unwashed and washed IG2, percentage passage and percentage blood-feeding in the tunnel test were significantly lower with the newly-colonised resistant Zenet strain as compared to the long-established susceptible Kisumu strain. While up to 70% of Kisumu blood-fed after penetrating the IG2 netting, none (0%) of the Zenet strain blood-fed through IG2. And while high mortality of Kisumu (up to 70%) was recorded with IG2, low mortality was recorded against unwashed and washed IG2 (36% and 26% respectively). This very much reflected the new adaptation of the Zenet strain to the tunnel test, possibly an avoidance or irritation of the treated net, or ‘reluctance’ to feed on guinea pigs. However, for those Zenet strain mosquitoes that did penetrate the netting, mortality inflicted by unwashed and 20 times washed IG2 was high, 58% and 80% respectively, and more closely resembled mortality in experimental huts.
This series of bioassay tests demonstrates that the chlorfenapyr component of IG2 LN makes the major contribution to controlling pyrethroid resistant An gambiae and An funestus. The tunnel tests were more predictive of efficacy in experimental huts whilst cone bioassays were less predictive.