Reduced Efficacy of Insecticide-treated Nets and Indoor Residual Spraying for Malaria Control in Pyrethroid Resistance Area, Benin

These tools may no longer be effective for malaria control in parts of Benin.

D uring the last decade, pyrethroid-treated mosquito nets have become the main method of malaria prevention in many malaria-endemic African countries (1,2). In a few notable exceptions, usually those with a more developed health infrastructure, such as South Africa, a longstanding practice of applying indoor residual spraying (IRS) has been successful (3). The 2 approaches to malaria prevention, insecticide-treated nets (ITNs) and spraying (IRS), are not mutually exclusive, and in malaria-endemic areas where ITN coverage is still limited, the feasibility of introducing IRS to reduce transmission is being considered, for example, by the President's Malaria Initiative Fund (4). Trials of IRS and ITNs have shown that in areas with pyrethroid-susceptible Anopheles gambiae the effectiveness of the 2 methods in controlling malaria does not differ (5). This comparability may not hold true for areas with pyrethroid-resistant populations. In southern Africa, for example, IRS with pyrethroid failed to control pyrethroid-resistant An. funestus and necessitated a switch to an alternative class of insecticide to which there was no resistance (6). During the last decade, pyrethroid resistance caused by the kdr mechanism has become widespread in An. gambiae in West Africa and is common in some areas (7). Whether kdr undermines the effectiveness of ITN in areas of high prevalence is unclear. An early experimental hut trial of ITNs in Côte d'Ivoire demonstrated a survival advantage of homozygotes for kdr resistance (8), whereas subsequent hut trials in adjacent resistant and susceptible populations showed no apparent difference in the effectiveness of ITNs between the 2 localities (9). Village randomized trials in Côte d'Ivoire showed that ITNs continued to prevent malaria despite a vector population that was kdr resistant (10). Whether kdr would undermine the effectiveness of IRS in the same way as resistance due to oxidases did against An. funestus in southern Africa (6) is unknown. To assess the practicability of applying IRS with pyrethroid in West Africa, we need to examine the effectiveness of this approach against a kdr-resistant population of An. gambiae. To get a clearer understanding of the influence of kdr resistance on the effectiveness of ITN, further experimental hut trials of ITNs against kdr-resistant populations need to be conducted. We describe 2 experimental hut trials in Benin. One compares the impact of IRS and ITN against a pyrethroid-resistant population in the southern part of the country; the other compares IRS and ITNs against a pyrethroid-susceptible population several hundred kilometers to the north.

Study Sites
Ladji is a large village on the outskirts of Cotonou, the capital of Benin. The village floods during the rainy season. An. gambiae comprises the Mopti (M) cytotype and shows resistance to pyrethroids and DDT; kdr is present at high frequency (11). The nuisance mosquito Culex quinquefasciatus is also present and shows resistance to pyrethroids. Five experimental huts belonging to the Centre de Recherche Entomologique de Cotonou (CREC) are situated in the village.
Malanville is in northern Benin, 800 km from Cotonou, in an irrigated rice-growing valley. The local An. gambiae comprises the M cytotype, but the kdr gene is almost absent and mosquitoes are susceptible to lambdacyalothrin and deltamethrin. Six experimental huts are present at Malanville.

Experimental Huts
The treated nets, residual spray treatments, and their respective untreated controls were evaluated in 4 experimental huts at each field site. Experimental huts are specially designed to test vector control products against freely entering mosquitoes under natural but controlled conditions. Huts were typical of the region. Each was made from concrete bricks, with a corrugated iron roof and a ceiling of thick polyethylene sheeting lined with hessian sackcloth on the interior surface, and each was built on a concrete base surrounded by a water-filled moat to exclude ants (12). Mosquito access was through 4 window slits, constructed from pieces of plywood fixed at an angle to create a funnel with a 1-cm gap, present on 3 sides of the huts. Mosquitoes had to fly upward to enter through the gaps and downwards to exit; this precluded or limited exodus through the aperture and enabled us to account for most entering mosquitoes. A veranda trap projected from the back wall of each hut. Movement of mosquitoes between a room and the veranda was unimpeded.

Mosquito Net Treatments
The nets were made of white, 100-denier polyester (SiamDutch Mosquito Netting Co., Bangkok, Thailand). Nets measured 2.0-m long, 1.6-m wide, and 1.8-m tall and had a surface area of 16.9 m 2 . To simulate badly torn nets, 80 holes, each measuring 2 × 2 cm, were cut in the sides and ends of each net.
The lambdacyhalothin application rates of 18 mg/m 2 for ITNs and 30 mg/m 2 for IRS were within the ranges recommended by the manufacturer. Indoor residual treatments were applied with a hand-operated compression sprayer equipped with a flat fan nozzle. The cement walls and sackcloth ceilings were sprayed uniformly after masking the veranda and window slits with protective coverings. The control hut was sprayed with water only. The treated huts were left for 1 week before evaluations were started.

Sleepers and Mosquito Collections
Preliminary experiments showed the huts to be evenly attractive to mosquitoes. The treatments were randomly allocated to the 4 experimental huts at each site. The main trials were conducted from April to June 2005 at the Ladji site and from September to November 2005 at the Malanville site. Eight adult men employed by CREC slept overnight in the huts and collected mosquitoes from the huts in the mornings. Informed consent to participate in the study was given beforehand, and chemoprophylaxis was provided during the trial. Ethical approval was granted by the London School of Hygiene and Tropical Medicine (LSHTM) and Benin national ethics committees.
The trial ran for 50 nights for 8 weeks at each site. The sleepers were rotated between huts to correct for possible variation in individual attractiveness. Each morning, mosquitoes were collected from the floors, walls, and ceilings of rooms, verandas, and nets with aspirators and torches. Mosquitoes were identified and scored as blood-fed or unfed and dead or live. Live mosquitoes were held in netted plastic cups and supplied with 10% honey solution for 24 h before delayed mortality was recorded. Male mosquitoes were not scored.
The entomologic impact of each treatment on mosquitoes was expressed relative to the control in terms of the following: deterrence, the proportional reduction in the number of mosquitoes entering a treated hut relative to that entering the control hut; induced exophily, the proportion of mosquitoes collected from the veranda trap of the treatment hut relative to the proportion in the veranda of the control hut; blood-feeding inhibition, the reduction in blood-feeding rate relative to the control hut; and mortality, the proportions of mosquitoes found dead in the hut at the time of collection and after a 24-h holding period.
If a treatment deters a considerable number of mosquitoes from entering the hut, the values given by proportion blood-feeding or proportion killed in the treatment hut may underestimate the full personal protective effect and overestimate the full insecticidal efficacy of the treatment. The personal protective effect of a treatment is better described by the reduction in the number of blood-fed mosquitoes in the treatment hut relative to the number blood-fed in the control hut: where B u is the total number of blood-fed mosquitoes in the untreated control huts and B t is the total number bloodfed mosquitoes in the huts with insecticide treatment.
The overall insecticidal effect of a treatment needs to take into account that a considerable number of mosquitoes might be deterred from entering the hut and hence not be killed by the treatment. A mass killing effect is desirable to reduce transmission. The overall insecticidal effect of a treatment relative to the number of mosquitoes that would ordinarily enter an untreated hut can be estimated by using the following formula and expressed as a percentage: where K t is the number killed in the treated hut, K u is the number dying in the untreated control hut, and T u is the total number collected from the control hut.

Residual Activity of Insecticide Treatments
To evaluate residual activity, World Health Organization (WHO) cone bioassays were undertaken monthly in the Ladji huts and bimonthly in the Malanville huts with a laboratory-susceptible strain of An. gambiae (Kisumu). An. gambiae females, 3-5 days old, were exposed within the cones to nets for 3 min or to sprayed walls and ceilings for 30 min. Approximately 50 mosquitoes in 5 replicates of 10 mosquitoes were tested on each substrate. Honey solution was provided during the 24-h holding period, and the temperature was kept at 25°C.

Biochemical Assays
Biochemical tests on individual mosquitoes were conducted to determine the activity of mixed function oxidases and nonspecific esterases present in pyrethroid-resistant and -susceptible samples of An. gambiae from the Ladji and Malanville sites. Tests were conducted on 3-day-old adult females (initially collected as larvae) in microtiter plates (13). Susceptible (Kisumu) and pyrethroid-resistant (Vkper) An. gambiae served as controls. Genotyping of An. gambiae was carried out to assess kdr frequency at both field stations (14).

Adult Bioassay Data
To determine whether a stronger pyrethroid resistance mechanism was present in the Ladji population than in the standard kdr strain Vkper, bioassays with 0.05% lambdacyalothrin-treated papers (18 mg/m 2 ) were conducted in WHO resistance test kits by using a range of exposure times on batches of 25 unfed An. gambiae females 2-5 days of age. One hundred mosquitoes per exposure period were tested. Deaths were scored 24 h later. Log-time mortality curves were generated, and lethal time to kill 50% (LT 50 ), estimated by using probit analysis.

Data Analysis
Proportional data from the hut trial (exophily, bloodfeeding, deaths) were analyzed by using logistic regression (STATA 6 software, Stata Corporation, College Station, TX, USA). Deterrence rates were analyzed by comparing the number of mosquitoes entering each hut by using the Wilcoxon rank sum test. Biochemical activity was analyzed with Kruskal-Wallis and Wilcoxon rank sum tests. The level of resistance to lambdacyalothrin in insecticide bioassays was analyzed by using probit analysis.

Insecticide Residual Activity
Residual activity on ITN as measured by cone bioassay tests showed no decline during the 8 weeks of the trial. Activity of the IRS wettable powder formulation on sackcloth and cement showed a decline in performance by week 4. This trend continued until the end of the trial ( Table 1).

Efficacy of Treatments in Huts
Over the 2-month trial, 1,395 An. gambiae, 3,070 Cx. quinquefasciatus, and small numbers of Mansonia uniformis, An. pharoensis, and Aedes aegypti were collected at Ladji. At Malanvile, 1,523 An. gambiae, 2,804 Mansonia sp., and smaller numbers of An. funestus and Ae. aegypti were collected. Only the malaria vector An. gambiae and the nuisance mosquito Cx. quinquefasciatus were analyzed further.
Fewer An. gambiae entered the ITN-and IRS-treated huts than the respective control huts. The treatment induced reduction in hut entry was more evident in the resistance area than in the susceptible area ( Table 2). The proportion deterred at each site did not differ between ITN or IRS treatments.
The untreated net was little or no barrier to bloodfeeding of An. gambiae at either field site owing to the large number of holes cut in each net. Treating the holed net with pyrethroid led to a 96% reduction in the number of mosquitoes blood-feeding at the susceptible site (Malanville) but to no reduction in blood-feeding at the resistant site (Ladji). Inhibition of blood-feeding by IRS at either the resistant or susceptible site was limited ( Table 2).
Natural mortality of An. gambiae occurred in both types of control huts but was notably higher at Ladji than at Malanville. Both modes of treatment were highly insecticidal at Malanville: ITNs treated with 18 mg/m 2 lambdacyhalothrin killed 99%, and IRS applied at 30 mg/m 2 killed 72% of An. gambiae that entered the huts. At Ladji, the proportions of An. gambiae killed in either the ITN-or IRS-treated hut did not exceed 30% ( Table 2).
The proportion of An. gambiae collected from the veranda traps in the mornings was greater at Malanville than at Ladji and greater in the huts with untreated nets than in the unsprayed control huts. Relative to the controls, lambdacyalothrin-treated nets and IRS induced little or no exophily of the pyrethroid-resistant An. gambiae into the verandas of the Ladji huts, despite high survival rate of mosquitoes in huts. At Malanville, pyrethroidinduced exophily by ITN or IRS hut was not evident and may have been obscured by the high death rates among the mosquitoes.
The personal protection derived from ITN was almost 100% in the susceptible area. Despite the low mortality rate and high rate of blood-feeding observed with ITN in the resistance area, the level of personal protection there was almost 50% because of the deterrent effect of lambdacyhalothin on mosquito entry into huts. The personal protective effect of IRS was low in both areas, and IRS was no barrier to blood-feeding. The overall insecticidal effect of pyrethroid-treated nets and IRS was negligible in the resistance area (<5.4%) but was considerable in the susceptible area (>55.8%). Table 3 breaks down the mortality data into 2-week blocks. Mortality associated with IRS treatments decreased week by week at both sites but started at a lower rate at the Ladji site because of the expression of resistance. Mortality associated with ITN treatments also showed a downward trend over time at Ladji but not at Malanville, where mosquitoes showed high susceptibility throughout the study.
Both ITN and IRS treatments at Ladji showed poor efficacy against Cx. quinquefasciatus (this species was not encountered in Malanville). Insecticide-induced deterrence was greater for ITN than for IRS (Table 4). Neither method killed many Culex nor stimulated repellency into verandas. The IRS treatment produced an unusually high level of blood-feeding inhibition.

Biochemical Assays and kdr Genotyping
An. gambiae from Ladji expressed a significantly higher level of oxidase activity than the standard susceptible (Kisumu) and the laboratory kdr (Vkper) strains, which had a similar level of oxidase activity. However, the pyrethroid-susceptible strain from Malanville showed a level of oxidase activity that was not significantly different from that of the Ladji strain. This finding would appear to rule out any contribution from oxidases to the pyrethroid resistance observed in An. gambiae from Ladji. The level of α-esterase activity in An. gambiae from Ladji was significantly higher than that expressed in Malanville or Kisumu strains, whereas the level of β-esterase activity in Ladji, Vkper, and Kisumu strains was similar and clearly played no part in resistance (Table 5). Overall, the mean level of esterase activity at Malanville was significantly lower than that of the susceptible reference strain (p<0.05). Genotyping data (Table 6) showed a high frequency of kdr resistance at Ladji (F [kdr] = 83%, n = 45) and low frequency at Malanville (F [kdr] = 6%, n = 45). The pyrethroid-resistant Vkper was fixed for the Kdr gene (F [kdr] = 100%, n = 47).

Adult Bioassays
The summary results of the exposure time mortality bioassays with lambdacyhalothin-treated papers in WHO cylinder kits are shown in Table 6. The slopes and LT 50 s of the probit regression curves were not significantly different for Ladji and Vkper strains. Tests on the Kisumu strain produced 100% mortality after only 1 min exposure. An LT 50 could not be calculated by using probit analysis, but the resistance factor in the Ladji and Vkper strains was at least 10-fold.

Discussion
A major loss of efficacy associated with pyrethroid resistance occurred in An. gambiae at Ladji, Benin. The reduction in efficacy affected IRS and ITNs equally: only 19% of mosquitoes in the ITN hut and only 22% in the IRS hut were killed after correction for natural mortality. By contrast, 98% of mosquitoes entering the ITN hut and 72% entering the IRS hut located in the susceptible north of Benin were killed by the lambdacyhalothin treatments after correction for natural mortality. These findings are the first clear evidence of pyrethroids' failing to control an An. gambiae population that contains kdr resistance at high levels. Whereas the loss of insecticidal effect was calculated to be >95%, a degree of personal protection associated with ITNs and IRS was still evident (45%-50%) relative to the untreated net or unsprayed hut owing to a partial deterrent effect of treatments on entry of mosquitoes rather than to any inhibition of blood-feeding once the insects were inside the huts. Indeed, on entering the huts, most mosquitoes did go on to blood-feed, and the deliberately holed ITN was no barrier to resistant mosquitoes. By contrast, in northern Benin, only 4% of the insecticide-susceptible mosquitoes that entered the hut fed through the holed ITN. The loss of personal protection and loss of mosquito mortality associated with resistance would presumably com-bine to make ITNs unattractive from the perspective of both the individual user and the malaria control manager. Incision of 80 holes per net is the standard for ITN trials in West Africa (8,9,12), and such nets have given a degree of personal protection in earlier trials. An ITN with no or few holes might be expected to give some protection against resistant mosquitoes from Ladji, but there were insufficient huts available to test this idea.
We sought evidence that other resistance mechanisms than kdr might be contributing to the reduced efficacy of pyrethroids at Ladji. Metabolic resistance due to mixed function oxidases (MFO) has, for example, undermined attempts at malaria control with deltamethrin residual spraying in southern Africa caused by An. funestus (6), and elevated MFO activity in a strain of An. gambiae from Cameroon reduced the efficacy of permethrin-treated netting in laboratory tests (20). The combined elevated activity of MFOs, glutathione S-transferase, and esterases resulted in a failure of the S. Mexican IRS program against An. albimanus (21). Our examination of enzymatic activi-ty in An. gambiae showed no evidence that MFO activity is any greater in mosquitoes from Ladji than in mosquitoes from Malanville, nor did esterase activity differ between Ladji and Vkper (kdr) strains. Thus, there was no evidence of metabolic resistance enhancing the resistance already caused by kdr in mosquitoes from Ladji. Lambdacyhalothin bioassay tests showed no evidence of resistance level differing between Ladji and Vkper strains, and we conclude that metabolic mechanisms made no contribution to the observations in Ladji.
In East Africa a different type of kdr based on a leucine-to-serine mutation, which confers resistance to permethrin and DDT (22), has been detected in several countries. However, no mosquitoes of this genotype were detected in tests on samples of An. gambiae from Ladji (23). The complete absence of efficacy of lambdacyalothrin against Cx. quinquefasciatus in Ladji merely confirms earlier findings involving other types of pyrethroid in experimental huts in West Africa (6,9,16,18).
The contribution of kdr to pyrethroid resistance in An. gambiae needs to be reappraised. While lambdacyhalothin-treated nets (reported here) and permethrin-treated nets reported earlier (24) were less effective in hut trials in the kdr area of Benin (Ladji) than in a corresponding area of Côte d'Ivoire (Yaokoffikro), pyrethroid-treated nets were more effective in the susceptible area of Benin (Malanville) than in the corresponding susceptible area of Côte d'Ivoire (M'Be) (9) for reasons that are presently unknown. Other differences between the biology of An. gambiae from Côte d'Ivoire and Benin exist. Ivorian An. gambiae with kdr is mainly of the S molecular form, whereas Beninoise An. gambiae is of the M form (V. Corbel, unpub. data). M and S forms differ in ecologic distribution and habitat. While mosquitoes of the M form with kdr might behave differently from those of the S form with kdr when exposed to pyrethroids, this is mere speculation. Moreover, the M form in Malanville showed higher vul-nerability to ITN than did the corresponding S form in Côte d'Ivoire, a finding that seems at odds with a behavioral hypothesis.
Our study provides persuasive evidence that pyrethroid resistance in Benin is capable of undermining control measures based on ITN. Nor is there reassurance to be taken from IRS, and any attempt to switch vector control strategy would seem doomed to fail. Whereas the earlier phase 3 malaria control trials of ITN in Côte d'Ivoire showed continuing effectiveness despite kdr at high levels (10), our phase 2 results from Benin give no grounds for optimism. However, only phase 3 can provide a definitive answer. Further phase 3 trials using pyrethroid-treated nets and IRS need to be undertaken in Benin in an area of pyrethroid resistance. The normal practice with phase 3 is to aim at complete community coverage. Coverage in real life is usually less than total, and the danger with the type of pyrethroid resistance found in Benin is that at lower levels of coverage the important mass protective effect of ITNs (25,26) may be lost and transmission may continue unabated among those who do not have ITNs. To establish whether this is true, phase 3 trials on resistant mosquito populations should ideally set the coverage level at <100%. If it is considered unacceptable to deny a section of the trial population access to ITNs, an alternative but much less rigorous approach would be to monitor malaria incidence among users and nonusers of long-lasting insecticide nets (LLIN) during the proposed scaling up of LLIN coverage in Benin currently being considered.
Pyrethroid resistance in Benin is far from homogeneous, and LLIN should give good protection wherever mosquito populations are susceptible. Use of LLIN should be encouraged but scale-up of treated nets may ultimately select for further resistance. The need to develop alternative insecticides to replace or supplement pyrethroids on nets is urgent and should be put on a par with the seeking of new antimalarial drugs or vaccines that have received far greater attention and resources in recent years.